Tetrahedron Computer Methodology Vol. 1, No. 3, pp. 237 to 246, 1988 0898 5529/89 $3.00+.00 Pnnted in Great Britain. © 1989 Pergamon Press plc

Computer-Assisted Automated Synthesis I: Computer-Controlled Reaction of Substituted

N-(Carboxyalkyl)amino Acids

Nobuyoshi Hayashi* and Tohru Sugawara

Central Research Division, Takeda Chemical Industries, Ltd., Juso Honmachi 2-chome, Yodogawa-ki, Osaka, 532, JAPAN

Received 17 October 1988, Revised 20 March 1989, Accepted 20 March 1989

Key words: Automated synthesis; Kinetic equations; Reaction control; Optimum conditions; Substituent effect

Abstract" Kinetic equations for controlling the consecutive reactions in the reduction of a Schiff base, formed by the equilibrium reaction of an amino acid with a 2-keto acid, have been developed. These equations have been integrated into a computer control program for automated synthesis of substituted N-(carboxyalkyl)amino acids. The equations were especially useful for obtaining optimum reaction conditions and on-line reaction control.

INTRODUCTION

New pharmaceuticals are often developed by introducing a large number of different substituents onto a particular skeleton which is known or expected to be biologically active. Although automatic systems for a batch-type reactors are available 1-3 and have reduced the man-power intensive nature in pharmaceutical research, quantitative prediction of the optimum reaction conditions is difficult, and these must be pre-determined by analytical techniques such as high performance liquid chromatography (HPLC). 4 Multi-step reactions, especially those with unstable or undetectable intermediates, are particularly difficult to optimize for automation.

We have recently developed an automated apparatus, equipped with a computer software, which can synthesize substituted N-(carboxyalkyl)amino acids. This paper deals with the development of the computer-assisted reaction control and the kinetic equations.

RESULTS AND DISCUSSION

The reaction of an amino acid tea-butyl ester acetic acid salt (1) with a 2-keto acid sodium salt (2) produces an unstable intermediate, Schiff base (3), which is reduced with sodium cyanoborohydride to give a substituted N-(carboxyalkyl)amino acid tea-butyl ester sodium salt (4).

237

238 N. HAYASHI and T. SUGAWARA

R - C H - N H 2 H O A c O = C - R '

OzBu t + I CO2Na

2

R - C H - N H - C H - R " I I

ButO2 C CO2Na

4

kl R - C H - N = C - R "

k_ 1 ButO2 / CO2Nal + H20 + AcOH

NaBH3CN

A. Method for Reaction Control s When the initial concentrations of 1 and 2 are identical, let their initial concentration be defined as A

and the Schiff base concentration at time t be X. The kinetic equation for the Schiff base formation can then be derived as follows:

dX kl(A_J02 k_.iX2 ~ - =

= [A~J~I- (~rT+ ~r~l)X] [A/~I - (/'k'~l- ~r'~l)X ]

f dX i A ~ l - ( ~ 1 + ~/'k~_l)X ] [ A/'k'~l - ( ~ 1 - ~ ) X ]

f dt

2 A k l ~ : 1 In A/k'l'l - (/k'~l - ~/-~11)X (1) t A~I - (~ l l+~l l )X

where k 1 and k_ 1 are rate constants in the forward and the reverse directions, respectively. If E is defined as the concentration of the Schiff base at equilibrium, the relationship between A and E is given by Eq. 2,

A -(1 +~-K~E -t J

(2)

where K is the equilibrium constant. Insertion of Eq. 2 in Eq. 1 gives rise to Eq. 3.

2Akl 1 1 I E + X ~ ' ) = - - In +

f f , 1 -7-g e - x (3)

Let concentrations of the Schiff base at time t 1 and t 2 be defined as X 1 and X 2, respectively. An equation can then be derived as follows:

/ +Xl / ( / t21n - + ~ - t l ln E + X 2 + ~ - ( t 2 - t l ) l n ( l + 4 " ~ = 0

X 1 / 5 - - X 2 (4)

Although the molar absorption coefficient of the Schiff base is unknown, it is possible to calculate the equilibrium constant using Eq. 4 based on measurement of optical absorbances as a function of reaction time.

Computer-assisted automated synthesis--I 239

After K is calculated, k t can be.obtained using Eq. 3. A computer program for calculating Eqs. 3 and 4 has tus been developed.

