Peptide Purification and HPLC

Every synthetic peptide leaves the resin as a crude mixture, and every peptide taken from a natural source comes with a great many other molecules. Purification separates the target from them, and high-performance liquid chromatography (HPLC), above all in its reversed-phase form, is the standard tool for the job[1][2]. The same technique is then used again to check how pure the product is[3].

The difficulty lies less in the amount of impurity than in its similarity. The by-products of synthesis are near-twins of the target: chains missing one residue, chains with one amino acid in its mirror-image form, chains with an oxidised methionine. A purification method has to tell apart molecules that may differ by a single carbon atom, and the purity figure at the end is only as good as the method's ability to see such differences. Both halves of the problem, separating and seeing, run through this page.

What has to be removed

Impurities fall into two groups. Process-related impurities come from the production method itself, such as salts or, for peptides made in cells, DNA fragments and cell debris. Product-related impurities are chemically similar to the target, and they are the hard part[1]. In synthetic peptides they include deletion sequences lacking one residue, epimers formed by racemisation, β-linked peptides formed when aspartic acid or serine rearranges, and peptides in which methionine, tryptophan or tyrosine has been alkylated or methionine or cysteine oxidised[1]. Most of them arise during chain assembly and cleavage, as described on the solid-phase synthesis page.

Peptides from natural sources pose a different problem: the target may be a tiny fraction of a tissue extract. Such extracts are first clarified, concentrated on small cartridges packed with reversed-phase material and roughly fractionated, for example by gel filtration, before HPLC is applied, often in several rounds[4].

How reversed-phase HPLC separates peptides

In reversed-phase HPLC the column is packed with porous silica particles whose surface carries hydrocarbon chains, most often of 18 or 8 carbon atoms (C18, C8). The peptide is loaded in a mostly aqueous solution and sticks to these chains through its hydrophobic side chains. An organic solvent, usually acetonitrile, is then added in a steadily rising proportion, and each peptide comes off when the solvent becomes strong enough to release it. Peptides therefore leave the column in order of increasing hydrophobicity[3][2]. Tryptophan, phenylalanine, leucine and isoleucine make a peptide stick most strongly, while charged and small polar residues add little[2].

Peptides behave differently from small molecules in one important respect. A small molecule touches the surface at one point, while a peptide or protein binds through several residues at once. Xindu Geng and Fred Regnier described retention as the displacement of a certain number of solvent molecules, Z, needed to release the molecule from the surface, and found that Z grows with molecular size. The larger Z is, the narrower the range of solvent concentration over which the molecule passes from firmly bound to freely moving[5].

For peptides the window is so narrow that high-resolution separation at a constant solvent composition can rarely be achieved: a small change in solvent takes a peptide from strongly retained to barely retained. Peptide mixtures are therefore routinely eluted with a gradient[3]. For most analytical separations, Mant and colleagues recommend a linear gradient from aqueous TFA to TFA in acetonitrile that rises by 0.5–2% acetonitrile per minute, at room temperature[2]. A longer, shallower gradient generally separates better, but it also keeps the peptide on the column for longer[3].

Standard conditions and the role of TFA

A typical starting method looks like this[3]:

Pore size matters because a molecule has to enter the particle to reach most of the surface. Pores of 100–300 Å suit peptides; for proteins they should be at least 300 Å. The guideline is that the pore should be roughly ten times wider than the molecule, so that it can diffuse in freely and reach the whole surface. Shorter C4 or C8 chains and other solvents such as 2-propanol change the selectivity, so two peptides that run together on one column may separate on another[3].

Trifluoroacetic acid does several jobs at once. It sets the pH at about 2, where the carboxyl groups of the peptide carry no charge and the silanol groups left on the silica surface, which become negatively charged above pH 3–4, cannot bind the peptide's positive charges. Its anion pairs with the positively charged groups of the peptide, an effect known as ion pairing. And it is volatile, so it evaporates when the collected fractions are freeze-dried[2]. Larger relatives of TFA, such as heptafluorobutyric acid (HFBA), pair more strongly and hold peptides with more positive charges for longer. This can change the order of elution and resolve peptides that TFA cannot. Mant's group found the best TFA concentration to be 0.2–0.25%, higher than the 0.05–0.1% traditionally used[2].

Temperature is a further handle. Raising it lowers retention and sharpens the peaks: in one test, two pairs of peptides that barely separated at 10 °C were fully resolved at 70 °C. Silica columns, however, tolerate only pH 2–8, so separations at higher pH need columns made of polymers[2].

Peptides are detected by the absorbance of the peptide bond at 210–220 nm, which every peptide has. Tryptophan and tyrosine also absorb near 280 nm, but many peptides contain neither[2]. For peptides that do, a symmetrical peak at both 214 and 280 nm is a good sign of purity[4].

Predicting and reading the order of elution

Because retention depends mainly on hydrophobicity, it can be predicted. Since the early 1980s the starting point has been to add up retention coefficients for the amino acids of a peptide[6]. Such sums are only a first approximation, because the same residue contributes differently depending on its position in the chain, its neighbours and whether the peptide folds[6]. A peptide that can form an α-helix often does so on the column, where the hydrophobic surface and the organic solvent favour the helix, and an amphipathic helix then binds with its hydrophobic face[2]. Oleg Krokhin's sequence-specific retention calculator, which allows for such effects, predicted the retention of about 2,000 peptides with a correlation of R² ≈ 0.98. In proteomics, predicted retention times serve as an extra check on peptides identified by mass spectrometry[6].

Very small differences can be resolved. In Mant's test mixtures, neighbouring peptides differed by a single carbon atom, and helical peptides carrying an L- or a D-amino acid at one position were separated once the right temperature was chosen[2]. That is the kind of difference that distinguishes a synthetic peptide from its racemised by-product.

