Árpád Furka

Combinatorial Chemistry

Principles and Techniques

Árpád Furka

Combinatorial Chemistry

Principles and Techniques

Published by Árpád Furka in electronic form Budapest 2007

© Árpád Furka, 2007

v

Preface

Combinatorial technologies that were invented in 1980s provided a possibility to produce new compounds in practically unlimited number. New strategies and technologies have also been developed that made possible to screen very large number of compounds and to identify useful components of mixtures containing millions of different substances. This dramatically changed the drug discovery process in the pharmaceutical industry and the way the researchers design their experiments. Instead of preparing and examining a single compound, families of new substances are synthesized and screened. In addition, combinatorial thinking and practice proved to be useful in areas outside the pharmaceutical research. Such area are, for example, search for more effective catalysts and materials research. Combinatorial chemistry became an accepted new branch within chemistry. It is the subject of numerous books, journals, international conferences and university courses. This book is written for university students and young researchers. The author feels it important to make it freely available for all potential readers. For this reason the book will be published exclusively in electronic form that can be downloaded from appropriate Web sites free of charge. The author wishes to express his appreciation to Dr. József L. Margitfalvi of the Central Chemical Research Institute of the Hungarian Academy of Sciences and Dr. György Kéri of Gedeon Richter Ltd, Budapest for reading parts of the manuscript and for their valuable suggestions. The mother tong of the author is Hungarian. Despite all efforts the text obviously contains grammatical errors. Correction of these errors, of course, would be important. The help of the readers in this respect would be highly appreciated. If you can help please contact the author by e-mail: afurka@szerves.chem.elte.hu. Árpád Furka

vi

vii

Table of Contents

Preface……………………………............................................................................ v Table of Contents....................................................................................................... vi 1. Introduction……………………………………………………………....…........ 1 1.1. Birth of the combinatorial approach................................................................... 2 1.2. The translated version of the document notarized in 1982................................. 5 1.3. Publication of the split-mix combinatorial synthesis.......................................... 13 References.................................................................................................................. 14 2. The solid phase synthesis....................................................................................... 15 2.1. Solid supports...................................................................................................... 18 2.1.1. Crosslinked polystyrene................................................................................... 18 2.1.2. Polyethylene glycol (PEG) grafted supports.................................................... 19 2.1.3. Inorganic ports................................................................................................. 20 2.1.4. Non-bead form supports.................................................................................. 20 2.2. Linkers, anchors.................................................................................................. 20 2.3. Protecting groups................................................................................................ 22 2.3.1. Protection of amino groups.............................................................................. 23 2.3.2. Protection of carboxyl groups.......................................................................... 24 2.3.3. Protection of other functional groups.............................................................. 24 2.3.4. Coupling reagents for peptide synthesis.......................................................... 25 2.4. Solid phase synthesis of organic molecules........................................................ 26 2.5. Solid phase reagents and scavenger resins in solution phase synthesis.............. 27 References.................................................................................................................. 28 3. Parallel synthesis. Synthesis of compound arrays based on saving reaction time............................................................................................... 29 3.1. The parallel synthesis.......................................................................................... 30 3.1.1. The multipin metod of Geysen........................................................................ 31 3.1.2. The SPOT technique of Frank......................................................................... 32 3.1.3. Other devices for parallel synthesis................................................................. 33 3.1.4. Parallel synthetic methods with reduced number of operations...................... 36 3.1.4.1. Synthesis of oligonucleotides on paper discs................................................ 36 3.1.4.2. The tea-bag synthesis.................................................................................... 37 3.2. The Ugi multicomponent reactions..................................................................... 37 3.3. Solution phase combinatorial synthesis.............................................................. 39 3.3.1. Dendrimer supported synthesis........................................................................ 39 3.3.2. Separations using fluorous tags and fluorous solvents.................................... 40 3.3.3. Application of solid phase reagents................................................................. 40 3.3.4. The use of scavengers in solution phase reactions........................................... 41 3.4. Automation in parallel synthesis......................................................................... 42

viii

3.4.1. Automatic parallel synthesizers....................................................................... 42 3.4.2. Quality control................................................................................................. 45 3.4.3. Parallel purification.......................................................................................... 47 3.4.4. Manufacturers of laboratory robots................................................................. 48 References.................................................................................................................. 52 4. Combinatorial synthetic methods.......................................................................... 55 4.1. Combinatorial synthesis on bead-form resin...................................................... 55 4.1.1. The split-mix synthesis.................................................................................... 55 4.1.1.1. The key features of the split-mix synthesis.................................................. 56 4.1.1.2. Encoding of beads in the synthesis of organic libraries............................... 61 4.1.1.3. Realization of the split-mix synthesis.......................................................... 64 4.1.1.4. Automation of the split-mix synthesis......................................................... 66 4.1.1.5. Preliminary considerations when planning experiments with peptide libraries......................................................................................................... 71 4.1.1.6. Full and partial libraries............................................................................... 75 4.1.1.7. Unusual partial libraries............................................................................... 79 4.1.1.8. Binary synthesis using the split-mix procedure........................................... 80 4.1.2. Combinatorial synthesis using amino acid mixtures....................................... 82 4.2. Combinatorial synthesis using soluble support.................................................. 83 4.3. Combinatorial synthesis on solid surface........................................................... 84 4.4. Combinatorial peptide synthesis by biological methods.................................... 86 4.5. Combinatorial synthesis using macroscopic solid support units........................ 87 4.5.1. Encoding by attached labels. The radiofrequency and optical encoding methods............................................................................................ 88 4.5.2. Units without labels. Encoding by position in space....................................... 92 4.5.2.1. The Encore technique................................................................................... 92 4.5.2.2. The String Synthesis..................................................................................... 93 4.5.2.3. String synthesis of cherry picked libraries................................................... 104 4.6. Examples............................................................................................................ 110 4.6.1. Split-Mix Synthesis of an encoded benzimidazole library............................. 110 4.6.2. Synthesis of a 10,000 member piperazine 2-carboxamide . library by Directed Sorting.............................................................................. 115 4.6.3. Synthesis of two libraries on one support........................................................ 117 References.................................................................................................................. 118 5. Screening methods................................................................................................. 121 5.1. High throughput screening of arrays of individual compounds.......................... 122 5.2. Screening of combinatorial libraries. Deconvolution methods........................... 124 5.2.1. Deconvolution methods for dissolved libraries............................................... 124 5.2.1.1. The iteration method..................................................................................... 124 5.2.1.2. Positional scanning....................................................................................... 129 5.2.1.3. Omission libraries......................................................................................... 133

ix

5.2.1.4. The amino acid tester libraries...................................................................... 135 5.2.1.5. Other methods for identification of the bioactive component of combinatorial libraries.................................................................................. 138 5.2.1.6. Dynamic combinatorial libraries................................................................... 138 5.2.1.7. Examples....................................................................................................... 138 5.2.2. Deconvolution methods of libraries tethered to the solid support................... 145 5.2.2.1. Screening of combinatorial libraries in tethered form.................................. 146 5.2.2.2. Screening of combinatorial libraries by releasing the content of individual beads intosolution........................................................................ 147 5.2.2.3. Examples...................................................................................................... 148 References................................................................................................................. 150 6. Combinatorial methods in materials and catalyst research................................... 151 6.1. Inorganic materials............................................................................................. 151 6.1.1. Preparation of thin film libraries...................................................................... 151 6.1.2. Screening......................................................................................................... 154 6.2. Heterogeneous catalysts...................................................................................... 155 6.2.1. Fabrication and testing of catalyst libraries..................................................... 156 6.2.2. Catalyst library design..................................................................................... 161 6.3. Polymers............................................................................................................. 164 References................................................................................................................. 168 7. Computational aspects of library design and synthesis......................................... 171 7.1. Software companies............................................................................................ 176 References.................................................................................................................. 182 Index.......................................................................................................................... 183

1

1. Introduction The discovery of new materials played an important role in the history of mankind. Many

discovered materials had effect on every day’s life. The impact of some of these materials was so definitive that they gave the name of long historical eras. So bronze gave the name for Bronze Age, for example, and iron for the Iron Age.

The life today is also largely affected by the materials we use. The standard of life could not be the same without semiconductors, insulators, adhesives, synthetic fibers, drugs, pesticides, paints etc. In order to improve our life, more and more useful materials and compounds need to be discovered. The question is how to do that? When we need a new bridge or want to build a skyscraper, for example, first these objects are designed then they are built according to the plans. Can we follow this route when we wish to make a new super conductor or a new drug? Certainly not. Our theoretical knowledge may be sufficient for designing a bridge or a skyscraper but is definitely not enough for designing a new more effective drug or designing a super conductor working at or near room temperature. We do not know exactly how the super conducting or other important properties of materials depend on their structure. The drugs exert their effects by interactions with proteins or other molecules found in living organisms. The rules governing these interactions, however, are largely unknown. The rational design of drugs had some successes. The drug candidates are designed in computers based on the already known three dimensional structures of target proteins. Both the ligand molecule and the protein itself can take up a practically unlimited number of conformations, and that leads to difficulties. The consequence is that mostly the traditional approach is followed: series of compounds are synthesized then the useful drug candidates are identified by trial and error. In practice, thousands of compounds are needed to be prepared and tested in order to find a drug candidate.

In the pharmaceutical research one of the bottlenecks was the synthesis of the very large number of compounds needed in the discovery process. Before 1980 the traditional approach was used. The compounds were prepared one at a time, and their testing were also carried out one by one. In the industry, however, sophisticated methods were developed and applied in order to improve productivity in the mass production of goods. It seems worthwhile to compare the production of compounds to that of automobiles. Compounds are mostly prepared step by step from the starting materials. The automobiles are also assembled from parts. The drug candidates are unique substances all differing from each other. The automobiles are also unique products since they can differ, for example, in their color, in their engines, in their transmission etc. They certainly differ from each other in their locks and keys.

The first car manufacturers in the world were Panhard & Levassor in 1889 and Peugeot in 18911. These French manufacturers did not standardize their car models, each car was different from the other. The first standardized car was the Benz Velo. Benz manufactured 134 identical Velos in 1895.

Ransome Eli Olds invented the basic concept of the assembly line in 1901 that was improved Henry Ford and installed it in his car factory in 1913. As a result, by 1927, 15 million Ford Model Ts had been manufactured. As a result of further improvements and application of

2

automation, today the streets are full of cars. This shows the power of organizing the process of production and application of automation. These methods that proved to be very successful in industry were not applied at all in the mass production of compounds. After 1980 the situation began to change. Several innovative papers were published which radically changed our theory and practice in designing and preparing new substances for pharmaceutical research and other areas of application. The new synthetic and screening procedures and, which is also very important, the new way of thinking introduced in these papers founded a rapidly growing new scientific field, Combinatorial Chemistry, revolutionized the pharmaceutical research and are gradually expanding to other areas within and outside chemistry. The new methods were developed in several laboratories. The way of thinking that led to these methods was probably different in all cases. The reasons that lad to the development of the combinatorial synthesis of peptides in the author's laboratory is described below. 1.1. Birth of the combinatorial approach

In 1964/65 I was a post doctoral fellow at the University of Alberta, Canada. I worked on

a project led by Professor L. B. Smillie which resulted in determination of the amino acid sequence of a pro-enzyme, chymotrypsinogen-B2. After returning to Budapest, I was wondering from how many sequence possibilities did we choose the right one. Since the protein comprised 245 amino acid residues and any of these positions could be occupied by any of the 20 different natural amino acids, the number of possible sequences, as expressed by a simple formula, amounted to 20245 (=5.65x10318) combinations. This certainly seemed to be a very big number, much-much bigger than, for example, the number of molecules in 1 mole substance (6.02x1023) but in order to really perceive its magnitude it had to be compared to something which was also very big. Finally I found an estimate of the mass of the universe based on an Einstein formula3. I also calculated the mass of a protein mixture in which each sequence variant of the 245 amino acids is represented by only a single molecule.

Estimated mass of the universe: ~1053 kg

Mass of the protein mixture: ~10295 kg The comparison showed that the mass of the protein mixture would exceed that of the

universe by more than two hundred orders of magnitude. The number of possible protein sequences also seems striking if it is compared to the estimated number of elementary particles in the visible universe4.

Number of sequences in the mixture: 5.65x10318

Number of elementary particles: ~1088 This was my first meeting with the immense diversity of molecules and the result shocked

me. Many years later I applied the same simple formula (20n, where n is the number of amino acid residues) to calculate the number of theoretically possible sequences in peptide families built up from the 20 natural amino acids. Such collections of compounds are named libraries. The results are listed in the second column of Table 1.1.

3

Although the figures expressing the number of components in peptide libraries were far from being as frightening as the number of the possible protein sequences, they seemed still very large if the possibility of their synthesis was considered. I thought that many useful bioactive peptides could - supposedly - be found among the largely unknown components of the libraries. For this reason the nonexistent peptide libraries reminded me of exceptionally rich gold reefs which await exploitation. Gold can be produced by mining out all the gold containing rock then separating the gold from the useless stone.

Table 1.1. The number of possible peptide sequences.

Number of residues

Name Number of sequences

2 Dipeptides 400 3 Tripeptides 8,0004 Tetrapeptides 160,0005 Pentapeptides 3,200,0006 Hexapeptides 64,000,0007 Heptapeptides 1,280,000,000

Exploitation of the peptide libraries could be achieved via the synthesis of all possible sequences followed by screening them against all potential targets. At that time, however, even the synthesis of all, say, pentapeptides seemed absolutely impossible. We usually prepared one peptide at a time mostly by solid phase synthesis (see later the details of this synthetic method) with an elongation rate of one amino acid a day.

Figure 1.1. The optimized synthesis of peptides from three amino acids (A, E and R). The solid

support is represented by . With this rate, the synthesis of all the 3.2 million pentapeptides would have taken 3.2x5=16 million days, that is, 43.8 thousand years of uninterrupted work.

A A AR

E E E R R R

AA

A E R

A E R

EA RA AE EE RE AR ER RR

A E R A E R A E R

4

The synthesis could have been optimized by reducing to an absolute minimum the number of necessary coupling steps. This can be achieved by using the already prepared peptides as starting materials in the synthesis of the longer ones. This is illustrated in Figure 1.1. A peptide library is prepared by solid phase synthesis using three amino acids (A, E and R). First the amino acids are attached to the solid support (resin). Then the resin containing one of the attached amino acid is divided into three portions and the synthesis is continued with the coupling of one of the amino acids to one of the resin portions and so on. In the first step 3 couplings are carried out, exactly the number of the formed products. In the second step 9 couplings are needed and 9 dipeptides are formed on the resin samples. In general, the number of coupling steps in such an optimized synthesis of a peptide library is the same as the total number of products formed in the whole synthetic process. If the 3.2 million pentapeptides are prepared the number of coupling steps is the sum of amino acids + dipeptides + tripeptides + tetrapeptides + pentapeptides.

20 + 400 + 8,000 + 160,000 + 3,200,000 = 3,368,420

Supposing again the rate of one coupling per day in order to get the necessary time in

years, the above figure is divided by 365. The result is 9,228 years. This shows that optimization of the synthesis reduces the time of the synthetic process from 43,800 years to 9,228 years, which is still too long to be realizable. I considered the accessibility of all peptide sequences to be very important, and around 1980 I began to think about potential solutions for their synthesis. It took only a short time to find one, which would work at least in principle. The idea was also based on the method of solid phase synthesis developed by professor Merrifield5. According to this first idea, the amino acids used in the solid phase preparation of peptides would be replaced by an equimolar mixture of 20 different amino acids in every coupling step of the synthesis. This would lead - at least in principle - to formation of a rapidly growing number of sequences and finally a full peptide library could be cleaved from the support in the form of a mixture. It was clear, however, that in such couplings the products are expected to form in unequal molar quantities as a consequence of the differences in the reactivity of the amino acids. The differences in molarities would be amplified in each successive coupling step leading to a mixture with uncertain composition. I felt that a better solution might exist and I was rethinking the problem again and again. In early spring in 1982 I spent a weekend in a little town in South-East of Hungary forgetting this time the whole diversity problem. To my great surprise, however, next morning I awoke with the perfect solution in my mind. The method based on this idea is known nowadays as the split-mix procedure.

The split-mix method opened the possibility for producing peptide mixtures containing millions of components. Such mixtures, however seemed unacceptable in the conventional drug discovery practice where single compounds were used in pure form. For this reason there was an urgent need to present in addition, an efficient strategy for identification of the bioactive substance that may be present in the complex synthetic mixture. This task, however, looked similar to finding the proverbial needle in a huge haystack. Nevertheless I could develop a theoretical solution in a very short time. I called it “synthetic back searching strategy” which later proved to be in principle identical with the "iteration strategy", published by others.

I was fully aware of the importance of the combinatorial approach in the pharmaceutical research but one of the leading Hungarian pharmaceutical companies I contacted showed no interest at all. In addition, the split-mix method was considered by the patent attorneys only as a potential research tool and for this reason it was judged not to be patentable. They suggested me,

5

however, to describe the method in a document and - in order to give me some support in potential future priority disputes - notarize it. I did so and the document written in Hungarian - in which the principles of combinatorial chemistry including both synthesis and screening were first clearly explained - was notarized in May, 1982. The photo of the first and last pages of the document is demonstrated in Figure 1.2.

Figure 1.2. The photo of the first and last page of the 1982 document The 1982 document, as shown in the Figure 1.2, was written in Hungarian. This is the first authentic document in which the principles of combinatorial chemistry are described. The translated version can be seen below. 1.2. The translated version of the document notarized in 1982 STUDY ON POSSIBILITIES OF SYSTEMATIC SEARCHING FOR PHARMACEUTICALLY

USEFUL PEPTIDES

Written by Dr. Árpád Furka, university professor Budapest, May 29, 1982

As exemplified, among others, by the peptide hormones discovered so far, the shorter-

lengthier peptides take part in a number of important functions in the living organism. It can be supposed, that only a small fraction is known of these biologically active peptides having

6

potential therapeutic effect. This fact motivates the intensive international and domestic research activity in this field.

Two, in principle different, approaches offer themselves for searching for peptides bearing new biological effects:

1. Isolation of peptides from living organisms based on their previously known biological effects. 2. Preparation of peptides by synthesis with post determination of their biological effects.

Until now the isolation procedure proved to be more effective in spite of the fact that this method is also very laborious. This may be explained by the fact that the number of possible peptides grows rapidly with the number of residues so even the synthesis of all tetrapeptides (160 thousands) seems to be a hopeless task. If we consider the 20 natural amino acids the dependence of the number (Nn) of possible peptides on the number of residues (n) is expressed by the following formula:

Nn = 20n If the n-residue peptides are synthesized stepwise and independently, the number of the required synthetic steps (Sn) can be calculated as follows:

Sn = (n-1) 20n It is noted, that a synthetic step means a complete coupling cycle, that is, in addition to the coupling step itself incorporates the operations connected with the protecting groups, too. With good organization, that is, choosing a systematic synthesis route the number of synthetic steps can be reduced. The minimum number of synthetic steps is:

n

i

inS

220

The synthesized peptides are supposed to be submitted to screening tests. Since several tests have to be done on each peptide, the total number of the required screening tests is hopelessly large. If the number of kinds of screening tests is denoted by t, the total number of screening tests is expressed by the following equation:

Tn = t 20n Table 1.2 shows the possible number of peptides depending on the number of residues, the number of synthetic steps required for their synthesis, and number of the screening tests, calculating with 10 different tests (t=10). The figures - which are rounded - clearly show, that even the synthesis and testing of all tripeptides would be an almost hopeless venture. Because of the very large number of possible peptides, the stepwise synthesis of all peptides - even in the case of small ones - is an unrealizable task. The large number of the screening experiments constitutes a further problem. The proposal to be outlined on the next pages will try to somewhat improve this almost hopeless situation.

7

Table 1.2. Possible number of peptides (Nn ) containing different number of residues (n), the number of synthetic steps required for their synthesis (Sn ) in an optimized

process, furthermore the number of screening experiments (Tn ) calculating with 10 different screening tests (t=10)

(the figures are rounded)

n Nn Sn Tn 2 4 hundred 4 hundred 4 thousand 3 8 thousand 8 thousand 80 thousand 4 160 thousand 168 thousand 2 million 5 3 million 3 million 30 million 6 64 million 67 million 640 million 7 1 billion 1 billion 13 billion 8 25 billion 26 billion 256 billion 9 512 billion 537 billion 5 trillion

10 10 trillion 10 trillion 102 trillion Systematic search for biologically active small peptides through synthesis and screening of

peptide mixtures

The proposal to be outlined here constitutes a research project which makes possible to search for biologically active peptides with much greater chance than before. When I write down this project I'm fully aware of its potential importance in industry. It is also clear, that it's realization is possible only through cooperation of different institutions. Primarily the participation of the pharmaceutical industry is desirable since the investments can be recovered through pharmaceutical industry.

The essence of the proposal is that instead of one by one synthesis of peptides, peptide mixtures should be prepared containing several hundred or several thousand peptides in approximately 1 to 1 molar ratio, and these peptide mixtures should be submitted to screening tests. It will be shown that on this way much labor can be saved both in the synthetic work and in the screening experiments. In the first stage one has to determine whether or not the mixture shows any biological effect. If biological effect is observed, of course, it has to be determined which component (or which components) are responsible for the activity.

Method for synthesis of peptide mixtures

Since not single peptides but rather mixtures of peptides are synthesized, post synthetic

purification and removal of by-products are out of question. Because of this, the classical method of synthesis (in solution) can not be used either. In the synthesis of peptide mixtures the solid phase method has to be applied. It is noted here, that in the syntheses not necessarily the 20 amino acids are used. In some cases more than 20 amino acids may be used, for example if - in addition - non-common amino acids are intended to be used as building blocks. Less than 20

8

amino acids may be used, for example, in decapeptides, since the synthesis of all peptides seems to be unrealistic and have to compromise with the use of fewer kinds of amino acids. Let denote by k the number of the amino acids intended to vary in the i-th position. The numbers of amino acids varied in the C-terminal and N-terminal position are k1 and kn , respectively. Realization of the synthesis

The resin is divided into k1 equal portions (that is to as many portions as many amino

acids are intended to vary at the C-terminal of peptides). Then each portion of resin is coupled with one of the k1 kinds of amino acids then the amino-protecting group is removed from every sample. A small quantity is removed from every sample and they are taken aside for later use, then the samples are thoroughly mixed. Then the mixture of aminoacyl resins is divided into k2 equal portions and each of them is coupled with one of the k2 kinds of protected amino acids then the amino-protecting groups are removed from each sample. Before mixing, again small samples are removed and taken aside. The mixture of dipeptides is cleaved from a small portion of the mixed resin to use it in biological tests. The rest of the mixed resin is divided into k3 equal parts and the amino acids intended to occupy the third position are coupled to them. Then the synthesis is likewise continued until the mixture of n-residue peptides is reached.

It is worthwhile to add some notes. As in an ordinary solid phase synthesis, one has to make an effort to achieve good conversion by applying the reagents in excess. Fortunately, however, conversions lower than 100%, or minor unwanted splitting reactions do not cause so serious problems like in ordinary syntheses. The labour requirement could be significantly reduced by using mixtures of properly protected amino acids in acylation reactions. This, however, does not seem to be an acceptable solution because of the differences in the reactivity of the activated amino acids which would lead to the formation of peptides in significantly different concentrations thus causing problems in the screening experiments. Formation of peptides in equal concentrations can only be assured by mechanical mixing of samples followed by dividing into equal portions. This makes possible a complete conversion for every amino acid component. Possibility of acylations with mixtures of several amino acids of identical reactivity might be a matter of further considerations. Smaller differences in reactivities could be compensated by properly selected molar ratios of the amino acid derivatives of the mixture. In the following calculations, however, the possibility of acylations with the mixtures of amino acid derivatives will be left out of considerations.

The number of peptides formed in the synthesis, that is, the number of components in the peptide mixtures - in a general case - can be calculated by the following formula:

Nn = k1.k2 . . . . . . kn-1.kn If the same number (k) of amino acids are varied in every position

Nn = kn The number of synthetic steps in the synthesis of a peptide mixture containing Nn peptides (considering the attachment of the first amino acid to the resin as separate step) is:

Sn = k1 + k2 + . . . . + kn-1 + kn

9

If the same number (k) of amino acids are varied in each position,

Sn = nk The formulae show the advantage of the synthesis of peptide mixtures: the number of the synthetic steps can be calculated by summing the numbers of the varied amino acids, while the number peptides is given by the product of the numbers of the varied amino acids. One example: the synthesis of the mixture of tetrapeptides prepared by varying the 20 kinds of amino acids, needs only 80 synthetic steps! It is noted, that in the same run all shorter peptides - that is the 400 dipeptides and the 8000 tripeptides - are formed, too. The traditional synthesis of these peptides would need 168 400 synthetic steps. A different comparison: in the traditional method with 80 steps only about 30 tetrapeptides can be synthesized.

Screening of peptide mixtures

Peptides mixtures - in the first approximation - are synthesized to determine whether or not they contain biologically active component. It is supposed - although it needs experimental verification - that screening experiments can be made with mixtures, too. This offers great advantage over the traditional method since the number of screening tests is reduced by a factor equal to the number of components of the mixture. For example, the mixture of the 8000 tripeptides can be examined by a single series of tests. If there is active peptide among them, one of the executable t tests gives positive result. If the number of active peptides is more than one, then, of course, more tests may give positive result. In the synthesis of the mixture of n-residue peptides it is wortwhile to test the shorter peptides, too. The synthesis is so designed to allow for this. Taking this requirement into account, and the number of kinds of tests being t, the total number of the executable tests is:

Tn = t(n-1)

Although this equation certainly holds, its realization in practice deserves some notes. There is - without any doubt - an upper limit in the number of components of the peptide mixtures to be submitted to screening tests. It is difficult to estimate this number without experiments. The mixtures may probably contain many thousands of components, and as it can be judged today, the method outlined above is rather limited by possibilities of screening tests than by the number of the required synthetic steps. If there are too many components in the mixture, too large samples have to be applied in the screening experiments to achieve observable effect for a single component. The mixture supposedly contains a number of more or less active analogs and their effect is probably summarized. Nevertheless, an unsurpassable limit in the number of components certainly exists. Therefore in certain cases may prove useful to examine the effect of the n-residue mixtures without final mixing. In other cases the synthesis should be designed so not to surpass the optimal number of components. "Back-searching" for the active peptide

If the peptide mixture is detected to contain active component, that is, if the mixture

shows a new type biological effect, then the further task is the isolation and structure determination of the active peptide followed by its synthesis. Once the mixture containing the

10

active component or components is in our hand the isolation can be carried out using the effective separation methods, since these make possible to separate the active compound even from thousands of inactive components. It is possible, however, to follow a different method, too. This will be outlined here. This approach to the identification of the active peptides is supposed to be less tedious then the isolation methods, moreover it supplies additional information concerning the structure-effect relationship. Applicability of the method requires a procedure for quantitative determination of activity. For the sake of simplicity let's suppose that the mixture contains a single effective component (besides analogs having the same kind of effect but smaller activity). Back-searching step No. 1

The experiments are started with the kn samples taken aside in the synthesis of the n-

residue peptides before final mixing. The mixtures of n-residue peptides are cleaved from each resin sample. The mixtures of peptides differ from each other only in the n-th (that is the N-terminal) residue of their component peptides. Each peptide mixture is submitted to a quantitative activity determination. This shows how the activity depends on the terminal amino acid residue, that is, this way we can determine the N-terminal residue of the active peptide, and in addition it will show the effect of its replacement by other amino acid residues. Let's suppose, for example, that the N-terminal residue in the sample showing the highest activity (as well as in the active peptide) is Phe (phenylalanine). It is noted here that if there are several samples showing equally high activity it is practical to choose as the N-terminal residue of the active peptide the cheapest or the synthetically less problematical amino acid. This note holds for the subsequent back-searching steps, too.

Back-searching step No. 2

The experiment is continued with the kn-1 samples taken aside in the synthetic stage of the

(n-1)-residue peptides. The amino acid determined before, that is Phe in our example, is coupled to each sample. Cleavage of the peptides from the support gives kn-1 different peptide mixtures. Their common feature is that every peptide has Phe in the N-terminal position. By submitting the peptide mixtures to quantitative screening experiments one can determine the amino acid residue occupying position n-1 (that is, the pre-N-terminal position) in the active peptide. This experiment also shows the effect on activity of substitution of this amino acids with other ones. Let's suppose that the pre-aminoterminal amino acid is Arg (arginine). It should be noted that in this back-searching step Phe is coupled to kn-1 samples and the same number (kn-1) of screening experiments have to be done. Not all of the t kinds of tests are required, only the one proved before to be positive. Consequently the number of the synthetic steps and the number of screening experiments are the same: kn-1. It is also noted that in the previous back-searching step only screening test are done (their number is kn) synthetic steps are not needed.

Back-searching step No. 3

This, and the subsequent back-searching steps may be realized using two different

approaches. The peptides in samples taken aside during the synthesis have to be elongated to contain n residues, in such way, to carry on their N-terminal section the amino acid residues assuring activity. This can be realized on two ways. Either by stepwise coupling with amino acids (in our example with protected Arg then Phe) or by coupling in a single step with a previously

11

synthesized oligopeptide having the required sequence (in our example Phe.Arg). The required synthetic steps in the two approaches significantly differ. The number of the screening experiments, however, are the same in both cases. Let's turn now to the No. 3. back-searching step.

Stepwise elongation

Let's take the kn-2 samples taken aside in the synthesis of (n-2)-residue peptides. Each

sample is coupled first with protected Arg then with protected Phe. After cleaving the peptides from the support each of the kn-2 peptide mixtures are submitted to activity tests to determine the amino acid residue occupying the third position counting from the N-terminal end. The number of screening tests to be executed is kn-2. The number of the required synthetic steps is: 2kn-2. The multiplying factor preceding k is the bigger the shorter are the peptides to be elongated. The numerical value of the factor is equal to the number of amino acids to be coupled with in the elongation process.

Elongation with oligopeptide

A previously synthesized dipeptide (in our example Phe.Arg) is coupled to each of the kn-2

samples taken aside and the process is continued as described above. The number of screening test is also kn-2. The number of synthetic steps (leaving out of consideration the synthesis of the oligopeptide) is also kn-2. This procedure seems to be more economical. In practice it means that the active peptide is synthesized in parallel with the screening tests using the classical method started from the N-terminus. Small fractions of the growing peptide are sacrificed in the back-searching steps. This back-searching method has the great advantage (in addition to the fact that it needs less synthetic steps) that when the back-searching procedure is finished the active peptide is synthesized, too.

Back-searching of more than one active peptide

In the synthetic peptide mixtures several active peptides may be present, showing

different effects. In these cases the number of back-searching steps will be bigger by a factor equal to the number of the differing active peptides. That is, if the number of the active peptides is "a" the values deduced above are multiplied by a. It is noted that the presence in the mixture of peptides having different effects may complicate the back-searching process especially in the case of peptides with opposing effects. This, however, is not treated in details.

The back-searching process ends when the sequences of all active peptides are determined by applying either the oligopeptide or the stepwise elongation method.

Total number of synthetic steps and screening tests summarized

for the whole synthetic backsearching process Number of synthetic steps using oligopeptide elongation

In synthesis:

n

iin kS

1

12

In back-searching:

1

1

n

iin kaS

Total in synthesis and back-searching:

1

1)1(

n

inin kkaS

If k amino acids are varied in each step: Sn = [n(a + 1) - a]k Number of synthetic steps using stepwise elongation.

In synthesis:

n

iin kS

1

In back-searching:

1

1)1(

n

iin knaS

Total in synthesis and backsearching:

n

i

n

iiin knakS

1

1

1)1(

If k amino acids are varied in each step:

1

1

n

in iaknkS

Number of screening tests equally valid using the oligopeptide and stepwise elongation In synthesis: Tn = t(n-1)

In back-searching:

n

iin kaT

1

Total in synthesis and back-searching:

n

iin kantT

1)1(

If k amino acids are varied in each step: Tn = t(n-1) + ank An example: preparation and screening of all pentapeptides N5 = 320000, n=5 k=20 t=10 a=1 Total number of synthetic steps Oligopeptide elongation 180 Stepwise elongation 300 Number of tests 140

13

Extension of the method to other types of compounds

Applicability of the method outlined before is not restricted for only the systematic

searching for active peptides. The same principle applies to all other sequential types of compounds, that is, when the compounds belonging to this type of compounds differ from each other only in their building blocks or the sequences of these building blocks. Among them may occur natural compounds like oligosaccharides or oligonucleotides but synthetic products may be taken into account, too. Among these later ones one may think about sequential copolymers or sequential polycondensates.

Dr. Árpád Furka university professor File number 36237/1982 I certify this stitched document comprising 14, that is, fourteen pages was subscribed in my presence by Dr. Arpad Furka, university professor, with his own hands. Budapest, 1982. Nineteen hundred and eighty two, June 15, (fifteen). Dr. Judit Bokai state notary public 1.3. Publication of the split-mix combinatorial synthesis The split-mix method was published in 1988 as posters on two international congresses. First on the 14th International Congress of Biochemistry in Prague6, then on the 10th International Symposium of Medicinal Chemistry, Budapest, Hungary7. The manuscript for publication in print was submitted in February 1990 to the International Journal of Peptide and Protein Research. The paper appeared in June 19918. It was unusual, however, that within the 16 month passed between submittance of the manuscript and appearance of the paper four patents were filed on the subject of the split-mix synthesis and the author of one of the patents was the Editor in Chief of the journal where the paper was submitted. Also soon after our paper appeared two other papers were published (in September 1991) on the same topic9,10 also with the Editor in Chief among the members of the authors of one of the publications10. More can be read on this subject in a paper published in Periodica Polytechnica Ser. Chem11 http://www.pp.bme.hu/ch/index.html (year 2004) and in the home page of the author of this book12. http://www.szerves.chem.elte.hu/furka

14

References 1. http://inventors.about.com/library/weekly/aacarsassemblya.htm 2. L. B. Smillie, Á. Furka, N. Nagabhushan, K. J. Stevenson, C. O. Parkes Nature 1968, 218,

343. 3. A. Einstein The meaning of relativity, Princeton University Press, 1955, 5th Ed., Princeton,

NY, p. 107. 4. A. Linde Scientific American 1994, November, p. 48. 5. R. B. Merrifield J. Am. Chem. Soc. 1963, 85, 2149. 6. Á. Furka, F. Sebestyén, M. Asgedom, G. Dibó, In Highlights of Modern Biochemistry,

Proceedings of the 14th International Congress of Biochemistry, VSP. Utrecht, The Netherland, 1988, Vol. 5, p 47.

7. Á. Furka, F. Sebestyén, M. Asgedom, G. Dibó Proceedings of the 10th International Symposium of Medicinal Chemistry, Budapest, Hungary, 1988, p 288, Abstract P-168.

8. Á. Furka, F. Sebestyén, M. Asgedom, G. Dibó Int. J. Peptide Protein Res. 1991, 37, 487. 9. R. A. Houghten, C. Pinilla, S. E. Blondelle, J. R. . Appel, C. T. Dooley, J. H. Cuervo Nature

1991, 354, 84. 10. K. S. Lam, S. E. Salmon, Hersh E. M, V. J. Hruby, W. M. Kazmierski, R. J. Knapp Nature

1991, 354, 82. 11. Á. Furka, I Hargittai PERIODICA POLYTECHNICA SER. CHEM. 2004, 48, No. 1, p. 13.

15

2. The solid phase synthesis

The solid phase synthesis is very important in combinatorial chemistry since most of the combinatorial synthetic procedures are based on this method. The solid phase synthesis was developed by Merrifield1 and demonstrated in the synthesis of peptides. Peptides and proteins are built up from α-amino acids. The structure of the α-amino acids is expressed by the following general formula where R is the side chain by which the α-amino acids differ from each other:

Table 2.1. The natural α-amino acids

Name Side chain -R

Three letter symbol One letter symbol

Alanine -CH3 Ala A Arginine -(CH2)3NH(C=NH)NH2 Arg R Asparagine -CH2CONH2 Asn N Aspartic acid -CH2COOH Asp D Cysteine -CH2SH Cys C Glutamine -(CH2)2CONH2 Gln Q Glutamic acid -(CH2)2COOH Glu E Glycine -H Gly G Histidine -CH2(4-imidazolyl) His H Isoleucine -CH(CH3)CH2CH3 Ile I Leucine -CH2CH(CH3)2 Leu L Lysine -(CH2)4NH2 Lys K Methionine -(CH2)2SCH3 Met M Phenylalanine -(benzyl) Phe F Proline Pro P Serine -CH2OH Ser S Threonine -CH(CH3)OH Thr T Tryptophane -CH2(3-indolyl) Trp W Tyrosine -(4-hydroxybenzyl) Tyr Y Valine -CH(CH3)2 Val V

The only exception is proline in which the side chain and the amino group form a ring.

H2N-CH-COOH

R

16

The twenty α-amino acids that are components of proteins are listed in Table 2.1. In the traditional way, peptides are synthesized in solution from properly protected amino acids.

The carboxyl group of one amino acid is protected (by protecting group B) while the amino group is free. The amino group of the other amino acid is protected (Z) and the carboxyl group is activated (X) in order to make it capable to acylate the other amino acid.

Figure 2.1. Solid phase synthesis of a dipeptide. 1: Attachment of the first N-protected amino acid to the solid support ( ). 2. Removal of the

protecting group (Z) from the amino group of the attached amino acid. 3: Coupling the second N-protected amino acid to the attached one. 4: Cleaving the dipeptide from the solid support and

removing the protecting group.

Z-NH-CH-COOX + H2N-CH-COOB Z-NH-CH-CO-NH-CH-COOBR1 R2 R1 R2

Cl-CH2- Z-NH-CH-COOH +

R

-CH2- Z-NH-CH-COO R

-CH2- NH2-CH-COO

R

-CH2- NH2-CH-COO

R

Z-NH-CH-COX +

R1

-CH2- NH-CH-COO

R

Z-NH-CH-CO- R1

NH-CH-COOH +

R

NH2-CH-CO-

R1

1

2

3

4

+

NH

COOH

H

17

In continuation of the synthesis the amino-protecting group of the dipeptide is removed then acylated with another amino-protected and activated amino acid. The product of every synthetic step of the synthesis is usually isolated from the solution then purified. This is mostly a tedious job. In the solid phase synthesis the first amino acid is attached by its carboxyl group to a polymer support then the next amino acid is coupled to the already attached one and the rest of the amino acids are coupled sequentially to the attached peptide chain. The solid support used by Merrifield was a styrene-divinylbenzene co-polymer in fine bead form. The resin was functionalyzed by introduction of chloromethyl groups that made possible to attach the first amino acid to the resin. The solid phase method is demonstrated in Figure 2.1 using the synthesis of a dipeptide as example. In the synthetic step No. 1 the first amino acid is attached to the resin. All functional groups of the amino acid are protected (Z) except the α-carboxyl group. In step No. 2 the protecting group is removed from the α-amino group of the attached amino acid in order to be able to couple the second amino acid to the first one. The second amino acid is also fully protected except again the α-carboxyl group which is properly activated (by group X). The coupling is realized in step No. 3. The steps No. 2 and No. 3 form a cycle that is sequentially repeated until the attachment of the last amino acid is finished. The closing step of the synthesis (step No. 4 in our example) is the cleavage of the product from the resin which usually involves the cleavage of the protecting groups from the functional groups of the peptide. It seems worthwhile to note that the synthesis can be started with a resin which already contains an attached amino acid. Such resins are commercially available.

Figure 2.2. Reaction vessel for solid phase synthesis The solid phase reactions can be carried out in a reaction vessel demonstrated in Figure 2.2. The glass or plastic tube has a grid at the bottom that keeps the resin in the vessel when the tap is open. The tube is usually mounted on a laboratory shaker. The resin (containing the attached amino acid) is placed in the reaction vessel then all the multi step operations are carried out in the tube. First the resin is swelled by adding solvent, usually a mixture of dimethyl formamide (DMF) and dichloro methane (DCM), then the group protecting the α-amino group is removed by adding a proper reagent. Following this operation the protected amino acid and the coupling reagent are added in dissolved form. During the coupling reaction, the reaction vessel is shaked. The solid phase reactions are generally slower than the solution phase counterparts and shaking shortens the reaction time. After all of the described operations the resin is thoroughly washed with solvent. The peptide can be elongated by repeating the above cycle. After finishing

Frit Resin

18

the coupling reactions a mixture of reagents is added which cleaves the peptide from the resin and removes the protecting groups. The synthesized peptide can be recovered from the filtrate. In the solid phase synthesis the amino acids and the reagents can be added in excess to drive the reactions to completion. The excess of the amino acids and reagents can easily be removed by filtration. The coupling step can even be repeated to ensure complete conversion. The traces of the reagents are removed by repeated washings and the product of coupling remains on the filter in pure form. As outlined above the elongation of the peptide chain on the support is realized in identical coupling cycles (of course the added protected amino acid may vary from cycle to cycle). This opens the possibility of automation. In fact Professor Merrifield and his colleagues constructed and published an automatic peptide synthesizer2. Today many kinds of solid phase peptide synthesizers are commercially available. In addition, solid phase automatic synthesizers have also been developed for preparation of other kinds of organic compounds, too (see later). 2.1. Solid supports Since the seminal publication of Merrifield in 19631 different types of solid supports have been developed. The solid support, of course, have to be insoluble in the solvents used in the solid phase synthesis and it is also a requirement not to react with the reagents applied in the synthesis. In the case of most solid supports the reactions take place both inside and at the surface of the solid particles. These supports are mostly used in the form of small resin beads that swell in the solvents applied in the synthesis. The reactions in other kinds of supports take place only at the surface. These supports are used as polymer or glass beads, rods, sheets etc. and (except their surface layer) do not swell in the solvents. The solid supports are usually composed of two parts: the core and the linker. The starting compound of the synthesis is attached to the support via the linker. The core ensures the insolubility of the support, determines the swelling properties, while the linker provides the functional group for attachment of the start compound and determines the reaction conditions for the cleavage of the product. The linker itself and the covalent bond formed with the start compound must be stable under the reaction conditions of the synthesis. 2.1.1. Crosslinked polystyrene

Cross-linked polystyrene resins are the most commonly used supports for solid phase

synthesis. The polystyrene resins are synthesized from styrene and divinylbenzene by suspension polymerization in the form of small beads. The ratio of divinylbenzene to styrene determines the density of cross links. Higher crosslink density increases the mechanical stability of the beads. Lowering the crosslink density, on the other hand, increases swelling and increases the accessibility of the functional groups buried inside the beads. In practice, mostly 1-2% divinylbenzene is used. Crosslinked polystyrene is very hydrophobic so it swells only in apolar

Core Linker Start compound

19

solvents. Table 2.2 shows the swelling factor (ml/g) of 1% crosslinked polystyrene in different solvents.