Since the reduction rate of keto acids is slow, while that of the Schiff base is rapid, compared to its formation rate, the formation of 4 may be controlled by the addition rate of sodium cyanoborohydride. Let te initial volume of the reaction mixture be defined as V 0, and the addition volumes of each aliquot of the ~ucing agent be V 1, V 2, and V i, respectively. The concentration of sodium cyanoborohydride is y, a is ~efined as X/E and 13 is the ratio of the stoichiometric coefficient of reduction (NaBH3CN/Schiff base). The 5chiff base concentration (SC1) before the first addition of the reducing agent is expressed by Eq. 5, and the first addition volume of reducing agent can be formulated on the basis of the stoichiometry. See Table 1 for te derivations of these equations.

SCI : t ( ' )

a~ A Vo vI-K V~ = (6)

y(1 +I/K)

The concentration of unreacted substrate (A1) is given by Eq. 7, which is corrected for the dilution caused by the first addition of the reducing agent.

A 1 = (1

Thus, the general equation of V i is written by

a~-K.) AV 0

1 + ~ Vo+V 1 (7)

Vi= y(1 +[K) (.1 1+ ~ : J (8)

where i denotes the number of the addition (for i = 1 to n). The total addition volume (V t) of reducing agent is '~en represented by Eq. 9.

V t = E V i (9) i=1

The reaction time before the first addition of reducing agent, t 0, is derived from Eq. 3.

to = + (10) 2Akl l + l / K - a

The addition interval (ti) after the first addition of reducing agent is given by

r l

i = t 1 l + a

2A.1 V o 1 - a (11)

and the total reaction time (T) is then obtained by the summation of t o and t i.

2 4 0 N . H A Y A S H I a n d T . S U G A W A R A

II

~g

0

¢.)

0

0 >

0

0

t'¢3 II

o ~

II

II ° ~

li

II

II

+-' c: 0 ~-~

~ N . .

7 .O-

I

..O-

+

+

.O- I

v O -

Il

~z

.O- I

v -O-

II

7 °V,I

..O-

II

+

+

II

ee~

7 .O-

I v ..O-

~I ~ II

4 ) - I

.O-

~ I ~ II

-o- + I

v -O-

e,I ~ II

O

v,..a

..O-

~ I ~, II

I ¢-4

II

.¢

I

II

::.<-

+

. .O- I

x_.,,

II

+

+

e,i O -

I

II

t O -

o

8~ "t::t

+

+

O - I

II

Computer-assisted automated synthesi~I 241

n

T=to + ~ ti i = 1

(12)

B. Relationship between K (or kl) and the Sterie Constant It is well known that the formation of a Schiff base by the reaction of an amine with a carbonyl

compound exhibits a bell-shaped pH-rate curve and proceeds in two steps, an addition step and a dehydration step that occurs by an acid-catalyzed pathway. 6,7 At neutral pH the equilibrium concentration of the carbinolamine addition compound is increased by electron-withdrawing substituents, but the rate of acid-catalyzed dehydration of the addition compound is increased by electron-donating substituents. Thus the observed rate, which is a resultant of these two steps, shows very little sensitivity to substituents, s'9 However substituted N-(carboxyalkyl)amino acids often have aliphatic substituents which make steric effects more significant.

In order to predict the pH in a methanolic reaction mixture, some typical amino acid tea-butyl ester acetic acid salts were tested. The range of pK a of I measured in 90% methanol was 7.5-7.9, and this was smaller than the range in aqueous solution as shown in Table 2. Thus the mixture of 1 and 2 in methanol was estimated to be an approximately neutral buffer solution.

Table 2. pK a Values of Amino Acid Tert-butyl Ester Acetic Acid Salts.

Amino acid tert-butyl ester pK a acetic acid salt in 1-120 in 90% MeOH

Alanine 8.30 7.90 Leucine 8.00 7.80 Norvaline 8.15 7.90 Threonine-O-tert-butyl ether 7.30 7.60 Phenylalanine 7.55 7.50 Glutamic acid 7.55 7.50

