Proximity-driven acceleration of challenging solid-phase peptide couplings

Significance

Solid-phase peptide synthesis (SPPS) is a foundational technology in applied and basic pharmaceutical sciences. While typical SPPS protocols are generally effective, the synthesis of peptides that tend to aggregate or have N-alkyl units remains challenging due to inefficient acylation reactions. Here, we explore a solution to this problem in which a nucleophilic pyridine, long known to be a catalyst of acylation reactions, is tethered to the resin along with the peptide substrate. Its reaction with an activated amino acid creates a high local concentration of a reactive acylating agent, resulting in more rapid coupling. We show that several difficult-to-synthesize peptides are produced in markedly higher yield with far fewer side products using this “proximity catalysis” strategy.

Abstract

Many important solid-phase synthesis reactions proceed slowly and inefficiently, for example, the coupling of an activated ester to a sterically hindered, bead-displayed amine, a critical step in the synthesis of many peptides. Forcing conditions are often required to achieve acceptable yields. We show here that such reactions can be accelerated by tethering to the bead a nucleophilic pyridine catalyst, which reacts with the activated ester, creating a high local concentration of a reactive acylpyridinium intermediate that couples efficiently to the bead-displayed amine. In many cases, the “cleanliness” of reactions carried out on resins equipped with an immobilized catalyst is substantially better than the analogous reactions in which the catalyst is added to the solution.

Solid-phase synthesis is used widely for the construction of peptides and other oligomeric compounds. Applications include both the large-scale production of individual bioactive molecules as well as the creation of combinatorial libraries. For all of these applications, efficient couplings are essential. This can be challenging for certain classes of peptides. For example, the synthesis of N-methyl or, more generally, N-alkyl peptides often requires forcing conditions using highly reactive amino acid derivatives (1, 2) due to the sterically hindered nature of the bead-displayed amine nucleophile. Even acylation of primary amines during the synthesis of certain peptide sequences exhibits poor coupling efficiency due to aggregation during chain elongation. This aggregation results from intra- or intermolecular hydrophobic interactions and/or hydrogen bonding and leads to poor swelling of the resin, limited diffusion of reagents, reduced accessibility of the N-terminal amine, and premature folding or sticking to the resin backbone, all of which hinder complete coupling and deprotection steps (3, 4).

Catalysts for difficult acylation reactions are well known. For example, electron-rich pyridines, such as 4-dimethylaminopyridine (DMAP) (5), or N-alkyl imidazoles (6–8) have been used to catalyze esterification reactions between activated esters and sterically hindered alcohols. Likewise, these species were found to be useful for accelerating amino acid couplings and suppressing racemization using a variety of activating agents (9, 10). DMAP is thought to function by reacting with the activated ester to form a highly reactive N-acyl pyridinium intermediate that is then attacked by an alcohol or amine to form the desired ester or amide (Scheme 1).

Scheme 1.

Proposed role of DMAP as a catalyst for acylation reactions.

However, as shown below, the addition of soluble DMAP to certain difficult solid-phase acylation reactions had little effect on the reaction rates or yields. We hypothesized that a more favorable outcome might be achieved by tethering the catalyst to the bead. In this case, the N-acylpyridinium ion formed upon reaction of the bead-displayed pyridine would be in close proximity to the nucleophile, and this high effective local concentration would perhaps result in an improved coupling yield and, hence, the formation of fewer side products. We show here that this is indeed the case and that several types of difficult acylation reactions proceed more rapidly and cleanly on beads functionalized with a nucleophilic catalyst.

Results

Resin-Displayed Catalyst Accelerates Peptide Coupling without Racemization.

