Anal. Chem. 1996, 68, 2361-2365
Racemization of Amino Acids in Solid-Phase Peptide Synthesis Investigated by Capillary Electrophoresis Daniel Riester,†,§ Karl-Heinz Wiesmu 1 ller,‡ Dieter Stoll,‡ and Reinhard Kuhn*,†
Institut fu¨ r Angewandte Forschung, FH Reutlingen, Alteburgstrasse 150, 72762 Reutlingen, Germany, and Naturwissenschaftliches und Medizinisches Institut, Eberhardstrasse 29, 72762 Reutlingen, Germany
The rate of racemization during solid-phase peptide synthesis was studied using capillary electrophoresis and 18-crown-6 tetracarboxylic acid as chiral selector. For this purpose, the tripeptide D-Tyr-L-Lys-L-Trp as a model compound was synthesized by solid-phase peptide synthesis. A separation method based on capillary electrophoresis was developed which allowed all eight optical isomers of the tripeptide to be separated in a single run. The separation method was validated and was found to be well suited for purity analysis, with a limit of detection of 0.05% of the major compound. The method was revealed to be highly sensitive even to small variations in the buffer pH. Capillary electrophoresis was also employed to prove the enantiomeric purity of the Fmocprotected amino acids used for peptide synthesis. A separation method based on micellar electrokinetic chromatography and γ-cyclodextrin was developed for this purpose. The formation of optical isomers during peptide synthesis was investigated in the final product without hydrolyzing the tripeptide. This strategy allowed the rate of racemization to be determined by activation of amino acids in coupling cycles and cleavage of the peptides from the resin and from side-chain protecting groups. The formation of stereoisomers could be verified and was 0.4% or less per synthesis cycle. The experimental data agreed well with theoretical considerations, showing that racemization takes place mainly at the carboxy-activated amino acid during coupling. Synthetic peptides are of growing importance for pharmaceutical or biological applications. Since the pioneering work of Merrifield,1 solid-phase peptide synthesis (SPPS) has been developed into a powerful tool for the synthesis of even big amounts of longer peptides. The problem of the purity of these synthetic peptides has been discussed in detail elsewhere.2 So-called deletion peptides, with one or more missing amino acids in the peptide sequence, often due to incomplete coupling, and sidechain reactions caused by cleavage from the resin and from protecting groups are the main reason for impurities in solid-phase †
Institute fu ¨ r Angewandte Forschung. Naturwissenschaftliches und Medizinisches Institut. § Present address: EVOTEC Biosystems, Grandweg 64, 22529 Hamburg, Germany. (1) Merrifield, R. B. J. Am. Chem. Soc. 1963, 85, 2149-2154. (2) Bayer, E. Angew. Chem. 1991, 103, 117-133; Angew. Chem., Int. Ed. Engl. 1991, 30, 113-129. ‡
S0003-2700(95)01151-6 CCC: $12.00
© 1996 American Chemical Society
peptide preparations.3 Also, racemization or isomerization of amino acids during SPPS has been observed frequently but is very difficult to detect. Thus, the requirement for enantiomerically pure peptides has sustained the development of efficient and sensitive analytical tools. Chromatographic techniques such as HPLC are well established and have proven to be efficient for the characterization of synthetic peptides.4,5 Studies on racemization require separation techniques for the discrimination of enantiomers. Armstrong and co-workers6-8 separated protected amino acids and peptides by HPLC using macrocyclic glycopeptide antibiotics and cyclodextrinbonded stationary phases. Chiral separations of di- and tripeptides by HPLC were also reported by Cram’s group9 and Esquivel et al.,10 using chiral crown ethers as stationary phases, as well as by Pirkle and co-workers,11 who used Pirkle-type stationary phases. Even