Complementary Strategies for Installation of Thioimidates into Peptide Backbones

Abstract

Thioimidates are a precursor and synthetic branch point to access either thioamide or amidine isosteres of the native amide (peptide bond). Previous syntheses of thioimidate-containing peptides were prone to side reactivity and required slow, cumbersome steps that were difficult to monitor. We describe a more efficient approach to directly couple thioimidates onto the growing peptide chain. This work also outlines optimal conditions for thioimidate formation on solid phase and identifies potential off-target sites for alkylation that impact the choice of protecting group.

Graphical Abstract

Introduction

Thioamide isosteres in peptides have been applied in a variety of settings to probe for intermolecular effects along the peptide chain. The electronic structure of the sulfur atom creates differences in both the hydrogen bonding preferences,^1-4 hydrophobic properties,^5,6 and stereoelectronic effects^7-12 surrounding the thioamide. Thioamides can also act as fluorescent quenchers to study conformational effects and proteolytic activity.^13-21 Alternatively, thioamides are more reactive than traditional peptide bonds, which enables site-selective conversion of thioamides into other functional groups.^22-27 Finally, thioamides occur naturally, suggesting they contribute unique and distinct properties from native oxoamide bonds that drove nature to evolve the bio-machinery necessary for their incorporation.²⁸

Thioamides are prone to a variety of side reactions during solid-phase peptide synthesis (SPPS), however, including loss of stereochemical integrity of the α-CH stereocenter on the thioamide-containing amino acid,^29-32 and acid-promoted chain scission during final deprotection and cleavage from the resin.^33-35 We introduced the thioimidate as a comprehensive reversible protecting group to resolve both of these synthetic pitfalls.^29,30,33

In contrast to thioamides, the biophysical behavior of amidines within peptides is relatively unexplored. Nature has also developed biosynthetic pathways to incorporate amidines into peptide backbones,^36,37 which compels investigation of their properties. The paucity of amidinopeptide examples,^27,38 however, likely derives from an historic lack of synthetic methods to access them. We recently demonstrated how thioimidates can serve as a site for conversion into amidines, representing the first general method for the incorporation of amidines into peptide backbones through SPPS,^22,39 and providing an avenue to finally explore this traditionally ignored peptide-bond isostere.

Thus, facile methods to insert thioimidates into peptides can expand the study and available sequence space of thioamide and amidine isosteres in peptides. The current state of the art in terms of thioimidate incorporation into peptides, however, first requires generation of the thioamide amino acid and subsequent conversion into the thioimidate (3a→4, Scheme 1 top), a process that is slow and prone to generation of unproductive side products. Here we outline two complimentary strategies to perform this critical step either before or after introduction of the thioamide on solid phase. The methods presented here should increase the fidelity and efficiency with which thiomidates may be implemented for peptide backbone mutagenesis.

Scheme 1:

Improving operational efficiency through direct coupling of thioimidates into peptides

Results and Discussion

The installation of thioamides into peptides during SPPS is most typically achieved using activated thioacylation reagents derived from commercially-available Fmoc amino acids (Figure 1A).^40-43 The 4-nitrobenzotriazole reagents (1-S) were originally developed by Rapaport and coworkers,⁴⁴ and later refined by Chatterjee and coworkers.⁴⁵ While these reagents allow thioamides to be added to the growing N-terminus of the resin-bound peptide (2-S, Figure 1A), they are not without problems. For example, the thioamide coupling step is not always quantitative, requiring Ac₂O capping in all cases to truncate unreacted chains. Critically, inclusion of significant quantities of oxoamide (2-O) can occur at these sites despite pure thioacyl reagent (Figure 1B and C, 1-O was not stable to isolation but comparison of the two LC-MS traces reveals that no 1-O is present within the sample of 1-S prior to coupling. Molecular identity confirmed by observance of the molecular ion in the peaks indicated).^3,22 Thus, a more robust reagent that can be activated with conventional coupling agents (i.e. HATU, DIC, etc.) would improve the fidelity of thioamide and thioimidate insertion during SPPS.

Figure 1:

Coupling of the thioamide to the growing peptide (A), despite pure thioacyl reagent (B), can lead to significant oxoamide impurities in the final peptide (C)

Our initial report of the introduction of thioimidates into peptides relied on first coupling the sometimes problematic thioacyl reagent (1-S) to resin-bound peptide (Scheme 1 top).^29,30 A lengthy alkylation step to convert the thioamide into the thioimidate then followed (3a→4), wherein the progress of the alkylation was slow, sequence dependent, and cumbersome to monitor. A beneficial and more efficient alternative would be to couple the thioimidate into the peptide directly, ideally without introducing more steps or special considerations compared to existing methods.

