Dissertation for the Degree of Doctor of Philosophy (Faculty of Medicine) in MedicalGenetics presented at Uppsala University in 2002

ABSTRACT

Mendel-Hartvig, M. 2002. Padlock probes and rolling circle amplification. New possibilitiesfor sensitive gene detection. Acta Universitatis Upsaliensis. Comprehensive Summaries ofUppsala Dissertations from the Faculty of Medicine 1175. 41pp. Uppsala. ISBN 91-554-5383-x

A series of novel methods for detection of known sequence variants in DNA, in particularsingle nucleotide polymorphism, using padlock probes and rolling circle replication arepresented. DNA probes that can be circularized – padlock probes – are ideal for rolling circlereplication. Circularized, but not unreacted probes, can generate powerful signal amplificationby allowing the reacted probes to template a rolling circle replication (RCR) reaction.However, when hybridized and ligated to a target DNA molecule with no nearby ends, theprobes are bound to the target sequence, inhibiting the RCR reaction is. This problem can besolved by generating a branched DNA probe with two 3’ arms such that the probes may becircularized while leaving the second 3’ arm as a primer for the RCR reaction. We describehow T4 DNA ligase can be used for efficient construction of DNA molecules having one 5’end but two distinct 3’ ends that extend from the 2’ and 3’ carbons of an internal nucleotide.An even stronger approach to circumvent the topological problem that can inhibit RCR is torestriction digest the template downstream of the padlock recognition site. By using Phi 29DNA polymerase with efficient 3’ exonuclease and strand displacement activity, the templatestrand can then be used to prime the RCR reaction. The amplified molecule is contiguous withthe target DNA, generating an anchored localized signal. The kinetics of the reaction wasinvestigated by following the reaction in real-time using molecular beacon probes. LocalizedRCR signal were obtained on DNA arrays, allowing detection of as little as 104-105 spottedmolecules, of either single- or double-stranded M13 DNA, in a model experiment. We havealso established a serial rolling circle amplification procedure. By converting rolling circleproducts to a second and even third generation of padlock probes the signal was amplifiedthousand-fold per generation. This procedure provides sufficient sensitivity for detection ofsingle-copy gene sequences in 50 ng of human genomic DNA, and large numbers of probeswere amplified in parallel with excellent quantitative resolution.

Key words: Padlock probes, RCR, in situ hybridization, branched DNA

Maritha Mendel-Hartvig, Department of Genetics and Pathology, Section of MolecularMedicin, Rudbeck Laboratory, Uppsala University, SE-751 85 Uppsala, Sweden

© Maritha Mendel-Hartvig 2002

ISSN 0282-7476ISBN 91-554-5383-xPrinted in Sweden by Fyris-Tryck AB, Uppsala 2002

MAIN REFERENCES

This thesis is based on following papers, which are referred to in the text by their Romannumerals:

I. Signal amplification of padlock probes by rolling circle replicationJohan Banér, Mats Nilsson, Maritha Mendel-Hartvig, Ulf LandegrenNucleic Acids Research (1998) 26: 5073-5078

II. Ligase-mediated construction of branched DNA: A novel DNA joining activitycatalyzed by T4 DNA ligase

Maritha Mendel-Hartvig*, Anil Kumar*, Jonas Jarvius, and Ulf Landegren

To be submitted to Nucleic Acids Research

III. Target-initiated rolling circle replication of padlock probes as a mechanism forlocalized mutation detection

Maritha Mendel-Hartvig, Johan Banér, Jonas Jarvius, Mats Nilsson, Ulf LandegrenManuscript

IV. PCR-independent multiplexed genotyping and quantitative expression analysis bya serial circle amplification procedureFredrik Dahl, Maritha Mendel-Hartvig, Johan Banér, Mats Gullberg, Ulf Landegren,Mats Nilsson.Manuscript

*Authors contributed equally to the work.

In addition, this thesis includes data not so far published.Reprints were made with permission from the publisher.

TABLE OF CONTENTS

1.INTRODUCTION 1

1.1 THE HUMAN GENOME – ARE WE ALL ALIKE? 1

1.2 SNPs - WHAT CAN THEY BE USED FOR? 3

1.3 METHODS TO DISTINGUISH KNOWN SEQUENCE VARIANTS 3

Important features to consider; specificity, selectivity, sensitivity 3

All those methods 3Restriction enzyme cleavage based methods 3Hybridization based methods 4Ligation based methods 4Polymerization based methods 5Invasive cleavage based method 5Methods for localized detection in situ 5

2. INTRODUCTION TO PRESENT INVESTIGATION 7

2.1 PADLOCK PROBES 8

Synthesis of padlock probes 8

Applications of circularized padlock probes 8

2.2 LIGATION REACTION 9

2.3 ROLLING CIRCLE REPLICATION 9

Rolling circle replication mechanism 9

Applications of rolling circle replication 10

3. PRESENT INVESTIGATION 11

I. Signal amplification of padlock probes by rolling circle replication 11

II. Ligase-mediated construction of branched DNA: A novel DNA 12

joining activity catalyzed by T4 DNA ligase

III. Target-initiated rolling circle replication of padlock probes as a mechanism 13

for localized mutation detection

Recent developments 16

How to make a break 16

Detection of allelic variants in BRCA1 gene 18

in interphase cell nuclei in situ

IV. Multiplexed genotyping and quantitative expression analysis by 22a serial circle amplification procedure

4. DISCUSSIONS AND CONCLUDING REMARKS 24

Padlock probes - a promising reagent in haplotyping 24

Padlock probes for localized detection on microarray 24

5. ACKNOWLEDGEMENTS 26

6. REFERENCES 28

1

1. INTRODUCTION

Nothing is more natural for me, when writing glimpses of the history of genetics, than tobegin with Gregor Mendel, by many seen as the father of genetics. He proposed the conceptof genes in 1865 (although he called them factors), and formulated his laws of heredity. Thehereditary information, he reported, is passed from parent to offspring as discrete factors anddifferent factors are responsible for distinct traits. If an individual inherits two factors for thesame trait with conflicting instructions, then one may dominate the other, but the other factorwould persist in latent, recessive, form. In 1902 and 1910 respectively, Walter Sutton linkedthe chromosomes to Mendelian heredity and Thomas Morgan and coworkers showed thatsome trait-determining genes are located near each other on specific chromosomes. OswaldAvery, Colin McLeod, and Maclyn McCarthy proved that DNA, not protein, is carrier ofhereditary information in most living organisms (1944), and Erwin Chargaff (late 1940s)measured the amounts of the four nucleic bases in DNA from different sources, anddiscovered that the amount of cytosine always equaled that of guanine, while the amount ofadenine equaled that of thymine. Chargaff’s finding was one important piece of informationfor constructing a model of the structure of DNA, another was Rosalind Franklin’s x-raydiffraction data for DNA (1953). With these clues on hand, James Watson and Francis Crickpublished their discovery of the double-helix structure of DNA as a letter in Nature andsuggested a possible copying mechanism for the genetic material (1). The definition of theDNA structure set off a rapid development of techniques to investigate nucleic acids.

With the invention of DNA sequencing using chain-terminating nucleotide analogsthe order of nucleotides in DNA sequences could be determined (2). This has led on tosequencing of whole genomes from a wide variety of organisms, and recently also man Homosapiens (3,4). Sequencing of the human genome offers invaluable insights into the geneticbasis of human physiology and disease.

1.1 THE HUMAN GENOME – ARE WE ALL ALIKE?

The most common form of sequence variation among individuals is single nucleotidepolymorphisms (SNPs) accounting for about 90% of the variation (5). SNPs are defined asallelic variants in single base pair positions in genomic DNA, where the least frequent variantoccurs at a frequency of at least 1% in a population. By comparing DNA sequences fromoverlapping clones sequenced in course of the human genome project, one finds SNPs at therate of approximately 1 per 1000 – 2000 basepairs (6). When the draft human genomesequence was published in Februari 2001, 1.42 million SNPs were also reported and suchsequence variants are collected in several public databases including NCBI(http://www.ncbi.nml.nih.gov/SNP/), ENSEMBL (http://www.ensembl.org) and HGBasedatabase (http://hgbase.cgr.ki.se/). Great efforts are made to unravel the sequence variationswithin the human genome and between different populations, and the number of SNPs indatabases increases. SNPs are distributed at different densities across the human genome e.g.,at lower frequency in protein coding versus noncoding DNA and higher towards chromosomeends.

2

1.2 SNPs - WHAT CAN THEY BE USED FOR?

Single nucleotides polymorphisms are the result of mutations and they are probably mostoften devoid of phenotypic consequences. Sequence variants that alter the structure of theprotein it codes for or that occur in the coding regions of a gene, are common in Mendelianinherited diseases such as cystic fibrosis (CF) and such variants may be screened for toidentify disease susceptibility (7). Mutations leading to monogenic diseases are usually rare ina population due to the process of natural selection (8), with the exception of some diseaseswith recessive inheritance, for instance CF (9). Human Medelian inherited diseases arecatalogued in the OMIM (Online Medelian Inheritance in Man) database(http://www.ncbi.nlm.nih.gov/omim/). In cancer, diagnostic sequence variants can be used todemonstrate genetic predisposition to disorder like breast and ovarian cancer. Anotherpromising area for genetic diagnostics is pharmacogenetics, where SNPs altering the primarystructure of a protein involved in drug metabolism may be screened for (10). Geneticheterogeneity is now recognized as a significant source of individual differences in drugresponses. These SNPs can provide information about our individual responses to drugs andthere is great hope from the pharmaceutical industry that by demonstrating sequences variantsinvolved in drug metabolism new targets and new tools will be found to enhance drugspecificity (11).

Genes for Mendelian disorders have successfully been mapped in genetic linkagestudies of pedigrees, but for complex diseases, diseases caused by multiple disease genes ofmodest effect in combination with environmental effects, other approaches may be necessary.An alternative approach is to locate genomic regions harboring genes contributing to geneticrisk of disease by association studies using SNPs as markers (6,12). These markers areabundant, widely dispersed, and easy to score, thereby making excellent markers in thegenome (13). Both functional and neutral SNPs, having no phenotypic influence, can be usedto identify gene variants that influence susceptibility in complex diseases by utilizing linkagedisequilibrium that can be expected to be strong between the mutation and the marker overshort genetic distances. SNPs are not only useful as markers in disease mapping, they are alsowidely used in population genetics and evolutionary studies (14).

By association studies, genetic and phenotypic variations are compared in largepopulation samples, such as case-control studies where a group suffering from a condition iscompared to a group of healthy controls. These studies assume a correlation between thedisease mutation and one or more marker SNPs. Recent results suggest that SNPs are co-inherited as discrete blocks of sequences, roughly 5 – 100 kb each, separated by hot spots ofrecombination, generating a few high-frequency haplotypes (15,16). Haplotypes are sets ofgenetic markers that are close enough on a chromosome to be inherited together and they arein general more informative and could simplify the search for genes contributing to complexdisease. As a consequence a HapMap project has been initiated with the aim to produce areference map of haplotypes region across the human genome.

3

1.3 METHODS TO DISTINGUISH KNOWN SEQUENCE VARIANTS

Important features to consider; specificity, selectivity, sensitivity

There are different approaches to distinguish known allelic variants, but some features are incommon. First of all the method must be sufficiently specific to identify the relevant targetsequence in the genome. The hybridization of an oligonucleotide of approximately 20nucleotides in general has not proven sufficiently specific to detect unique sequences incomplex genomes (17). Higher specificity may be achieved by using longer hybridizationprobes, and such probes are used in genetic library screening, Southern blots (18), and in situhybridization (19). Another means to achieve higher specificity is to require hybridization bytwo or more independent short probes to a unique sequence in order to obtain a combinatorialeffect. This is utilized in, e.g., the polymerase chain reaction (PCR) (20,21), the ligasereaction (OLA, LCR) (22,23), and the Invader reaction (24).