Table 2.2. Swelling factor of 1% crosslinked polystyrene

Solvent Swelling factor Solvent Swelling factor Tetrahydrofurane 5.5 Acetonitrile 4.7 Toluene 5.3 Dimethylformamide 3.5 Dichloromethane 5.2 Methanol 1.8 Dioxane 4.9 Water -

Functional groups can be introduced into the resin by two approaches: either by post-

functionalization of the aromatic rings of polystyrene, or by using functionalized styrene in polymerization. The bead size of the resin is an important factor that has to be considered in solid phase synthesis. The reactions are faster when small beads are used, but application of very small beads may cause problems in filtration. The bead size is characterized either by the diameter of the beads or by the inversely proportional mesh size. In practice most often the 200-400 mesh (35-75 micron) or the 100-200 mesh (75-150 micron) bead sizes are used. The bead size distribution also deserves consideration. A narrow bead size distribution is advantageous. The capacity of the polystyrene beads is around 0.5 mmol/g. 2.1.2. Polyethylene glycol (PEG) grafted supports

PEG-grafted polystyrene has a 1-2% crosslinked polystyrene core and to its aromatic rings, polyethylene glycol chains are covalently attached. Its commercial name is Tentagel3 (Figure 2.3).

Figure 2.3. Structure of PEG-grafted polystyrene (n ~ 70). X is a functional group (Br, OH, SH or NH2) for attachment of the substrate

The advantage of the PEG-grafted polystyrene is that the substrate at the end of a flexible

chain is more accessible to reagents. It behaves like being in a solution-like environment. The PEG-chains gives a hydrophilic character to the resin and swells well in water and methanol but poorly in ether or ethanol.

O O

X n

20

2.1.3. Inorganic supports Glass beads with controlled pore size can be manufactured and are commercially available. The glass beads can be functionalized and can be used as supports in solid phase synthesis. The mechanical stability of such glass beads surpass that of the resin beads but do not swell in solvents. Functionalized ceramics can also be used as supports. 2.1.4. Non-bead form supports Polymers can be used as supports for solid phase synthesis not only as microscopic beads but also in the form of macroscopic objects if their surface can be functionalized with groups that can serve as anchors to hold the substrate in a reasonable quantity. Using appropriate monomers like styrene or others, polyolefin chains can be grafted by radiation into the surface of the objects and the chains can be functionalized.4 Such grafted macroscopic solid support units as SynPhase crowns and SynPhase lanterns are commercially available in different sizes at Mimotopes, Australia.5 2.2. Linkers, anchors The initial building block of the compound to be prepared by solid phase synthesis is covalently attached to the solid support via the linker. The linker is a bifunctional molecule. It has one functional group for irreversible attachment to the core resin and a second functional group for forming a reversible covalent bond with the initial building block of the product. The linker that is bound to the resin is called anchor.

The anchor can also be considered as a protecting group of one of the functional groups of the final product and, as such, it determines the reaction conditions by which the product can be cleaved from the support. A large variety of the commercially available resins contain the already built in anchor. A series of selected examples are found below.

Merrifield resin. The Merrifield resin can be used to attach carboxylic acids to the resin. The product can

be cleaved from the resin in carboxylic acid form using HF. Trityl chloride resin. The trityl chloride resin is much more reactive than the Merrifield resin. It can be used for

attachment of a vide variety of compounds like carboxylic acids, alcohols, phenols, amines,

Resin + Linker Resin Anchor

CH2-Cl

21

thiols. The products can be cleaved under mild conditions using a solution of trifluoroacetic acid (TFA) in varying concentrations (2-50%).

Hydroxymethyl resin. The resin can be applied for attachment of activated carboxylic acids and the cleavage

conditions resemble that of the Merrifield resin. Wang resin. The resin is used to bind carboxylic acids. The ester linkage formed has a good stability

during the solid phase reactions but its cleavage conditions are milder than that of the Merrifield resin. Usually 95% TFA is applied. It is frequently used in peptide synthesis.

Aminomethyl resin. Carboxylic acids in their activated form can be attached to the resin. Since the formed

amide bond is resistant to cleavage, the resin is used when the synthesized products are not cleaved from the support; they are tested in bound form.

Rink amide resin. The Rink resin is designed to bind carboxylic acids and cleave the product in

carboxamide form under mild conditions. The amino group in the resin is usually present in protected form. For attachment of the substrate first the protecting group is removed then it is reacted with the activated carboxylic acid compound. Cleavage of the product in carboxamide form can be performed with dilute (~1%) TFA.

CH2-OH

CH2-OH O

O-CH3

CH2-NH2

O C l

22

Photolabile anchors. Photolabile anchors have been developed that allow cleavage of the product from the

support by irradiation without using any chemical reagents. Such anchors, like the 2-nitro-benzhydrylamine resin below, usually contain nitro group that absorbs UV light.

Traceless anchors. The initial building block of a multi-step solid phase synthesis needs to have one

functional group (in addition to others) for its attachment to the solid support. It may happen that in the end product this group is unnecessary and needs to be removed. For this reason anchors have been developed that can be cleaved without leaving any functionality in the end product at the cleavage site. These traceless anchors usually contain silicon based linkers. 2.3. Protecting groups If a chemist wants to carry out a reaction on only one functional group of a multi-functional group compound, the reactivity of the rest of the functional groups needs to be suppressed. This can be achieved by application of protecting groups. A protecting group is reversibly attached to the functional group to convert it to a less reactive form. When the protection is no longer needed, the protecting group is cleaved and the original functionality is restored. A large number of protecting groups were developed for use in peptide synthesis since the amino acids are multi-functional compounds. It is an important requirement for a protecting group to be stable under the expected reaction conditions and to be cleavable - if possible - at mild reaction conditions. The stability/cleavage conditions of a protecting group are considered relative to those of the others. Two protecting groups are said to be orthogonal if either of them can be removed without affecting the stability of the other one. Some of the protecting groups most widely used in peptide synthesis are described below.

O CH NH2

OCH3

OCH3

CH NH2

NO2

23

2.3.1. Protection of amino groups The benzylcarbonyl (Z) group. Bergmann and Zervas suggested the benzyloxycarbonyl group for amino-protection in peptide synthesis in 1932 and this important protection type is still in use. The Z group can be introduced by the reaction of the amino group containing compound with benzylchloroformate under Schotten-Bauman conditions. The Z protection is stable under mildly basic conditions and nucleophilic reagents at ambient temperature. Cleavage can be brought about by HBr/AcOH, HBr/TFA or catalytic hydrogenolysis. The t-butoxycarbonyl (Boc) group. An alternative choice for amino group protection is the Boc group. Its advantage is that can be removed under milder conditions than the Z group. The Boc group is completely stable to catalytic hydrogenolysis and as such is orthogonal to the Z group. Basic and nucleophylic reagents are no effect on the Boc group and its removal can be carried out by TFA at room temperature. The most convenient reagent that can be used in the protection reaction is the Boc anhydride (Boc2O). The 9-fluorenylmethoxycarbonyl (Fmoc) group. The Fmoc group differs from both Z and Boc groups since it is very stable to acidic reagents.

CH2-O-C-Cl

O

H2N~

CH2-O-C-NH

O

~

+

Me3C-O-C-NH~ O

24

HCH2-O-C-NH-

O

The Fmoc group can be removed under basic conditions. Usually 20% piperidine dissolved in DMF is used as reagent. One of the reagents for introducing the Fmoc group is the FmocCl. 2.3.2. Protection of carboxyl groups Carboxyl groups are most often protected by converting them to benzyl esters or t-butyl esters.

The benzyl esters are cleaved by saponification, HBr/AcOH, HF and catalytic hydrogenation but not by TFA. Their response to acids is similar to that of the Z groups but somewhat less sensitive.

The t-butyl esters, unlike benzyl esters, are stable to bases or nucleophilic attack. The properties of t-butyl esters are somewhat similar to those of the Boc groups although they are less sensitive to acidolysis. They can be cleaved by TFA.

2.3.3. Protection of other functional groups The alcoholic and phenolic hydroxyl groups are protected by converting them to benzyl ether or t-butyl ether. The former protecting group can be cleaved by HF, HBr/AcOH or by catalytic hydrogenolysis and the latter one by TFA. Thiol groups can also be protected by benzyl ether formation or by tritylation. The guanidino group (present in arginine) can be protected by nitration or by arylsulphonyl groups. The nitro group resists HBr/AcOH and can be cleaved by liquid HF. Among the arylsulphonyl groups the tosyl (Tos) group can be cleaved by liquid HF or sodium in in liquid ammonia. Two other arylsulfonyl groups are more sensitive to acidic conditions. The 2,2,5,7,8-pentamethylchroman-6-sulphonyl (Pmc) group can be cleaved by TFA under conditions

~COO-CH2- ~COO CCH3

CH3

CH3

benzyl ester t-butyl ester

25

similar to the removal of the Boc group. The 4-methoxy-2,3,6-trimethylbenzenesulphonyl (Mtr) group is also cleaved by TFA but is less sensitive and requires a few hours for cleavage. The amide groups (in side chains of asparagine and glutamine) can be protected by tritylation. The trityl protecting group is stable to base, catalytic hydrogenolysis, very mild acid but is cleaved with TFA. It is used in conjunction with the Fmoc amino group protection strategy. The NH group of the imidazole ring (in the side chain of histidine) is protected in conjunction with the Fmoc strategy by tritylation. The trityl protecting group can be removed by TFA at room temperature. The indole ring (in tryptophane) can be protected by Boc group that can be removed by TFA. 2.3.4. Coupling reagents for peptide synthesis In the coupling reactions of peptide synthesis the carboxyl group of the acylating amino acid is activated. Care should be taken in selecting the activation method to avoid racemization. One of the choises is 1,3-diisopropylcarbodiimide (DIC) with addition of N-hydroxybenztriazole (HOBt) in order to reduce racemization.

NH-C-NH~

O

O

Me

MeMe

SO2

MeMe

NH-C-NH~

O

Me

MeMe

Me

SO2

Pmc protection Mtr protection

NH-C-NH~

O

O2N-

Protection by nitro group

N

N

CPh3

Protection by trityl group

Protection by Boc group

N

Boc

26

1,3-Diisopropylcarbodiimide (DIC) N-Hydroxybenztriazole (HOBt) Another very often used coupling reagent is O-benztriazole-N,N,N’,N’-tetramethyl-uronium-hexafluoro-phosphate (HBTU) that is known not to cause racemization.

N

N

N

O

NCH3H3C

N

CH3

CH3

PF6_

+

O-benztriazole-N,N,N’,N’-tetramethyl-uronium-hexafluoro-phosphate

(HBTU) One of the bases applied in the coupling reactions is N,N'-Diisopropylethylamine

(DIPEA). N,N'-Diisopropylethylamine (DIPEA) 2.4. Solid phase synthesis of organic molecules The vast number of reactions developed for the synthesis of organic molecules were optimized for solution phase. In the decades following its introduction by Merrifield,1 of the solid phase method was mainly used in peptide chemistry. Except the studies of Lezenoff,6-9 Camps,10 Frechet11-13 , Crowley and Rapoport,14 little experience has been accumulated in its application in the synthesis of organic molecules. The advent of combinatorial chemistry, however, induced radical changes and initiated a fast expanding research in the field. The classes of chemical reactions developed for solid phase as a result of this research are showed below:

Anchoring reactions

Amide bond forming reactions

Aromatic substitutions

Condensation reactions

N=C=N CHCHH3C

H3C

CH3

CH3

N

N

N

N

OH

27

Cycloaddition reactions

Organometallic reactions

Michael additions

Heterocyclic forming reactions

Multi-component reactions

Olefin forming reactions

Oxidation reactions

Reduction reactions Substitution reactions

Protection/deprotection reactions

Cleavage from supports

Other types of solid phase reactions

Excellent compilations of these reactions were prepared by Hermkers et al.15,16, Á. Furka17 and W. M. Bennett.18

2.5. Solid phase reagents and scavenger resins in solution phase synthesis Solid phase additives are successfully applied in many solution phase synthetic reactions. In solid phase reactions the substrate is bound to the solid phase carrier and the reagents are in solution. In solution phase reactions both the substrates and the reagents are in solution. In some solution phase reactions, however, the reagent is bound to resin. The advantage of such reagents is that the by products of the reagent remains bound to the resin and can be easily removed from the reaction mixture by filtration. One example is the polymer bound HOBt that is used in amide formatting reactions.

N

N

N

OH

Polymer bound HOBt More solid phase reagents and examples of their applications are found in the already mentioned compilations.15-18 Different types of resins can also be used in solution phase reactions for removal of the excess of reagents, substrates or by products. Those resins that can be used for such purposes are named scavenger resins. One example is formylpolystyrene19-20 which is used for removal of primary amines from reaction mixtures.

28

H

O

Formylpolystyrene

Other examples of scavenger resins and their applications are found in the above mentioned compilations.15-18 References

1. R. B. Merrifield J. Am. Chem. Soc. 1963, 85, 2149. 2. R. B. Merrifield J. M. Steward,N. Jernberg Anal. Chem. 1966, 38, 1905. 3. W. Rapp In G. Jung (Ed) Combinatorial Peptide and Nonpeptide Libraries 1996, VCH,

New York, 425. 4. H. M. Geysen, R. H. Meloen, S. Barteling Proc Natl Acad Sci USA 1984, 81, 3998. 5. http://www.mimotopes.com 6. C. C. Lezenoff Acc. Chem. Res. 1978, 11, 327. 7. P.M. Worster, C. R. McArthur, C. C. Lezenoff Angew. Chem. 1979, 91, 255. 8. C. C. Lezenoff, V. Yedidia Can. J. Chem. 58, 1980, 287. 9. V. Yedidia, C. C. Lezenoff Can. J. Chem. 1980, 58, 1144. 10. E. Camps, J. Cartells, J. Pi Anales de Quimica 70, 848 (1974). 11. J. M. J. Frechet Tetrahedron 1981, 37, 663. 12. M. J. Farrall, J. M. J. Frechet J. Org. Chem. 1976, 46, 3877. 13. J. M. J. Frechet, C. Schuerch, J. Am. Chem. Soc. 1971, 93, 492. 14. J. I. Crowley, H. Rapoport Acc. Chem. Res. 1976, 9, 135. 15. P. H. H. Hermkens, H. C. J. Ottenheijm, D. Rees Tetrahedron 1996, 52, 4527. 16. P. H. H. Hermkens, H. C. J. Ottenheijm, D. Rees Tetrahedron 1997, 53, 5643. 17. Á. Furka In Combinatorial & Solid Phase Organic Chemistry 1998, Advanced

ChemTech Handbook, Louisville, 35. 18. W. M. Bennett In H. Fenniri (Ed) Combinatorial Chemistry 2000, Oxford University

Press, Oxford, New York, 139. 19. K. G. Dendrinos, A. G. Kalivretenos J. Chem. Soc. Perkin Trans. 1998, 1, 1463. 20. M.V. Creswell, G. L. Bolton, J. C. Hodges, M. Meppen Tetrahedron 1998, 54, 3983.

29

3. Parallel synthesis. Synthesis of compound arrays based on saving reaction time

Execution of the chemical reactions used in the synthesis of compounds or in the analytical methods take time like the food cooking processes applied in the kitchen. The house wives know for a long time how to organize their work economically. A typical house wife, for example, begins cooking the food then starts the washing machine and while the soup is boiling and the washing machine runs the program she occupies herself with vacuum cleaning the carpet. That is, she is doing different activities in parallel in order to be more effective. The idea of making our activities more efficient by doing operations in parallel is really old. The Tibetan monks, for example, mechanized their praying by applying praying mills and they operate these praying machines in parallel that makes praying very efficient.

Figure 3.1. Praying mills1 In chemistry, the concept of parallel work was introduced relatively late. The Hungarian microbiologist, Gy. Takátsy was the pioneer of this type of activity. Takátsy organized his microbiological analytical work into parallel operations.2 He developed a radically new method for serological titrations in 1955. Among the new tools that made the parallel work very easy he introduced the microtiter plates that gained wide application in both biology and chemistry. The mictotiter plates are plastic plates with holes drilled into them. The holes serve as reaction vessels. The standard plate has 96 reaction vessels and their arrangement is shown in Figure 3.2.

30

Figure 3.2. Standard microtiter plate: 8 rows and 12 columns 3.1. The parallel synthesis The principle of parallel synthesis is the same as that applied by the house wives in the kitchen and the Tibetan monks in praying. Execution of the chemical reactions takes time and during that time not only one but a series of reactions can be realized. Each synthetic reaction is started in a different reaction vessel and all the necessary operations are executed in parallel. Figure 3.3. shows the principle of the synthesis of five different trimers, for example tripeptides, in parallel.

Figure 3.3. Parallel synthesis of five trimers in five (numbered) reaction vessels. The Black, gray and white circles represent building blocks, for example amino acids

The five trimers are synthesized on solid support (P) in reaction vessels 1 to 5. At the end of the synthesis, each trimer is individually cleaved from the support and collected in one of the

Side view

Top view

P P P 4 4 cleavage P

P P P 2 2 cleavage P

P P P 1 1 cleavage P

P P P 5 5 cleavage P

P P P 3 3 cleavage P

31

five vessels designated for storing the end products. The figure demonstrates that in parallel synthesis the number of reaction vessels is the same as the number of compounds to be prepared. The number of operations is practically the same as in the one by one synthesis of the same compounds since the solvents and reagents have to be serially transported into each reaction vessel. The real advantage is that the reaction time for the in synthesizing the 5 compounds is about the same as preparing a single one. The series of compounds prepared by the parallel and the other combinatorial methods are called compound libraries. 3.1.1. The multipin metod of Geysen

The first example of parallel synthesis was published by Geysen and his colleagues.3 They synthesized series of peptide epitopes in an apparatus developed for this purpose (Figure 3.4.). In the multipin apparatus the authors used the microtiter plate introduced by Takátsy for reaction vessels (Figure 3.4./b) and a cover plate with mounted polyethylene rods fitting into the wells (Figure 3.4/a). The end of polyethylene rods (pins) were coated with derivatized polyacrilic acid (marked by black).

Figure 3.4. The multipin apparatus

Figure 3.4. The multipin apparatus The amino acids used in building the peptides and the coupling reagents were dissolved and added to the wells. The coated ends of the pins were immersed into solution and kept there until the coupling reactions ended. Washings and deprotection of the peptides were also executed in the apparatus. The formed peptides were attached to the pins. The peptides prepared in the Geysen’s experiment were screened - after deprotection - without cleaving them from the pins.

Figure 3.5. The well row 4/column 9

Pin

Solution

a

b

1 2 3 4 5 6 7 8 9 10 11 12

1

2

3

4 5 6

7

8

32

The sequence of the peptide formed on a pin depended on the order of the amino acids added to the particular well. The amino acids or their order (or both) were different for each well so a different peptide formed on each pin. The wells, as well as the pins, were characterized by their coordinates: rows and columns. By recording the order of the added amino acids into each well, the expected sequences of the peptides could be determined from the position occupied in the plate.

If the order of added amino acids in the well row 4/column 9 is Gly, Gly, Arg, Phe, for example (Figure 3.5), then the sequence of the tetrapeptide formed on the pin row 4/column 9 is Phe.Arg.Gly.Gly (taking into account that numbering of the amino acids in peptides stars at the N-terminus). The multipin method is still used and the multipin apparatus is a commercially available product. In such apparatus, however, not coated plastic rods are used as pins. The coated head of the roads is replaced by SynPhase crowns or SynPhase lanterns (Figure 3.6) mentioned in chapter 2.

Figure 3.6. Pins with crown (a) and lantern (b)

The multipin procedure was applied by Ellman and his colleagues4 in pioneering the preparation of organic libraries by parallel synthesis.

Scheme 3.1.

Derivatives of 1,4-benzodiazepines were constructed from 2-aminobenzophenones, amino

acids and alkylating agents (Scheme 3.1). The Fmoc protected 2-aminobenzophenones were first attached to an acid labile linker (L) then through the linker to the pins (P). After removal of the protecting group it was coupled with a protected amino acid (1). This was followed by the removal of the Fmoc protecting group and cyclization (2), then by alkylation of the ring nitrogen to introduce R4 (3). Finally the product was cleaved from the support (4). 3.1.2. The SPOT technique of Frank The SPOT method introduced by Frank5,6 and his group was also developed for preparing

O

R 2

N

N O

R 1

R 3

O

R 2 R 4

N

N O

R 1

R 3

O

R 2 R 4

N

N O

R 1

R 3

H

H O

R 2

O

NHFmoc

R 1

O

R 2

O

NH

R 1

O

NHFmoc

R 3

1 2 3 4

L L L L P P P P

a b

33

peptide arrays. The synthesis is carried out on cellulose paper membranes derivatized to serve as anchors for the first amino acids of the sequences to be prepared. Small droplets of solutions of protected amino acids dissolved in low volatility solvents and coupling reagents are pipetted onto predefined positions of the membrane (Figure 3.6). The spots thus formed can be considered as reaction vessels where the conversion reactions of the solid phase synthesis take place. An array of as many as 2000 peptides can be made on an 8x12 cm paper sheet. The peptides can be screened on the paper after removing the protecting groups. 3.1.3. Other devices for parallel synthesis De Witt and co-workers7 also developed an apparatus for parallel synthesis. It was designed for the synthesis of small organic molecules. The solid support was placed in porous tubes immersed into vials containing solutions of reagents which diffused into tubes. The temperature of the reaction mixtures could be controlled by heating or cooling the reaction block.

Figure 3.6. The spot synthesis Another inexpensive device was described by Meyers et al.8 Beckman polypropylene deepwell plates were modified by drilling a small hole in the bottom of each well. A porous polyethylene frit was fitted into the bottom of the wells to allow removal of the solutions and solvents by vacuum. At the bottom, a rubber gasket prevented the leakage of the wells during the reactions.

Figure 3.7. The LabMate manual synthesizer (photo: www.aapptec.com)

34

A simple commercially available parallel device, offered by the firm aapptec, is demonstrated in Figure 3.7. It is a manual synthesizer allowing the parallel synthesis of 25 different compounds in 0.05 to 2 mmol quantity. The content of the reaction vessels can be mixed and heated in 4 different zones. Cooling is also possible if connected to a circulating chiller. Another parallel synthesizer is offered by BÜCHI. The Syncore® Reactor (Figure 3.8.) can accommodate 24 to 96 reaction vessels. It can be used for both solid phase and liquid phase parallel synthesis. Among the features vortex mixing, heating and cooling and parallel evaporation of the samples can be mentioned.

Figure 3.8. The Syncore® Reactor of BÜCHI (photo: www.buchi.com)

Preparation of many organic compounds needs heating. Since the mid-1980s the use of microwave heating began to spread in chemical laboratory practice9. This kind of heating raised considerably the speed of chemical reactions in both solution and solid phase. The reaction times are typically reduced from days or hours to minutes or second often followed by increased yields, too. This type of heating is also applied in the synthesis of combinatorial libraries in order to save time. In microwave heating the energy is not transferred by conduction or convection so the reaction vessel is not heated only the solvent and the reactants. The energy is absorbed by dipolar molecules. Molecules that have larger dipolar momentum absorb better. For this reason solvents with large dielectric constant are preferred. Although examples for application of domestic microwave ovens in parallel synthesis of combinatorial libraries are found in the literature10 in practice rather specially constructed heaters are applied. The experiments carried out in domestic ovens are often difficult to reproduce because of the uneven electromagnetic field distribution, the pulsed irradiation and the unpredictable formation of hot spots. Two kinds of specially constructed commercial reactors are available that work either in multimode or in monomode operation.11,12

35

Figure 3.9. In the XP-1500 Plus™ system of CEM up to 12 samples can be heated simultaneously (photo: courtesy of CEM)

The multimode reactor has a large cavity like the domestic oven but reflection by the walls and a mode stirrer ensures a nearly homogenous distribution of the electromagnetic field. In this reactor the samples can be heated in parallel. A parallel reactor for heating up to 12 samples is demonstrated in Figure 3.9.

Figure 3.10. The software controlled ExplorerPLS® system of CEM

(photo: courtesy of CEM)

The ovens operating in monomode, on the other hand, heat only one sample at a time. The vials containing the reactants are delivered serially into the oven. Such system, the ExplorerPLS® of CEM is demonstrated in Figure 3.10. The Explorer handles all of the routine tasks necessary to execute a large number of reactions each day. The system has a sample deck with interchangeable racks.

36

3.1.4. Parallel synthetic methods with reduced number of operations As already mentioned the parallel synthetic methods are based on simultaneous execution of a series of reactions and, as a consequence, are saving a lot of time in the synthesis of compound arrays. The number of operations that are executed in these reactions, however, is more or less the same as in the one by one preparation of the same compounds. In some parallel methods attempts were made to reduce the number of operations, too. 3.1.4.1. Synthesis of oligonucleotides on paper discs Ronald Frank and his colleagues who introduced the spot synthesis also developed a method for the parallel preparation of oligonucleotides on paper discs which involves reduction of the number of operations necessary in the synthesis.13 The method was demonstrated by simultaneous preparation of two octamers using, as solid support, Whatmann 3MM paper discs (diameter 2 cm) labeled by pencil. The synthesis was carried out according the principle outlined in Figure 3.11: whenever the two growing chains had to be elongated with the same nucleotide the two discs were transferred into the same reaction vessel and the elongation was realized in a single coupling cycle. The sequences of the two nucleotide are seen at the bottom of Figure 3.7 together with the coupling order of the nucleotides. The sequences differ only at coupling positions 3 and 6. At these positions the couplings were carried out separately in two reaction vessels while in the remaining six coupling positions (1, 2, 4, 5, 7 and 8) the two discs were placed into the same reaction vessel and each of these elongations (including the attachment of the first nucleotide to the support) was realized in a single coupling cycle.

Figure 3.11. Flow diagram of the synthesis of two oligonucleotides At coupling positions 3 and 6 the discs were placed in separate reaction vessels and the couplings were executed separately. The total number of the executed coupling cycles in the synthesis of the two octamers was 10. In the case of normal parallel synthesis of the same two compounds 16 coupling cycles would have been needed. This shows that the method is capable to significantly reduce the number of the necessary operations. It seems worthwhile to note that

21

T 21

A

1 2

1 2

T C

3

21A

4

21

T

5

21A

7

1 2

A G

6

21T

8

1TAATATTA 2TAGTACTA

8 7 6 5 4 3 2 1

37

in a single coupling cycle 18 different operations had to be executed including washings and dryings. 3.1.4.2. The tea-bag synthesis A parallel synthetic method based on the principle suggested by Frank but using a different solid support was developed by R. A. Houghten.14 Series of peptides were synthesized on bead form polymer support. The polymer beads were enclosed into visibly labeled permeable plastic bags (tea bags Figure 3.12.). A different peptide - according to pre-decided sequences - was prepared in each bag. In a coupling step the bags were grouped according to the amino acid appearing in their assigned sequence at that coupling position then placed into the same reaction vessel for coupling with the same amino acid. For example in coupling step 3 all bags in which the assigned sequence contained Ala in position 3 were grouped and transferred into the reaction vessel where Ala was used in the elongation reaction. Before the next coupling step the bags were manually regrouped after reading their label - again according to the sequences assigned to the bags. All operations, including removal of protecting groups, couplings, washings and even the cleavages were performed on the solid supports enclosed into the same bags. This procedure has the same advantage that was pointed out at the Frank method: less number of operations is needed than in a normal parallel synthesis and the number of reaction vessels is also less than the number of the synthesized compounds.

Figure 3.12. The tea bag method 3.2. The Ugi multicomponent reactions One family of organic reactions is particularly suitable for parallel execution. These are the Ugi multicomponent reactions. In multicomponent reactions several starting compounds can be combined in a one step reaction to give a complex product. One of the most extensively used such reactions are the Ugi four component reactions15,16 symbolized by U-4CR. In this reaction an acid, very often carboxylic acid, a primary or secondary amine, an aldehyde or ketone and an isonitrile transform into a single product (Figure 3.13). By varying R1, R2, R3 and R4 large series of compounds can very easily be prepared. The rection is driven by the high reactivity of the isonitrile component.

tea-bag r e a c t i o n v e s s e l s w i t h b a g s

38

R1 COOHR2 NCR3 CHOR4 NH2

R1 N

HN

R2

O

R4

R3

O

Figure 3.13. The Ugi four component condensation Althopugh the multicomponent reactions were first realized in 193817 their practical importance was recognized only after the advent of combinatorial chemistry. A vast number of compounds can be and are prepared by the Ugi and other multicomponent reactions. This can be exemplified by Figure 3.14. that shows how the products can be varied by replacing the carboxylic acid component by other acids.18

R5 N

HN

R

R5-COOHR1-CO-R2R3-NH-R4

R-NC

HX

HN3

R3

NN

NNR4

R1 R2 R

R3

NNH

R

XR4

R1 R2

N

NH

X

R3

NR

R1R2

H2S2O3H2O HSCN

HOCNO

R3

R1 R2

O

X = O or SX = O or S

Figure 3.14. Products of U-4CRs carried out with different acids. HX is replaced by acids seen on

the arrows The variability of the products can be further increased by post condensation transformations. An acid catalysed cyclization of the condensation product19 is demonstrated in Figure 3.15. The cyclization is brought about by the removal of the protecting Boc group. The Ugi reactions can be realized in both solution and solid phase. Any of the four reactants can be linked to the resin. Mostly good conversations can be achieved.

R1 CHO

R2 NH2

NC

OH

NBoc

R3

O

R4

HN

R4

O R1

ON

R2

R3Boc

N

N

R4

O R2

R1

OR3

10% TFA

Figure 3.15. Post cyclization of the primary Ugi product

39

3.3. Solution phase combinatorial synthesis Before introduction by Merreifield20 of solid phase synthesis, the organic compounds were generally prepared in solution. The use of the solution phase synthesis in combinatorial chemistry has some advantages and also serious disadvantages. The main advantage is that the overwhelming majority of synthetic procedures recorded in the literature are realized in solution phase. The disadvantage, on the other hand, is that in a multi-step reaction the products need to be isolated and purified in each step that is often tedious and time consuming. Nevertheless, solution phase synthetic methods are applied in combinatorial chemistry, too. There approaches, however, that make possible to reduce the disadvantages and so to make the solution phase procedures competitive and applicable besides the solid phase methods. 3.3.1. Dendrimer supported synthesis.21 Dendrimers are branching oligomers. They are built up in stepwise manner from monomers that result in branching at every coupling position.

These oligomers are soluble, relatively large molecules their size considerable exceeds those of the building blocks and reagents used in combinatorial syntheses. To the ends of their branches linkers can be attached so they can serve as soluble supports for combinatorial synthesis (Figure 3.16.).

Figure 3.16. Two step synthesis on dendrimer

Because the large size of the dendrimer molecules, the products of each coupling step can easily be separated from the excess of reagents by size exclusion chromatography. After cleaving the small molecule products from the support, the size exclusion chromatography also makes possible to separate them from the dendrimer molecules. This kind of separation is usually much faster than the conventional separation and purification processes but it is much slower than the simple filtration in the solid phase procedures.

: linker, : building block

40

3.3.2. Separations using fluorous tags and fluorous solvents. The fluorous solvents are immiscible with most organic solvents and water.22 This fact is exploited in using fluorous-organic liquid-liquid extraction for separation of products of solution phase combinatorial syntheses from reagents.23 This separation works if fluorous tags are attached to the reagents or to the products. The attachment of the fluorous tags may occur before the combinatorial reaction step or after it. An example is demonstrated in Figure 3.17. A Stille coupling is carried out using a fluorous reactant and a fluorous solvent, the commercially available FC-72, consisting mainly C6F14 isomers.24 At the end of coupling, the product was extracted into dichloro methane and the by product (Cl-Sn(Ch2CH2C6F13)3) was found in CF-72.

MeO

Sn(CH2CH2C6F13)3

BrMeO

+PdCl2(PPh3)2

LiCl/DMF/THF+ Cl-Sn(CH2CH2C6F13)3

Figure 3.17. Stille coupling using fluorous reactant A second example is outlined in Figure 3.18. An aromatic urea derivative is synthesized from an amine and an isocyanate that was applied in excess. After completion of the reaction, a fluorous quencher was added also in excess by which the excess of the isocyanate was transformed into a fluorous adduct. Finally an organic/fluorous liquid-liquid extraction was carried out. The fluorous adduct and the excess of the fluorous quencher separated into the fluorous solvent (FC-72) and the product was isolated by evaporation of the organic solvent (THF). This approach was applied in the synthesis of a 9 member library.25 3.3.3. Application of solid phase reagents. In organic reactions the reagents are often used in excess in order to drive the transformations to completion. If the reaction is carried out in solvent with reagents attached to resin, the excess of the reagents can easily be removed by filtration at the end.

HH2

F3C+

Br NH

NH

OCNF3C

Br

O

excess

NCO

HNSi(CH2CH2C6F13)3

Si(CH2CH2C6F13)3

HN N

Si(CH2CH2C6F13)3

Si(CH2CH2C6F13)3

O

+

Fluorous quencher in excess

Figure 3.18. Fluorous quenching

41

In Figure 3.19. an example is found showing acylation of a secondary amine with a solid phase active ester.26 The product can be isolated from the filtrate while the by product and the excess of the active ester remains on the filter.

O

NO2

O

R1

O

+ HN

R2

R3

OH

NO2

O

N

R2

R3

+

CH3-CN,Et3N

70o

active ester productR1

O

Figure 3.19. Acylation with solid phase active ester In another example shown in Figure 3.20. the coupling reagent, 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide (EDC), is used in insoluble form (P-EDC) for coupling an acid (R1COOH) with a secondary amine (R2R3N). The transformed form of the reagent can be filtered out and the product can be isolated from the filtrate. Not only reagents but catalysts can also be used in solid phase form.25

N NC

NCl

+ HO R1

O

N NH

CCl

N

O R1

O

P-EDC

N NH

CCl

NH

O

R3

NH

R2

R1 NR2

R3

O

+

product

Figure 3.20. Amide bond formation with solid phase carbodiimide 3.3.4. The use of scavengers in solution phase reactions. Solid phase scavengers are often applied in reactions carried out in solution in order to remove the remains of reagents used in excess. In Figure 3.21. a reaction is outlined in which a solid phase catalyst (borohydride) and a solid phase scavenger (aldehyde) is used.25 In this reaction an aldehide (R2CHO) is reacted with an excess of a primary amine (R1NH2) to form an imine which is reduced with the solid phase borohydride catalyst to a secondary amine (product). Although the catalyst can be filtered out, the product is still contaminated with the excess of the primary amine. This reactant is then removed from the reaction mixture in the form of solid phase imine by the added aldehyde scavenger. The clean product can be isolated from the filtrate.

42

R1 NH2 + R2

O

H

NR3 BH4

excess

R1N R2

R1 NH2

O

H

filtered out

H

R1N R2

H

clean productproduct andremaining amine

Figure 3.21. The use of solid phase catalyst and scavenger in one reaction

3.4. Automation in parallel synthesis Like many other important developments in combinatorial chemistry, automation also began in the field of peptide chemistry. Introduction of the solid phase peptide synthesis procedure by Merrifield19 in 1963 opened the possibility for automation. Merrifield not only invented the new synthetic technology but he also built the first solid phase peptide synthesizer.27 In the last half century numerous companies developed and commercialized automatic synthesizers. 3.4.1. Automatic parallel synthesizers Automatic parallel synthesizers are developed for both peptide and organic synthesis. The peptide synthesizers generally do not need heating or cooling. The organic syntheses, on the other hand, often need heating or cooling or special atmosphere or elevated pressure. So the organic synthesizers are more complex than the peptide synthesizers. Figure 3.22. shows a peptide synthesizer made and sold by the firm aapptec. It can be used to prepare simultaneously up to 96 peptides by solid phase synthesis. A Teflon reaction block that can be shaken contains the reaction vessels closed by septum. The reactors have a frit at their bottom to hold the resin and to make possible to remove the liquids by applying vacuum. The resin can be filled into the reaction vessels by volume using a simple tool. The solutions of the reagents and amino acids are stored in containers closed by septum. The solvents are stored in bottles. The solvents and solutions are automatically transferred into reaction vessels by needle like probe that can penetrate through septum. The probe is fitted to an arm that can be moved in x, y and also in z direction. The Apex 396 is available with one or two arms and also with two kinds of reactors. The synthesized peptides can be cleaved automatically from the resin. The quantity of the synthesized peptides may vary from 0.005 to 1 mmol. The synthesizer like other automatic machines runs under computer control. Once the resin is placed into the reaction vessels, the solutions of the protected amino acids and reagents are filled into their containers and the solvent bottles are also filled the started machine executes the synthesis in an unattended manner following step by step a pre-prepared program. The software supplied with the machine allows easy programming.

43

Figure 3.22. The Apex 396, a solid phase peptide synthesizer (photo: www.aaptec.com)

The peptides cleaved from the resin are in dissolved form. The solvent needs to be evaporated or lyophilized. A simple module developed for this purpose is seen in Figure 3.23. It is designed to prevent liquid bumping. It can be connected to vacuum pump equipped with cold trap. The module can also be used to evaporate or concentrate fractions after purifications.

Figure 3.23. Parallel evaporation/lyophilization module (photo: www.aaptec.com)

Another automatic synthesizer is demonstrated in Figure 3.24. This is Model 384 Ultra High Throughput Synthesizer is also manufactured and sold by aapptec.

44

Figure 3.24. The is Model 384, Ultra High Throughput Synthesizer for solid phase applications (photo: www.aaptec.com)

Figure 3.25. The Solution, a solution phase parallel synthesizer (photo: www.aaptec.com)

The synthesizer has four reactor blocks each containing a maximum of 96 reaction vessels. So a maximum of 384 different peptides or other compounds can simultaneously be prepared. The maximum number of compounds that can be prepared depends on the volume of the reaction vessels that can vary from 3 to 35 ml. Heating and cooling is optional. An automatic machine, the Solution, designed for solution phase synthesis is shown in Figure 3.25. but the instrument can be used for solid phase synthesis, too. It can accommodate fifteen 80 ml reactors or ninetysix smaller reaction vessels (3 or 10 ml). The machine automatically performs liquid-liquid extractions and can transfer the products to titer plates.

45

Another machine, the Sophas HTC (Figure 3.26), is manufactured by Zinzer Analytic. It can be used for both solution phase and solid phase parallel preparations. 600 compaunds can be synthesized in a parallel procedure. The reaction vessels can be heated up to 150 oC and can be cooled to -40 or -80 oC. Inert reaction conditions are assured. Slurry distribution and sample picking for HPLC during the synthesis are also possible.

Figure 3.26. The Sophas HTC automatic parallel synthesizer (photo: www.zinsser-analytic.com)

3.4.2. Quality control The components of compound libraries prepared by parallel synthesis may contain by products. For this reason they are usually submitted to quality control and, depending on the result, they may needed purification before screening. The quality control usually involves HPLC separations or mass spectroscopy or both. After their synthesis, the library components are usually available in evaporated dry form. Before submitting them to quality control the dry samples need to be dissolved. Samples are then taken from the solutions that are serially submitted to liquid chromatography (LC). The chromatogram usually shows if impurities are present in the samples. The synthesized products as well as their impurities can be identified by taking samples from the LC eluents end analyzing them by mass spectrometry (MS). In this process automatic liquid handlers do the job that ends at sample loading. One of the available liquid handlers is the 223 Sample Changer, the product of Gilson (Figure 3.27.). This is a programmable sampler for automated sample preparation and transfer. It is used for serial dilutions, sampling into vials and tube to tube transfers.

46

Figure 3.27. The 223 Sample Changer (photo: www.gilson.com)

Another liquid handler is also a Gilson product: the Multiple Probe 215 Liquid Handler/Injector (Figure 3.28.). This device is a large-capacity multiple-probe liquid handler that processes four or eight samples simultaneously. It also performs injection into four or eight parallel systems simultaneously. The instrument is ideal for parallel injection into HPLC and LC/MS systems.