After I was mixed with 2 in methanol, optical absorbances of the Schiff base formed were measured at 24°C (UV-235 nm). The K and kl were calculated using Eqs. 4 and 3, respectively. Figs. 1 and 2 show logarithmic plots for over-all equilibrium and forward rate constants against the steric constants of the 2-substituents in 2. The steric constants used were E s values for alkyl components of ester hydrolysis. 1° The /( and k I values for the Schiff base formation were fout~d to increase with increasing steric constants in a linear manner, but no significant influence of 1 was found in the series of tert-butyl ester acetic acid salt of alanine, norvaline, valine, leucine, isoleucine, phenylalanine and glutamic acid. The rough linear relationship in Fig. 1 agrees, not with E s values for acyl components of ester hydrolysis, 11 but with E s values for alkyl components. This result suggests that the steric influence of substituents in the reaction of 1 with 2 is large on the branched a-carbon and small on the branched 13-carbon. Since the structural difference in the amino acids used appears at the B-position from the nitrogen atom, the relative difference of the steric influence may be compensated in errors of measurement and in the large steric influence on the branched a-carbon of amino acids. The reaction constants (p) were calculated by the least squares method and determined to be 9=0.614 for K and 9=2.607 for k 1, respectively. Thus the relationships between K (or k 1) steric constant were expressed by Eqs. 13 and 14.

K = 0.47 exp (0.614 E s / 0.4343) (13)

K = 9.57 exp (2.607 E s /0.4343) (14)

242 N. HAYASHI and T. SUGAWARA

100

.~50

4 0 J

(a)

/ / - - - - . . . . . . . . . . . . . . . .

L/ . . . . . . . . . . . . . . . . . . . ; ; , I

/ i

I

1

60 120 Reaction time (min)

180

Figure 1. Logarithmic plots for K against steric constants of keto acid substituents.

1.0

~ 0.5 ~'

Bu n o

pr I

Bu s 0.1

0.05 I i , , | I i i m

-1.0 -0.5 0.0 Sterlc constant

Figure 2. Logarithmic plots for k I against steric constants of keto acid substituents.

C. The Dropping Rate of Reducing Agent The dropping rate of the reducing agent was calculated using Eqs. 8, 10 and 11. As a typical example,

the computed result for the controlled reaction of norvaline tert-butyl ester acetic acid salt with 2- ketoisocaproic acid sodium salt in methanol is shown in Table 3. The optimum mole ratio of NaBH3CN:Schiff base for the reduction of the Schiff base was found to be 2:3, and ~t in Eq. 8 was empirically set at 0.95 for convenience. It is not clear why the optimum ratio of NaBH3CN:Schiff base differs from the stoichiometric ratio of 1:3.12 To judge the effectiveness of the reaction control, a rate study

Computer-assisted automated synthesis--I 243

was performed with solutions of the same concentration as those in Table 3, using an HPLC analytical procedure. Chemical yield curves of the product, N-(3-methyl-l-sodioxycarbonylbutyl)norvaline tea-butyl ester, under various reaction conditions are presented in Fig. 3.

Table 3. Reaction Control table of Narvaline Tea-butyl Ester Acetic Acid Salt With 2-Ketoisocaproic Acid Sodium Salt in Methanol.

Input data Time (min) Reagent volume (ml)

Equilibrium constant (K): 0.25 Coefficient (ct =X/E): 0.95

Forward-rate constant (kl): 0.64 (1.mol-l.min -1)

RAW MATERIAL Initial concentration: 0.27

(mol.1 "1) Initial volume (ml): 15.0 REDUCING AGENT Concentration (mol.l-1): 0.2 Optimum mole ratio: 0.67

(NaBH3CN/Schiff base)

4.7 4.3 13.6 2.9 28.5 2.0 52.4 1.4 89.3 0.9

145.2 0.6 229.1 0.4 353.7 0.3 538.2 0.2 810.0 0.1

The dotted line (c) indicates the chemical yield versus reaction time when sodium cyanoborohydride was added to the reaction mixture at a rate of 0.23 ml.min "1. The dotted line (d) shows the chemical yield versus reaction time when the reactants and sodium cyanoborohydride were mixed at once. When the addition rate of sodium cyanoborohydride was rapid compared to the formation rate of the Schiff base, the yield (dotted line (d)) levelled off at relatively low value. 13 This fact indicates that the formation of the Schiff base was interfered with by the large excess of sodium cyanoborohydride, and it is suggested that a competitive equilibrium, with a complex formed between the keto acid and the sodium cyanoborohydride, may be responsible. However, the yield curve for the reaction controlled by the kinetic approach in Table 2, the dotted line (b), was similar to the calculated yield presuming consumption of a stoichiometric amount of reducing agent, the solid line (a) in Fig. 3. It is thus clear that our computer-control program, based on a kinetic approach, is of great use for on-line reaction control to maintain optimum reaction conditions.

