Native chemical ligation asks an N-terminal cysteine to attack a peptide thioester in water at neutral pH, and the reversible transthioesterification that opens the reaction runs on thiolate, not thiol. Selenocysteine changes that arithmetic at the source: the selenol side chain has a pKa of 5.2, against 8.5 for a typical thiol, so at neutral pH the Sec residue is essentially fully deprotonated and the nucleophile is a selenolate rather than a mostly-protonated thiol [1]. Ligation of an N-terminal Sec peptide with a peptide thioester proceeds through the same mechanism as the parent cysteine reaction, with selenium simply substituted for sulfur at the entry step [2].

How much faster, and under what conditions

The gain is real but strongly pH-dependent, and the numbers are worth reading rather than the headline. Rate constants tabulated for the reaction of free cysteine and free selenocysteine with a model chromogenic thioester — the free amino acids generated by reducing cystine or selenocystine with an equimolar amount of TCEP — give kobs = 2.5 × 10−6 s−1 for the thiol and 2480 × 10−6 s−1 for the selenol at pH 5.0, a 992-fold difference. At pH 8.0 the same pair give 665 × 10−6 s−1 and 9500 × 10−6 s−1, a 14.3-fold difference [1]. The advantage collapses by nearly two orders of magnitude across three pH units because raising the pH is simply another way of buying thiolate; selenium's contribution is largest exactly where the thiol route is worst.

Two limits on what that measurement licenses. It was run on free amino acids against a small chromogenic electrophile, so it isolates the intrinsic nucleophile effect and carries no information about the steric penalty of a real peptide junction. And it describes the first, reversible step only — the S→N or Se→N acyl shift that commits the product is a separate matter.

Selenoesters: changing the electrophile too

The step that made selenium a practical assembly method rather than a curiosity was pairing the better nucleophile with a better acyl donor. Peptide selenoesters react with peptide dimers bearing an N-terminal selenocystine in aqueous buffer to give native amide bonds without any additive; the ligations are complete in minutes even at sterically hindered junctions, and can be run in concert with one-pot deselenisation [3]. Several mechanistic pathways were proposed for the transformation and probed by a combination of experiment and computation in that first report [3], which stopped short of declaring any one of them definitive.

Note what the substrate is. The incoming fragment is supplied as the oxidised diselenide dimer, and the reaction consumes it directly — peptide selenoester and peptide diselenide are simply mixed in ligation buffer with no reducing agent required [4]. That is the operational inversion relative to ligation at Sec with a thioester, where the selenol is the nucleophile and has to be kept reduced (see below). The method is generally called diselenide–selenoester ligation, DSL.

How much of the acceleration survives into a real ligation depends on the format. Under peptide nucleic acid templation, where both fragments are held on a complementary strand, the selenocystine–selenoester reaction was measured as 10 times faster than templated native chemical ligation at cysteine and operated at nanomolar reactant concentrations [5]. That is a templated comparison at high dilution, not a bulk one; treat it as evidence that the selenium pair keeps its advantage when concentration is the limiting variable rather than as a general rate ratio.

Junctions a thioester cannot reach

The clearest demonstration of the electrophile effect is the proline–proline junction. Acylated proline at a peptide C-terminus is deactivated both sterically and electronically — an n→π* interaction lowers its reactivity — and the Pro–Pro motif had not been constructed by ligation at all before prolyl selenoesters were combined with a trans-γ-selenoproline nucleophile in a DSL–deselenisation manifold, which delivered it and was applied to the one-pot assembly of two proline-rich polypeptides [6]. If a thioester junction has stalled on steric grounds, this is the specific reason to look at selenium rather than at longer reaction times.

Dilution is the other axis, and here the additive-free character is traded away deliberately. A reductive variant, rDSL, ligates peptide fragments down to low nanomolar concentrations without solubility tags or hybridising templates, and was applied to lipidated polypeptides and palmitoylated variants of the transmembrane lipoprotein phospholemman (FXYD1) — a fragment class that is difficult to hold in solution at conventional ligation concentrations [7].

