Native chemical ligation runs on an activated acyl donor: a peptide whose C-terminal carboxyl is carried as a thioester, which the thiol of an N-terminal cysteine can attack in water at neutral pH. Making that thioester is where a ligation project can stall. The group has to sit in a narrow window: stable enough to survive chain assembly, cleavage, HPLC and lyophilisation, reactive enough to hand its acyl group to a thiol under mild aqueous conditions. Every route to it is an attempt to hit that window. The ligation step that consumes this material is treated separately in the native chemical ligation guide; this article is about getting the thioester into your hands, and about the two losses running in the background of every route: epimerisation at the residue carrying the thioester, and hydrolysis of the thioester itself.
Why Fmoc chemistry is hostile to a thioester
Fmoc removal uses piperidine, a secondary amine, in DMF. A thioester is an acyl donor and piperidine is a nucleophile, so every deprotection cycle is an opportunity for the two to react. Reviewing the o-aminoanilide linker family, Blanco-Canosa and co-workers state the problem directly: because Fmoc-SPPS is the most adopted methodology for peptide synthesis, the direct preparation of peptide thioesters is poorly efficient, which necessitates other approaches that overcome the piperidine-mediated aminolysis of the peptide–COSR [1]. The loss is not a slow background decomposition to be tolerated; it is a competing reaction with a nucleophile present in large excess, once per residue.
That leaves two structurally different escapes. Assemble the chain by a chemistry that never sees piperidine, or assemble it by Fmoc chemistry on a group that is not yet a thioester and convert it afterwards. Both are in routine use, and the choice is forced by what else is on the peptide. A 2011 review of the field frames the trade-off: Boc-based solid-phase synthesis provides very reliable access to peptide thioesters, but the acid lability of many peptide modifications and the requirements of most parallel peptide synthesizers call for the milder Fmoc-based synthesis, which is more cumbersome for thioesters and typically proceeds with lower yields than the synthesis of peptide acids and peptide amides [2]. Read that as a planning statement, not a verdict on which chemistry is better.
The Boc route, and what in-situ neutralisation changed
Boc chemistry removes the temporary protecting group with TFA, generating a protonated N-terminal ammonium salt that must be neutralised before the next coupling. The in-situ neutralisation protocols of Schnölzer and colleagues perform that neutralisation simultaneously with coupling rather than as a separate step, using high concentrations of Boc-amino acid-OBt esters (above 0.2 M) plus base for rapid coupling, 100% TFA for Boc removal, and a single 30 s DMF flow wash between deprotection and coupling; single 10 min coupling times were used throughout, giving overall cycle times of 15 min manual and 19 min machine-assisted [3]. The reported gain was specifically on difficult sequences arising from sequence-dependent chain aggregation, and the reported syntheses included two roughly 50-residue segments and the corresponding 99-residue protein [3].
Two details of that work matter for thioesters in particular. First, the authors report that no racemisation was detected in the base-catalysed coupling step, and that they identified and eliminated several side reactions, among them pyrrolidonecarboxylic acid formation from Gln in hot TFA-DMF and chain termination by reaction with excess HBTU [3]. Second, the whole protocol assumes a peptide that can tolerate repeated 100% TFA and, at the end, the strong acid used to cleave a thioester resin.
Safety-catch sulfonamide linkers
The safety-catch idea is to anchor the peptide through a linkage that is chemically inert during assembly and is switched on only at the end. An acylsulfonamide is the classic example: as the N-acyl sulfonamide anion it is stable to piperidine and to TFA, and it becomes a competent acyl donor only after N-alkylation converts it into an activated imide, which thiolysis then converts to the thioester.
Loading is the weak point: acylating a sulfonamide nitrogen is harder than acylating an amine, and the harsher conditions used to force it are the ones that risk racemising the first residue. Fmoc-protected amino acid fluorides were found to be excellent reagents for acylating sulfonamide safety-catch linkers suitable for subsequent preparation of peptide C-terminal thioesters, giving high loadings on different types of resins with low levels of epimerisation [4]. Note the wording: low levels, not none. If you cannot afford to scramble the C-terminal residue, quantify the epimer content on a short model before committing a long chain to the same loading conditions.
A variant keeps the safety-catch activation but changes what attacks it. Alkylating the sulfonamide linker with a protected 2-mercaptoethanol derivative installs the thiol on the linker itself, and after chain assembly the thioester is generated on the solid phase as a consequence of an intramolecular N,S-acyl shift; depending on the TFA stability of the spacer separating the sulfonamide from the resin, treating the peptidyl resin with TFA gave either a soluble or a support-bound deprotected thioester [5]. This is the point where the safety-catch family and the acyl-shift family stop being separate ideas.
N-acylurea and N-acyl-benzimidazolinone linkers
The Dbz approach starts from a 3,4-diaminobenzoic acid linker. Its o-aminoanilides are stable synthetic intermediates that can be efficiently transformed into an aromatic N-acylurea, and the activation is done by acylating the free aromatic amine with nitrophenylchloroformate followed by addition of base, which promotes intramolecular attack of the anilide to form the resin-bound benzimidazolinone [6]. The design argument, in the authors' terms: the linker is a stable amide during chain assembly, and the key activation step uses the most robust reaction in solid-phase peptide synthesis, the acylation of an amine [6]. The resulting N-acyl-benzimidazolinone (Nbz) is not itself the species that ligates. It is a mildly activated acyl donor that exchanges with an exogenous thiol to give the thioester, and in the Dbz system that transthioesterification is primarily rate-limiting [1].