To determine the effect of coimmobilization of a substrate and a DMAP-like catalyst on resin (henceforth referred to simply as DMAP), we carried out a kinetic study of the rate of acylation of a bead-displayed secondary amine with a glycol-containing, Fmoc-protected amino acid NHS ester (called PEG₂ in Fig. 1). Three parallel on-resin experiments were conducted, all using the same PEG₂-NHS stock solution to keep ester concentration and conditions identical. In the first case, no catalyst was present (a Boc protecting group was present on the lysine side chain). In the second, the beads carried the DMAP catalyst, and in the third, 10 equivalents of soluble DMAP were added to the reaction using the resin lacking an immobilized catalyst. Aliquots were withdrawn at defined time points and, following release of the compounds from the resin, conversion to product was quantified by LC–MS.

Fig. 1.

Effect of coimmobilized or soluble catalyst on the rate of coupling of an NHS ester to a secondary amine. (A) The on-resin reactions analyzed in this experiment. (B) The solution phase reaction analyzed in this experiment. (C) Rate constants for PEG₂–NHS coupling to the secondary amine on- or off-resin and in the presence or absence of catalyst.

By fitting the raw data to a first-order decay model (ln([SM]_t/[SM]₀) vs. time), the rate of the on-resin uncatalyzed reaction was found to be k = 0.197 h⁻¹ (t_1/2‚ ≃ 210.6 min). The presence of the immobilized DMAP catalyst increased the rate 3.2-fold (k = 0.625 h⁻¹, t_1/2 ≃ 66.6 min). In contrast, addition of soluble DMAP had only a small effect on the reaction rate (k = 0.244 h⁻¹, t_1/2 ≃ 170.6 min). These data demonstrate that the immobilized catalyst accelerates acylation of the bead-displayed secondary amine significantly and that the effect is substantially greater (2.6-fold) than can be achieved by the addition of soluble DMAP.

An interesting question is whether the acyl transfer reaction from the pyridinium ion to the secondary amine substrate occurs exclusively between moieties on the same strand (intrastrand) or, in the crowded environment of the bead, an amine on one strand can attack an acylpyridinium ion on a different strand (strand-to-strand). To probe this point, we carried out the analogous three reactions in solution (no catalyst, catalyst linked to the peptide substrate, and catalyst added to the solution). If the on-resin acyl transfer occurs through a mostly intrastrand mechanism, we would expect to see a similar rate acceleration for the analogous molecule in solution. This was not the case. As shown in Fig. 1C, the rate of acylation of the compound tethered to the catalyst was only 1.3-fold faster than that of the molecule lacking the catalyst and almost identical to the rate of the reaction in the presence of soluble DMAP. We conclude that the bulk of the on-resin acylation reaction occurs in a strand-to-strand fashion where the acyl-pyridinium ion from one molecule reacts with the amine of a nearby molecule (Scheme 2). (See SI Appendix, sections 2–4 for full kinetic study data). This result suggests that the effect of the coimmobilized catalyst will be greatest at high bead loading density and taper off at lower loadings, though we have not probed this issue experimentally.

Scheme 2.

Graphical depiction of intrastrand and strand-to-strand mechanisms for catalyst-aided acylation. The black bar represents the peptide. See text for details.

Synthesis of catalyst-free peptides.

Having established that the coimmobilized DMAP catalyst accelerates peptide coupling, we turned to developing a resin that would not “contaminate” the product with the appended catalyst. This was done in two ways.

In the first approach, a lysine residue was added to TentaGel, and the catalyst was appended to the side chain amine, as before. A RAM linker was then appended to the α-amino group to generate resin 11 (Fig. 2A). After Fmoc removal and chain extension, treatment of the resin with TFA releases the desired product and leaves the catalyst behind on the bead. To evaluate the utility of this strategy for large-scale (200 µmol) peptide synthesis, we made the tripeptide Ile-Pro-Pro (IPP), a small, but challenging-to-synthesize, peptide that is used as an inhibitor of angiotensin-converting enzyme (11, 12). The synthesis employed resin 11 or commercially available TentaGel RAM under the following conditions: 10 equivalents amino acid, 10 equivalents DIC/Oxyma, 2 h, 37 °C. After release from the resin by treatment with TFA, the products were analyzed by LC-MS. The compound made on the DMAP-containing resin was significantly cleaner than the peptide made on TentaGel-RAM, in which a significant amount of side product resulting from a missed proline coupling was evident (Fig. 2). Following HPLC purification of the product and lyophilization, IPP was derived in 57% yield from the catalyst-containing resin and 41% yield from TentaGel-RAM resins.

