Reversed-Phase HPLC for Peptides: Column, Gradient, Mobile Phase and Detection Wavelength

A reversed-phase method for a peptide is built from six decisions: stationary phase and pore size, column dimensions and particle size, the acidic mobile-phase modifier, the gradient programme, column temperature, and the injection and detection conditions. Each decision changes which species are separated from the main peak, and therefore changes the purity figure the method will later report. This page covers how those decisions are made. What a finished purity figure means, and what it omits, is covered separately in the handbook's article on HPLC purity; this one begins earlier, with an empty column and a sample of unknown behaviour.

The worked assumptions throughout are generic: a synthetic peptide of about 20 residues, isolated as a trifluoroacetate salt, analysed for purity and related impurities on an ordinary binary-pump HPLC with a diode-array detector. The principles are drawn from the standard review literature on peptide chromatography 1 and from the pharmacopoeial and ICH texts that govern chromatographic procedures 4 5.

Stationary phase, pore size and particle size

Retention in reversed-phase chromatography is governed by hydrophobic contact between the peptide and the bonded alkyl chains. For peptides the contribution of each residue to retention is approximately additive, which is why retention coefficients for individual amino acids could be derived from model synthetic peptides and used to predict elution order 2. The practical consequence is that the column is chosen for the size and overall hydrophobicity of the sequence rather than for any single residue.

Pore size matters as much as chain length. A peptide must diffuse into the pores to interact with most of the bonded surface; if the pores are narrow relative to the molecule, access is restricted, peaks broaden and recovery falls. Standard 100–130 Å phases serve short and mid-length peptides well. Wide-pore 300 Å phases are the conventional choice once chains are long enough, or folded enough, for restricted diffusion to matter — commonly somewhere beyond 30 to 40 residues 1.

Column dimensions follow the detector and the sample. A 4.6 × 150 mm column at 1.0 mL/min is the robust default for ultraviolet purity work. A 2.1 × 100 or 2.1 × 150 mm column at 0.2–0.4 mL/min suits mass-spectrometric detection, where lower flow improves ionisation and lower sample consumption matters. Once a method exists, the harmonised pharmacopoeial chromatography text permits limited changes to column length and particle size, holding the ratio of length to particle diameter within −25% to +50% of the original, provided system suitability is still met 5.

Mobile phase and the acidic modifier

The mobile phases are water (A) and acetonitrile (B), each carrying the same acidic modifier. The modifier does three jobs: it holds the pH near 2 so that acidic side chains and the C-terminus are protonated and uncharged, it suppresses interaction between basic residues and residual silanols on the silica, and — in the case of trifluoroacetic acid — it pairs with the protonated basic groups to form a more hydrophobic, better-retained complex. Without a modifier, basic peptides tail badly and retention drifts 1.

Trifluoroacetic acid at 0.1% (v/v), about 13 mM, gives the sharpest peaks and remains the reference choice for ultraviolet purity methods. Its cost appears at the detector. It absorbs at low wavelength, so the baseline rises as the proportion of B increases; some methods reduce the concentration in B slightly to flatten the drift. More seriously, it suppresses electrospray ionisation, through ion pairing in the droplet and changes in surface tension, lowering mass-spectrometric signal substantially 3. Formic acid at 0.1% is the usual choice for LC-MS: weaker ion pairing, broader peaks for basic sequences and somewhat different selectivity, but far better ionisation. Difluoroacetic acid is used as a compromise between the two.

A consequence for method design: a purity method developed with trifluoroacetic acid and an identity method run with formic acid will not show the same elution order in every case. Where the two must be compared peak for peak, confirm the assignment with a spiked or stressed sample rather than by assuming the order is preserved.

Gradient design

Peptide retention changes steeply with the proportion of organic solvent: a shift of a few per cent in B can move a peptide from strongly retained to unretained. Isocratic elution is therefore impractical except for the simplest separations, and nearly all peptide methods use a linear gradient 1. The slope of that gradient, in % B per minute, is the single largest influence on how well the main peak is separated from its close relatives, most of which differ from the target by one residue or one modification.

