121
Letters in Peptide Science, 4 (1997) 121–127 ESCOM
LIPS 165
Rapid semi-on-line monitoring of Fmoc solid-phase peptide synthesis by matrix-assisted laser desorption/ionization mass spectrometry Gert Talbo*, John D. Wade, Nicola Dawson, Mary Manoussios and Geoffrey W. Tregear Howard Florey Institute of Experimental Physiology and Medicine, University of Melbourne, Parkville, VIC 3052, Australia Received 31 December 1996 Accepted 29 January 1997 Keywords: Monitoring amino acid acylations and Nα-deblocking; Protected peptides; Side reactions
SUMMARY A simple yet highly effective application of matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) for the rapid monitoring of Fmoc solid-phase peptide synthesis is described. A few beads of the resin are removed at any desired step during synthesis, the fully protected peptide is cleaved from the resin and an MS spectrum of the analytes present is produced. Some standard side-chain protecting groups may be cleaved off during sample preparation for MS analysis; however, these cleavages are readily identified. Using this approach, incomplete amino acid acylations are readily detected in approximately the same time as by traditional tests such as ninhydrin. The semi-on-line method also lends itself to ready optimization of synthesis protocols and to the examination of resin-bound peptide side reactions which may not be detectable by chemical means.
INTRODUCTION Modern methods of solid-phase peptide synthesis (SPSS) allow the assembly of larger and increasingly complex peptides. Unfortunately, parallel methods for monitoring the progress of synthesis have not developed as rapidly; hence, minor impurities caused by side reactions, double acylations or incomplete couplings or deblockings may not be detected until after final cleavage and deprotection. Numerous methods for the monitoring of the progress of solid-phase synthesis have been developed over the years. All have their
*To whom correspondence should be addressed. 0929-5666/$ 6.00 + 1.00 © 1997 ESCOM Science Publishers B.V.
advantages as well as limitations. These methods include colorimetric analysis of resin-bound peptides with ninhydrin, isatin, picric acid, bromophenol blue dye and quinoline yellow dye [1], continuous conductance measurement of reaction effluents [2], spectroscopic monitoring of Fmoc chromophore in continuous-flow synthesis [3], and Edman sequencing [4]. Gel-phase NMR spectroscopy has also been employed [5]. More recently, high-resolution probe technology together with 1H NMR spectroscopy has been used to identify chemical compounds covalently bound to a single solid-phase synthesis bead [6]. However, spectra
122 148
h
g
f e d c
b
a
168
S-S-R-N-E-Y-M-R-S-G-L-Y-S-T-F-T-I-Q-S-L-Q Fig. 1. Part of the human transcription factor Elk-1 (S148–Q168) [14].
from samples of this size can be complicated by large signals from the solvent background and the broad signals from the resin’s polymer core and the test takes a relatively long time. The emergence of mass spectrometry (MS) as an analytical tool has had a major impact on synthetic peptide research. Its sensitivity and accuracy have been exploited for the analysis of crude, unprotected, peptide samples and the identification of impurities arising from the assembly itself. Several examples attest to the usefulness of this method for the detection of modifications which would have otherwise been unobserved. These include incomplete Nin-deprotection of the formyl group on tryptophan and its migration to lysine residues [7,8]. Plasma desorption MS (PDMS) was used to survey problems in solid-phase synthesis [9]. However, each of these applications have depended on the post-synthesis cleavage and deprotection of resin samples to enable a qualified assessment by MS of the integrity of the synthetic product. The usefulness of MS during peptide assembly is obviously limited by the time taken to effect cleavage and deprotection of a resin sample. In the case of multi-arginine peptides, such a treatment could take hours of adequate deprotection before analysis can be undertaken. This would have the effect of unduly extending synthesis time by many hours or even days. What is required is a simple yet rapid means of monitoring synthesis efficiency in approximately the same time as by present colour development methods. Matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) lends itself perfectly to the task because it is well suited for mixture analysis and it tolerates the high level of impurities present [10]. The relatively low mass accuracy achieved with highly impure samples is rarely a problem as the nature of the analytes is known. During the course of the synthesis of a number
of so-called ‘difficult’ peptides, we observed that both colour monitoring with ninhydrin and TNBSA and continuous-flow UV absorbance of the reaction column effluent did not completely reveal the occurrence of synthesis difficulties. We decided to examine the utilization of MALDI-MS to analyse Fmoc solid-phase peptide synthesis peptide-resin samples directly as a means of providing a precise analysis of synthetic progress and efficiency. At the same time as this work was in progress, a similar application of MALDI-MS to the analysis of Boc-polystyrene solid-phase synthesis efficiency was described [11]. A dual linker strategy was employed in which a photolabile linker allowed direct cleavage during laser irradiation and subsequent direct MALDI-MS analysis of the resulting fully protected peptide. MATERIALS AND METHODS Peptide synthesis was performed manually on a Cambridge Research Biochemicals synthesizer as previously described; however, coupling times were of 30 min unless otherwise specified [12]. The 1% trifluoroacetic acid (TFA) labile linker, 4(4-hydroxymethoxy)butyric acid, was used as linker between the growing peptide and the resin. All chemicals for peptide synthesis were bought from Auspep, Melbourne, Australia. Mass spectra were acquired using a MALDI mass spectrometer (Bruker, Biflex, Bruker-Franzen, Germany). The acceleration voltage was set to 19.5 kV. The spectra shown are the average of 30–50 single spectra. The MALDI matrix used was a saturated solution of α-cyano-4-hydroxy cinnamic acid [13] dissolved in dichloromethane (DCM). The DCM used was washed intensively (6 times) with equal volumes of water obtained from a Milli-Q water purification system.