Fig. 2.

Solid-phase synthesis of catalyst-free peptides. (A) Graphical illustration of the “pre-RAM” strategy in which an acid-cleavable RAM linker is inserted between the immobilized catalyst and the peptide. (B) Chromatograms of the crude product mixtures from syntheses of Pro-Pro-Ile-NH₂ carried out on TentaGel-RAM resin (-Catalyst) or using resin 11 shown in part A (+ Catalyst). (C) Mass spectra of the crude product mixtures derived from the products shown in (B). The spectra were taken over the entire peak, which is likely broadened due to the presence of conformational isomers created by slow rotation around the tertiary amide bonds. (D) Chromatogram of the crude product at 65 °C using an optimized gradient. See text and SI Appendix, Fig. S7 for details.

While mass spectrometry (MS) analysis across the product peak resulting from synthesis of IPP revealed only the desired product (Fig. 2 B and C), the broadness of the peak raised the concern that there might be minor impurities present that did not ionize well or whose ionization might be suppressed and thus would not be detected. Additionally, we suspected that some of the peak broadening might be due to slow cis–trans isomerism of the tertiary amide bonds in the peptide. To probe these issues, we investigated a variety of other chromatographic conditions and heated the column to increase the rate of cis–trans peptide bond isomerism. This resulted in the chromatogram shown in Fig. 2D, which was recorded at 65 °C. Two minor peaks were separated from the IPP major product. One was the result of trifluoroacetylation of the N terminus of IPP (labeled as 12+TFAc in Fig. 2D), which occurred during TFA cleavage of the product from the resin. The other is a nonpeptidic impurity of unknown structure. Moreover, the increased temperature indeed resulted in a sharper major peak, consistent with accelerating the interconversion of amide bond isomers. These data (see SI Appendix, Fig. S7 for details) strongly support the contention that the synthesis of IPP on resin 11 indeed proceeded cleanly and efficiently to produce the desired product.

We next turned to the synthesis of a peptide representing residues 65 to 74 of the Acyl Carrier Protein (ACP). This is a well-established model system for the evaluation of new techniques for synthesizing difficult peptide sequences (13–15). Each coupling was performed with 10 equivalents of amino acid and DIC/Oxyma at 37 °C for 2 h on resin that did or did not include the immobilized catalyst. As shown in Fig. 3, LC analysis of the crude material produced in each synthesis revealed a single dominant peak, but the peak from the material produced using TentaGel-RAM resin lacking the catalyst appeared to be broader than that produced from the beads carrying the catalyst, suggesting the presence of coeluting byproducts. Indeed, mass spectrometry analysis showed only the desired product 13 for the peptide synthesized on the DMAP-functionalized resin 11. In contrast, several contaminants with masses consistent with missed couplings (specifically Valine 1 and Glutamine 2) were apparent in the mass spectra of the material produced on resin lacking the catalyst. A similar result was obtained when the synthesis was carried out on TentaGel-RAM resin with DMAP in solution (Fig. 3 B2 and C2).

Fig. 3.

Effect of soluble or immobilized DMAP on the Solid-phase synthesis of the ACP (65 to 74) peptide 13. (A) Synthetic scheme for the reaction employing the pre-RAM strategy. For the reactions lacking the immobilized catalyst, the beads lacked the 1-functionalized lysine residue but were otherwise identical. (B) Chromatograms of the crude products after release from the resin. (B-1) Material synthesized on resin 11 with the immobilized catalyst. (B-2) Material synthesized on TentaGel-RAM resin with DMAP added to the solution. (B-3) Material synthesized on TentaGel-RAM resin with no catalyst. (C-1, C-2 and C-3) Display MS spectra of the crude product mixtures corresponding to the chromatograms shown in B-1, B-2 and B-3.