Figure 3.28. The Multiple Probe 215 Liquid Handler/Injector (photo: www.gilson.com)

The MALDILC™ System of Gilson is designed to perform a microbore HPLC with fraction collection and simultaneous matrix addition onto MALDI plates. The plated fractions can then be analyzed repeatedly by MALDI-TOF MS.

47

Figure 3.29. The MALDILC™ System (photo: www.gilson.com)

3.4.3. Parallel purification The automatic parallel chemical synthesizers produce a large number of products. In order to purify them in reasonable time the parallel approach needs to be applied in the purification stage, too. An example of parallel purification instruments is the Quad 3+ system (Figure 3.30.) that is suitable to purify 4, 6 or 12 samples in parallel while recording their UV spectra.

Figure 3.31. The Quad 3+ system of Biotage

(photo: www.biotage.com)

48

3.4.4. Manufacturers of laboratory robots The companies listed below are engaged in laboratory automation, manufacture and commercialize such products. In addition to the companies that offer automatic compound synthesizers other companies manufacture automatic chemical analyzers that are also important in laboratories applying combinatorial chemistry. Besides the hardware, most companies offer software, too. aaptec (www.peptideprotein.com) Advanced Automated Peptide Protein Technologies (aaptec) is the successor of Advanced ChemTech and offers a family of automatic of peptide, protein and other chemical synthesizers. accelab (www.accelab.de). The company focuses on integrated solutions for productivity enhancement in research laboratories. The accelab automation platforms are highly compact and flexible and allow real unattended 24-hour operation. They use SCARA and gantry robots which handle the major part of the required automation tasks, leaving the parallel routine sample processing to the integrated standard laboratory instruments. AutoDose (www.autodose.ch). The company develops leading edge technologies for high precision powder dispensing. POWDERNIUM™ automates the very traditional yet tedious weighing process of solids/powders. POWDERNIUM is available in several models to accommodate volumes ranging from 0.2mg to a few hundred grams and can run unattended. The accuracy can be as high as 0.1 mg. Beckman Coulter (www.beckman-coulter.com). The company offers a complete range of automation tools for your research applications, from modular liquid handling systems to integrated robotic systems in genetic analysis or drug discovery. Bio-Automation (www.bio-automation.com). The company provides automation solutions to the Life Science Industry. Its Bio-Bot is a laboratory robot that is designed to provide walk away time to single instrument or multiple instrument workstations. The software provides a simple, yet powerful control. BioDot, Inc. (www.biodot.com). The company manufactures a wide line of material handling, dispensing, and processing modules provide engineering solutions that can be customized to fit precise requirements. Biotage (www.biotage.com). The company is currently producing automated systems for the parallel purification and screening of the high number of compounds generated by combinatorial chemistry. Automated solutions include the basic Quad3 which can purify twelve fractions with twelve separate solvents in less than thiry minutes to the flagship Parallex and FLEX systems which purify and collect combinatorial arrays using an intelligent fraction collection system based on UV peak

49

characteristics. Biotage also offers automated and semi-automated systems for production-scale HPLC and FLASH chromatography processes. BUCHI Corporation (www.buchi.com) Buchi produces and offers a large array of laboratory equipments. Caliper Technologies (www.caliperls.com ). The company designs and manufactures Labchip™ devices and systems that enable high-throughput screening. The LabChip™ systems replace entire chemical laboratories. Caliper's microfluidic LabChip™ devices function like "liquid integrated circuits." They process fluid - containing DNA, proteins, or cells - like semiconductors process electrons, executing biological tests in seconds. Genes can be analyzed within minutes. Promising drug compounds can be tested within days instead of months. Carl Zeiss Jena (www.zeiss.de). The company has developed a screening system for ultra high throughput screening (UHTS) for pharmaceutical drug research. The high-performance multimode readers offer 96-channel parallel detection of fluorescence, absorption and luminescence in 96, 384 and 1,536-well microtiter plates. The compact workstations and systems containin a new robust technology for the transport of microtiter plates with a throughput of > 100,000 specimens a day. The user-friendly software offer simple assay programming. Cartesian Technologies, Inc. (www.cartesiantech.com). The company manufactures equipment for pharmaceutical and agricultural research. The equipment helps automate and increase the process efficiencies in areas such as drug screening, genomics, and combinatorial chemistry. Cellomics, Inc. (www.cellomics.com ). Cellomics Inc.’s mission is to improve the efficiency of the drug discovery process by delivering a cell-based screening platform that automates target validation and lead optimization using fluorescence-based assays. Today, the Company’s integrated platform consists of proprietary fluorescence assays, a proprietary, cell-based High Content Screening (HCS) system, and bioinformatics software. CyBio (www.cybio-ag.com). The company offers modular technology platforms for automated drug research, high throughput screening, liquid handling, luminescence readers. Genetix (www.genetix.co.uk). The company offers a multi-tasking robot, offering Colony Picking, Gridding and Liquid Handling. The 'Q' BOT is an invaluable addition to any laboratory engaged in high throughput Pharmaceutical, Genomic or Bioresearch screening.

50

Genomic Solutions (www.genomicsolutions.com). The company provides instrumentation, software and related products and services. It offers to life science researchers in biopharmaceutical companies, universities and government institutions with integrated, flexible systems. Gilson (www.gilson.com). The company is specialized in analytical instrumentation for scientific research and industrial markets. It has developed software and instrumentation to keep pace with the most sophisticated HPLC, LC and sample preparation technology available for the laboratory. Gyros (www.gyros.com). Gyros AB offers pharmaceutical, biotechnology and diagnostic companies access to a unique, proprietary technology platform. Routine or non-routine laboratory processes are miniaturized and integrated into application-specific CD microlaboratories. Hundreds of samples can be processed in parallel on the disposable CDs. Integrating different laboratory steps onto a single CD microlaboratory offers the potential to reassess and redesign traditional working procedures. Hamilton Company (www.hamiltoncomp.com). Hamilton offers robotic instruments. MICROLAB automated precision liquid handling systems increase the speed, throughput and productivity of sample preparation procedures. HEL (www.helgroup.co.uk). The company that manufactures reactor and calorimeter systems for process screening and optimization. The systems runs on proprietary software and hardware and incorporate robotic work-stations for liquid handling and automated sampling to HPLCs and other analytical tools. The systems performs chemistry from a 5 ml to 100 ml scale with up to 32 reactions in parallel. The systems are available in glass or metal and have wide operating temperature and pressure ranges. IRORI (www.irori.com). The company provides combinatorial chemistry technology to the pharmaceutical industry. The AccuTag™-100 Combinatorial Chemistry System is used at large and small pharmaceutical and drug discovery companies. It offers microreactors, miniature electronic tags, and automated sorting instrument for combinatorial directed sorting. KBiosystems (www.kbiosystems.com). The company manufactures robots for medium to high throughput laboratory automation including the Duncan high throughput PCR thermal cyclers, the K-Core 2D Gel cutter, the Preptide proteomics robot and the K2 and K3 colony pickers, arrayers, re-arrayers and replicators. Labman Automation (www.labman.co.uk). Labman designs and builds instruments for high-throughput screening and microarrayers. Also deals with powder feeding and dispensing and weighing/labelling.

51

LEAP Technologies, Inc. (www.leaptec.com). provides Front-End automation for chromatography, mass spectroscopy, elemental analysis, and other analytical techniques. We specialize in applications that demand reliability, flexibility, precision, and high throughput. Custom solutions are available for non-standard sample preparation and loading problems. We work with major chromatography and mass spectroscopy companies to provide total integrated solutions for critical automation applications. Mitsubishi (www.mitsubishitoday.com). The company offers PA-10, a powerful super-lightweight 10kg robotic arm with 7-axis redundancy control that can be controlled by a PC or a built in control unit. The PA-10 is a useful addition in any research lab. Important features that make the PA-10 useful are its innovative open system and its flexibility with human arm like maneuverability. PerkinElmer Life Sciences (www.lifesciences.perkinelmer.com). The company supplies products, services and technologies for functional genomics, high throughput screening and drug discovery as well as for clinical screening. Personal Chemistry (www.personalchemistry.com). The company developed the Coherent Synthesis™ and the Coherent Synthesis™ used in pharmaceutical research. Protedyne (www.protedyne.com). The company manufactures the Computer Integrated Laboratory Automation (CILA) systems for the biotechnology and pharmaceutical markets. REMP, Switzerland (www.remp.com). The company is specialized in laboratory automation and robotics. REMP supplies solutions for compound storage and retrieval, compound cherry-picking, plate replication and re-formatting, automated powder dosing, environmental conditioning, and plate heat sealing and piercing applications. Robocon Inc. (www.robocon.com). The company specializes in laboratory automation for pharmaceutical research, biotechnology and medical/veterinarian diagnostics. Sias AG (www.sias.ch). The company is producing robotic laboratory automation for liquid handling and analysis that is available as a free standing workstation. ST Robotics (www.strobotics.com). The company produces "bench top robots" R16 and R17 and the bench top version of the Cartesian R15.

52

TECAN (www.tecan.com). The company produces a large portfolio of instruments and systems the Robotic Sample Processors for automated liquid handling, the Microplate equipments such as Readers and Washers. TekCel (www.tekcel.com). The company manufactures a family of robotic workbenches, called TekBenches™. These products include liquid handling and assay development, automated microplate sealing/resealing, storage and retrieval system. Tomtec (www.tomtec.com). The company manufactures a complete line of liquid handling systems including harvesters, 96-well pipetters, 384-well pipetters, plate washers, and robotic components and systems. Zinsser Analytic (www.zinsser-analytic.com) The company is specialized in developing, producing and distributing innovative laboratory solutions for liquid handling and automation including systems for combinatorial chemistry and tools for drug discovery. Zymark Corporation (www.zymark.com). The company is a is a designer and installer of workstation-based laboratory automation products. References

1. J. Kehnscherper, G. Kehnscherper, A. Hausen, W. Mochmanní A világ vallásai, Tessloff & Babilon, 1999.

2. Gy. Takátsy Acta Microbiologica Acad. Sci. Hung. 1955, 3, 191. 3. H. M. Geysen, R. H. Meloen, S. J. Barteling Proc. Natl. Acad. Sci. USA 1984, 81, 3998. 4. B. A. Bunin, J. A. Ellman J. Am. Chem. Soc. 1992, 114, 11997. 5. R. Frank, S. Güler, S. Krause, W. Lindenmayer,In Peptides 1990, E. Giralt, D. Andreu

(Eds), 1991, ESCOM, Leiden, 151 6. R. Frank, Tetrahedron 1992, 48, 9217. 7. S. H. De Witt, J. S. Kiely, C. J. Stankovic, M. C. Schroeder. D. M. R. Cody, M. R. Pavia

Proc. Natl. Acad. Sci. U. S. A. 1993, 90, 6909. 8. H. V. Meyers, G. J. Dilley, T. L. Durgin, T. S. Powers, N. A. Winssinger, H. Zhu, M. R.

Pavia, Molecular Diversity, 1995, 1, 13. 9. P. Lidström, J. Tierney, B. Wathey, J. Westman Tetrahedron, 2001, 57, 9225. 10. B. M. Glass, A. P. Combs In I. Sucholeiki (Ed) High-Throughput Synthesis. Principles

and Practices, Marcel Dekker Inc. 2001, 123. 11. O. Kappe, A. Stadler In G. A. Morales, B. A. Bunin (Eds) Methods in Enzymology,

Combinatorial Chemistry Part B, 2003, Elsevier Academic Press, 197. 12. B. M. Glass, A. P. Combs, S. A. Jackson In G. A. Morales, B. A. Bunin (Eds) Methods in

Enzymology, Combinatorial Chemistry Part B, 2003, Elsevier Academic Press,223. 13. R. Frank, W. Heikens, G. Heisenberg-Moutsis, H. Blöcker Nucleic Acid Research 1983,

11, 4365.

53

14. R. A. Houghten Proc. Natl. Acad. Sci. USA 1985, 82, 5131. 15. I. Ugi Isonitrile chemistry, Academic Press, 1971, 1. 16. I. Ugi Proc Estonian Acad Sc. Chem. 1995, 44, 237. 17. A. Laurent, C. F. Gerhardt Liebigs Ann. Chem. 1838, 28, 265. 18. I. Ugi, A. Dömling, B. Ebert In G. Jung (Ed) Combinatorial Chemistry. Synthesis,

Analysis, Screening, Wiley-VCH, 1999, 125. 19. C. Hulme, H.Bienamé, T. Nixey, B. Chenera, W. Jones, P. Tempest, A. L. Smith In G. A.

Morales and B. A. Bunin (Eds) Methods in Enzymology, Combinatorial Chemistry Elsevier Academic Press, 2003, 369, 469.

20. R. B. Merrifield J. Am. Chem. Soc. 1963, 85, 2149. 21. N. K. Terrett Combinatorial Chemistry, Oxford University Press, 1998, 64. 22. I. T. Horváth, J. Rábai Science, 1994, 266, 72. 23. D. P. Curran Angew. Chem. Int. Ed. Engl. 1998, 37, 1174. 24. D. P. Curran, M. Hoshino J. Org. Chem. 1996, 61, 6480 25. B. Linclau, D. P. Curran In I. Sucholeiki (Ed) High-Throughput Synthesis. Principles and

Practices, Marcel Dekker Inc. 2001, 135. 26. S. W. Kaldor, M. G. Siegel Current Opinion in Chem. Biol. 1997, 1, 101. 27. R. B. Merrifield J. M. Steward,N. Jernberg Anal. Chem. 1966, 38, 1905.

54

55

4. Combinatorial synthetic methods The parallel synthetic methods described in the previous paragraph considerably speed up preparation of compounds used for different purposes. The advent of the real combinatorial processes accelerated the preparation of new compounds to never dreamed speeds. Although there are compounds that can be prepared in a single reaction step most substances are synthesized in a multi-step process. Such substances are built up stepwise from building blocs. The building blocks of a peptide, for example, are the amino acids that are linked together step by step. The real combinatorial methods can be used in multi-step processes and their most important characteristic feature is that make possible to prepare in a single run all structural derivatives that can be theoretically deduced from the structures of the building blocks. The number of operations and number of reaction vessels relative to the number of synthesized compounds are drastically reduced. Most combinatorial synthetic methods are based on solid phase approach but there are versions that can be realized in solution, too. Among the solid phase carriers most often the bead form resin is applied but surfaces of solid materials are also used. 4.1. Combinatorial synthesis on bead-form resin 4.1.1. The split-mix synthesis The split-mix method introduced by Furka and his colleagues1-3 is based on Merrifield's solid phase procedure and originally it was demonstrated by synthesis of peptides. The principle is described here in a simplified version, using only three different protected amino acids as building blocks that are represented in Figure 4.1 by red, yellow and blue circles. The same concept is valid regardless of the number or types of monomer units or other kinds of blocks involved. The synthesis is executed by repetition of the following three simple operations that form a cycle: 1. Dividing the solid support into equal portions; 2. Coupling each portion individually with only one of the different amino acids; 3. Mixing and homogenizing the portions.

In the first round (Figure 4.1.a) the amino acids are coupled to equal portions of the resin and the final product - after recombining and mixing the portions - is the mixture of the three amino acids bound to resin.

In the second cycle, this mixture is again divided into three equal portions and the amino acids are individually coupled to these mixtures. In each coupling step, three different resin bound dipeptides are formed, so the end product is a mixture of 9 dipeptides. In Figure 4.1.a. the divergent, vertical and convergent arrows indicate dividing, coupling (with one kind of amino

56

acid) and mixing, respectively.

Figure 4.1. The split-mix synthesis. a: Preparation of a library of nine dipeptides on solid support. Divergent arrows: dividing into equal portions; vertical arrows: coupling; convergent arrows: mixing and homogenizing, b: 27

tripeptides, c, d and e: 81 tetrapeptides. Green circle represent resin, red yellow and blue circles are amino acids or other organic monomers

A third dividing , coupling and mixing step that is not demonstrated in the figure would lead to the formation of a mixture of 27 resin bound tripeptides (Figure 4.1.b.) and a fourth cycle would produce 81 tetramers (c, d and e in Figure 4.1.). 4.1.1.1. The key features of the split-mix synthesis

The split-mix synthesis has several key features that are crucial to the method's utility in the pharmaceutical discovery process.

Efficiency. By examining Figure 4.1. it can be seen that starting with a single substance

(the resin, used as the solid support), after each coupling step the number of compounds is

a b c d e

57

tripled: first 3x1=3 resin bound amino acids, then 3x3=9 resin bound dipeptides and, if the process is continued, in the third cycle 3x9=27 resin bound tripeptides and in the fourth cycle 81 tetrapeptides are formed. This means that the number of peptides increases exponentially after each coupling step (Table 4.1.).

This is the reason why the split-mix method is so productive. Table 4.1. also shows that while the number of the products increases exponentially in each step the number of coupling cycles, that can be considered as a measure of the invested labor, remains constant.

Table 4.1. Exponential increase of the number of peptides in split-mix synthesis

Step number

Number of amino acids

Number of reaction vessels

Number of cycles in one step

Total number of cycles

Number of peptides

1 3 3 3 3 31=3 2 3 3 3 6 32=9 3 3 3 3 9 33=27 4 3 3 3 12 34=81 As the synthesis proceeds, the invested labor increases only linearly. Linear increase of

labor and exponential growth of the number of products: this is the reason of the exceptionally high efficiency of the method.

Table 4.1. also shows the number of reaction vessels that are needed to execute the synthesis. This is also only three is each step: one reaction vessel for each amino acid or other kind of building block. The number of the reaction vessels do not depend on the number of compounds formed in the process.

If 20 different amino acids are used in the synthesis, the number of peptides in each coupling step is increased by a factor of 20. The number of peptides (Np) can be expressed by the following formula where n is the number of the coupling steps, that is, the number of amino acids forming the peptides.

Np = 20n

After executing 5 coupling cycles with each of the 20 amino acids, for example, more

than 3 million peptides are present in the mixture. Such a synthesis does not need millions of reaction vessels neither thousands of years for their preparation. It is enough to use 20 reaction vessels, one for each amino acid, and the pentapeptide library can be prepared in a couple of days. The number of the executed coupling cycles (Nc), as expressed by the following formula, increases only linearly with the length of peptides.

Nc = 20n

Formation of all possible sequences. Another feature of the split-mix synthesis is that all possible combinations of amino acid building blocks are represented in the synthesized peptides. This is clearly shown by the simple example outlined in Figure 4.1. No more sequential orders of the red, yellow and blue circles can be deduced than those present in the dimmers, trimers and

58

tetramers demonstrated in Figure 4.1. This combinatorial nature of the composition of the mixtures synthesized by the split-mix method is reflected in their name: "combinatorial libraries." This combinatorial feature of the split-mix synthesis holds for preparations of non-sequential (e.g. cyclic and other) libraries, too. Formation of all possible sequences is the consequence of equally dividing the resin mixtures into the reaction vessels of the next coupling step. As a result of this operation all products formed in any reaction vessel are evenly distributed among the reaction vessels of the next reaction step. This can be considered as the combinatorial distribution rule that governs the product formation in the combinatorial process.

Formation of the products in one to one molar quantities. Peptides are considered to be

natural compounds although certainly not all peptide sequences are found in nature. Peptide libraries are most often prepared in order to find biologically active substances among them. Other kinds of organic libraries are also synthesized for the same purpose. In the identification process, or in screening of the libraries, the goal is to find the biologically most effective component of the mixture. Serious problems may arise in screening if the peptides are not present in equal quantities in the mixture. A low activity component, for example, if it is present in a large amount, may show a stronger effect than a highly active component present in a much lower quantity. Therefore, it is important to prepare libraries in which the constituents are present in equal molar quantities. The split-mix method was designed to comply with this requirement.

After each round of couplings, the resin is thoroughly mixed. This ensures that before dividing the resin the mixture is nearly homogenous. If the resin is divided into equal portions the previously formed peptides can be supposed to be present in equal number and in equal molar quantities in each portion. The coupling of any portion of the resin with an amino acid does not alter the number or the molar ratio of the peptides originally present in the mixture; simply adds the same amino acid to each sequence. Consequently, the molar ratio of the newly formed peptides is expected to be the same as the molar ratio of originally present ones. That is, the new sequences are formed in equal molar ratio.

In addition to the execution of couplings with equal portions of resin samples it is also important that the couplings are carried out on spatially separated samples adding a single amino acid to each sample. This makes possible to use appropriate chemistry to drive each coupling reaction to completion regardless of the reactivity of the amino acids. As a result, both the number of peptides and their equimolar ratio is preserved in every portion and in each step. It is worthwhile to note that the equimolarity could be altered at will if for some reason it would be advantageous. Simply unequal portions should be used in some couplings. Applying a larger portion of resin in one of the couplings, for example, would result in formation in larger molar quantity of a subgroup of products. The parallel nature of the split-mix synthesis and formation of individual compounds. The split-mix procedure has another intrinsic feature which plays an important role in screening and gives a unique character to the method: in any individual bead of the solid support, only one kind of peptide is formed. This may seem surprising at first glance, but becomes quite understandable upon closer examination. In Figure 4.2. the fate of a randomly selected bead is followed in a three step coupling process.

59

Let’s suppose that the bead in the first reaction step randomly finds itself in the reaction vessel where the coupling is done with the “red” amino acid. Consequently, to all of the functional groups of this bead – and to those of all of the other beads in the same vessel – the “red” amino acid is attached. In the next step the bead is in the vessel where the “yellow” amino acid is added. To all coupling sites this amino acid is coupled. In the third step, for similar reasons, all dipeptides are elongated with the “blue” amino acid. Thus all peptide molecules that form in the bead are the same. The sequence of all peptides is “blue-yellow-red” (the reversed order of couplings).

The beads behave in the process like independent reaction vessels. The content of these reaction vessels is not interchanged with those of the other ones. Any selected bead randomly travels through the successive reaction vessels and the final sequence stores the information about the route the bead traveled in the course of the synthesis.

Figure 4.2. Formation of a single substance in each bead. Small yellow, red and blue circles: amino acids or other kinds of organic monomers; the large green circle: an arbitrarily selected bead of the solid support randomly appearing in different

reaction vessels in the three coupling steps. The formation of one substance in each bead is a very important feature of the split-mix

synthesis. If the products are cleaved individually from separated beads, then they can be examined as individual substances like those produced by the parallel synthesis. Furthermore, if the formed compounds are not cleaved from the beads only the protecting groups are removed the products can also be tested as individual substances. The libraries in which the products remain

60

bound to the beads are called tethered libraries. Such libraries can be prepared by attaching to the resin the first amino acid by a cleavage resistant bond. The possibility of screening the products as individual compounds like those produced in the parallel synthesis ensure an enormous advantage in applications.

When comparing the products of the split-mix synthesis to those produced in a parallel process attention have to be called to an important difference. In parallel synthesis not only individual substances form in the reaction vessels, but the position of the reaction vessel in the reaction block unambiguously determines the identity of the product. The coordinates (row and column) identify the expected products since the synthetic history of each well, that is the added reagents and their order is exactly known. The situation in split-mix synthesis is different. Although each bead contains a single product it is not possible to easily identify the content. All beads look the same and the synthetic history of the beads is unknown. This means that if we determine is some way or other that the content of a selected bead, say a peptide, has a useful property all that we know is only the length of the peptide and this is the same for all components of the library. Neither the amino acid composition nor the sequence of the amino acids is known. If we want do know these data they have to be determined in a separate process that is called deconvolution. In the case of peptide libraries this can be done by scarifying at least a part of the content of the bead for sequence determination.

If a component of a tethered peptide library is examined, the best choice is to submit the bead to sequence determination using a peptide sequencer.4 On the other hand, if the peptide is examined in cleaved form the appropriate choice is to determine the sequence by mass spectrometry.5

After carrying out the synthesis of a peptide library, all peptides can be cleaved from the support and this way a mixture of free peptides is formed. These libraries are called soluble peptide libraries. In such library millions of peptides may be present and finding a bioactive peptide among them seems, at first glance, like finding a needle in a hay stack. Nevertheless, appropriate strategies have been developed to solve the problem. These strategies will be described later.

Applicability of the split-mix method in the synthesis of organic libraries. Although the

split-mix method was developed with intention to prepare large number of peptides, it was clear from the beginning that the method would be applicable for the synthesis of different families of other kinds of organic compounds, too. The series of non-peptide compounds are usually called organic libraries. Since most organic compounds are prepared by multi-step synthesis, it is quite obvious that the split-mix synthesis can be used for preparation of organic libraries. It has to be made clear that the split-mix method can only be applied in the synthesis of organic libraries if the chemistry of the process is well developed. The advent of the combinatorial era brought to light the importance of the solid phase organic reactions and, as a result of an intensive development, a large number of previously described solution phase organic reactions have been optimized to solid phase (see the second chapter). These reactions are applied in both parallel and split-mix approach. From the point of view of the pharmaceutical research and many other applications, the organic libraries are very important. Peptides are not the most preferred drug candidates because of their high susceptibility to enzymatic degradation. The ideal drug leads are small organic compounds due to their, in general, more favorable pharmacodynamic properties.

The use of organic libraries prepared by the split-mix method brings about a problem that does not occur with peptide libraries. Identification of a library component formed in a bead is not so easy than in the case of peptide libraries. Determination of the structure of an organic

61

compound in most cases is more complicated than that of a peptide. In order to circumvent this problem, different encoding methods have been developed.

4.1.1.2. Encoding of beads in the synthesis of organic libraries As already explained in the previous section, when peptide libraries are prepared the sequence of the peptide formed in a bead depends on the synthetic history of the bead (Figure 4.2.). The structure of an organic compound formed in a bead is also determined by the synthetic history of the bead, that is, by the route the bead traveled through the reaction vessels during the synthetic process. Methods have been developed in order to chemically record, in parallel with the synthesis of the library components, the route of all beads. Encoding chemical tags are be attached to the beads in processes different from those that are applied for coupling of the building blocks (the two reactions need to be orthogonal). Similarly, at the end of the synthesis the tags have to be cleavable from the beads separately from the products. The chemical tags that carry the information of the route of the beads also need to be easier analyzed than determining the structure of the synthesized compounds. Two different approaches were suggested for chemical encoding.

Encoding with sequences Binary encoding

When encoding by sequences, the encoding tags are either oligonucleotides6-8 or peptides.9,10 Their sequences encode both the identity of organic building blocks coupled to the bead and the order of their coupling. Figure 4.3a shows a bead with organic molecules and the encoding tags at the surface. The white-black-gray-white, squares encode the “yellow”, “blue” and “red” organic monomers and their yellow-blue-red-yellow coupling order. The code can be read by determining the amino acid sequence of the peptide or the nucleotide sequence of the oligonucleotide tag. The use of oligonucleotide tags has advantages when compared to the peptide ones. Because of the possibility of amplification, their sequence determination needs much less quantity than peptides do.

Figure 4.3. Beads encoded by sequence (a) and binary code (b) The encoded synthesis of organic libraries follows the general route of the split-mix method with one exception. A fourth operation is added to the usual three ones of the coupling cycle: coupling the units of the code to the beads.

a b

62

1. Dividing the solid support into equal portions; 2,3. Coupling each portion individually with only one of the organic building blocks; 2,3. Coupling each portion with the encoding unit; 4. Mixing and homogenizing the portions. The encoding unit can be coupled either before (2) or after (3) the organic building block. Figure 4.4. shows the first cycle of such synthesis.

Figure 4.4. First cycle of an encoded synthesis. Green cycles: support, yellow, blue and red

cycles: organic building blocks, white, grey and black squares: units of the code. In the binary encoding system the coding units are different organic molecules and their combination forms the code. In one of the binary encoding method the encoding molecules are halobenzenes carrying a varying length hydrocarbon chain (pink structures in Figure 4.3.b) attached to the beads through a cleavable spacer.

O O O

ONO2

HOOC Ar( )n

Figure 4.5. Structure of a binary encoding molecule

The structures of some aryl groups that appear in the electrophoretic tags are shown below.

Linker Electrophoretic Tag

63

Cl Cl

ClCl

Cl H

Cl

HCl

Cl H

F

HCl

Cl

It is characteristic for the binary encoding technique that the coding units do not form a sequence. It is simply their presence which codes for the organic building blocks and their position. In their original paper Ohlmeyer and his colleagues11, demonstrated the method for encoding peptide sequences. By 18 different coding units arranged according to a binary coding format the authors were able to code all sequences in a 117,649 member peptide library, formed by varying 7 amino acids (D, E, I, K, L, Q and S) in six positions. The presence of the coding units could be determined after cleavage in a single step by electron capture gas chromatography. Table 4.2. shows a simple example constructed in order to demonstrate the principle of binary encoding. The nine different tags (T1-T9) are used to encode the structure of 343 organic molecules synthesized by using the building blocks A1-A7, B1-B7 and C1-C7 in the first, second and third coupling step, respectively.

Table 4.2. Binary encoding

Coupling #1 Blocks Codes

Coupling #2 Blocks Codes

Coupling #3 Blocks Codes

A1 T1 B1 T4 C1 T7 A2 T2 B2 T5 C2 T8 A3 T3 B3 T6 C3 T9 A4 T2T1 B4 T5T4 C4 T8T7 A5 T3T1 B5 T6T4 C5 T9T7 A6 T3T2 B6 T6T5 C6 T9T8 A7 T3T2T1 B7 T6T5T4 C7 T9T8T7

It can be seen that that the tags T1, T2 and T3 and mixtures formed from them are encoding A1 to A7. Similarly T4, T5 and T6 encode B1 to B7 and T7, T8 and T9 for C1 to C7. It can be read from the table that the code for the compound formed from the building blocks A1B1C3, for example is T9T4T1. Similarly A2B3C4 and A7B7C7 are encoded by T8T7T6T2 and T9T8T7T6T5T4T3T2T1, respectively. The table shows that a building block in most cases is coded by more than one tag. These tags are attached to the beads in a single operation using the mixtures of the tags as reagents. The binary encoding system proved to be very successful in practice. Encoding tags other than halobenzenes have also been proposed and successfully used in practice.

64

4.1.1.3. Realization of the split-mix synthesis In order to experimentally realize the split-mix synthesis a simple device has been constructed in the author’s laboratory that is still in use. The photo of this device is seen in Figure 4.6.

Figure 4.6. Manual device for split-mix synthesis. Vertical and tilted position. The device is an aluminum tube mounted on a laboratory shaker. On one side of the tube there are two rows of altogether 20 holes to which reaction vessels can be attached. The reaction vessels that are normally used for solid phase synthesis (Figure 2.3.) were inserted into the holes and tightened by applying plastic rings. The unused holes were stopped. One end of the aluminum tube was attached to a waste container and the system could be evacuated by a water pump. The tube could be twisted around its axis. The Figure shows the tube in two positions. The left and right photo shows the reaction vessels in nearly vertical and tilted positions, respectively. The vertical position of the reaction vessels was used when the resin was portioned into them, when reagents or solvents were added and when solutions were removed. The reaction vessels stayed in tilted position and shaking was applied during the coupling reactions and the removal of protecting groups. In the synthesis of peptide libraries 200-400 mesh resin (capacity 0.5 mmol/g) was used and swelled in DMF prior to portioning. This resin contains about 10 million beads per gram. The following operations were typical in a coupling cycle. Portioning. The resin was suspended in DMF/DCM (2:1 v/v) in a round bottom container and was continuously mixed by bubbling nitrogen though it. The density of the solvent mixture was very near to that of the resin so the slurry could be kept homogenous during the portioning operation which was carried out by pipetting equal volumes of the slurry into the reaction vessels. After the first round of pipetting a small volume remained in the flask. This was diluted with the solvent and the pipetting was repeated in order to transfer all of the resin into the reaction vessels.

65

Removal of the terminal protecting group. The protecting group was removed by shaking with a solution of TFA (Boc strategy) or piperidine (Fmoc strategy) then washed several times with solvents. Coupling. A DMF solution of protected amino acid containing HOBt and a solution of DIC was added then shook for about one hour. After removing the solution, the resin was washed several times with DMF and DCM. Mixing. After addition of DCM/DMF (2:1 v/v) the reaction vessels were removed from their place and their content was poured into the round bottom container. The remainder was also washed into the container where the slurry was mixed by nitrogen bubbling. At the end of the synthesis a deprotecting cocktail was added to the thoroughly washed resin and after shaking then the solution was separated from the resin by filtration and dried. Productivity. In the synthesis of peptide libraries from 20 amino acids (although cystein was usually omitted) and using the manual device (Figure 4.6.) one elongation step, that is one coupling with each of the 20 amino acids, could easily be realized in one day. Taking this speed as standard, the number of synthesized peptides is shown in Table 4.3.

Table 4.3. Productivity in peptide synthesis

Peptides Number of peptides Number of days Dipeptides 400 2 Tripeptides 8,000 3 Tetrapeptides 160,000 4 Pentapeptides 3200000 5 Hexapeptides 64,000,000 6 Heptapeptides 1,280,000,000 7

The data of Table 4.3. are striking. In as short time as a week more than 1 billion peptides can be prepared. This really shows the exceptionally high productivity of the method. Before using the method, we did not even dream about anything comparable to this. It is worthwhile to note that during the synthesis of a peptide library of a given length all the libraries of the shorter peptides are also formed. Of course, if needed, samples of all these libraries can be separated from the resin mixtures. If a pentapeptide library is synthesized in 5 g resin, and in the tripeptide and tetrapeptide phase 12.5 mg and 250 mg resin is removed, respectively, we end up with a tripeptide, a tetrapeptide and a pentapeptide library and in these libraries the components are present in practically the same molar quantities. Identification of the components in mixtures containing thousands or millions of peptides is impossible. Therefore, when the split-mix synthesis was first tried experimentally, very simple libraries were prepared containing only 9 to 180 peptides. The components of the synthesized peptide mixtures were identified by two dimensional high voltage paper electrophoresis. In order to facilitate the identification, a software was developed. Using this software, the sequences of the expected peptides could be generated by computer. Based on the sequences the computer also

66

calculated the molecular weights, the electric charges of the peptides in two different (pH 2, and pH 6.5) buffers. Based on these data, the relative electrophoretic mobilities were derived and transformed into two dimensional electrophoretic maps. The computer predicted maps were compared with the experimental ones so the products of the synthesis could be identified. The software made possible to generate all components of huge peptide libraries. These were the first examples of what are called today virtual libraries. Figure 4.7. shows the predicted electrophoretic map of the haxapeptide library containing 64 million components. Migrations in horizontal and vertical directions are supposed to occur at pH 6.5 and pH 2, respectively. YYYYYY is the last generated sequence.

Figure 4.7. Predicted two dimensional electrophoretic map of 64 million hexapeptides 4.1.1.4. Automation of the split-mix synthesis Quite shortly after publishing the split-mix method in 1991, an American company, Advanced ChemTech Inc. constructed and manufactured an automatic machine that was capable to carry out all the operations of the split-mix synthesis automatically under computer control. This ACT 357 machine is the only as yet commercialized automatic split-mix synthesizer. At present the synthesizer is produced and commercialized by aapptec. The device was designed to be used for preparation of peptide libraries but can also be applied for the synthesis of organic libraries if the reactions can be run at room temperature and at atmospheric pressure. The photo of the synthesizer is seen in Figure 4.8. The machine has:

(i) A Teflon reaction block (1) with 36 reaction vessels and one collection vessel for combining and mixing the resin.

(ii) A rack (2) for the bottles for the solutions of monomers that can be either protected amino acids or other kinds of building blocks.

(iii) Two arms each moving in x,y directions and holding a probe. The needle like probe of Arm 1 (3) transfers solvents, solutions of the monomers and solutions of reagents into reaction vessels. This probe is able to spray the solvent and so it can be used to wash

67

the walls of the reaction vessels and that of the collection vessel. The probe of Arm 2 (4) has a wide tip and transfers slurries of resin from the reaction vessels to the collection vessel and back. This probe is also capable to transfer solvent into the collection and reaction vessels.

(iv) Five small bottles to hold solutions of reagents (5) (v) The synthesizer has 3 bottles (6) for storing solvents and a waste container (7). (vi) The computer seen in the photo controls all operations.

The computer can easily be programmed to control the synthesis of different kinds of libraries using different kinds and different number of monomers in each step, applying reagents in different molar concentrations. Double or triple coupling and different coupling times etc. can also be programmed.

Figure 4.8. The ACT 357 automatic split-mix synthesizer The arrangement of the tabletop of the machine can be better viewed in Figure 4.9. The Teflon reaction block (1) contains the conic collection vessel (2) and the reaction vessels (3).

Figure 4.9. The tabletop of the ACT 357 synthesizer

1

2

3 R4

R5 R2

R32

R1

5 6 4

2 1

4

3

6

7

5

68

Both the collection vessel and the reaction vessels are equipped with frits at the bottom so their liquid content can be removed by applying vacuum. The whole reaction block can be shaken at adjustable speeds by an orbital shaker.

The solutions of the monomers, usually protected amino acids, are placed into bottles (4) that are stored in places defined by a rack. The bottles are closed by septum. A group of 5 bottles is also found in defined places in the table top. The coupling reagents are stored in these bottles (R1-R5) and they are also closed by septum. There are also two cleaning-waste stations in the table top: one for Arm 1 (5) and another one for Arm 2 (6). The stand by position of the arms is above the center of these stations. In the left cleaning station (5) the tip of the probe of Arm 1 can be cleaned by solvent. This is very important. When Arm 1 transfers solution of a building block into a reaction vessel the needle like probe penetrating through the septum of the container is immersed into the solution, removes the programmed volume of solution and transfers it into the programmed a reaction vessel. In order to avoid cross contamination both the inside and the outside of the probe must be washed. This happens at the station (5).

Before starting to work with the machine both arms must be calibrated: Arm 1 for reaction vessels, collection vessel, monomer bottles, reagent bottles and the left cleaning station (5) and Arm 2 for reaction vessels, collection vessels and the right cleaning station (6). These are the places that are visited by the two arms in the course of the synthesis. As a result of calibration the exact x,y coordinates of the reaction vessels, collection vessel, reagent bottles, monomer bottles and the cleaning-waste stations are stored in the memory of the computer. The z coordinates of the arms also need to be calibrated.

The content of each monomer container must also be defined as well as the content of the 5 reagent bottles and the 3 system fluids placed in the 3 solvent bottles.

The computer can be instructed to initialize the stepwise operations of the synthetic procedure by commands that are entered into the software (ChemFiles). Sequential execution of all commands of the ChemFile result in fully automatic realization of the synthetic process.

Commands.

Flush. The command is used to clean and prime the system fluid lines at the beginning of a synthesis or when a system fluid bottle is changed.

Split. The probe of Arm 2 removes equal volumes from the resin slurry present in the collection vessel and transfers them into the defined reaction vessels. The volumes and the repetition of the whole process can be specified. Combine. By this command the probe of Arm 2 removes a defined volume of resin slurry from a defined group of reaction vessels and delivers it into the collection vessel for mixing. The command can also be used to combine the liquid from several reaction vessels into a single reaction vessel. In this case the number of the destination vessel also needs to be entered. Mix. The command is used to shake the whole reaction block. It allows mechanical (vortex) mixing and in addition nitrogen bubble mixing for the collection vessel. Dispense sequence. A defined volume of liquid is dispensed from the containers of a source rack into the containers of a destination rack.

Dispense system fluid. The command allows a specified amount of a selected system fluid to be delivered to a range of vessels defined as destination rack. Transfer. The command provides for transfer of a specified volume of reagent (from a defined reagent bottle) to a range of reaction vessels. Also liquid can be moved between any calibrated position on the tabletop.

69

Empty. The empty command establishes a time value in hours, minutes or seconds for applying vacuum for emptying the liquid from the reaction vessels or the collection vessel. Wash. The wash command is usually applied to wash down the resin from the walls of the collection and reaction vessels after mixing. The liquid is sprayed to the walls and the vessel is simultaneously emptied. Wait for. The command offers a timer with a range from seconds to hours, during which all operations are paused. Repeat. The repeat command allows the user to develop loops within the ChemFiles in order to repeat an operation or a sequence of operations. Besides construction of the ChemFile, preparation for the synthesis of a peptide library involves the assignment of the amino acids to the bottles of the rack and filing the bottles with the solutions of the protected amino acids containing HOBt. The reagents also need to be assigned to their bottles and fill into them. An example of the assignment is shown in Table 4.4.

Table 4.4. Assignment of reagents to bottles

Bottle Reagent Solvent R1 DIC NMP R2 HBTU DMF R3 DIEA NMP R4 MeOH R5 Piperidine DMF

The system fluids (solvents) also have to be assigned and filled into the bottles. A possible

assignment is seen it Table 4.5.

Table 4.5. Assignment of solvents to their bottles

Bottle number Solvent (SF) 1 DMF 2 DMF/DCM 2:1 3 DCM

A peptide library is built up from amino acids. In construction of the ChemFile the

“Library Builder” function of the software can be used to assign the amino acids to the different coupling positions and define the reaction vessels to where they are delivered for coupling. Coupling position 1 means the C-terminus of the peptides. The amino acids occupying this position are coupled first to the support. In order to prepare a full library the same amino acids need to be assigned to all coupling positions. Table 4.6. shows the data of the Library Builder when a partial pentapeptide library is prepared. It can be seen that the number of amino acids assigned to the different coupling positions is different. In the coupling positions 1 and 4, for

70

example, 18 and 14 amino acids are used, respectively. Although it is not seen in Table 4.6. the library builder calculates the expected number of peptides in the library (884,520) and based on the quantity and mesh size of the resin also gives the number of beads per peptide.

