Difficult sequences: why some peptides resist assembly on the resin
Solid-phase peptide synthesis is usually described as a repeating cycle, and the description implies that every cycle is equivalent: deprotect, wash, couple, wash, repeat. If each step ran to completion the length of the target would be a scheduling question rather than a chemical one. In practice, coupling efficiency is not a constant of the chemistry. It is a property of the particular chain being assembled at the particular moment it is being assembled, and for certain sequences it collapses — not gradually across the synthesis, but abruptly, at a specific residue, after which every subsequent step underperforms. The peptide that comes off the resin is then a mixture in which the target is one component among many, and no amount of downstream purification recovers material that was never made.
These are the sequences the literature has called “difficult” since the term was formalized in the late 1980s, and the label is more precise than it sounds. It does not mean long, and it does not mean chemically exotic. Some twelve-residue sequences are far harder to assemble at high purity than some forty-residue ones. What distinguishes them is a structural event that happens on the resin, to the growing chains, while they are still attached.
The failure is physical before it is chemical
A coupling reaction requires that the activated amino acid reach a free N-terminal amine. In solution that is a diffusion problem with a trivial answer. On a resin bead it is not, because the growing peptide chains are not isolated strands waving in solvent. They are tethered at high local concentration — often in the tens to hundreds of millimolar within the swollen bead — and at that density the backbone amides of adjacent chains can hydrogen-bond to one another. When they do, the chains organize into β-sheet structure inside the polymer matrix.
The consequences follow directly from that structure. In a β-sheet the backbone amide protons are engaged in interchain hydrogen bonds, the chains are packed, and the N-terminal amine is no longer freely accessible to an incoming activated species. The bead itself changes: solvation collapses, swelling decreases, and the resin’s apparent volume in DMF can drop noticeably over the space of two or three cycles. Reagent access to the interior falls with it. Deprotection slows as well, because piperidine must reach the same buried positions the coupling reagent could not, so the next cycle begins with a fraction of the chains still carrying an Fmoc group that will never be removed in time.
The signature of this is characteristic. Coupling proceeds cleanly for the first several residues, then a threshold is crossed — commonly somewhere between residue five and residue fifteen — and the Kaiser or chloranil test suddenly reports incomplete coupling where it had been reporting completion. The onset is sharp because sheet formation is cooperative. Once nucleated, it propagates.
Sequence predicts it, imperfectly
The compositional risk factors are well described. Runs of hydrophobic residues with small, unbranched or β-branched side chains aggregate most readily: alanine, valine, isoleucine, leucine, phenylalanine. Polyalanine stretches are the textbook case. Glutamine and threonine contribute through side-chain hydrogen bonding in addition to backbone effects. Sequences containing repeated Val–Val, Ile–Ile, or Ala–Ala motifs are flagged routinely in synthesis planning.
What breaks any purely compositional rule is that aggregation depends on register and context, not on amino acid inventory alone. Two sequences with identical composition and different order can behave entirely differently, because β-sheet formation requires that the hydrogen-bonding partners line up. Proline is the clearest illustration of the inverse: it has no backbone amide proton to donate, so it cannot participate in the sheet, and a single proline placed in an otherwise aggregation-prone run frequently suppresses the problem outright. The same structural logic that makes proline a conformational disruptor in a folded peptide — discussed in the context of cis–trans isomerization — makes it an aggregation breaker on the resin.
Predictive algorithms exist and are used, but they are heuristics. The practical position in most laboratories is that a sequence is flagged as probably difficult before synthesis and confirmed as actually difficult during it, by monitoring.
Pseudoproline dipeptides: building the disruption in
If proline suppresses aggregation and the target sequence has no proline, the chemical solution is to install a temporary one. Pseudoproline dipeptides do exactly this. A serine, threonine, or cysteine residue is condensed with an aldehyde or ketone to form a five-membered oxazolidine ring (from Ser or Thr) or thiazolidine ring (from Cys) that incorporates the side-chain oxygen or sulfur and the backbone nitrogen. The result is structurally a proline mimic: a cyclic, tertiary-amide residue that introduces a kink in the backbone and removes the amide proton from the hydrogen-bonding inventory.
