Aspartimide Formation, And Why Asp-Gly Is A Difficult Neighbour

In Fmoc-based solid-phase synthesis, every residue addition is followed by a base step to strip the Fmoc protecting group from the new amine. The standard condition is 20 percent piperidine in dimethylformamide, typically two exposures of 5 to 10 minutes at ambient temperature. The growing chain therefore sits in a strongly basic environment once per residue, and any structure that is base-labile gets many chances to react.

An aspartate side chain protected as a tert-butyl ester is one such structure. The amide nitrogen of the residue immediately after it can be deprotonated by piperidine and then attack the side-chain ester carbonyl. That closes a five-membered succinimide ring, expelling the tert-butyl ester group and forming what is conventionally called the aspartimide.

The reaction is a function of the neighbour. Glycine after aspartate is the worst case, because glycine has no side chain to get in the way and its backbone nitrogen is the most accessible of the twenty. Asp-Gly is the sequence that appears in every discussion of this problem for that reason.

The mass bookkeeping

Ring closure is a dehydration relative to the free acid form, so the aspartimide species is 18.011 Da lighter than the intended peptide on a monoisotopic basis. That is a visible shift by electrospray mass spectrometry and is usually the first clue.

The imide ring is electrophilic, and piperidine is present in large excess, so it is also attacked by piperidine itself. That opens the ring and leaves a piperidide adduct: the aspartimide at minus 18.011 Da plus piperidine at 85.089 Da, for a net plus 67.078 Da against the target. A peak at plus 67 Da in an electrospray spectrum of a crude Fmoc synthesis containing aspartate is close to diagnostic.

Why hydrolysis of the ring is the real problem

If the imide ring is opened by water rather than by piperidine, it can open at either carbonyl. Opening at one carbonyl restores the normal alpha peptide linkage and the correct sequence. Opening at the other produces a beta-aspartyl linkage, in which the backbone now runs through the side-chain carboxyl and the alpha carboxyl hangs free.

Both products have the same elemental composition. They differ in connectivity, in chromatographic retention and in conformation, and not at all in mass. Reported ratios from base-mediated ring opening of model aspartimide peptides generally favour the beta product, in the region of 2 to 3 parts beta to 1 part alpha, measured by reverse-phase HPLC peak area on separated standards.

Racemisation at the aspartate alpha carbon is a further consequence, because the ring carbon is flanked by two carbonyls and the proton on it is acidic. The result is a family of four species from one residue: alpha and beta linkages, each in L and D form.

The failure mode: a beta isomer counted inside the main peak

The failure that matters is not a visible impurity peak. It is a beta-aspartyl isomer that elutes close enough to the target to sit inside the same integrated peak on a routine gradient, contributing its area to the reported purity figure and its mass to the reported identity confirmation.

How it shows up: a purity figure in the high nineties, a clean electrospray spectrum matching the expected mass, and a shoulder on the main peak that changes shape when the gradient is made shallower or the column temperature is changed. Re-running the same sample on a 1 percent per minute acetonitrile gradient instead of 2 percent per minute is often enough to split the peak into two. Nothing about the original chromatogram was wrong. The integration boundary was simply drawn around two compounds.

What synthesis chemists do about it

The mitigations all work by making the side-chain ester harder to attack or by shortening the base exposure. Bulky side-chain esters are the most common answer: 3-methylpent-3-yl and 2-methylpent-3-yl esters hinder the approach of the neighbouring nitrogen far more than a tert-butyl ester does. Additives that buffer the deprotection also help, with 0.1 molar hydroxybenzotriazole or a formic acid additive in the piperidine solution both reported to reduce imide formation.

• Sterically hindered aspartate side-chain esters in place of tert-butyl.
• Acidic additives in the Fmoc removal solution, for example 0.1 molar hydroxybenzotriazole.
• Shorter base contact time and ambient rather than elevated temperature, since heating an Asp-Gly sequence accelerates the reaction sharply.
• Backbone amide protection on the residue after aspartate, which removes the nitrogen that does the attacking.

Elevated-temperature synthesis deserves a separate line. Microwave and heated synthesis methods shorten coupling times considerably and are widely used, and they also accelerate this particular side reaction. A sequence that behaves at 25 degrees Celsius can become problematic at 75 degrees Celsius with unchanged base conditions.

What Aurum's published figures do and do not resolve

Purity is assayed by reverse-phase HPLC with ultraviolet detection at 214 nm, and identity is assayed by mass spectrometry. Those are the specifications Aurum publishes, and neither of them can exclude a beta-aspartyl isomer. Mass is blind to it by definition, and HPLC only resolves it if the method was developed to do so.

Isomer-specific analysis is not among the specifications Aurum publishes. Distinguishing alpha from beta linkage requires methods outside an identity and purity panel: enzymatic digestion with a linkage-specific protease, nuclear magnetic resonance, or a chromatographic method deliberately optimised against authentic isomer standards. Where a sequence contains an aspartate followed by glycine, no published figure speaks to the isomer question.

This is a specific case of a general point about what a percentage from a chromatogram represents.

Common questions

Which sequences are most affected?

Aspartate followed by glycine is the reference worst case. Asp-Asn, Asp-Ser and Asp-Ala are also noted as susceptible. Aspartate followed by a beta-branched residue such as valine or isoleucine is the least susceptible, because the neighbouring nitrogen is shielded.

Does glutamate do the same thing?

The equivalent cyclisation at glutamate would close a six-membered ring, which is entropically and geometrically less favourable. It is reported but at far lower rates, which is why the problem is discussed as an aspartate problem.

Is the plus 67 Da species removable?

The piperidide adduct differs enough in mass and usually in retention to be separated by preparative chromatography. It is the isobaric beta isomer, not the adduct, that survives purification unnoticed.

Can a chromatogram alone rule this out?

Only if the method was shown to resolve the isomers, using authentic standards of both. A single symmetrical peak on an unvalidated gradient is not evidence of a single compound.

Does the imide survive cleavage from the resin?

The strong acid cleavage conditions used at the end of an Fmoc synthesis will open the ring, which is where the alpha and beta mixture is generated. The imide is largely an on-resin intermediate; the isomers are what ends up in the vial.