Secondly, the method must be sufficiently selective to distinguish between closelysimilar sequence variants. Allele specific oligonucleotide (ASO) hybridization has been usedin Southern blot hybridization to detect the sickle-cell mutation in the beta-globin gene (25),but both specificity and signal strength have proved limiting. By using pair of oligonucleotideprobes, one of which is allelic specific in ligation reactions or as primers in the PCR analysisand thereby amplifying the segment of interest, excellent selectivity and also sensitivity havebeen obtained. However, only a limited number of target sequences can be analyzed perreaction, and the methods are poorly suited for localized detection.

Thirdly, the method must be sufficiently sensitive to detect the limited number oftarget sequences present in a reasonable aliquot of genomic DNA. This problem can beobviated by target amplification, e.g. by PCR (20,21), or by increasing the signal throughamplification of the probe, such as the Invader technology using (24). The sensitivity of anassay is limited by the detection systems physical limitations and background level, e.g. thesignal to noise ratio. Unspecific binding of probes or unspecific signal amplification has to beminimized in order to generate as low background as possible. Using probes that fluoresceonly when hybridized may reduce the background signals (26-28). There are also severalinstruments that can detect a single fluorescent molecule, but they are still in there’s infancy.The problem with single molecule detection is to distinguish between true and false signalwith high accuracy due to small detection area and sample volumes giving statistical problemsas the number of targets is to low.

All those methods

There is a large number of methods to analyze known sequences variants. Most of them arebased on hybridization or denaturation of a probe strand and the complementary target, oftenin combination with enzymatic reactions, e.g. by nucleic acid-specific cleaving(endonucleases), joining (ligases), or synthesizing (polymerases) enzymes, for reviews (29-31).

4

Restriction enzyme cleavage based methods: Restriction enzymes can be used to discriminatebetween allelic variants if only one of the sequence variants can be recognized by a specificrestriction enzyme. The limitation of these assays is the demand of a restriction site in thesequence of interest. According to Hunkapiller, less then half of all point mutations give riseto gain or loss of a restriction enzyme cleavage site (22). However, restriction enzymecleavage sites may also be introduced by slightly altering the sequence around the sequencevariation (32).

Hybridization based methods: Hybridization assays take advantage of the differences inprobe-stability between a perfectly matched contra a mismatched oligonucleotide probe, suchas in allele specific oligonucleotide (ASO) hybridization (33,34). The reaction is dependent ofthe melting temperature (Tm) of the DNA duplex, and the stability increases with increasingGC content (35,36). Also nucleotides around the target SNP are of importance and this makesit hard to predict the optimal reaction condition in advance (37). Tm differences betweenprobes depending on GC content, may be reduced by adding tetra-methyl-ammonium-chloride, TMAC, in the hybridization mixture (38), or by create artificial mismatches byincorporate base-analogs in the probe (39). The ASO approach may be enhanced bymonitoring the hybridization of an ASO probe and its target over a temperature gradient toensure that a maximally discriminating temperature interval can be located, such as indynamic allele specific hybridization (DASH) (40), or melting curve analysis with TaqMan orLightCycler (41). A similar effect can be achieved by using high-density arrays consisting ofall possible sequence variants at and around the site of the SNPs screened for, and thereby getaround the problem with prediction of the optimal ASO design (42).

Ligation based methods: The ligase approach has been shown to be highly discriminativeamong single sequence variants in that a 5’phosphorylated DNA fragment is only efficientlyjoined covalently to a 3’-hydroxylated DNA fragment in a fully matched DNA duplex(22,43,44), or DNA/RNA duplex (45,46). Thermostable ligases increase even further thespecificity in the DNA ligase assay and have been used in ligase chain reaction (LCR) assays(23,47). The procedure is suitable for automatation, since the reaction can be preformed understandardized conditions. Allele-specific ligation assays have been adapted to different formatsand detection strategies (48-50).

5

Polymerization based methods: Sanger sequencing (2), the method used to decipher thehuman genome, is seen as the gold standard in SNP detection. DNA sequencing defines boththe location and nature of the changes (51,52). However, even with the development ofcapillary sequencing machines, the method is time consuming and the sample throughput islimited. To increase sample throughput, methods like minisequensing, have evolved (53,54),where the primer is extended by incorporation of a single or a few allele specific nucleotidesduring a polymerase extension reaction. The reaction conditions are similar for all SNPs andcan be adapted into a microarray format (53-55). In pyrosequencing, short segments of DNAaround a SNP can be determined. In the pyrosequencing detection system the release ofpyrophosphate upon extension is monitored luminometrically (56). The incorporation ofnucleotides by the polymerase releases pyrophosphate, which is converted to ATP. In thepresence of adenosine 5’-phosphosulfate, ATP catalyzes a luciferase reaction, generating adetectable flash of light, which thus reflects incorporation of a nucleotide (57-60).

DNA polymerases have also been used in allele-specific PCR where the DNApolymerase only extend the primer that is hybridized and completely match the variableposition (61,62). Recently, homogenous methods have evolved that combine targetamplification with probe hybridization. All reagents are assembled at the start of the reaction,so the reaction tubes never have to be opened, minimizing carry-over contaminationproblems. Several different methods are in use, but most involve fluorescent-labeled probesthat uses fluorescence resonance energy transfer (FRET). In FRET the fluorescence spectrumof one fluorophore (the donor) overlaps with the excitation spectrum of another fluorophore(the acceptor). The excitation of the donor induces fluorescence of the acceptor, while its ownfluorescence decreases. The distance between the donor and the acceptor is of importance andthis quality is used in the Molecular Beacon method. A molecular beacon probe is anoligonucleotide forming a steem-loop structure in solution. One end of the probe has afluorophore, a donor and the other end has an acceptor that quenches the energy from thedonor. Upon hybridization to its target the donor and accepter come apart and the donorthereby fluoresce (26).

In the TaqMan assay (63), the FRET labeled probe (TaqMan probe) is quenchedwhile intact, but will emit fluorescence during the PCR reaction due to the polymerase 5’nuclease activity digesting the hybridized TaqMan probe during the extension reaction(64,65). The results are monitored in real-time during the amplification and somemultiplexing can be preformed. However, the number of different fluorophores that can bediscriminated in the same reaction is an important limiting factor.

Invasive cleavage base method: The Invader technology is also a FRET-based method,containing two target specific probes, the signal probe with a 5’ region that is non-complementary to the target sequence and an invader probe. When hybridizing and fullymatched a 5’-endonuclease (Flap) recognize this strucure with the non-complementary 5’region and cleave at the branch point. The cleaved oligonucleotide can next cycle function asan invader probe and a linear increase of signals is thereby obtained (24,66,67).

Methods for localized detection in situ: For localized sequences detection in interphase andmetaphase chromosome preparations, fluorescence in situ hybridization (FISH) has becomethe predominant technique (68,69). The method is sensitive enough to detect single-copygenes (19) and can be used to simultaneously detect different target sequences by usingdifferently labeled probes (70-72). As discussed earlier, the specificity is achieved by using

6

relatively long hybridization probes, but this precludes discriminating closely similarsequences variants (73).

The sequence selectivity and sensitivity of detection of short oligonucleotide probeshybridizing in situ can be improved by incorporation of labeled nucleotides during thepolymerization reaction in the primed in situ (PRINS) reaction (74,75). The method can beused to distinguish single nucleotide variants in situ as a short unlabeled primer will primeand extend the target molecule only if the 3’nucleotide is correctly matched to variable targetposition. This method has not yet proven to be sensitive enough for detection of single copygenes in human genomic DNA. Other methods aiming to improve both specificity andsensitivity is in situ PCR (76) and in situ PCR in combination with PRINS (77), however thesignals tends to be spurious with some loss of localization and thereby resolution.Other means to increase the sensitivity is by amplifying the detection molecule by enzymaticdeposition of different substrates, such as peroxidase-mediated deposition of hapten- orfluorochrome-conjugated tyramide (78-80).Progresses are also made to improve sensitivity by finding new detection labels, such asquantum-dot (81), and up converting phosphor (82), or by denser labeled probes, e.g.branched DNA probes (82).

7

2. INTRODUCTION TO PRESENT INVESTIGATION

This thesis presents a series of novel methods for detection of known sequence variants inDNA using padlock probes and rolling circle replication. The aim of this work has been todevelop methods for distinguishing sequence variants in single copy genes and rare targets,both in solution and localized in situ using padlock probes, ligation reactions, and rollingcircle replication, and I will describe the advantages of the reagents and the methods used.

2.1 PADLOCK PROBES

Circularizing oligonucleotide probes - padlock probes - detect target sequences with very highspecificity and selectivity (83-85). These linear oligonucleotide probes have target recognitionsequences situated at both the 5'- and 3'-ends, connected by an intervening sequence that caninclude sequence elements useful for detection. When hybridized to a target molecule the twoends are brought adjacent to each other, and by the use of a ligase they are covalently joined(intra-molecular ligation). The circularized probe is wound around the target strand in amanner similar to padlocks, driven by the helical nature of double-stranded DNA.

Several features conspire to make the padlock probe excellent reagents for nucleic acidsequence discrimination. The requirement for simultaneous hybridization of two differentprobe segments to target molecules provides high specificity of detection in complexpopulations of nucleic acids. Moreover, the ligation reaction guarantees selectivity amongsimilar target sequence variants, because of the ligases’ inability to covalently joinoligonucleotides that are mismatched, especially at the 3’terminus (22,23,86). In PCRreactions eventually cross-reactions between different probe pairs can give rise to falseamplification products (87). The risk of nonspecific amplification in multiplex PCR isestimated to increase to the square of the number of primer pairs that are combined in thereaction (88). Due to the intra-molecular nature of the ligation reactions, problems with cross-reaction between padlock probes are minimized, and as a result large sets of padlock probes

8

can be combined in the same reaction (89). Finally, covalently circularized padlock probes arelocked around the target sequence and thereby resist extreme washing conditions, whichdecrease background in in situ analysis. They also resist exonuclease treatment, resulting in afurther reduction of the nonspecific signals in genetic assays (83,89).

Synthesis of padlock probes

Padlock probes are oligonucleotides typically 70–100 nt in length, but even longer padlockprobes may be used for particular applications (90). The oligonucleotides can be chemicallysynthesized by phosphoroamidite chemistry. Such long oligonucleotides are usually gel orHPLC purified. However, it is important to analyze the quality and the function of thesynthesized oligonucleotides, since it is important that they are of high quality for properfunction. Both the 3’ and 5’ end must be intact to undergo ligation, and the 5’ end must bephosphorylated. During synthesis the proportion of imperfect products accumulates with thelength of the oligonucleotide, leading to lower yield and attendant purification problems. Thisdifficulty has been addressed by Kwiatkowski el al. constructing a novel support for DNAsynthesis (91). Long oligonucleotides are relatively expensive to synthesize, especially if theycontain detectable functions such as haptens or fluorophores. A PCR-based method tosynthesize padlock probes has been presented (90,92). The protocol describes enzymaticsynthesis of padlock probes in the range of a few hundred nucleotides, labeled byincorporation of modified nucleotides during synthesis, as an alternative for flexible small-scale production. For applications where several different modifications have to be introducedin the probe, it may not be possible to use PCR or chemical synthesis. The yield is typicallylow and some of the modifications are incompatible with the DNA polymerases or withchemical synthesis. One way to solve this problem is to synthesize short oligonucleotides, thatcan be assembled to a full-length padlock probe by templated ligation. This is a more timeconsuming approach and there is a need for purification after the ligation reaction. This hasbeen done by preparative gel electrophoresis. Another way to purify the ligation-productcould be by using biotinylated oligonucleotides as the template for the ligation. After theligation reaction the biotinylated oligonucleotides are captured on streptavidine solid phase,e.g. Dynabeads, eluting full-length padlock probes.