Two features matter operationally. First, the pseudoproline is introduced as a preformed dipeptide building block — Fmoc-Xaa-Ser(ψMe,Mepro)-OH and its analogues — rather than as a single residue, because the ring nitrogen is too hindered to acylate efficiently in a standard coupling. This constrains placement: a pseudoproline can only go where a Ser, Thr, or Cys already exists in the target sequence, and the residue preceding it is fixed by the available building block. Second, the modification is transient. The oxazolidine or thiazolidine ring opens under the acidic conditions of TFA cleavage, regenerating the native Ser, Thr, or Cys with no residual modification. The final peptide is the intended sequence; the disruption existed only while it was needed.
Placement is usually recommended at intervals of roughly five to six residues through an aggregation-prone region, positioned ahead of where the sheet would nucleate rather than after it has formed.
Backbone amide protection and the process levers
Where no Ser, Thr, or Cys is available, the amide proton can be masked directly. 2-Hydroxy-4-methoxybenzyl (Hmb) and 2,4-dimethoxybenzyl (Dmb) groups attach to the backbone nitrogen, eliminating the hydrogen-bond donor and forcing a conformational change, and both are removed during TFA cleavage. The cost is steric: acylating the nitrogen adjacent to an N-alkylated residue is slow, and the coupling onto a backbone-protected residue often requires forcing conditions or a preformed dipeptide of its own. Hmb has a partial workaround in that its phenolic hydroxyl can accept the incoming acyl group and transfer it intramolecularly, but the chemistry is more demanding than pseudoproline substitution, which is why it tends to be the second option rather than the first.
Beyond the building blocks, the process variables are the ones most often adjusted first because they require no change to the synthetic plan:
- Solvent and additives. Chaotropic salts, DMSO co-solvent, or a shift from DMF to NMP can disrupt interchain hydrogen bonding and restore swelling. The effect is real but sequence-dependent and rarely sufficient alone for a severely aggregating stretch.
- Resin architecture. Low substitution levels reduce the local chain concentration that drives aggregation in the first place, at the cost of yield per gram. PEG-based supports solvate polar peptide chains more effectively than polystyrene and are standard for known-difficult targets.
- Elevated temperature. Heating, including microwave-assisted protocols, increases coupling rates and can overcome a partially aggregated state. This carries a specific analytical consequence: elevated temperature accelerates base-mediated epimerization, particularly at cysteine and histidine, so a thermal solution to an aggregation problem can produce a chiral purity problem in its place.
- Double coupling and capping. Repeating the coupling step raises conversion; capping unreacted amines with acetic anhydride converts what would become an internal deletion sequence into a truncated chain that is easier to separate chromatographically.
What reaches the certificate
None of this appears on a certificate of analysis, but its consequences do. Incomplete couplings in an aggregated region generate deletion sequences — chains missing one internal residue — and these are the impurity class least likely to be resolved by reversed-phase chromatography, because removing a single residue from the middle of a long hydrophobic stretch can leave retention nearly unchanged. A sequence that aggregates during synthesis therefore tends to carry a specific and partly co-eluting impurity burden, one whose origins are covered in more detail in the survey of where synthetic peptide impurities come from, and one that argues for identity and orthogonal methods rather than a single purity figure.
It also explains a pattern that otherwise looks like inconsistency. Two suppliers producing the same sequence to the same stated purity may have used different resins, different pseudoproline placements, or different thermal profiles, and their impurity fingerprints will differ accordingly — not because one process failed, but because the same structural obstacle admits several partial solutions with different residual signatures. Lot-to-lot variation within a single supplier has the same root whenever a synthesis runs near the edge of where its aggregation control is adequate.
The useful reframing is that a difficult sequence is not a manufacturing defect but a structural property of the target, present in the sequence before any chemistry is attempted. The peptide is trying to fold while it is being built. Synthesis strategy for these targets consists largely of preventing that folding until assembly is complete — and the analytical profile of the finished material is, in part, a record of how well that prevention worked.