An alternative cleavable linker strategy, based on the work of Abboud et al. (16), is shown in Fig. 4. It is designed for the synthesis of peptides with a C-terminal Cys residue, which are useful as native ligation substrates or for selective labeling on the cysteine thiol. Starting with resin 14, disulfide exchange using 2-amino-1,1-dimethyl-ethane-1-thiol produced 15. This disulfide is stable to the conditions employed for Fmoc-based SPPS (16). The side chain amine was acylated with acid 1 to generate 16. Upon removal of the Fmoc group, the IPP tripeptide was synthesized, again using the same conditions (10 equiv. amino acid, 10 equiv. DIC/Oxyma, 2 h, 37 °C). The beads were then treated with TCEP to remove the catalyst through cleavage of the disulfide bond and washed thoroughly. TFA treatment then released the desired peptide CIPP (17). As shown in Fig. 4, a significantly cleaner product was obtained when the catalyst was on the resin than when it was present in solution or absent.

Fig. 4.

Synthesis of CIPP peptides. (A) Disulfide linker-based strategy. The RAM linker between the bead and the bead-displayed amine is not shown for clarity. (B) Synthetic scheme for CIPP synthesis with the disulfide-linked catalyst present. (C-1, C-2, C-3) Chromatograms of the crude product synthesized under the conditions indicated after release from the resin. (C-1) Material synthesized on resin with the immobilized catalyst. (C-2) Material synthesized on TentaGel-RAM resin with DMAP added to the solution. (C-3, D-1, D-2, D-3) Mass spectra of the crude product mixtures corresponding to chromatograms shown in C-1, C-2 and C-3.

An important technical point with respect to using this scheme is that, in earlier work, we found that addition of the activated ester of 1 to Fmoc-protected peptides can sometimes result in unwanted removal of the Fmoc group and addition of the catalyst to the N terminus (not shown). Therefore, we routinely attached the catalyst to substrates bearing a different protecting group at the N terminus. However, because N-terminal Mtt-, Dde-, Mmt-, or Alloc-protected forms of Cys(Npys) are unavailable, the practical option is Fmoc-Cys(Npys)-OH and we thus developed modified conditions to ensure clean formation of the desired product [coupling conditions outlined in SI Appendix, Table S1 (entry 5)].

Degree of racemization in proximity-driven peptide coupling reactions.

Racemization is a concern when highly reactive acylation intermediates are formed. Therefore, we carried out coupling of an amino acid sensitive to configurational instability in the presence and absence of the coimmobilized catalyst. First, we added the activated ester (DIC/Oxyma) of Fmoc-protected L-phenylglycine to beads displaying L-proline in the presence or absence of the coimmobilized DMAP-like catalyst. Separately, the diastereomeric dipeptide was synthesized by using Fmoc-D-phenylglycine in the absence of catalyst. After cleavage from the resin, the diastereomeric Fmoc-protected products were readily resolved by LC (the Fmoc group was not removed in order to increase the retention time of the product and improve separation between the diastereomers). The chromatograms showed no detectable production of L-proline-D-Fmoc-phenylglycine when the L-activated ester was added to the resin (SI Appendix, Fig. S13).

Use of different resins.

While our laboratory typically employs TentaGel resin for solid-phase synthesis, other supports, such as polystyrene (PS) and acrylamide-polyethylene-copolymer (PEGA), are also utilized widely. Therefore, we evaluated the utility of the coimmobilized catalyst for the synthesis of difficult peptides on these resins. In each case, the resin was modified using the pre-RAM strategy to leave the catalyst behind after TFA cleavage.