Applications of circularized padlock probes

Padlock probes have been applied in situ to analyze single-nucleotide variants of a repeatedmotif in human centromeric alpha satellite DNA (84,85). After highly stringent washes to getrid of all unreacted padlock probes, the probes were detected via biotin, digoxigenin ordinitrophenyl residues introduced in the linker segment, followed by detection viafluorescence labeled antibodies. The sensitivity of the detection reaction is insufficient todetect single copy gene sequences in situ. By doing PCR reaction across the ligation point onthe ligated padlock probe, Zhang et al. reported detection of viral RNA sequences with anincreased sensitivity compared to reverse transcription PCR procedures (93), and Thomas etal. were able to detect mutant target sequences in a mixture of a 500-fold excess of normaltarget copies (94). Another way to improve the detection sensitivity is to use the circularizedpadlock probe as a template in a rolling circle replication reaction, resulting in anamplification of the specific signal (95-97). Banér et al combined large sets of padlock probes

9

in the same reaction and were able to genotype sequence variants in the human ATP7B gene(89). This shows that padlock probes are well suited for parallel gene analysis. Padlock probeshave also been used to invade double-stranded DNA by a triple-helix complex (98) or byusing a pair of peptidenucleic acids (PNA) openers to force the two target DNA strand apartand irreversibly attached a padlock probe to one of the complementary strands (99).

2.2 LIGATION REACTION

Ligases are enzymes whose function is to repair nicks in DNA duplexes in vivo and theligation reaction has been thoroughly investigated (100-102). The ligation reaction needs thecofactors magnesium ions and ATP, except for ligases of eubacterial origin that require NAD+.The enzyme is activated by the addition of an AMP group, whereafter the activated adenylatedenzyme binds to the nicked DNA and transfer of the adenylyl group to the 5’ phosphate of thenicked DNA. The enzyme then guides the attack of the activated phosphate by the nearby 3’-hydroxyl group, resulting in the formation of a posphodiester bond between the two adjacentstrands, and release of AMP. The joining by a DNA ligase of two oligonucleotides hybridizedto a target DNA sequence is quite sensitive to mismatches close to the ligation site, especiallyat the 3’ end (22,103). Luo et al. improved the ability of the Tth DNA ligase to identifycorrectly base paired ligation substrates by site-directed mutagenesis (47). The joining by T4DNA ligase of difficult substrates, such as blunt ends of duplex DNA molecules or DNAstrands hybridized to an RNA template (46,104,105), tend to result in the interruption of theligation process. This leads to accumulation of oligonucleotides where the 5’ phosphate to bejoined by ligation has been adenylated. This tendency can be reduced, by using substantiallylower amounts of the cofactor ATP (45,46).

2.3 ROLLING CIRCLE REPLICATION REACTION

Rolling circle replication mechanism

The rolling circle replication mechanism is used by many plasmids and bacteriophages forreplication of their circular genomes. The replication process is isothermal and usuallyrequires DNA polymerase, a primer to initiate the replication, DNA nucleotides and alsoDNA binding and unwinding proteins. The rolling circle replication (RCR) mechanism, oralso called rolling circle amplification (RCA), can also function well with small DNA or RNAcircles as templates in in vitro reactions, producing many copies in tandem of a sequencecomplementary to the initial circle (106-108). Fire & Xu used several commercially availableDNA polymerases and DNA circles as small as 26 nucleotides and generated RCR productsconsisting of many copies of the complementary circular template. RCR products longer than12000 nucleotides were obtained, which means that the enzyme traveled at least 280 timesaround the circle before dissociating. This rolling circle replication does not require DNAbinding or helix-unwinding proteins in the polymerization process. Most likely, the circle is

10

too small to form a double helix all the way around the circle and the newly synthesized DNAstrand is forced to unwind at a point remote from where polymerization is ongoing. It is ofinterest to consider the size of the circle relative the DNA polymerase. A perfect circle of 26nt circumference would have a diameter of approximately 40 Å or half that of the polymeraseenzyme itself (107). The DNA polymerase is thus too large to travel around the axis of thetemplate DNA strand during the polymerization reaction. Instead it is believed that theenzyme sits on the surface of the template, while the template is rotated along the periphery ofthe template strand (106).

Applications of rolling circle replication

Several groups are using the RCR reaction for signal amplification where small circularoligonucleotides serve as the template (95,96). The signal amplification methods aresomewhat different. The RCR primer has been attached to an antibody to increase sensitivityin immunochemical reactions (109-112) or the RCR primer has dual functions, both as RCRsignal amplifier and discriminator by being complementary to a DNA target sequence(93,95,97,113-115). The RCR reaction is a linear amplification of signal over time and can bemonitored in real-time for kinetic studies and quantification of circular probes in the reaction,for instance in combinations with molecular beacon probes (116). By introducing a secondprimer complementary to the RCR product a so-called hyberbranched RCR (HRCA) reactionwill be generated (95,117). The signal amplification in a HRCA reaction is much faster andthis technique have been used in combination with DNA intercalating dyes and by introducinga FRET labeled primer (94,114,118). HRCA is also in use as an alternative method to PCRfor DNA amplification (119,120). I will return to RCR amplification for sensitive detection inthe discussion of my work.

11

3. PRESENT INVESTIGATIONS

I. Signal amplification of padlock probes by rolling circle replication

Padlock probes have been applied in situ to detect and distinguish sequence variants inalphoid sequences in metaphase chromosomes (83-85). These sequences are composed ofhundreds or thousands of units that are repeated in tandem, and they span segments of severalthousand kilo basepair (kb) in the centromeric region of the chromosomes. Pairs ofdifferentially labeled padlock probes can be used to distinguish single nucleotide differencesamong the repeat motifs (84,85). However, single motifs, recognized by single reactedpadlock probes, are difficult to detect due to problems with nonspecifically bound but notreacted probes and limitations of current oligonucleotide labeling and fluorescence imagingtechnology. By synthesizing the padlock probes via PCR, it is possible to label them moredensely (85,92), but also such probes have so far failed to yield reliable single copy genedetection in situ. Detection of small numbers of probes may be enhanced by enzyme-catalyzed precipitation of substrate, for example using the horseradish peroxidase (HRP) todeposit tyramid (78), but sensitivity is limited since all non-specifically bound probe thatwithstands denaturing washing conditions can give rise to false positive signals. In order todetect single or very low copy number targets, localized, molecular amplification of reactedprobes is crucial. The circularized padlock probe is ideally suited to template a rolling circlereplication reaction (106,107). The utility of rolling circle replication has been used forlocalized detection of single-copy genes in human genomic DNA samples (95,121). However,there is some doubt if the published methods are sufficiently robust.

In this paper we demonstrate that circularized, but not unreacted probes can generatepowerful signal amplification by templating a rolling circle replication reaction. Replicationof any unligated padlock probes that may remain after stringent washes can only result in thesynthesis of a short base-paired sequence, which is not available for detection via probehybridization. A comparison between the Klenow fragment of DNA polymerse I and Phi 29DNA polymerase showed that Phi 29 DNA polymerase, having strong strand displacementactivity and extremely high processivity (122), was better suited for replication of circularizedprobes. Unlike the Klenow fragment, Phi 29 DNA polymerase could displace circular probehybridized to a target DNA strand, a requirement for efficient rolling circle replication. Thenucleotide incorporation rate was estimated to 1000 nucleotides per minutes with a half-life ofapproximately eleven hours, yielding a thousand-fold probe amplification per hour. Thissignal enhancement should be quite sufficient for signal detection in situ of probes havingreacted with single copy target sequences. The RCR product is single-stranded andcontiguous, which should facilitate subsequent localized detection by hybridization withdetection probes complementary to the RCR-product. The linear amplification should alsorender the process suitable for precise quantitative studies (116).

By comparing the efficiency of rolling circle polymerization reactions, templated bypadlock probes that had reacted with different template molecules, it became evident that afree target end close to the site where the padlock probe is attached is a prerequisite forefficient RCR. Specifically, when closed circular single-stranded M13 DNA was used as atarget, no RCR product was obtained. This, we believe, is a consequence of the complextopological situation that arises when a polymerase attempts to replicate a circular DNAstrand, catenated to another DNA strand. In the two following papers we have addressedapproaches to solve the topological problem in order to generate an RCR-amplified localizedsignals in situ.

12

II. Ligase-mediated construction of branched DNA: A novel DNA joining activitycatalyzed by T4 DNA ligase

Rolling circle amplification is a promising method for signal enhancement in localizeddetection reactions of allelic variants in single copy genes. It is likely that growing RCRproducts rapidly become unable to diffuse from the site of the target recognition during in situdetection and the long, contiguous single stranded RCR product is available for hybridizationdetection by many copies of labeled probes during the reaction. However, as we have shown,the RCR reaction is greatly inhibited if the probes are bound to target sequence with nonearby free ends.

In this paper we present a method to overcome this topological problem bygenerating branched padlock probes having two 3’ arms. One of the 3’ arms serve todistinguish sequence variants in a target-dependent ligation reaction, and the other 3’ arm actsas a primer for an RCR reaction. Unreacted padlock probes can be efficiently removed byhighly stringent washes, minimizing background signals. The RCR can be initiated from thesecond 3’ arm of the padlock probes using small, externally supplied circularizedoligonucleotides as templates for the RCR reaction. For this purpose it is of importance thatthe two 3’ arms have different nucleotide sequences, one 3’ arm being complementary to theinvestigated target and the other complementary to the circularized oligonucleotide used forenhanced detection.

In order to attempt ligase-mediated construction of branched DNA we synthesized anoligonucleotide with an internal 2’-5’ ribonucleotide residue. We found that T4 DNA ligasecould be used to attach an extra 3’ strand to a predefined internal nucleotide residue via aregular 3’-5’ phosphodiester bond, if this internal nucleotide leaving a free internal 3’hydroxyl group. The extra 3’ strands was positioned to undergo enzymatic joining at theinternal nucleotide by hybridizing to a template strand (Fig. 1A). The ability of T4 DNA ligaseto create this type of branched product has not previously been demonstrated. Rossi andWestern have shown that short duplex DNA molecules with a single-stranded extensions canbe turned into a hairpin-loop structures by T4 DNA ligase under reaction conditions similar tothose we used in the branched ligation reaction (Fig. 1B) (123). To prove that the reactionproduct we observed was indeed a branched DNA molecule, three alternate oligonucleotidesof distinct sizes were used to template the reaction. Their sizes had no influence on themigration of the ligation product, clearly showing that they do not take part in the ligationproduct. Only strands containing a 2’-5’ linkage with a free 3’ hydroxyl were substrates for theligation reaction. No ligation was observed with oligonucleotides having a regular 3’-5’linkage at the corresponding position, whether or not there was a free 2’ hydroxyl present. Thegeometric requirements for a nicked DNA duplex to be joined by DNA ligases have beendiscussed by Pritchard et al (124). Any disruption of the duplex that will change the anglebetween the 3’ hydroxyl, the 5’ phosphate and the AMP leaving group of the activated ligase,may decrease the rate of ligation.

13

Figure 1: A. Branched ligation, B. Intra molecular loop ligation

We sought to optimize reaction conditions for the construction of branched oligonucleotides.By lowering the ATP concentration and using an excess of enzyme over substrates the yieldof branched ligation products reached 70-80%, similar to levels observed in a control nickligation reaction. The requirement for large/stoichiometric/ amounts of ligase indicates thatthe enzyme is turned-over very poorly in the reaction or not at all. The reason for this is notunderstood. We investigated the possibility that the enzyme is stalled on the branchedsubstrate during or after the ligation reaction by analyzing the interaction of the ligase with itssubstrate by electrophoretic mobility shift assays (125). No bandshift was observed, indicatingthat the ligase is not strongly bound to the substrate after the ligation.