123 At any desired step during peptide assembly, a few beads of peptide-resin were removed, transferred to an Eppendorf tube and rinsed with washed (aq) DCM to remove contaminants, e.g. DMF and salts. A solution of 5% TFA in washed DCM (2–3 µl) was added to cleave the protected peptide from the resin, and the matrix was then added to the solution. Approximately 0.7 µl of the resulting solution was immediately applied to the target prior to MALDI-MS analysis. Accurate dispensing of small volumes of DCM is not possible as the solvent either evaporates rapidly or spreads over a relatively large area. RESULTS AND DISCUSSION Elk-1 is a transcription factor which may be involved in the pathogenesis of certain tumours
[14]. We undertook to synthesize the sequence from positions 148 to 168 inclusive as seen in Fig. 1. The peptide was found to be difficult to synthesize by normal automated SPPS chemistry. Mass spectrometric analysis of the crude, cleaved product showed several peaks due to many incomplete coupling and deblocking steps in the synthesis. In order to appropriately modify the synthesis protocol to successfully obtain the correct peptide, we undertook to develop a MALDIMS method to accurately monitor the coupling/ deblocking steps during synthesis. The simplest means of achieving this was to use an acid labile linker which would enable the release of a fully protected peptide by a brief treatment with dilute acid. Preliminary efforts used 4-hydroxy-3-methoxyphenoxyacetic acid, but it was observed that
Fig. 2. MALDI-MS spectrum obtained after coupling and deblocking at Arg155, resulting in the following sequence: R(Pmc)-S(tBu)-[Hmb]G-L-Y(t-Bu)-S(t-Bu)-T(t-Bu)-F-T(t-Bu)-I-Q(Trt)-S(t-Bu)-L-Q(Trt) (calculated MW 2845.5 Da). The molecular ions are sodium and potassium adduct ions, respectively. The ‘-PmcNa/K’ peak corresponds to the peptide which has lost the Pmc sidechain protecting group and, thus, is protonated instead of ionized by alkali ions. The remaining peaks correspond to the loss of Hmb and t-Bu from either ‘MNa+/K+’ or ‘-PmcNa/KH+’, respectively.
124
Fig. 3. MALDI-MS spectrum showing incomplete acylation. The peptide has the following sequence: N(Trt)-E(t-Bu)-Y(t-Bu)-MR(Pmc)-S(t-Bu)-G-L-Y(t-Bu)-S(t-Bu)-T(t-Bu)-F-T(t-Bu)-I-Q(Trt)-S(t-Bu)-L-Q(Trt).