The ChemFile is a series of commands arranged in the order of execution. Table 4.7. shows the first ten rows of a ChemFile. The execution involves the swelling and washing of the resin, diluting the resin with solvent and the first round of splitting the resin transferring 2.5 ml of slurry into each of 18 reaction vessels.

Table 4.6. The Library Builder

C o u p l i n g p o s i t i on Number of Reaction vessels

8 7 6 5 4 3 2 1

1 A A A A A 2 R R R R R 3 N N N N N 4 D D D D D 5 K Q Q Q Q 6 L E E E E 7 M G G G G 8 F M H H H 9 P F I I I 10 S P K K K 11 T S L L L 12 V T M S M 13 W V F T F 14 W P V P 15 S W S 16 T T 17 V V 18 W W 19 20

The synthesis normally ends with the resin combined in the collection vessel in addition

to this there are other choices, too. In order to be able to calculate the weight of full peptide libraries, let's suppose that only

one peptide is responsible for the biological activity. Let's also arbitrarily fix the quantity of this peptide (and therefore all peptides in the mixture) to 1 pmol. The real quantity requirement, depending on the sensitivity of the screening experiment and other factors, can easily be deduced from this quantity.

71

Table 4.7. Example of commands in a ChemFile CV: collection vessel, SF system fluid

Commands 1 Fill CV with 75.0 ml of SF1 2 Mix for 3 minutes on speed 2 3 Empty collection vessel 4 Fill CV with 75 ml of SF2 5 Mix for 10 minutes on speed 2 6 Empty CV 7 Wash CV 8 Fill CV with 63.0 ml of SF2 9 Mix for 3 minutes on speed 2 10 Split 1 thru 18 using 2500 µl

4.1.1.5. Preliminary considerations when planning experiments with peptide libraries Full peptide libraries are often prepared from 19 amino acids leaving out cysteine. The forthcoming considerations are based on such peptide libraries. The peptide libraries have an intrinsic feature that is advisable to take into account when planning experiments with them: the number of their components increases exponentially with the number of the varied positions and, as a consequence, both the weight of the libraries and the weight of the solid support needed for their preparation also increase exponentially. The effect of the number of the varied positions on the number of components of the full libraries is shown in Table 4.8.

Table 4.8. Number of peptides in libraries depending on the number of varied positions

Number of varied positions

Number of peptides

2 361 3 6,859 4 130,321 5 2,476,099 6 47,045,881 7 893,871,739 8 16,983,563,041 9 322,687,697,779 10 6,131,066,257,801

In order to be able to calculate the weight of full peptide libraries, let's suppose that only one peptide is responsible for the biological activity. Let's also arbitrarily fix the quantity of this peptide (and therefore all peptides in the mixture) to 1 pmol. The real quantity requirement, depending on the sensitivity of the screening experiment and other factors, can easily be deduced

72

from this quantity. Table 4.9 shows the weight of full libraries depending on the number of their varied positions. It can be seen that if the number of the varied positions is near 10, the weight of the libraries is so high that solubility problems may arise in screening. As a consequence, the weights of the libraries are certainly one of the factors that need consideration in planning.

Table 4.9. Approximate weight of libraries containing each peptide in 1 pmol quantity

Number of varied positions

Weight Units

2 92 ng 3 3 µg 4 65 µg 5 2 mg 6 35 mg 7 765 mg 8 17 g 9 353 g

10 7 kg The quantity of the resin that is needed for the synthesis is expected to be - and really is - even a bigger problem. Table 4.10. shows the weight of the resin needed to prepare all peptides in 1 pmol quantity. In practice, these quantities are expected to be even higher than indicated in Table 4.10 because the libraries are usually prepared not for a single but for a series of experiments and the screening tests may also have lower sensitivity. Problems may occur in handling such large quantities of resin and, if the number of the varied positions is high enough, it is practically impossible to carry out the synthesis. Consequently the weight of the resin needs to be considered carefully.

Table 4.10. Approximate weight of the resin needed to prepare libraries containing each peptide in 1 pmol quantity

Number of varied positions

Sum of moles Units Weight of resin Units

2 361 pmol 720 ng 3 7 nmol 14 µg 4 130 nmol 261 µg 5 2 µmol 5 mg 6 47 µmol 94 mg 7 894 µmol 2 g 8 17 mmol 34 g 9 323 mmol 645 g

10 6 mol 12 kg

73

Another problem which deserves consideration before beginning a synthesis is the ratio of the number of beads of the resin to the number of the expected peptides. Since only one peptide forms in each bead, the maximum number of peptides is limited by the number of beads. Furthermore, two essential operations of the method, mixing and portioning, are influenced by probability. As a consequence, if the number of the beads is equal to the number of peptides not all peptides are expected to form and also deviations from the equimolarity are expected. For this reason, formation of all expected peptides, as well as their near equimolarity, is ensured only if the number of beads well exceeds the number of peptides. A ten fold excess of the beads can be considered quite safe. For reasons outlined above, when very complex libraries are prepared, it is desirable to choose as small bead size as possible, for example, 200-400 mesh (diameter: 38-75 µm) resin. Each gram of this resin contains about 10 million beads. Table 4.11. shows the quantity of resin needed if the number of beads equals or exceeds 10 times the number of peptides. The data in Table 4.11. clearly demonstrates that, due to practical reasons, the number of varied positions in full libraries is limited to about 6 or 7. The difficulties arising from the overwhelmingly large number of peptides in some full libraries can be circumvented by preparing their partial libraries. One may follow two different approaches for doing this: 1. Reducing the number of the varied amino acids; 2. Reducing the number of the varied positions. It is, of course, possible to combine the two approaches. It seems worthwhile to consider in some detail both possibilities.

Table 4.11. Approximate weight of the resin if 1 or 10 beads are assigned to each peptide

Number of varied positions

Weight (1 bead)

Units Weight (10 beads)

Units

2 36 µg 361 µg 3 686 µg 7 mg 4 13 mg 130 mg 5 248 mg 2 g 6 5 g 47 g 7 89 g 894 g 8 2 kg 17 kg 9 32 kg 323 kg

10 613 kg 6 t Table 4.12. shows that the number of components in the libraries can effectively be reduced by reducing the number of the varied amino acids .

74

Of course the chemist is not restricted to use the same number of amino acids in all positions. An example of an octapeptide library is demonstrated in Table 4.13. that is constructed by varying different number of amino acids in different positions.

Table 4.12. Number of peptides in partial libraries depending on the number of varied amino acids

Nnumber of varied positions

N. of amino acids 5

N. of amino acids 10

N. of amino acids 15

2 25 100 225 3 125 1,000 3,375 4 625 10,000 50,625 5 3,125 100,000 759,375 6 15,625 1,000,000 11,390,625 7 78,125 10,000,000 170,859,375 8 390,625 100,000,000 2,562,890,625 9 1,953,125 1,000,000,000 38,443,359,375 10 9,765,625 10,000,000,000 576,650,390,625

Table 4.13. Partial octapeptide library deduced by varying different number of amino acids in different positions

Position Number of varied

amino acids 1 10 2 8 3 12 4 9 5 4 6 19 7 4 8 12

Total number of peptides 31,518,720 Intuition plays an important role when one decides which amino acid can be omitted in the synthesis. One has to be aware, however, that if a partial library is prepared and an amino acid critical to the activity of the potential active peptide happens to be among the omitted ones the active peptide and the activity of the whole library is lost. It is very convenient to prepare less complex libraries by reducing by one or two or even more the number of the varied positions. Each fixed position reduces the number of peptides by a factor of 19. The partial heptapeptide library of Table 4.14, for example, that has three non-varied positions has only 130,321 components.

75

Considerations of the abovementioned features of peptide libraries may, perhaps, help the potential user to be aware of the limitations of the library method, formulate a realistic research plan and, when possible, circumvent the difficulties.

Table 4.14. Partial heptapeptide library with 3 non-varied positions

Position Number of varied amino acids 1 19 2 1 3 19 4 1 5 1 6 19 7 19

Number of peptides 130,321 The examples in the considerations made in this section were peptide libraries. The conclusions, however, hold for organic libraries, too. 4.1.1.6. Full and partial libraries Although peptide libraries that are constructed from 20 or 19 amino acids in all coupling positions are usually considered full libraries, precise definition for such libraries that is generally accepted not yet exists. One may also consider any library synthesized by the split-mix method to be full library. Any other library that contains a smaller or larger fraction of the components of the full library – and no extra components – is a partial library. It is not easy to give an exact and at the same time short description of a library. For exact description all building blocks used in all coupling positions have to be indicated. This can be done in the form of a table. This is shown for a simple peptide library in Table 4.15. containing 400 components.

Table 4.15. Description of a tetrapeptide library

Coupling position Amino acids 1 A, F, G, H, R 2 H, I, K, L 3 D, E, F, T 4 A, G, K, T, W

If a full pentapeptide library, composed in every step from the twenty natural amino acids, is represented as shown in Figure 4.10. the sequences of the peptides can be read along the lines

76

drawn through one of the amino acids found in the five rows of the figure. In the case of pentapeptides 3.2 million different lines can be drawn each representing one of the theoretically possible 3.2 million pentapeptide sequences. Both libraries represented in Table 4.15. and Figure 4.10. can be considered full libraries. Any library in construction of which the same amino acids are used but their number is reduced in one or more coupling positions relative to those found in Table 4.15. or Figure 4.10. can be considered a partial library.

Figure 4.10. Pentapeptide sequences represented by lines The library of Table 4.16. for example, is a partial library of the full library of Table 4.15. In the synthesis of the partial library of Table 4.16. R and F are omitted in coupling positions 1 and 3, respectively and so the number of components is reduced from 400 to 240.

Table 4.16. Partial library of the full library of Table 4.15.

Coupling position Amino acids 1 A, F, G, H 2 H, I, K, L 3 D, E, T 4 A, G, K, T, W

The library of Figure 4.11. is a partial library of the full library represented in Figure 4.10. In the synthesis of this library in the coupling steps 1, 3, 4 and 5 all of the 20 amino acids are varied. In coupling step 2, however, a single amino acid glycine is used that is coupled to the resin without previous portioning. Coupling position 2 is a non-varied position and glycine is the amino acid occupying coupling position 2 in all peptides. All sequence lines cross glycine. The number of sequences, and consequently the number of possible sequence lines, is only 160,000. This is the number of the components of the full library divided by 20. As it will be shown later the partial libraries that have a single non-varied position play an important role in screening. They are often called sub-libraries.13 The synthetically easiest accessible and at the same time the simplest sub-libraries are

A C D E F G H I K L M N P Q

R S T V W Y 1

A C D E F G H I K L M N P Q

R S T V W Y 2

A C D E F G H I K L M N P Q

R S T V W Y 3

A C D E F G H I K L M N P Q

R S T V W Y 4

A C D E F G H I K L M N P Q

R S T V W Y 5

77

those ones that form at the end of a split-mix synthesis omitting the final mixing. Their non-varied position is the last coupling position. If 20 amino acids are varied in the synthesis, a single split-mix run need to be carried out (without the last mixing) and the process ends up with 20 sub-libraries. These sub-libraries have another interesting feature: if they are mixed a full library is formed. As 4.12. shows, this feature is the same for any full sets of sub-libraries having the same non varied position. The non-varied positions in sets a, b and c are coupling positions 1, 2 and 3, respectively, and it can be seen that each of the three sets form a full library.

Figure 4.11. A partial pentapeptide library with one non-varied position

Figure 4.12. Full and sub-libraries of a 27 component tripeptide library. The non-varied positions are at coupling position 1 (a), coupling position 2 (b) and coupling

position 3 (c), respectively.

A C D E F G H I K L M N P Q

R S T V W Y 1

G 2

A C D E F G H I K L M N P Q

R S T V W Y 3

A C D E F G H I K L M N P Q

R S T V W Y 4

A C D E F G H I K L M N P Q

R S T V W Y 5

a b c

78

As it was pointed out, the synthesis is a single run process of the sets of sub-libraries in which the non-varied position is the last coupling position, like in set c of Figure 4.12. In sets like a and b the non-varied positions are in first and intermediate coupling positions, respectively. The simplest way to prepare the components of these sets is to synthesize them separately, one by one. There are sub-library sets that do not form a full library. An example is demonstrated in Figure 4.13. In sub-libraries c, b and a the “blue” amino acid occupies the first, second and third coupling position, respectively. Both the “all yellow” and the “all red” sequences (marked by an arrow), for example, and also other ones would be missing from the mixture of the three sub-libraries, while other trimers are present in duplicates or triplicate (“all blue”).

Figure 4.13. A set of sub-libraries that do not form a full library As already pointed out, when the number of components in a full library is too large to synthesize it in a single run it is practical to prepare it in portions. It has to be taken into account, however, that some partial libraries are unpractical to prepare because their completion to a full library requires too much work.13 It is unpractical for example to prepare L2 as a portion of the full library L1 because L3 does not complete it to L1. The total number of components in L2 and L3 is only 32 while L1 contains 256 tetramers. Several other libraries would have been needed to be prepared in order to complete L2+L3 to L1. L4 and L5, however, are two reasonable choices for portions of L1. Both L4 and L5 have 128 components that add up to 256.

Coupling position 3 2 1

a b c

79

L1 L2 L3 L4 L5 A,D,E,F A,D E,F A,D E,F A,D,E,F A,D E,F A,D,E,F A,D,E,F A,D,E,F A,D E,F A,D,E,F A,D,E,F A,D,E,F A,D E,F A,D,E,F A,D,E,F 256 16 16 128 128 4.1.1.7. Unusual partial libraries As previously showed, in a split-mix synthesis all those compounds form that can be deduced by combination of the applied building blocks in their preparation. This is an important advantage of the method. It may occur, however, that not the whole library is needed only an arbitrarily selected subset. Such series of compounds, of course are not combinatorial libraries, can not be directly prepared by the method. This is a disadvantage. It is possible, however, to design and synthesize a combinatorial library that contains all the wanted compounds and in addition other components. Construction of a combinatorial library that contains an arbitrary set of compounds is demonstrated by a simple example. Suppose that the components of the set are the following five pentapeptides.

54321 ADKLL ADMLG GIFGP GIFLP GDMGL

Next, the amino acids appearing in the five coupling positions are recorded. Pos. 1 L, G, P Pos. 2 L, G Pos. 3 K, M, F Pos. 4 D, I Pos. 5 A, G A combinatorial synthesis using these amino acids as building blocks produces a library that contains all of the five pentapeptides. The number of components of the library is 3x2x3x2x2=72. This means that in addition to the 5 wanted peptides 67 extra peptides are formed. The number of coupling cycles in this synthesis is 3+2+3+2+2=12. If the 5 peptides are prepared by parallel synthesis the number of the coupling cycles is 25. Cohen and Skiena showed49 that the total number of components of the libraries can be reduced by increasing the number of the coupling cycles and by properly designing the synthesis. They developed software that makes possible to optimize the total number of components of the

80

libraries that contain the wanted compounds vs the number of the coupling cycles needed in the synthesis. One of their examples is the synthesis of an arbitrary set of 496 pentapeptides. One may think whether for the preparation of such a set the parallel method or the split-mix synthesis is more advantageous. The parallel synthesis of such a set needs 2480 coupling cycles that involves much work. They showed that by application of their optimization method the synthesis of a 20,000 member pentapeptide library that contains all of the 496 arbitrarily selected pentapeptides, needs only 324 coupling cycles. One may decide what is better: the reduction of the number of the coupling cycles by a factor of ca. 8 and accepting the presence of additional ca. 19,500 components in the mixture or preparation of the individual compounds in parallel by investing several times more in labor. The choice certainly depends on additional factors, too. The use of modified versions of the split-mix synthesis that applies macroscopic solid support units (see later) offers a much better choice for the preparation of arbitrarily selected series of compounds. 4.1.1.8. Binary synthesis using the split-mix procedure The concept of the binary synthesis was introduced by Fodor and his colleagues in a paper describing a new technique for combinatorial synthesis14 that will be discussed later. The split-mix method also proved to be applicable for carrying out binary synthesis.15 In order to realize a binary peptide synthesis the operations of a cycle of the split-mix procedure need to be modified as follows:

Figure 4.14. Flow diagram of a four step binary peptide synthesis

Resin

Mix No coupling

E

1/2 1/2

Mix No coupling

R

1/2 1/2

Mix No coupling

G

1/2 1/2

Mix No coupling

L

1/2 1/2

81

Divide the resin into two equal parts Couple an amino acid to one part of the resin Mix the two parts This cycle of operation is repeated in the process a pre-determined number of times using in each coupling step a different amino acid. It has to be emphasized that only one of the two portions of the resin is submitted to coupling and nothing is done with the second part. Figure 4.14. shows the scheme of a simple binary peptide synthesis coupling successively with the following amino acids: E, R, G and L. In the steps of the binary synthesis one part of the resin and all the peptides already formed in the resin remain unchanged. In each synthetic step only the second part of the resin undergoes coupling, and in this process, all the peptides formed in the previous steps are elongated with one amino acid. Table 4.17. shows the products formed in the process shown in Figure 4.14.

Table 4.17. Peptides formed in a four step binary synthesis

Coupl. step

Peptides in no coupling part Coupl. with

Peptides in coupling part after coupling

1 0 L L 2 0, L G GL 3 0, L, G, GL R R, RL, RG, RGL 4 0, L, G, GL, R, RL, RG, RGL E E, EL, EG,EGL, ER, ERL, ERG, ERGL

The zero in the table means unchanged empty resin. The final mixture contains the products found in the last row of the table including a fraction of the resin that remains unchanged. One of the products is ERGL, that is, the peptide formed from the four amino acids used in the synthesis and its sequence reflects the coupling order of the amino acids. The other products can be derived from the sequence of this tetrapeptide. All the sequences and amino acids that can theoretically be derived from this root tetrapeptide sequence by deletion of partial sequences or amino acids are found in the mixture. The components of the mixture are present in equimolar quantities. As it was pointed out by Fodor an his colleagues, the number of peptides formed in the binary synthesis (N) can be calculated from the number of coupling cycles (c) according to a simple formula.

N=2c This indicates a very high efficiency since N grows exponentially with the number of coupling cycles. In 22 coupling cycles, for example, more than 4 million components form. The number of components in the groups of peptides of different lengths, follow a binomial distribution. When c=10 the total number of components is 1024. The length of peptides varies from 0 to 10. The number of peptides belonging to different length is indicated in brackets: 0(1), 1(10), 2(45), 3(120), 4(210), 5(252), 6(210), 7(120), 8(45), 9(10) and 10(1).

82

One has to note that the N=2c formula calculates the maximum number of components that are formed only when in the root sequence every amino acid is represented only once. In many cases this requirement can not be maintained, for example, when the 20 L-amino acids are used as building blocks in more than 20 coupling cycles. Multiple occurrences of amino acids in the root sequence have two consequences:

(i) the number of components in the synthesized library is less than calculated from the formula and,

(ii) the equimolarity of components is no longer maintained since some compounds form in multiple molar quantities.

If an amino acid appears more than once in the root sequence, some peptides may form from more than one source. This happens, for example, if the binary synthesis is carried out on the basis of the following root sequence: EGGL. The products derived from this sequence are:

0, L, G, GL, G, GL, GG, GGL, E, EL, EG, EGL, EG, EGL, EGG, EGGL The products indicated in bold appear twice in the list so their quantity relative to the other components of the mixture is doubled. In the paper of Sebestyén et al.15 tricks are described how to avoid formation of the products in non-equal quantities. The binary synthesis may prove useful for exploration whether or not deletions in a region of a longer peptide lead to bioactive fragment(s). 4.1.2. Combinatorial synthesis using amino acid mixtures As pointed out in Chapter 1 Section 2. replacement of the single amino acids in the coupling cycles of the Merrifield solid phase peptide synthesis by a mixture of the 20 amino acids would, in principle, lead to the formation of a peptide library containing all the theoretically expected components. Geysen and his colleagues16 published such synthesis in 1986. They used amino acid mixtures in every coupling step of their solid phase synthesis and successfully applied the “Iteration method” (see details in Chapter 5) in the analysis of the library. The method is even more efficient than the split-mix procedure. In every coupling step of the synthesis only a single coupling operation is executed in a single reaction vessel in contrast with the 20 reaction vessels and 20 coupling operations needed in the split-mix procedure. In the split-mix procedure a total of 100 couplings are needed to prepare the 3.2 million pentapeptides. The amino acid mixture method needs only 5 coupling steps. There are, however, disadvantages too. One of the disadvantages was pointed out in Chapter 1 Section 2. It is known that the coupling rates of the amino acids differ from each other. As a consequence, formation of the peptides in 1 to 1 molar ratio can not be assured. Some peptides form in significantly higher molar quantity then others and some peptides do not even form. Efforts have been made to compensate these differences. Rutter and Santi described in their patent17, that the differences can in part be compensated by proper adjustment of the concentrations of amino acids in the coupling mixtures. The amino acids showing a slower coupling rate were represented in higher concentrations in the mixture. Full compensation, however, can not be achieved because in addition to the acylating components, the coupling rates are also effected by the acylated peptides themselves.

83

Another disadvantage relative to the split-mix method is that the one bead one product feature is completely lost. Since mixtures of amino acids are used in every coupling step, instead of single products, mixtures are formed in the beads. The synthetic history of every bead is the same. As a consequence, within the limits of statistics, the content of the beads is the same. This means that every bead contains all components of the library. The loss of the one bead one product feature is a very significant disadvantage. The library can be analyzed only as a mixture. The individual components of the library are absolutely not accessible. The method has been applied mainly in preparation of peptide libraries but non-peptide libraries have also been synthesized.19 4.2. Combinatorial synthesis using soluble support The very large majority of the classical methods developed for preparing organic compounds work in solution phase. The advent of combinatorial methods induced a fast development in the area of solid phase synthetic procedures. Nevertheless, most organic synthetic methods that are found in the literature are still applicable only in solution phase. In addition, the solid phase reactions are significantly slower than those in solution. A dissolved reagent molecule that is outside resin beads can react with a molecule attached to the solid support inside a resin bead only after diffusion into the solvent bound within the particle. The diffusion is a slow process so the solid phase reactions take a considerable longer time than those in solution phase. There were attempts by Shemyakin and others20-22 to substitute the cross linked polystyrene support introduced by Merrifield (Chapter 2) by linear soluble polystyrene to achieve homogeneous phase in reactions while preserving some advantages of the solid phase approach. Han et al.23 applied polyethyleneglycol (PEG) as support in synthesis of peptide libraries following the split-mix approach. One of the two hydroxyl groups of the polymer is blocked by a methyl group. The first amino acid or other building block can be attached directly or via a linker to the remaining free hydroxyl group at the other end of the polymer chain.

MeO-CH2-CH2-O-(CH2-CH2-O)n-CH2-CH2-OH PEG proved very suitable for this purpose since it is soluble in a wide variety of aqueous and organic solvents and its solubility provides homogeneous reaction conditions even when the attached molecule itself is insoluble in the reaction medium. Separation from the reaction medium of the polymer and the synthesized compounds bound to it can be achieved by precipitation and filtration. The precipitation requires concentrating the reaction solutions then diluting with diethyl ether or tert-butyl methyl ether. Under carefully controlled precipitation conditions the polymer precipitates in crystalline form. The above cited authors prepared 1024 pentapeptides in order to show the applicability of their method. Other types of compounds like polysaccharides and oligonucleotides have also been synthesized on PEG.24 In addition to providing conditions for faster and smoother reactions, application of the soluble supports has another advantage. The support does not have a bead form it is rather represented by a collection of individual molecules. As a consequence, the reactions are absolutely unaffected by statistics.

84

The method has disadvantages, too. The very important feature of the split-mix method that a single compound forms in one bead, is completely lost. In addition, the separation of the support from the reaction mixture is not as simple as filtering out the bead form resin. 4.3. Combinatorial synthesis on solid surface A very remarkable combinatorial synthesis was developed by Fodor and his colleagues by combining the solid phase synthesis with the photolithographic procedure applied in the fabrication of the computer chips. The method was published in 1991 under the title “The light-directed, spatially addressable parallel chemical synthesis”.25 The method makes possible to prepare an array of peptides or other kinds of molecules on the surface of a small glass slide. At the beginning the full surface is functionalized with aminoalkyl groups that are protected by the photo-labile 6-nitroveratryloxycarbonyl (Nvoc) groups. These protecting groups can be removed from definite regions of the surface by irradiation. The deprotected amino groups can be acylated with N-protected amino acids. The α-amino groups of the amino acids are also protected by the photo-labile Nvoc groups. The principle of the method is demonstrated in Figure 4.15. The example shown in Figure 4.15. demonstrates the synthesis of nine dipeptides from the amino acids A, G and K. Before each coupling step one or more areas of the slide are irradiated through a mask in order to remove the protecting groups from those areas. Then the slide is submitted to coupling with the indicated amino acid. This can be done by immersing the slide into the solvent containing the protected amino acid and the coupling reagent. Although the full slide is submitted to coupling reaction, coupling occurs only in the irradiated areas where the free amino groups are found. By completing a coupling cycle the full area of the slide becomes again protected. Before the next coupling cycle a new area have to be irradiated in order to produce free amino groups.

Figure 4.15. Formation of nine dipeptides in the light directed synthesis

In Figure 4.15. the irradiated areas are white and those shadowed by the mask are gray. The synthesis of the 9 dipeptides is completed in 6 cycles irradiation and coupling (a to f). It is

A

b

G

b

K

c

A

d

G

e

K

f

h i j

AA

GG

AG AK

GA GK

KA KG KK

g

85

remarkable that more peptides (32=9) form than the number of the executed coupling cycle (6) that is characteristic in the combinatorial processes. The dipeptides are formed in the locations shown in Figure 4.15/g. It is worthwhile to note that in fact only two different masks are needed in the synthesis: those two shown in Figures 4.15/a and 4.15./b. Mask positions d, c and f can be produced by rotation of mask a by 90o, 180o and 270o, respectively. If the synthesis is continued, in each elongation step finer masks need to be used and in each elongation step the number of the components of the library increases exponentially as in the split-mix synthesis. In the next elongation step, for example the masks 4.15./h and 4.15./i would be needed and the other mask positions could be presented by rotation of these two. Mask position j, for example, could be brought about by rotation of mask 4.15./h by 180o. After completing the next 6 coupling cycles with the amino acids A, G and K (at mask positions h, i, j, and at those positions brought about by rotating these by 90o, 34=81 tetrapeptides would form. As expected in a combinatorial synthesis, among these 81 tetrapeptides all sequences would be represented that can be deduced as a result of inserting the three amino acids (A, G and K) into coupling positions 3 and 4. After completing the couplings, before the library undergoes testing, the protecting groups, of course, have to be removed. In the testing experiments the products remain attached to the slide. The light directed synthesis in some respects is very similar to the split-mix method. In the synthesis of peptide libraries the invested work, that is the number of the executed coupling steps (Nc), linearly increases with the lengths of the peptides (n, the number of amino acids in the peptides), while the number of the components of the library (Np) increases exponentially with the length. If 20 amino acids are used in every step of the synthesis, the following formulae express the invested work and the number of the peptides formed in the process. Nc = 20n Np = 20n There are also differences relative to the split-mix method. One of the differences is that in the light directed method the couplings of one elongation step can not be executed in parallel, like in the split-mix procedure. The couplings with the single amino acids need to be carried out serially, one after the other. Furthermore, the light directing method is, of course unsuitable to prepare the libraries in large quantities. Another difference provides a very significant advantage for the light directed method. The identity of every product formed on the surface of the slide is exactly known. There is no need for a separate analytical process in order to identify the products. If the masks, their positions and the order of their application as well as the coupling order of the amino acids are known, the identity of the products in every location of the slide can be deduced. This makes application of the libraries in the testing experiments very simple. It has already been mentioned that the binary synthesis was first introduced and demonstrated in conjunction with the light directed synthesis. Its principle is that in each elongation step only half of the slide is submitted to coupling the other half remains unchanged. Figure 4.16. demonstrates the synthesis based on the same ERGL root sequence that was used in demonstrating the binary synthesis realized by the split-mix method. The white regions of the slide are irradiated then coupled with the indicated amino acid. The gray regions remain unchanged. Four elongation steps are executed (a, b, c and d) as shown in Figure 4.16. The

86

products are found in Figure 4.16/e. It can be seen that the products are the same as those formed in the split-mix binary synthesis (row 4 in Table 4.17).

Figure 4.16. Binary synthesis on the basis of the root sequence ERGL using the light directed method

In the introductory publication around one thousand peptides were synthesized on the surface of the slide. Since that time the number of the substances produced on a slide was very significantly increased. In practice, the method is applied for making oligonucleotide chips that are extensively used in nucleic acid analysis.26 On the surface of a chip less than 1.5 cm2 about 500,000 different oligonucleotides can be synthesized and a single silicon wafer may contain 49 to 400 different oligomer arrays. The light directed synthesis was developed at an American company, Affymax, and the chips are manufactured and commercialized by Affymetrix. More details of the method can be found in the home page of the company.27 4.4. Combinatorial peptide synthesis by biological methods In 1990 three different research groups introduced a new biological approach for producing peptide sequence libraries28-30. This approach is briefly exemplified by phage display libraries.31 First, an oligonucleotide library is synthesized chemically by a series of couplings with equimolar nucleotide mixtures. The formed oligonucleotides are then inserted into the DNA of phages. In the next stage the phages infect the host bacterium (usually Escherichia coli) and replicate together with the inserted “foreign” DNA segment. A library of phage clones forms. Each clone carries in its DNA a different “foreign” sequence segment which is expressed as a partial sequence of its coat protein. Every phage particle carries a many identical coat protein molecules with the same (foreign) peptide sequence fused to the outer end. In this respect the phages resemble to the beads in PM synthesis each containing an individual compound. The

L K R R

K

E

E

EL

GL

EGLL

RGL

L

ERGL

ERL

RL

ERG

RG

EG

G

ER

R

E

a b c d

e

G

87

DNA of the phage can be considered as an encoding tag since the sequence of the peptide can be determined (after ampflification) by sequencing the proper portion of the DNA. 4.5. Combinatorial synthesis using macroscopic solid support units The split-mix combinatorial synthesis, as it was shown, is a very efficient procedure and in addition it produces individual compounds since each resin bead contains a single compound. The quantity of the compound, however, that forms in a bead may prove too low (it is in the range of nanogram-microgram quantities) when compared to the needs. The total quantity of a compound that is formed in a split-mix synthesis is not necessary low because usually a large number of beads contain the same compound. Since, however, the content of the individual beads is not known, it is absolutely impossible to select those of them that contain the same product from millions of beads, cleave the product from the selected group of beads and so produce a larger quantity of the desired substance. In a split-mix synthesis a very large number of compounds are formed. We exactly know what compounds are present in the library and, again, it is not possible to pick out a desired compound or a pre-determined set of compounds for testing. These disadvantages of the split-mix method stimulated the efforts to modify the procedure. The goal of the modifications was to preserve the high productivity of the method and besides that

(i) produce the individual compounds with known identity and (ii) produce the compounds in large (multi-milligram) quantities.

In order to fulfill this goal the microscopic solid support units (beads) applied in the original method had to be replaced by macroscopic ones. In addition these macroscopic units had to be labeled somehow in order enable the experimenter to identify the product formed in the units. This means that the number of the support units applied in the synthesis need to be the same as the number of the products. Before starting the synthesis a product have to be assigned to each unit and properly labeled them. Assignment of the product involves assignment of the building blocks and their coupling order. In addition, during the synthetic process the units have to be distributed one by one into the reaction vessels according to the structure assigned to the products. The simplest way to label the units is to assign numbers to them from 1 to n where n is the number of compound to be prepared. The building blocks and their order can be recorded in a list or in a computer. Once the number of the unit is known, the operator can read from the list or from the computer which building block need to be coupled into the unit in a given reaction step. In other words the operator can identify the reaction vessel into which the unit has to be transferred in the given phase of the synthesis. This is demonstrated in figure 4.17. Box a in the figure shows 9 dipeptides assigned to the units 1 to 9. The sequences to be synthesized in the units 1, 2 and 3, for example, contain G in the first coupling position consequently they have to be placed into reaction vessel b where in the first synthetic step G is coupled into all units. The amino acid G coupled into the units is indicated in bold in the sequences. The arrows show where the units need to be transferred for the second coupling. The numbers on the arrows show the numbers of the units. So the units 1, 2 and 3 are transferred into the reaction vessels e, f and g, respectively. Similarly, the units 4, 5 and 6 from reaction vessel c are transferred into reaction vessels e, f and g, respectively.

88

Two essentially different approaches have been developed to solve the problem of the identification of the units. In one of the approaches physical labels are attached to the units. In the other approach no labels are used. The units are encoded by the position they occupy in space.

Figure 4.17. Sorting of macroscopic units

4.5.1. Encoding by attached labels. The radiofrequency and optical encoding methods The first approach for the modified split-mix synthesis using macroscopic solid support units was developed independently in two laboratories.32,33 In the suggested methods the solid support units are permeable capsules containing resin and the labels are radiofrequency tags that are also enclosed into the capsules. The method was commercialized by IRORI34. A capsule enclosing the resin and the radiofrequency (Rf) tag is demonstrated in Figure 4.18. The capsules are made of polypropylene and named MicroKans at IRORI. Their length and diameter are 18 mm and 7 mm, respectively. They can enclose 25-30 mg resin making possible to produce in them around 25 µmol compound. The Rf tag is a small microelectronic device in glass cover, its length is 13 mm and the diameter is 3 mm. They have a permanent 40 bit code etched into their memory and can receive and emit radiofrequency signals. When placed in radiofrequeny field they re-emit their code.35 The MicroKans are available in 5 different sizes. Their volume varies from 250 µL to 660 µL.

Figure 4.18. Permeable capsule enclosing resin and a radiofrequency tag.

GL-4 LL-5 AL-6

GA-7 LA-8 AA-9

GG-1 LG-2 AG-3 GL-4 LL-5 AL-6 GA-7 LA-8 AA-9

2

3 4 5

6 7

8 9

a

GG-1 LG-2 AG-3 G L A

b c d

GG-1 GL-2 GA-3

LG-1 LL-2 LA-3

AG-1 AL-2 AA-3

G L A

e f g

1

Rf tag

Resin

89

Other kinds of support units have also be introduced. One type of them is the Micro Tube. Figure 4.19. demonstrates a MicroTube that is a plastic tube also containing an Rf tag. The

surface of the plastic tube is covered with a radiolitically grafted and functionalized polystyrene layer. The length of a Micro Tube is 15 mm and the diameter is 6 mm. The capacity is about 30

µmol.

Figure 4.19. Micro Tube A third kind of support unit carries an optical coding system: the "Laser Optical Synthesis Chips". The supports are 1x1 cm polystyrene grafted square plates. The medium carrying the code is a 3x3 mm ceramic plate in the center of the synthesis support (Figure 4.20.). The code is etched into the ceramic support by a CO2 laser in the form of a two dimensional bar code that can be read by a special scanner.36

Figure 4.20. Support units labeled by a two dimensional bar code

Sorting of the units can be done either manually or by using an automatic sorting machine. The sorting process named “Directed Sorting” is guided by the software named “Synthesis Manager” that tracks all the units during the synthesis. In manual sorting the reading of the code and transferring the units into the proper reaction vessel are done manually. Figure 4.21. demonstrates a manual sorting step. During the chemical synthetic step the units are in the reaction vessels A-E. After finishing this step the units are pooled for washing.

After washing each unit is scanned. The Synthesis Manager identifies one of the reaction vessels F-J where the unit has to be delivered. In the reaction vessels F-J the units are coupled with a different building block identified by the Synthesis Manager. After sorting all units, the next chemical step can be started. The Synthesis Manager continues tracking the units even after the synthesis is completed. It determines the place of the units in the cleavage station, too.

Plastic tube covered with grafted polystyrene

Rf tag

90

Cleavage stations are also available at IRORI together with other items that are needed in the synthesis including, for example, a device that makes possible to easily fill the MicroKans with exact quantities of dry resin.

Figure 4.21. Directed Sorting manually The key operation in the synthesis is sorting. Since every unit has to be scanned and delivered separately, the manual sorting process is relatively slow. Only several hundred or a maximum of 1000 compound is usually prepared using this method. Definitely does not make possible to prepare in a single run thousands of compounds. The automatic sorting machine developed at IRORI solves this problem. The principle of the automatic sorting is outlined in Figure 4.22. After each chemical step the capsules are transferred from the reaction vessels into a larger vessel and thoroughly washed. The pooled and washed units are then further transferred into the vibratory bowl (D) of the sorter. Vibration of the bowl then forces the units into a tube (E). At the solenoid gate (B) the antenna (C) reads the code and the computer (A) determines the destination of the unit. The destination is one of the containers (F) that represents a reaction vessel, into which the capsule need to be delivered for the next synthetic step. The delivery is executed by the X-Y movement of the delivery mechanism. After sorting, of course, the capsules collected in each vessel are reacted with a different monomer. The automatic sorter can accommodate up to 10,000 units that can be sorted into a maximum of 48 containers. The sorting speed is 1000 units per hour.

Pool and wash

A B C D E

F G H I J

Scan then deliver into a reaction vessel

91

Figure 4.22. Automatic sorting machine. A: computer, B: solenoid gate, C: antenna, D: vibratory bowl, E: tube, F: containers

The radiofrequency tagging and the visual coding of the support units are used in manual sorting systems developed by other companies, too. The Australian company, Mimotopes, offers two kinds of solid support units shown in Figure 4.23: SynPhase Crowns (a) and SynPhase Lanterns (b). Their surfaces are grafted and functionalized. Both kinds of units can be coded by attaching to them Rf tags (c and d). One end of the Rf tag fits into the holes in the crowns and lanterns and so it can be firmly attached to them. Scanning of the units goes as described at the IRORI method. A color tagging system has also been developed at the company that uses 8 different colors in the form of colored rings. The color system can be applied to both crowns and lanterns. The stems are firmly attached to the crowns and lanterns by inserting them into their holes. These stems hold the code forming rings (Figure 4.23. e and f).

Figure 4.23. Radiofrequency and color coding of crowns and lanterns: a: SynPhase Crown, b: SynPhase Lantern, c: SynPhase Crown with Rf tag, d: SynPhase Lantern

with Rf tag, e: color coded SynPhase Crown, f: color coded SynPhase Lantern

Position of the ring on the stem encodes the reaction step and its color encodes the building block. A list needs to be prepared in advance in which building blocks are assigned to the positions and colors of the rings. The codes of the units are read visually and are distributed manually among the reaction vessels of the next reaction steps according to the data of a list. The 8 colors in 4 reaction steps allow encoding 84=4096 units.

B

C

F A

g

D E

a b c d e f

92

4.5.2. Units without labels. Encoding by position in space Methods developed independently in two laboratories by Smith et al.37 and Furka et al.38 made the use of physical labels on the solid support units unnecessary. In both methods the macroscopic solid support units were stringed and the position of the units on the string encoded the identity of the unit. In the method of Smith et al. in addition to the position on the strings, colors and reaction vessels were also parts of the code and the authors termed their technique Encore for Encoding by Necklace, Color and Reaction vessel. The method of Furka et al. is termed “String Synthesis”. The units are identified by the string number and their position on the strings. 4.5.2.1. The Encore technique The described version the Encore technique39 is suitable for preparing up to 960 compounds using 10, 8 and 12 building blocks in the first, second and third synthetic step. The solid support units are SynPhase Lanters.

Figure 4.24. The Encore technique. a: reaction vessels of the first coupling step, b and d: strings labeled with color rings, c: the

stringing tool, e: reaction vessels of the second coupling step showing the color of the strings, f: one of the 12 reaction vessels of the third coupling step.

