Proline cis–trans isomerization: when two chromatographic peaks are one peptide

A purity chromatogram is read as a census of molecules: one peak, one species, and the area percentages distribute the sample among them. That reading is usually correct and occasionally badly wrong. A peptide containing proline can present as two well-separated peaks, or as a peak with a pronounced shoulder, while being chemically a single compound of a single mass and a single sequence. The two peaks are two conformations of the same molecule, interconverting too slowly to average out during the time the sample spends on the column. Integrating them as parent and impurity understates purity, sometimes badly, and no amount of care in the integration will fix a misinterpretation that occurred before the integration began.

The phenomenon has a specific structural cause, and proline is almost always responsible.

Why proline is the exception

The amide bond that links residues in a peptide backbone is not a simple single bond. Delocalization of the nitrogen lone pair into the carbonyl gives the C–N linkage substantial double-bond character, which flattens the amide unit into a plane and imposes a considerable barrier to rotation about that bond. Two planar arrangements are possible: trans, with the two flanking alpha-carbons on opposite sides of the C–N axis, and cis, with them on the same side.

For an ordinary secondary amide — any residue followed by anything other than proline — this is not a real competition. The cis arrangement forces the two alpha-carbons and everything hanging off them into close contact, and the resulting steric penalty is severe enough that the cis population in a typical peptide sits well below one percent. Backbone geometry is effectively all-trans, and structural models can assume it without much risk.

Proline breaks the assumption because it is not a secondary amide. Proline’s side chain loops back and bonds covalently to its own backbone nitrogen, forming the five-membered pyrrolidine ring and leaving the nitrogen with three carbon substituents rather than two carbons and a hydrogen. The amide preceding a proline is therefore tertiary. The consequence is a near-cancellation of the steric argument that governs every other position: in the trans arrangement the preceding residue’s alpha-carbon is close to proline’s ring carbon at the delta position, and in the cis arrangement it is close to proline’s own alpha-carbon. Either way a carbon is in the way. The energetic gap that overwhelmingly favours trans elsewhere narrows to something comparable to thermal energy.

The result is real, measurable populations of both isomers. In short unstructured peptides the cis fraction at a given Xaa–Pro bond is commonly reported in the range of roughly five to thirty percent, depending on the preceding residue, the solvent, and temperature. Aromatic residues immediately before proline — tyrosine, phenylalanine, tryptophan — are consistently associated with elevated cis populations, attributed to stacking between the aromatic ring and the pyrrolidine ring that is geometrically accessible in the cis form and not the trans. An aromatic–proline pair flags a candidate for conformational doubling before anything is injected.

The timescale that makes it visible

Population alone would not produce two peaks. Every peptide in solution samples many conformations, and chromatography averages them into a single peak because interconversion is fast relative to the separation. What distinguishes the proline amide is not that two states exist but that the barrier between them is high.

Rotation about a peptide amide bond carries a barrier on the order of twenty kilocalories per mole, which corresponds to interconversion times measured in seconds to minutes near room temperature — slow in molecular terms by an enormous margin. Ordinary bond rotations happen on the picosecond scale; side chains reorient in nanoseconds; even large-scale rearrangements in folded proteins are typically microseconds to milliseconds. A process taking tens of seconds is, from the perspective of a chromatographic run, nearly frozen.

That is the whole mechanism. A molecule that enters the column in the cis form largely stays cis for the duration of its transit, and a molecule that enters trans stays trans. The two forms differ in shape, therefore in the hydrophobic surface they present to the stationary phase, therefore in retention. They elute at different times as separate bands. Molecules that do happen to interconvert mid-column elute between the two positions, which is why conformational doubling frequently appears not as two clean peaks but as two peaks bridged by an elevated baseline — the chromatographic signature of on-column exchange.

The same slow exchange gives separate resonance sets in solution-state NMR rather than a single averaged set, and it is the reason proline isomerization is often rate-limiting in protein folding. Cells maintain a dedicated enzyme family, the peptidyl-prolyl isomerases — cyclophilins, FKBPs, and parvulins among them — to accelerate a rotation that would otherwise be a kinetic bottleneck. That such enzymes exist is the clearest evidence that the barrier is biologically significant, not a laboratory curiosity.

Distinguishing conformers from impurities

The practical question is how to tell conformational doubling from a genuine related substance, because the two look similar on a chromatogram and the purity figure that results is very different.