We began with oligopeptides containing only a single type of amino acid. These sequences, especially those rich in hydrophobic residues, such as oligo-leucine, tend to aggregate and are notoriously difficult to synthesize (15, 18). A synthesis of Leu₆ on polystyrene resin bearing the DMAP-like catalyst proceeded efficiently and cleanly, providing an ≈80% yield of the desired product whereas the analogous synthesis on polystyrene resin lacking the catalyst provided only a ≈ 15% yield of Leu₆ along with a mixture of truncated sequences lacking 1, 2, or 3 Leu residues and an unidentified product (Table 1; see SI Appendix, Fig. S10A for structures and primary data).

Table 1.

Effect of the coimmobilized catalyst on the yield of the peptides shown on various solid supports

Entry	Resin	Catalyst	Sequence	Purity
1	Polystyrene	+	LLLLLL	80%
2	Polystyrene	−	LLLLLL	15%
3	PEGA	+	PVNIIGRNLLTQIGCTLNF	60%
4	PEGA	−	PVNIIGRNLLTQIGCTLNF	50%
5	TentaGel	+	CPRTTpSFAES	55%
6	TentaGel	−	CPRTTpSFAES	25%
7	PEGA	−	GILTVSVAV	0%
8	PEGA	+	GILTVSVAV	0%

For entries 1 to 6, each step was carried out using four equivalents of amino acid/DIC/Oxyma for 10 min at 75 °C (microwave irradiation). For entries 7 and 8, each step employed 10 eq. of amino acid, DIC, and Oxyma and was carried out at 37 °C for 2 h.

Next, a fragment of HIV protease [HIV-1 PR (81 to 99), entries 3 and 4] frequently used by researchers to evaluate new synthesis methods (15) was constructed on PEGA resin that did or did not contain the catalyst. Again, the resin displaying the catalyst provided a higher yield, though in this case the increase was modest (Table 1; See SI Appendix, Fig. S10B for structures and primary data).

Surprisingly, synthesis of the nine-residue peptide GILTVSVAV failed to provide the desired product on PEGA resin in the presence or absence of the catalyst. As shown in Fig. 5, a seven-residue peptide resulting from two missed couplings (of a valine and the serine) was the major product in both cases. However, the reaction carried out in the absence of the catalyst also produced significant amounts of material resulting from a third missed coupling, which were absent in the “+catalyst” synthesis, demonstrating that even in this suboptimal case, the presence of the coimmobilized catalyst was beneficial.

Fig. 5.

HPLC trace of the crude product in the synthesis of NH₂-GILTVSVAV-CONH₂ on “pre-RAM” PEGA resin in the presence or absence of the coimmobilized catalyst. The asterisks denote unidentified peaks whose masses do not correspond to missed coupling side products.

Phosphorylated peptides are known to be synthetically challenging. The incorporation of phosphorylated amino acids is often inefficient, and the presence of a protected phosphate group can interfere with subsequent steps (17). Thus, we evaluated the efficiency of synthesis of a phospho-peptide corresponding to a segment of the phosphorylated GSK3β protein, commonly used in studies of Glycogen Synthase Kinase 3 beta function and interactions (19), on TentaGel RAM resin that did or did not display the catalyst. Only ≈25% of the desired full-length product was produced on the resin lacking the catalyst (Table 1, entry 5) and the product mixture was dominated by peptides lacking the phosphorylated serine and the following residue. In contrast, the catalyst-modified resin provided an approximately 55% yield of the desired phospho-peptide and significantly suppressed these specific deletion impurities (Table 1, entry 6; see SI Appendix, Fig. S10C for structure and raw data).

We next asked if the presence of the immobilized DMAP catalyst is beneficial to the synthesis of a longer peptide. Specifically, the 20-residue MUC1-derived peptide (NH₂-APDTRPAPGSTAPP AHGVTS-CONH₂); residues 301 to 320 of the human protein). This sequence and related MUC1 tandem-repeat fragments are frequently used as model substrates in SPPS (20, 21). This synthesis employed the same conditions listed for entries 1 to 6 in Table 1. No sequence-specific optimization was undertaken, including efforts to suppress aspartimide formation.

