HPLC Analysis for Peptide Purity

High-performance liquid chromatography is the workhorse purity test for synthetic peptides because most research sequences are ionizable, UV-active at the peptide bond, and separable on reverse-phase silica. The number on a CoA is only as good as the method that produced it. This article explains how that method is built and how to look at the trace you receive.

ChinaPeptides reports HPLC as part of quality control for research lots. The same chromatogram is used during purification to decide where to cut. Analytical and preparative methods are related but not identical; a shallow analytical gradient can reveal shoulders that a fast prep gradient hid.

Why reverse-phase C18 (and relatives) dominate

Peptides interact with alkyl-bonded silica through hydrophobic face contacts and with the mobile phase through ionic and hydrogen-bonding terms. Acetonitrile/water gradients with an ion-pair reagent resolve most linear research peptides. C18 is the default for mid-size sequences. Wider-pore (300 Å) media are preferred as molecular weight grows so the chain can enter the pores. C4 or C8 can reduce destructive retention of very hydrophobic peptides. HILIC or mixed-mode methods appear when reverse-phase fails, but they are not the first method on a typical peptide CoA.

Column temperature changes selectivity. A few degrees can split a shoulder or merge two peaks. If you compare your in-house trace to the manufacturer’s, match temperature as well as gradient.

Ion pairing: TFA, formic acid, and the UV trade-off

Trifluoroacetic acid at about 0.05–0.1% is the traditional ion-pair additive. It protonates carboxylates, pairs with basic residues, and sharpens peaks for many sequences. It also suppresses electrospray ionization if the same eluent is sent to a mass spectrometer, which is why LC-MS methods often use formic acid instead. Formic acid methods can show different selectivity; a peptide that is a single peak in TFA/ACN may split or merge in formic acid/ACN.

If your assay is TFA-sensitive, that is a salt-exchange problem, not a reason to abandon TFA HPLC for release. You can purify in TFA, exchange to acetate or HCl, and still use a TFA analytical method for comparison—or validate a formic-acid method if that is what your lab runs. State the need on the synthesis specification.

Building a gradient that tells the truth

A generic 5–95% acetonitrile run in ten minutes will make most peptides look like one peak. A release method should spend time in the region where the peptide actually elutes. Chemists often run a scouting gradient, then a shallow segment around the product. Related deletions that differ by one leucine or isoleucine need that shallow segment; they will not separate on a steep screen.

Detection at 214 or 220 nm is standard. 280 nm highlights Trp/Tyr-containing species and can make a tryptophan-rich impurity look larger than a Trp-free deletion. If the CoA shows only one wavelength, do not assume the impurity inventory is complete for a multi-chromophore mixture such as a dye-labeled peptide. Labeled peptides need a wavelength for the peptide and one for the dye; free dye is a real impurity in fluorescent peptide lots.

Method choice	Effect on the chromatogram	When to change it
Steep generic gradient	Single fat peak, hidden shoulders	Never as the only release method for a close analog
Shallow segment at elution	Resolves related deletions	Default for ≥98% claims
TFA ion pair	Sharp peaks, MS-unfriendly	Standard UV purity
Formic acid	MS-friendly, different selectivity	LC-MS identity, orthogonal check
300 Å pore, C4/C8	Less trapping of large/hydrophobic chains	Long or lipidated peptides

Reading the release chromatogram

Start with the method header: column, pore size, flow, temperature, A/B solvents, gradient table, wavelength, and integration events. Then look at the peak: is the main component symmetric? A fronting or tailing peak can be overload, a secondary interaction with silanols, or an unresolved companion. Integration start and stop lines should not clip a shoulder that is obviously present.

Area percent assumes everything that counts is in the integration window and is UV-active. Early solvent fronts and late column-wash peaks need a consistent policy. Some labs exclude the injection disturbance; others include small early peaks as polar impurities. If you are qualifying a supplier, ask which policy they use rather than arguing about 0.4%.

Compare the HPLC result to the mass spectrum. A clean single UV peak with two masses is a co-elution. Two UV peaks with the same mass can be conformational isomers, cis-trans proline, or aspartimide-related isomers. The LC-MS characterization article covers that disagreement in more detail.

Artifacts that are not “the peptide is dirty”

Overload produces a triangular peak and fake shoulders. A dirty injector or carryover from a previous hydrophobic peptide creates ghost peaks that vanish on a blank. Peptide adsorption to old vials changes apparent purity if you re-inject a dilute sample after it sat. Acetonitrile evaporation in a poorly sealed vial concentrates late-eluting material. None of these are synthesis defects, but they can appear if you rerun a CoA sample carelessly.

Conformational broadening is common for proline-rich and partially structured peptides. Raising temperature or adding a small amount of isopropanol can sharpen the peak without changing covalent purity. Do not “purify away” a conformational box by throwing away half the peptide unless you have proof of a covalent isomer.

What to request and what to archive

For a research lot, request the analytical method summary and the chromatogram of the released pool, not only the number. If the peptide is a standard you will use for months, archive the method so your in-house QC can be compared. If you change columns, expect a small shift in area percent; that is method change control, not lot failure.

HPLC does not replace identity testing, solubility checks, or common sense about which purity grade you ordered. It is the purity language the field uses. Learn to read the dialect.