N-Terminal Selective Peptide Modification: Chemistry and Proof

A peptide carrying a single lysine hands any acylating reagent two primary aliphatic amines and no reason to prefer one. Both are nucleophilic nitrogens on a saturated chain, both give the same amide, and both give the same mass shift. The one intrinsic difference is basicity. Every N-terminal-selective method is therefore built on one of two moves: exploit that basicity difference under tight pH control, or abandon it and find a transformation the α-amine can undergo that a lysine ε-amine cannot.

The pK_a gap is real, and narrower than it is usually described

State the numbers before designing around them. In alanine pentapeptides — model compounds with no folded structure to perturb anything — the measured pK is 8.0 for the N-terminal α-amino group and 10.4 for the lysine side-chain amine ^[1]. A compilation of 541 measured pK values across 78 folded proteins gives averages of 7.7 ± 0.5 for the N-terminus (16 measurements) and 10.5 ± 1.1 for lysine (35 measurements), and the authors note that neither protein average differs significantly from its model-peptide counterpart ^[1]. So the working separation is about 2.4 pK units, and folding does not erase it on average.

Averages are the wrong statistic for a selectivity argument. The same compilation reports a measured range of 6.8–9.1 for N-termini and 5.7–12.1 for lysine side chains in folded proteins ^[1]. A lysine at the low end of that range is the better nucleophile at neutral pH, and an average tells you nothing about which lysine you have. Short synthetic peptides sit much closer to the model-compound values than folded proteins do, since there is little tertiary structure to perturb them — but a lysine flanked by acidic residues is under no obligation to sit at 10.4.

For one residue the argument fails outright. An N-terminal proline is a secondary amine whose basicity is similar to that of a lysine side chain, so pH control cannot separate them ^[2]. Any protocol that says run it at pH 6.5 and you get the N-terminus is silently assuming position 1 is not proline.

Activated esters: what pH control actually gets you

The classical answer to the two-amine problem is to acylate at a pH where most ε-amines are still protonated. Sélo and co-workers defined the conditions for preferential incorporation of biotin at the N-terminal α-amino group of peptides as pH 6.5, a 24 h reaction, and a 1:5 peptide-to-biotin ratio ^[3]. Read the cost inside that sentence: a full day and a fivefold reagent excess. Suppressing the ε-amine suppresses the α-amine too, and the compensation is time and excess — the two variables that most reliably produce a second acylation.

Chan and co-workers measured how far this gets you on a defined peptide set. Modifying the unprotected pentapeptides XSKFR, with X varied over all 20 natural amino acids, using an alkyne-functionalised N-hydroxysuccinimide ester gave internal lysine-modified peptides for 5 of the 20 sequences and only moderate-to-low N-terminal selectivity, between 5:1 and 1:4, for 13 of the 20 ^[4]. A 1:4 ratio is not selectivity; it is a preference for the lysine. On whole proteins the same NHS ester produced significant or major amounts of di-, tri- or tetra-modified species ^[4]. If you are labelling a lysine-containing peptide with an activated ester and calling the product N-terminal, that library is the result you have to argue against.

Ketenes: a less basicity-dependent electrophile

The same study's positive result is the more useful half. An isolated, alkyne-functionalised ketene reacting with the XSKFR library at room temperature and pH 6.3 gave a ratio of modified α-amino to modified ε-amino greater than 99:1 for 13 of the 20 sequences, and moderate-to-high selectivity between 4:1 and 48:1 for six of the seven remaining sequences ^[4]. Do the arithmetic: 13 excellent, 6 acceptable, and one sequence the report places in neither band. Nineteen of twenty is a strong reagent, not a guarantee, and the residue at position 1 decides which band you land in. Insulin, lysozyme, RNase A and BCArg were each selectively N-terminally modified with that ketene at room temperature, in contrast to the multiply modified products the NHS ester gave on the same targets ^[4]. The alkyne the ketene installs was then carried straight into a copper-catalysed cycloaddition with a dansyl azide, with no intervening step ^[4] — the standard division of labour, in which the N-terminal step fixes the position and the cycloaddition fixes the payload, set out in the companion guide on CuAAC and strain-promoted click chemistry.