We settled on the synthesis and coupling of thioimidate dipeptides (Scheme 1 bottom), where the overall synthesis and coupling of the dipeptide to resin required the same number of steps to reach the final on-resin thioimidate as the conventional thioacyl route. In practice, because the thioacyl approach (Scheme 1, top) requires coupling of the alanine to the resin prior to coupling with thioacyl reagent, this represents an additional, albeit routine, step that is not accounted for in the comparison between both approaches.

The operational efficiency of the thioimidate dipeptide process, however, would be higher because a slow on-resin alkylation step is replaced with a faster and more easily monitored solution alkylation (6→7, Scheme 2, example synthesis). The only remaining hurdle was liberation of the final C-terminal carboxylate, which required a protecting group that could be unveiled without the use of acid (to preserve any side-chain functional group protection) or base (to preserve the N-terminal Fmoc protecting group). The allyl ester protecting group fit these needs as the Pd-catalyzed deprotection conditions are orthogonal to both Fmoc and side-chain protecting groups.⁴⁶ However, the compatibility of Pd catalysts for allyl ester deprotection in the presence of thioimidates was unknown, and Pd has been shown to oxidatively add to electron-deficient C─S bonds.^47-49

Scheme 2:

Example synthesis of thioimidates for direct coupling (4 steps).

Our initial investigations into Pd-catalyzed deprotection of the allyl ester in 7 employed Pd₂(dba)₃ as the Pd(0) source (Table 1). Ligand denticity was observed to play a key role in the successful generation of 9 (entries 1–3 versus 4–6). Regardless of electronic and steric variation, monodentate ligands resulted in poor conversion and low yields of the desired product (entries 1–3). Alternatively, bidentate ligands afforded full conversion and a significant increase in yield (entries 4–6).

Table 1:

Optimization table

Entry	Precatalyst	Exogeneous Ligand	PhSiH3	Additive	%Yield	%Conversion
1	5 mol% Pd2(dba)3	20 mol% PPh3	2 eq.	–	22	34
2	5 mol% Pd2(dba)3	20 mol% PCy3	2 eq.	–	32	54
3	5 mol% Pd2(dba)3	20 mol% P(OEt)3	2 eq.	–	15	29
4	5 mol% Pd2(dba)3	10 mol% dppe	2 eq.	–	54	100
5	5 mol% Pd2(dba)3	10 mol% Xant-Phos	2 eq.	–	59	100
6	5 mol% Pd2(dba)3	10 mol% dppf	2 eq.	–	62	100
7	5 mol% Pd2(dba)3	10 mol% dppf	1.2 eq.	–	39	100
8	5 mol% Pd2(dba)3	10 mol% dppf	8 eq.	–	64	100
9	5 mol% Pd2(dba)3	10 mol% dppf	2 eq.	3 eq. NEt3	65	100
10	1 mol% PdCl2(dppf)	–	2 eq.	1 eq. NEt3	66	100
11	0.1 mol% PdCl2(dppf)	–	2 eq.	1 eq. NEt3	69	100
12	5 mol% Pd2(dba)3	10 mol% dppf	2 eq.	3 eq. HCO2H	37	100

^aReactions were analyzed by LC-MS and species were identified by their molecular ion.

Yields and conversion were determined through comparison of peak integration based on UV-vis absorption.

Two equivalents of PhSiH₃, the nucleophile for the Pd-allyl intermediate, were deemed necessary for the reaction (entry 6 versus 7), but little benefit was observed for higher loading of the silane (entry 8). We screened several other silicon, carbon, and nitrogen nucleophiles, but none showed any advantage over PhSiH₃ (see SI Table S5).

Thioimidates are stable to trialkyl bases used commonly in peptide synthesis. Unsurprisingly, the presence of NEt₃ had no significant impact on the yield of 9 (entry 9), but provided an avenue to simplify the reaction set up. Using NEt₃ as an in-situ reductant, the air-stable precatalyst PdCl₂(dppf) could be used in place of a Pd(0) source (entry 10), even at reduced catalyst loading (entry 11). Finally, the identity of the solvent and reaction temperature had little impact on the yield (see SI Tables S6 and S7).

Thioimidates are known to break down to thioesters in the presence of acid in hydroxylic solvents,^33,50 and thioesters were observed in all reactions described in Table 1. Thus, we hypothesized that the modest yield of 9, despite complete conversion of 7, was due to the instability of 9 under the conditions of LCMS analysis. The internal carboxylic acid group in 9 would exacerbate this side-reactivity, leading to unavoidable instability of the product. This hypothesis was supported by our observation that the inclusion of HCO₂H during the reaction reduced the yield of 9 (entry 12).