Shuman and colleagues have demonstrated that adenylation of the ligase is aprerequisite for stable complex formation between the ligase and substrate DNA (126).Adenylated ligase has a much higher affinity for nicked DNA than unadenylated enzyme,ensuring that only the activated enzyme binds at nicks (127). The enzyme also requires a 5’phosphate group at the nick in order to bind. The yield of branched ligation products decreasewith increased ATP concentration in the reaction while the adenylated 5’-phosphorylatedoligonucleotide increase. According to Cherenapov et al (128,129) T4 DNA ligase has twoATP-binding sites; a high-affinity catalytic ATP binding site and another, low-affinity non-covalent nucleotide binding site. In the presence of 5 mM Mg2+ and absence of double-stranded DNA they showed that ATP and Mg2+ formed a complex, MgATP2-, that couldoccupy the DNA binding site. This could be a possible explanation for the inhibitory effect byATP during branch ligation, if the MgATP2- complex binds with higher affinity than thebranched substrate. An alternative explanation could be that the ligase binds very weakly tothe branched structure and falls off after adenylation of the substrate. In the absence of ATP,adenylated DNA is covalently joined by a non-activated ligase, and thereby the ligationreaction is completed. At higher ATP concentrations the ligase may be reloaded before theligation reaction can be completed. Thereby ATP would have an inhibitory effect on theligation reaction, resulting in the observed accumulation of adenylated ligation sustrates. Thelatter explanation is contradicted by the observations of Doherty et al. that only activatedligases bound to nicked DNA, but they did not attempt to ligate adenylated nicked DNA (130).Several investigators have shown that isolated adenylated DNA complexes form proper

14

phosphodiester bonds in the presences of the enzyme and in the absences of the cofactors,NAD+ or ATP (102).

Branched DNA molecules generated by enzymatic ligation, as we describe, couldalso be used to generate biallelic padlock probes, branched padlock probes with 3’ armsspecific for each of two allelic sequence variants. Most polymorphisms in the mammaliangenome are biallelic and probes like this could have advantages in sequence discrimination.For example, biallelic padlock probes could be immobilized on a surface for genotyping. Thetwo 3’ arms of the padlock probes could be differently labeled and contain enzymatically orchemically cleavable units in the 3’ arms. After the ligation reaction and cleavage, followed bystringent washes, only the ligated arm will be attached to the surface, giving rise to a signal(131). An alternative means to detect which 3’ arm has been ligated is by PCR amplificationacross the ligation point (89). Another way to amplify the signal would be to perform an RCR,templated by the ligated branched padlock probe. We were thereby interested to see ifbranched DNA molecules could serve as templates for replication by DNA polymerases.Lorsch et al., have shown that reverse transcriptase is able to read through a 2’-5’ linkage in atemplate (132). To get rid of the unligated 3’ arm that could be inhibitory for the RCRreaction. Several DNA polymerases have been tested for their ability to replicate across thebranch site, but without any success, so far.

A generic branched DNA molecule could also be used to construct different locus-specific probes by attaching specific sequences to the 5’ and the two 3’ ends via templatedDNA ligation. It is also possible to use the procedure we describe to generate branchedmolecules with more than two branches, and to construct branched DNA structures fornanoengineering purposes.

III. Target-initiated rolling circle replication of padlock probes as a mechanism forlocalized mutation detection

In this study we demonstrate an even more powerful approach to elicit local signalamplification from specifically reacted probes, circumventing the topological problemdiscussed in paper I. A labeled padlock probe should give localized signal, but in solutionphase experiments we have demonstrated that during denaturing washes padlock probes areable to slide along the template strand for at least several hundred nucleotides, resulting in arisk of losing the localization of signals in in situ analyses (83). Of greater concern is the riskthat also nonspecifically bound probes can be detected. By digest the template downstream ofthe padlock recognition site and using Phi 29 DNA polymerase with efficient 3’ exonucleaseand strand displacement activity (122), we found that the template strand could be used toprime the RCR reaction. The advantage of this approach is two fold: First, since no externalprimer is used the topological problem of replicating a circular DNA molecule catenated to atarget DNA strand is avoided. Second, the amplified signal that forms is contiguous with thetarget DNA, generating an anchored localized signal, which can be conveniently detected byhybridizing a labeled complementary probe to repeated segments of the RCR product.

Two different approaches to open the target DNA were demonstrated. After theligation of padlock probes to denatured target DNA, a short oligonucleotides was hybridizedacross a restriction endonuclease recognition site on the target allowing this to be cleaved byrestriction digestion. By taken advantages of the 3’ exonuclease activity of Phi 29 DNApolymerase (122), the single-stranded target DNA was digested until the enzyme reached theduplex between target and padlock probe and an RCR reaction was initiated. Unfortunately itis not always possible to find a restriction endonuclease recognition site sufficiently close to

15

the target sequence. However, by using a strategy with an adapter-primer containing class-IISrestricion endonuclease the target can be cleaved wherever it is desired by designing adapter-primers complementary to the target (133,134). The 5’ end of the primer formes a hairpin,containing the recognition site for a type IIS restriction endonuclease. These enzymesrecognize asymmetric sequences and cleave these sequences at a defined distance. Theenzyme used in this study, FokI from Flavobacterium okeanikoites, digests the two strands 9and 13 nucleotides, respectively, away from the recognition site.

Figure 2: Different approaches to open the target strand. A. By restriction digest of excistingrestriction endonuclease recognition site. B. By adapter-primer containing class-IIS restrictionendonuclease. C. By padlock probe containing class-IIS restriction endonuclease site.

We investigated how far the padlock probe could be located from the 3’ end of the targetstrand without preventing an efficient RCR. In paper I, the rolling circle product wasmonitored by incorporation of radiolabeled nucleotides. Also 3’ ends of remaining unligatedpadlock probes may prime extension reactions. To be able to specifically identify RCRproducts, a complementary oligonucleotide was added and the extension products weredigested with a restriction endonuclease, generating a fragment the size of the padlock probe.Real-time monitoring with molecular beacon probes offers a convenient method to monitorRCR reactions (116). Molecular beacons are probes that emit fluorescence when hybridized.The molecular beacon probes used were made of 2’-O-Me-RNA to withstand the 3’exonuclease activity of Phi 29 DNA polymerase. Moreover, one of the stem sequences wascomplementary to the RCR product to avoid inter molecular hybridization, quenching thefluorescence. Single-stranded M13 DNA was digested with a set of three different restrictionendonucleases or using two primer-adapters, generating in total five distinct lengths of 3’overhang. The reactions were followed in real-time using molecular beacon probes. The timefor the RCR reaction to reach threshold fluorescence increased with the length of the 3’overhang. As shown in paper I, target molecules extending 3.6 kb downstream of where thepadlock probes bound and circular templates did not generate RCR products. In contrast toour results Kuhn at el. report no difficulties generating primer initiated RCR reaction productsfrom padlock probes catenated to one strand of a duplex DNA sequence, so called earringprobes (115). Also Lizardi et al. generated primer-initiated RCR signals in situ without anymeasures avoid a topological block (95). In both papers the detection efficiency appears low.Interphase nuclear DNA immobilized on glass is likely to be broken relatively frequently(135), so it is possible that by chance a free end could be close by a circularized probe,making RCR reaction feasible, or some probes could be pseudo-topologically linked, that isnot catanated to the target strand and thereby free to act as the template in RCR reaction(136). Kuhn et al. on the other hand, circularize probes to “endless” DNA molecules,dumbbell DNA (dbDNA) and by treatment with exonuclease they ensure that all target DNAis closed circles (115). They found that Sequenase and Vent exo- DNA polymerase incombination with single-stranded binding protein (SSB) conducted efficient RCR reactions on

16

such earring probes. On the other hand they found that Phi 29 DNA polymerase did not workat all. Their conclusion is that by choosing the right enzyme there will be no problem withcatenated padlock probes in RCR reactions. We could not repeat these observations using thesame conditions except that circular closed M13 was used as ligation target. The discrepancybetween results is not understood. Moreover, Kuhn et al. are using a primer to initiate theRCR reaction, which means that even though they are able to generate an RCR product, thissignal may not be localized. In the method described in our paper III, the RCR reaction isinitiated from the target strand and contiguous with the target strand, generating a truelocalized signal.

The combination of localized and specifically amplified RCR signals should beadvantageous in high-throughput assays. DNA or RNA samples from different patients couldbe spotted in a microarray format, and probed with one or several padlock probessimultaneously in order to detect allelic variation in single copy genes or in transcripts. In afirst attempt to test this approach 10-fold dilutions of single- and double-stranded M13 DNAwere spotted on SigmaScreen glass slides, followed by target-dependent ligation of padlockprobes and generation of RCR signals. A ligation-dependent RCR signal was obtained onlywhere the single- or denatured double-stranded M13 DNA template had been first cleaved.These results look very promising, and detectable signals were obtained with as little as 104-105 target molecules per spot. It is likely that only a minority, maybe 10% of the spottedmolecules were immobilized on the surface and, that not all of these are accessible forhybridization, suggesting that even fewer molecules were detected.

Recent developments

How to make a break

I have been working with three different approaches to generate specific nicked target DNAof which two are presented in paper III. A third possible approach is to design the padlockprobe so that it contains the type IIS restriction enzyme site, see figure 2C. In order to protectthe padlock from digestion, the padlock probe would contain modified nucleotides that resistthe enzymatic digestion, while permitting digestion of the opposite strand. It is also ofimportance that the modified padlock probe is able to function as template in the RCRreaction. This method would have the advantage, in comparison with the adapter-primerapproach, that only a standard oligonucleotide hybridizing to the padlock probe is needed fordigestion. This sequence is not involved in target recognition and thereby the sameoligonucleotide could be used for all padlock probes. DNA strands containingphosphorothioates have been shown to resist digestion by several restriction endonucleases,for instance RsaI and HincII, while permitting digestion of the opposite strand. This is used inthe strand displacement amplification (SDA) method (137). To test the feasibility of ourapproach, we constructed a model system to investigate if phosphorothioates were able toprotect the padlock probe from being cleaved and, moreover, if FokI was able to cleave acrossa three way junction where an oligonucleotide contained thio-modifications. Sets of shortoligonucleotides of different length, radiolabeled one at the time, were hybridized to form athree-way junction to mimic a padlock probe hybridizing on a target. Comparison was donebetween common oligonucleotides and thio-modified oligonucleotides and the results wereanalyzed by denaturing gel electrophoresis. As can be seen in figure 3, phosphorothioates doprotect from digestion.

17

Figure 3: Phosphorothioates protect against digestion by FokI. Reactions containing two pmolradiolabeled oligonuclotide (56nt) with a FokI recognition sequence at the 3’ end, 1 pmol template(71nt), and 3 pmol oligonuclotide (23nt) complementary to the 3’ end of the radiolabeledoligonucleotide were incubated at 65º C for 10 min in 20 µl reaction buffer (10 mM TrisOAc, pH 7.5,10 mM Mg(Ac)2, 50 mM KAc, BSA 0.1 µg/µl), and allowed to cool to room temperature for at least10 min. Thereafter 5 units of FokI was added and the reaction were incubated at 37º C for 1 hour. Anequal volume of loading buffer was added and the samples were separated in a denaturing 10%polyacrylamide gel and analyzed on a phosphorimager. If the radiolabeled strand lacked thio-modification then a fragment, of the expected size of 19nt, was generated by digestion with FokI, butthis fragment was missing from strands where phosphorothioates were present in the cleavage site.The fragment of 26nt is depending of unspecific digestion.

We synthesized a thio-modified padlock probe, target-specific to positions 6263-6292 inM13mp18 + strand, and by using same approach as in the model system we showed that thetarget strand was cleaved and the padlock probe was left intact (data not shown). Moreover,we shown that Phi 29 DNA polymerase is able to read through the thio-modifications, butattempts to initiate target templated RCR reactions have failed so far. Structural studies ofFokI bound to DNA have shown that the enzyme is dimerized and requires binding to asecond recognition sequence (138,139). This has to be considered when developing solidphase methods, such as DNA microarrays and in situ detection methods. It is of importance tooptimize the time and enzyme concentration in the reaction, to avoid unspecific digestion.FokI has also been shown to cut imprecisely, and we have seen that some oligonucleotideshave been cut even outside the thio-modifications, see figure 3 lane 2 and 4. Experiments withthe restriction enzyme BpmI, known to generate precise cuts, seem more promising in these

18

respects. Thio-modification of oligonucleotides was done during the synthesis with theoxidation steps replaced by sulfurization. By thio-oxidation one of the two non-bridgingoxygen is substituted by a sulphur molecule giving a mixture of the two diastereomers,differently tolerated by different enzymes (140). Enzymatic synthesis to introduce thio-modifications using alpha-thiodeoxynucleotides in a DNA polymerase extension reaction willgenerate only one of the two possible configurations (141), and may thereby be a betteralternative for introducing thio-modifications. There are also methods to exchange thebridging oxygen, one or both, in the phosphodiester bond (142-144), but these modificationshave not been tried.