LC–MS analysis of the crude product synthesized in the absence of catalyst showed the full-length peptide together with one major impurity: an Asp-derived species and the corresponding −18 Da aspartimide product (−18 Da relative to the desired peptide), consistent with modification within the Asp–Thr region. A small amount of a product due to a missed coupling of Ala1 was also seen (Fig. 6). The catalyst-modified resin produced the desired product as well, again with a small (but in this case larger) amount of the “-Ala1” impurity. The crude purity was higher for the catalyst-modified synthesis (≈85%) compared to the noncatalyst reaction (≈70%) but this was almost entirely due to a reduced level of the aspartimide-derived product. The couplings were efficient in both cases. Whether or not the catalyst is generally effective in suppressing aspartimide formation will require testing of more substrates.

Fig. 6.

LC traces of the crude products for the microwave-assisted synthesis of NH₂-APDTRPAPGSTAPPAHGVTS-CONH₂ in the presence or absence of a coimmobilized DMAP-like catalyst. Asi represents a product due to aspartimide formation, not a missed coupling. The other minor peaks were not assigned and do not have masses corresponding to products due to missed couplings.

Synthesis of multiply N-methylated peptides.

We next examined the utility of the proximity catalysis effect for the synthesis of multiply N-methylated peptides (NMPs). As mentioned above, acylation of the sterically hindered secondary amine of an N-methyl amino acid is difficult, proceeding 10 to 100 times more slowly than the acylation of a primary amine (8). Peptide 19 (Fig. 7), a sequence chosen randomly, was employed as an initial target. We utilized commercially available Fmoc-protected N-methyl amino acids and employed the same conditions for each coupling (10 equivalents of amino acid, 10 equivalents of DIC/Oxyma, for 2 h at 37 °C). To allow a direct comparison of the purity of the product in the presence and absence of the DMAP-like unit, the product of the “no catalyst” protocol 19 was acylated with 1 in the last step of the synthesis, whereas in the “+ catalyst” synthesis, this was done prior to synthesis of the tetrapeptide. As shown in Fig. 7C, chromatographic analysis of the crude product demonstrated peptide 19 was produced in much greater purity when the DMAP catalyst was present throughout the synthesis. Next, the pre-RAM strategy (i.e., using resin 11) was used to produce tetrapeptide 21 under the identical conditions outlined above. The purity of the crude product was then compared to the products obtained by synthesizing 21 on TentaGel RAM resin in the presence or absence of one equivalent of DMAP in solution. Once more, compound synthesis with the immobilized catalyst produced the desired product in greater purity, with the main impurities being deleted peptides lacking N-Me-Alanine2 and N-Me-Glycine3 (Fig. 7).

Fig. 7.

Effect of catalyst on the synthesis of N-methylated peptides 19 and 21. (A) Synthesis of 19 on TentaGel RAM resin prefunctionalized with the catalyst. (B) Synthesis of peptide 19 on TentaGel RAM with the addition of 1 in the final step. (C) Synthesis of 21 using “pre-RAM” addition strategy. (D) Synthesis of compound 21 using standard SPPS. (E) Synthesis of compound 21 using standard SPPS with the addition of 1 equiv. of DMAP in solution at each coupling step. Box: (A-E2&3) Chromatograms of the crude products after release from the resin. (A-2) Peptide 19 synthesized on resin with the immobilized catalyst. (B-2 and B-3) Peptide 19 synthesized on TentaGel-RAM resin lacking catalyst, then acylated with 1 in the final step. In (B-2) DMAP was present in solution, in (B-3) it was not. (C-2, D-2, and E-2) HPLC chromatograms of crude compounds 21. MS-based identification of the major impurities is provided in SI Appendix, sections S6-1. In A, B, D, and E, the RAM linker is not shown for clarity.