96 96 96 96 96 96 96 96 96 96 a

b

12 12 12 12 12 12 12 12

12 numbered reaction vessels

d

e

f

c

93

The Encore technique is demonstrated in Figure 4.24. The 960 lanterns are evenly distributed among 10 reaction vessels (a). The content of each reaction vessels is reacted with one of the 10 building blocks of the first reaction step. At the end of this step the content of all the 96 lanterns placed in a reaction vessel is the same. Before coupling with the second building block the lanterns are stringed on stainless steel or polyethylene stringing tools (c). Each string contains 10 lanterns and a labeling color ring. The positions of the lanterns are counted from the ring. Each lantern of a string comes from a different reaction vessel (b). This way 96 identical strings are formed. Each string contains 10 lanterns and the content of each lantern is different. The strings are labeled by color rings. There are 8 different colors. The 96 strings are distributed into 8 groups of 12 strings labeled with the same color. The figure shows two examples (b and d). The 12 strings having the same color are transferred into the same reaction vessel in order to react in step 2 with the same building block. As a result, each of the 8 reaction vessels contains 12 strings (e). After the second synthetic step the strings are rearranged into 12 numbered reaction vessels (f). One string from each of the 8 reaction vessels is transferred into one numbered vessel. This way each of the 8 strings of one reaction vessel carries a different color. After the third synthetic step the lanterns are transferred into ten 96 well plates for cleaving the formed compounds from the support. One lantern goes into each well of the plate. The content of each well is identified from three recorded data: position of the lantern on the string, color of the string and the number of the third step reaction vessel. The position on the string, the color of the string and the number of the reaction vessel identify the first, second and third building block, respectively. The Encore technology has been commercialized and a number of tools were developed and made available for simplifying the operations. 4.5.2.2. The String Synthesis The String Synthesis introduced by Furka et al.38,40,41, like the Encore technology, also uses stringed macroscopic solid support units and the units are identified by their position occupied on the string. The other aspects, however, are entirely different. First of all only one string is assigned for every building block in the synthesis. Consequently, the content of the strings coming out from a synthetic step must be redistributed into the strings of the next step. The units are not pooled. The redistribution follows the combinatorial distribution rule (see 4.1.1.1.): all products formed in a synthetic step are equally divided among all reaction vessels of the next synthetic step. This means that the units of any string that contain the same product have to be evenly distributed among the strings of the next step. This can be done without pooling the units that would result in loosing the information embodied in the strings. The units are directly transferred from the old (source) strings into the new (destination) ones. This has two important consequences:

(i) If the redistribution follows a predetermined pattern the information stored in the sequences of the units is preserved during the whole synthetic process and the route of every unit can be tracked by computer.

(ii) The units can be transferred in groups (except the last redistribution) that make the process much faster than the one by one sorting.

94

(iii) The computer controlled redistribution offers a possibility for automation of the sorting process.

The inventors of the method considered different shapes for the units, different patterns for their redistribution and different sorting devices.40 In the following pages the use of two kinds of units and two kinds of manual sorting devices are described applying a fast redistribution pattern named Semi-Parallel Sorting.

The support units are Mimotopes SynPhase Crowns and SynPhase Lanterns demonstrated in Figure 4.25 (see also Figure 4.23. a and b). The crowns are used attached to stems (a and c) that makes possible stringing. The stems are available at Mimotopes in different colors. Lanterns can also be used attached to stems (d) but they can also be used in themselves (g) since they have a hole in their center that provides possibility for stringing. The commercial stems are modified. They have a drilled hole to allow the string to be passed, and they are carved to keep the holes parallel and facilitate threading while they are in the sorting device (b). The modified stems can be used repeatedly. An empty stem (e) is used to label the head of the strings and a half stem (f) at the tail of the strings.

Figure 4.25. Solid supports used in String Synthesis. a: stem, b: carved stem, c and d: crown and lantern attached to carved stem, respectively,

e: full stem with scratches to label the head and the number of the string, f: half stem for labeling the tail of the strings, g: lantern used without stem.

Figure 4.26. Stringed crowns and lanterns. Their position is counted from the head of the strings

a b c d e f g

Tail

25 20 1 5 10 15

Head

1 5 10 15

95

Figure 4.26. shows stringed crowns and lanterns with full stem labels at the heads and half stems at the tails. The two ends of the string must be distinguishable. Labeling at least one end of the strings makes possible to unequivocally define the position of the crowns. Position of the units on the string is counted from head to tail.

The number of strings that are formed in a synthetic step depends on the number of building blocks used in that step since every building block needs a different string. Each string is placed into a different reaction vessel for coupling. The strings themselves must be numbered or otherwise labeled. The simplest way to label the strings is to make visible scratches on the stem marking the head (Figure 4.25./e). Using colored stems is also a possibility.

The string itself must be resistant to solvents and other reaction conditions occurring in the synthesis. In preparation of peptide libraries a polyethylene fishing line proved applicable.

The use of the String Synthesis is demonstrated with preparation of a tripeptide library on SynPhase Crowns. Five amino acids are used as building blocks in each coupling step. Consequently 5 strings need to be formed in each step and the number of the expected tripeptides is 5x5x5=125. The number of crowns on one string is 25. After threading the crowns, each string is placed into a reaction vessel carrying the same number as the string (Figure 4.27.). In each reaction vessel the coupling is done with a different amino acid.

Figure 4.27. Strings in reaction vessels showing below them scratched full stems that indicate the number of the strings

Manual devices for redistribution of the units. The strings coming from the reaction vessels after completed couplings are named source strings. Their support units, crowns or lanterns, are then redistributed into the strings of the next reaction step denoted as destination strings.

Figure 4.28. Devices for sorting crowns and lanterns

1 2 3 4 5

Source tray Destination tray

Source tray Destination tray

Crown sorter

Lantern sorter

96

Redistributions can be carried out using very simple devices that can be easily made by a machine shop. Two different devices are constructed for sorting: one for crowns and another one for lanterns. Both devices operate on the same principle and both contain two identical pieces: a source tray and a destination tray (Figure 4.28.). Both pieces of the crown sorter are metal plates with several numbered parallel slots and bent at the two edges. The pieces of the lantern sorter are polymer plates with numbered grooves. Before sorting, the crowns hang in the slots and the lanterns stay in the grooves of the source tray as shown in Figure 4.29.

Figure 4.29. Crowns and lanterns in the sorting device In the sorting process the crowns or lanterns are pushed into the slots and grooves,

respectively, of the destination tray. It is important to note that in this operation the units preserve their positions relative to each other when they are redistributed in groups. Figure 4.30 shows the top view of the crown sorter after the delivery of a group of 5 crowns from the slot 5 of the source tray into the slot 1 of the destination tray.

Figure 4.30. Top view of the crown sorter The slots and groves of the source and destination trays of the sorter are numbered. It is

important to place each source string into the slot of the source tray carrying the same number. The heads and tails of the source strings must be positioned into the slots of the source tray as indicated in the figure otherwise the software (see later) can not be used. After sorting, the units are restrung.

The destination strings must be numbered according to the numbers of the destination slots or grooves and render their heads and tails to the heads and tails of the destination slots or grooves.

Crowns Lanterns

54321

54321

Source tray Destination tray

Head Tail

Head Tail

97

Figure 4.31. Crowns in the slot of the source plate. a: after loading, b: after removing the string

The units are loaded into the sorter in stringed form. Figure 4.31. shows the crowns

loaded into the slots while still attached to the string (a). When the units are in their place in the source plate the string is cut and removed (b).

The units are sorted in a string free form. After sorting they are found in the destination plate. Before the delivery into the reaction vessels they are restrung (Figure 4.32.).

Figure 4.32. Stringing the crowns in the slot of the destination tray

Semi-Parallel Sorting (SPS). In the string synthesis one solid support unit is assigned for

each product. Except the last elongation step, the products form in groups (product groups) and occupy a defined region on the string. The number of units in each product group is the same. The units of each group need to be evenly distributed among the strings of the next step. Except the last distribution step the units are also transferred in groups (delivery groups). In order to get the number of units of the delivery groups the number of units of the product groups is divided by the number of the destination strings. This calculation is done by computer.

Figure 4.33. Semi-Parallel Sorting. Sorting the units of 3 source strings into 3 destination

ones in 5 relative positions of the source and destination trays.

Source trays

Destination trays1 2 3 4 5

a b

Head Tail

98

The simplest way of the distribution would be to first transfer all the units of one source string to the destination strings and then follow with the next source string. The Semi-Parallel Sorting, however, outlined in Figure 4.33. is faster.

In positions 1, 2, 3, 4 and 5 of the figure, one, two, three, two and one slots of the source and destination trays are in alignment (indicated by enhanced lines). From each aligned slot of the source tray one delivery group of units is transferred into the corresponding slot of the destination tray. The deliveries in these positions are repeated until all units are transferred.

A B C D E F G H I Starting Data for Semi-Parallel Sorting SORTING: Delivery in every cycle starts from the highest number (rightmost) source slot into the first (leftmost) destination slot. Run: Ctrl + S The number of monomers and their symbols (A,C,L etc.) must be entered The calculated sequences are reversed, they reflect the coupling positions from left to right Use the red numbers in the sorting process Do not delete blue cells! Enter data only into yellow cells!

Number of slots Crowns in slots

Number of building blocks in coupling steps (maximum number is 52)

Sort Number Source Destin. Source Destin.

Identical Crowns

Crowns to

move CP 1 5 1 5 5 25 25 25 5 CP 2 5 2 5 5 25 25 5 1 CP 3 5 3 0 0 0 0 1 0 CP 4 4 0 0 0 0 0 0 CP 5 5 0 0 0 0 0 0 CP 6 6 0 0 0 0 0 0 CP 7 7 0 0 0 0 0 0 CP 8 8 0 0 0 0 0 0 CP 9 9 0 0 0 0 0 0

CP 10 Maximum number of building blocks in one step 20 Total number of crowns 125 Maximum number of sorter slots 20 Number of coupling Steps: 3 Maximum number of crowns 1,000 Pause (in seconds): M O N O M E R S I N C O U P L I N G S String number 1 2 3 4 5 6 7 8

CP 1 I F L V G CP 2 E F W Y S CP 3 E F W Y S CP 4 CP 5 CP 6 CP 7 CP 8 CP 9

CP 10

Figure 4.34. Datasheet of the Excel Book where the starting data are entered.

99

The software. The software is written in Visual Basic and the data appear in Microsoft Excel sheets. It can handle up to 1000 crowns, up to 20 reagents (building blocks) and up to 9 reaction steps. Figure 4.34. shows the datasheet of the Excel book where the starting data are entered. Among the starting data are the number and symbols of the building blocks (monomers) used in the coupling steps. The symbols are single letter abbreviations. In the case of peptide synthesis the symbols correspond to the respective amino acids. In the Excel sheets the areas of data entrance are yellow. Several data are instantly calculated and appear in the blue regions of the screen.

Figure 4.35. The flow diagram of the synthesis

First coupling First sorting Second coupling Second sorting Third coupling Products

Str 1 Str 2 Str 3 Str 4 Str 5

I F L V G

String 1 String 2 String 3 String 4 String 5

E F W Y S

Str 1 Str 2 Str 3 Str 4 Str 5

E F W Y S

Str 1 Str 2 Str 3 Str 4 Str 5

100

Among the instantly calculated data are the total number of crowns (or lanterns) needed in the synthesis and the number of coupling steps (column B), the number of source and destination slots (or grooves) used in the first and subsequent sorting steps (D and E) and the number of crowns occupying these slots (F and G). The number of crowns in product groups, that is the number of units that contain the same product appear in H. The number of crowns in a delivery group that have to be moved in every sorting cycle from a source to a destination slot can be seen in column I.

The program can be started by pressing together Ctrl and S. The result of calculations appears (depending on the number of redistributions) in sheets Sort #1 through Sort #9. The sheets show a block containing the products present in the crowns of the source slots and, below these, a second block showing the content of the crowns sorted into the destination slots. Positions of the crowns are counted downward from the top. The number of sheets showing the results of couplings and sortings is equal to the number of sortings plus one. The last sheet contains the predicted product distribution on the final strings. The software is free and can be downloaded via the Internet from the following Web site:

http://szerves.chem.elte.hu/furka

by clicking on the title Excel Book appearing on the lower part of the main page. The software is available only for those who have Excel installed in their computer. Experimental example. Synthesis of a library of 125 tripeptides. The synthesis was carried out using 125 Mimotopes SynPhase Crowns (capacity 5.3 µmol each) derivatized with Fmoc-Rink amide linker. The procedure was started with the formation of 5 strings by threading 25 crown units on Berkley Fire Line fishing line. Five Fmoc protected amino acids were used in each coupling position. The flow diagram is demonstrated Figure 4.35. The symbols of amino acids used in the couplings are those found in the Datasheet demonstrated in Figure 4.34. and that are also indicated in Figure 4.35. below the reaction vessels.

Coupling. Couplings were carried out with strings placed in 100 ml flasks. The protecting groups were removed by adding 10 ml 1:1 v/v DMF-piperidine then mixed on an orbital mixer for 30 minutes. After the cleaving the protecting groups the solutions were decanted from the strings then washed with 3x15 ml DMF, 15 ml DCM, 15 ml DMF, 15 ml DCM and 2x15 ml DMF. The deprotection operation was once more repeated then finally washed with 2x15 ml DCM. The strings were dried then 10 mmol Fmoc amino acid, 10 mmol HOBt and 15 mmol DIC was added in 10 ml NMP solution then mixed on orbital mixer for 2 hours. The solution was then decanted and washed with 3x40 ml DMF, 40 ml DCM and 2x40 ml DMF. The above coupling operation was once more repeated. The strings were finally washed with 2x40 ml DCM. The crowns, still on the strings were dried in an oven then submitted to sorting.

Sorting. After entering the starting data into the Datasheet (Figure 4.34), it can be read from the last column, that in the first sorting the crowns need to be moved from each slot in groups of 5. In the second sorting the crowns are moved one by one. The first sorting is demonstrated in Figure 4.36.

The redistribution of the 125 crowns was completed in the nine stages each representing different relative positions of the source and destination trays. In the stages 1, 2, 3, 4, 5, 6, 7, 8 and 9 the number of the transferred groups (of 5 crowns) was 1, 2, 3, 4, 5, 4, 3, 2 and 1, respectively.

101

Figure 4.36. First sorting

The second sorting is demonstrated in Figure 4.37. In this case the delivery groups contained only a single crown, so one crown was transferred from each slot. The redistribution was completed in 5 cycles each containing 9 stages. Figure 4.37. shows only the first cycle. In the 9 stages of the first sorting cycle altogether 25 crowns were delivered into the destination tray. The rest of the crowns were redistributed in additional 4 cycles not shown in the figure.

Cleavage. The crowns were separately placed in numbered test tubes. The string numbers, positions on the strings and the numbers of the test tubes were simultaneously recorded. 1 ml 1:1 v/v piperidine-DMF was added to each test tube and left to stand for 30 minutes. The crowns were then washed with 3x2 ml DMF, 2 ml DCM, 2 ml DMF and 2x2 ml DCM. After adding 1 ml 95% TFA/H2O the tubes were allowed to stand for 30 minutes. The solutions were decanted into vials numbered according to the numbers of the test tubes. The crowns were washed with 1 ml 95% TFA/H2O and the solutions added to the same vials then dried in a rotawap.

Product distribution. The product distribution on the strings during the synthesis predicted by the computer appeared in sheets Sort #1, Sort #2 and Sort #3 of the Excel Book. Some of the predictions concerning the String No. 1 are summarized in Table 4.18. It can be seen that after coupling 1 on String 1 - as expected - all units contained I. After the first redistribution, String 1 contained five products in groups of five crowns. The product distribution in the rest of the strings not shown in the figure was exactly the same.

Stage Source tray Destination tray Stage Source tray Destination tray

54321

54321

54321

54321

54321

54321

54321

54321

6 7 8 9

54321

54321

54321

54321

54321

54321

54321

54321

54321

1

2

3

4

5 54321

102

Figure 4.37. Second sorting. The first cycle After coupling 2, String 1 contained five dipeptides in groups of five crowns. The rest of

the strings (not shown) differed from the first one since different amino acids were coupled into them. After the second sorting, as the table shows, all products in crowns of String 1 were different. Again, the product distribution in the rest of the strings (not shown) were exactly the same. It is typical in String Synthesis that after redistributions the product distribution on all strings is the same.

It is also typical that after couplings the strings are different. After the third – that is the last – coupling not only the strings differ from each other but the content of the crowns within the strings is also different. Positions of the formed tripeptides on the five strings after the third coupling are shown in Table 4.19. Since the redistribution process is directed by computer, the String Synthesis is suitable for automation. Although no automatic machine has yet been constructed, an automatic sorter designed according to Figure 4.38. would be capable to sort very fast tens of thousands of support units placed in vertical source tubes (a) arranged circularly.42 In the sorting process the units would be dropped through computer controlled electronic gates into the destination tubes (b) stepwise rotated. This arrangement of the tubes would make possible to transfer the units simultaneously from all tubes (parallel sorting).

54321

54321

54321

54321

54321

54321

54321

54321

54321

54321

54321

54321

54321

54321

54321

54321

54321

54321

1 2 3 4 5

6 7 8 9

Stage Source tray Destination tray Stage Source tray Destination tray

103

Table 4.18. Content of strings No. 1 (Str.1) after first and second coupling (Cpl. 1 and Cpl.2) and first and second sorting (Sort 1 and Sort 2)

Posi-tion

Cpl. 1 Str. 1

Sort 1 Str. 1

Cpl. 2 Str. 1

Sort 2 Str. 1

1 I I EI EI 2 I I EI FI 3 I I EI WI 4 I I EI YI 5 I I EI SI 6 I F EF EF 7 I F EF FF 8 I F EF WF 9 I F EF YF

10 I F EF SF 11 I L EL EL 12 I L EL FL 13 I L EL WL 14 I L EL YL 15 I L EL SL 16 I V EV EV 17 I V EV FV 18 I V EV WV 19 I V EV YV 20 I V EV SV 21 I G EG EG 22 I G EG FG 23 I G EG WG 24 I G EG YG 25 I G EG SG

Figure 4.38. Parallel sorting

3 1 2

1 2 3

5 4

5 4

6

6

7

7

a b

104

Table 4.19. Position of products on the final strings

Str. 1 Products

Str. 2 Products

Str. 3 Products

Str. 4 Products

Str. 5 Products

EEI FEI WEI YEI SEI EFI FFI WFI YFI SFI EWI FWI WWI YWI SWI EYI FYI WYI YYI SYI ESI FSI WSI YSI SSI EEF FEF WEF YEF SEF EFF FFF WFF YFF SFF EWF FWF WWF YWF SWF EYF FYF WYF YYF SYF ESF FSF WSF YSF SSF EEL FEL WEL YEL SEL EFL FFL WFL YFL SFL EWL FWL WWL YWL SWL EYL FYL WYL YYL SYL ESL FSL WSL YSL SSL EEV FEV WEV YEV SEV EFV FFV WFV YFV SFV EWV FWV WWV YWV SWV EYV FYV WYV YYV SYV ESV FSV WSV YSV SSV EEG FEG WEG YEG SEG EFG FFG WFG YFG SFG EWG FWG WWG YWG SWG EYG FYG WYG YYG SYG ESG FSG WSG YSG SSG

4.5.2.3. String synthesis of cherry picked libraries43 The software developed to guide sorting in String Synthesis can be used only when complete combinatorial libraries are prepared. Very often, however, only selected components of the full libraries are needed. These non-complete combinatorial libraries are often called cherry picked libraries. In order to make possible preparation of such libraries by String Synthesis, modified software has been constructed. Like the software described in the previous section this software is also written in Visual Basic and can be downloaded free of charge via the Internet from the same address: http://szerves.chem.elte.hu/furka by clicking on the title Excel Book 2 appearing in the lower part of the main page. When using the software first of all the sequences of the cherry picked (input) library have to be entered into the computer (e.g. copy the sequences into column A of the Input Sheet).The software then analyses the sequences then generates a virtual library that is in fact a full combinatorial library that contains all the actual members of the input library. This is

105

followed by rearranging the components of the input library according to their order in the virtual library then distributed into the starting source strings. Table 4.20 shows a part of an input library, the generated virtual library and the rearranged input library.

Table 4.20. Sequences in the input, virtual and the rearranged input libraries

Cherry P. Virtual Cherry P. Input Library Rearranged

1 CITW CITW CITW 2 CITA CITA CITA 3 CITF CITF CITF 4 DGPV CITV CITV 5 DGPW CITG CIPW 6 DGPA CIPW CIPF 7 DGPG CIPA CIPV 8 DGPF CIPF CIPG 9 DGRV CIPV CIRW 10 DGRW CIPG CIRF

Since the library to be synthesized is not a complete combinatorial library, the delivery of

the support units from the source strings into the destination ones can not occur in equal groups. The software generates tables that guide the redistribution operations in every phase of the synthetic process. They also provide possibility to check for potential errors of the operator.

The building blocks of the library are coded using both the lower case and the capital letters of the English alphabet (all together 52 symbols). The sequences of these letters encode the compounds to be synthesized. The order of coupling positions - that is the order of the characters in the sequences - go left to right. These are practically inversed peptide sequences.

When entering the input sequences into the computer there are no restrictions concerning the order of library members but no gaps in column A are allowed. Column A can accept a maximum of 15,000 sequences. The order of characters in the input sequences can be reversed (if for example peptide sequences are used) by pushing Ctrl + i (Figure 4.39.)). The maximum number of characters in the sequences, that is the maximum number of building blocks, is 10.

Execution of the sorting program can be started by pushing Ctrl + e (Figure 4.39). The execution time depends on the size of the library and on the speed of the computer. In the execution process the sequences are assigned to support units then the units are grouped into strings. The tables guiding the redistributions are then calculated and displayed. Execution of the program stops at Sheet 13, showing the position of the products on the final strings. Figure 4.40. shows a part of Sheet 13. Rearranging the order of the components of the input library. The order of the components of the input library in column A is usually accidental. For this reason they can not be directly arranged into strings that are submitted to coupling with the same building blocks.

106

A B C D E F G H I J L M N O 1 CITW I N P U T S H E E T 2 CITA Copy the library sequences into column A 3 CITF Execute sorting: Ctr+e Clear input column A: Ctr+j 4 CITV Invert sequence: Ctr+i Save original library: Ctr+b Sort the virtual library: Ctr+s

5 CIPW Number of compounds 206 Number of monomers in sequences 4 6 CIPF M O N O M E R S (BLOCKS) O F T H E L I B R A R Y 7 CIPV Num 1 2 3 4 5 6 7 8 9 10 11 8 CIPG CP1 5 C D E F A 9 CIRW 49 46 25 32 54

10 CIRF CP2 3 I G H 11 CIRV 61 76 69 12 CISW CP3 4 T P R S 13 CISA 50 45 62 49 14 CISV CP4 5 W A F V G 15 CISG 52 33 48 46 27

Figure 4.39. Part of the Input Sheet.

The strings that need to be formed usually do not even contain the same number of units. As a consequence, the order of the components of the input library has to be rearranged into a form that allows regular redistributions.

PRODUCTS ON THE STRINGS Unit

number Str. 1

Str. 2

Str. 3

Str. 4

Str. 5

1 AHSW AHSA AHSF AHSV AHSG 2 AHRW AHRA AHRF AHRV AHRG 3 AHPW AHTA AHPF AHPV AGSG 4 AHTW AGSA AHTF AHTV AGRG 5 AGSW AGRA AGSF AGSV AGPG 6 AGRW AGPA AGRF AGRV AISG 7 AGPW AGTA AGPF AGPV AIRG 8 AGTW AISA AGTF AGTV AIPG 9 AISW AIRA AISF AISV FHRG 10 AIRW AITA AIRF AIRV FGPG

Figure 4.40. Products on the strings

A partial tetrapeptide library is used to illustrate the operations executed by the program. By analyzing the input library the program first determines the crucial starting data and displays

107

them in the Input Sheet (Figure 4.39.): (i) The number of input sequences (F5) (ii) The length of the sequences (N5) (iii) The number of amino acids used in the different coupling positions (CP1 to CP4) (iv) Codes of the amino acids (section Monomers of the library). (v) The number of units into which a particular amino acid has to be coupled

(displayed below the code of the amino acid).

Based on the above data, a full (virtual) peptide library is generated in which all components of the input library are present. The number of components in the virtual library is limited to 30,000. The sequences of the input library are then arranged into the order they appear in the virtual library. Sheet 13 shows the sequences of the virtual library and those of the input cherry picked and the rearranged cherry picked libraries. The first 10 sequences of these libraries are reproduced in Table 4.20. The components of the original cherry picked library in column A of the Input Sheet is replaced by the rearranged library. All further manipulations are based on this rearranged library: first the sequences are assigned to support units and then the units are distributed into the starting strings. The occupancy of the starting strings appears in Sheet 3. The same sheet, in its lower part, contains the guiding tables for the first redistribution. In the experimental realization of the synthesis the starting strings need to be formed manually by placing the indicated number of support units on the strings then submit them to the first coupling step. The symbols of the amino acids that need to be coupled into strings appear in red. Those of the other amino acids in the sequences are black. The sets of destination strings that are formed in redistribution steps occupy one of the Sheets 4 to 12. The products appear in Sheet 13.

Third coupling with monomers: T P R S

N u m b e r of u n i t s o n t h e s t r i n g s 50 45 62 49

Third sorting Str 1 Str 2 Str 3 Str 4

CITV CIPG CIRV CISG 1 CITF CIPV CIRF CISV 2 CITA CIPF CIRW CISA 3 CITW CIPW CGRG CISW 4 CGTV CGPG CGRV CGSG 5 CGTF CGPV CGRF CGSV 6 CGTA CGPF CGRA CGSF 7 CGTW CGPW CGRW CGSA 8 CHTV CHPV CHRG CGSW 9 CHTF CHPF CHRV CHSG 10

Figure 4.41. Sequences on the strings undergoing the third coupling step

108

As an example, Figure 4.41. shows a part of Sheet 5. This sheet demonstrates the 4 strings that undergo the third coupling. The figure includes the first 10 tetrapeptide sequences from each string. The codes of amino acids that need to be coupled with the respected strings appear at the top and the number of units in each string are found below them. The numbers in the last column show the position of the units on the strings. The sequences of the strings are printed in three different colors: The codes of amino acids already coupled into the units in the previous coupling steps are blue. The amino acids of the actual coupling steps appear in red and the codes the amino acids that need to be coupled into the units in the forthcoming coupling steps are black.

The products of the synthesis and their positions on the strings appear in Sheet 13 (Figure 4.40). The sequences of the products are also shown in reversed form in the lower part of the sheet (not shown in the table). This makes possible to read the orders of the building blocks as peptide sequences, too.

Guiding tables for redistribution experiments. As already mentioned, in the case of the cherry picked libraries some components that are present in a full (or virtual) library are missing.:

No. of units to deliver from source

Cycle/stop position

Units in source troughs

Units in destination troughs

1 2 3 1 2 3 1 2 3 4 4 5 / 6 206 0 0 0 50 45 62 49 3 5 5 / 5 4 0 0 50 45 62 45 4 5 5 5 / 4 7 5 0 50 45 59 40 4 4 4 5 / 3 11 10 5 50 41 54 35 4 3 5 / 2 15 14 9 46 37 50 35 4 5 / 1 15 18 12 42 34 50 35 2 4 / 6 15 18 16 38 34 50 35 5 5 4 / 5 17 18 16 38 34 50 33 2 5 3 4 / 4 22 23 16 38 34 45 28 3 5 5 4 / 3 24 28 19 38 32 40 25 4 3 4 / 2 27 33 24 35 27 35 25 4 4 / 1 27 37 27 31 24 35 25 3 / 6 27 37 31 27 24 35 25 3 3 / 5 27 37 31 27 24 35 25 2 3 / 4 30 37 31 27 24 32 25 2 4 4 3 / 3 30 39 31 27 24 30 25 4 3 3 / 2 32 43 35 25 20 26 25 3 3 / 1 32 47 38 21 17 26 25

Figure 4.42. Part of an Experiment guiding table

Data for guiding redistribution Data for checking potential errors in sorting

109

This has two consequences

(i) The numbers of support units belonging to the strings usually differ. This is clearly seen, for example, in Figure 4.41 (number of units on the strings).

(ii) (ii) The number of units within the groups of identical products may also differ. For these reasons the transfers in redistributions can not be realized in equal groups. Even empty groups may occur. As a consequence, in order to be able to execute the redistributions the number of units of every delivered group has to be calculated and displayed by the computer. The “Experiment guiding tables” are presented in the lower parts of Sheets 3 to 12. A part of the table found in Sheet 4 is demonstrated in Figure 4.42. The table guides the redistribution after the second coupling. This is the second sorting process when the units of three source strings are redistributed into four destination strings. The guiding data are found in the columns below the title: Data for guiding redistribution. The third sorting is realized in 5 cycles. The figure shows only those guiding numbers that belong to cycles 4 and 5. In each cycle the deliveries occur at 6 different relative positions (stops) of the two trays of the manual sorter. The 6 relative tray positions of the six stops of a cycle are demonstrated in Figure 4.43. Taking cycle 5 as example, the cycle/stop positions change from 5/1 to 5/6. The same relative tray positions are repeated in all cycles. The support units (crowns or lanterns) are delivered from the slots/troughs of the upper source tray into those of the lower destination tray. The slots/troughs of the trays appear as vertical lines. The enhanced lines represent slots/troughs from which and to which the units are delivered in a particular stop position.

Figure 4.43. The 6 relative tray positions in the 6 stop positions of cycle 5

Figure 4.42. shows one column for each of the three source strings from which the units need to be transferred. The cycle/stop numbers are found in the fourth column. The start position (stop position 1) is at the bottom in all cycles. The numbers of support units that have to be transferred at a stop position from the source slots/troughs into the destination ones are found in the columns of the strings in the same row where the cycle/stop numbers are found. For example in the stop position 2 of cycle 5 (5/2) 4 and 3 units are transferred from the slots/troughs 2 and 3,

5 / 1 5 / 2 5 / 3 5 / 4 5 / 5 5 / 6 Cycle/stop positions

1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7

1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7

110

respectively. These transfers are made from slots/troughs 2 and 3 into the destination slots/troughs 1 and 2, respectively (see Figure 4.43). The program also provides possibility to check the accuracy of the redistribution in every phase of execution and discover a potential error made by the operator. This is made possible by displaying the number of units that have to remain in the source slots and appear in the destination slots after successful transfers. These numbers are found in Figure 4.42. below the title: Data for checking potential errors in sorting. These numbers are also found in the same row where the cycle/stop positions are. After the mentioned transfers in the stop position 5/2, for example, 14 and 9 units remain in the source slots/troughs 2 and 3, respectively and 46 and 37 units appear in the destination slots/troughs 1 and 2, respectively. The software developed to guide sorting in the synthesis of cherry picked libraries also provides a possibility to automate the redistribution process. 4.6. Examples 4.6.1. Split-Mix Synthesis of an encoded benzimidazole library.44 The library synthesized at Affymax (an American company) had three diversity positions using 36 building blocks in each position. The structure of the components can be described by the following general formula.

N

N

HN

R1

R2

R3

O Since 36 building blocks were used in three positions the number of components was 36x36x36=46,656. R1 and R2 were built into the structure by using amines as building blocks; their structure is demonstrated in Figure 4.44.

R

NH2

n

R'

NH2

n

R N

NH2

nNH2

n

O,S,N

NH2

nNH2

H3C

n

Figure 4.44. Structure of amines used in the synthesis of the benzimidazole library. Total number: 71

R3 was introduced by aldehide building blocks. Their structures are represented in Figure 4.45.

111

CHOR CHOR'

R N

CHO

O,S,N

CHO H3C

NH2

n

Figure 4.45. The aldehide building blocks. Total number: 35 The beads were encoded by a special binary type encoding developed at Affymax. The encoding tags were secondary amines that were built in using the Alloc (allyloxycarbonyl) protected monomers shown below. R and R’ are various length alkyl chains.

R'N

NOH

R

OO O

O

Encoding tags were used only at the first and second diversity positions. At the end of the synthesis the samples were not mixed so encoding at this stage was unnecessary. When a peptide library is prepared amino acids are used as building blocks. Their reactivity is well known as well as the optimal reaction conditions. This is not the case when a non-peptide library is synthesized. The reactivity of all building blocks has to be carefully checked and the reaction conditions also need to be optimized. Some otherwise favorite building blocks have to be excluded because their poor reactivity. It is not uncommon to spend much more time with the pre-synthetic studies than with the synthesis itself. In the case of the benzimidazole library synthesis the reaction conditions were also carefully optimized and the selected building blocks showed good reactivity. The synthesis was carried out using Tentagel HL NH2 resin as solid support. The first step was conversion of a part of amino groups of each bead to be suitable to attach to them the coding tags and convert the remaining amino groups for the acceptance of the first building block of the product. For this reason a part of the amino groups were protected by Fmoc groups and the rest was blocked by Boc protecting groups (Figure 4.46).

NH2

NH2

HN

NH

Fmoc

Boc

NH2

NH

Boc

Figure 4.46. Introduction of Fmoc and Boc protecting groups

This reaction was carried out with 72 g resin (27.4 mmol amine) in DCM in presence of

112

DIEA. The reagent was a mixture of 22.7 g ( 104 mmol) Boc2O and 0.79 g (3.1 mmol) Fmoc-Cl. The resin was finally treated with piperidine that removed the Fmoc protecting groups and made available a part (about 1/9 part) of the amino groups for attachment of the first coding tags.

Table 4.21. Encoding mixtures for the first and second diversity position

RV Code1 Code2 RV Code1 Code2 RV Code1 Code2 RV Code1 Code2 1 A U 10 AE UY 19 DE XY 28 ACF UWZ 2 B V 11 AF UZ 20 DF XZ 29 ADE UXY 3 C W 12 BC VW 21 EF YZ 30 ADF UXZ 4 D X 13 BD VX 22 ABC UVW 31 AEF UYZ 5 E Y 14 BE VY 23 ABD UWX 32 BCD VWX 6 F Z 15 BF VZ 24 ABE UVY 33 BCE VWY 7 AB UV 16 CD WX 25 ABF UVZ 34 BCF VWZ 8 AC UW 17 CE WY 26 ACD UWX 35 BDE VXY 9 AD UX 18 CF WZ 27 ACE UWY 36 BDF VXZ

Encoding. Six different tags were used to encode the 36 resin samples at coupling No. 1 (Code1), and another six ones at coupling No. 2 (Code2). Their stock solutions were labeled A, B, C, D, E, F and U, V, W, X, Y, Z for Code 1 and Code 2, respectively. Attachment of the Code1 tags for the R1s (Figure 4.47.). The resin was divided into 36 portions and place into reaction vessels RV1 to RV36. Samples of A, B, C, D, E, F solutions were added to RV1 to RV6. To the rest of the reaction vessels (RV7 to RV36) mixtures were added according to Table 4.21. The coupling reagents for the acylations were DIC and HOBt.

HN

NH

Tag1-Alloc

Boc

NH2

NH

Boc

A

Figure 4.47. Encoding at the first diversity position

Coupling the linkers to the resin. Two linkers (L1 and L2) were used in the synthesis. Structure of both are seen in Figure 4.47. L1 is an acid-labile linker from which the product can be cleaved off as an unsubstituted amide. In other words R1 in the product is H. The other linker (L2) makes possible to attach to the resin the primary amines of Figure 4.44. using reductive amination then cleave the product as substituted amides (R1≠H).

113

O

O OOH

O

HNFmoc

O

O O

O

OHO

Figure 4.48. The linkers First the Boc protecting groups were removed in all the 36 reaction vessels with a solution of 50% TFA (Figure 4.49/A). L1 was coupled only to RV1 DIC and HOBt. L2 was coupled to RV2 to RV36 also by DIC HOBt (Figure 4.49/B).

HN

NH

Tag1-Alloc

Boc

A HN

NH2

Tag1-Alloc

BHN

NH

Tag1-Alloc

Linker

Figure 4.49. Coupling with the linkers. A: removal of the Boc protecting group, B: coupling with the linker

Reductive amination with R1 amines (Figure 4.50). The R1 groups were built into the products by submitting the content of RV2 to RV36 to reductive amination with amines selected from those in Figure 4.44. RV1 was left unchanged since the L1 linker itself holds the amino group in Fmoc protected form. The reductive amination was carried out by adding solutions of the 35 amines and NaCNBH3 to RVs 2 to 36 and keeping the solution at 50o for 12 hrs.

HN

NH

Tag1-Alloc

Linker

HN

NH

Tag1-Alloc

Linker NH

R1

Figure 4.50. Introduction of the R1 groups by reductive amination

L1 L2

114

Building in a scaffold. The next synthetic step was the attachment of a substituted benzene scaffold by acylating the amine nitrogen with 4-fluoro-3-nitrobenzoic acid (Figure 4.51).

HN

NH

Tag1-Alloc

Linker NH

R1

HN

NH

Tag1-Alloc

Linker N

R1

O

F

NO2

Figure 4.51. Acylation with 4-fluoro-3-nitrobenzoic acid Since the amino group in the L1 linker was protected by Fmoc group the content of RV1 was treated with piperidine to remove the protecting group. Then the couplings were carried out in all the 36 reaction vessels by adding solutions of 4-fluoro-3-nitrobenzoic acid and DIC. After the acylation the 36 resin samples were combined in solvent then mixed with mechanical stirring and nitrogen bubbling. After washing the resin was dried then divided into 36 equal portions. Nucleophilic displacement of fluorine by R2 amines Figure (4.52). One of the 36 amines (in 24x molar excess), solvent and DIEA were added to each of the reaction vessels, kept at 50o for 12 hrs then washed.

HN

NH

Tag1-Tag2-Alloc

Linker NR1

ONH

NO2

R2

HN

NH

Tag1-Tag2-Alloc

Linker NR1

ONH

NH2

R2

Figure 4.52. Displacement of fluorine by R2 amines Attachment of the second set of tags for encoding the R2 amines (Figure 4.53). For encoding in the second diversity position a different set of encoding N-Alloc-Tag monomers were used. Their labeled U, V, W, X, Y and Z. They were used individually and as mixtures according Table 4.21. First the Alloc protecting groups were removed from the R1 encoding tags. To each reaction vessel a solution of 1 M TBAF and TMSN3 was added followed by addition of a solution of Pd(PPh3)4. After rapid mixing, the solution was left to stand at room temperature. The liberated secondary amines of Tag1-s were acylated in the presence of DIAE and HATU. After washing the 36 resin samples were combined. Before combining, however, usually 5 beads of each sample were decoded to ensure the fidelity of coding.

115

HN

NH

Tag1-Alloc

Linker N

R1

O

NH

NO2

R2

HN

NH

Tag1-Tag2-Alloc

Linker N

R1

O

NH

NO2

R2

HN

NH

Tag1

Linker N

R1

O

NH

NO2

R2

A B

Figure 4.53. Second encoding. A: removal of Alloc protecting group. B: coupling the encoding tags for the second diversity

position Reduction of the nitro group and benzimidazole formation with the R3 aldehides. The combined samples were thoroughly mixed, submitted to reduction with SnCl2 then divided again into 36 equal samples. To each of 35 samples a different aldehide was added in 15 fold molar excess to facilitate the ring formation. One sample was reacted with trimethylorthoformate to make R3=H in the product. The reaction mixtures were heated at 50o for 12 hrs. The samples were finally washed and dried in vacuo without mixing them.

HN

NH

Tag1-Tag2-Alloc

LinkerNR1

ONH

NH2

R2

HN

NH

Tag1-Tag2-Alloc

LinkerNR1

ONH

NO2

R2

HN

NH

Tag1-Tag2-Alloc

Linker NR1

ON

N

R2

R3

A B

Product

Figure 4.54. Reduction and ring formation. A: reduction, B: rection with the aldehides of Figure 4.45. and ring formation

Before screening the products were cleaved from individual beads. The products were used in the screening process and their identity could be determined by decoding the remaining beads. The beads were treated with 6M HCl in a glass tube. The released secondary amines were dansylated and identified by HPLC.45,46 4.6.2. Synthesis of a 10,000 member piperazine 2-carboxamide library by Directed Sorting47 Application of the radiofrequency encoding method and automatic sorting described in chapter 4.5.1. is exemplified by the synthesis of a large organic library containing 10,000 discrete components represented by the following formula.

116

N

NHN

O R3

R1

R2 Representative members of the arrays of the building blocks used in the three diversity positions are found in Figure 4.55.

H2N NO

H2NH2N

H2H O

OH2N NN H2N NHBoc

Cl O

O

Cl O

O

HO

O

N FFF

CONCO HO

ONHBoc

NHO

O

HO

O

NNH HO N

OSOO

ClHO O

O O

NHBoc

Cl O

O

HO

O

OHO

O

OH

O OHSOO

ClO OCl

OSO O

Cl

CF3

CF3

N

N

N

C

CO

O

HO NH

O

O O

R1

R2

R2

R3

R3HO

O

NNH

BocHN

Figure 4.55. Building blocks used in the synthesis of the piperazine 2-carboxamide library The scaffold was built in by coupling with the orthogonally protected piperazine-2-carboxylic acid.

N

NHN

O R3

R1

R2 The steps of the synthesis can be followed in Figure 4.56. The procedure was started with 10,000 MicroKans filled with resin and also containing the RF-tag. The linker was already attached to the resin. After splitting the MicroKans into portions directed by the Synthesis Manager, the first combinatorial step (Figure 4.56/A) was attachment of the primary amines (R1) by reductive amination in the presence of NaBH(OAc)3. The MicroKans were pooled then the previously formed secondary amines were acylated in the presence of HBTU and DIEA with the protected piperazine-2-carboxylic acid (Figure 4.56/B). After attachment of the scaffold the Fmoc groups were removed by piperidine (Figure 4.56/C) then MicroKans were sorted again.

117

Sorting was followed by the second combinatorial step (Figure 4.56/D) in which the deprotected nitrogen of the ring was reacted with the R2 building blocks, sulfonyl chlorides, isocyanates, chloroformates and carboxylic acids using properly selected reagents and solvents. The MicroKans were pooled again and the Alloc protecting groups were removed (Figure 4.56/E) with Pd(Ph3)4. After sorting the third combinatorial (Figure 4.56/F) step was executed using the R3 building blocks to functionalize the second amino group of the ring. Reaction conditions were similar to the second combinatorial step. The products were removed from the resin (Figure 4.56/G) by treating the MicroKans with 50% TFA-DCM.