cleavage times with 1% TFA in DCM were occasionally too long to be useful, i.e. more than 5 min. We then used the more labile linker 4-(4hydroxymethoxy)butyric acid. The types of ions and analytes detected The sample is a fully protected peptide and, hence, it has no natural protonation site, except after deblocking a free amino group at the Nterminus. Therefore, the sample is ionized by sodium and potassium like other samples without a protonation site, e.g. carbohydrates. The Nterminus exposed after deblocking, however, may not be sufficiently basic to facilitate the protonated ion species. The analyte used in Fig. 2 has been deblocked, but the sodium and potassium adduct ions are still dominant and appear in the spectrum as a double molecular ion peak. At the same time as the protected peptide is cleaved from the resin, partial side-chain protect-
ing group removal may occur. Although it is possible to avoid these cleavages by carefully adjusting the conditions, i.e. concentration of the acid and reaction time, this is contrary to the desired speed and simplicity of the method. The use of 5% TFA ensures an almost instantaneous and complete cleavage of the peptide from the resin. The peaks, due to the missing protecting groups, appear at calculable positions and, hence, do not interfere with the peaks in the region of interest of the spectrum. In Fig. 2, three examples of side-chain protecting group loss can be seen. Descending from the molecular ion peak, the first peak at −56 Da corresponds to the loss of a tertbutyl (t-Bu) group. Further below, the peak corresponding to the loss of N-(2-hydroxy-4-methoxybenzyl) (Hmb) can be seen at −135 Da. The same pattern is repeated at 288 Da lower mass. This offset is due to the loss of the 2,2,5,7,8-pentamethylchroman-6-sulphonyl (Pmc) side-chain pro-
125 tecting group of arginine and the accompanying peaks at lower mass correspond to the loss of tBu or Hmb, respectively, as described above. The loss of the protecting group from arginine exposes a very basic protonation site. Therefore, the offset corresponds to the loss of Pmc and sodium/potassium and, thus, the peaks are the protonated species. Adjustments to the solid-phase synthesis protocol Samples analysed after acylation of residues Thr161 and Tyr159 at positions ‘a’ and ‘b’ (for denotation see Fig. 1) showed that the reactions had proceeded to completion within the standard 30 min coupling time. However, MALDI-MS analysis following the coupling of Ser156 at position ‘c’ revealed that serine was missing on approximately one-fifth of the peptide. Acylation was complete after recoupling for an additional half hour reaction time. The coupling of Arg155
in position ‘d’ was observed to be very slow. Consequently, three small-scale pilot acylations were carried out. Each test method used different activation agents, O-(7-azabenzotriazol-1-yl)1,1,3,3-tetramethyluronium hexafluorophosphate (HATU), O-(benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU), and N,N′-dicyclohexylcarbodiimide (DCC) plus the catalyst 1-hydroxybenzotriazole (HOBT). The base used with HATU and HBTU was diisopropylethylamine (DIEA). The remaining synthesis parameters were unchanged. After 30 min acylation, MS analysis of the peptide following coupling of arginine activated with HATU indicated the reaction to be complete. In contrast, the coupling with HBTU was shown to be almost complete while the DCC-activated coupling was even less successful. Based on these results, the remainder of the peptide-resin was acylated with arginine activated with HATU using DIEA as the base.
Fig. 4. MALDI-MS spectrum indirectly showing incomplete acylation. The peptide has the following sequence: M-R(Pmc)-S(t-Bu)G-L-Y(t-Bu)-S(t-Bu)-T(t-Bu)-F-T(t-Bu)-I-Q(Trt)-S(t-Bu)-L-Q(Trt) (calculated MW 3063.7 Da).
126
Fig. 5. MALDI-MS spectrum of the crude peptide following changes to the synthesis protocol showing the assembly to have been satisfactorily performed (calculated MW 2467.7 Da).
However, despite the result of the pilot acylation a double coupling was still necessary. The deblocking up to this stage was performed by 20% piperidine in DMF. However, after the Arg155 acylation, deblocking was only complete when piperidine was replaced by 2% 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) in DMF [15]. This base was used as deblocking agent for the remainder of the synthesis. Following the above-described adjustments to the coupling and deblocking procedure, the correct molecular weight (MW) up to this point (calculated 2845.5 Da) was obtained as seen in Fig. 2. Spectrum ‘g’ in Fig. 3 shows an example of incomplete coupling. The molecular ion appears as a double peak. The peak at 3828.6 Da corresponds to the sodium adduct ion and the peak at higher mass corresponds to the potassium adduct ion (calculated MW 3824.3 Da). The double peak at ‘-N’ corresponds to the sodium and potassium
