Side-Chain Anchoring: Keeping the C-Terminus Free for Modification

In a conventional Fmoc synthesis the α-carboxyl of the first residue is the attachment point, so it is committed from the first coupling to the last, and what the cleavage cocktail hands back is whatever the linker leaves behind: a free acid, or a primary amide. Anything else at that position — a secondary amide, an ester, an aldehyde, a thioester — has to be assembled while the chain is still on the support, which is impossible if the α-carboxyl is the bond holding the chain there. The protocol literature on backbone amide linkers puts the constraint plainly: most solid-phase routes to C-terminal modified peptides have limited generality because they rely on the C^α-carboxyl for attachment to the solid support, and anchoring instead through a backbone nitrogen is what allows significant flexibility for chemical modification of the C-termini ^[1]. Side-chain anchoring reaches the same freedom through a different bond. The library's guide to resin and linker selection describes the ordinary α-carboxyl case; what follows is what changes when that carboxyl is no longer the anchor.

Loading the residue through its side chain

The manoeuvre is to attach the eventual C-terminal residue by a functional group on its side chain, carrying its α-carboxyl through the whole synthesis under a protecting group that can be removed later without disturbing the anchor. The oldest worked example is also the clearest. Attempts to anchor Fmoc-asparagine or Fmoc-glutamine as p-alkoxybenzyl esters were described as fraught with difficulties; instead, N^α-Fmoc-C^α-tert-butyl aspartate or glutamate was attached quantitatively through its free ω-carboxyl to a tris(alkoxy)benzylamino (PAL) support, chain elongation proceeded normally by standard Fmoc chemistry, and treatment with trifluoroacetic acid in dichloromethane for 90 min at 25 °C released the peptides in greater than 95% yield without side reactions at the C-terminus ^[2].

Read what that anchor did to the residue. The ω-carboxyl was joined to an amine on the handle, so acidolysis of that bond returns a primary carboxamide, and the products are peptides with C-terminal asparagine or glutamine rather than aspartate or glutamate. The handle chemistry writes the identity of the anchoring residue into the product, and an ester-type handle on the same ω-carboxyl would return the free acid instead. That is the first thing to settle when planning a side-chain-anchored synthesis, and it is settled before any coupling.

Other side chains work, with different motivations. For a C-terminal cysteine the case against α-carboxyl anchoring is that esterification onto the support is known to occur sometimes in low yields, has attendant risks of racemisation, and may also result in conversion to a C-terminal 3-(1-piperidinyl)alanine residue as the chain grows by Fmoc chemistry; converting N^α-Fmoc, C^α-tert-butyl cysteine to an S-xanthenyl preformed handle derivative and attaching that to amino-containing supports gave anchoring without racemisation, and the absence of the piperidinylalanine was inferred from retention of the chains on the support throughout the cycles of Fmoc chemistry ^[3]. Note the word inferred: that is a mass-balance argument on the bead, not a direct measurement of a by-product.

Amine side chains are anchored differently again. Ornithine and lysine whose side chain is protected with an N-Boc group can be converted efficiently to the isocyanate and immobilised on Wang-type resin in almost quantitative yield, giving a carbamate anchor; four cyclic peptides were built on the ornithine anchor, and the authors place the method at least on a par with previously reported methodologies for yield and purity of the final products while arguing it is operationally more straightforward ^[4]. A reductive-amination route has been run at preparative scale: caged aminoluciferin peptides were assembled by side-chain anchoring of the P₁ residue, the method was validated and optimised for up to more than 100 g of resin, and two libraries containing 719 novel peptide substrates in total were made with P₁ as lysine or arginine ^[5]. The openly accessible full text of that paper gives the anchoring step as reductive amination with sodium cyanoborohydride onto a (3-formylindolyl)acetamidomethyl resin, and cleavage as trifluoroacetic acid in dichloromethane at 1:1, with 2.5% triisopropylsilane and ethanedithiol added for trityl-containing sequences.

It is not a purity strategy

A review that surveys these strategies and supplies its own experimental comparison reports that C-terminal and side-chain syntheses give similar results in terms of purity, that the stability of the two bonds which anchor the peptide to the polymer may determine the overall yield, and that this should be considered for large-scale production; the same authors add that resins and linkers not subject to side reactions can be preferred for some peptides ^[6]. That is the correct calibration. A side-chain anchor buys a free α-carboxyl and the chemistry that follows from it. On the evidence in that comparison it does not buy a cleaner crude.

Three protection systems that must not overlap

The arrangement only works if the anchor, the α-carboxyl protecting group and the Fmoc cycle are mutually orthogonal. The Fmoc cycle is fixed — piperidine in dimethylformamide, once per residue, tens of times over — so the other two have to be chosen around it, and the α-protection additionally has to come off on the resin under conditions the anchor ignores.

