No solid phase, no place to bind
When a peptide goes onto a bonded-silica column and less mass comes off than went on, no change of gradient recovers it. Counter-current chromatography attacks that failure mode by removing its cause: there is no adsorbent, no bonded ligand and no pore network anywhere in the flow path. Ito's step-by-step account of high-speed counter-current chromatography (HSCCC) conditions frames the technique on exactly that basis, describing it as free of irreversible adsorption and offering resolution comparable to column chromatography [1]. Both phases are liquids drawn from a single biphasic solvent system: one is held inside the column by a centrifugal force field while the other is pumped over it, and the concentration of a compound in the stationary phase divided by its concentration in the mobile phase — the partition coefficient — is what does the separating [2].
Be precise about what that buys. Removing the solid support removes one loss mechanism. It does not remove solubility limits, it does not remove interfacial behaviour, and it does not tell you that a separation exists. Everything downstream depends on a biphasic solvent system that has to be found before the instrument is switched on, and for peptides that search is the hard part.
Hydrodynamic and hydrostatic instruments
Two hardware families hold the liquid stationary phase by different means. Hydrostatic columns — centrifugal partition chromatographs, CPC — are built from disks of interconnected channels and use rotary seals. Hydrodynamic columns are bobbins of coiled open tubing with no rotary seals [3]. Support-free liquid–liquid chromatography of either kind is used mainly in phytochemistry, but both families have been applied to synthetic peptide purification [4].
Whichever family you are on, the parameter under scrutiny is the same: Sf, the fraction of column volume that remains filled with stationary phase while mobile phase is flowing. Retaining that liquid inside the column is described as a major challenge of the technique [3].
Stationary phase retention and the resolution equation
Sf is not a housekeeping number, because the CCC resolution equation depends on both column efficiency and the stationary phase retention ratio. Berthod and Faure worked the consequence through for hydrodynamic columns. A coiled tube retains more liquid stationary phase with a larger bore, but at constant column volume a smaller bore is longer and therefore generates more efficiency; an optimum bore size therefore exists, and its value depends on the solute's partition coefficient and the mobile phase flow rate. Their theoretical treatment also showed that larger bores allow dramatically reduced experiment durations: in reversed-phase CCC, hydrophobic solutes separated at high flow rate and reduced Sf can come off in retention times similar to those of polar solutes separated at smaller flow rates and much higher Sf [3]. Because Sf sits in the resolution equation, losing stationary phase mid-run costs resolving power, not only retention time.
Temperature is easy to leave uncontrolled, and it is not a neutral variable. Roehrer and Minceva varied column temperature between 25 and 40 °C on a non-aqueous n-hexane/ethyl acetate/acetonitrile system and an aqueous-organic n-heptane/ethyl acetate/methanol/water system, and analysed the effect on stationary phase retention, column efficiency and separation resolution. Because temperature changes the phase composition itself, a shift can even turn a biphasic system into a single-phase system; they concluded that appropriate temperature control of the column is essential to avoid destabilising temperature-sensitive systems, and used the stability and settling time of the biphasic systems as the indicators of whether temperature susceptibility has to be handled during method development [5].
Peptide work tends to land on the awkward systems. Knight reported that some of the solvent systems required for hydrophobic synthetic peptides have low interfacial tension, and identified two instruments — an eccentric multi-layer hybrid coil planet centrifuge and a spiral disk planet centrifuge — that were able to retain the stationary phase of those systems [6]. Aqueous polymer two-phase systems are harder again, and the instructive instrument comparison there was run on proteins rather than peptides: Shibusawa and Ito loaded 10–200 mg each of cytochrome c, myoglobin, ovalbumin and haemoglobin in an aqueous–aqueous two-phase polymer system and compared a cross-axis coil planet centrifuge with a horizontal flow-through coil planet centrifuge. The best results came from the cross-axis instrument at 750 rpm and a flow-rate of 2.0 ml/min using a multilayer coil of small helical diameter (β = 0.25–0.60), which resolved the four proteins in less than 5 h [7].
