Thiol-independent peptide ligation: A review of mechanisms

Abstract

The mechanisms of several types of thiol-independent peptide ligation reported over the past 14 years have been reviewed.

Introduction

Ligation in the peptide field has come to be recognised as the linking of a side-chain unprotected N-terminal peptide to a side-chain unprotected C-terminal peptide by a prior tethering process of the two fragments such that the formation of the new peptide link takes place intramolecularly. The first reported example of chemical peptide ligation, dubbed native chemical ligation (NCL), was reported by Dawson et al. ¹ in 1994. They used the SH group of an N-terminal cysteine (Cys) of one peptide as the site of tethering and an activated carboxyl residue of a C-terminal amino acid to facilitate its attack by the SH group of the N-terminal cysteine. Specifically, an alkyl thioester (–COSR) of a side-chain unprotected C-terminal peptide was reacted with the SH group of the Cys residue of the side-chain unprotected N-terminal peptide to afford, via a thioester exchange, a (linked) thioester which generated the new peptide linkage via an S to N-acyl transfer. This Cys-based ligation has been widely adopted over the past 25 years or so to synthesise peptides and proteins, some carrying post-translational modifications (e.g. N/O-glycosylation, phosphorylation). The mechanism is well-known and not in dispute; indeed, it can be said that the concept of the tethering methodology was mechanism-based. For this reason, the mechanism of NCL will not be considered in this review.

In response to the low abundance of Cys in proteins (ca. 1.4%), several other methods have been developed over the years, among which β- or γ-thiol-containing unnatural amino acids were used as Cys surrogates. In the pursuit of thiol-independent peptide ligation methods, 16 years after the introduction of Cys-based ligation, Li et al. ² in 2010 described a prototype of serine/threonine-based ligation (STL) and in 2013 Zhang et al. ³ published a full description. (In 2018, Liu and Li ⁴ published a comprehensive review of the process.) Potentially, this could extend considerably the scope of chemical ligation, since these amino acids are very abundant in natural proteins (ca. 12.7% in total). In STL, the OH group of an N-terminal serine (Ser) or threonine (Thr) of an unprotected peptide was used as the site of tethering in an analogous way to that used by the Cys-based ligation, but instead of the use of a C-terminal thioester, a C-terminal salicylaldehyde (SAL) ester of an unprotected peptide was used.

Prior to the report of STL, ² Bode et al. ⁵ in 2006 reported an important variant in which the ligation was effected by a decarboxylative condensation between chemically modified natural α-amino acids, a C-terminal α-keto-acid and an N-terminal hydroxylamine. This became known as the α-keto-acid-hydroxylamine amide-forming (KAHA) ligation. Here, there is no prior tethering and the success of the selective bimolecular reaction is due to the intrinsic reactivity as a nucleophile of the hydroxylamine which reacts rapidly with the C=O group of the α-keto-acid. There are two closely related processes, one uses an unsubstituted N-terminal hydroxylamine (type I) and the other uses an N-terminal O-benzoyl hydroxylamine (type II).

In 2017, de Figueiredo et al. ⁶ reported an innovative procedure for dipeptide synthesis that uses an inverse activation strategy (e.g. amine activation) and it occurs with no epimerization. The activation of the amino group is effected by conversion into an N-acylimidazole derivative. It should be said that this method does not compare in scale with the STL and KAHA ligations, since it simply describes dipeptide syntheses and does not use unprotected amino acids. Indeed, de Figueiredo et al. ⁶ make no claim that the method is a ligation. However, because the mechanism of linking two amino acids is novel and interesting, I have included it, perhaps taking a liberty with the use of ‘ligation’.