Allyl esters are the usual answer, because palladium(0) removes them under conditions that are neither strongly acidic nor basic. The practical difficulty is delivery rather than chemistry. Standard allyl removal uses a suspended palladium catalyst, an approach not feasible on batch and continuous-flow peptide synthesisers because delivery of the insoluble catalyst is problematic; solvent conditions were examined and optimised to solubilise the catalyst, prevent undesired Fmoc deblocking, and be compatible with sensitive amino acids, specifically tryptophan and methionine, and with glyco- and sulfopeptides ^[7]. The Fmoc-deblocking clause is the orthogonality failure being engineered out: if the removal cocktail takes the N-terminal Fmoc off as well, the resin carries two free nucleophiles at once and the step is no longer selective.

tert-Butyl esters are the other option, but they are normally removed by trifluoroacetic acid, which is also what most acid-labile anchors respond to; that is why the PAL example above carries its α-tert-butyl group through to final cleavage rather than unmasking it on the resin. A Lewis-acid route narrows the overlap. Iron(III) chloride has been used for selective on-resin deprotection of the side-chain tert-butyl-protected aspartic and glutamic acid of a peptide, in a method its authors describe as mild, cost-effective and Fmoc-chemistry compatible, and which allowed on-resin incorporation of amides, esters and thioesters in good yield ^[8]. Be clear about what was in that reactor: the peptides were held by the usual C-terminal attachment and the ester being unmasked was a side-chain ester, so the work establishes the deprotect-then-acylate sequence rather than a side-chain-anchored synthesis. The deprotection chemistry is the transferable part.

Thioesters have to be installed last

The reason a thioester cannot simply be carried through an Fmoc synthesis is that C-terminal peptide thioesters cannot survive the repetitive piperidine deprotection steps of Fmoc chemistry ^[9]. Side-chain and backbone anchoring strategies were used for exactly that reason to prepare the required partially protected and activated peptide intermediates spanning the length of bovine pancreatic trypsin inhibitor, with thioesters established at the C-termini by selective deprotection and coupling steps carried out while the peptides remained bound to the supports, and pilot native chemical ligations pursued directly on-resin as well as in solution after cleavage and purification ^[9].

The generalised version immobilises a variety of Fmoc-amino acids by their side chains with their C-termini protected, builds the chain by the Fmoc protocol, unmasks the C-terminal carboxylate, then couples either thiols or amino acid thioesters to give peptide and glycopeptide thioesters in high yields; the authors report that a significant proportion of the proteinogenic amino acids could be incorporated as C-terminal residues, and demonstrated the route on a 28-amino-acid glycopeptide thioester that was elaborated to an N-terminal fragment of the glycoprotein erythropoietin by native chemical ligation ^[10]. Hold on to the hedge. A significant proportion is not all, and the C-terminal residue of a side-chain-anchored peptide has to be one whose side chain will sit on a handle, which is a real restriction on what you can target.

Aldehydes are generated on the bead

A C-terminal aldehyde can be neither carried through a synthesis nor used as an anchor, so it has to appear late and in place. In one efficient solid-phase route to C-terminal tyrosine peptide aldehydes the C-terminal residue was anchored as Dde-tyrosinol through its side chain, the chain was assembled by standard Fmoc protocols, and the aldehyde was produced on the support by sulfur trioxide–pyridine oxidation of the corresponding resin-bound protected peptide alcohol; the efficiency and reliability of the oxidation step was dramatically improved by the incorporation of a small poly(ethylene glycol) spacer between the linker and the solid support, and the tetrapeptides were obtained after acidolytic cleavage and reversed-phase HPLC in high yield and purity ^[11]. The spacer detail is worth keeping. An on-resin oxidation is a heterogeneous reaction on a group that has to be reachable inside a swollen gel, and the fix those authors reported was to move the anchored residue further from the polymer.

Aspartimide, with the resin as the leaving group

Anchor through aspartate and the β-carboxyl is an ester or an amide to the support for the entire synthesis, sitting through every piperidine treatment. That is the arrangement succinimide chemistry needs. Two results mark the risk. Where the aspartic acid side chain was protected as the allyl ester during the synthesis of a macrocyclic peptide lactam, the imide analogue was a major side-reaction product; extensive two-dimensional NMR analysis was used to characterise the structural identity of both the lactam and the imide, and a straightforward protocol modification was developed to minimise aspartimide formation ^[12]. Where the side chain carried the Dmab group instead, a 15-residue galanin fragment analogue containing the Asp-Ala motif showed a strong tendency to form aspartimide with unusually high efficiency, which the authors then probed with different resin linkers, a variety of Fmoc-deprotection conditions and several coupling methods on a five-residue fragment ^[13].