The partition coefficient window, and the shake flask that sets it
Since K is the stationary-to-mobile concentration ratio [2], a value near zero means the solute stays with the mobile phase and leaves near the solvent front, while a large value means it sits in the stationary phase and elutes late and dilute. Useful separations happen in a window between those extremes, and finding a solvent system that puts your compound there is the step that decides whether there is a separation at all.
Wang and colleagues call solvent system selection the stumbling block to wider application of HSCCC. They built five linear prediction models for the Arizona solvent system family, which the paper abbreviates HEMW, using fourteen compounds of different structures across five HPLC columns of different brands, and proposed two selection routes — one for a targeted single compound, one for multi-component separation of a complex sample — both operating on a partition coefficient window of 0.25 < K < 2.5. The predicted systems for five known compounds were then checked by HPLC analysis together with a shake-flask test [8]. The paper describes them only as fourteen compounds of different structures, and validates the method on five known compounds and two Chinese herbal extracts; nothing in it concerns peptides, so the fitted models carry no information about peptide partitioning. What transfers is the procedure: K is measured in a test tube before a column is committed. Ito's condition-selection sequence puts determination of K immediately after the choice of two-phase solvent system and before elution mode, flow rate and rotation speed are fixed [1].
Solvent systems that have actually worked on peptides
Peptides needed their own solvent systems from the start. Knight and co-workers found one that does not exhibit solvent carryover at high speed flow and centrifugation conditions: tert-butyl methyl ether, n-butanol, acetonitrile and aqueous trifluoroacetic acid, with the volume ratios and the acid percentage adjustable. A cross-axis coil planet centrifuge was used for various dipeptides; superior resolution was obtained with a multi-layer coil planet centrifuge, which separated six dipeptides and carried out the preparative purification of two synthetic peptides [9].
Hydrophobic sequences need harsher phases. Knight later reported that synthetic peptides with many aromatic, aliphatic and especially acidic amino acid residues are not very soluble and require strong solvents for useful partitioning. A chloroform–methanol–acidic solvent system fractionated neutral and basically charged 26-mers to provide high yields, and an insoluble 15-mer with five Trp residues and 60% overall hydrophobic amino acid content was purified in pyridine–acetic acid and in a basic tert-butyl methyl ether/n-butanol/acetonitrile system, with high recovery [6].
Protected synthetic intermediates are the limiting case, and the clearest reason to reach for a liquid stationary phase in the first place. Amarouche and colleagues start from the observation that protected synthetic peptide intermediates are often hydrophobic and not soluble in most common solvents, which makes them difficult to purify by the preparative reversed-phase HPLC usually used for industrial production. They developed two biphasic systems covering a wide range of polarity — heptane/tetrahydrofuran/acetonitrile/dimethylsulfoxide/water and heptane/methyl-tetrahydrofuran/N-methylpyrrolidone/water — and used them for the CPC purification of a 39-mer protected exenatide and an 8-mer protected bivalirudin intermediate, reporting phase compositions by 1H NMR alongside the viscosity, density and interfacial tension of each system [4].
What a shake-flask K does not tell you
Knight and co-workers determined a partition coefficient for a 16-residue myristylated peptide amide in the n-butanol–acetic acid–water (4:1:5) system, then ran chromatography in that system and in one modified with ethyl acetate. The peptide material showed anomalous elution behaviour, which they attributed to its surfactant properties. The fix was procedural rather than thermodynamic: loading the sample exclusively in the stationary phase gave satisfactory retention of the compound, and conditions using the upper phase as the mobile phase then separated the impurities [10].
Read that as a limit on what the pre-run measurement licenses. A partition ratio measured in a settled tube is a bulk equilibrium number, and it supports a prediction about where the solute sits at equilibrium. It does not support a prediction about what a surface-active peptide does at the interface inside a spinning coil, and for lipidated or strongly amphipathic sequences the two can come apart.
Loading, and the displacement modes
Because the column holds a volume of liquid rather than a surface, capacity is governed by how much peptide the two phases will dissolve. That is why the peptide CCC literature argues about solvent strength and solubility rather than about isotherm shape [4, 6].