Because of the compatibility of Pd-catalyzed methods with SPPS and amino-acid activation chemistry,⁵¹ we sought to avoid any deleterious reactivity of the thioimidate dipeptide and further increase the efficiency of our strategy by combining the Pd-catalyzed deprotection and activation of the carboxyl group with HATU into a single operation to realize quantitative formation of amide 10 (Scheme 3). This experiment lends further support to our hypothesis that modest yields in Table 1 were due to decomposition of 9 during analysis. This one-pot procedure directly transferred to coupling onto solid-phase (Scheme 2 and Table 2).

Scheme 3:

One-pot procedure for allyl-deprotection and coupling.

Table 2:

Compatibility of Pd-catalyzed de-allylation conditions with thioimidates that can be converted into thioamides and amidines

entry	Xaa	Yaa	R	yield (%)
1	Phe	Ala	Azb	92
2	Phe	Ala	Me	>95
3	Ala	Glu(tBu)	Me	>95
4	Asp(tBu)	Gly	Azb	90
5	Cys(Trt)	Tyr(tBu)	Me	>95
6	Trp(Boc)	Ala	Me	89
7	Glu(tBu)	Ile	Me	93
8	Arg(Pbf)	Ala	Me	90a

Yield determined by coupling 1-pyrenebutyric acid onto the N-terminus after dipeptide coupling and Fmoc removal. Peptides were then cleaved and analyzed by LCMS monitoring at 330 nm. Resin was 2-chlorotrityl Tentagel, cleavage with 20% HFIP/DCM.

^aDeprotected with 10 mol% Pd(dppf)Cl₂ for 2 hours.

Each allyl-protected dipeptide in Table 2 was stable to storage, and was coupled to the test sequence using the one-pot procedure described above and then cleaved for analysis. In general, we observed excellent yields of coupled thioimidate with minimal oxoamide impurity (based on LCMS analysis of crude cleavage mixtures. See SI for further details). Thus, because the progress of dipeptide construction and thioimidate protection can be readily monitored and characterized during small-molecule synthesis, this new method of directly coupling protected thioamides is both more efficient and highly preferable to existing methods.

The methyl and 4-azidobenzyl thioimidates in Table 2 were selected for examination because they are the only two thioimidates that can be harnessed for further synthetic utility.^22,29,30,39 While Pd-catalyzed methods have already been demonstrated to be compatible with all 20 amino acids and side-chain protecting groups,^46,51 the reactivity of Pd(0) with thioimidates was unknown. Thus, the principle conclusions of the results in Table 2 was that the methyl and p-azidobenzyl thioimidate protecting groups are compatible with the Pd-catalyzed chemistry under consideration. We found excellent compatibility with a variety of amino acids, though noted that arginine required higher catalyst loading and reaction time.

The dipeptide-coupling approach to install preformed thioimidate linkages into peptides discussed above provides protected thioamides or sites for further elaboration into amidines without any subsequent steps needed before continuing conventional elongation of the peptide through SPPS. The drawback of this dipeptide approach, however, is that it decreases the sequence flexibility and diversity that is possible through coupling of a single thioacyl amino acid. For this reason, methods to efficiently convert thioamides into thioimidates on-resin are still valuable, yet a systematic investigation of the reaction conditions and potential pitfalls for this transformation has not been reported.

We therefore sought to probe the model reaction described in Table 3 to identify conditions to increase the efficiency with which thioamides within peptides could be alkylated to produce thioimidates (Scheme 1, 3→4). What was immediately clear, and not shown in Table 3, was that N,N-diisopropylethylamine (DIEA) was the most effective base for the transformation compared with weaker bases such as triethylamine and N-methylmorpholine (see Table S3). This result is unsurprising given DIEA is one of the strongest organic bases for SPPS that does not elicit deprotection of the N-terminal Fmoc protecting group. Accordingly, subsequent investigation was only carried out with DIEA.

Table 3:

Efficiency of alkylation to form thioimidate on-resin

entry	R-X	solvent	yield (%)
1	Me–I	DMF	66
2	Bn–Br	DMF	24
3	Allyl–Br	DMF	16
4	4-N3–BnBr	DMF	56
5	4-N3–BnBra	DMF	92
6	4-N3–BnBr	DCM	54
7	4-N3–BnBr	MeCN	46
8	4-N3–BnBr	acetone	35
9	4-N3–BnBr	MeOH	30
10	4-N3–BnBr	THF	18

^aAddition of TBAI at 5 mM. Percent yield calculated based on integration of LC peaks (330 nm, R’ = 1-pyrenebutyric acid) corresponding to 11 and 12 (12/(11+12)). Resin was 2-chlorotrityl Tentagel. Cleavage with 2% TFA in DCM.