Both were measured on a side-chain-protected aspartate, not on a side-chain-anchored one, so neither supplies a rate for the anchored case. What they license is the warning: leaving the β-carboxyl esterified or amidated through dozens of base treatments is a documented route to the imide, and a resin anchor is such a bond. On the standard mechanism the ring closure displaces whatever occupies the β-carboxyl, so the expected consequence in a side-chain-anchored synthesis is loss of the chain from the bead — a failure that reports as yield rather than as a new mass in the crude chromatogram, and is therefore harder to attribute. The library's article on aspartimide formation covers the sequence dependence and the protecting-group countermeasures.

Pyroglutamate, with the resin as the leaving group again

Anchor through glutamate and there is a window, exactly one deprotection long, in which the anchored residue has a free α-amine and a γ-carbonyl bonded to the support, positioned to close a five-membered ring. The measurements that exist were made on the analogous N-terminal glutamine problem. The extent of cyclisation from H-Gln-X-OCH₂-resin was determined by a quantitative reaction and found to be quite pronounced with X being asparagine, leucine and glycine, which establishes that the rate depends on the neighbouring residue ^[14]. Separately, formation of pyrrolidone carboxylic acid residues from amino-terminal glutaminyl residues was shown to be catalysed by weak acids but not by strong acids; during carbodiimide-mediated coupling reactions the N^α-protected amino acid reagent accelerated the cyclisation and resulted in a significant amount of chain termination, and the side reaction could be minimised by accelerating the coupling reaction while reducing the time of exposure to weak acids, most effectively by coupling in dimethylformamide with the preformed symmetric anhydride ^[15].

Both used peptides anchored through the α-carboxyl with the side-chain carboxamide free; neither tested a glutamate whose γ-carbonyl is the anchor, so neither tells you how good a leaving group the support is in that step. What transfers is operational. The exposed window is the interval between deprotecting the anchored residue and acylating it, and both papers indicate that window is closed by coupling fast and keeping weakly acidic additives out of it.

The anchor sets the cleavage conditions

Because the bond to the support is now a side-chain bond, the acid that breaks it also decides which side-chain protecting groups leave at the same moment, and that gives the strategy a second use. From the xanthenyl-anchored cysteine work: fully deprotected peptides, including free sulfhydryl peptides, were released in excellent yield using cocktails containing a high concentration of trifluoroacetic acid, 80–90%, plus appropriate thiols or silanes as scavengers, while high-yield release of partially protected peptides was achieved with cocktails containing only 1–5% trifluoroacetic acid; in peptides with two cysteines the intramolecular disulfide was obtained either by oxidising the released dithiol in solution, by adding dimethyl sulfoxide to the cleavage cocktail as a mild oxidant, or by concomitant cleavage and co-oxidation using iodine or thallium tris(trifluoroacetate) in acetic acid ^[3]. The low-acid option is what makes a side-chain anchor useful for fragment work, and it is the capability the thioester syntheses depend on when the acyl donor has to come off the bead partially protected ^[9].

Head-to-tail cyclisation on the resin

Once the α-carboxyl is unmasked and the N-terminal Fmoc is removed, both reacting groups sit on the same resin-bound molecule and the ring can be closed before cleavage. The automated palladium protocol described above was developed for precisely that use, on a continuous-flow synthesiser, for head-to-tail cyclic peptides, branched peptides and multiple antigenic peptides ^[7]. The isocyanate-derived ornithine and lysine anchors were likewise designed for on-resin cyclisation within standard Fmoc synthesis ^[4]. The aspartimide caveat is not hypothetical in this context, because the allyl ester that makes on-resin unmasking possible is the same aspartate side-chain protection involved in the imide result described above ^[12].

Choosing between a side chain and the backbone

Both routes free the C-terminus; they differ in what they constrain. A backbone amide linker anchors the growing chain through a backbone nitrogen and so places no requirement on the identity of the C-terminal residue, which is the basis of the claim that in effect any peptide containing C-terminal variations can be prepared in overall good purity and yield, with minimal side reactions, using one or more of three variations of that strategy ^[1]. Side-chain anchoring instead requires a C-terminal residue with an attachable side chain and, as the aspartimide and pyroglutamate cases show, inherits that residue's own failure chemistry. Against that, it uses handles and cleavage cocktails already on the bench, and the cases it serves are the ones in which α-carboxyl attachment is documented to fail or fall short: cysteine ^[3], asparagine and glutamine ^[2], and thioester acyl donors ^[9]. The comparison that matters at scale is not purity but the durability of the anchoring bond over a long synthesis ^[6] — and, like the choice of resin and linker itself, it is fixed before the first residue goes on.