Detection of allelic variants in BRCA1 gene in interphase cell nuclei in situ

Breast cancer is the most common cause of cancer form among women in the Western world(Socialstyrelsen, Cancer incidence in Sweden 1994, Stockholm, 1997) and it is estimated thatone woman in eight will develop breast cancer during her lifetime (145). Over the past 10years, the incidence of breast cancer has remained unchanged, although mortality rates havesteadily decreased, due to earlier detection and more effective therapy. Approximately 5-10%of breast cancers have a clear hereditary component (146), involving germ line mutations inspecific genes. Two genes predisposing for breast and ovarian cancer, BRCA1 (chromosome17) and BRCA2 (chromosome 13), have been isolated by linkage analysis (147,148), togetheraccounting for 70% of inherited breast cancers. More than 100 unique disease-causingmutations have been found in the BRCA1 breast cancer susceptibility gene (149), and themutations are distributed throughout the coding sequence with no apparent clustering or hotspots (150). This complex mutation pattern complicates development of general diagnostictests for breast cancer. Today, population based mutations, such as founder mutations in theAshkenazi population where 2% of the population carry one of three BRCA1 and BRCA2mutations (151), and known mutations in families with several affected family members arediagnosed.

In research there is an interest to evaluate the relevance of newly found mutations inthe etiology of an investigated disease. Often there exist an archive with patient materialcoupled with relevant clinical data. A common way to collect tumor samples is in paraffin-embedded blocks of formalin fixed tissue. Slices of the blocks can be investigated long afterthe tumor was collected. A clever approach to speed up the sample throughput by preparingso-called tissue microarray (TMA) has been invented by Kononen et al (152). By punchingcore biopsies from a donor block and depositing the core in a recipient block in an arrayformat, sections can be prepared that permit up to 1000 individual tissue specimens to beanalyzed simultaneously for a variety of molecular targets at the DNA, RNA, or protein level.Moreover, thousands of replicas of TMA slides can be made from the same block, facilitatinganalysis of different questions asked on the same patient cohort (153). Unfortunately, presentprobing strategies lack the sensitivity and specificity required detecting the location ofspecific DNA sequence variants in such arrays.

In collaboration with Dr. Minna Tanner, Institute of Medical Technology, Universityof Tampere, Finland, a project has been initiated to analyze point mutations in the BRCA1gene, in TMA slides as well as metaphase spreads and interphase cells. Padlock probes,targeting wild-type and mutant, position 3172 and 1806 in the BRCA1 gene, 2 foundermutations in a Swedish patient cohort (154,155), were constructed with appropriate adapter-primers and preparatory investigations on metaphase chromosomes were done. Theadvantages to do in situ reactions on metaphase chromosomes are that signals generated can

19

be evaluated by reference to the position on DAPI stained chromosomes. Unfortunately, veryfew signals were generated, so we chose to focus our work on interphase nuclei DNA insteadwhere DNA sequences may be more accessible. The strategy used was to hybridize and ligatepadlock probes, followed by digestion of template DNA with adapter-primers, and target-initiated RCR reaction. By hybridizing haptenized probes, e.g. digoxygenin or biotin,complementary to the RCR-product, followed by an immunochemical enzymatic reactionwith horseradish-peroxidase (HRP) conjugated antibodies and fluorescent labeled tyramide(TSA), strong signals were generated. This detection system has been shown to be verysensitive, but also nonspecifically bound antibodies can generate signals. We therefore madeconsiderable efforts to optimize the reaction conditions in order to reduce background. Ininterphase nuclei it is not possible to relate points signals to chromosome morphology. Toshow if a signal was indeed target dependent, the RCR signal was colocalized by hybridizingfluorescent FISH probes to loci situated near the padlock probe on the very samechromosome. We first investigated the distribution of hybridized BRCA1 FISH probes ininterphase nuclei in combination with FISH probes to TOPO II gene (17q12) approximately 2megabases centromeric to BRCA1. We got strong signals and low background giving us anidea of the distribution of TOPO II FISH probe in comparison with BRCA1 FISH probe,figure 4A. Usually wild-type cells were used and the expected numbers of signals were 2 to 4depending of where in the cell cycle the cell were. As a negative control the reaction wasperformed by omitting the ligase or by using padlock probe for the mutant target on wild-typecells. Figure 4B shows DAPI stained interphase nuclei with RCR-product detected byrhodamine TSA reaction and localized FITC labeled TOPO II probe. By counting signals incells treated with and without ligase and only counting signals that were colocalized with theFISH probe, a slightly higher number of signals were obtained from cells treated with ligase.Experiments with fluorescent labeled probes complementary to RCR-products instead of TSAdetection reaction have not succeeded so far for single-copy gene detection. We believe thatwe have positive RCR signals, but with very low efficiency.

20

Figure 4.

21

Figure 4: Cells from human cancer cell line MDA-453 were cultured and collected in PBS (0.1 MPhosphate buffer, pH 7.5) with a cell density of approx. 650 cells /µl. An aliquot of 150 µl werepipetted into cytospin funnel, and centrifuged for 8 min at 800 rpm in a cytospin centrifuge. The cellswere fixed with Carnoy solution, 75% methanol and 25% acetic acid, in a series from 50, 75, and100% Carnoy solution, air-dried at RT over night, followed by incubation in 2xSSC (1xSSC is 0.15 MNaCl, 0.015 M Na-citrate), at 37º C for 30 min, wereafter the cells were dehydrated on slides througha graded ethanol serie and air-dried. For pretreatment, the slides were incubated at 37º C for 15 min in40 µg/µl RNase A in 2xSSC, followed by dehydration through an ethanol series and air-dried. Thecells were denatured in 70% formamide and 2xSSC at 70º C for 2 min. A. Hybridization of FISH-probes to BRCA1 (red) and TOPO II gene (green). The hybridization of FISH-probes was performedin 10% dextran sulphate, 2xSSC, 32% formamide, and 0.01 µg/µl Cot-1 DNA at 37º C over nightincubation. The microscope slides were washed in 0.4xSSC and 0.3% NP-40 at 70º C for 2 min,followed by washing in 2xSSC, and 0.1% NP-40 at ambient temperature for 1 min. After rinsing indistilled water, the cells were counterstained by 0.1µg/µl DAPI in Vectashield. B. Colocalization ofRCR BRCA1 (red) and FISH-TOPO II (green) signals. The hybridization of padlock probe andadapter-primer, 0.5 µM respectively, was performed in 10% dextran sulphate, 2xSSC, and 0.5 µg/µlsonicated single-stranded herring sperm DNA at 37º C over night incubation, followed by ligationreaction in 20 mM TrisHCl, pH 8.3, 25 mM KCl, 10 mM MgCl2, 0.5 mM NAD, 0.01% Triton X-100,8.7 % glycerol, 0.1 µg/µl sonicated single-stranded herring sperm DNA, and 0.25 units/µl Ampligaseat 50º C for 20 min. The FokI digestion was performed in 10 mM TrisHCl, pH 7.5, 10 mM MgCl2, 1mM DTT, 50 mM NaCl, 0.1µg/µl BSA, and 0.8 units//µl FokI at 37º C for 60 min, and followed byRCR reaction in 50 mM TrisHCl, pH 7.5, 10 mM MgCl2, 20 mM (NH4)SO4, 1 mM DTT, 0.15 µg/µlBSA, and 40 ng/µl Phi 29 DNA polymerase at 37º C for 3 hours. Slides were washed three times, firstin 2xSSC, 30% formamide at 42º C for 10 min, followed by 2xSSC at 55º C for 10 min, and 2xSSC atRT for 5 min, respectively, and then incubated in FISH hybridization solution (same solution as in A),containing 0.1 µM RCR-complementary probe and TOPO II-FISH probe, at 37º C over night. TheDIG-labeled RCR-probe was detected by tyramide amplification reaction (TSA) according to therecommendations from the supplier (NEN). Anti-DIG-HRP was diluted 1:2000and tyramide-rhodamine conjugate was diluted 1:400. The slides were washed as described in A and counterstainedwith DAPI in Vectashield.All hybridizations and enzymatic reactions were performed under cover slip and after the reactions,the microscope slides were washed in 2xSSC to get rid of the cover slip, followed by dehydrationthrough an ethanol series and air-dried. Long incubations were performed in moister chamber and thecover slip was sealed with glue.

Theoretically an RCR signal from circular padlock probe should be strong enough to bedetected in the system we use. FISH is a demanding method with many steps that must beunder control. First of all the specimen must be in good condition. There must be sufficienttarget for the hybridization assay. Raap et al. has shown that losses of DNA during fixationand denaturation is approximately 20% per step, and in combination with rapid renaturation,but also depending on attachment of target DNA to the surface, the probe accessibility to thetarget sequence is greatly reduced (135,156). Next, when a padlock probe has beensuccessfully hybridized and topological linked to the target by the ligation reaction, the RCRreaction must be efficient. Lizardi et al. also had problems with RCR amplification of padlockprobes on metaphase chromosomes, but were able to generate RCR signals on salt-extractednuclei variable detection efficiency (95). More recently they have used a different strategyinvolving ASO probes for allele specific hybridization, followed by priming of RCR of apreformed circular probe. The ASO probes consist of an oligonucleotide with two 3’ ends,one targeting the sequence variant looked for and the other 3’ end complementary to the small

22

circular oligonucleotide templating the RCR reaction. With this approach they were able todetect sequence variants both in interphase cells in situ and on DNA fibers (157). Christian etal. have demonstrated detection of gene copy number in situ with padlock probes and RCRsignal amplification (121). By pretreating cells with a combination of restriction enzymes andexonuclease, they first made the target DNA single-stranded. Thereafter a padlock probe washybridized and ligated, followed by an RCR reaction initiated by an external primer. Theyfound that exonucleolytic removal of the non-targeted DNA strand significantly increased theefficiency of the process. To ensure that the correct strand was removed by the 3’ exonucleasetreatment, they used a restriction enzyme site situated 5’ of the site on the strand, 20 or sonucleotides away, where the padlock probe is located. In paper I we are discussing thedifficulties for a RCR reaction to occur when there is no free target end near where thepadlock probe is catenated. Christian and coworkers wanted to show that the importance forefficient RCR reactions is the target accessibility for the probe and that a catenated padlockprobe can template the RCR reaction by using external primers. When they treated the cellswith restriction enzyme and omitted the exonuclease step, no signal increase was obtainedcompared with heat denaturation and no enzymatic treatment at all. This, they mean, rules outthe possibility that the restriction cut in it self is responsible for an RCR reaction to occur. Ifthey instead had cleaved 3’ of the padlock probe and used a 5’ nuclease, they might have hadtarget initiated RCR signals, primed by the target strand and the signals had been continuouswith the target strand generating a true localized signal. Some of the signals they present lookvery fuzzy and some of them also resembles of nucleolus formation often seen in interphasenuclei. When doing studies on interphase nuclei, where the signal cannot be confirmed by theposition in the substrate, it is even more important to have relevant controls. It is also ofimportance to analyze the result blinded, not to be biased in the judgment of the results. Wenow have a microscope with a spot finding program and this will facilitate furtherinvestigations.