O

O

OO

AO

O

NHO

R1

O

O

O

CB

N

R1

O

N

N

Alloc

Fmoc

O

O

ON

R1

O

N

N

Alloc

R2

O

O

ON

R1

O

N

N

R3

R2

N

NHN

O R3

R1

R2

O

O

ON

R1

O

NH

N

AllocD

E

O

O

ON

R1

O

N

HN

R2

F G

Figure 4.56. Scheme of the synthesis of the piperazine 2-carboxamide library 4.6.3. Synthesis of two libraries on one support A synthesis of a very interesting library has been outlined in a patent application of Geysen.48 According to the patent application, the library is prepared on a resin that has a built in arm with two branches (Figure 4.57. 1 and 2).

Figure 4.57. Synthesis of two libraries on one support

1

2 Protecting group

2

A1-10B1-10C1-10 1

2 Protecting group

A1-10B1-10C1-10 1

3 steps

3 steps

2

A1-10B1-10C1-10 1

X1-10Y1-10Z1-10

118

Both branches have appropriate functional group for attachment of building blocks. The functional group on branch 1 is free that on branch 2 is protected. A three step split-mix synthesis is executed using 10 building blocks in each step (building blocks A1-10, B1-10 and C1-10 in steps 1, 2 and 3, respectively). Thus 10x10x10=1,000 different trimers are formed on branch 1. After mixing in the final combinatorial step the protecting group is removed from branch 2 then a second three step split-mix synthesis is executed using again 10 blocks (X1-10, Y1-10 and Z1-10) in each step. As a result a 10x10x10=1,000 component library forms on branch 2 of the support. In the final product to branches 1 of the beads one component of the A, B, C library is attached. Similarly, branches 2 hold one component of the library X, Y, Z. In both split-mix processes a single compound forms in each bead. As a consequence – if the beads are present in large excess – all possible combinations of pairs of components are formed. Since both libraries have 1,000 components the total number of different pairs is 1,000x1,000=1,000,000. Of course to prepare such a library at least of 10,000,000 beads are needed. Such a library can be used for different purposes. If the two molecules are located at appropriate distances (depending on the length of the branches) the potential interaction of the two molecules can be studied. If one of the two molecules is a catalyst then the effect of the catalyst on the other molecule can be tested. If both molecules are catalysts the experiments may show which combination of the two catalysts is most effective on added substrate. References

1. Á. Furka, F. Sebestyén, M. Asgedom, G. Dibó, In Highlights of Modern Biochemistry, Proceedings of the 14th International Congress of Biochemistry, VSP. Utrecht, The Netherlands, 1988, Vol. 5, p 47.

2. Á. Furka, F. Sebestyén, M. Asgedom, G. Dibó Proceedings of the 10th International Symposium of Medicinal Chemistry, Budapest, Hungary, 1988, p 288, Abstract P-168.

3. Á. Furka, F. Sebestyén, M. Asgedom, G. Dibó Int. J. Peptide Protein Res. 1991, 37, 487. 4. Peptide sequencer is an instrument in which the amino acids are cleaved stepwise from

the peptides starting at the N-terminus and the removed amino acids are identified. 5. R. Süβmuth, A.Trautwein, T. Richter, G. Nicholson, G. Jung In G. Jung (Ed)

Combinatorial Chemistry 1999, Wiley-VCH, Weinheim, 499. 6. S. Brenner and R. A. Lerner Proc. Natl. Acad. Sci. USA 1992, 89, 5381. 7. M. C. Needels, D. G. Jones, E. H.Tate, G. L. Heinkel, L. M. Kochersperger, W. J. Dower,

R. W. Barett, M. A. Gallop Proc. Natl. Acad. Sci. USA 1993, 90, 10700. 8. J. Nielsen, S. Brenner, K. D. Janda J. Am. Chem. Soc. 1993, 115, 9812. 9. V. Nikolaiev, A. Stierandova, V. Krchnak, B. Seligman, K. S. Lam, S. E. Salmon, M.

Lebl Pept. Res. 1993, 6, 161. 10. J. M. Kerr, S. C. Banville, R. N. Zuckermann J. Am. Chem. Soc. 1993, 115, 2529. 11. M. H. J. Ohlmeyer, R. N. Swanson, L. W. Dillard, J. C. Reader, G. Asouline, R.

Kobayashi, M. Wigler, W. C. Still Proc. Natl. Acad. Sci. USA 1993, 90, 10922. 12. 12.Á. Furka, F. Sebestyén, J. Gulyás In Proc. 2nd Int. Conf. Biochem. Separations,

Keszthely, Hungary, 1988, 35. 13. Á. Furka Drug Development Research 1994, 33, 90. 14. S. P. A. Fodor, J. L. Read, M. C. Pirrung, L. Stryer, A. Tsai Lu, D. Solas Science, 1991,

251, 767. 15. F. Sebestyén, G. Dibó, A. Kovács, A. Furka Bioorg. & Med. Chem. Letters 1993, 3, 413.

119

16. H. M. Geysen, S. J. Rodda, T. J. Mason Mol. Immunol. 1986, 23, 709. 17. W. J. Rutter, D. V. Santi U.S. Pat. 5,010,175 (1991). 18. J. M. Ostresh, J. M. Winkle, V. T. Hamashin and R. A. Houghten Biopolymers 1994, 34,

1681. 19. T. Carell, E. A. Winter and J. Rebek, Jr., Angew. Chem. Int. Ed. Eng. 1994, 33, 2061. 20. M. M. Shemyakin, Yu. A. Ovchinnikov, A. A. Kiryushkin, I. V. Kozhevnikova

Tetrahedron Letters 1965, 2323. 21. F . Cramer, R . Helbig, H . Hettler, K. H. Scheit, H .Seliger Angew Chem Int Ed 1966, 5,

601. 22. H . Hayatsu, H.G. Khorana J Am Chem Soc 1966, 88:3182–3183. 23. H. Han, M. M. Wolfe, S. Brenner, K. D. Janda Proc Natl Acad Sci USA 1995, 92:6419. 24. D. J. Gravert, K. D. Janda Trends Biotechnol 1996, 14, 110. 25. S. P. A. Fodor, J. L. Read, M. C. Pirrung, L. Stryer, A. T. Lu and D. Solas Science 1991,

251, 767. 26. S. P. A. Fodor Laboratory Automation News 1997, 2, 50. 27. http://www.affymetrix.com 28. J. K. Scott and G. P. Smith Science 1990, 249, 404. 29. S. Cwirla, E. A. Peters, R. W. Barrett and W. J. Dower Proc. Natl. Acad. Sci. USA 1990,

87, 6378. 30. J. J. Devlin, L. C. Panganiban and P. E. Devlin Science 1990, 249, 404. 31. G. P. Smith and V. A. Petrenko Chem. Rev. 1997, 97, 391. 32. E. J. Moran, S. Sarshar, J. F. Cargill, M. Shahbaz, A Lio, A. M. M. Mjalli, R. W.

Armstrong J. Am. Chem. Soc. 1995, 117, 10787. 33. K. C. Nicolaou, X –Y. Xiao, Z. Parandoosh, A. Senyei, M. P. Nova Angew. Chem. Int.

Ed. Engl. 1995, 36, 2289. 34. http://www.irori.com 35. Xiao-Yi Xiao, K. C. Nicolaou In H. Fenniri (Ed) Combinatorial Chemistry 2000, Oxford

University Press, Oxford, 75. 36. X.-Y. Xiao, C. Zhao, H. Potash, M. P. Nova Angew. Chem. Int. Ed. Engl., 1997, 36, 780. 37. J. Smith, J. Gard, W. Cummings, A. Kaniszai, V. Krchňák J. Comb. Chem. 1999, 2, 368. 38. Á. Furka, J. W. Christensen, E. Healy, H. R. Tanner, H. Saneii J. Comb. Chem. 2000, 2,

220. 39. V. Krchňák, V. Paděra In G. A. Morales and B. A. Bunin (Eds) Methods in Enzymology,

Combinatorial Chemistry Elsevier Academic Press, 2003, 369, 112. 40. Á. Furka, Comb. Chem. & High Throughput Screening 2000, 3, 197. 41. Á. Furka, J. W. Christensen, E. Healy In G. A. Morales and B. A. Bunin (Eds) Methods in

Enzymology, Combinatorial Chemistry Elsevier Academic Press, 2003, 369, 99. 42. Á. Furka US Patent 7/16/2002. 43. Furka, G. Dibó, N. Gombosuren Drug Discovery Technologies 2005, 2, 23. 44. D. Tumelty, L-C. Dong, K. Cao, L. Lee, M. C. Needels In I. Sucholeiki (Ed) High

Throughput Synthesis, Principles and Practices, Marcel Decker Inc. 2000, 93. 45. D. Maclean, J. R. Schullek, M. M. Murphy, Zhi-Jie Ni, E. M. Gordon, M. A. Gallop Proc.

Natl. Acad. Sci. 1997, 94, 2805. 46. Z. J. Ni, D. Maclean, C. P. Holmes, B. Ruhland, M. M. Murphy, J. W. Jacobs, E. M.

Gordon, M. A. Gallop J. Med. Chem. 1996, 39, 1601–1608. 47. F. Herpin, G. C. Morton In G. A. Morales and B. A. Bunin (Eds) Methods in Enzymology,

Combinatorial Chemistry Elsevier Academic Press, 2003, 369, 75.

120

48. H. M. Geysen WO 01/40148 A2. 49. B. Cohen, S. Skiena J. Comb. Chem., 2000, 2, 10.

121

5. Screening methods Compound arrays of individual compounds, combinatorial compound libraries as well as arrays of new materials are prepared in order to find among their components new pharmaceuticals, new insecticides, new fungicides, new plastics, new semiconductors etc. The new useful compounds or materials can be found by examining the libraries looking for components having pre-determined properties. This process is called screening. In order to be able to do screening we need assays that unequivocally show the presence or absence of components having the desired property. The development of the assay methods itself is an area of intensive research, dealing with this subject, however, is not within the scope of this book. The results of the assays often appear as changes in color, fluorescence, radioactivity, conductivity etc. Life on earth is largely dependent on pairs of molecules that perfectly fit together like enzymes and substrates, antigens and antibodies, hormones and receptors (Figure 5.1). Detection of binding of a component of a synthesized library (red in the figure) to a large target molecule (green) is an often applied screening procedure. The binding can be detected by changing a color appearance of fluorescence or radioactivity etc.

Figure 5.1. Fitting and non-fitting

There is an important difference between screening arrays of compounds produced by parallel synthetic methods and the combinatorial libraries prepared by the split-mix procedure. Components of the compound arrays are known compounds that are individually tested in parallel format each component occupying a different test tube (Figure 5.2). In combinatorial libraries also individual compound form but their identity is not known. So in the screening process the identity of the active compound also have to be determined. There is another difference: the whole library can be tested in a single experiment. It is the target molecule itself that selects from the mixture the active component that is fitting to it (marked by an arrow in Figure 5.3).

122

Figure 5.2. Parallel screening of components of compound arrays

Figure 5.3. Screening a combinatorial library 5.1. High throughput screening of arrays of individual compounds The introducing by Takátsy1 of the parallel approach in his microbiological experiments was a very important development in the history of the analytical methods. The use of his microtiter plates made possible to carry out the analytical assays in parallel format and exploit its advantages in improving the efficiency. The later huge increases in the speed of the experiments and the appearance of the high throughput screening (HTS) methods are also based on the

123

application of his microtiter plates. Improvements in the sensitivity of the assay methods made possible to further increase productivity by replacing the 96-well plates with 384 and even 1,536-well ones and by applying automation. Appearance automatic high performance work stations made possible screening of well over hundred thousand compounds per day. The SAGIAN™ Core Systems (Figure 5.4.) is one example of a standardized integrated system. Both liquid handling and reading of the results of assays are fully automated. Figure 5.5 shows a plate reader.

Figure 5.4. A high performance automatic work station. The SAGIAN™ Core Systems

(Photo: www.beckman-coulter.com)

The above core systems can be integrated with devices produced by other companies. Figure 5.5. shows the Analyst® GT of Molecular Devices Corp., a microplate reader optimized for HTS, integrated into Sagian Core Systems.

Figure 5.5. Analyst® GT integrated into Sagian Core Systems (Photo: www.moleculardevices.com)

124

5.2. Screening of combinatorial libraries. Deconvolution methods The methods applied in screening of combinatorial libraries substantially differ from those applied when dealing with libraries prepared by parallel synthesis. Parallel synthesis produces arrays of individual compounds so their components – as described in the previous paragraph – can be examined either individually or as individual components of arrays submitted to high throughput screening processes. Combinatorial libraries, on the other hand, are mixtures of a large number of compounds. When the concept of combinatorial synthesis was born, finding a single useful component in a mixture of millions of compounds seemed an unrealizable task. As it later turned out the problem could be solved by different experimental strategies that apply a logically devised series of operations in order to identify the wanted components of the mixtures. The first such strategy was described by the author in 1982 (see paragraph 1.2.). The strategies by which the useful components of multi-component mixtures can be identified are called deconvolution methods. The combinatorial libraries are prepared in two different forms. In one form the synthesized libraries are cleaved from the support. In this case the components form real mixtures that are examined in solution. The combinatorial libraries can also be prepared in tethered form. The libraries are not cleaved from the support. The components remain on the beads of the resin and can be examined as individual compounds. The deconvolution methods applicable for tethered libraries substantially differ from those developed for dissolved libraries. 5.2.1. Deconvolution methods for dissolved libraries The libraries cleaved from the support then dissolved may contain thousands or even millions of compounds. The deconvolution methods developed for dissolved libraries makes possible to unequivocally identify the library component that is responsible for a biological property. Of course, these deconvolution methods work only if an appropriate biological or biochemical assay method is at hand for detection of the biologically active component in the mixture. 5.2.1.1. The iteration method The iteration method is a strategy that was developed for screening combinatorial libraries cleaved from the support. These libraries are real mixtures and are screened in solution. The strategy was first conceived by the author in 19822,3 and the same concept was published by Geysen at al. in 19844, by Houghten et al. in 19915, and Janda in 19946. The task of identifying a compound having a certain property in a mixture of millions of other compounds can be compared to that the police face when have to identify a criminal among millions of individuals that – at least in principle – can be considered as potential suspects. Using the data supplied by an eyewitness, for example, the police can gradually reduce the list of suspects to the one who committed the crime. On the basis of a single data, for example, that the criminal is a man, the list of suspects can be reduced by 50%. This kind of approach is demonstrated by a simple example. Let’s suppose that three features of the faces are used for identification: hair, mustache and beard (Figure 5.6.).

125

Figure 5.6. Three features of faces: hair (1), mustache (2) and beard (9)

Each of the features has three variants: hairless, medium hair and long hair; no mustache, small mustache and large mustache; no beard, small beard and large beard. From all the variants of the features applied to the head of Figure 5.6, the 27 different faces of Figure 5.7. can be deduced. This figure is the same as the number of components in a tripeptide library synthesized by using the same three amino acids in all three coupling positions.

Figure 5.7. The 27 different faces Let’s see now how a suspect can be identified. Suppose the witness says that the criminal has medium hair. Based on this, all suspects of groups 1 and 3, respectively can be omitted from the list. The reason: everybody in group 1 is hairless and nobody in group 3 has medium hair. The perpetrator has to be found among those in group 2. If the witness saw no mustache, the faces of columns 2 and 3 (group 2), respectively, can also be excluded remaining the three faces of column 1 (group 2). If the witness remembers a man with large beard, the criminal can unequivocally be identified as the suspect occupying row 3 in column 1 of group 2. The iteration strategy follows essentially the same approach that was described above and is also demonstrated with a simple example: finding a bioactive component in a tripeptide library of 27 peptides synthesized from the same three amino acids in all of the three coupling positions.

1

2

3

1

2

3

1

2

3

1 2 3 1 2 3 1 2 3 Group 1 Group 2 Group 3

Head

1

2

3

126

In the optimized procedure the synthesis is modified the following way. Before mixing, in each coupling step a sample is removed and preserved for later use (Figure 5.8). After removal of the samples the products are mixed as normal. As an alternative approach, instead removing samples after the last coupling operation before mixing, one may choose to leave unmixed the final three products of the last coupling step. It is a good practice to determine before beginning the iteration whether or not the library contains the desired bioactive component. For this reason the full library is cleaved from the final mixed product of the synthesis and the solution of the peptides is then tested for the presence of the desired bioactive component. The iteration experiments, of course, are executed only if the test is positive. Iteration step No. 1. Determination of the amino acid occupying coupling position 3 in the bioactive peptide. The iteration procedure begins with determination of the amino terminal residue of the bioactive peptide. For this reason the samples removed in the final (third) coupling step are separately submitted to cleavage. The components of the resulting three tripeptide mixtures are demonstrated in the three columns of Figure 5.9.

Figure 5.8. Unmixed products in the split-mix synthesis of a tripeptide library and the samples removed from them.

Large circles: resin, small circles: amino acids. The color of the circles in the boxes of removed samples illustrates the terminal amino acid of the peptides in the samples.

The peptides occupying the same row in the columns differ only in the amino acid occupying coupling position 3 (the N-terminal position). All the peptides of a column have the same N-terminal amino acid and this makes possible to identify the terminal residue of the active

Products of the coupling operations Removed samples

Step 1 Step 2 Step 3

127

peptide. If the sample marked by + shows activity in screening, it means, that the bioactive peptide has to be among the peptides of the marked column otherwise there would be no activity. Figure 5.9. Testing the three sub-libraries cleaved from the three samples removed after the third

coupling. The + sign shows the bioactive sub-library

Since all peptides of this column have “red” amino terminal, consequently the N-terminal amino acid of the bioactive peptide is also the “red” one. The other two amino acids are, of course, not yet identified. This is analogous to choosing the medium hair group in Figure 5.7. Iteration step No.2. Determination of the amino acid occupying coupling position 2 in the bioactive peptide. The amino acid residue occupying the coupling position 2 in the bioactive peptide can be determined in three steps outlined in Figure 5.10.

Figure 5.10. Determination of the amino acid occupying coupling position 2 in the bioactive peptide.

+

Cleavage Cleavage

+

Coupling Coupling

Testing Testing

Samples removed after second coupling

Step 1 Step 2 Step 3

128

The samples removed after the second coupling operation contain dipeptides still attached to the support. All peptides within a sample contain the same (“yellow”, “blue” or “red”) amino acid at their terminal position. In the first step the “red” amino acid – which is known to occupy

the coupling position 3 in the bioactive peptides – is separately coupled to the three samples. In the second step the peptides are separately cleaved from the support. The figure shows that the three groups of peptides are differing from each other only by the amino acid (“yellow”, “blue” or “red”) occupying the coupling position 2. In the third step the three peptide mixtures are separately tested. If the test shows activity in the mixture marked by the plus sign, for example, the bioactive peptide has to be in this mixture (this is similar to selecting column 1 of group 2 in Figure 5.7.). Since all components of the mixture have “blue” amino acid in coupling position 2 the bioactive peptide also has the “blue” amino acid in this coupling position. As a result of iteration step No. 2 the second amino acid of the bioactive tripeptide has been identified leaving only one amino acid to be determined. Iteration step No.3. Determination of the amino acid occupying coupling position 1 in the bioactive peptide. The so far unknown third amino acid of the bioactive peptide can be determined in four steps demonstrated in Figure 5.11. The three samples removed after the first coupling operation are submitted to a two step elongation. First the “Blue” amino acid is attached that is known to occupy coupling position 2 in the bioactive peptide. This is followed by attachment of the “red” amino acid identified as the amino terminal amino acid in the sequence of the active tripeptide. In the third step the peptides are separately cleaved from the support. As the figure shows each of the three product samples contains a single tripeptide. These samples are finally tested. If the sample marked by + sign is proves to be the active one, then the so far unknown amino acid is the “yellow” one. The “yellow” amino acid occupies the C-terminal position of the tripeptide. The amino acid sequence of the bioactive amino acid is: “Red-Blue-Yellow”. This step is analogous to identifying the suspect’s face as that of column 1, row 3 of group 2 in Figure 5.7.

Figure 5.11. Determination of the amino acid occupying coupling position 1 in the bioactive peptide.

Coupling Coupling

Coupling Coupling

Cleavage Cleavage

Testing Testing +

Step 1 Step 2 Step 3 Step 4

129

In a real case the number of iteration steps depends on the length of the peptides in the synthesized library. The number of the samples that need to be tested in an iteration step is the same as the number of amino acids used as building blocks in the corresponding coupling step. A synthesized library may contain a number of bioactive or other useful components that can be identified. It may prove useful to take this into account when deciding the quantity of the library to be prepared. The number and kinds of tests as well as their sensitivity have to play a definitive role in planning. The quantity of the samples removed in the synthetic process as well as the quantity of the final mixed full library is also a question that needs to be decided. It is advisable to remove the samples in each synthetic step in the same molar quantity and leave - after the last sample removal – the same molar quantity for mixing. Since in the course of the synthesis the number of components increases in each step their molarity is accordingly reduced. Table 5.1. The quantity of samples removed in the synthesis of a tetrapeptide library for iteration

experiments

Coupling number Quantity in % 1 0.006 2 0.12 3 2.44 4 48.7

Left for mixing 48.7 If 20 amino acids are used in each step, the molarity of the components decreases by a factor of 20. As a consequence, in order to keep the molarity constant in the removed samples, their proportion have to be increased about 20 times in each step (not exactly 20 times since the total quantity decreases after each sample removal). This is exemplified by a tetrapeptide library synthesized using 20 amino acids in each step. The quantities of the removed samples are shown in Table 5.1. As already mentioned, when dealing with combinatorial libraries made by the split-mix method, less labor is needed not only in the synthesis but also in screening. This is the case doing the deconvolution by the iteration method, too. The efficiency can be shown by a simple example. A tetrapeptide library made from 20 amino acids has 160,000 components. If this library is prepared by the parallel method the screening process needs 160,000 experiments since all components have to be tested separately. The iteration procedure for the same library needs 20 experiments after each coupling operation plus an experiment with the full library, altogether only 81 experiments. 5.2.1.2. Positional scanning When using the iteration strategy, besides the biologist who makes the screening tests, a chemist is also needed for doing the chemical elongation steps on the removed samples according to the intermediate results of the screening experiments. There was a need to develop, if possible, a different screening strategy that could be used by biologists without the help of chemists. This

130

could have been achieved by pre-preparing sets of libraries that without any further modification make possible identification of the bioactive component. The screening strategy fulfilling the requirement described above was independently developed in two laboratories. The principles and realization of the strategy later named positional scanning, was first described in a patent application filed by Furka et al. in May, 19927 then published and practiced by Pinilla et al.8 The sub-libraries described in Chapter 4 and demonstrated in Figures 4.11. and 4.12. can be considered as candidates for being components of pre-prepared sets of libraries in positional scanning.

Figure 5.12. Sub-library (B) and two components (A and C) of a full trimer library (Figure 4.12.)

The reason is demonstrated in Figure 5.12. The sub-library B is one of the 9 sub-libraries of the full library of Figure 4.12. A sub-library is a special partial library of a full library. It is prepared by using a single amino acid in one coupling position in the synthesis. In all other coupling positions all those amino acids are varied that are used in the synthesis of the full library. As a consequence, all those components are present in a sub-library that contain the same (non-varied) amino acid in the non-varied position. The “non-varied position” in the sub-library of Figure 5.12. is coupling position 3, and the “non-varied amino acid” is the “red” one. In the sub-library B all those peptides are present which contain the “red” amino acid in coupling position 3 (for example trimer A) and no peptide is present that has other amino acid in this position (compare to Figure 5.13.). Trimer C, for example is not present in the sub-library because it contains the “yellow” amino acid in coupling position 3. If a bioactive peptide of the full library happens to contain the non-varied amino acid in the non-varied coupling position (like the “red” amino acid in coupling position 3 of the trimer A) then the bioactive peptide has to be found among the components of the sub-library. Consequently, the sub-library gives a positive response in the screening test. On the other hand, if a different amino acid occupies the non-varied position a negative result is expected. In order to be able to do positional scanning, all sub-libraries of a full library have to be prepared and tested. Figure 5.13. shows the full set of sub-libraries and their compositions of the full library A. The number of sub-libraries in a set is the same as the total number of couplings executed in the synthesis of the full library. In the synthesis of library A, three couplings are

3 2 1

3 2 1 3 2 1

A B C

131

executed in each of the three coupling positions so the number couplings as well as that of the sub-libraries is nine.

Figure 5.13. Sub-libraries of the full trimer library A. The non-varied amino acids in columns B, C and D are the “yellow”, “blue” and “Red” amino

acids, respectively. The indices show the non-varied coupling positions. If a full pentapeptide library is made using 20 amino acids in all coupling positions the set of set of sub-libraries contains 100 sub-libraries. A practical way for denotation of the sub-libraries is to use a figure for indication of the non-varied coupling position followed by a one letter symbol for the non-varied amino acid9 as demonstrated in Figure 5.14.

Figure 5.14. Components of a positional scanning kit in the case of a peptapeptide library.

A B1 B2 B3 C1 C2 C3 D1 D2 D3

1A 1C 1D 1E 1F 1G 1H 1I 1K 1L 1M 1N 1P 1Q 1R 1S 1T 1V 1W 1Y 2A 2C 2D 2E 2F 2G 2H 2I 2K 2L 2M 2N 2P 2Q 2R 2S 2T 2V 2W 2Y 3A 3C 3D 3E 3F 3G 3H 3I 3K 3L 3M 3N 3P 3Q 3R 3S 3T 3V 3W 3Y 4A 4C 4D 4E 4F 4G 4H 4I 4K 4L 4M 4N 4P 4Q 4R 4S 4T 4V 4W 4Y 5A 5C 5D 5E 5F 5G 5H 5I 5K 5L 5M 5N 5P 5Q 5R 5S 5T 5V 5W 5Y

132

The first step in positional scanning is testing the full library. If the result is positive then all components of the kit are tested. From the result, the amino acid sequence of the bioactive peptide can be deduced. If the assays show that the sub-libraries marked by boxes in Figure 5.14. are bioactive then the coupling positions 1, 2, 3, 4 and 5 are occupied in the bioactive peptide by E, L, V, R and T, respectively. Taking into account that the order of amino acids in peptide sequences is opposite of their coupling order, the sequence of the bioactive peptide is:

T-R-V-L-E

As it will be shown below, preparation of all sub-libraries, for example the 100 sub-libraries of Figure 5.14., needs too much work and long time to prepare them for screening with a single target. A company, however, can synthesize them in larger quantities and divide them into smaller equimolar portions. The collections formed from these smaller quantities could be sold as kits ready for use by biologists. The synthesis of a single sub-library of Figure 5.14. needs 81 couplings: one coupling in the non-varied coupling position plus 20 couplings in each of the remaining 4 positions. Since there are 100 different sub-libraries in the figure, the total number of the required couplings is 8100! The synthesis of the 100 sub-libraries, however, can be optimized in order to be able do the preparation in less number of amino acid couplings. The optimization can be achieved by doing as many couplings as possible with the combined form of the sub-libraries under preparation. This is briefly described bellow. Step 1. Preparation of the 1A to 1Y sub-libraries. The resin is divided into 20 equal portions then a different amino acid is coupled to each portion. No mixing. One fifth part of each portion is removed then four full split-mix cycles (split-couple-combine) are executed on each of the 20 removed portions to get the 20 1X type sub-libraries. The total number of couplings is:

20 + 4x20x20 = 1620

Step 2. Preparation of the 2A to 2Y sub-libraries. The 20 portions remaining in Step 1 are mixed, divided into 20 equal samples then each of them is coupled with a different amino acid. No mixing. One fourth part of each sample is removed. The removed samples are separately submitted to three full split-mix cycles. The result is the 20 2X type library. The total number of couplings is:

20 + 3x20x20 = 1220

Step 3. Preparation of the 3A to 2Y sub-libraries. The 20 samples remaining in Step 2 are mixed, divided into 20 equal portions then a different amino acid is coupled to each one. No mixing. After coupling one third part of each sample is removed then separately submitted to two full split-mix cycles. The product is the 20 3X type library. The number of the executed amino acid couplings is:

20 + 2x20x20 = 820

Step 4. Preparation of the 4A to 4Y sub- libraries. The 20 samples remaining in Step 3 are mixed, divided into 20 equal parts then a different amino acid is coupled to each part. No mixing. Half of each sample is removed then one full split-mix cycle is executed on every sample. As a

133

result, the 20 4X type sub-libraries are formed. The total number of the executed amino acid couplings is:

20 + 20x20 = 420

Preparation of the 5A to 5Y sub-libraries. The 20 samples remaining in Step 4 are not mixed. Each of them is coupled with a different amino acid. The products are the 20 5X type sub-libraries. The total number of couplings is:

20

The total number couplings executed in the whole process leading to the 100 sub-libraries of the positional scanning kit of pentapeptides is: 4100. Compare this to the 8100 couplings needed in the non-optimized process. The number of couplings could be reduced by almost 50%. The fact that the synthesis of a positional screening kit is still to laborious prompted us to think about other possibilities to circumvent the need for preparation of full positional scanning kits. The result was the development of the omission libraries and the amino acid tester libraries. 5.2.1.3. Omission libraries. The potential applicability of omission libraries in screening was first independently realized in two laboratories.10,11 Peptide omission libraries can be prepared using the split-mix method like in the synthesis of full libraries with an important difference: one amino acid is omitted in all coupling positions. It is important: the same amino acid is omitted in all coupling positions. This is illustrated in Figure 5.15. Figure 5.15. Amino acids used in the synthesis of an omission peptapeptide library. Histidine (H)

is omitted in all the five coupling positions. As a consequence of the omission of histidine in the synthesis, no histidine containing peptide forms. All histidine containing peptides that otherwise would be present in the full library are missing. In the figure no H containing lines can be drawn. A short symbol can be used to denote the omission libraries: a minus sign followed by the one letter symbol of the omitted amino acid. The symbol of a histidine omission library, for example, is -H. An optional figure can be appended to the symbol indicating the number of the amino acid building blocks in the peptides (length). Accordingly, the symbol of a histidine omission pentapeptide library is: -H5.

A C D E F G I K L M N P Q

R S T V W Y 1

A C D E F G I K L M N P Q

R S T V W Y 2

A C D E F G I K L M N P Q R S T V W Y 3

A C D E F G I K L M N P Q

R S T V W Y 4

A C D E F G I K L M N P Q

R S T V W Y 5

134

The number of peptides in omission libraries synthesized from 20 amino acids, as well as the number of peptides missing from them is summarized in Table 5.2. Table 5.2. Number of peptides in full and omission libraries and the number of peptides missing

from the omission libraries

Length Full library Omission library Missing peptides 2 400 361 39 3 8,000 6,859 1141 4 160,000 130,321 29,679 5 3,200,000 2,476,099 723,901 6 64,000,000 47,045,881 16,954,119

Figure 5.16. Composition of omission libraries

Using a very simple example, Figure 16. shows how the composition of omission libraries can be derived from that of the full one. The fact that all the peptides of the omitted amino acid are missing from the library makes the omission libraries applicable in the identification of

Omitted amino acid „Yellow” „Blue” „Red”

Full library

+ + - -

Amino acids in the bioactive peptide:

135

bioactive peptides. If the bioactive peptide contains the omitted amino acid, for example, the omission library gives negative result in testing, since the bioactive peptide is not present in the omission library. On the other hand, if the bioactive peptide does not contain the omitted amino acid the omission library gives positive test since all peptides except those of the omitted amino acid are present. Based on these properties the omission libraries can be used for determination of the amino acid composition of the bioactive peptide. This is demonstrated in Figure 5.16. If two omission libraries (“blue” and “red”) give negative result in screening test this means that the bioactive peptide contains “blue” and “red” amino acids. The “yellow” amino acid is not present because the test with the “yellow” omission library is positive. By use of omission libraries the amino acid composition of the bioactive peptide can be determined. Nothing is known, however, about the coupling position of these amino acids. This has to be determined by additional experiments. This task, however, is much less complicated than the original one. This can be illustrated by a simple example. Suppose we deal with a full pentapeptide library and the result of using the 20 omission libraries is that the bioactive peptide contains the following four amino acids: A, G, R and H. Then a much less complex library can be defined that is built up using these 4 amino acids in all coupling positions.

1 A G R H 2 A G R H 3 A G R H 4 A G R H 5 A G R H

If this simpler library – that contains only 256 components instead of the 3.2 million ones in the in the original pentapeptide library – is prepared the bioactive peptide must be present in it. This means that the original task is reduced to a much simpler one. This simpler library can be named “occurrence library”. The amino acid sequence can be determined by application of the iteration or the positional scanning method to the occurrence library. Practical example will be shown in a separate paragraph. The synthesis of the omission libraries is simple. The number of libraries in a kit is 20 if the 20 amino acids are used in preparation of the full library. The number of components of the kit does not depend on the length of the peptides. Preparation of the kit is simpler and less time consuming than, for example, the synthesis of the 100 components of the positional scanning kit. 5.2.1.4. The amino acid tester libraries The peptide mixtures missing from omission libraries contain all peptides of the omitted amino acid irrespective of its number or position occupied in the sequences. In addition, there is no peptide in the mixture that does not contain the omitted amino acid at least in one position. As pointed out by Câmpian et al.12,13 these properties make possible to use these mixtures, as an alternative method, for determination of the amino acid composition of bioactive components of peptide libraries. The amino acid libraries form a kit. The number of components in an amino acid tester kit is the same as the number of amino acid building blocks used in the synthesis of

136

the full library. If 20 amino acids are used in the synthesis, the number of components is 20 like in the case of the omission library kit. An amino acid tester library gives an opposite result in screening compared to an omission library. An alanine tester library, for example, that comprises all alanine containing peptides shows activity in screening only when alanine is present in the sequence of the bioactive peptide. If the test is carried out with a tester library of an amino acid that is not present in the bioactive component no or a reduced activity is expected. Figure 5.17. shows the composition of simple amino acid tester libraries that can be compared to the full one. If the libraries marked by the plus sign prove to be active then the amino acid composition of the bioactive peptide is: “yellow”, “red”.

Figure 5.17. Composition of amino acid tester libraries Figure 5.17. shows that the components of the amino acid tester libraries may contain the amino acid to be tested in one, two, three etc. positions. Table 5.3. shows how the components of an alanine tester library can be derived in the case of a full tripeptide library prepared using 20 amino acids. The components of the tester library are divided into 7 groups. The group 1 peptides, for example, contain alanine in coupling position 1, and in the positions 2 and 3 the remaining 19 amino acids are varied. In group 2 and 3 peptides the single alanine occupies position 2 and 3, respectively. In groups 4, 5 and 6 alanine is found in two positions. Group 7 comprise a single peptide with alanine in all three positions. The

Tester library for the amino acid „Yellow” „Blue” „Red”

Full library

+ + - + Amino acids in the bioactive peptide:

137

total number of peptides in the seven groups is 1141, in accordance with the corresponding figure of Table 5.2. Considering the possibility of the synthesis, the groups of 1, 4, 5, 7 and groups 5, 6 of Table 5.3. can be amalgamized into groups 1 and 2, respectively of Table 5.4. Group 3 of Table 5.3. remains alone and is transferred into Table 5.4. as group 3. The total number of peptides of course remains unchanged: 1141.

Table 5.3. Groups of peptides occurring in a tripeptide alanine tester mixture

Coupling position Group 1 Group 2 Group 3 Group 4 Group 5 Group 6 Group 7 1 A 19 19 A A 19 A 2 19 A 19 A 19 A A 3 19 19 A 19 A A A

Number of peptides 361 361 361 19 19 19 1 The synthesis of the alanine tester library needs the preparation and mixing of the three partial libraries represented in Table 5.4. as groups 1 to 3. Although not demonstrated in the tables, the number of component libraries to be prepared in the case of tetrapeptides and pentapeptides is four and five, respectively.

Table 5.4. Groups of peptides to be synthesized

Coupling position Group 1 Group 2 Group 3 1 A 19 19 2 20 A 19 3 20 20 A

Number of peptides 400 380 361 As mentioned in paragraph 4.1.1.6 preparation of the missing part of some partial libraries, by which the partial library can be completed to a full one, is very complicated and time consuming. The amino acid tester libraries that are the missing part of the omission libraries clearly exemplify this. The amino acid tester libraries are less complex than the omission libraries, they contain less number of components (see Table 5.2.). This may be advantageous in screening. The more complicated synthesis is, of course, a disadvantage. In the 5.2.1.6. paragraph an example will be presented to show that the component libraries of an amino acid tester library can be prepared in a single run using the automatic synthesizer aapptec 357.

138

5.2.1.5. Other methods for identification of the bioactive component of combinatorial libraries In addition to the deconvolution strategies described above other methods were also published that makes possible the identification of the bioactive components of dissolved combinatorial libraries. These methods allow separation of the bioactive molecules from the rest of the components of libraries based on the specific binding of the bioactive component to the target molecule. In one of such approaches the target molecule is covalently attached to an insoluble matrix. A chromatography column is filled with the matrix then the dissolved library is passed through the column. The active components are retarded in the column since they bind to the target while the inactive components pass. Like in other affinity approaches, after washing the column the active molecules are eluted. The separated compounds are then submitted to structure determination. In another approach both the target protein molecule and the components of the library are dissolved in the same medium. The active molecules are allowed to bind to the target protein then the mixture is submitted to size exclusion chromatography. The large protein molecule holding the attached active molecule readily separates from the inactive small molecule library components. Applying appropriate conditions the ligand dissociates from the receptor and its structure can be determined. 5.2.1.6. Dynamic combinatorial libraries The dynamic combinatorial libraries introduced by Professor J.-M. Lehn25 differ from the so far described “static” libraries. They are combinatorial libraries since they contain all components that can be derived by combination of the building blocks. It is very important, however, that their components are in equilibrium. The constituents undergo continuous interconversion by recombination of their building blocks. This dynamic interconversion of the components gives the specialty of these libraries and, at the same time, determines their applicability. Addition of a target molecule to the mixture of molecules forming the equilibrium favors the formation of the component that binds to the target. Binding of the component to the target creates a strong driving force towards its formation. In principle, this method is capable to accelerate the identification of lead compounds in drug discovery. It should be noted, however, the applicability of the method in the pharmaceutical research is limited since the use of the majority of the combinatorial synthetic methods leads to formation of “static” libraries. 5.2.1.7. Examples The applicability of the screening strategies described before is demonstrated by a few model experiments. The task in these experiments was to determine whether or not a synthesized tripeptide library has a component that inhibits binding of LHRH14 to its antibody. The amino acid sequence15 of the hormone is shown below:

pGlu-His-Trp-Ser-Tyr-Gly-Leu-Arg-Pro-Gly-NH2

139

The LHRH polyclonal antibody as well as the radioactively labeled LHRH were the products of Advanced ChemTech. The competitive inhibition of LHRH to its antibody was determined by radioimmunoassay16 (RIA). Testing the full trippeptide library.11 Since the LHRH is a decapeptide amide, a tripeptide amide library was prepared and used in screening. In the split-mix synthesis of the tripeptide amide library 19 amino acids were used in the first and second coupling position (cysteine was omitted) and, because LHRH has pyroglutamic acid at the N-terminal position, pyroglutamic acid was added to this set in coupling position 3. The library was prepared on Rink amide resin using the F-moc strategy. The tripeptide amide library was added in increasing concentrations to the mixture of radioactive LHRH and its antibody and their binding was determined by RIA. The result is demonstrated in Figure 5.18. It can be seen that binding is strongly reduced by increasing the concentration of the library. This makes probable that the library has component/s that inhibit binding, that is, it is worthwhile to make further experiments in order to identify this component. The result also suggests that the optimal concentration for the binding experiments should be around 50 microgram/ml. In all the further experiments the libraries were applied in molarities equivalent to this concentration.

Figure 5.18. Effect of concentration of the full tripeptide library on binding of LHRH to its antibody

Iteration experiment.17 Before the final mixing in the split-mix synthesis of the full tripeptide amide library, equal samples were removed and cleaved from the support. These mixtures were suitable to demonstrate the first step in the iteration strategy.

140

-100

102030405060708090

100

3A 3D 3E 3F 3G 3H 3I 3K 3L 3M 3P 3Q 3R 3S 3T 3V 3W 3Y 3p

100

- LH

-RH

Bin

ding

%

FiFigure 5.19. Inhibitory effect of sub-libraries used in the first iteration step.

3p denotes pyroglutamic acid in the N-terminal position Figure 5.19. shows how strong the inhibitory effect of the sub-libraries is. It can be clearly seen that sub-library 3R exhibits the far strongest effect. This means that the amino acid occupying the coupling position 3 in the inhibitory tripeptide amide is arginine, R. Application of omission libraries.10 The omission libraries were derived from the tripeptide amide full library described above. Thus in the synthesis of 19 omission libraries (-A to –Y) one amino acid was omitted in all the three coupling positions. The remaining 18 amino acids were built in into all positions. Pyroglutamic acid was also present in all omission libraries in coupling position 3. Pyroglutamic acid omission library could not be prepared since this amino acid can be inserted only into coupling position 3 of the tripeptides. It was also important, however, to test whether or not this amino acid is present in the active peptide. For this reason a full tripeptide amide library was prepared by using 19 amino acids in each coupling position and the pyroglutamic acid was omitted from position 3 (denoted by –p). The importance in inhibition of the amide groups of the peptides was also tested. In order to do this one part of the full library of tripeptides was cleaved from the support in the form of carboxylic acids instead of amides and tested in this form (denoted by -a). When tested the omission libraries, the full tripeptide amide library (denoted by X) was also included. The result is demonstrated in Figure 5.20. It can be seen that the omission libraries that less reduces the competitive binding are: -G, -P, -R and –a. This means that the amino acid composition of the inhibitory tripeptide is glycine (G), proline (P) and arginine R. The peptides that do not have amide groups are not effective inhibitors. This means that the amide group is also essential part of the inhibitory tripeptide.