Mass is the first test and an incomplete one. Both conformers have identical elemental composition, so mass spectrometry across the two peaks returns the same value. That immediately excludes the most common synthesis-derived impurities — deletion sequences, truncations, incomplete side-chain deprotection, oxidation — all of which shift the mass. But identical mass does not prove conformational identity, because one important impurity class shares it. A D-amino acid epimer arising from racemization during coupling has exactly the same mass as the parent and can resolve chromatographically. Same mass, two peaks, is consistent with both explanations.

Temperature is the discriminating test. Raising the column temperature accelerates interconversion exponentially, and at sufficient temperature exchange becomes fast relative to the separation. Two conformer peaks respond by broadening, drawing together, and eventually coalescing into a single averaged peak. Epimers do not do this. A D-amino acid at a stereocentre is a distinct covalent configuration that cannot interconvert with its L counterpart under chromatographic conditions at any accessible temperature; raising the temperature will change the epimer separation somewhat, as it changes every separation, but the two peaks will not merge into one. Running the same sample at two or three column temperatures and watching whether the peaks coalesce separates the two hypotheses cleanly, and it is a diagnostic that requires no additional sample and no additional instrumentation.

Two supporting observations are worth having. Re-injection of a collected fraction is informative: material collected from one conformer peak will, after standing long enough for equilibrium to re-establish, re-inject as both peaks in the original ratio, because the isolated conformer relaxes back to the equilibrium mixture. A collected impurity peak re-injects as itself. And the peak ratio is a thermodynamic quantity rather than a manufacturing one — it should be reproducible across lots of the same sequence made by the same route, whereas a racemization or deletion impurity varies with the specifics of the synthesis. An “impurity” appearing at an identical 12% in every lot from every supplier is not an impurity.

Sequence inspection remains the cheapest screen. No proline in the sequence makes conformational doubling of this kind unlikely; a proline preceded by an aromatic residue makes it a strong prior.

Consequences beyond peak shape

Proline’s conformational behaviour propagates into several other areas that intersect with peptide manufacture and characterization.

In synthesis, the cis-accessible geometry at Xaa–Pro drives a specific failure mode. When proline occupies the second position from the C-terminus, the resin-bound dipeptide can adopt the geometry required for the free N-terminal amine to attack the ester linking the peptide to the support, cyclizing to a diketopiperazine and cleaving the chain from the resin. The result is early loss of material and a truncated impurity profile — a known hazard of X-Pro C-terminal sequences that shapes resin and protecting-group strategy for such targets.

In enzymatic characterization, proline blocks proteolysis. Trypsin, the workhorse of peptide mapping, cleaves after lysine and arginine but not when the following residue is proline. A map assuming every K and R is a cleavage site will predict fragments that never appear, and the resulting coverage gap can be mistaken for a modification. The rule belongs in fragment prediction rather than being discovered from the data.

In secondary structure, proline is a helix breaker on two counts. Its ring constrains the backbone dihedral angle to a narrow range incompatible with the interior of an alpha-helix, and it has no backbone amide hydrogen to donate to the helical hydrogen-bonding network. Its enrichment in turns follows from the same geometry. Extended runs of proline adopt the distinctive polyproline II conformation, an all-trans left-handed helix that is common in regions that would otherwise be called disordered and that carries a characteristic signature in circular dichroism spectra — a useful reminder that “no ordered structure by CD” and “polyproline II” are not the same finding.

And in conformationally constrained design, proline and its N-alkylated relatives are used deliberately for exactly the property described here. Restricting backbone freedom is the point; the isomerization equilibrium is a lever rather than a nuisance.

What this means for reading a certificate

None of this appears on a certificate of analysis. A certificate reports a purity percentage and, at best, a list of individual impurities above a reporting threshold. It does not report whether the method development that produced those numbers included a temperature study, and it does not distinguish a conformer from a related substance. Two laboratories analyzing the same proline-containing peptide, one running at ambient column temperature and one at sixty degrees, can generate materially different purity figures from identical material — a variable that sits alongside the other method parameters that move a purity number without any change in the sample.

The general lesson is one this site returns to often: a chromatographic peak is an operationally defined object, not a molecular one — a band of material that eluted in a particular window under particular conditions. Most of the time that band corresponds to a chemical species, which is why the shorthand works. Proline is a standing reminder that the correspondence is a convention rather than a law, and that the interesting analytical questions tend to live precisely where the convention fails.

Further reading

Research use only. This post is for educational and reference purposes on peptide synthetic and analytical chemistry. It does not constitute medical, veterinary, or dosing guidance.