The first-generation linker had a documented failure mode. Synthesis of some glycine-rich sequences revealed limitations, such as diacylated products that cannot be converted into N-acylurea peptides — the free aromatic amine gets acylated during chain assembly, and a branched species results [7]. The second-generation linker bearing an o-amino(methyl)aniline (MeDbz) moiety was introduced to enable more robust chain assembly; its generality was illustrated by a pentaglycine sequence under different coupling conditions including microwave heating at coupling temperatures up to 90 °C, affording the unique and desired N-acyl-N'-methylacylurea (MeNbz) product, and the method was extended to all 20 natural amino acids with their ligation reactions [7]. Kinetic analysis of those ligations with model peptides showed the MeNbz peptide rapidly converts to arylthioesters that are efficient at ligation [7]. If you are assembling a Gly-rich segment on a first-generation Dbz resin and seeing an unconvertible mass, that is the diagnosis.
The same o-aminoanilide can be pushed further. Chemoselective oxidation of a peptide–Dbz–CONH2 with sodium nitrite at acidic pH gives a benzotriazole (Bt) species, and the resulting peptide–Bt undergoes transthioesterification with a low energy barrier, with the N-to-S acyl transfer not rate limiting and thiol exchange complete in under 5 min for all C-terminal residues tested; with benzeneselenol, LYRAF–Bt exchanged to the phenyl selenoester in under a minute [1]. That speed is not free; the price is below.
Devices built on an intramolecular N-to-S acyl shift
These routes install a β-amino thiol next to the C-terminal carbonyl and let the molecule activate itself. Because the shift is intramolecular and reversible, the acyl group sits mostly on nitrogen — where it is an unreactive amide during Fmoc assembly — and is drawn onto sulfur under the conditions chosen for conversion.
The bis(2-sulfanylethyl)amido (SEA) group is among the best characterised of these. A peptide carrying a bis(2-sulfanylethyl)amino group at its C-terminus reacts with a cysteinyl peptide in water at pH 7 and 37 °C to give chemoselective and regioselective formation of a native peptide bond, and the same report describes a solid support allowing straightforward synthesis of peptide SEA fragments using standard Fmoc/tert-butyl procedures [8]. The underlying equilibrium was then measured: SEA peptides equilibrate in aqueous solution with the corresponding S-2-(2-mercaptoethylamino)ethyl thioester through an N,S-acyl shift, and HPLC was used to follow the rate of equilibration for different C-terminal amino acids and the position of the equilibrium as a function of pH [9]. The thioester form participates efficiently in thiol–thioester exchange with 5% aqueous 3-mercaptopropionic acid, with the highest reaction rate obtained at pH 4, conditions the authors describe as significantly less acidic than those previously reported for related systems; the method was validated by synthesis of a 24-mer peptide thioester [9]. Two consequences follow. The C-terminal residue changes the equilibration rate, so a rate measured on one segment does not transfer to another. And pH is a control knob, not a fixed recipe: it sets both where the equilibrium sits and how fast the exchange runs.
That pH control is what makes the SEA group useful at a junction alkyl thioesters handle badly. Ligation of C-terminal peptidyl prolyl alkylthioesters with N-terminal cysteinyl peptides usually shows poor kinetic rates compared with other C-terminal residues, and the reaction is accompanied by formation of a deletion side product which is minimised by using a SEA thioester surrogate at mildly acidic pH [10]. Read that as scoped to the prolyl junction and that side product, not as a general claim that acidic pH improves ligations.
The cysteinyl-prolyl ester (CPE) device works by the same shift but with a different downstream step. A peptide bearing a CPE moiety at the C-terminus is transformed into a diketopiperazine thioester by intramolecular N-S acyl shift and then used for ligation, and the stereochemistry of the two residues decides where activation lands: with L-Cys-L-Pro or D-Cys-D-Pro the desired diketopiperazine thioester forms at the preceding amino acid residue, whereas with D-Cys-L-Pro or L-Cys-D-Pro a thiolactone forms at the C-terminal prolyl ester and ligation occurs at the C-terminal Pro instead [11]. Build a CPE segment with the wrong relative configuration and you do not get a failed reaction; you get a clean product joined at the wrong residue.
Two further variations show the range of the same mechanism. An N-alkyl cysteine (NAC) carried at the C-terminus after conventional Fmoc assembly was converted to the aryl thioester by treatment with 4-mercaptophenylacetic acid without significant side reactions, and the resulting thioester was used to prepare a 95-residue chemokine by a silver-free thioester method [12]. An N-4,5-dimethoxy-2-mercaptobenzyl (Dmmb) auxiliary placed on the backbone amide drives the N-S shift efficiently under mildly acidic conditions after Fmoc assembly, with the thioester obtained by treating the resulting S-peptide with sodium 2-mercaptoethanesulfonate; the approach was also amenable to on-resin preparation and was demonstrated on a 41-mer, a phosphorylated peptide and a 33-mer carrying a trimethylated lysine [13].