141

0

20

40

60

80

100

120

X -A -D -E -F -G -H -I -K -L -M -N -P -Q -R -S -T -V -W -Y -p -a

Bin

ding

of L

H-R

H %

Figure 5.20. Effect of omission libraries on binding. X and -a mean full tripeptide amide and full tripeptide acid libraries, respectively, while -p

denotes the library from which pyroglutamic acid was omitted. The other omission libraries are represented by a minus sign followed by the one letter symbol of the

omitted amino acid The results of the experiments carried out with omission libraries gave no indication about the position of the amino acids within the sequence of the active peptide. Despite this, the information gained by only 21 screening experiments is very valuable. They define an „amino acid occurrence library” that can be synthesized by varying only three amino acids, Gly, Pro, and Arg in all of the three coupling positions. The inhibitory tripeptide is expected to be present among the 27 components of this tripeptide amide library. In other words, by screening with omission libraries, the complexity of the library in which the active peptide is found could be reduced from the original 7220 to only 27. The positions of the identified amino acids could be determined by using one of the following three possibilities:

1. Preparation by parallel synthesis and screening of the 27 components of the occurrence library.

2. Application of positional scanning to the occurrence library (preparation and screening of nine sub-libraries).

3. Positional scanning with nine sub-libraries of the full library (if available). Preparation and use of amino acid tester libraries.12 Amino acid tester libraries offer an alternative choice besides omission libraries for determination of the amino acid composition of active peptides. The set of libraries used in the experiments were derived from the full libraries prepared from 19 amino acids plus pyroglutamic acid in coupling position 3. Each library of the set contains 1064 tripeptide amides and can be formed like the alanine tester library in Table 5.5. by mixing the groups 1 to 3. Instead of preparing separately the three groups, however, they were synthesized then mixed in a single optimized process using the ACT 357 automatic synthesizer.

142

Table 5.5. Components libraries of the alanine tester library

Coupling position Group 1 Group 2 Group 3 1 A 18 18 2 19 A 18 3 20 20 A

Number of peptides 380 360 324 The flow diagram of the synthesis of the alanine tester mixture is demonstrated in Figure 5.21. The procedure was started with 1.064 g Rink resin distributed in the reaction block (Figure 5.22.) according to the number of peptides in Group 1 (0.380 g in one of the reaction vessels) and Groups 2 plus Group 3 (0.684 g in the collection vessel). The Group 1 resin was first coupled with alanine, distributed into 19 samples and each was coupled with one of the 19 amino acids (including alanine), and then mixed. The Group 2 + 3 resin was distributed into 18 equal parts, coupled with one of the 18 amino acids (no alanine among them) then mixed. The mixture was split into two parts in quantities corresponding to the number of peptides in Group 2 (0.360 g) and Group 3 (0.324 g). The Group 2 resin was coupled without portioning with alanine then mixed with the resin sample containing the Group1 peptides. The Group 3 resin was portioned into 18 equal parts, coupled with one of the 18 amino acids (no alanine among them), mixed then coupled with alanine without portioning.

Figure 5.21. Flow diagram of the synthesis of the alanine tester library (ATL). A: Coupling with alanine; 18, 19, 20: Portioning, then coupling with 18, 19 and 20 different

amino acids, respectively, then mixing.

18

18 19

20

A

A

A

0.684

0.360 0.324

0.380

ATL

A

Group 1

Group 2 Group 3

1.064

Groups 1+2

143

The Groups 1 + 2 resin was distributed into 20 equal parts, each coupled with one of the 20 amino acids (including alanine and pyroglutamic acid) then mixed. Finally, the Groups 1+2 resin was mixed with the Group 3 sample to give the alanine tester library (ATL). All operations were preprogrammed and at the end the product was accumulated and mixed in the collection vessel. The applicability of the synthesized amino acid tester libraries in determination of the amino acid composition of active peptides was tested under reaction conditions described at omission libraries. Figure 5.23. shows that the inhibition of binding of LHRH to its antibody is strongest in the case of the glycine (G), proline (P) and arginine (R) tester libraries. Consider that on the y axis not LHRH binding% but 100- LHRH binding% is plotted. Consequently the active tripeptide amide contains glycine, proline and arginine.

Figure 5.22. The reaction block of the ACT 357 Synthesizer and the quantities and places of the

resin at the start of the process Applicability of the amino acid tester libraries is the same as that of the omission libraries.

01020304050607080

A D E F G H I K L M N P Q R S T V W Y

100-

LHR

H B

indi

ng%

Figure 5.23. Effect of amino acid tester libraries on binding of LHRH to its antibody

Synthesis and use of positional scanning libraries. Screening with both omission and amino acid tester libraries led to the same result: binding of LHRH to its antibody is inhibited by a tripeptide amide having composition glycine, proline and arginine. As outlined before, based on

Reaction vessel 0.38 g resin Collection vessel

0.684 g resin

144

this result an occurrence library can be defined. If this library is synthesized, it contains among its components the inhibitor tripeptide amide (Table 5.6.).

Table 5.6. Building blocks of the occurrence library

Coupling step Amino acids 1 G P R 2 G P R 3 G P R

The position of the three amino acids in the sequence of the inhibitor tripeptide amide is determined by synthesizing and testing of the nine component omission library kit of the positional scanning library. The synthesis is optimized to make possible to prepare the nine component libraries of the kit (1G, 2G, 3G, 1P, 2P, 3P, 1R, 2R and 3R) in a single run on the automatic synthesizer ACT 357. The solid support was again Rink resin. The synthesizer was pre-programmed that made possible to execute the whole process automatically. The flow diagram is demonstrated in Figure 5.24. The starting resin, placed into the collection vessel was first divided into three portions then coupled with glycine, proline and arginine, respectively.

Figure 5.24. Flow diagram of the synthesis of the 9 positional scanning sub-libraries of the occurrence library

Coupling with G

Coupling with P Coupling with R

Mix

Split

Mix & Split

2G 2P 2R 1G 1P 1R

3G 3P 3R

Start

1/3 1/3 1/3

1/2 1/2 1/2

145

Before mixing, 1/3 part of each reaction product was transferred into a separate reaction vessel then each of them was individually submitted to two consecutive portioning-mixing cycles coupling in each cycle with glycine, proline and arginine, yielding 1G, 1P, and 1R as end product. The remainder was mixed, divided into three parts then each part coupled with one of the three amino acids. Again, before mixing, 1/2 part of each sample was transferred to a separate reaction vessel then individually submitted to a full portioning mixing cycle, using again glycine, proline and arginine in couplings. These operations resulted in formation of 2G, 2P and 2R. The remainder was mixed, divided into three portions then each coupled with one of the three amino acids. The three products were 3G, 3P and 3R.

The synthesized nine first order sub-libraries were used to determine the position of glycine, proline and arginine in the tripeptide responsible for competitive inhibition of binding of LH-RH to its antibody (Figure 5.25).

0

20

40

60

80

100

1R 2R 3R 1G 2G 3G 1P 2P 3P

100

- LH

-RH

bin

ding

%

Figure 5.25. Positional scanning by sub-libraries of the occurrence library

Since on the y axis not LHRH binding% but instead 100- LHRH binding% is plotted, Figure 5.25. shows that the inhibition of binding of LHRH to its antibody is strongest in the case of the 1G, 2P and 3R sub-libraries. Consequently, arginine, proline and glycine occupy the coupling positions 3, 2 and 1 in the tripeptide, respectively. The sequence of the inhibitor tripeptide is Arg-Pro-Gly-NH2. This sequence happens to be identical with the C-terminal sequence of LHRH. 5.2.2. Deconvolution methods of libraries not cleaved from the solid support The components of the tethered libraries are found in the beads of the solid support as individual compounds. Consequently, they can be tested as individual substances. It has to be taken into account, however, that the structure of the compounds present in any particular bead is unknown. For this reason the deconvolution process has to solve two problems:

Identify the bead that contains the component showing the wanted property Identify the compound tethered to the bead.

146

The beads containing the individual components of the combinatorial libraries can be, and are, tested in two different ways:

The components of the libraries are tested in tethered form The screening tests are carried out with compounds cleaved from individual beads.

Both approaches have to make possible identification of the bead containing the active compound and determination of the structure of the wanted compound. As it will be shown both deconvolution processes need less number of assays then are needed in the determination of the activities of compound arrays prepared by parallel synthesis. 5.2.2.1. Screening of combinatorial libraries in tethered form When the beads are immersed into solution the tethered compounds are available to interaction with dissolved molecules. They can specifically bind to receptors, antibodies, enzymes, viruses, etc. For this reason binding tests offer themselves as assays in deconvolution processes. The first example of carrying out binding tests with individually synthesized peptides tethered to the beads of the solid support was described by Smith and his colleagues.18 After the split-mix combinatorial synthetic method became available19 this approach was applied to tethered peptide libraries.20

Figure 5.26. Identification of the beads that specifically bind to target molecules. The beads containing specifically binding peptide are colored

In application of the method the beads containing the full tethered peptide library, or a fraction of it, is immersed in the solution of a target molecule. The beads containing peptide that binds to the target are identified, separated from the rest of the beads then the peptide they contain is sequenced. The binding have to be visualized somehow. This can be achieved by labeling the target molecule before or after the binding experiment. The target protein can be

147

labeled by attaching to it a color, fluorescent or radioactive residue. If the target protein is labeled by color and the beads are examined through a microscope the binding beads can easily be distinguished from rest of the beads by their color as shown in Figure 5.26. The beads are colored because the target molecules are colored that are attached to the peptide molecules of the bead. The colored beads are manually separated from the rest of the beads then by washing with an appropriate solvent the attached labeled target protein is removed from them. After washing, the sequence of the peptides is determined using automatic sequencing machine. A more effective and faster selection process can be applied if the target molecule is labeled by fluorescence. The beads can be sorted by fluorescence-activated cell sorting instrument. A special automatic machine was developed by Morten21 for sorting fluorescent beads. A different approach, infrared termography, was applied by Taylor and Morken22 to identify catalysts in non-peptide tethered libraries synthesized by the split-mix procedure. The method is based on the heat that is evolved in the beads that contain a catalyst when the tethered library immersed into a solution of a substrate. The heat increases the temperature of the catalyst containing beads. In the beads that do not contain catalyst no heat is evolved so their temperature remains unchanged. Although the increase of the temperature is very low, when the beads are examined through an infrared microscope the catalyst containing beads appear as bright spots as demonstrated in Figure 5.27. The screened library was an organic one and the beads were encoded. The bright beads were separated and the identification process ended with determination of their code.

Figure 5.27. Identification of catalyst containing beads by infrared thermography

5.2.2.2. Screening of combinatorial libraries by releasing the content of individual beads into solution An alternative way of screening the tethered libraries is releasing the content of the individual beads into solution then the libraries can be screened like the compound arrays prepared by the parallel synthetic methods. If the bead containing the active compound is identified, however, the procedure has to be continued by determination of the structure of the released compound. If only a fraction of the content of the bead is released for screening the rest

148

can be used for structure determination. If encoded libraries are used the synthesized compounds and the codes can be released separately. It is also a possibility to release the content of the beads and identify the bead containing the active compound in several stages. The procedure developed at Pharmacopeia23 uses this latter possibility for screening libraries of small organic compounds. The libraries are prepared applying the binary encoding technique and using a photolabile linker which allows a two stage release of the organic substance. After portions of beads are distributed into small containers (Figure 5.28/A), the first portions of the substances are released by irradiation. The content of each vessel is then submitted to screening. If one of them proves to be active (marked by + sign in the figure), the beads of this container are re-distributed into vessels each containing a single bead (Figure 5.28/B). After releasing the second portion of the substances, a second screening identifies the bead which carried the active substance (marked by + sign). Finally the encoding molecules are released from the identified bead and determined by electron capture gas chromatography thus determining the structure of the organic molecule responsible for the biological activity.

Figure 5.28. Identification of the bead containing the bioactive component in two stages. It is worthwhile to note that this two stage process needs less number of screening experiments then does the one stage process when all beads are tested individually. The 11 containers of Figure 5.28/A contains a total of 110 beads. The one stage process would need 110 screening experiments. The two stage process needs – as shown in the figure – only 21 assays. 5.2.2.3. Examples The application of the screening methods developed for libraries not cleaved from the solid supports are demonstrated by examples described in the literature. Identification of bioactive peptides with enzyme-linked colorimetric assay.24 The experiment is carried out with tethered peptide libraries synthesized on TentaGel resin. Binding of the target molecule to the beads containing the active peptide is indicated by the blue color of an indigo derivative that forms from 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) dephosphorylated by the enzyme alkaline phosphatase. The target protein is biotinylated before the binding experiment and the alkaline phosphatase is derivatized with streptavidin. The target protein binds strongly the reporter enzyme that converts BCIP to the blue indigo. If the protein binds to a bead containing the active peptide then a solution of BCIP is added, then insoluble

+

+

A B

149

indigo forms on the bead staining its surface to blue. The experiment is carried out with 5,000 to 50,000 beads at room temperature or 4oC. The goal of the experiment is to identify only those beads that bind the target protein by strong specific interaction. Weak, non-specific interactions may also occur. In order to make these interactions invisible, the beads are incubated with gelatin that coats the beads by weak, non-specific interaction. The strong specific binding of the target molecule to the beads is expected to displace the gelatin from the surface. Non-specific binding, however, may also occur with the alkaline phosphatase, too. In order to eliminate the misleading effect of this the beads are prescreened with streptevidin-alkaline phosphatase. The beads are incubated with streptevidin-alkaline phosphatase and after washing a incubated with a BCIP solution. As a result of interaction of the streptavidin-alkaline phosphatase with some peptides, blue beads appear that are manually removed. Following the prescreening and washings, the beads are ready for final screening. The beads are incubated with a solution of the biotinylated target protein for 1-2 hours when the target molecules displace the gelatin from some beads and bind strongly to the their peptides. This is an invisible process. In order to make it visible the beads, also after washings, are first incubated with streptavidin-alkaline phophatase which is attached to the immobilized target molecules via the strong biotin-streptavidin interaction. Finally, the beads are washed with BCIP solution. Blue color develops on some beads. The beads are examined in Petri dish under microscope. The blue beads are removed manually. The target protein-enzyme complex is washed away from the beads with urea solution. The amino acid sequences of the peptides that show specific binding property can be determined without cleaving them from the beads. Automatic micro-sequencing machines that are based on the well known Edman degradation can be used for this purpose. Whole cell binding assay.24 In addition to target proteins, binding experiment can be carried out with intact cells, too. This approach can be used to study the cell surface receptors and their ligands. Sterilized tethered peptide libraries are used in the experiments. The beads surrounded by binding cells can be distinguished from the inactive ones as seen in Figure 5.29. The active beads are again manually separated and sequenced after removal of the attached cells. Figure 5.29. Cell binding. The large circles are beads, the smaller ones are cells. The cell binding

beads are marked by arrow

150

References

1. Gy. Takátsy Acta Microbiologica Acad. Sci. Hung. 1955, 3, 191. 2. Furka Á (1982) Tanulmány, gyógyászatilag hasznositható peptidek szisztematikus

felkutatásának lehetöségéröl (Study on possibilities of systematic searching for pharmaceutically useful peptides). Unpublished theoretical study written in Hungarian for internal use, describing the PM synthesis and an iteration strategy for screening of soluble libraries. Notarized on June 15, 1982, file number 36237/1982. See also paragraph 1.2.

3. Á. Furka Drug Discovery Today 2002, 7, 1. 4. H. M. Geysen, R. H. Meloen, S. J. Barteling Proc. Natl. Acad. Sci. USA 1984, 81, 3998. 5. R. A. Houghten, C. Pinilla, S. E. Blondelle, J. R. . Appel, C. T. Dooley, J. H. Cuervo

Nature 1991, 354, 84. 6. K. D. Janda Proc. Natl. Acad. Sci. 1994, 91, 10779. 7. Á. Furka, F. Sebestyén WO 93/24517. 8. C. Pinilla, J. R. Appel, R. A. Houghten, In C. H. Scneider, A. N. Eberle, (Eds) Peptides

1992, 1993, ESCOM, Leiden, 65. 9. Á. Furka Drug Development Research 1994, 33, 90. 10. T. Carell, E. A. Winter, J. Rebek Jr. Angew. Chem. Int. Ed. Engl. 1994, 33, 2061. 11. E. Câmpian, M. Peterson, H. H. Saneii, Á. Furka Bioorg. & Med. Chem. Letters 1998, 8,

2357. 12. E. Câmpian, J. Chou, M. L. Peterson, H. H. Saneii, Á. Furka, R. Ramage, R. Epton (Eds)

In Peptides 1996, 1998, Mayflower Scientific Ltd. England, 131. 13. E. Câmpian, H. H. Saneii and Á. Furka PharmaChem 2003, April, 43. 14. A. V. Schally et al. J. Biol. Chem. 1971, 246, 7230. 15. H. Matsuo et al. Biochem. Biophys. Res. Commun. 1971, 43, 1334. 16. C. Patrono, B. A. Peskar, (Eds), Radioimmuoassay in Basic and Clinical Pharmacology,

1987, Springer -Verlag, Heidelberg. 17. E. Câmpian, J. Chou, Á. Furka unpublished results. 18. J. A. Smith J. G. R. Hurrel, S. J. Leach Immunochemistry 1977, 14, 565. 19. Á. Furka, F. Sebestyén, M. Asgedom, G. Dibó, In Highlights of Modern Biochemistry,

Proceedings of the 14th International Congress of Biochemistry, VSP. Utrecht, The Netherlands, 1988, Vol. 5, p 47.

20. K. S. Lam, S. E. Salmon, E. M. Hersh, V. J. Hruby, W. M. Kazmierski, R. J. Knapp Nature 1991, 354, 82 and its correction: K. S. Lam, S. E. Salmon, E. M. Hersh, V. J. Hruby, W. M. Kazmierski, R. J. Knapp Nature 1992, 360, 768.

21. M. Meldal Biopolymers (Peptide Science) 2002, 66, 93. 22. S. J. Taylor, J. P. Morken Science 1998, 280, 267. 23. http://www.pharmacopeia.com 24. K. S. Lam, A. L. Lehman, A. Song, N. Doan, A. M. Enstrom, J. Maxwell, R. Liu, In G. A.

Morales, B. A. Bunin (Eds) Methods in Enzymology, Combinatorial Chemistry Part B, 2003, Elsevier Academic Press, 298.

25. J.-M. Lehn Chem. Eur. J. 1999, 5, 2455.

151

6. Combinatorial methods in materials and catalyst research

Besides the organic compounds utilized as drugs, pesticides etc. there exists another important group of materials that have definitive effect on our every-days life. These are the solid inorganic materials and polymers. These materials substantially differ from organic compounds that have well defined molecular structure. In the inorganic solid materials the elementary composition is not always stochiometric, the proportion of the component elements may be very different. In addition a considerable part of the elements of the periodic table may occur among their constituents. The polymers also differ from the small molecular organic compounds. They usually contain a large but undefined number of building blocks. For this reason preparation and examination of this class of materials needs special methods. Nevertheless discovery of the new materials that have useful properties also require preparation and testing of a very large number of samples. In order to speed up the research, application of the combinatorial thinking and the combinatorial methods seems to be a realistic choice. In 1970 J. J. Hanak1 reported a new methodology for fast screening of new electronic materials. Hanak smoothly varied the concentration of components of the mixtures so the effects of a large number of composition differences could be measured on a single sample. His approach was disregarded for a very long time and only the advent of the combinatorial methods2-5 speeded up the production and testing of the new materials. It was only in 1995 when Xiang et al.6 prepared and tested parallel samples of materials. They also demonstrated that by application of the combinatorial methods, the already known high temperature superconducting compounds could be readily identified. The field of combinatorial materials research is still rapidly expanding. The main research areas that are using the combinatorial high-throughput approach are inorganic materials, catalysts and polymers. Mapping of composition-property relationships, optimization by using of phase diagrams and parameter space are an integral part of research. 6.1. Inorganic materials The classes of inorganic materials that can be studied by combinatorial methods include semi- and superconductors, dielectrics, phosphors, superalloys, magnetoresistive materials and others. The main areas where the new inorganic materials may find application are electronic devices, displays, memory devices, photonic devices, magnetic and optical data storage. The inorganic solid materials are best investigated in the form of thin films. A number of methods have been developed to fabricate these films. A few of these methods is outlined below. 6.1.1. Fabrication of thin film libraries In the most inorganic libraries the primary combinatorial variable is the chemical

152

composition. Additional parameters, however, are also very important in library fabrication like temperature, the atmosphere and the pressure that also should be varied. One of the main methods in thin film library fabrication is the vapor deposition technique. In these techniques like sputtering, pulsed laser deposition, electron beam evaporation or laser molecular epitaxy, the target material is evaporated by a high energy source (ion gun, electron gun, laser) then is deposited on the substrate. Another fabrication approach is to deliver the components of the film into small wells in dissolved form then to evaporate the solvent. The physical vapor deposition technique makes possible formation of two kinds of libraries. In one kind of libraries the components are discrete films each having a definite composition that differs from film to film. The other kind of library is formed in a single film so that the composition is smoothly varied across the film. The composition of such films is different in all of its points. The delivery of components in dissolved form into wells leads to the formation of series of discrete films. In one of the fabrication methods7 spatially addressable deposition can be performed that in principle resembles to the earlier described light-directed, spatially addressable parallel chemical synthesis invented by Fodor et al.8 Both methods are based on the use of masks that cover parts of the solid surface at predetermined places. In fabrication of thin film libraries the mask prevents deposition of the vapor on the covered parts of the solid surface on which the films are made.

Figure 6.1. Quaternary masking system.9 Masks (a to d) and positions of the 1024 library components (e)

a b c d d

e

153

The 5 masks (a trough d) of a quaternary masking system are shown in Figure 6.1. The deposition process begins with the use of mask a in position shown in the figure. After the first deposition the mask is rotated by 90o and the deposition is continued. This is followed by two more rotations with deposition after each rotation. Then the deposition is continued by using the remaining 4 masks each with 3 rotations after the depositions. By proper variation of the precursors and their quantity (thickness of their deposited layers) in the deposition cycles a library of 45=1024 discrete thin films is formed and the composition of the films differ in every position (Figure 6.1/c). Such a thin film library can be composed on a 2.4x2.5 cm silicon plate. Composition of a library by application of the masks is a fully combinatorial process and for this reason is fast. Each deposited film is composed of 5 layers that may differ in composition and/or thickness. The multilayer films are postannealed at intermediate temperature (200-500o) to homogenize the composition then heated at elevated temperatures (800-1000o) to promote reaction of the constituents. A parallel approach can also be applied to fabricate discrete film libraries: the constituents of the films are transferred in solution into small wells formatted on plates. The solutions that have different composition or concentration are delivered by automatic fluid dispensers or ink jetting.10 The series of liquid samples are first heated to a moderate temperature to evaporate the solvent then an elevated temperature is applied to bring about the reaction of the constituents and finish the formation of the films.

Figure 6.2. Fabrication of discrete films by dispensing the constituents in solution

In Figure 6.3. a pulsed-laser deposition system is outlined.10 There are three different targets (1, 2 and 3) from which the constituents of the film can be transferred into the substrate (film). As the figure shows the targets can be rotated. The constituents of the target (1) are evaporated by laser beam and deposit into the substrate. Evaporation of the targets 2 and 3 can be executed after 120o and 240o rotations, respectively.

Figure 6.3. Pulsed-laser deposition system with three rotatable targets (1,2,3)

Laser beam

Substrate

1 2 3

Targets 1,2,3

120o

154

In this deposition system the constituents of the films are deposited in layers that need to be homogenized by post-annealing. There are in use other multi-target deposition systems in which deposition from the targets occur simultaneously. In these co-deposition systems a homogenous layer is formed. As already mentioned it is possible to fabricate a multi-component library in a single multi-layer film, too. The composition in each layer of the film is continuously changing along a gradient.11 In Figure 6.4. such deposition system is outlined.

Figure 6.4. Gradient deposition on a triangle in three steps (a, b, c) using three targets (CaCO3, SrCO3, BaCO3).

The substrate is an equilateral triangle shaped LaAlO3 piece (height 2.5 cm). The bottom precursor is TiO2.In the first deposition step (a) the target is CaCO3. A shutter moves at constant speed across the triangle in the direction of the arrow and gradually covers it. The thickness of the deposited CaCO3 layer is continuously increasing (from 0 to 1225 A) and the maximum is at the corner 1. Before SrCO3 deposition in the second step (b) the triangle is rotated anticlockwise by 120o otherwise the process is the same. In the last step (c) BaCO3 is deposited again after 120o rotation. In the final film (d) the composition is smoothly changing from point to point and the maximum thickness of the CaCO3 (1225 A), SrCO3 (1475 A) and BaCO3 (1647 A) layers are at the corners 1, 3 and 2, respectively. The chip is heated for homogenization at 400o for 24 hours for homogenization and at 900o for 1.5 hours for crystallization. 6.1.2. Screening The methods used for screening of inorganic film libraries are very different since very different properties have to be measured. The very large number of different compositions present in a single chip, as well as the small size of the films represent special difficulties. Special screening methods have been developed and are under development in order to face the problems involved in the experiments. The non-destructive physical methods are preferred like optical measurements. For example X-ray microbeam techniques available at synchrotron radiation facilities are used with spot size of 3x20 µm.

1

2 3

2

3 1

3

1 2

1

2 3

Shutter

Shutter

Shutter

CaCO3 SrCO3 BaCO3

Ca

Sr Ba

a b c

120o 120o

d

155

Some measurements are executed in serial mode. These are relatively slow since each film has to be measured separately. Other screening determinations can be carried out in parallel. These methods are much faster since in parallel mode, a large number of films, often the entire library, can be measured simultaneously. This technique can be used, for example, in screening of phosphor libraries. The new film fabrication and screening methods are subjects of intensive research. Appearance of even better and faster methods are expected in both synthesis and screening. But according to same opinions even the use of the existing methods make the materials research about 10,000 times faster than the conventional ones. Detailed characterization of the different screening technologies falls outside the scope of this book. 6.2. Heterogeneous catalysts The heterogeneous catalysts belong to a very important class of materials since they are used in the manufacture of a large number (about 7000) of chemicals and for this reason they significantly contribute to the economy and to our living standards. Catalysts are used in about 60% of chemicals productions.12 Catalysts are complex materials. According to estimations about 50-70 elements of the periodic table can be regarded as suitable components for heterogeneous catalysts. The activity and specificity of catalysts depends not only on the elements they are built from but also on the proportion of the elements and on the conditions of preparation. Despite the invested research efforts in this area we still can not predict how the properties of a catalyst depend on composition. For this reason the discovery of new catalysts can only be accomplished by trial and error like that of other new materials and pharmaceuticals. This is the reason why the combinatorial methods need to be applied in this area. The “multi-sample concept” proposed by Hanak1 and his pioneering work in the 1970s and 1980s were not followed but the advent of the combinatorial methods in the pharmaceutical area2-4 initiated very intensive research for adaptation of the combinatorial methods to the catalyst research. Today the principles of combinatorial approach are already accepted and widely applied.

Figure 6.5. The three kinds of activities in catalyst research

When thinking about preparing new catalysts it has to be taken into account that the possible number of different compositions of the elements is immensely high. It can easily be

Combinatorial catalyst research

Computation methods Information mining

High throughput catalyst preparation

High throughput testing

156

calculated, for example, that if 6 elements are used as catalyst components and each of them in any of 10 different concentrations, the total number of possible compositions is 106, one million and this multiplied if different preparation conditions are applied. For this reason it is absolutely impossible to consider testing all (50-70) elements in a catalyst search. There are too many parameters that can be varied. This can be expressed by stating that the parameter space is usually very high in catalyst research. As a consequence, computational methods need to be applied in order to reduce the number of executable experiments. In combinatorial catalyst research usually three kinds of activities are amalgamated (Figure 6.5.). Catalyst discovery is a multi-step iterative process. It starts with library design that involves data-mining from the literature and considers many variables like precursor materials and their relative concentrations, support materials, mixing conditions, calcination temperatures, the reactor applied in testing and analytical tools. The designed library is then synthesized and tested. In these phases automated processes are usually applied. In a single step only a small fraction of the huge parameter space can be explored. For this reason the experimental data coming out from the first step are usually considered as preliminary results that are the basis of further iterations. The further experiments are guided by computational methods that can make predictions based on the already existing experimental data (Figure 6.6.).

Figure 6.6. Iterative scheme of catalyst library design, fabrication and testing In the informatic platforms used so far following methods, some of them developed in the area of artificial intelligence, have been applied: Artificial neural network13 Genetic algorithm14 Holographic research strategy15 Support vector machines30 Decision trees31

Library design Activity

Information mining

Preparation Testing

157

6.2.1. Preparation and testing of catalyst libraries There are different methods for preparation of catalyst libraries. The fabrication methods depend on the form in which the catalysts are tested. Thin film arrays. Like other solid materials, catalysts can also be prepared and tested in the form of thin film arrays. In their preparation the same or similar methods can be used as those described paragraph 6.1.1. for thin film materials. One approach is the use of vapor deposition techniques: application of sputtering methods and masks. Parallel deposition of the components in dissolved form is also used. The solutions are usually transferred to their place by automatic liquid dispensers then are evaporated. Both kinds of thin films are tested in either calcinated or reduced forms. One analytical method applied in testing thin film libraries is mass spectroscopy. As Figure 6.7. shows, each catalyst film of the library is sequentially heated by a CO2 laser beam to the desired reaction temperature and the reactant gases are transported to the catalyst site by larger diameter probe and the deflected reaction products go through the inner tube to the mass spectrometer for analysis. By moving the probe and heating in x-y directions over the plate (or by moving the plate itself) containing the films, the probe and the laser beam heating can be positioned to every catalyst site. The advantage of this approach is that not only the activity of the catalyst can be tested but its selectivity too, since the concentration of the reaction products can be determined by this method. It is a disadvantage, however, that the determinations can be done only sequentially, one library member at a time. A similar method was applied for analysis of the activity of catalyst powder libraries deposited on addressable positions of a heated substrate.17

Figure 6.7. Testing the activity of thin film catalyst library by analysis of the reaction products by mass spectrometry

A different approach is demonstrated in Figure 6.8. The oxidizing activity of 16 catalysts are tested that are deposited in small wells of a reactor in the format of pellets. The gas submitted to the catalytic process is in contact with all catalysts and the activities are detected using infrared thermography. The active catalysts appear as bright spots. See the explanation in paragraph 5.2.2.1.18

Heating laser beam Catalyst film

Rection gas in

To MS

158

Figure 6.8. Detection by IR thermography of the catalytic activity of Ir and Pd in a library of 16

components.

The advantage of using the IR thermography is that the activities of the library members can be determined in parallel, that is, in a single experiment. This is much faster than the serial determination. The disadvantage is that nothing is known about the products and the selectivity of the catalysts. Microreactors. The traditional devices for testing new catalysts are single laboratory reactors. These single reactors in combinatorial catalyst research are replaced by a parallel array of microreactors. The catalysts to be tested are prepared in series of small vials using automatic liquid dispensing systems. After evaporation of the solvents the residues are calcined, grinded then filled into the microreactors. A possible arrangement of 16 microreactors is demonstrated in Figure 6.9. The reactant gas mixture is distributed among the 16 heated microreactors. A constant gas stream flows through all reactors and finally leaves at the outlet. A probe is sequentially positioned to one of the reactors and direct the reacted gas mixture to the analyzer. The analysis is usually done by mass spectrometry, gas chromatography or by combination of both.

Figure 6.9. Schematic arrangement of 16 microreactors This arrangement is advantageous because both the activity and the selectivity of the catalysts can be determined by MS and/or GC. The effect of other parameters like that of the

Ir

Ti

Fe

Pt

V

Er

Gd Zn

Co Pd

Bi

Rh

Ag

Cr

Cu

Ni

Inlet

Outlet

To MS, GC

Reactors, Side view end view

Heating Positionable probe

159

temperature can also be tested. The disadvantage of this experimental set up is that the analysis can be executed only sequentially that is relatively slow. A different experimental arrangement and application of a different analytical method makes possible the parallel analysis of the products of all reactors (Figure 6.10.).19 The device consists of a bundle of 49 catalyst cartridges each attached by a gold plated nozzle to a 20 cm long analysis tube. The nozzle has at its center a small hole that allows the gases to pass. At the end part of the bundle of the analysis tubes a CaF2 window, transparent to IR radiation, is mounted. The distance from the end of the tubes is only 1 mm. Outside, at the CaF2 window there is a semitransparent mirror that directs the IR radiation into the analysis tubes. The gaseous reaction mixture is fed into the catalyst cartridges through a common inlet. The gas passes through the catalyst cartridges then enters through the nozzles into the analysis tubes and finally leaves the apparatus in combined form at the outlet. The products are analyzed while they are in the analysis tubes. The IR radiation reflected by the gold coated nozzles is simultaneously determined by a focal plane array (FPA) detector.20 The FPA detector contains an array of small detectors. The density is several thousand elements in a few square millimeters and each element can record a full IR spectrum.

Figure 6.10. Schematic view of 49 parallel reactors analyzing the products with IR measurement in parallel mode. a: catalyst cartridge (1 cm), b: analysis cell (20 cm, diameter 4 mm), c: inlet of

gaseous reaction mixture, e: bundle of 49 catalyst cartridges, f: heater, g: bundle of analysis tubes, h: outlet of gaseous products, i: CaF2 window, j: semitransparent mirror, k: IR beam

The parallel microreactor systems can be applied in monolith form, too.21 The catalyst active layer is deposited on the walls of the monolith by a wash-coat procedure. Catalysts on beads. As described by Schunk and his colleagues the catalysts can be deposited on beads and can be tested in a new type of microreactor.22,23 As carriers of catalysts uniform α-Al2O3, γ-Al2O3, SiO2 or TiO2 beads are used. Their diameter is 1 mm and the each bead carries a catalyst of different composition. The catalysts are deposited on the beads by wet impregnation method using automated liquid dispensing. The beads are dried at 80 oC for 16 h and calcinated at 420 oC for 3 h in air.

Gold-coated nozzle

a b

c

d

e g h i

j k f

160

The catalysts carried by the beads are tested in a special “single bead reactor” developed for this purpose. The reactor has two identical parts: the base and top (Figure 6.11.). The wells in the upper and lower part of the microreactor are composed of silicon membranes. The wells are etched into the membranes then the membranes are combined by silicon fusion bonding. The number of wells in a microreactor is 384 or 625. After filling the reactors with beads the upper and lower parts of the reactor are pressed together or are permanently bonded.

Figure 6.11. Beads in the wells of the single bead reactor

Figure 6.12. The one bead microreactor in the flange. a: top view, b: schematic side view

The reactors are mounted into a stainless steel flange (Figure 6.12.). The flange system provides sealing and heating up to 450 oC. It is connected to the reaction gas feeding system and provides a continuous flow of the gas through all individual reactor wells. It also contains a sampling capillary that is sequentially and automatically positioned to the outlet of the reactor wells and transfers the samples to a scanning mass spectrometer for analysis. The analysis time per bead is about 25-80 sec. The practice of preparation of catalyst libraries on inorganic beads offered the possibility of speeding up the process by using the split-mix method.2-4 The first such approach was patented in 200024 and applied for testing the single bead microreactor.25 A Mo-Bi-Co-Fe-Ni library was prepared on 3000 γ-Al2O3 beads using the following solutions: (NH4)6Mo7O24.4H2O,

Top Base

Bead

Inlet

To MS

Positionable sampling capillary

a

b

161

Bi(NO3)3.5H2O, Co(NO3)2.6H2O, Fe(NO3)2.9H2O and Ni(NO3)2.6H2O in molar concentrations of 0.025, 0.075, 0.25, 0.25 and 0.1, respectively. The catalyst precursors were applied in the following order: Bi, Co, Fe, Ni. Before addition of the precursor solutions the beads were split into 4 equal samples placed into small dishes. The precursors were added in four different concentrations: 0, 0.002, 0.1 and 1 weight %. After the impregnation period the solutions were evaporated the four samples of beads were dried at 80 oC for 16 h, calcinated at 400 oC then mixed (Figure 6.13.).

Figure 6.13. The first cycle of the split-mix process

In continuation three similar cycles were executed by adding sequentially CoII, FeII and NiII in the same concentrations. Finally, the library was tested in the one bead microreactor. The split-mix procedure was also used for preparation of catalyst libraries by others.26 6.2.2. Catalyst library design In the field of combinatorial catalysis different approaches are used for library design. Industrial companies, like Symix, Avantium, hte GmbH, are using their own proprietary methods. In academic research the Genetic Algorithm (GA) is widely applied. The combination of GA with Artificial Neural Networks (ANNs) has also been reported. Recently, a new approach, the Holographic Research Strategy (HRS) and its combination with ANNs have been described.

The holographic research strategy (HRS). HRS was developed with intention to make possible the exploration of the huge parameter space in catalyst research and find the best catalyst composition by a reasonable number of experiments. This approach has been developed by a Hungarian group and the strength of this approach has been demonstrated using both hypothetical and real experimental data.15

BiIII 0.0 0.002 0.1 1.0

Drying Drying Drying Drying

Calcination Calcination Calcination Calcination Drying

162

HRS is based on a special arrangement of multi-dimensional data in a 2D space. The compositions of the catalysts are represented in two dimensions by plotting the substrates and their quantities on the two (x,y) axes of a sheet. Before discussing HRS different possibilities for plotting repeating quantities, as shown in Figure 6.14., has to be shown.

0

1

2

3

4

0

1

2

3

4

Figure 6.14. Two possibilities for plotting repeating quantities along an axis The same quantities are arranged in the two plots but in different order. In plot a there are large changes in the quantities along the horizontal axis. In plot b, however, the quantities are changing smoothly in a wave-like manner. Any given quantity differs from its neighbors by only one unit. In HRS this second plotting form is applied. Lets suppose that in the hypothetical experiment six precursors are used represented by A, B, C, D, E and F. Their concentration levels are shown in Table 6.1.

Table 6.1. The concentration levels of the precursors

Level A B C D E F 1 0.0 0.0 0.0 0.0 0.0 1.0 2 0.2 0.2 0.2 0.2 0.2 0.2 3 0.4 0.4 0.5 0.5 4 0.6 1.0

The precursors A, B, C, D, E, F are used in 3, 2, 4, 4, 2 and 3 concentration levels, respectively. So the number of catalyst compositions is 3x2x4x4x2=576. In HRS the different catalyst compositions can be represented in a “holographic” sheet demonstrated in Figure 6.15. Each rectangle corresponds to one of the 576 catalyst compositions. The parallel lines above and left to the axes represent the concentration levels of the precursors incorporated into the catalysts represented by the rectangles below and right to the lines, respectively. The order of precursor plotting is read upward and right to left at horizontal and vertical axes, respectively (Figure 6.15.). As shown in Figure 6.15 moving from left to right along the X-axis the value of A increases in 3 steps to its maximum thus forming a half wave. At the first level of A the level of B increases in the possible two steps to maximum then at the second level of A decreases in two steps to minimum. At the third level of A in additional two steps B increases again to maximum forming altogether one and a half wave. Analogously, the four possible levels of the C are plotted

a b

163

“above” a single fixed level of B so forming three full waves. The full combination of the levels of variables along the X-axis leads to 24 data points along the X-axis. The remaining D, E and F variables are similarly plotted along the Y-axis leading to 24 data points. Eventually all of the 574 catalyst compositions appear in the 2D holographic representation.

Figure 6.15. Holographic representation of catalyst compositions by HRS. The A, B and C precursor concentrations are plotted along the horizontal axis, those of D, E and F along the vertical axis. The different concentrations are represented by the lines parallel with

the axes.

HRS helps to find the best catalytic performance by testing as less as possible number of compositions from the potential 576 ones. The initial step is the determination of the first catalyst library as shown in Figure 6.15. Its members are highlighted by gray rectangles. After determination of the best catalytic performance a special two dimensional transformation is applied. This means that the order of the precursors is changed in plotting. The A, B, C and D, E, F order of the precursor plotting is replaced by C, B, A and F, D, E, respectively (Figure 6.16/a). As a result of the transformation the best composition (hit) has a new neighborhood. In the close vicinity of the hit there are compositions that have not been tested. As the next step of iterations an area of 5x5 rectangles – a so called experimental window - around the black rectangle (showed by the enhanced large rectangle) is assigned for experimental testing. The best catalyst coming out from these experiments is shown by the small enhanced rectangle. The two dimensional transformation is repeated again: orders of C, B, A and F, D, E are replaced by C, B, A and F, E, D, respectively. The black rectangle in Figure 6.16/b shows the new position of the best catalyst. In the next iteration again a 5x5 catalyst composition is tested and the result is represented by the enhanced small rectangle in the figure. The results of two more iterations are seen in Figures 6.16/c and d. The small enhanced rectangle appearing in Figure 6.16/d is supposed to represent the global maximum of the catalyst activities. By applying HRS, the maximum catalyst activity can be determined by testing only a fraction of the full parameter space. However, the algorithm can be stacked in a local optimum if the experimental window is too small. This potential difficulty can be overcome by selecting not only the single hit, but also the best second and third composition. Additionally, the efficiency of the optimization can be enhanced by combined

C B A

F E D

164

application of artificial neural networks and HRS.