D. Computer Program The computer program for the automated synthesis apparatus was written primarily in NEC

N88-BASIC(86). The physical constants were stored on a magnetic floppy disk; with the equilibrium and f0rward-rate constants, for use in the reaction control, being referred to by the substituent name of the 2-keto acid. If a physical constant was not already recorded, the information such as K and k 1, a steric constant, or experimental data of optical absorbances as a function of reaction time, was requested for input and then recorded. If necessary, K and k 1 values are computed using the calculation subroutines and stored with the appropriate substituent name on the magnetic floppy disk. The user inputs information for reaction control through dialogue with the interactive program prior to beginning operation, and then all experimental operations proceed automatically.

244 N. HAYASHI and T. SUGAWARA

;Y p i / ] Bun

l Bu I

- 1 . 0 | i i

- 0 . 5

10.0

5.0

1.0

0.5

0.1

0.05

0.0

!

4 I •

"6 B

° °

r - I

¢fl

0 0

m

I ' 0

m

0 k ~

Steric constant

Figure 3. Chemical yield curves under various reaction conditions for the reaction of norvaline tert-butyl with 2 ketoisocaproic acid sodium salt. Solid line (a): calculated yield presuming consumption of the stoichiometric amount of reducing agest. Dotted lines (b), (c) and (d): observed yields.

CONCLUSIONS

The equations 8-14, incorporated in a computer program, have been applied to control an automated synthesis, and all synthetic processes were automatically performed from the mixing of reactants to the isolation of the pure products as powders. The automated apparatus has been run continuously for 24 hours, giving an average rate of synthesis of three compounds per day. 14,15

EXPERIMENTAL

Analytical Procedure M.p.'s were determined with a Koefler hot-stage apparatus. I.r. and u.v. spectra were measured with a

Hitachi 260-10 and a Hitachi 557 spectrometers, respectively. 1H n.m.r, spectra were recorded using JEOL JNM-GX-400 FT(400 MHz) spectrometer, tetramethylsilane as an internal standard. Absorbances of the Schiff base (intermediate) formed were measured at g 235 nm using an optical waveguide u.v. spectrum analyzer, model 100-3, Guided Wave Inc. The instrument used for the chromatographic analysis was a Shimadzu LC-5A HPLC and operated as follows: wavelength g 235 nm, column Nucleosil 5C-18 (4 x 250 mm, M.Nagel), flow rate 0.7 ml-min "1, mobile phase 0.05 M sodium acetate: methanol=3:7 v/v, chart speed 2.5 mm-min -1. The pK a values of 1 were measured with a Radiometer TTT-80 titrator.

Computer-assisted automated synthesis--I 245

,V.(3-Methyl-l-sodioxycarbonylbutyl)norvaline tert-butyl ester (5) Under computer control, racemic norvaline tert-butyl ester acetic acid salt (932 mg, 4 mmol) in

methanol (5 ml) and 2-ketoisocaproic acid sodium salt (608 mg, 4 mmol) in methanol (10 ml) were mixed. The addition rate of the methanolic solution of sodium cyanoborohydride (0.2 mol-1-1) was controlled as shown in Table 3. The consecutive reaction was terminated by the computer program, when the chemical yield decreased by less than 0.1% per minute. Thus the reaction was stopped after 146 minutes. After evaporation to dryness under reduced pressure, 0.8 mol solution of sodium hydroxide (10 ml) was added to ~e residue and it was injected into the purification unit equipped with a XAD-2 column (2 x 50 cm). The eluent (mobile phase, water:methanol=l:l; flow rate, 6 ml.min "1) was monitored with an u.v. detector (~. 235 nm) and automatically separated into two fractions, which were identified as diastereoisomers, 5a and 5b, at ~tention times 52.0 and 62.3 minutes, respectively. The two fractions eluted were freeze-dried. The combined synthetic yield of both 5a and 5b was 78%. The diastereoisomers, 5a and 5b, were further purified in a similar manner as described above, and analyzed.

For 5a: m.p. 145-147°C (Found: C, 56.94; H, 9.03; N, 4.50%. C15H2804N Na 0.41-120 requires C, 56.91; H, 9.17; N, 4.42%); Vmax.(KBr): 3420 (NH), 1725 (ester CO), and 1590 cm-l(COO-); ~.(H20): 225 and 235 nm(¢ 367 and 203 dm-3mol.-lcm-1); 8r1(400 MHz,D20): 0.905 (3H, d, J 6.6, Me(Me)CH), 0.918 (3H, d, J 6.6, Me(Me)CH), 0.910 (3H, t, J 7.2, MeCH2), 1.34 (2H, m, MeCH2), 1.390 (2H, t, J 7.3, CH2CH(Me)2), ca. 1.6 (1H, m, CH(Me)2), ca. 1.63 (2H, m, ETCH2), 3.038 (1H, t, J 7.1, CHCO2Na), 3.151 (1H, t, J 6.8, CHCO2But), 1.493 (9H, s, But).