From analysis to purification

For a synthetic peptide the usual sequence is an analytical run to see how complex the crude mixture is, a preparative run to collect the product, and a final analytical run, on the same or a different system, to check its purity[3]. Preparative work need not mean industrial equipment: Mant's group purified 200 mg of a crude 26-residue antimicrobial peptide in one step on a semi-preparative column of standard size[2].

Their rule of thumb for the gradient is simple. The crude peptide is first run on an analytical column with a gradient of 1% acetonitrile per minute, and the acetonitrile concentration at which the product elutes is noted, 42% in their example. The preparative run then climbs quickly to 12% below that value, here 30%, slows to 0.1% per minute across the product region, and speeds up again to wash off the hydrophobic impurities. Fractions were collected every 2 minutes and each one was analysed; only one fraction at each edge of the product peak contained overlapping impurities. With a larger load, the product began to elute earlier and spread over more fractions[2].

Analysing each fraction, rather than pooling by the shape of the UV trace, is what makes such a run reliable. The pure fractions are then freeze-dried. Purified peptides tend to stick to charged surfaces, so polypropylene tubes are better than glass or polystyrene, especially as the peptide becomes purer[4]. At industrial scale another problem appears: acetonitrile, the main solvent for reversed-phase purification, is costly and toxic, and greener alternatives such as ethanol, 2-propanol and dimethyl carbonate, as well as supercritical carbon dioxide, are being tested[1].

When reversed phase is not enough

Reversed-phase HPLC separates by hydrophobicity alone. An impurity with nearly the same hydrophobicity as the target will run with it however good the column is, yet it may still differ in charge, size or polarity. A second step based on a different property therefore helps more than a second run of the same kind[2][1]:

Hydrophilic interaction chromatography, named by Andrew Alpert in 1990, turns reversed phase around: a polar column is run with a mostly organic mobile phase, and the more hydrophilic a peptide is, the longer it stays on the column[7]. Mant's group combined it with cation exchange, so that peptides elute in groups according to their charge and, within each group, according to their hydrophilicity[2]. In Alpert's later variant, ERLIC, peptides containing isoaspartic acid, the β-linked form produced when aspartic acid rearranges, eluted slightly later than the same peptides with normal aspartic acid[8][1].

What a purity figure means

Purity is usually judged from the chromatogram itself[3]: the larger the main peak compared with all the others, the purer the peptide. The figure is useful, but it has limits worth knowing.

• Only what is separated is seen. An impurity that elutes under the main peak adds to it. Peak shape, a second detection wavelength[4] and a second, different separation method[2] help to reveal such hidden components.
• Absorbance is not a count of molecules. At 214 nm the signal comes mainly from peptide bonds, so a shorter impurity contributes less per molecule than a longer one.
• Salts and water are invisible. Cationic peptides purified with TFA end up as trifluoroacetate salts. In one study the octapeptide lanreotide, which has two positive charges, carried 4.5 TFA molecules per peptide molecule. Residual TFA can alter a peptide's behaviour, interfere with infrared spectroscopy and affect biological experiments. Re-running the peptide by reversed-phase HPLC with acetic acid removed only about two-thirds of it, while an ion-exchange resin or a deprotonation and reprotonation cycle replaced more than 95% of it with acetate[9]. A weighed sample of peptide therefore contains less peptide than its weight suggests, which matters when solutions are made up with the peptide calculator.

A purity figure is therefore always a statement about the method as much as about the peptide.

References

1. Al Musaimi O, Jaradat DMM (2024). Advances in therapeutic peptides separation and purification. Separations 11:233. DOI: 10.3390/separations11080233
2. Mant CT, Chen Y, Yan Z, Popa TV, Kovacs JM, Mills JB, Tripet BP, Hodges RS (2007). HPLC analysis and purification of peptides. In: Fields GB (ed.) Peptide Characterization and Application Protocols. Methods in Molecular Biology 386:3–55. Humana Press. DOI: 10.1007/978-1-59745-430-8_1
3. Aguilar MI (2004). Reversed-phase high-performance liquid chromatography. In: Aguilar MI (ed.) HPLC of Peptides and Proteins: Methods and Protocols. Methods in Molecular Biology 251:9–22. Humana Press. DOI: 10.1385/1-59259-742-4:9
4. Shaw C (1994). Peptide purification by reverse-phase HPLC. In: Walker JM (ed.) Basic Protein and Peptide Protocols. Methods in Molecular Biology 32:275–287. Humana Press. DOI: 10.1385/0-89603-268-X:275
5. Geng X, Regnier FE (1984). Retention model for proteins in reversed-phase liquid chromatography. Journal of Chromatography 296:15–30.
6. Krokhin OV (2006). Sequence-specific retention calculator. Algorithm for peptide retention prediction in ion-pair RP-HPLC: application to 300- and 100-Å pore size C18 sorbents. Analytical Chemistry 78:7785–7795. DOI: 10.1021/ac060777w
7. Alpert AJ (1990). Hydrophilic-interaction chromatography for the separation of peptides, nucleic acids and other polar compounds. Journal of Chromatography 499:177–196. DOI: 10.1016/S0021-9673(00)96972-3
8. Alpert AJ (2008). Electrostatic repulsion hydrophilic interaction chromatography for isocratic separation of charged solutes and selective isolation of phosphopeptides. Analytical Chemistry 80:62–76. DOI: 10.1021/ac070997p
9. Roux S, Zékri E, Rousseau B, Paternostre M, Cintrat JC, Fay N (2008). Elimination and exchange of trifluoroacetate counter-ion from cationic peptides: a critical evaluation of different approaches. Journal of Peptide Science 14:354–359. DOI: 10.1002/psc.951