Deselenisation, and why the chemoselectivity is the point

Sec at the junction is a means, not usually an end. Desulfurisation converts a ligation-site cysteine to alanine but does so globally, which forces protection of every other cysteine in the target. The result that opened up selenium was that selenocysteine can be deselenised chemoselectively, without concomitant desulfurisation of cysteine residues in the same molecule [2]. In the original demonstration, treatment of a selenylsulfide peptide with a 50-fold excess of TCEP at pH 5.5 gave the deselenised alanyl peptide, and an optimised two-stage protocol — reduction of the selenylsulfide with 9.4 equivalents of DTT for 1 h, then addition of only 2.2 equivalents of TCEP — gave clean deselenised product in 2 h; a 38-mer Grx3(1–38) construct carrying two Sec residues and one Cys was converted to the bis-alanine product with the cysteine untouched, and deselenisation of a stereochemical probe was complete within 2 h giving only the L-enantiomer [8]. Those conditions and yields come from the full text, which is openly accessible.

The selectivity has a thermodynamic basis: the Se–H bond is much weaker than S–H, with the bond dissociation energy difference experimentally estimated at about 13 kcal mol−1 [9]. The accepted picture is radical: the weaker C–Se bond favours homolytic cleavage, phosphine reacts with a selenium-centred radical to give a phosphine selenide and an alanyl radical, and that radical abstracts a hydrogen atom from a donor to regenerate alanine [4]. Practically, deselenisation needs no radical initiator, whereas the desulfurisation it displaces is driven by a radical initiator (VA-044) in combination with TCEP [8]; that difference in required forcing conditions is what the chemoselectivity rests on.

Oxygen decides the product. Under anaerobic conditions the reaction is finished in one minute and gives alanine, with the alanine product accounting for more than 95% when run in degassed buffer in an anaerobic chamber; under ambient aerobic conditions a serine product appears in significant amounts (20%), and under oxygen-saturated conditions selenocysteine is converted to serine [9]. This is not only a side reaction to suppress. Sec can be converted to Ser in a single high-yielding step, and coupling that oxidative deselenisation to ligation at Sec gives a programmed route to a serine junction [10]. So one masked residue serves two native outcomes, and which one you get is set by how rigorously you exclude oxygen. Both electrophiles have been run into the same one-pot oxidative deselenisation: native chemical ligation at Sec with peptide thioesters and additive-free selenocystine ligation with peptides bearing phenyl selenoesters, applied to the synthesis of a MUC5AC-derived glycopeptide [19].

The junction repertoire extends further. DSL–deselenisation has been run at β-selenoaspartate and γ-selenoglutamate, giving native Asp and Glu at the ligation site [11], and other selenol-derived amino acids have been developed on the same logic [2].

Handling selenols

The selenol has strong reducing power and is oxidised rapidly in air to the diselenide (selenocystine), which is why native chemical ligation at Sec has required excess reducing agents — TCEP, MPAA, thiophenol — to maintain a productive selenol concentration [4]. Assume oxidation as the default state of the material: single-Sec test peptides were always isolated as their corresponding diselenide dimers, and dual-Sec sequences exclusively as intramolecular diselenides [12]. Purified Sec peptides are therefore handled and weighed as the oxidised form, and the mass you observe on isolation is the dimer's, not the free selenol's.

TCEP appears on both sides of the ledger — as the reductant that keeps a selenol available for thioester ligation [4] and as the reagent that strips selenium in deselenisation [8]. The two operations are therefore staged rather than run together: the ligation is taken to completion first, then deselenisation reagents are added to the same vessel [11]. DSL avoids the conflict at the ligation step entirely by using the diselenide as the substrate [3, 4].