Reactivity is set by the thiol leaving group
A thioester's ligation rate is governed by how good a leaving group its thiol is, which is why the leaving group is routinely swapped in the ligation buffer rather than built in. Peptide α-thioalkyl esters are commonly used because they are easy to prepare, and they are rather unreactive, so the ligation is catalysed by in-situ transthioesterification with thiol additives [14]. With the older catalysts — a thiophenol/benzyl mercaptan mixture, or the alkanethiol MESNA — ligations typically took 24–48 h [14]. Screening thiols identified the property that matters: substituted thiophenols with pKa above 6 best combine the ability to exchange rapidly and completely with thioalkyl esters and to then act as effective leaving groups in reaction with the cysteine thiol, and (4-carboxymethyl)thiophenol (MPAA), a non-malodorous water-soluble thiol, gave an order of magnitude faster reaction in model studies and in the synthesis of the small protein OMTKY3 [14].
Two things follow. The thioester you isolate need not be the one that reacts, so a mediocre alkyl thioester is often the right thing to purify and store. And the pKa criterion is a compromise between two opposed requirements in the same molecule — the additive must be nucleophilic enough to displace the alkyl thiol and weakly basic enough to leave again — which is why picking the most acidic aryl thiol available does not help.
Epimerisation at the C-terminal residue
The stereocentre next to an activated acyl group is the one at risk, but the published measurements say the risk is conditional rather than constant, and the conditions matter more than the residue. In a study of silver-ion-activated thioester segment condensation, no substantial epimerisation of the C-terminal amino acid was detected during preparation of the peptide thioesters by the Boc solid-phase method; epimerisation was observed during a thioester–thiol exchange reaction and during segment condensation in DMSO in the presence of a base, whereas thioester–thiol exchange reactions in aqueous solution gave no epimerisation [15]. Using a less polar solvent suppressed it substantially: in a condensation between a 45-residue thioester segment and a 44-residue amino component, the epimer content of the 89-residue coupling product fell from 27% to 6% [15].
Hold the referent of those numbers. They were measured on that specific 45 + 44 condensation, carried out with silver-ion activation in organic solvent with base present — not on an aqueous ligation, and the same study found aqueous thiol exchange epimerisation-free [15]. The general position in aqueous ligation is stated with its own hedge: most ligations proceed without epimerisation, while β-branched (for example Ile, Thr, Val) and Pro-derived C-terminal thioesters react slowly and can be susceptible to significant epimerisation and hydrolysis [16]. So the practical rule is not that thioesters epimerise, but that activation with base in organic solvent does, and a slow junction is where the cited work reports that susceptibility [16].
Hydrolysis is the competing loss
Water is the other nucleophile in the buffer, and it is present at 55 M. Every gain in acyl-donor reactivity is also a gain in hydrolysis rate, which is why moderately reactive species are the norm. The benzotriazole route makes the trade-off measurable: LYRAG–Bt was studied for stability at pH 7 and showed a half-life of roughly 16 min with about 28% peptide remaining after 0.5 h, much lower stability than the corresponding LYRAG–Nbz [1]. That is the cost of the sub-5-minute transthioesterification quoted earlier for the same species [1]. Ligations in that work were run at pH 6.7–7.0 in 6 M guanidinium chloride with 0.2 M sodium phosphate at 22 °C [1], and a species whose half-life at pH 7 is of that order has to be generated and consumed in one operation, not isolated.
Greater donor reactivity therefore only helps when the cysteine segment is already present and concentrated. If a ligation is slow for a separate reason — a β-branched C-terminal residue, a poorly soluble segment, a low concentration forced by material limits — pushing to a more activated donor moves loss from the ligation column into the hydrolysis column without improving yield.
Choosing between the routes
The decision is mostly made by what the segment carries. Acid-stable segments and available HF capability point to Boc assembly on a thioester resin, where the in-situ neutralisation protocols also address the aggregation-driven coupling failures that limit long segments [3]. Acid-sensitive modifications point to an Fmoc-compatible precursor, and there the sub-choice is between an activated-linker route (Dbz/Nbz, MeDbz/MeNbz) and an acyl-shift device (SEA, CPE, NAC, Dmmb). Glycine-rich sequences are a specific argument for the methylated second-generation linker over the first [7]. A prolyl junction is a specific argument for SEA at mildly acidic pH [10]. The hydrazide route, in which peptide hydrazides are made by Fmoc chemistry and converted by in-situ sodium nitrite activation and thiolysis, occupies the same slot and avoids HF entirely [17].
Whichever route you take, characterise the thioester before you use it. Confirm the mass of the thioester rather than the free acid, since hydrolysis gives a species differing by the mass of the thiol. Establish the epimer content of the C-terminal residue under your own activation conditions rather than importing a figure measured on a different residue in a different solvent. And decide deliberately whether the thioester is a compound you isolate or an intermediate you generate in the ligation vessel, because for the most activated species that choice has already been made for you.