Figure 6.16. The holographic presentations after the two dimensional transformations. Black rectangle: location of the best catalyst in the transformed representation, large enhanced rectangle: catalyst compositions selected for the next catalyst generation, small enhanced

rectangle: best experimental catalyst composition within the catalyst generation 6.3. Polymers Polymers are an important class of materials. Their application is so widespread that our life today could not be imagined without them. They are used as structural, packaging and coating materials, they are components of our clothes and they are applied even in microelectronics and nanotechnology. Their properties depend not only on composition but to a high degree on conditions of their processing that is effected by a large number of variables. The combinatorial methods that are introduced and used in this area help to faster determine the influence of the mentioned variables. In this respect the polymer libraries prepared in the form of continuous thin films are very important. The dependence of the properties of the films, like

E D F a

B A C

C B A

F E D d

A B C

b D E F

B C A

E F D c

165

dewetting, phase behavior, surface morphology and crystallization, can be studied by optical means. Application of studies of continuous gradient films instead of conventional approaches makes the research faster, cheaper and more successful. Below, fabrication of some thin film libraries is described.

Fifure 6.17. Principle of the flow coating process generating continuous gradient thickness films.

S: substrate, α: blade angle, G: height of the blade above the substrate (typically from tens of microns to hundreds of microns), H: thickness of the wet film, h: thickness of the dry film. The

substrate is moving in the direction of the white arrow. (Reprinted with permission from C. M. Stafford et al. Rev. Sci. Instrum. 2006, 77, 023908)

Generating thickness gradient libraries. Thickness is an important factor in the behavior of thin films. Figure 6.17. shows a velocity-gradient knife coater that can be applied to prepare coatings and thin films containing continuous thickness gradient.27 . The substrate can be, for example, polished silicon wafer or glass slide. As the figure shows a polymer solution is spread on a substrate under a knife edge by moving the substrate at constant acceleration. The result is a dried film with controllable thickness. The thickness is controlled by the instantaneous velocity of the substrate relative to the blade. Lower velocities generate thinner film, while high acceleration results in a relatively steep gradient. The thickness of the films is varying in one dimension. The thickness at different positions can be determined by using a UV-visible interferometer.28 Composition-gradient libraries. The properties of thin polymer films largely depend on composition. Preparation and examination of composition-gradient thin film libraries makes possible to speed up the determination of composition - property relationship. The composition-gradient libraries are prepared in a process involving three steps illustrated in Figure 6.18.28 The first step is preparation of the gradient (Figure 6.18/a). Before starting preparation the small vial is filled with polymer solution B. After starting preparation two syringe pumps begin to operate. One of them introduces polymer solution A at rate v1 and the other one withdraws polymer solution at rate v2. As a result, the composition in the vial continuously changes. A small amount of the solution is continuously extracted with an automated sample syringe. The composition of the solution entering into the syringe is also continuously changing. At the

S

166

beginning the solution is rich in component B and at the end is rich in polymer A. So at the end the syringe contains a gradient along its length.

Figure 6.18. Preparation of a composition-gradient library from polymers A and B. a: mixing A and B and extracting the gradient, b: deposition of the gradient, c: film spreading

In the second step (Figure 6.18/b) the content of the syringe is deposited on a substrate as a stripe. The composition of the deposited stripe forms a continuous gradient. The third step (Figure 6.18/c) is spreading the composition-gradient stripe on the substrate orthogonal to its direction by using a knife-edge coater. After the solvent evaporates a continuous linear gradient film remains on the substrate. The remaining solvent is removed under vacuum during annealing.

Figure 6.19. Introducing the temperature gradient

A B

B-rich

A-rich

stirrer

a b c

v1 v2

substrate

sample with thickness gradient

heating cooling

temperature gradient

thickness gradient

167

Temperature-gradient libraries. Two dimensional gradients.28,29 The thin polymer films deposited on the substrate can be submitted to annealing at continuously changing temperatures. If the film is a thickness- or composition-gradient library then the direction of the temperature gradient is chosen to be orthogonal to the primary one. This way two dimensional gradient libraries can be formed. One of the two parameters is changing in x the other one in y direction. The film differs from point to point by at least one of the two parameters and so it can be the source of a very large number of experimental data. Figure 6.19. shows the how a thickness-temperature-gradient library can be formed. The direction of the temperature gradient is perpendicular to the direction of the thickness gradient. The device is an aluminum T-gradient stage. It uses a heat source and a heat sink to produce a linear gradient ranging between adjustable end point temperatures (160 to 70 oC over 40 mm). The end point temperatures can be adjusted within the limits of the heater, cooler and the maximum heat flow through the aluminum plate.

Figure 6.20. High-throughput screening by computer controlled microscope (Reprinted with permission from J.C. Meredith et al. Macromolecules 2000, 33, 9747

Copiright (2000) American Chemical Society) High-thoughput screening. The simplest way to study the one or two dimensional gradient libraries is optical microscopy. Figure 6.20. exemplifies this. A digital camera coupled to an optical microscope takes 1024x1024 pixel images and sends them to computer for analysis. The computer also controls the x-y movement of the sample stage over a predetermined grid that divides the sample area into a virtual array of cells. The cells are photographed in serial manner and the magnified images are sent to computer.

168

Figure 6.21. Composite images of a thickness-temperature gradient polystyrene library (Reprinted with permission from J.C. Meredith et al. Macromolecules 2000, 33, 9747, Copiright

(2000) American Chemical Society) As an example, Figure 6.21. shows the result of a dewetting experiment carried out with a thickness-temperature gradient polystyrene library. The thickness range was from 33 to 90 nm. The endpoint temperatures were 135.0 ± 0.5 and 75.0 ± 0.1 oC over 40 mm (gradient 2 oC/mm). Figure 6.22. shows a composite picture. Figure 6.22. shows magnified images of the A, B and D boxed regions of Figure 6.22. These photos illustrate how the structures within the library depend on thickness and temperature.

Figure 6. 22. Close-up images of the boxed areas in Figure 6.22. (Reprinted with permission from J.C. Meredith et al. Macromolecules 2000, 33, 9747, Copiright

(2000) American Chemical Society) The methods demonstrated above represent only a few examples of the numerous combinatorial approaches applied in the field of polymer research. Even these few examples convince the reader about the applicability of the combinatorial methods in this area.

169

References

1. J. J. Hanak J. Mater. Sci. 1970, 5, 964. 2. Á. Furka, F. Sebestyén, M. Asgedom, G. Dibó, In Highlights of Modern Biochemistry,

Proceedings of the 14th International Congress of Biochemistry, VSP. Utrecht, The Netherlands, 1988, Vol. 5, p 47.

3. Á. Furka, F. Sebestyén, M. Asgedom, G. Dibó Proceedings of the 10th International Symposium of Medicinal Chemistry, Budapest, Hungary, 1988, p 288, Abstract P-168.

4. Á. Furka, F. Sebestyén, M. Asgedom, G. Dibó Int. J. Peptide Protein Res. 1991, 37, 487. 5. H. M. Geysen, R. H. Meloen, S. J. Barteling Proc. Natl. Acad. Sci. USA 1984, 81, 3998. 6. X.-D. Xiang, X. Sun, G. Briceno, Y. Lou, K.-A. Wang, H. Chang, W. G. Wallace-

Freedman, S.-W. Chen, and P. G. Schultz Science 1995, 268, 1738. 7. J. Wang, Y. Yoo, C. Gao, I. Takeuchi, X. Sun, H. Chang, X.-D. Xiang, P. G. Schultz

Science 1998, 279, 1712. 8. S. P. A. Fodor, J. L. Read, M. C. Pirrung, L. Stryer, A. T. Lu and D. Solas Science 1991,

251, 767. 9. X.-D. Xiang In I. Sucholeiki (Ed) High Throughput Synthesis, Principles and Practices,

Marcel Decker Inc. 2000, 231. 10. J. D. Hewes, D. Kaiser, A. Karim, E. Amis Combinatorial Chemistry

http://polymers.msel.nist.gov/. 11. H. Chang, X.-D. Xiang In I. Sucholeiki (Ed) High Throughput Synthesis, Principles and

Practices, Marcel Decker Inc. 2000, 251. 12. Selim Senkan Angew. Chem. Int. Ed. 2001, 40, 312. 13. Z. Hou, Q. Dai, X. Wu, G. Chen Appl. Catal. A.: General 1997, 161, 183. 14. V. Nissen Evolutionäre Algoritmen, Deutscher Univeritatsverlag, Bamberg, 1994. 15. L. Végvári, A. Tompos, S. Göbölös, J. L. Margitfalvi Catal. Today 2003, 81, 517. 16. P. Cong, R. D. Doolen, Q.Fan, D. M. Giaquinta, S. Guan, E. W. McFarland, D. M.

Poojary, K. Self, H. W. Turner, W. H. Weinberg Angew. Chem. Int. Ed. 1999, 38, 484. 17. M. Orschel, J. Klein, H. W. Schmidt,W. F. Maier Angew. Chem. Int. Ed. 1999, 38, 2791. 18. F. C. Moates, M. Somani, J. Annamalai, J. T. Richardson, D. Luss, R. C. Willson Ind.

Eng. Chem. Res. 1996, 35, 4801. 19. P. Kubanek, O. Busch, S. Thomson, H. W. Schmidt, F. Schüth* J. Comb. Chem. 2004, 6,

420. 20. C. M. Snively, G. Oskarsdottir, J. Lauterbach Angew. Chem., Int. Ed. Engl. 2001, 40,

3028. 21. M. Lucas, P. Claus Appl. Catal. A.: General 2003, 254, 35. 22. S. A. Shunk, C. Baltes, J. Klein OIL GAS European Magazine 2/2005, 77. 23. T. Zech, G. Bohner, O. Laus, J. Klein Rev. Sci. Instrum. 2005, 76, 062215-1. 24. WO 002002043860A2 25. J. Klein, T. Zech, J. M. Newsam, S. A. Schunk Appl. Catal. A.: General 2003, 254, 121. 26. Y. Sun, B. C. Chan, R. Ramnarayanan, W. M. Leventry, T. E. Mallouk, S. R. Bare, R. R.

Willis J. Comb. Chem., 2002, 4, 569. 27. C. M. Stafford, K. E. Roskov, T. H. Epps III, M. J. Fasolka Rev. Sci. Instrum. 2006, 77,

023908. 28. J. C. Meredith, A. Karim, E. J. Amis MRS Bulletin April 2002. 29. J. C. Meredith, A.P. Smith, A. Karim, E.J. Amis, Macromolecules 2000, 33, 9747. 30. L. A. Baumes, J. M. Serra, P. Serna, A. Corma J. Comb. Chem., 2006, 8, 583.

170

31. L. A. Baumes, M. Moliner, A. Corma, QSAR & Comb. Sci., 2007, 26, 255.

171

7. Computational aspects of library design and synthesis The appearance of combinatorial and HTS methods made the synthesis and screening of millions of compounds a reality. This generated a so huge amount of data that conventional bookkeeping proved unable to handle. In order to overcome this situation new software companies were founded that produced many products to serve the need of combinatorial and HTS methods. Since application of combinatorial chemistry began in pharmaceutical research the computational tools were developed keeping in mind the needs of drug research. Some software are used in library design, others help chemistry and again others proved helpful in data recording, analysis and data retrieval. Drug research is a long and expensive process. The chemical part of the discovery of a drug usually begins after the therapeutic target has been identified. First a lead compound has to be discovered that shows at least a limited effect on the target. Then comes the optimization process when, by modifying the structure of the lead, a more effective compound has to be found and, at the same time, the unwanted side effects has to be reduced to a possible minimum. Although candidate molecules isolated from natural sources are still very important, the synthetic small organic molecules are even more important. Taking into account the high effectiveness of the combinatorial synthetic processes they are the preferred preparation methods that are applied in both the lead discovery and optimization. The best approach to find a lead compound is to synthesize compound libraries and screen them. It is still a big problem, however, what to synthesize. In other words how to design the libraries that are then prepared using the combinatorial methods. In contrary to the peptide libraries that have a limited number of components that can easily be calculated, the potential number of the small organic compounds is very difficult, if not impossible to calculate. Their number is estimated1 to be in the range of 10200. It is very likely that if one could synthesize and test them all one or even more effective molecules could be found for every potential therapeutic target. The problem is that even if all the matter of the visible universe converted into single molecules only a very small fraction of the total number of the possible structures could be accessible. To put it in a different aspect by synthesizing every year 1,000 libraries containing 1 million components each, the synthesis would require 10191 years while our universe is about only 1010 years old. Taking into account the problem outlined above one concludes that the libraries to be prepared need to be carefully designed. A generally accepted approach is that in the lead discovery phase highly diverse libraries are generally prepared. In the lead optimization phase, on the other hand, the preferred library components are structurally similar to the lead molecule. The problem is that neither the molecular diversity nor the structural similarity can be exactly defined. Another problem is that structural similarity in many cases does not mean biological similarity. In many cases a very slight modification in the structure of the lead results in complete loss of the biological activity. In the early phase of the pharmaceutical discovery process usually a high diversity compound library is needed because this kind of library offers a good possibility to find a lead. In the optimization stage, on the other hand, the similarity to the lead of the components is important and not the high diversity of the library.

172

A considerable theoretical effort has been devoted to make the characterization of molecular diversity and similarity computable. Hundreds of indices and descriptors have been introduced for this purpose. The 2D descriptors,2 for example, are based on predefined structure fragments. The presence or absence of these fragments in a molecule is expressed as a binary bitstring. The 3D descriptors3 by using distances between atoms and bond angles reflect the 3 dimensional shapes of molecules, and as such, is expected to better model the biological behavior then the 2D descriptors. The conformational flexibility of the molecules, are, however, source of serious difficulties in applications. Other descriptors like topological indices,4 2D and 3D fingerprints,5 steric fields,6 atom pair fingerprints7,8 and others are also in use. The software developed to help library design apply these descriptors and indices. In library design two main strategies can be distinguished:

1. Reactant based approach 2. Product based approach.

In reactant based approach an optimized subset of building blocks are selected considering primarily the experimental possibilities of the synthesis and not the properties of the products. Using this approach the synthesis usually produces combinatorial libraries. In product based approach, on the other hand, the properties of the produced molecules are taken into account and the reactants are accordingly selected. Typically huge virtual combinatorial libraries are enumerated and then the compound arrays to be synthesized are selected by applying different filters to remove the unwanted components. Since the selection is a cherry picking process the resulting compound array is not a combinatorial type library and can not be efficiently synthesized by the split-mix method. As examples, two applications are mentioned: DiverseSolutions and Selector developed by Tripos. DiverseSolutions assesses the chemical diversity of libraries, selects diverse or representative subsets and compares their diversity. It also identifies those molecular metrics that best distinguish differences between compounds. Selector characterizes, compares and samples sets of compounds. The available descriptors include fingerprints and atom pair distances. Clustering tools identify relationships between compounds based on their similarity. Compound selection includes Tanimoto Dissimilarity and the Reciprocal Nearest Neighbor approach. Drug-likeliness of the library components is again a preferred requirement. Despite that its exact definition is not possible. Physical and chemical properties of the molecules like solubility, hydrophobicity, number of H-bond donor and acceptor centers, reactivity biological relevance and others are also taken into account. Lipinski and his colleagues analyzed the structure of a large number of marketed drugs and summarized their finding in the very important Lipinski’s Rule9 that has to be considered in library design:

Molecular weight > 500 Number of hydrogend bond donors > 5 Number of hydrogen bond acceptors > 10 ClogP >5 or MlogP > 4.15

According to the rule if any two of the above conditions are satisfied it indicates a poor absorption or permeation of the compound.

173

The brain blood barrier penetration is again another property that needs to be considered. Methods of quantitative structure activity relationship (QSAR) including artificial neural networks and a genetic algorithm based approaches are also widely applied. Other software that are used in library design and drug search are: CombiLibMaker, ChemEnlighten, TOPKAT and ChemSpace. CombiLibMaker performs virtual combinatorial chemistry. Libraries can be defined and enumerated with full control of stereochemistry. The generated virtual libraries can be stored in databases for subsequent searching and retrieval. The libraries can be submitted to virtual screening including docking. CombiLibMaker reads and writes all of common 2D and 3D structural database formats. ChemEnlighten is a decision support program for scientists who work with lists of compounds to set priorities for synthesis, screening and purchase, provides access to vital information and analysis tools. Different databases can be searched and standard descriptors can be calculated as molecular weight, hydrogen donors and acceptors, and works with ClogP/CMR to calculate log P, molar refractivity, molecular connectivity, shape and topology metrics. Different subsets can be quickly selected. TOPKAT offers quantitative structure-toxicity relationship (QSTR) models that predict toxicity of a compound solely from its structure. ChemSpace helps to decide which chemistry should be used in the synthesis of a library and which chemistry will most likely result in activity. A typical virtual library contains at least 50 million compounds. They can be screened according their physicochemical properties, novelty, drug-likeness, diversity, therapeutic relevance and synthetic feasibility. There are other kinds of software that help the combinatorial chemist in practical realization of synthesis of libraries. What synthesis route to choose? Are the starting materials available? Where to buy them at reasonable price? These are important questions. Software of MDL Information Systems, Inc. helps to solve these problems. Available Chemical Directory (ACD) is a very important database where the starting materials and reagents are found. There is a list of 435,000 chemicals in the ACD that can be purchased from 680 suppliers. The database is continually updated. Once the list of the starting compounds and reagents is selected ACD helps to identify and locate the commercial sources and side-by-side comparisons can be made concerning purity, quantity and price. By use of MDL ISIS ACD Finder the compounds can be searched by structure, name and formula. Figure 7.1. shows 3-ethyl butylacetate as an example. The ACD also links to Pure Substance Database that provides safety, hazard and regulatory information. The structures can be entered using ISIS Draw. This is a structure drawing program that can be used in construction of the library, too. It can be downloaded free of charge from the home page of MDL. Other software of MDL provides fast access to synthetic methodology information by connecting directly to chemical literature.

CrossFire Beilstein provides access to the most important collection of preparation of organic compounds. The huge amount of data – like those mentioned below - is easily searched by computer.

CrossFire Gmelin preparation of inorganic compounds collected over two centuries. ChemInform Reaction Library contains new and novel methodologies. Current Synthetic Methodology comprises the most innovative and significant new

reactions since 1992.

174

Figure 7.1. Search in Available Compounds Database

175

MDL Reference library of Synthetic Methodology contains functional group transformations, chiral chemistry, metal-mediated transformations and heterocyclic chemistry.

MDL Solid PhaseOrganic Reactions is an important source of new synthetic methodologies using solid support including description of polymer and other solid supports, linkers and protecting groups.

ORGSYN Database contains those general and verified methods that are published in Organic Syntheses series.

Comprehensive Asymmetric Catalysis is a major reference work focusing on reviewing catalytic methods for asymmetric organic syntheses.

Comprehensive Organic Functional Group Transformations is a reference work focusing on construction, introduction and interconversion of functional groups.

Encyclopedia of Reagents of Organic Syntheses is a major reference work on preparation, handling and use of reagents in organic chemistry.

Science of Synthesis is the world’s most comprehensive major reference work covering functional group transformations and syntheses of compound classes.

There are also available software developed for searching pharmacology, safety, metabolism and toxicology information:

MDL Drug Data Report contains current bioactivity findings and newly launched developmental drugs

MDL Comprehensive Medicinal Chemistry contains searchable 3D models plus important biochemical properties including drug class, logP and pKa.

OHS Hazard Communication contains full-service tools for employee safety etc. MDL Metabolite Database is the world’s largest and most comprehensive xenobiotic

transformations compiled from literature. MDL Toxicity Database contains the complete content of the Registry of Toxic Effects of

Chemical Substances database. PubChem, a freely available component of National Institute Health's Molecular Libraries Roadmap Initiative (http://pubchem.ncbi.nlm.nih.gov/) also provides a wide range of information on small molecules: structure-activity analysis, structure clustering tools, search of unique chemical structures using names, synonyms or keywords. Links to available biological property information are provided for each compound. Among others, PubChem also provides a fast chemical structure similarity search tool. The literature published in the field of combinatorial chemistry is very important for the combinatorial chemist. A very useful compilation of papers and books published from the beginnings through 2003 can be found at http://www.5z.com/divinfo/. Software for storage and retrieval of compound libraries are also important for the practicing combinatorial chemists. Software developed to operate on automatic machines is exemplified by Odyssey of RoboDesign (now Rigaku).

176

Odyssey is a high-throughput sample storage and retrieval system. The system is equipped with an identification system and barcode tracking, enabling programmed access to each microwell plate, and can process over 400,000 plates a year (Figure 7.2.).

Figure 7.2. the Odyssey storage and retrieval system (photo: www.rigaku.com)

An LCD touch-screen provides an intuitive user interface. The systems are fully automatic and include comprehensive safety features assuring reliable and safe operation. Data is accessed using a customized database. The Odyssey is available in 3 configurations suitable for storage and retrieval of 2,500, 5,000 and 10,000 plates. 7.1. Software companies The companies listed below are engaged in developing software and commercialize such products. Beside the name of the companies the addresses of their home pages are also indicated. Aber Genomic Computing (www.abergc.com). It is an informatics company based in Wales, UK. AberGC provides novel data mining, scheduling and predictive modeling solutions based upon evolutionary computing, machine learning and other supervised learning techniques. Their new product, gmax-bio , is the first commercial package to fully utilize these techniques and is designed for all aspects of drug discovery research. Gmax-bio is a novel informatics package based on genomic computing techniques which uniquely utilize Darwinian methods of natural selection to evolve mathematical algorithms to rapidly solve complex data-mining and predictive modeling problems. Accelrys (www.accelrys.com). It is a leading provider of software for biologists, chemists, and materials scientists. Covers computation, simulation, and the management and mining of scientific data. Afferent Systems (www.afferent.com).

177

It offers an integrated solution for combinatorial chemistry informatics, including instrument control and product data generation, access, and storage, all optionally using an Oracle-based enterprise-wide database. Anacapa (www.anacapagrp.com). It specializes in software development for process automation. Proficiencies include customization, documentation and legacy porting. Anacapa Group, Inc. also specializes in product marking and handling solutions for pharmaceutical manufacturing and research applications. They are experts in barcoding and data capture solutions for tubes and vials, microtiter plates/trays, syringes, ampules and other laboratory vessels. Automation Developers (www.automationdevelopers.com). It provides a wide range of custom software solutions. From start in the field of laboratory automation for the pharmaceutical and analytical chemistry industry, to integrated Microsoft office solutions for business. Automation Partnership (www.automationpartnership.com). The offered systems help to reduce drug discovery timelines, resolve process bottlenecks and are in use at major pharmaceutical companies worldwide including sample management software and inventory data handling at any scale of operation. HomeBase is a narrow aisle based automated system for management of sample libraries. BergenShaw International (www.bergenshaw.com). Its Focus software product enables high throughput laboratories to rapidly accelerate their response to change by quickly identifying those process factors and combination of factors associated with yield loss and yield gain. Bioreason (www.bioreason.com). It is in the business of developing proprietary data mining tools and applications for chemical and biological information. Currently is focused on developing data mining applications for drug discovery from high throughput screening (HTS) data. The software systems are developed by combining existing and novel data mining techniques with chemical information and molecular modeling techniques. Bioreason will work with the data from pharmaceutical and biotechnology firms to discover and optimize drug leads in their HTS data. Chemical Computing Group (http://www.chemcomp.com/) Among the solutions offered by the company are focused combinatorial library design, diverse combinatorial library design and combinatorial library enumeration. Other software for calculation electrostatic maps, probabilistic contact potentials, ligand-receptor docking, multi-fragment search and molecular surface & maps are also available. Columbus Molecular Software (www.columbus-molecular.com). The company was founded in 1997. Develops and markets decision support tools for use by life scientists engaged in drug discovery. Helps organizations to effectively extract and visualize the valuable knowledge contained in research data.

178

CombiChem, Inc. (www.combichem.com). It engages in collaborations and partnerships to accelerate the discovery process for pharmaceutical, biotechnology and agrochemical organizations. CCI's proprietary and unique discovery approach is a platform technology which integrates proprietary molecular design technology and rapid synthesis with chemistry expertise. dataBasyx (www.databasyx-inc.com). The company is a laboratory instrumentation tracking, service and supports application. Datect (www.datatect.com). By data validation and data processing technology the company increases the automation of scientific data analysis by alleviating the need for time consuming manual examination of all data and results. Datect’s technology also standardizes data validation and data analysis decision-making, eliminating any potential variability in individual judgment. DoubleTwist, Inc. (www.doubletwist.com). The company is a leading provider of genomic information and bioinformatics analysis technologies. It provides research environments that leverage information technology and the World Wide Web to simplify and accelerate genomic research. EMAX Solution Partners (www.emax.com). It integrates chemical information systems to speed productivity and compliance for major corporations. The company specializes in solutions for both research and operations. In the competition to find breakthrough products, discovery research groups generate huge data stores through the use of combinatorial chemistry and high throughput screening. OPTIMA from EMAX® can accelerate the drug discovery process by integrating proprietary and commercial substance information into a software infrastructure for rapid access. The complete OPTIMA solution is an advantage in the race to increase promising drug leads and feed the new product pipeline. Galactic (www.galactic.com/galactic/index). The company has been dedicated to designing and developing state-of-the-art scientific software for the spectroscopy and chromatography communities. It offers open architecture software solutions compatible with virtually any laboratory instrument. The unique approach provides a single software platform to integrate all laboratory data in one common package for archival storage, data viewing, processing, plotting, and management. GeneData (www.genedata.com). The company specializes in the computational analysis of genomes, transcriptomes, proteomes, and metabolomes as well as compound libraries, and offers a network of communicating software modules, each of which addresses a critical step in the product development cycle of life science companies. For the analysis of high-throughput screening (HTS) data the company provides a modular and scaleable software solution, GeneData ScreenerTM, that improves the understanding of biological activity of compounds,

179

geneticXchange (www.genetixchange.com). It is a software product company that produces the K1 System Data Integration Middleware Platform for any Biotech needing a solution to the biological data integration challenge. Gensym Corporation (www.gensym.com). The company is a leading supplier of software products and services for intelligent real-time systems that help organizations manage and optimize complex dynamic operations. Common applications include quality management, process optimization, dynamic scheduling, network fault management, energy and environmental management, and abnormal situation management. IBM (www-3.ibm.com/solutions/lifesciences/solutions.html). The company offers technology infrastructure in high-performance computing, data integration, knowledge management, storage, e-business, and information services. Today, IBM systems include the most advanced storage management; and a world-renowned computational biology center. IDBS (www.idbs.co.uk). Its specialized applications are used to acquire, manage, integrate and visualize chemical and biological data ranging from the large amounts of data generated in High Throughput Screening and combi-chem programs, through multiple IC50 determinations and profiling, to the complex experimental protocols of toxicology studies. Labtronics (www.labtronics.com). The company is recognized for laboratory automation and instrument interfacing. Labtronics has an Innovative Software Solution. LabVantage Solutions (www.labvantage.com). The company is a provider of state-of-the-art software, implementation services, and consulting to leading discovery-oriented, conventional research, and quality control laboratories. It offers configurable, industry-specific solutions for in a variety of industries including high throughput screening, genomics, proteomics, pharmaceuticals, oil and gas, process chemicals, food and beverage, environmental, and forensics. Managed Ventures (www.menagedventures.com). The company has developed custom application components in Java for High Throughput Screening (HTS), compound registration, inventory and proteomics. Services include the implementation of solutions for drug discovery using pre-built components to rapidly deliver working web-based applications. HTS and other informatics applications have been integrated for Managed Ventures clients in less than 8 weeks using open source Java, XML, SOAP and any JDBC-compliant database (Oracle, DB2, SQL Server, mySQL). MatriCal (http://www.matrical.com/). The company specializes in microwell plates and automated sample management and storage solutions. The MatriStore™ is a compact, economical, climate controlled compound management system that supports multiple sample formats, including 96 and 384 mini-tubes and 96 to 1536 microplates, and others. A storage capacity from 750k to 40 million samples with automatic sample retrieval in plate or individual sample format

180

MDL Information Systems Inc. (www.mdli.com). MDL is the leading provider of integrated scientific information management systems, databases, and services used worldwide in pharmaceutical, chemical, agrochemical, and biotechnology research and development, and in other industries that use chemical products. NuGenesis® Scientific Data Management System (www.nugenesis.com). It is an application-independent software and database platform that automatically creates a common electronic format for laboratory data and scientific information that can be unified, managed and shared throughout the enterprise. NuGenesis Technologies also offers comprehensive consulting, support and training programs. Odysis (www.odysis.com). It supplies scheduling software technologies and services for building more reliable, productive and flexible automated systems for a broad range of industrial, scientific and commercial applications. Pharma Algorithms (http://ap-algorithms.com). The company provides combi-chem and HTS strategies, a comprehensive QSAR approach to resolve and accelerate the process of data evaluation contributing to early lead optimization. The software introduces the unique capability of developing and building computational algorithms in-house utilizing comprehensive yet flexible statistical and fragmental approaches. Process Analysis & Automation (www.paa.co.uk). Their software has been written to control the equipment for the National DNA Database at the Forensic Science Service in the UK. OVERLORD is available with over 100 drivers and with an installed user base of 100 systems worldwide. OVERLORD is the independent software control system of choice. It also has robot drivers for Hudson Plate Crane, Zymark Twister I & II, Hamilton SWAP/Labsystems Relay, Sands Technology and Mitsubishi robots and pipetting workstation drivers for Hamilton, PE Life Sciences (Packard), Tecan and Qiagen. OVERLORD Developer is available for in-house developers, and incorporates a full licensed version of VBA (as in Microsoft EXCEL, Word and Access) for the user to write the extensions required for their application. Pangea Systems, Inc. (www.pangeasystems.com). The company is a provider of software for advanced bioinformatics, the application of information technology to life science research and development. It provides unique computational tools, an open computing environment, and value-added professional services that integrate the collection, organization and analysis of biological and biochemical information. Partek Pro 2000 (www.partek.com). It is a comprehensive data visualization and pattern recognition system that is being widely adopted in the market. This is due to unique combination of interactive visualization and strong statistical, neural, and other numerical analysis techniques. It is also being used in cheminformatics, combinatorial chemistry, high-throughput screening, and in analysis of clinical data.

181

Prelude Computer Solutions, Inc. (www.preludenic.com). The company provides a wide variety of consulting services for regulated industry. Prelude specializes in 21 CFR Part 11 compliant document management, including integration with LIMS systems and electronic publishing systems. Additionally, Prelude's business partners include IBM, Lotus, Microsoft, and CDC Solutions. QUMAS (www.qumas.com). QUMAS is dedicated to the design and development of Enterprise Compliance Management Solutions for companies in regulated industries in the pharmaceutical, medical device, biotechnology, and contract research industry sectors. QUMAS enterprise compliance solutions offer immediate payback through; pre-packaged compliance with FDA, EMEA and other international regulations in particular 21 CFR part 11 (electronic signature and records), rapid deployment, validation, user training, and by automating critical document processes to dramatically reduce document cycle times. REMP (http://www.remp.com) REMP, a Tecan Group company, develops and produces devices, consumables, software and fully automated sample processing and storage systems mainly used in research and development ReTiSoft (www.retisoft.ca). The company is a software research and development enterprise. Products include a software framework Genera which simplifies instrument integration, scheduling software Supra, 3D simulation viewer SimView, and instrument testers Clones for the laboratory automation market. Rigaku Corporation (www.rigaku.com) The company is engaged in analytical and industrial instrumentation, automation and software production. Spotfire, Inc. (www.spotfire.com). It provides software solutions that empower scientists and engineers-and their enterprises-to make decisions in eTime that get products to market first. The Web-based offerings are used by life- and material-sciences companies worldwide for such activities as high-throughput screening, genomics, lead optimization, combinatorial chemistry, formulation development, and bioinformatics. Structural Bioinformatics Inc. (www.strubix.com). The company is a world leader in proteomics-driven drug discovery – the large-scale generation and use of protein structural information to accelerate the discovery and optimization processes. Symyx (www.symyx.com). The company is developing high-speed combinatorial technologies for the discovery of new materials. The Renaissance (TM) software and database platform provides tools to enable the implementation of complete workflows for high throughput design, synthesis and screening of materials.

182

System Services Inc. (www.sysservices.com). The company’s Labstar is a comprehensive PC based laboratory automation software package designed to simplify data handling needs. The package includes VisualMicroplate graphical display, robust file import/export, graphing, calculations, reporting, and data sorting. Laboratory data from HTS machines and other sources can be easily managed to help detect anomalies, compare data visually, identify errors and sort large volumes of information. TAL Technologies (www.taltech.com). It is committed to providing quality software tools to simplify data acquisition and bar code data collection. Titian Software (www.titian.co.uk). The company specializes in sample management software for the life sciences. The software tracks sample inventory, manages workflows and seamlessly integrates with laboratory workstations. Tripos (www.tripos.com) Tripos is a software company that offers a wide variety of products that help pharmaceutical research. Molecular modeling, ligand and receptor based design, library design, bio- and cheminformatics are the main areas area of their activities. Viaken (www.viaken.com). Viaken is an application service provider (ASP) that provides, manages and supports Bioinformatics applications for the small to mid-size Biotechnology, Pharmaceutical and Agriculture company. Viaken offers complete application solutions in the areas of Genome Informatics, Chem Informatics and Pharmaco Informatics. The Viaken solution includes architecture design, implementation, secure hosting, network infrastructure, 24x7 user support, application and server management. Viaken's turn-key solutions, called "Technology Templates TM" * fulfill the research and development needs of its customers by providing solutions that can be rapidly implemented to meet common R&D IT needs. Additionally, Viaken can design and implement custom solutions to meet any of its customers' IT needs. References

1. A.W. Czarnik Org. Chem. 1995, 8, 13. 2. J. M. Barnard J. Chem. Inf. Comput. Sci. 1993, 33, 532. 3. G. Moreau P. Broto Nouv. J. Chem. 1980, 4, 359. 4. L. B. Kier, L. H. Hall In Molecular Connectivity and DrugRresearch, Academic Press

1976. 5. Software developed at Tripos. Inc. 6. R. D. Cramer, D. E. Patterson, J. E. Bunce J. Am. Chem. Soc. 1988, 110, 5959. 7. R. E. Carhart, D. H. Smith, R. J. Venkataraghavan J. Inf. Comput. Sci. 1985, 25, 64. 8. R. P. Sheridan, R. B. Nachbar, B. L. Bush, J. Comp. Aided. Mol. Design 1994, 8, 323. 9. A. Lipinski, E. Lombardo, B. W. Dominy, P. Feeney Adv. Drug. Delivery Rev. 1997, 23,

3.

Index 223 Sample Changer, 46 A ACT 357, 67 Analyst® GT, 123 anchors, 22 Apex 396, 43 Available Compounds Database, 173-174 B Benz, 1 C catalysts on beads, 159 ChemEnlighten, 173 ChemInform Reaction Library, 173 ChemSpace, 173 Cohen, B., 79 color encoding, 91 CombiLibMaker, 173 Comprehensive Asymmetric Catalysis, 175 Comprehensive Medicinal Chemistry, 175 Comprehensive Organic Functional Group Transformations, 175 CrossFire Beilstein, 173 CrossFire Gmelin, 173 Current Synthetic Methodology, 173 D De Witt, S. H., 33 deconvolution, 60, 124-138 DiverseSolutions, 172 diversity descriptors, 172

diversity molecular, 171 Drug Data Report, 175 drug-likeliness, 172 E Einstein, 2 electrophoretic map, 66 encoding, 61-63 by position in space, 92 optical, 89 by radiofrequency, 115 Encyclopedia of Reagents of Organic Syntheses, 175 ExplorerPLS system, 35 F fabrication of films by evaporating solutions, 153 of polymer films, 164 of thin films, 151-154 fluorous separations, 40 Fodor, S. P. A., 80, 81, 84, 152 Ford, H., 1 Frank, R., 32, 36, 37 Furka, Á., 5, 13, 27, 55, 92, 93, 100, 104, 130 G Geysen, M., 31, 82, 117, 124 gradient deposition, 154 H Hanak, J. J., 151, 155

184

high throughput screening, 122-123 holographic research strategy, 161-164 Houghten, R. A., 37, 124 I ISIS Draw, 173 iteration method, 124-129, 139-140 J Janda, K. D., 124 K Kéri, Gy., v L LabMate, 33 Levassor, 1 library amino acid tester tester application, 143 amino acid tester tester preparation, 141-143 amino acid tester, 135-137 benzimidazole, 110-115 catalyst prepared by split-mix, 161 catalyst, 156-164 cherry picked, 104-110 combinatorial dynamic, 138 composition-gradient, 165 design product based approach, 172 design reactant based approach, 172 design, 171-176 dynamic combinatorial, 138 inorganic, 151-155 occurrence, 144 omission application of, 140-141 omission, 133-135, 140-141 organic, 61-63

partial, 63-80 piperazine 2-carboxamide, 115-117 polymer, 164-168 library temperature-gradient, 166 thickness gradient, 164-165 unusual, 79 virtual, 66, 104-108, 172 phage display, 86 thin film, 151-155 light directed synthesis, 84-86 linker, 20 Lipinski, A., 172 Lipinski’s Rule, 172 M M 384 Ultra HTS Synthesizer, 44 MALDILC™ System, 47 manual sorting, 95-97 manufacturers, 48-52 Margitfalvi, J. L., v masking, 152-153 materials, 151-155, 164-168 inorganic, 151-155 Meredith, J.C., 167, 168 Merrifield, R. B., 4, 15, 17, 18, 20, 21, 26, 42, 55, 82, 83 Metabolite Database, 175 Meyers, H. V., 33 microreactors, 158-160 microwave heating, 34 multicomponent reaction, 37-38 Multiple Probe 215 Liquid Handler, 46 O Ohlmeyer, 63 OHS Hazard Communication, 175 Olds, Ransome Eli, 1 one bead-one product, 1

185

ORGSYN Database P Panhard, 1 Parallel evaporation module, 43 PEG, 19 Peugeot, 1 Pinilla, S. E, 130 planning experiments, 71-75 polymers, 164-168 positional scanning, 129-133 application, 145 preparation of catalyst libraries, 157 protecting group, 22 Alloc, 114-117 BOC, 23 Fmoc, 32 nitro, 25 Nvoc, 84 trityl, 24 Z, 23-24 PubChem, 175 pulsed-laser deposition, 153 Pure Substance Database, 173 Q Quad 3+ system, 47 R Reference library of Synthetic Methodology, 175 resin hydroxymethyl, 21 Merrifield, 20-21 Rink amide, 21 trityl chloride, 20 Wang, 21 Tentagel, 19

S SAGIAN™ Core Systems, 123 Saneii, H., v scavenger, 27 Science of Synthesis, 175 screening by affinity chromatography, 138 catalyst using MS, 157, 158 catalysts by GC, 158 catalysts using IR, 159 screening inorganic films, 154-155 methods, 121-138 of polymer films, 166-168 tethered libraries, 146-148 tethered peptide libraries, 148-149 using IR thermography, 147, 157-158 using size exclusion, 138 Selector, 172 similarity, molecular, 171 Skiena, S., 79 Smillie, L. B., 2 Smith, J., 91 software companies, 176-182 solid phase reagents, 40 synthesis, 15-18 Solid PhaseOrganic Reactions Solution parallel synthesizer solution phase synthesis Sophas HTC sorter crown, manual, 95 IRORI automatic, 91 IRORI manual, 90 lantern, manual, 95 sorting directed, 89, 115-117 parallel, 103 semi-parallel, 97-102 split-mix synthesis, 55-61, 161

186

Stafford, C. M., 164 Stille coupling, 40 string synthesis, 93-104 sub-library, 76-79 preparation, 143-145 support solid surface, 20, 84-86, 89 solid, 18-19 soluble, 83 Syncore Reactor, 34 SynPhase crowns, 32, 91, 94, 95, 100 SynPhase lanterns, 20, 91,94 synthesis binary, 80-82 synthesis dendrimer supported, 39 efficiency, 56-57 encore, 91-93 guiding tables, 108 of cherry picked libraries, 104-110 parallel, 30-37 solid phase, 15-18 split-mix, 55-61, 161 spot, 32-33 string, 93-104 tea-bag, 37 with amino acid mixtures, 82-83 synthesizer automatic, 42-45, 66-71 manual, 33-34, 64 system Odyssey, storage and retrieval, 175-176 pulsed-laser deposition, 153 quaternary masking, 152-153 T Takátsy, Gy., 29, 31, 122 Tanimoto, 172 thin film catalyst arrays, 157 TOPKAT, 173 Toxicity Database, 175

Two dimensional gradient films, 166 two libraries on one support, 117-118 U Ugi, I., 37, 38 V vapor deposition technique, 152 X Xiang, X. -D., 151 XP-1500 Plus system, 35