though a huge number of stationary phases for liquid chromatography are commercially available, which solves most separation problems, method optimization can be time-consuming, and the results occasionally show poor efficiency. Recently, capillary electrophoresis (CE) has been introduced as a powerful separation technique for the resolution of peptides.12-14 We described enantio-separations of di- and tripeptides by CE using an optically active crown ether as chiral selector.15,16 In this study, we describe the resolution of eight optical isomers of a tripeptide. This method was used to investigate the rate of racemization during solid-phase peptide synthesis. (3) Metzger, J. W.; Kempter, C.; Wiesmu ¨ ller, K.-H.; Jung, G. Anal. Biochem. 1994, 219, 261-277. (4) Hearn, M. T. W., Regnier, F. E., Wehr, C. T., Eds. High Performance Liquid Chromatography of Proteins and Peptides; Academic Press: London, 1983. (5) Lottspeich, F., Henschen, A., Hupe, K., Eds. High Performance Liquid Chromatography in Protein and Peptide Chemistry; De Gruyter: Berlin, 1981. (6) Zukowski, J.; Pawlowska, M.; Nagatkina, M.; Armstrong, D. W. J. Chromatogr. 1993, 629, 169-179. (7) Armstrong D. W.; Liu, Y.; Ekborgott, K. H. Chirality 1995, 7, 474-497. (8) Hilton, M.; Armstrong, D. W. J. Liq. Chromatogr. 1991, 14, 3673-3683. (9) Cram, D. J. Science 1988, 240, 760-762. (10) Esquivel, B.; Nocholson, N.; Peery, L.; Fazio, M. HRC&CC-J. High Resolut. Chromatogr. Chromatogr. Commun. 1991, 14, 816-823. (11) Pirkle, W. H.; Alessi, D. M.; Hyun M. H.; Pochapsky, T. J. Chromatogr. 1987, 398, 203. (12) Gaus, H.-J.; Beck-Sickinger, A. G.; Bayer, E. Anal. Chem. 1993, 65, 13991405. (13) Kuhn, R.; Hoffstetter-Kuhn, S. Capillary Electrophoresis: Principles and Practice; Springer: Berlin, 1993. (14) Grossmann, P. D.; Colburn, J. C.; Lauer, H. H. Anal. Chem. 1989, 61, 11861194. (15) Kuhn, R.; Erni, F.; Bereuter, T.; Ha¨usler, J. Anal. Chem. 1992, 64, 28152820. (16) Kuhn, R.; Riester, D.; Fleckenstein, B.; Wiesmu ¨ ller, K. H. J. Chromatogr. A 1995, 716, 371-379.
Analytical Chemistry, Vol. 68, No. 14, July 15, 1996 2361
EXPERIMENTAL SECTION Instrumentation. Experiments were carried out using a HP 3DCE instrument (Hewlett-Packard, Waldbronn, Germany) or an ABI 270A capillary electrophoresis system (Perkin-Elmer/Applied Biosystems, Foster City, CA). If not otherwise stated, separations were performed in untreated, open-tube fused silica capillaries (50 cm (effective length) × 75 µm i.d. or 50 µm i.d.), applying a potential of 30 kV. The capillary temperature was maintained constant at 30 °C. For the preparation of the buffer, a solution of 30 mM Tris and 10 mM 18C6H4 was titrated to pH 2.0 with citric acid. An appropriate amount of each sample (usually 0.05% w/v) was dissolved in water by sonication and injected by pressure for 2 s at 50 mbar (HP 3DCE) or vacuum for 2-4 s at 5 in. Hg (ABI 270A). The sample was detected by UV absorbance at 210 nm. The elution order of the enantiomers was determined by spiking the mixture with the pure enantiomers. Chemicals. All reagents were of analytical grade if not stated otherwise. Diethyl ether, dimethylformamide (DMF), methanol, tert-butyl alcohol, sodium dodecyl sulfate (SDS), piperidine, and disodium tetraborate were from Merck (Darmstadt, Germany). Acetic acid, citric acid, 18-crown-6-tetracarboxylic acid (18C6H4, purity > 98%), γ-cyclodextrin, dichloromethane (DCM), diisopropylcarbodiimide (DIC), 1-hydroxybenzotriazole (HOBt), phenol, thioanisole, trifluoroacetic acid, tris(hydroxymethyl)aminomethane (Tris), sodium hydroxide, and urea were from Fluka (Buchs, Switzerland). p-(Benzyloxy)benzyl alcohol resin was from Rapp Polymere (Tu¨bingen, Germany). N-R-(9-Fluorenylmethoxycarbonyl)-O-tert-butyl-L-tyrosine (Fmoc-L-Tyr(tBu)), N-R-(9-fluorenylmethoxycarbonyl)-N--(tert-butyloxycarbonyl)-L-tryptophan (FmocL-Trp(Boc)), and