For 5b: m.p. 154-156°C (Found: C, 57.03; H, 9.22; N, 4.60%. C15H2804 N Na 0.4 1-120 requires C, 56.91; H, 9.17; N, 4.42%); vmax.(KBr): 3420 (NH), 1720 (ester CO), and 1590 (COO-)cm-1; ~.(H20): 225 and 235 nm (~ 246 and 104 dm.3mol.-lcm.-1); 8 H (400 MHz; D20): 0.912 (3H, t, J 7.3, MeCH2), 0.915 (6H, d, J 6.6, (Me)2CH), ca. 1.3 (2t-1, m, MeCH2), ca. 1.42 (2H, m, CH2CH(Me)2 ), 1.63 (2H, m, ETCH2), 1.477 (9H, s, But), ca. 1.62 (1H, m, CH(Me)2), 3.086 (1H, t, J 7.1, CHCO2Na), 3.215 (1H, t, J 6.5, CHCO2But).

REFERENCES

1. Winicov, H.; Schainbaum, J.; Buckley, J.; Longino, G.; Hill, J.; Berkoff, C.E. "Chemical Process Optimization by Computer - A Self-directed Chemical Synthesis System". Anal. Chim. Acta, 1978, 103, 469-476.

2. Legrand, M.; Bolla, P. "A Fully Automatic Apparatus for Chemical Reactions on the Laboratory Scale". J. Auto. Chem., 1985, 7, 31-37.

3. Chodosh, D. F.; Wdzieckowski, F. E.; Shainbaum, J.; Berkoff, C. E. "Automated Chemical Synthesis. Part 2: Interfacing Strategies". J. Auto. Chem., 1983, 5, 99-102.

4. Frisbee, A. R.; Nantz, M. H.; Kramer, G. W.; Fuchs, P. L. "Robotic Orchestration of Organic Reactions: Yield Optimization via an Automated System with Operator-specified Reaction Sequences". J. Am. Chem. Soc., 1984, 106, 7143-7145.

5. Hayashi, N.; Sugawara, T. "Computer-Controlled Reaction of Substituted N-(Carboxyalkyl)amino Acids". Chem. Lett., 1988, 1613-1616.

6. Westheimer, F. H. "Semicarbazone Formation in Sixty Per Cent Methylcellosolve". J. Am. Chem. Soc., 1934, 56, 1962-1965.

7. Jencks, W. P. "Studies on the Mechanism of Oxime and Semicarbazone Formation". J. Am. Chem. Soc., 1959, 81,475-481.

8. Anderson, B. M.; Jencks, W. P. "The Effect of Structure on Reactivity in Semicarbazone Formation". J. Am. Chem. Soc., 1960, 82, 1773-1777.

9. Cordes, E. M.; Jencks, W. P. "On the Mechanism of Schiff Base Formation and Hydrolysis". J. Am. Chem. Soc., 1962, 84, 832-837.

10. Jones, R. W. A.; Thomas, J. D. R. "Steric Influence of the Alkyl Component in the Alkaline Hydrolysis of Acetate and Propionates". J. Chem. Soc., 1966, (B), 661-664.

246 N. HAYASHI and T. SUGAWARA

11. Taft, R. W. "Polar and Steric Substiment Constants for Aliphatic and o-Benzoate Groups from rates of Esterification and Hydrolysis of Esters". J. Am. Chem. Soc., 1952, 74, 3120-3128.

12. Ohfune, Y.; Kurokawa, N.; Higuchi, N.; Saito, M.; Hashimoto, M.; Tanaka, T. "An Efficient One-step Reductive N-monoalkylation of a-amino acids". Chem. Lett., 1984, 441-444.

13. Although the yield levelled off, it slowly increased while the reaction mixture was standing overnight. 14. Hayashi, N.; Sugawara, T.; Shintani, M.; Kato, S. "Computer-Assisted Automated Synthesis II:

Development of a Fully Automated Apparatus for Preparing Substituted N-(Carboxyalkyl)amin0 Acids". Submitted to J. Auto. Chem..

15. Hayashi, N.; Sugawara, T.; Kato, S.; Shintani, M. "Computer-Assisted Automated Synthesis IlI: Automated Synthesis of Substituted N-(Carboxyalkyl)amino Acids". In preparation.