Recently, in 2019, a novel chemoselective, high-yielding α-aminonitrile ligation was reported by Canavelli et al. ⁷ which exploits only prebiotically plausible molecules – hydrogen sulphide, thioacetate and ferricyanide – to yield α-peptides in water. The ligation, which is carried out at room temperature (RT), is extremely selective for α-aminonitrile coupling and tolerates all of the unprotected 20 proteinogenic amino acid residues. Here, ligation does not have the meaning that it had with STL and KAHA ligation in which large unprotected fragments are linked. Rather, Canavelli et al. ⁷ describe the ligation as iterative in that unprotected amino acid-derived units are added one at a time.

The mechanisms of the above ligation processes will be discussed below. Several of them have useful variants, but only the basic mechanism of each process will be considered. Each also, to varying extents, has seen many applications with impressive syntheses of large peptides and proteins, but these will not be enumerated here.

Two well-known peptide ligations are not considered here for the reasons given below. Peptide ligation via click chemistry, first reported in 2005 by Franke et al., ⁸ generates unnatural peptides containing 1,4-substituted [1,2,3]-triazole linkers. They are produced by a Cu-catalysed reaction between peptide fragments terminating in an azide and an alkyne, respectively. This ligation was ruled out because a new peptide bond is not formed. So too was Staudinger peptide ligation which was first reported in 2000 by Nilsson et al. ⁹ There the original process ⁹ started from an alkyl thioester of a peptide which was converted by a transthiolation reaction with o-(diphenylphosphino)-benzenethiol into a thioester of a peptide with a C-terminal –CO–S–C₆H₄–(o-diphenylphosphino) group. This was reacted with a peptide with an N-terminal azido group whereupon reaction between the phosphine group and the azide group in a ‘traceless’ Staudinger process produced, via a rearrangement in which extrusion of the phosphine fragment took place, a new peptide link. The dependence of this process on a thio group exempted it from consideration here.

The purpose of this review is to describe mechanisms of several types of thiol-independent peptide ligation reported over the past 14 years that have not hitherto been brought together and reviewed. These methodologies are becoming important as alternatives to the well-known native peptide ligation introduced in 1994 involving the use of an N-terminal cysteine. Moreover, the wider community of bioorganic chemists would be interested in the often ingenious methods that are being developed to link unprotected peptides together. Moreover, an understanding of the mechanisms of these methods can be very useful in the search for new ligation methodologies. Also a few current drugs and several in development are small peptides and convenient methods for their synthesis are always in demand.

The four ligation processes discussed here can be classified into two main categories, amino acid activation and aminonitrile activation, the former further sub-classified into acid activation, of which there are two variants, and amine activation in which there is just one. The mechanisms of the four ligation processes will be discussed under those headings. Each section will have two self-explanatory sub-headings: Process and Mechanism.

Amino acid activation

Acid activation

STL

Process

As will be seen, the chemoselective STL process, being an O-analogue of Cys-based ligation (NCL), is also mechanism-based. The key to the tethering process is the reaction of a free amino group of a side-chain unprotected N-terminal amino acid with the aldehyde group of a side-chain unprotected C-terminal SAL ester at the C-terminal carboxyl group. The reaction is carried out in a pyridine/acetate buffer and produces an oxazolidine as an intermediate which upon acid treatment (trifluoroacetic acid/water) yields the ligated product (Scheme 1). ³

Scheme 1.

The chemoselective serine/threonine ligation (STL) process.