We also conducted two reactions that were almost identical to that shown in Fig. 7C, except one or two copies of the “PEG₂” unit (shown in Fig. 1) were inserted between the lysine and the DMAP unit 1. As shown in SI Appendix, Fig. S12, the major product remained the desired peptide 21 but the amount of side products was affected by the linker length. In both cases, the longer linkers increased the efficiency of the Gly3 coupling and almost entirely suppressed the formation of the “21-Gly3” impurity. However, when two of the PEG₂ spacers separated the catalyst from the lysine unit, a significant amount of side product resulting from a missed coupling of Ala2 appeared. These data suggest that the spacing between the catalyst and the resin-displayed, nucleophilic amine likely plays a role in the efficiency of the acylation reaction.

Discussion

While the solid-phase synthesis of peptides is a relatively mature field, there remains the fundamental problem of achieving high yields of acylation when the amine nucleophile is sterically hindered or sequestered in aggregated structures, making many peptides challenging targets. For decades, the standard response to this issue has been to use highly reactive acylating agents to drive recalcitrant couplings (1, 22). Along these lines, DMAP has been used as an acylation catalyst, producing a reactive acylpyridinium intermediate from the activated ester or acid chloride (9, 10).

Here, we chose to explore a different strategy in which a DMAP-containing unit is coimmobilized on the resin along with the growing peptide. The hypothesis was that reaction of the catalyst with soluble active ester would result in a very high local concentration of the acylpyridinium ion in the bead environment, resulting in more rapid and efficient acylation of nearby amines. This was shown to be true. The presence of the immobilized catalyst accelerated the acylation of a secondary amine by more than threefold (Fig. 1), a significant, if not enormous boost in rate. As demonstrated here in several different contexts, the presence of an immobilized DMAP-like catalyst generally has a salutary effect on the outcome of the peptide coupling reaction and it is markedly superior to having the catalyst in solution. In some cases, for example, in the synthesis of Leu₆, the yield was dramatically higher when the immobilized DMAP was present (Table 1). In others, a more modest increase in yield was realized (Table 1).

In all cases reported here, we employed DIC and Oxyma as the activating agents. In our experience, these conditions generally provide the best results for acylation of bead-displayed nucleophiles, particularly secondary amines (23). This is important. We would not expect to see a major catalyst-dependent improvement in overall yield if the uncatalyzed reaction proceeds in high yield. Indeed, for the microwave-assisted synthesis of the 20-residue Mucin-derived peptide, which is longer but does not require any difficult couplings, the presence of the immobilized DMAP-like catalyst provided only a modest boost (Fig. 6). We have not tried to engineer conditions to provide the greatest possible difference between the catalyzed and uncatalyzed reactions (for example, by limiting time or using less reactive coupling agents) since we are interested in the practical issue of achieving the best result in any particular synthesis.

Our data suggest that most of the accelerated acylation occurs via a strand-to-strand mechanism, where an acylpyridinium intermediate from one strand “reaches over” to an amine on a different strand (Scheme 2). Thus, this proximity effect is unique to reactions carried out in the crowded environment of a resin. It is conceivable that a similar effect might also be realized in the context of a soluble polymer, but this has yet to be investigated.

It is important to point out that, even from the modest suite of examples reported in this initial study of proximity catalysis, it is clear that the presence of the immobilized DMAP unit is not a “magic wand” that ensures success in any synthesis. For example, we were surprised to find that the attempted synthesis of NH₂-GILTVSVAV-CONH₂ on DMAP-containing resin gave a very clean product but one in which two of the couplings failed to occur (Fig. 5). This highlights that for certain sequences careful optimization will still be required to provide a high yield, though the conditions reported here provide a good starting point. Indeed, even though the desired peptide was not seen in the Fig. 5 experiment, the result was significantly better than in the absence of the catalyst. In summary, our data so far indicate that the presence of the immobilized DMAP unit is generally beneficial but is particularly helpful for difficult couplings in relatively short peptides, which is our primary concern since we aim to use this technology to make high quality one bead one compound libraries of cyclic N-alkylated peptides and related species.