IV. PCR-independent multiplexed genotyping and quantitative expression analysis by aserial circle amplification procedure

In this paper we describe a new amplification method, a serial isothermal rolling circlepolymerization reaction (serial RCR) that utilizes the described advantages of the padlockprobe in combination with RCR amplification and microarray format. By converting therolling circle product to a second or third generation of circularized probes we are able toamplify the signal billion-fold. After ligation of a target-specific padlock probe withspecificity for sequence variants an RCR reaction is initiated with an oligonucleotidecomplementary to the RCR product or by target-proming. Next an oligonucleotide is added todirect digestion of the RCR product to monomer units and to subsequently allow these toligate to form a second generation of circular molecules capable of acting as a template fornext generation of RCR reaction primed by the same added nucleotide. We demonstrated thatthis procedure provides sufficient sensitivity for detection of single-copy gene sequences in50 ng human genomic DNA by hybridization to tag oligonucleotide arrays, and that largenumbers of probes could be amplified in parallel with no problem of cross-reactivety. It iscrucial that the quantitative representation of circularized probes is preserved during theamplification. By following the RCR reactions with molecular beacon probes in real-time, wecould demonstrate that the efficiency of amplification of distinct padlock probes were quitesimilar.

23

For many applications it is of particular value that the amplification products are alinear single-stranded DNA molecules and composed of copies of the padlock probe or itscomplement. Lizardi et al. have described a hyperbranched rolling circle amplification(HRCA) technique (95) In this procedure two primers of opposite polarity rapidly generate anumber of physically unlinked and mainly double-stranded DNA of variable length. TheHRCA reaction proceeds much faster than linear amplification reactions and it has been usedby several groups (94,95,97,114,158,159). This reaction has been shown to yield someamplification products in the absence of circularized probes apparently interfering withdetection of low-copy number targets (118). In serial RCR there is a risk that primers andunligated padlock probes between generations can give rise to HRCA. By using radiolabelednucleotides in the RCR reaction, we showed that the ligation efficiency to circularize themonomeric RCR product is at least 99%. A comparison between serial RCR and HRCAreactions performed in the presence of DNA intercalating dyes showed that only in the HRCAreaction was double stranded DNA produced. This shows that serial RCR produce single-stranded DNA and thereby is suited for multiplexed amplification reactions for quantitativereal-time or DNA microarray analysis.

24

4. DISCUSSIONS AND CONCLUDING REMARKS

The majority of mutations in genetic disorders are due to single base pair changes or smallinsertions or deletions. With the knowledge of the sequence of the human genome and theprogress of databases cataloguing sequences variations, it is desirable to develop DNAmethods that are highly sensitive and specific, and suitable to detect the localization of DNAsequence variants singly or for large sets of target sequences. A robust method for in situnucleotide discrimination in single copy gene sequences could reveal haplotypes of genevariants without access of family data. For these purposes it is crucial to use highly specificdetection probes that can be efficiently scored, preferably from even single target molecules.

Padlock probes are excellent reagents for nucleic acid sequence discrimination andthe circularized probe can be used as a template in an RCR reaction offering thousand-foldsignal enhancement with preserved specificity (paper I Banér 98). By converting the rollingcircle product to a second and even third generation of padlock probes the signal can beamplified thousand-fold per generation (paper IV). Padlock probes have been proven suitablefor detection of sequence variants of DNA samples and can be amplified in parallel withquantitative resolution (89). The methods described in paper II and III have the potential todetect single copy genes in situ in interphase cells, in metaphase chromosomes, or even alongDNA fibers.

Padlock probes - a promising reagent in haplotyping

The process to distinguish which combination of alleles resides on a particular chromosome,or haplotyping, is a technical challenge. Although the availability of parental genotypessometimes reveals the haplotypes of the offspring, this information is often not available.Recent investigations have shown that the grouping and interaction of several SNPs inhaplotypes may confer considerably greater resolving power than individual SNPs. Drysdaleet al. reported an individual’s genetic response to sabutamol (a drug for asthma) that iscorrelated with the interaction of multiple SNPs within a haplotype. They suggested that thehaplotype approach might be essential in recognizing genetic factors that create a particulardisease or pharmacogenetic phenotype (160). Since the human genome is diploid and containstwo copies of each chromosome, with the exception of the X and Y chromosome, it may bedifficult to establish the haplotypes and often it must be inferred by determining the genotypesin several generations. Direct determination of haplotypes is possible by cloning the twochromosomes independently to reveal which SNPs appear along the same piece of DNA.Methods exist that isolate complete human chromosome completely to a haploid state throughfusion between human and rodent cells. These methods are laborious and time consuming. Itis also possible to statistically infer haplotypes based on the occurrence of SNPs in patients(160).

Precise positional information for specific sequence variants within chromosomes isof great interest. The methods described in paper II and III might have the potential toinvestigate single copy genes in DNA fibers. Zhong et al. have demonstrated allelediscrimination on human genomic DNA fibers with hybridization probes targeting DNAsequences as small as 50 nucleotides (157). Padlock probes may have further advantagesgenerating lower background, compared with hybridization probes, either by being moreresistant during stringent washes as reported so only reacted padlock probe remains asdescribed in paper II, or as in paper III, only reacted padlock probes can act as templates inRCR reactions. Both in the method of Zhong et al. and in paper II, any remaining unreacted

25

probes can still give rise to signals in RCR reactions. By using sequence-coded padlockprobes and combinatorial multicolor technology this approach could also permit thesimultaneous detection of many different specific point mutations.

Padlock probes for localized detection on microarray

The focus of human genetic research has gradually shifted from monogenetic disorders tocomplex or multifactorial diseases. They include the most common human diseases such asdiabetes, hypertension, asthma, common cancers, and neuropsychiatric diseases. ForMendelian diseases caused by mutations in a single gene, mapping disease loci by pedigreestudies has been very successful. However, for multifactorial diseases this strategy is usuallynot practical. With the development of SNPs markers hopes are increasingly pinned onassociation studies to unravel underlying causes for these kind of diseases. Extremely efficientmethods will be required to analyze of large numbers of sequence variants in many patientsamples to identify genetic risk factors.

Padlock probes may prove useful to discriminate sequence variants in sets of arrayedDNA samples. Different techniques have been developed to construct arrays of samples, suchas by transferring cDNA fragments in 2D arrays by stamping (161,162) or ink-jet printing(163). In most of these methods, the sample of interest has to be amplified by PCR or RT-PCR, before immobilization to the support. In tissue microarray (TMA) material from tumorsare mounted in array format, whereupon a patient cohort can be examined for one or severaltarget sequences (152). A similar approach has been patented by Landegren et al. wherearrays are obtained by fixing together carrier element that can have different immobilizedDNA samples. The bundle can be sectioned, like slicing of sausages, generating perhapsthousands of DNA arrays of representing the same set of samples (164).

Multiplex padlock assays could greatly increase throughput in genotyping. Padlockprobes are well suited for parallel gene analysis because cross-reactive ligation eventsbetween different probes, which may arise when a large number of probes are combined,generate linear products that can easily be distinguished from circularized probes. Thecombination of padlock probes and localized signal amplification via RCR should allowdetection of large sets of target DNA sequences. The multiplexing capability of padlockprobes has recently been demonstrated (89). Moreover, by converting the rolling circleproduct to a second and third generation of padlock probes, followed by assortment of thereaction products on a so-called tag array, an even more sensitive assay has been established(paper IV). The advantage with this approach is that large numbers of genetic variants can beanalyzed in parallel at limited expense and with minimal consumption of samples.

26

5. ACKNOWLEDGEMENTS

This study was carried out at the Department of Genetics and Pathology at UppsalaUniversity. I am grateful to a number of people who, in various ways, have helped me duringthe course of this work, first at BMC and now at the Rudbeck Laboratory.

I should like to particular express my sincere gratitude to:

Ulf Landegren, my supervisor, for your optimism and strong belief in your ideas. It has been areal challenge to try to make them come true. I wanted to be dedicated to research, and so Ihave been.

Ulf Pettersson, for nice scientific atmosphere and for being member in my industrial scientificcommittee and sharing my excitement over results.

Jeanette Backman for always being so helpful and friendly. You certainly left a void behindyou when you left for your dream job (which you really deserve).

The molecular medicine group, wow, what a group!!! It is just the kind of group you feelcomfort within, so many nice people at the same place. I will miss you! Anette (systrami), foreverything that we have shared during this time. DOA, now with New York style, myconferences pal. Mats N. for sharing my excitement over odd ligations, and for reading mymanuscript. Johan with your funny letters about the result in “gel-ligan” and never endingfalling leaves. I will also miss all inspiring lunch discussions about science, but also aboutcars and motorbikes, and the music in the lab, Mats G. for “prog” music (tralla la la la lej, detär så kul att vara dej), Anders for samba and introducing betting in football. I think a lot ofpeople have come to know each other in that way. Jonas for motorbike-rock and introducingFriday cake together with Mårten. Sigrun for marvelous recorder music, - I´m reallyimpressed. The boys in “drängstugan” with guests, Simon, Fredrik, Jonas, Johan, alwaysdiscuss science (when you are not taking a break playing computer games). Marek for healthylunch walks, when the food is better somewhere else, Ola for positivism (and for just brokenmy synthesis record, congratulations!) and Maryam, always with a smile, trying to keep thelab in order (I wish you good luck!)

Marie-Louise, Britt-Marie, Kicki, Maria, Lena, Anh-Nhi, for sharing lab with me at BMC as anewcomer, especially Kicki for even sharing lab bench. (Yes, someday we will probably dothat again.)

The girls, “stödstrumporna”, for drinking beer and sharing the ups and downs in the scientificworld.

People at the Clinical Genetics that generously let me use their microscope, especially I wouldlike to thank Lore Zech, my female mentor, encouraging me to take every opportunity tospeak at seminars, and Gisela for lunch discussions and standing my Spanish speakingattempts.

Minna, Jorma, Pävi, Sari and all others in Tampere, for having patience with me occupyingthe lab and emptying the storages of your reagents. It was a pleasure to work in your well-

27

ordered facilities and take advantage of your knowledge. I will especially thank Minna whonearly spend both days and nights at the lab with me, when we were trying to get results.

Anil Kumar, for figuring out how to make a branched DNA molecule and having the trust inme to fulfill the idea. It was really fun and it worked!

I am also very grateful to my colleges at Amersham Biosciences, especially to Björn Ekström(today at Pyrosequencing) that gave me the opportunity to begin my PhD studies. During thistime several of you have joined my industrial scientific committee and it has always been apleasure to show you my results. I would also like to thank Björn Nilsson (today at Karobio)that let me finish my work. Watch out, I’m coming back and it will be fun to join my newgroup to 100%. (What shall I do with the rest of my time?)

I also want to acknowledge the most important people in my life, my own pedigree. Myparents that probably started my scientific interest with Christmas books in science; - sorryyou had to wait so long for this (just joking). My children Åsa and Mikael, the best synthesis Iever have done (but not alone), I hope that I haven’t frighten you from doing the same (guesswhat!). Thank you for all your care. And last my synthesis partner, Ibbe - what can I say? –You are simply the best.