Mechanism

The detailed mechanism ³ (Scheme 2) involves reaction between the free NH₂ group of an N-terminal Ser or Thr of a side-chain unprotected peptide (2) and the aldehyde group of a C-terminal SAL ester of a side-chain unprotected peptide (1) to generate initially an imine (3). This suffers an intramolecular attack by the OH group of Ser or Thr at C=N to form an oxazolidine (4). Both the steps are likely reversible. Next, a nucleophilic attack by the secondary amino group of the oxazolidine upon the phenolic ester triggers an O to N [1,5]-acyl transfer to yield, via an irreversible step, an N,O-benzylidene acetal-linked product (5), which upon acid treatment is converted into a natural peptide linkage (6). Importantly, since no reaction on the C-terminal C=O group occurred during the ligation, no epimerization was expected and none was found. The chemoselectivity of STL is achieved by the 1,2-hydroxylamine bi-functional group present at the N-terminal serine/threonine residue. Neither the free amino group of a lysine (Lys) residue nor the N-terminal amine of the peptide SAL ester affected the ligation process because the imine formation from each of these NH₂ groups is fast, reversible and non-productive. The lack of a vicinal hydroxyl group to each means that no O to N-acyl transfer can occur to afford a stable intermediate, and hence the unprotected NH₂ group of an internal Lys and the free N-terminal amine of the peptide SAL ester is unlikely to compete with N-terminal Ser/Thr for reaction with the C-terminal SAL ester. The key to the use of the SAL ester in STL is that the resultant N,O-benzylidene acetal can be readily removed by acidolysis to restore the natural peptide linkage. Here, the labile benzylic C–O and C–N bonds ensured a rapid hydrolysis.

Scheme 2.

The mechanism of the chemoselective serine/threonine ligation (STL) process.

α-Ketoacid-hydroxylamine amide-forming (KAHA) ligation

Process

In 2006, Bode et al. ⁵ described a novel ligation process that involves the coupling of hydroxylamines and α-ketoacids to form amides which proceeds in the presence of reactive functional groups, requires no reagents or catalysts and produces only water and CO₂ as by-products There are two variants of the KAHA ligation reaction, each requiring no reagent or a catalyst (Scheme 3). The type I KAHA ligation (a) relies on the reaction in DMSO or MeOH at 40 °C of an α-ketoacid (7) and an O-unsubstituted hydroxylamine (8) to produce the amide (9) with CO₂ and H₂O as by-products, whereas the type II KAHA ligation (b) involves the reaction in aqueous buffer at 40 °C between an α-ketoacid (7) and an O-benzoylhydroxylamine (10) to produce the amide (9) with CO₂ and benzoic acid, instead of water, as by-products.

Scheme 3.

The two variants, (a) type I and (b) type II of the α-ketoacid-hydroxylamine amide-forming (KAHA) ligation reaction.

Mechanism

In 2012, Pusterla and Bode ¹⁰ reported their investigations into the mechanism of KAHA ligation. They carried out three ¹⁸O-labelling experiments using ¹⁸O-phenalkylhydroxylamines and ¹⁸O-phenalkyl-α-ketoacids, the results of which (Table 1) ruled out some mechanisms for both type I and type II. Surprisingly, in the type I process, the oxygen of the amide product originated from the hydroxylamine (entry 2) rather than from the ketonic C=O group of the α-ketoacid (entry 1). However, in the type II process, the oxygen of the amide product did originate from the ketonic C=O group of the α-ketoacid (entry 3).

Table 1.

The results of three ¹⁸O-labelling experiments using ¹⁸O-phenalkylhydroxylamines and ¹⁸O-phenalkyl-α-ketoacids carried out in studies of the mechanism of the type I and type II α-ketoacid-hydroxylamine amide-forming (KAHA) ligation reaction.

Entry		Label transfer (%)
1		0
2		>80
3		>80

Their proposed mechanism for the type I KAHA ligation that took account of those results involved lactone and oxiridine intermediates. That mechanism will not be reproduced here, since a subsequent theoretical study showed that it was energetically less favoured than an alternative pathway. The labelling pattern obtained with the type II process indicated a quite different mechanism which was not delineated at that time.