DMAP is only one of several reported catalysts of peptide bond coupling and it is likely that coimmobilization of different catalysts on resin would also be beneficial and perhaps even superior to DMAP for certain reactions. For example, Oxyma is a very good organocatalyst for further activation of the activated ester formed by the reaction of DIC and the amino acid.

Indeed, while this paper was under review, a report from Wei et al. appeared showing that coimmobilization of derivatives of Oxyma or HOBt on resin greatly improves the synthesis of a variety of peptides (24) when DIC alone or DIC/HOBt are used as soluble activating agents. These data support and extend the model proposed here for proximity catalysis of peptide bond formation. It is difficult to compare the efficacy of the reactions reported in that paper with those presented here since the Wei, et al. study did not compare the catalyzed reactions to uncatalyzed reactions using DIC and Oxyma as the soluble activating agents. This will be a point of investigation in future studies.

Some caveats are worthwhile mentioning here. First, while we refer to the immobilized DMAP unit as a catalyst, we have no evidence that a single DMAP unit on the resin mediates multiple acylation reactions (the molar ratio of the DMAP unit and the substrate is ≈1:1). We use the term “catalyst” loosely based on its historic use as a catalyst in solution. Second, all of the reactions carried out in this study were open to the air, which, in South Florida and coastal North Carolina, tends to be humid, and were carried out in solvents that have not been dried rigorously. It is possible that thorough elimination of low levels of water, which can react with the acylpyridinium intermediate, might improve the results of reactions employing soluble DMAP catalyst, which were markedly inferior to the results achieved using immobilized catalyst. However, we are largely interested in using this process for the construction of one bead one compound combinatorial libraries (25) and other applications where the maintenance of rigorously anhydrous conditions is difficult. Third, while a virtuous “proximity catalysis” effect was evident in all of the reactions explored in this study, it remains to be determined whether this will always be the case, for example, in the synthesis of larger N-methylated peptides, especially macrocycles, that are of keen interest as ligands for “undruggable proteins (26, 27). A potential concern, at least using the architecture reported in this study, is that as the N terminus of the peptide moves farther away from the C-terminally anchored catalyst, acyl transfer may become less efficient. Indeed, the data shown in Fig. 7 and SI Appendix, Fig. S12, where the insertion of linkers of different lengths between the lysine side chain and the catalyst resulted in differences in the side products produced from missed couplings, are consistent with the idea that the orientation of the catalyst and the reactive amine is important. In future iterations of resin design, it might be advantageous to include multiple copies of DMAP per substrate chain spaced at different intervals from the N terminus. This is under investigation. It is interesting to note that in the aforementioned study by Wei et al., the best results were obtained with four equivalents of the Oxyma catalyst per substrate (24).

Finally, it will be interesting to see if immobilized DMAP or other nucleophilic catalysts may accelerate other sluggish coupling reactions beyond the synthesis of peptides, such as the formation of sulfonamides from amines and sulfonyl chlorides (28) or Dakin–West transformation of carboxylic acids into ketones (29). This work is also in progress.

Materials and Methods

Optimal Synthetic Protocol for the Construction of TentaGel Beads Displaying a DMAP-Like Catalyst.

TentaGel-NH₂ (Rappe Polymere GmbH) beads were incubated with four equivalents (relative to the amino groups on the resin; 2.5 mM) each of Mtt-Lys(Fmoc)-OH, DIC, and Oxyma in DMF for 2 h at 37 °C. The Fmoc protecting group on the side chain was removed by two successive treatments with 20% piperidine in DMF for 5 min each. DMAP-containing acid 1 was dissolved in DMSO and the mixture heated gently until full dissolution of the acid. This was then mixed with an equimolar ratio of DIC and Oxyma in DMF so as to produce a solution in which the concentration of each reagent was 2.5 mM and the DMSO/DMF ratio was 1:4. This solution was added to the resin along with three equivalents of collidine and stirred at 37 °C for 2 h.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All of the study data are included in the article and/or SI Appendix.