Getting Sec into an Fmoc chain

Sec is installed by ordinary Fmoc SPPS with a side-chain-protected building block. An early derivative for this purpose, Fmoc-Se-p-methoxybenzylselenocysteine, was prepared to allow SPPS of a peptide with an unprotected N-terminal Sec, which was then ligated to a peptide thioester to give a 17-mer corresponding to the C-terminus of ribonucleotide reductase [13]. Choice of protecting group is the main practical decision:

  • Mob/Meb (benzyl-type). Removed with 2,2'-dithiobis(5-nitropyridine); Sec(Mob) and Sec(Meb) are extremely labile under DTNP conditions, and in bis-Sec(Mob) peptides the same treatment also cyclises them to the diselenide, while bis-Sec(Meb) models required additional manipulation [14].
  • Xan and Trt. TFA-labile side-chain protection, so the group comes off in the standard cleavage cocktail; Fmoc-Sec(Xan)-OH was found to be bench-stable for extended timeframes, while Fmoc-Sec(Trt)-OH appeared to detritylate slowly when not stored at -20 °C [12].
  • DEAMC (7-diethylamino-3-methyl coumarin). A photolabile group removed with visible light at 450 nm, reagent-free, without deleterious deselenisation of the Sec residue — which is what allows the next ligation to be run without an intervening purification [15].
  • Selenazolidine (Sez). Masks an N-terminal Sec entirely so that a middle segment can be acylated at its own C-terminus first; unmasking then exposes the Sec for the next ligation [16].

The last two exist for the same reason: iterative one-pot assembly. With DEAMC, one-pot syntheses of 60- and 80-residue fragments of mucin-1 and of apolipoprotein CIII were completed in 2–4 hours [15]. With Sez, the first sequential native chemical ligation/deselenisation was executed on three unprotected peptide segments to give human phosphohistidine phosphatase 1 (PHPT1), a 125-residue protein whose three cysteines are positioned unfavourably for ligation [16] — the case that motivates selenium chemistry most directly, since the native cysteines can be left free throughout.

Making the selenoester

Aryl selenoesters are the more reactive acyl donor, and that rate advantage is what the method is built on [4]. Solid-phase synthesis via a side-chain anchoring strategy gives peptide selenoesters in high yield and without epimerisation[17].

There is a direct bridge from hydrazide chemistry. In expressed protein selenoester ligation, protein acyl hydrazides derived from recombinant intein fusion precursors are converted to protein aryl selenoesters, ligated with synthetic modified selenopeptides, and then chemoselectively deselenised; the ligations run at high dilution and the sequence was used to make ubiquitinated polypeptides, lipidated GTPases and a phosphorylated chaperone protein [18]. If you already generate hydrazides for thioester access, the selenoester is one activation step away and buys the rate profile above.

What the method delivers, and what it costs

The published protocol covers Fmoc-SPPS of the peptide diselenide and selenoester fragments, their fusion by DSL, and chemoselective deselenisation of the ligation products, demonstrated on the post-translationally modified collagenous domain of adiponectin via DSL–deselenisation at selenocystine and on two tick-derived thrombin-inhibiting proteins via DSL–deselenisation at β-selenoaspartate and γ-selenoglutamate. Each protein was synthesised within 15 working days from resin loading, and each synthesis was performed independently by an academic and an industrial laboratory with comparable yields [11]. That reproducibility check is unusual and is the most useful single datum for deciding whether to adopt the chemistry.

Scale is not the constraint it once was, and DSL does not have to carry a whole target on its own. A 31.5 kDa phosphorylated insulin-like growth factor binding protein (IGFBP-2) of 290 amino acid residues, bearing a phosphoserine modification and nine disulfide bonds, was assembled using two ligation manifolds in the same synthesis — diselenide–selenoester ligation and native chemical ligation [20]. That is the realistic pattern: use the selenium junction where the thioester route is deactivated or slow, and the conventional junction everywhere else.

The costs are specific. Both halves of the junction have to be made first: the protected Sec derivatives are synthetic targets in their own right [12, 14], and the selenoester needs its own solid-phase route [17]. Deselenisation then destroys the selenium you installed, so if the target is a genuine selenoprotein that step must be suppressed and the Sec protected or masked instead [16]. And if oxygen is not excluded, a fraction of the product will be the serine variant rather than the alanine one [9] — a substitution that mass spectrometry will resolve but that a quick HPLC check may not.

The decision rule that follows from all of this is narrow. For an unhindered Xaa–Cys or Xaa–Ala junction with a cooperative thioester, native chemical ligation with desulfurisation remains the simpler route. Selenium earns its cost at deactivated or sterically congested junctions, at low fragment concentrations, and above all where the target carries native cysteines you do not want to protect — because deselenisation leaves those thiols alone [2].