N-R-(9-fluorenylmethoxycarbonyl)-N--(tertbutyloxycarbonyl-L-lysine (Fmoc-L-Lys(Boc)) were purchased from Novabiochem (La¨ufelfingen, Switzerland). Fmoc-D-Tyr(tBu), FmocD-Lys(Boc), and Fmoc-D-Trp were obtained from Neosystem Laboratoire (Strasbourg, France). All amino acids were certified by the suppliers to be enantiomerically pure, with a limit of detection of at least 0.1%. In addition, the enantiomeric purity of all protected amino acids was tested by capillary electrophoresis. No chiral impurities could be detected in these samples. Peptide Synthesis. Eight stereoisomers of the tripeptide TyrLys-Trp, containing both D- and L-amino acids, were simultaneously synthesized17 utilizing a robotics system equipped for multiple peptide synthesis (Syro, MultiSynTech, Bochum, FRG), following the “Fmoc strategy”. Preloaded (p-benzyloxy)benzyl alcohol resins (30 mg, loaded with 0.41 mmol/g Fmoc-L-Trp(Boc) or 0.60 mmol/g Fmoc-D-Trp) were added to 2 mL polypropylene filter tubes. Side-chain protection groups were tBu (tyrosine) and Boc (lysine, L-tryptophan). Fmoc deprotection was achieved by treating the resins twice with piperidine (40% in DMF) for 3 and 12 min, followed by seven washings with DMF. DIC (1.5 M in DMF) was added for activation in order to facilitate the coupling of amino acids. Solutions of 0.5 M Fmoc-L-amino acids or 0.5 M Fmoc-Damino acids were dissolved in 0.5 M HOBt in DMF. Couplings were carried out using a 5-fold excess of amino acids and coupling reagents within 50 min allotted reaction time. The tripeptides were cleaved from the resins, and the side chains were deprotected by adding trifluoracetic acid (1 mL) containing 5% scavenger (thioanisole/1,2 ethanedithiole/phenol, 10:5:15 v/v/w). After 30 min, the cleavage solution was filtered from the resin and was
stored for an additional 2 h in order to ensure complete sidechain deprotection. The peptides were precipitated at -20 °C by the addition of 5 mL of diethyl ether/n-heptane (1:4 v/v), washed twice, and lyophilized. The identity and purity of the peptides were verified by mass spectrometry (Sciex API III, Thornhill, Canada) and RP-HPLC (Gynkotek, Germering, FRG). Purities of higher than 92% were obtained for peptides prepared on L-Trp(Boc) resins, and purities higher than 65% were determined (HPLC) for the tripeptides from the D-Trp resins. Loading of the Resin. The coupling of Fmoc-L-Trp(Boc) (1.2 g, 2.4 mmol) or Fmoc-D-Trp (1.02 g, 2.4 mmol) to p-(benzyloxy)benzyl alcohol resin (1 g, 0.8 mmol) was carried out in 3-fold excess with 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium tetrafluoroborate (TBTU; 770 mg, 2.4 mmol) and a 4-fold excess of diisopropylethylamine (DIPA; 0.55 mL, 3.2 mmol) in the presence of 1-hydroxybenzotriazole (0.37 g, 2.4 mmol). The resin was washed in a polypropylene filter tube three times with DMF (6 mL each), and Fmoc-amino acid, HOBt, and TBTU were added as solid powders to the swollen resins. Next, 0.5 mL of DIPA and 1 mL of DMF were added under vigorous shaking to obtain a viscous suspension, which was mixed in a horizontal shaker for 3 h. DIPA (0.55 mL, 3.2 mmol) was added, and the reaction mixture was filtered after 2 h. The resin was washed three times with DMF and twice with DCM. Remaining hydroxyl groups on the resin were capped with acetic acid anhydride (0.75 mL, 8 mmol) and pyridine (0.64 mL, 8 mmol) in 3 mL of DCM within 30 min. Finally, the resin was washed three times with DCM and twice with DMF and methanol and dried in vacuo. The resin loading was obtained from quantitative Fmoc determination.18 Validation of the Analytical Method. The optimized method was validated in terms of precision of peak area and migration time, accuracy, linearity, limit of detection, limit of quantitation, and ruggedness. In all experiments, peak areas were corrected by corresponding migration times by dividing the