Some 5 years later in 2017, Patil ¹¹ reported his computational investigations of the mechanisms of both types of KAHA ligation. For the type I process, the energetically favoured mechanism that he proposed differed markedly from that of Pusterla and Bode, ¹⁰ yet still took account of the results of their ¹⁸O-labelling experiments (Table 1). The model compounds used in the calculations for the type I process were phenylpyruvic acid (7; R¹ = PhCH₂) and N-phenethylhydroxylamine (11; R² = PhCH₂). His revised mechanism of the type I process (Scheme 4) involved initial attack by the hydroxylamine (11) upon the C=O group of the α-ketoacid (7) to give a tetrahedral intermediate (TI) (12) which suffered a loss of H₂O, the protonation of the departing OH group being provided by the adjacent CO₂H group. The product was a protonated nitrone (13) which decarboxylated to a carbanion (14). A hydroxyl group transfer yielded a hydroxyimine (15) which tautomerised to the product amide (16). A coherent mechanistic course was also delineated for the type II process. The model compounds used in the calculations were phenylpyruvic acid (7; R¹ = PhCH₂) and O-benzoyl-N-phenethylhydroxylamine (17; R² = PhCH₂). The proposed pathway (Scheme 5) involved attack by the O-benzoyl hydroxylamine (17) upon the C=O group of the α-ketoacid (7) with concomitant proton transfer from the CO₂H group to give a zwitterionic TI (18) which, in a manner unspecified by the author, ¹¹ yielded, via loss of CO₂ and benzoic acid, the same hydroxyimine (15) formed in the type I pathway, tantomerisation of which yielded the product amide (16). A possible mechanism for the hydroxyimine-forming step, the transformation of (18) to (15), involves the concerted loss of CO₂ and benzoic acid in which the proton attached to nitrogen is transferred to the departing benzoate ion, as shown in (19).

Scheme 4.

The mechanism of the type I α-ketoacid-hydroxylamine amide-forming (KAHA) ligation reaction.

Scheme 5.

The mechanism of the type II α-ketoacid-hydroxylamine amide-forming (KAHA) ligation reaction.

Amine activation

N,Nʹ-carbonyldiimidazole (CDI)-based ligation

Process

Classical amide synthesis methods involve the activation of the carboxylic acid to a reactive acyl derivative in order to allow facile reaction with an amine, and the foregoing methods all apply this methodology for the synthesis of peptides. Now, a recently reported innovative procedure for peptide synthesis goes through an inverse activation strategy (e.g. amine activation) and it occurs with no epimerization. ⁶ The activation of the amine is by conversion to an N-acylimidazole derivative.

Mechanism

In a typical dipeptide synthesis (Scheme 6), the activation of the amino group of an α-amino ester (20) involves reaction with CDI to convert it into an N-acylimidazole derivative (21). The carboxylic acid group of a second N-protected α-amino acid (22) attacks the C=O group of (21) displacing imidazole to yield a mixed carboxylic-carbamic anhydride (23). The attack by the NH group of (23) upon the distal C=O group initiates an intramolecular 1,3-acyl transfer which generates the dipeptide (24) with the loss of CO₂. ⁶

Scheme 6.

The process and mechanism of a dipeptide synthesis using the N,Nʹ-carbonyl-diimidazole (CDI)-based ligation.

Aminonitrile activation/ligation

Process

Addition of an amino acid to a peptide (25) is initiated by conversion of the terminal CO₂H group into a cyano group thus forming an α-aminonitrile (26) which is the starting-point for the three-step iterative cycle (Scheme 7). The first step is thiolysis of the nitrile group of (26) with H₂S to form a thioamide (27) and the second is its hydrolysis to form a thioacid (28). In the third step, oxidation of the thioacid (28) and reaction with an incoming α-aminonitrile monomer (29) yield a chain-extended nitrile (30) available to undergo another ligation cycle. A conceptual difference between this scheme, reported by Canavelli et al., ⁷ and NCL chemistry is the fact that the elongating peptides, not the monomers, are electrophilically activated.

Scheme 7.

The α-aminonitrile ligation process.

