As a target gets longer, stepwise solid-phase synthesis gives lower yields and a more complex impurity profile, because the small losses at each cycle compound. Where that becomes limiting depends on the sequence, the scale and the purity required rather than on length alone, but beyond roughly fifty residues it is often more efficient to make shorter fragments and join them. The answer is to make two clean fragments and join them — but joining two unprotected peptides in water, at one specific place, with everything else untouched, is a genuinely hard problem. Three methods solve it in different ways, and each leaves something behind at the junction. Choosing between them starts with deciding where in the sequence a scar is acceptable.
The 1994 method that made chemical protein synthesis practical. One fragment carries a C-terminal thioester, the other an N-terminal cysteine. The cysteine thiol attacks the thioester to give a transient thioester-linked intermediate, which then rearranges spontaneously to a native amide bond at the junction1.
Two features make it work on unprotected material. The first step — thiol–thioester exchange — is reversible, so unproductive reactions with internal cysteines simply undo themselves; only an N-terminal cysteine can proceed to the irreversible rearrangement2. And because the chemistry is specific to that pair of functional groups, native chemical ligation can often join fragments that are otherwise unprotected — though some sequences, and some subsequent steps, still call for temporary protection. Under suitable conditions the reaction can consume both starting materials almost completely and give a single main product peak3.
The constraint is the cysteine. A ligation site has to be a cysteine, or become one. Where that is unacceptable, the standard escape is desulfurisation after ligation: the junction cysteine is converted to alanine, which widens the choice of site from the roughly 2% of positions that are cysteine to the much commoner alanine. The same logic has been extended to thiol-derived versions of other amino acids4. Selenocysteine can also be used in place of cysteine2.
Where one fragment is large or awkward to make synthetically, it can be produced recombinantly and furnished with a thioester by intein chemistry — expressed protein ligation, which lets a synthetic segment carrying a modification be joined to a suitable recombinant protein fragment2.
Sortase A is the most widely used enzymatic route. Sortase A recognises a C-terminal LPXTG motif, cleaves between the threonine and the glycine, and joins the threonine carbonyl to the N-terminal amine of an oligoglycine on the other fragment5. Its appeal is that it works on folded proteins in buffer, under mild conditions, with no thioester to prepare.
Its two limitations are worth knowing before committing to it. The transpeptidation is reversible — ligation to an oligoglycine acceptor regenerates an LPXTG motif in the product, so the enzyme can cleave what it has just made. In one well-documented system with equimolar substrates the reaction reaches equilibrium near 50% conversion5; that figure is not a universal ceiling, since the position of the equilibrium depends on the substrates, the leaving group, the concentrations and the conditions. Conversion can be improved with an excess of the acceptor nucleophile, by removing the released fragment, or by using a depsipeptide substrate whose ester leaving group makes the step effectively irreversible7. And the catalytic efficiency of the wild-type enzyme is low. Measured against an LPETG substrate at 22.5 °C, S. aureus sortase A gives kcat of 1.5 s−1 and a KM of 7.6 mM, for a kcat/KM of about 200 M−1s−16 — the high KM being the main problem. In practice that means high enzyme loadings, often 0.1 to 1.0 molar equivalents, and long reactions. Evolved variants are substantially faster: directed evolution of sortase A has produced mutants with up to a 140-fold improvement in LPETG-coupling activity6.
Both are addressable. A large excess of the glycine-bearing nucleophile pushes the equilibrium; engineered sortase variants improve the kinetics; and a depsipeptide substrate — where the leaving group is attached through an ester rather than an amide — makes the reaction effectively irreversible, allowing catalytic enzyme loading with an excess of substrate7. The junction is not freely designable. Sortase cleaves between the threonine and the glycine, so the product carries the donor sequence through LPXT joined to the incoming acceptor; which residues remain depends on the acceptor design, and the product can regenerate a recognition sequence and so stay susceptible to further cleavage.
Butelase 1, from Clitoria ternatea, and the related plant asparaginyl endopeptidases recognise a much shorter C-terminal motif. For butelase 1 the requirement is Asx-His-Val, where Asx is asparagine or aspartate: the enzyme cleaves between the Asx and the histidine, releasing the His-Val dipeptide, and joins the remaining Asx to the N-terminus of the incoming fragment8. Butelase 1 is the fastest peptide ligase described, with catalytic efficiencies reported up to 542,000 M−1s−18, and in the published protocol a butelase-mediated reaction requires 100- to 1000-fold less enzyme than one with sortase A, with around 0.005 molar equivalents used9. Those comparisons belong to the enzymes, substrates and conditions tested rather than being a general rule, and the asparaginyl ligase family is not uniform: recognition and acceptor preferences differ between butelase 1, the OaAEP and VyPAL enzymes, and engineered variants.
There are constraints in return, though they are preferences more often than absolute rules. For butelase 1 the acceptor fragment can begin with any of the natural amino acids except proline, and the published work reports a marked preference for a bulky hydrophobic residue such as isoleucine, leucine or valine at the following position9; reported scope varies with the enzyme variant and the reaction format, so it is worth checking against the specific system. The released His-Val dipeptide also competes as a nucleophile and has to be managed10. And obtaining the enzyme is itself non-trivial, which is why engineered recombinant variants are an active area.
Desulfurisation is not selective for the junction. Converting the ligation-site cysteine to alanine is useful when alanine is the native residue at that position, but the common desulfurisation conditions act on free thiols generally, so other cysteines in the product are at risk too. If the target contains cysteines that must survive, they need temporary protection, or the design needs a selective alternative — which is a reason to settle the ligation site early in a cysteine-rich target rather than late.
In practice the decision is best taken in this order, with the choice of chemistry following from the answers rather than preceding them.
| Native chemical ligation | Thioester + N-terminal Cys; irreversible; no enzyme; leaves a Cys |
| Desulfurisation | Converts the junction Cys to Ala, widening site choice |
| Sortase A | LPXTG + oligoGly; reversible, so conversion depends on design and conditions; junction retains LPXT |
| Depsipeptide trick | Makes the sortase reaction effectively irreversible |
| Asparaginyl ligase | Asx-His-Val; far less enzyme; constrained second residue |
LifeTein synthesises the fragments these routes need: peptide thioesters and hydrazide precursors, fragments with a free N-terminal cysteine, and sequences carrying LPXTG or oligoglycine extensions at a defined terminus. If you are planning a two-fragment assembly, send us the full target sequence and we will suggest where the junction can sit before anything is made.