adduct of the peptide lacking asparagine, respectively. The peaks in between the molecular ion peaks and ‘-N’ correspond to the loss of sidechain protecting groups as described above. Two repeat couplings were required to effect complete acylation of the peptide (data not shown). It can be argued that the peaks corresponding to a missing amino acid residue may be masked by the peaks derived from the loss of side-chain protecting groups. However, this may not be a major concern as shown in the spectrum obtained at position ‘e’ (Fig. 4). The figure shows the molecular ion peak, the loss of side-chain protecting group peaks and the peaks at ‘-MPmcNa/K’. The peak corresponding to the peptide lacking methionine is hidden under the ‘-PmcNa/K’ peak and cannot be seen, but the peak corresponding to the combined mass of the ‘lacking methionine’ and loss of ‘-PmcNa/K’ is seen at ‘-MPmcNa/K’. The peak at lower mass relative to ‘-MPmcNa/K’ cor-
127 responds to the further loss of t-Bu. The assignment as a ‘-Pmc’ peak is supported by the fact that it is a single peak due to the exposure of the strong protonation site on the arginine side chain. The sample analysed at positions ‘f’ and ‘h’ showed complete coupling following the normal 30 min reaction cycle. In summary, the adjustments to the standard synthesis protocol as a result of the on-line MS results are: (1) HBTU was substituted by HATU in the coupling of arginine in step ‘d’. (2) Piperidine was replaced by DBU as deprotecting agent after step ‘d’. (3) Double coupling was employed at steps ‘c’ and ‘d’. (4) Triple coupling was necessary at steps ‘e’ and ‘g’. When the final MALDI-MS monitoring spectrum showed the peptide synthesis to be complete, the full batch of peptide-resin was cleaved, purified and mass analysed. An MS spectrum of the resulting crude peptide showed the synthesis to be satisfactory, as seen in Fig. 5. The results demonstrate that it is possible to adjust the standard synthesis protocol to obtain a better product. However, we do not envision MALDI-MS monitoring as a standard method because of the associated costs. Although the time consumed for the mass measurement is very short, the MS operator must be almost on constant standby to take full advantage of the method. Further, the nature of the analyte means that the sample is ionized by alkali ions giving rise to two peaks 16 Da apart, MNa+ and MK+, respectively. These two peaks may become difficult to resolve at molecular weights above 5 kDa due to the nature of the sample preparation method. CONCLUSIONS We have shown how rapid semi-on-line MALDI-MS monitoring of the individual steps in peptide synthesis can assist the assembly of difficult peptides. In the example described in this
paper, the coupling activation agent and the deblocking agent had to be changed in one step. Further, double and triple coupling were necessary at three other steps. After making these adjustments to the synthesis protocol, a highly pure peptide of correct MW was obtained. The method lends itself very much to the detection of abnormalities or side reactions during synthesis when there is still time and a chance to correct the error. ACKNOWLEDGEMENTS The work described herein was supported by an Institute Block Grant from the National Health and Medical Research Council of Australia. REFERENCES 1 Fields, G.B. and Noble, R., Int. J. Pept. Protein Res., 35 (1990) 161. 2 McFerran, N.V., Walker, B., McGurk, C.D. and Scott, F.C., Int. J. Pept. Protein Res., 37 (1991) 382. 3 Dryland, A. and Sheppard, R.C., J. Chem. Soc. Perkin Trans. I, (1986) 125. 4 Tregear, G.W., Van Rietschoten, J., Sauer, J., Niall, H.D., Keutmann, H.T. and Potts Jr., J.T., Biochemistry, 16 (1977) 2817. 5 Epton, R., Goddard, P. and Ivin, K.J., Polymer, 21 (1980) 1367. 6 Keifer, P.A., Baltusis, L., Rice, D.M., Tymiak, A.A. and Shoolery, J.N., J. Magn. Reson., A119 (1996) 65. 7 Merrifield, R.B., Singer, J. and Chait, B.T., Anal. Biochem., 174 (1988) 399. 8 Chowdhury, S.K. and Chait, B.T., Anal. Biochem., 180 (1989) 387. 9 Fontenat, J.D., Ball, J.D., Miller, M.A., David, C.M. and Montelaro, R.C., Pept. Res., 4 (1991) 19. 10 Hillenkamp, F., Karas, M., Beavis, R.C. and Chait, B.T., Anal. Chem., 63 (1991) 1193A. 11 Carrasco, M.R., Fitzgerald, M.C. and Kent, S.B.H., Abstract P300, 24th European Peptide Symposium, Edinburgh, Scotland, U.K., 1996. 12 Wu, C.-R., Wade, J.D. and Tregear, G.W., Int. J. Pept. Protein Res., 31 (1988) 47. 13 Beavis, R.C. and Chait, B.T., Rapid Commun. Mass Spectrom., 3 (1989) 432. 14 Rao, V.N., Huebner, K., Isobe, M., Ar-Rushdi, A., Croce, C.M. and Reddy, E.S.P., Science, 244 (1989) 66. 15 Wade, J.D., Bedford, J., Sheppard, R.C. and Tregear, G.W., Pept. Res., 4 (1991) 194.