Displacement modes push loading considerably further. pH-zone-refining CCC adds a retainer to the stationary phase and an eluter to the mobile phase at a given molar ratio, forming a sharp retainer rear border that moves through the column at a desired rate regardless of the composition of the two-phase solvent system. Set against conventional HSCCC, Ito lists the characteristics as follows: sample loading capacity is increased over 10 times, fractions are highly concentrated near saturation level, and yield is improved by increasing the sample size [11]. Applied to peptides, Ma and Ito separated acidic and basic oligopeptide derivatives containing up to three amino acids, and demonstrated the preparative ability of the method by separating multigram quantities on a standard separation column with a total capacity of 325 ml [12].
Ion exchange does something comparable for a longer sequence. Boudesocque and colleagues purified a synthetic 17-amino-acid peptide by ion exchange centrifugal partition chromatography using the biphasic methyl tert-butyl ether/acetonitrile/n-butanol/water (2:1:2:5, v/v) system in the descending mode, di(2-ethylhexyl)phosphoric acid as cation exchanger at an exchanger-to-peptide mole ratio of 100, and Ca2+ ions in aqueous solution as displacer. Critical impurities were efficiently eliminated and the peptide was recovered at 69% yield and 98% purity, with a productivity of 2.29 g per litre of stationary phase per hour [13].
Crude mixtures and salt-rich phases
Two properties make the technique attractive on crude material: nothing in the sample can foul a bed that does not exist, and the aqueous phase can be salt-rich, because it is a solvent rather than a mobile phase passing over a bonded surface — with the salt content set by the composition that keeps the system biphasic.
From plant material, HSCCC separated cyclic peptides from an extract of Vaccaria segetalis seeds with petroleum ether–ethyl acetate–methanol–water at an optimised volume ratio of 0.5:3.5:1:5, giving 38.0 mg of segetalin B and 28.5 mg of segetalin A from 190 mg of crude extract, at HPLC purities of 98.1% and 95.6% respectively. Note the workflow: that separation followed an initial clean-up step on silica gel [14]. From microbial material, Cheel and colleagues enriched Desmonostoc muscorum CCALA125 extracts by polymeric resin treatment, ran high-performance CCC to obtain three cyclic lipopeptides, and then repurified each by semi-preparative HPLC to reach 86%, 92% and 90% purity [15].
For salt tolerance the clearest published demonstration is on an adjacent analyte class rather than on peptides. Vega and colleagues isolated mycosporine-like amino acids from two red macroalgae and a marine lichen on a biphasic system of ethanol, acetonitrile, saturated ammonium sulphate solution and water (1:1:0.5:1), running eight separation cycles at 1 g of Pyropia columbina extract per cycle. The salt that made the system biphasic then had to come back out: the enriched fractions were desalted by precipitation with methanol and permeation on a Sephadex G-10 column [16]. Aqueous–aqueous polymer systems shift the difficulty elsewhere: in the protein comparison above, which instrument geometry and which coil helical diameter were used decided how well the separation ran [7].
Where it sits next to preparative reversed-phase
Preparative reversed-phase HPLC is the route normally used for industrial production of synthetic peptides, and counter-current chromatography is not a general replacement for it [4]. The case for a liquid stationary phase is specific: a peptide whose recovery from a solid phase is poor; a protected intermediate that will not dissolve in what a reversed-phase gradient requires [4]; an ionisable derivative where a displacement mode buys an order of magnitude in loading [11]; a crude mixture you would rather not put on a bed at all.
The published workflows are almost all hybrids, and that is worth reading plainly rather than around. Silica clean-up before the CCC step [14]; resin enrichment before and semi-preparative HPLC after [15]; desalting after a salt-rich system [16]. What counter-current chromatography contributes is material, in quantity, from a column that has no surface capable of keeping any of it. What it does not contribute is the purity verdict: in the natural-product separations above, purity was assigned by HPLC and identity by high-resolution mass spectrometry and NMR, not by the preparative run itself [14, 15]. Plan that analysis alongside the separation rather than after it.