With the base additive established, we next investigated the rate of thioimidate formation with several alkylating reagents. We identified MeI and the 4-azidobenzyl bromide as providing the most efficient thioimidate formation after 6 h (Table 3, entries 1 & 4). The increased efficiency of 4-azidobenzyl bromide (entry 4) relative to benzyl bromide (entry 2) is not immediately obvious as the electron-donating ability of the azide group is chameleonic,⁵² but certainly appears to be beneficial in this context. Gratifyingly, the alkylation with 4-azidobenzyl bromide was greatly accelerated through an in-situ Finkelstein reaction with tetrabutylammonium iodide (entry 5). Finally, DMF and DCM appeared to be the best solvents for the reaction (Table 3, entries 4 & 6–10).

The rapid reactivity of both MeI and 4-azidobenzyl bromide is certainly advantageous for thionopeptide synthesis, as deprotection conditions to convert the respective thioimidates back to the thioamide are well established. Restoration of the thioamide from 4-azidobenzyl thioimidates can be achieved through reductive conditions using mild reagents such as PPh₃ or DTT.³⁰ Methyl thioimidates must be deprotected using H₂S, a toxic gas,²⁹ but methyl thioimidates are also useful as reactive handles for the site-specific insertion of amidines into peptide backbones.²²

Extended exposure to alkylating reagents introduced the risk of alkylating reactive side-chain functional groups. For example, the imidazole of histidine is often protected with a Boc or trityl protecting group, but has been shown previously to alkylate with loss of the protecting group (Figure 2A).⁵³

Figure 2:

Synthesis of thiono S-Tag, displaying flexibility and functional group tolerance of thioimidate protection. Side products arising from methionine oxidation, overalkylation, and Edman-like degradation are all avoided. (i) 500 mM p-azidobenzyl bromide, 50 mM DIEA, 5 mM TBAI. (ii) Routine coupling and acetylation (see SI). (iii) 75/20/2.5/2.5 TFA/DCM/TIPSH/m-cresol, 1 h. (iv) 0.1 M DTT, 0.1 M DIEA, DMF, 2 h.

To investigate the potential for undesired alkylation elsewhere in the peptide, we synthesized S-tag, derived from RNase A,^54,55 with a thioamide between Glu₉ and Arg₁₀ (13, Figure 2B). Conversion to the thioimidate at this point in the synthesis allowed us to probe the alkylation potential of three specific amino acids of concern—Arg₁₀, His₁₂, and Met₁₃. Notably, off-target alkylation does not appear to be an issue for amino-acid side chains protected as an ether, ester, amide, or carbamate,^22,29,30,39 but the protected versions of Arg, His, and Met have not received considerable attention in previous work and these amino acids present side-chain functional groups that could pose a risk of reacting with the alkylating reagent.

Upon exposure to conditions to convert the thioamide to the thioimidate (13→14), we only observe alkylation of His₁₂ (signified by concomitant loss of the Trt side chain protecting group as expected⁵³). We did not observe any evidence of alkylation of Arg₁₀ and Met₁₃. The undesired alkylation of His₁₂ did not affect subsequent elongation of the remaining peptide (14→15). Moreover, the reduction with DTT to liberate the protected thioamide also returned the unalkylated His in the final cleavage and isolation (15→16). Thus, one must be cognizant of the potential for His to be alkylated during conversion to thioimidate, but the overall process is invisible upon final work up. Of course, we also point out that any concerns regarding undesired alkylation of natural and unnatural amino acid side chains can be removed altogether by employing the thioimidate dipeptide strategy described above (Scheme 1 bottom).

Conclusion

Thioamides represent a single-atom chemical tool to perturb and study hydrogen-bonding, hydrophobic, stereoelectronic, and pharmacological effects in peptide biomolecules. The installation and preservation of thioamides in peptides, however, is far from straightforward. Protection of thioamides as thioimidates has been shown to dramatically reduce side-products generally associated with thionopeptide synthesis. Additionally, recent reports further demonstrate how thioimidates can serve as reactive handles for the site-selective insertion of amidines,²² an understudied peptide bond isostere. Thus, this work expands the options by which thioimidates may be implemented during the synthetic design of peptides with backbone mutations to the native oxoamide. The methods described here therefore increase the scope of thioimidates, thioamides, and amidines in peptide synthesis.

Data availability statement

The data underlying this study are available in the published article and its Supporting Information.