28

6. REFERENCES

1. Watson, J. and Crick, F. (1953) Nature, 171, 737-738.2. Sanger, F., Nicklen, S. and Coulson, A.R. (1977) Proc Natl Acad Sci U S A, 74, 5463-

5467.3. Venter, J.C., Adams, M.D., Myers, E.W., Li, P.W., Mural, R.J., Sutton, G.G., Smith,

H.O., Yandell, M., Evans, C.A., Holt, R.A. et al. (2001) Science, 291, 1304-1351.4. International Human Genome Sequencing Consortium. (2001) Nature, 490, 860-921.5. Collins, F.S., Brooks, L.D. and Chakravarti, A. (1998) Genome Res, 8, 1229-1231.6. Risch, N. and Merikangas, K. (1996) Science, 273, 1516-1517.7. Stern, H.J., Carlos, R.D. and Schutzbank, T.E. (1995) Clin Biochem, 28, 470-473.8. Schafer, A.J. and Hawkins, J.R. (1998) nature Biotechnology, 16, 33-39.9. Rotter, J.I. and Diamond, J.M. (1987) Nature, 329, 289-290.10. Evans, W.E. and Relling, M.V. (1999) Science, 286, 487-491.11. Athanasiou, M.C., Malhotra, A.K., Xu, C. and Stephens, J.C. (2002) Psychiatr Genet,

12, 89-96.12. Collins, F.S., Guyer, M.S. and Chakravarti, A. (1997) Science, 278, 1580-1581.13. Chicurel, M. (2001) Nature, 412, 580-582.14. Hacia, J.G., Fan, J.B., Ryder, O., Jin, L., Edgemon, K., Ghandour, G., Mayer, R.A.,

Sun, B., Hsie, L., Robbins, C.M. et al. (1999) Nat Genet, 22, 164-167.15. Daly, M.J., Rioux, J.D., Schaffner, S.F., J., H.T. and Lander, E.S. (2001) Nature

Genetics, 29, 229-232.16. Johnson, G.C.L., Esposito, L., Baratt, B.J., Smith, A.N., Heward, J., Di Genova, G.,

Ueda, H., J., C.H., Eaves, I.A., Dudbridge, F. et al. (2001) Nature Genetics, 29, 233-237.

17. Wallace, R.B., Shaffer, J., Murphy, R.F., Bonner, R.F., Hirose, T. and Itakura, K.(1979) Nucleic Acids Research, 6, 3543-3557.

18. Southern, E.M. (1975) J.Mol. Biol, 98, 503-517.19. Landegent, J.E., Jansen in de Wal, N., van Ommen, G.-J.B., Baas, F., de Vijlder,

J.J.M., van Duijn, P. and van der Ploeg, M. (1985) Nature, 317, 175-177.20. Saiki, R.K., Scharf, S., Faloona, F., Mullis, K.B., Horn, G.T., Erlich, H.A. and

Arnheim, N. (1985) Science, 230, 1350-1354.21. Mullis, K.B. and Faloona, F.A. (1987) Methods in Enzymology, 155, 335-350.22. Landegren, U., Kaiser, R., Sanders, J. and Hood, L. (1988) Science, 241, 1077-1080.23. Barany, F. (1991) Proc. Natl. Acad. Sci. USA, 88, 189-193.24. Lyamichev, V.I., Mast, A.L., Hall, J.G., Prudent, J.R., Kaiser, M.W., Takova, T.,

Kwiatkowski, R.W., Sanders, T.J., de Arruda, M., Arco, D.A. et al. (1999) NatureBiotechnology, 17, 292-296.

25. Wallace, R.B., Schold, M., Johnson, M.J., Dembek, P. and Itakura, K. (1981) NucleicAcids Res, 9, 3647-3656.

26. Tyagi, S. and Kramer, F.R. (1996) Nature Biotechnology, 14, 303.27. Dubertret, B., Calame, M. and Libchaber, A.J. (2001) nature Biotechnology, 19, 365-

370.28. Nazarenko, I.A., Bhatnagar, S.K. and Hohman, R.J. (1997) Nucleic Acids Research,

25, 2516-2521.29. Landegren, U., Nilsson, M. and Kwok, P.-Y. (1998) Genome Research, 8, 769-776.30. Syvänen, A.-C. (2001) Nature Reviews Genetics, 2, 930-942.31. Shi, M.M. (2001) Clinical Chemistry, 47, 164-172.32. Cohen, J.B. and Levinson, A.D. (1988) Nature, 334, 119-121.

29

33. Conner, B.J., Reyes, A.A., Morin, C., Itakura, K., Teplitz, R.L. and Wallace, R.B.(1983) Proc Natl Acad Sci U S A, 80, 278-280.

34. Tyagi, S., Bratu, D.P. and Kramer, F.R. (1998) Nature Biotechnology, 16, 49-53.35. Wetmur, J.G. and Davidson, N. (1968) J. Mol. Biol., 31, 349-371.36. Wetmur, J.G. (1991) Critical reviews in biochemistry and molecular biology, 26, 227-

259.37. Maskos, U. and Southern, E.M. (1992) Nucleic Acids Research, 20, 1675-1678.38. Bains, W. (1994) Genet Anal Tech Appl, 11, 49-62.39. Guo, Z., Liu, Q. and Smith, L.M. (1997) Nature Biotechnology, 15, 331-335.40. Howell, W.M., Jobs, M., Gyllensten, U. and Brookes, A.J. (1999) nature

Biotechnology, 17, 87-88.41. Aoshima, T., Sekido, Y., Miyazaki, T., Kajita, M., Mimura, S., Watanabe, K.,

Shimokata, K. and Niwa, T. (2000) Clinical Chemistry, 46, 119-122.42. Hacia, J.G., Sun, B., Hunt, N., Edgemon, K., Mosbrook, D., Robbins, C., Fodor,

S.P.A., Tagle, D.A. and Collins, F.S. (1998) Genome Research, 8, 1245-1258.43. Alves, A.M. and Carr, F.J. (1988) Nucleic Acids Res, 16, 8723.44. Wu, D.Y. and Wallace, R.B. (1989) Genomics, 4, 560-569.45. Nilsson, M., Barbany, G., Antson, D.-O., Gertow, K. and Landegren, U. (2000)

Nature Biotechnology, 18, 791-793.46. Nilsson, M., Antson, D.-O., Barbany, G. and Landegren, U. (2001) Nucleic Acids

Research, 29, 578-581.47. Luo, J., Bergstrom, D.E. and Barany, F. (1996) Nucleic Acids Research, 24, 3071-

3078.48. Nickerson, D.A., Kaiser, R., Lappin, S., Stewart, J., Hood, L. and Landergren, U.

(1990) Proc. Natl., Acad. Sci. USA, 87, 8923-8927.49. Samiotaki, M., Kwiatkowski, M., Parik, J. and Landegren, U. (1994) Genimics.50. Iannone, M.A., Taylor, J.D., Chen, J., Li, M.-S., Rivers, P., Slentz-Kesler, K.A. and

Weiner, M.P. (2000) Cytometry, 39, 131-140.51. Kwok, P.Y., Deng, Q., Zakeri, H., Taylor, S.L. and Nickerson, D.A. (1996) Genomics,

31, 123-126.52. Nickerson, D.A., Taylor, S.L., Weiss, K.M., Clark, A.G., Hutchinson, R.G., Stengard,

J., Salomaa, V., Vartiainen, E., Boerwinkle, E. and Sing, C.F. (1998) Nat Genet, 19,233-240.

53. Syvänen, A.-C., Aalto-Setala, K., Kontula, K. and Söderlund, H. (1990) Genomics, 8,684-692.

54. Pastinen, T., Raitio, M., Lindroos, K., Tainola, P., Peltonen, L. and Syvänen, A.-C.(2000) Genome Research, 10, 1031-1042.

55. Shumaker, J.M., Metspalu, A. and Caskey, C.T. (1996) Hum. Mutat., 7, 346-354.56. Alderborn, A., Kristofferson, A. and Hammerling, U. (2000) Genome Research, 10,

1249-1258.57. Nyren, P., Pettersson, B. and Uhlen, M. (1993) Anal Biochem, 208, 171-175.58. Ronaghi, M., Karamohamed, S., Pettersson, B., Uhlen, M. and Nyren, P. (1996) Anal

Biochem, 242, 84-89.59. Ahmadian, A., Gharizadeh, B., Gustafsson, A.C., Sterky, F., Nyrén, P., Uhlén, M. and

Lundeberg, J. (2000) Analytical Biochemistry, 280, 103-110.60. Pourmand, N., Elahi, E., Davis, R.W. and Ronaghi, M. (2002) Nucleic Acids Res, 30,

e31.61. Wu, D.Y., Ugozzoli, L., Pal, B.K. and Wallace, R.B. (1989) Proc Natl Acad Sci U S

A, 86, 2757-2760.

30

62. Newton, C.R., Graham, A., Heptinstall, L.E., Powell, S.J., Summers, C., Kalsheker,N., Smith, J.C. and Markham, A.F. (1989) Nucleic Acids Research, 17, 2503-2516.

63. Holland, P.M., Abramson, R.D., Watson, R. and Gelfand, D.H. (1991) Proc. Natl.Acad. Sci. USA, 88, 7276-7280.

64. Livak, K.J., Flood, S.J., Marmaro, J., Giusti, W. and Deetz, K. (1995) PCR MethodsAppl, 4, 357-362.

65. Livak, K.J. (1999) Genet Anal, 14, 143-149.66. Mein, C.A., Barratt, B.J., Dunn, M.G., Siegmund, T., Smith, A.N., Esposito, L.,

Nutland, S., Stevens, H.E., Wilson, A.J., Phillips, M.S. et al. (2000) GenomeResearch, 10, 330-343.

67. Mein, C. (2002) Nucleic Acids Res, 30, e53.68. Tanke, H.J., Florijn, R.J., Wiegant, J., Raap, A.K. and Vrolijk, J. (1995) Histochem J,

27, 4-14.69. van Ommen, G.J., Breuning, M.H. and Raap, A.K. (1995) Curr Opin Genet Dev, 5,

304-308.70. Nederlof, P.M., van der Flier, S., Wiegant, J., Raap, A.K., Tanke, H.J., Ploem, J.S. and

van der Ploeg, M. (1990) Cytometry, 11, 126-131.71. Dauwerse, J.G., Wiegant, J., Raap, A.K., Breuning, M.H. and van Ommen, G.J.

(1992) Hum Mol Genet, 1, 593-598.72. Speicher, M.R., Gwyn Ballard, S. and Ward, D.C. (1996) Nat Genet, 12, 368-375.73. Raap, A.K. (1998) Mutation Research, 400, 287-298.74. Koch, J., Kolvraa, S., Pertersen, K., Gregersen, N. and Bolund, L. (1989)

Chromosoma, 98, 259-265.75. Koch, J. (1996) Methods: A companion in Enzymology, 9, 122-128.76. Nuovo, G.J., Hohman, R.J., Nardone, G.A. and Nazarenko, I.A. (1999) The Journal of

Histochemistry & Cytochemistry, 47, 273-279.77. Harren, T., Schwinger, E. and Mennicke, K. (2001) Human Mutation, 17, 131-140.78. Bobrow, M.N., Harris, T.D., Shaughnessy, K.J. and Litt, G.J. (1989) J Immunol

Methods, 125, 279-285.79. Raap, A.K., van de Corput, M.P.C., Verenne, R.A.W., van Gijlswijk, R.P.M., Tanke,

H.J. and Wiegant, J. (1995) Human Molecular Genetics, 4, 529-534.80. Speel, E.J.M., Hopman, A.H.N. and Komminoth, P. (1999) The Journal of

Histochemistry & Cytochemistry, 47, 281-288.81. Han, M., Xiaohu, G., Su, J.z. and Nie, S. (2001) nature Biotechnology, 19, 631-635.82. van de Rijke, F., Zijlmans, H., Li, S., Vail, T., Raap, A.K., Niedbala, R.S. and Tanke,

H.J. (2001) nature Biotechnology, 19, 273-276.83. Nilsson, M., Malmgren, H., Samiotaki, M., Kwiatkowski, M., Chowdhary, B.P. and

Landegren, U. (1994) Science, 265, 2085-2088.84. Nilsson, M., Krejci, K., Koch, J., Kwiatkowski, M., Gustavsson, P. and Landegren, U.

(1997) Nature Genetics, 16, 252-255.85. Antson, D.-O., Mendel-Hartvig, M., Landegren, U. and Nilsson, M. (2002) to be

submitted.86. Luo, J. and Barany, F. (1996) Nucleic Acids Research, 24, 3079-3085.87. Mullis, K.B. (1991) PCR Methods and applications, 1, 1-4.88. Landegren, U. and Nilsson, M. (1997) The Finnish medical Society DUODECIM, Ann

Med, 29, 585-590.89. Banér, J., Nilsson, M., Waldenström, E., Isaksson, A., Jarvius, J. and Landegren, U.