Transamination: change the functional group, then stop worrying about amines

Acylation asks a reagent to discriminate between two amines. Transamination sidesteps the question by converting the α-amine into a group no other residue carries. A pyridoxal 5'-phosphate-mediated reaction site-specifically oxidises the N-terminal amine of a protein to a ketone, which is then used to attach a reagent of choice through oxime formation ^[5]. Selectivity now lives entirely in the first step; the second step is chemoselective for a carbonyl the peptide did not previously contain, so payload choice at that stage is unconstrained.

The first step is emphatically not sequence-agnostic, and that caveat is the one usually dropped. The N-terminal sequence significantly influences the overall success of the strategy ^[5]. A combinatorial peptide library screen across N-terminal amino acid combinations identified a highly reactive alanine-lysine motif, which was confirmed to promote modification of peptide substrates with PLP; encoding that tripeptide on two protein substrates produced substantial increases in reactivity relative to their wild-type termini ^[5]. That is a genuine advance and a genuine restriction at once: for reliable conversion you are expected to install a particular terminal sequence, which puts transamination nearer a tag-based method than a universal one.

2-Pyridinecarboxaldehyde

2PCA compresses those two steps into one. MacDonald and co-workers reported selective modification of protein N termini with 2-pyridinecarboxaldehydes that requires no genetic engineering of the target, demonstrated on 12 different proteins including the soluble domain of the human estrogen receptor; the function of that protein was confirmed after modification through binding of a fluorescent estrogen mimic ^[6]. Confirming retained function is the control that separates a site-selective reaction from one that has quietly acylated a binding surface, and it is worth designing into the experiment rather than adding later.

2PCA has since become an N-terminomics reagent as much as a bioconjugation reagent. A commercially available alkyne-modified 2PCA, combined with Cu(I)-catalysed azide–alkyne cycloaddition and chemically cleavable linkers, gives selective N-terminal biotinylation and efficient enrichment of protein N termini without synthesising bespoke probes ^[7]. The transferable point for a bench user is not the proteomics application: it is that a community which depends on this reagent had to characterise its sequence preference before it could trust it, and how they did that is the subject of the last section.

When the terminal residue is itself the handle

The alternative to fighting the pK_a gap is to choose a first residue whose side chain participates, which turns a two-amine selectivity problem into a one-reaction chemoselectivity problem.

Glycine. An N-terminal glycine can be labelled at a single site by generating a latent nucleophile from the N-terminal imine, which reacts with an aldehyde to deliver an aminoalcohol under physiological conditions; the method differentiates N-terminal Gly from the other proteinogenic amino acids, and installs an orthogonal aldehyde group primed for later tagging with an affinity tag, a ¹⁹F NMR probe or a fluorophore ^[8]. Reported on isolated proteins and extended to lysed cells, with the reaction conditions mild enough not to alter protein structure or function ^[8].

Cysteine. An N-terminal cysteine is a long-established handle. In native chemical ligation, chemoselective reaction of two unprotected peptide segments gives an initial thioester-linked species whose spontaneous rearrangement yields a full-length product with a native peptide bond at the ligation site ^[9]. The requirement for an N-terminal cysteine to mediate ligation is the method's principal limitation; thiol-mediated ligation followed by desulfurisation has extended the accessible junction residues to Phe, Val, Leu, Thr, Lys, Pro and Gln, using mercapto-containing unnatural amino acids as cysteine surrogates ^[10]. For labelling rather than ligation, the same 1,2-aminothiol condenses with cyanobenzothiazole probes: a single amino-acid tag — the N-terminal cysteine — labelled with a high degree of specificity under physiological conditions, both in vitro and at the surface of live cells ^[11]. Ligation itself is covered in the native chemical ligation guide.

Serine and threonine. The 2-amino alcohol of an N-terminal Ser or Thr is oxidised very rapidly by periodate at pH 7 to generate an aldehyde in the peptide, which then condenses with a hydrazide to give a hydrazone conjugate ^[12]. Two practical points come straight from the original work: a low molar ratio of periodate to peptide minimises the potential for side reactions while the desired oxidation stays rapid and highly specific, and the hydrazones formed were stable at pH 6–8 for at least 12 h at 22 °C but labile at more acidic pH ^[12]. That acid lability is a property of the linkage rather than of the targeting step, and it belongs in the same design conversation as the guide on conjugate linker chemistry and payload release.