1. Run a blank gradient first. Record the baseline at 214 nm so that system peaks are not later integrated as impurities.
2. Run a scouting gradient: 5% to 65% B over 60 minutes, which is 1% B per minute, on a 4.6 × 150 mm column at 1.0 mL/min.
3. Note the % B at which the main peak elutes, after correcting for the dwell volume between mixer and column.
4. Design a focused gradient that starts about 10% B below the elution point and ends about 10% B above it, at 0.25–0.5% B per minute.
5. Add a wash step at 90–95% B for at least five minutes, so that strongly retained material leaves the column instead of appearing in the next run.
6. Re-equilibrate at the starting composition for about ten column volumes — roughly 15 mL for a 4.6 × 150 mm column.
7. Inject a deliberately stressed sample and confirm that the new peaks resolve from the main peak. If they do not, flatten the slope or change the stationary phase before anything else.

Dwell volume is the most common reason a gradient method fails on a second instrument. The delay between the programmed change in composition and its arrival at the column differs between systems, and on narrow-bore columns it can equal several minutes of run time. Measure it on each instrument and state it in the method.

Temperature, flow and injection

Column temperature is set, not left to the room. Raising it from ambient to 40–60 °C lowers mobile-phase viscosity, sharpens peaks by improving mass transfer and can change selectivity between closely related species 1. The limit is the stability of the sample and of the column: check that a sample held on the column at the chosen temperature does not generate new peaks, and stay within the manufacturer's limit for the phase at low pH.

Injection conditions distort peaks more often than any other avoidable cause. Dissolve the sample in mobile phase A, or in water with no more organic solvent than the starting composition. A sample dissolved in strong solvent and injected in a large volume travels partly unretained, giving split or fronting peaks. On a 4.6 mm column, 10–20 µL of a solution around 0.5–1 mg/mL is a typical starting point; on a 2.1 mm column, 1–5 µL. Keep the main peak well inside the detector's linear range, commonly below about one absorbance unit, or its area is under-reported and every impurity ratio is inflated.

Detection wavelength

For purity, detect at 214 nm (or within 210–220 nm), where every peptide bond absorbs and closely related impurities respond roughly in proportion to chain length. Detection at 280 nm responds only to tryptophan and tyrosine and is used for identity support or concentration work, not for purity. Configure the diode-array detector to store the full spectrum across the run, typically 200–300 nm, with the reference wavelength switched off so that it cannot subtract real signal. A stored spectrum lets a later analyst check peak homogeneity across the main peak and re-examine a region at another wavelength without re-running the sample.

A reference method

The table sets out a starting method for a generic 20-residue peptide. It is a point of departure for development, not a validated procedure; the focused gradient in particular must be rebuilt around the observed elution point.

Recording the method and knowing when it is finished

ICH Q14 treats method development as a documented exercise: define what the procedure must measure and to what performance, identify the parameters that affect that performance, and record the ranges within which the method was shown to work 4. For a small laboratory the proportionate version is a one-page development record listing each parameter tried, what changed, and why the final value was chosen. That record is what makes later robustness claims and system suitability limits defensible.

Regulatory expectations for synthetic peptides make the same point about sensitivity and selectivity. The European guideline on synthetic peptides asks that impurity methods be able to meet a reporting threshold of 0.1%, and warns specifically about the risk of co-eluting impurities given the structural similarity of peptide-related species 6. A research laboratory is not bound by that guideline, but it describes what a capable purity method looks like.

1. State the purpose: analyte, sample type, and whether the method must quantify impurities or only compare profiles.
2. Record the column by chemistry, dimensions, particle size, pore size, manufacturer and lot.
3. Record both mobile phases with the modifier and its concentration, and how they are prepared.
4. Record the gradient table in full, including wash, re-equilibration and the measured dwell volume.
5. Record flow rate, column temperature, autosampler temperature, injection volume and sample diluent.
6. Record detection wavelength, bandwidth, reference setting and data rate.
7. Record the integration parameters, so that two analysts integrate the same trace the same way.
8. Attach the chromatograms of the blank, a typical sample and a stressed sample.
9. Define the system suitability criteria before the method is used on real samples.

A method is ready for validation or verification when a stressed sample shows its degradation products resolved from the main peak, a blank shows nothing that could be mistaken for an impurity, replicate injections agree, and every parameter above is written down. Until then it is a set of conditions, not a method.