(2002) submitted.90. Antson, D.-O., Isaksson, A., Landegren, U. and Nilsson, M. (2000) Nucleic Acids

Research, 28, e58.

31

91. Kwiatkowski, M., Nilsson, M. and Landegren, U. (1996) Nucleic Acids Research, 24,4632-4638.

92. Myer, S.E. and Day, D.J. (2001) Biotechniques, 30, 584-588, 590, 592-583.93. Zhang, D.Y., Brandwein, M., Hsuis, T.C.H. and Li, H. (1998) Gene, 211, 277-285.94. Thomas, D.C., Nardone, G.A. and Randall, S.K. (1999) Arch Pathol Lab Med, 123,

1170-1176.95. Lizardi, P.M., Huang, X., Zhu, Z., Bray-Ward, P., Thomas, D.C. and Ward, D.C.

(1998) Nature Genetics, 19, 225-232.96. Banér, J., Nilsson, M., Mendel-Hartvig, M. and Landegren, U. (1998) Nucleic Acids

Research, 22, 5073-5078.97. Hatch, A., Sano, T.S., Misasi, J. and Smith, C.L. (1999) Genetic

Analasis:Biomolecular Engineering, 15, 35-40.98. Escudé, C., Garestier, T. and Hélène, C. (1999) Proc. Natl. Acad. Sci. USA, 96, 10603-

10607.99. Kuhn, H., Demidov, V.V. and Frank-Kamenetskii, M.D. (1999) Angew. Chem. Int.

Ed., 38, 1446-1449.100. Lehman, I.R. (1974) Science, 186, 790-797.101. Higgins, N.P. and Cozzarelli, N.R. (1979) Methods in Enzymology, 68, 50-71.102. Engler, M.J. and Richardson, C.C. (1982) The enzymes, XV, 3-29.103. Wu, D.Y. and Wallace, R.B. (1889) Gene, 76, 245-254.104. Kleppe, K., van de Sande, J.H. and Khorona, H.G. (1970) Proceedings of the National

Academy of Sciences, 67, 68-73.105. Fareed, G.C., Wilt, E.M. and Richardson, C.C. (1971) The Journal of Bioological

Chemistry, 246, 925-932.106. Fire, A. and Xu, S.-Q. (1995) Proc. Natl. Acad. Sci. USA, 92, 4641- 4645.107. Liu, D., Daubendiek, S.L., Zillman, M.A., Ryan, K. and Kool, E.T. (1996) J. Am.

Chem. Soc., 118, 1587-1594.108. Daubendick, S.L., Ryan, K. and Kool, E.T. (1995) J. Am. Chem. Soc., 117, 7818-7819.109. Zhou, Y., Calciano, M., Hamann, S., Leamon, J.H., Strugnell, T., Christian, M.W. and

Lizardi, P.M. (2001) Exerimental and Molecular Pathology, 70, 281-288.110. Schweitzer, B., Wiltshire, S., Lambert, J., O'Malley, S., Kukanski, K., Zhu, Z.,

Kingsmore, S.F., Lizardi, P. and Ward, D.C. (2000) PNAS, 97, 10113-10119.111. Gusev, Y., Sparkowski, J., Raghunathan, A., Ferguson, H., Montano, J., Bogdan, N.,

Schweitzer, B., Wiltshire, S., Kingsmore, S.F., Maltzman, W. et al. (2001) AmericanJournal of Pathology, 159, 63-69.

112. Schweitzer, B., Roberts, S., Grimwade, B., Shao, W., Wang, M., Fu, Q., Shu, Q.,Laroche, I., Zhou, Z., Tchernev, V.T. et al. (2002) nature Biotechnology, 20, 359-365.

113. Ladner, D.P., Leamon, J.H., Hamann, S., Tarafa, G., Strugnell, T., Dillon, D., Lizardi,P. and Costa, J. (2001) Laboratory Investiation, 81, 1079-1086.

114. Faruqi, F.A., Hosono, S., Driscoll, M.D., Dean, F.B., Alsmadi, O., Bandaru, R.,Kumar, G., Grimwade, B., Zong, S., Knott, T. et al. (2001) BMC Genomics, 2.

115. Kuhn, H., Demidov, V.V. and Frank-Kamenetskii, M.D. (2002) Nucleic AcidsResearch, 30, 574-580.

116. Nilsson, M., Gullberg, M., Dahl, F., Szuhai, K. and Raap, A.K. (2002) Nucleic AcidsResearch, 30, 1-6.

117. Nallur, G., Luo, C., Fang, L., Cooley, S., Dave, V., Lambert, J., Kukanski, S., Lasken,R. and Scheitzer, B. (2001) Nucleic Acids research, 29, e118.

118. Pickering, J., Bamford, A., Godbole, V., Briggs, J., Scozzafava, G., Roe, P., Wheeler,C., Ghouze, F. and Cuss, S. (2002) Nucleic Acids Research, 2002, e60.

32

119. Dean, F.B., Nelson, J.R., Giesler, T.L. and Lasken, R.S. (2001) Genome research, 11,1095-1099.

120. Dean, F.B., Hosono, S., Fang, L., Wu, X., Faruqi, F., Bray-Ward, P., Sun, Z., Zong,Q., Du, Y., Du, J. et al. (2002) PNAS, 99, 5261-5266.

121. Christian, A.T., Patte, M.S., Beth, C.M., Reed, B.E., Sorensen, K.J. and Tucker, J.D.(2001) PNAS, 98, 14238-14243.

122. Blanco, L., Bernard, A., Lázaro, J.M., Martin, G., Garmendia, C. and Salas, M. (1989)The Journal of Biological Chemistry, 264, 8935-8940.

123. Western, L.M. and Rose, S.J. (1991) Nucleic Acids Research, 19, 809-813.124. Pritchard, C.E. and Southern, E.M. (1997) Nucleic Acids Researchs, 25, 3403-3407.125. Rossi, R., Montecucco, A., Ciarrocchi, G. and Biamonti, G. (1997) Nucleic Acids

Research, 25, 2106-2113.126. Sriskanda, V. and Shuman, S. (1998) Nucleic Acids Research, 26, 525-531.127. Doherty, A.J. and Dafforn, T.R. (2000) J. Mol. Biol, 296, 43-56.128. Cherepanov, A.V. and de Vries, S. (2001) Biophysical Journal, 81, 3545-3559.129. Cherepanov, A.V. and de Vries, S. (2002) The Journal of Biological Chemistry, 277,

1695-1704.130. Doherty, A.J. and Suh, S.W. (2000) Nucleic Acids Research, 28, 4051-4058.131. Landegren, U. and Nilsson, M. (1997) Ann Med, 29, 585-590.132. Lorsch, J.R., Bartel, D.P. and Szostak, J.W. (1995) Nucleic Acids Research, 23, 2811-

2814.133. Podhajska, A.J. and Szylbalski, W. (1985) Gene, 40, 175-182.134. Kim, S.C., Podhajska, A.J. and Szybalski, W. (1988) Science, 240, 504-506.135. Raap, A.K., Vrolijk, M.J. and van der Ploeg, M. (1986) Cytometry, 7, 235-242.136. Kool, E.T. (1997) Chem. Rev., 97, 1473-1487.137. Walker, G.T., Little, M.C., Nadeau, J.G. and Shank, D.D. (1992) Proc. Natl. Acad.

Sci. USA, 89, 392-396.138. Wah, D.A., Hirsch, J.A., Dorner, L.F., Schildkraut, I. and Aggarwal, A.K. (1997)

Nature, 388, 97-100.139. Bitinaite, J., Wah, D.A., Aggarwal, A.K. and Schildkraut, I. (1998) Proc. Natl. Acad.

Sci. USA, 95, 10570-10575.140. Taylor, J.W., Schmidt, W., Cosstick, R., Okruszek, A. and Eckstein, F. (1985) Nucleic

Acids Research, 13, 8749-8764.141. Lutz, S., Ostermeier, M. and Benkovic, S.J. (2001) Nucleic Acids Res, 29, E16.142. Beigelman, L., Karpeisky, A., Matulic-Adamic, J., Haeberli, P., Sweedler, D. and

Usman, N. (1995) Nucleic Acids Res, 23, 4434-4442.143. Cummins, L., Graff, D., Beaton, G., Marshall, W.S. and Caruthers, M.H. (1996)

Biochemistry, 35, 8734-8741.144. Xu, Y. and Kool, E.T. (1998) Nucleic Acids research, 26, 3159-3164.145. Eng, C. and Vijg, J. (1996) Nature Biotechnology, 15, 422-426.146. King, M.C., Rowell, S. and Love, S.M. (1993) Jama, 269, 1975-1980.147. Hall, J.M., Lee, M.K., Newman, B., Morrow, J.E., Anderson, L.A., Huey, B. and

King, M.C. (1990) Science, 250, 1684-1689.148. Miki, Y., Swensen, J., Shattuck-Eidens, D., Futreal, P.A., Harshman, K., Tavtigian, S.,

Liu, Q., Cochran, C., Bennet, L.M., Ding, W. et al. (1994) Science, 266, 66-71.149. Couch, F.J. and Weber, B.L. (1996) Hum Mutat, 8, 8-18.150. Shattuck-Eidens, D., McClure, M., Simard, J., Labrie, F., Narod, S., Couch, F.,

Hoskins, K., Weber, B., Castilla, L., Erdos, M. et al. (1995) Jama, 273, 535-541.151. Hartge, P., Struewing, J.P., Wacholder, S., Brody, L.C. and Tucker, M.A. (1999) Am J

Hum Genet, 64, 963-970.

33

152. Kononen, J., Budendorf, L., Kallioniemi, A., Bärlund, M., Schraml, P., Leighton, S.,Torhorst, J., Mishatsch, M.J., Sauter, G. and Kallioniemi, O.-P. (1998) NatureMedicine, 4, 844-847.

153. Kallioniemi, O.-P., Wagner, U., Kononen, J. and Sauter, G. (2001) Human MolecularGenetics, 10, 657-662.

154. Håkansson, S., Jaohannsson, O., Johansson, U., Sellberg, G., Loman, N., Gerdes, A.-M., Holmberg, E., Dahl, N., Pandis, N., Kristofferson, U. et al. (1997) Am. J. Hum.Genet., 60, 1068-1078.

155. Staff, S., Nupponen, N.N., Borg, Å., Isola, J.J. and Tanner, M.M. (2000) Genes,Chromosomes & Cancer, 28, 432-442.

156. van de Rijke, F.M., Florijn, R.J., Tanke, H.J. and Raap, A.K. (2000) The Journal ofHistochemistry & Cytochemistry, 48, 743-745.

157. Zhong, X.-B., Lizardi, P.M., Huang, X.-H., Bray-Ward, P.L. and Ward, D.C. (2001)PNAS, 98, 3940-3945.

158. Hafner, G.J., Yang, I.C., Wolter, L.C., Stafford, M.R. and Giffard, P.M. (2001)Biotechniques, 30, 852-856, 858, 860 passim.

159. Zhang, D.Y. (2001) Gene, 274, 209-216.160. Drysdale, C.M., McGraw, D.W., Stack, C.B., Stephens, J.C., Judson, R.S.,

Nandabalan, K., Arnold, K., Ruano, G. and Liggett, S.B. (2000) PNAS, 97, 10483-10488.

161. Schena, M., Shalon, D., Davids, R.W. and Brown, P.O. (1995) Science, 270, 467-470.162. Pinkel, D., Segraves, R., Sudar, D., Clark, S., Poole, I., Kowbel, D., Collins, C., Kuo,

W.-L., Chen, C., Zhai, Y. et al. (1998) Nature Genetics, 20, 207-211.163. Stimpson, D.I., Cooley, P.W., Knepper, S.M. and Wallace, D.B. (1998) Biotechniques,

25, 886-890.164. Landegren, U., Khorlin, A., Mendel-Hartvig, M. and Öhman, O. (1995) Sweden patent

no. PCT/SE95/01420.