Mechanism

In early model experiments reported by Canavelli et al., ⁷ a key step in the efficiency of the process was identified as the conversion of an α-amino acid nitrile (AA-CN) (31) into its N-acetyl derivative (Ac-AA-CN) (33) via oxidative acetylation with potassium ferricyanide/thioacetic acid (32) (Scheme 8(a)). This allows two controlled reactions at pH 9 which proceed essentially quantitatively (Scheme 8(b)). First, reaction of (Ac-AA-CN) (33) with H₂S forms the corresponding thioamide (Ac-AA-SNH₂) (34), the mechanism involving attack by HS⁻ upon the nitrile group of (33) to give an intermediate iminosulfide (–(HS)C=NH) which tautomerises to (34). The mechanism of the hydrolysis of (34) involves attack by HO⁻ upon the C=S group of (34) and departure of ammonia to form the N-acetylated thioacid (Ac-AA-SH) (35). An α-aminonitrile (AA-CN) does not undergo a similar sequence of reactions with H₂S and H₂O, instead giving a mixture of products at the hydrolysis step. The N-acetylation of AA-CN is crucial because the electron-withdrawing effect of the acetyl group in Ac-AA-CN labilises the nitrile group to thiolysis to Ac-AA-SNH₂, and more importantly, to hydrolysis of the thioamide to give a single product, the thioacid Ac-AA-SH (Scheme 9). (The mechanism of the oxidative acetylation, which may be effected by several oxidising agents besides ferricyanide, was suggested more than 20 years ago by Lui and Orgel. ¹² In the first step (Scheme 10), oxidation of the thioacid (36) yields a diacyl compound containing an S–S bond (37). This is a good electrophile which an amine (38), in the second step, can attack to form the corresponding amide (39) and expel RCOS-SH (40) as the leaving group.)

Scheme 8.

(a) The conversion of an α-aminonitrile (AA-CN) into an α-amidonitrile (Ac-AA-CN) via oxidative acetylation with potassium ferricyanide/thioacetic acid and (b) the mechanism of thiolysis of an α-amidonitrile (Ac-AA-CN) to an α-amidothioamide (Ac-AA-SNH₂) and of its hydrolysis to an α-amidothioacid (Ac-AA-SH).

Scheme 9.

A schematic showing the sequential thiolysis and hydrolysis of an α-aminonitrile (AA-CN) and of an N-acetyl-α-aminonitrile (Ac-AA-CN) in which only the latter yields a unique hydrolysis product, an N-acetylated thioacid (Ac-AA-SH).

Scheme 10.

The general mechanism of the conversion of a compound containing an amino group into its N-acyl derivative via oxidative acylation with potassium ferricyanide/thioacetic acid.

The quantitative nature of these reactions was demonstrated by following the four-step conversion of Gly-CN (50 mM) to Ac-Gly₂-CN (Scheme 11) by ¹H NMR (600 MHz, H₂O:D₂O::98:2, 25 °C). ⁷ In the first step, Gly-CN was converted into Ac-Gly-CN by quantitative oxidative acetylation using thioacetic acid/potassium ferricyanide at RT. Thiolysis (H₂S) of Ac-Gly-CN at pH 9 in water at RT for 10 min to Ac-Gly-SNH₂ was also quantitative, but hydrolysis in water at pH 9 at 60 °C to Ac-Gly-SH was only 80% complete after 24 h. The crucial chain-lengthening process in which Ac-Gly-SH is converted to Ac-Gly₂-CN using Gly-CN/potassium ferricyanide at RT was quantitative after 20 min.

Scheme 11.

The demonstration of the quantitative nature of the α-aminonitrile homologation by following the four-step conversion of Gly-CN to Ac-Gly₂-CN by ¹H NMR.

Impressively and importantly, the iterative ligation cycle in which an N-acetyl-α-amino acid nitrile (41) successively undergoes thiolysis to an N-acetyl-α-aminothioamide (42), hydrolysis to an N-acetyl-α-aminothioacid (43) and homologation via reaction with an α-amino acid nitrile (44) and an oxidising agent, neatly summarised by Canavelli et al. ⁷ (Scheme 12), tolerates all of the 20 proteinogenic amino acid residues.

Scheme 12.

The iterative ligation cycle in which an N-acetyl-α-amino acid nitrile is converted to a higher homologue.