peak area by the migration time. Precisions of peak area and migration time for the LDD isomer (spiked to 1% to the DLL tripeptide) were better than 1.7 and 1.2% RSD (n ) 3), respectively. Linearity was measured in the concentration range from 0.0005 to 0.5% (w/v) of the major tripeptide. Linear regression analysis gave a correlation coefficient of 0.9992, with the origin lying within the 95% confidence interval of the straight line. The limit of detection was measured to be 0.05% (signal/noise ) 3), and the limit of quantitation was 0.2% of the major compound. The method was rugged with respect to small variations in temperature in the range of 25 to 30 °C, as well as variations in crown ether (5-10 mM) and Tris (20-40 mM) concentration. However, the system was sensitive to small variations in the buffer pH. Thus, proper adjustment of the pH was important for obtaining reproducible results.
(17) Beck-Sickinger, A.; Jung, G. Angew. Chem., Int. Ed. Engl. 1992, 31, 367383.
(18) Meienhofer, J.; Wakin, M.; Heimer, E. P.; Lambros, T. S.; Makofske, R. C.; Chang, C.-D. Int. J. Pept. Protein Res. 1979, 13, 35-42.
2362
Analytical Chemistry, Vol. 68, No. 14, July 15, 1996
RESULTS AND DISCUSSION Theoretical Considerations. To investigate the rate of racemization, the tripeptide D-Tyr-L-Lys-L-Phe (DLL) was chosen as a model compound. Based on the three chiral centers of the molecule, eight optical isomers can be derived. In general, a peptide that consists of n amino acids needs n coupling cycles for its synthesis. Supposing that racemization occurs as independent events, Ai, during each cycle and that only the activated
amino acid at the N-terminal end of the peptide chain can racemize, the total probability of the formation of a stereoisomer is given by the mathematical product of the individual probabilities, P(Ai), according to eq 1:
P(A1A2‚‚‚An) ) P(A1)P(A2)‚‚‚P(An)
(1)
Thus, the formation of an isomer strongly depends on the number of positions where racemization occurs. The concentration of a diastereomer racemized in only one position should be given by the rate of racemization of that particular amino acid. Further, the proportion of a diastereomer, which is formed by isomerization in two positions, can be calculated theoretically as the mathematical product of both racemization rates. Accordingly, the antipode (LDD) of D-Tyr-L-Lys-L-Trp is supposed to be formed only in minute amounts, because the probability of its formation is the product of the racemization rate of each of the three cycles. Provided other byproducts are not considered, the amount of the desired tripeptide will decrease in the same rate as the optical isomers are formed. A similar approach was described by Chang et al.19 in a study on enantiomeric purity of amino acids for peptide synthesis. Investigation of the Enantiomeric Purity of the Protected Amino Acids. Knowledge of the enantiomeric purity of the protected amino acids is crucial for the study of the racemization. Although all amino acids were certified by the suppliers to be enantiomerically pure with a limit of detection of 0.1%, appropriate methods for direct chiral separation were developed to prove the purities. Figure 1 shows the chiral separation of the Fmocprotected amino acids by CZE. Fmoc-DL-Lys(Boc) could not be separated using cyclodextrin as chiral selector. Therefore, the Boc protection group was selectively cleaved from the side chain of lysine by diluted trifluoroacetic acid, leaving a primary amine function. Chiral separation of this partially deprotected lysine was successful when the crown ether system described in the Experimental Section was used. No enantiomeric impurities could be detected in any of the three amino acids (LOD 0.1% for FmocTrp and Fmoc-Lys and 0.2% for Fmoc-Tyr). Considering our results and those of the supplier, enantiomeric impurities should be