The same two residues also support a ligation. An O-salicylaldehyde ester at the C-terminus of one segment reacts with an N-terminal serine or threonine on the other; the coupling runs through an N,O-benzylidene acetal intermediate that is readily converted into a natural Xaa-Ser/Thr peptidic linkage at the ligation site ^[13].

Proline. Because a secondary amine forms a highly electrophilic iminium ion with an aldehyde, the residue that defeats pH control can be addressed by a Petasis three-component reaction in which that iminium is intercepted by a nucleophilic organoboronate under biocompatible conditions ^[2]. Peptide screening showed the reaction to be highly selective for N-terminal proline, and it is stereoselective (>99% de), delivering doubly functionalised conjugates in a single step ^[2].

Enzymatic routes have specificity, not universality

Sortase A cleaves an LPXTG motif between threonine and glycine and joins the threonine carboxyl to an amino group of pentaglycine; applied in vitro, a protein or peptide carrying LPXTG is specifically ligated to an aminoglycine protein or peptide through an amide bond ^[14]. The N-terminal constraint therefore sits on the incoming nucleophile, which has to present an oligoglycine α-amine — one more reason to design an N-terminal glycine in. Scope, equilibrium behaviour and tolerated sequences are covered in the sortase guide.

Subtiligase acylates the α-amine directly, and its sequence specificity has been mapped at proteome scale. Weeks and Wells characterised ligation efficiency for more than 25,000 enzyme–substrate pairs using proteome-derived peptide libraries, identified a family of 72 mutant subtiligases with activity toward N-terminal sequences that had previously been recalcitrant to modification, and applied individual mutants for site-specific bioconjugation of purified proteins, including antibodies ^[15]. Note what that concedes: stringent enzyme–substrate specificity limits the utility of enzymatic peptide ligation, and the field's answer has been to enumerate that specificity and engineer around it rather than to find one universal enzyme ^[15]. The protease-catalysed ligation guide treats subtiligase in depth.

A mass shift cannot tell you where the label went

This is where most reported N-terminal conjugates are weakest. Intact mass tells you that a modification occurred and how many copies were added. On a peptide with one α-amine and one lysine, the α- and ε-acylated products are constitutional isomers with identical monoisotopic mass; no improvement in mass accuracy on an intact measurement will separate them. Even after fragmentation the answer is not automatic. Search engines identify modified peptides with a defined measure of reliability, but their localisation of the site or sites of modification is often arbitrary and unreliable, and the field continues to lack a widely accepted false-localisation-rate metric that describes site certainty in published data sets ^[16]. Site localisation is a second inference with its own error rate, and it has to be reported as one.

What that means at the bench: digest the conjugate with a protease that cuts between the terminus and the nearest lysine, run LC-MS/MS on the fragments, and require site-determining fragment ions — ions that bracket the modified residue on both sides — before claiming a position. An observed missed cleavage at a lysine is suggestive, since trypsin will not cut after an acylated lysine, but a missed cleavage is consistent with several explanations and is not localisation on its own. Where two candidate residues survive the analysis, quantify the isomer ratio from extracted ion chromatograms of the localised peptides instead of reporting a single site for what is a mixture.

The scaled-up version of the same logic is now standard for characterising the reagents themselves. Proteome-derived peptide libraries combined with LC-MS/MS are highly diverse and allow modification efficiencies for tens of thousands of N-terminal sequences to be analysed in a single experiment, and published protocols apply exactly this to both 2PCA and subtiligase ^[17]. Notice also that the selectivity ratios quoted earlier are properties of an assay design as much as of a reagent: XSKFR carries one lysine, at a fixed position, in a five-residue chain that survives direct analysis ^[4]. Your sequence is not that sequence.

Choosing

• No lysine in the peptide: the ambiguity disappears and an activated ester is unambiguous by construction — but say that is why, rather than implying a selectivity you never demonstrated.
• Lysine present and the terminus fixed: ketene or 2PCA, and expect the identity of residue 1 to matter ^{[4, 6]}.
• Terminus under your control: pick Cys, Ser/Thr, Gly or Pro and use the dedicated chemistry ^{[2, 8, 11, 12]}.
• A native amide bond needed at the junction rather than a triazole, oxime or hydrazone: native chemical ligation, Ser/Thr ligation, sortase or subtiligase ^{[9, 13, 14, 15]}.

Whichever route you take, the deliverable that supports the claim is not the intact mass spectrum of the conjugate. It is the peptide map underneath it.