Anal. Chem. 2001, 73, 4924-4936
Capillary Electrochromatography Analysis of Hormonal Cyclic and Linear Peptides Karin Walhagen and Klaus K. Unger
Institut fur Anorganische Chemie und Analytische Chemie, Johannes Gutenberg-Universita¨t, Duesbergweg 10-14, D-55128 Mainz, Germany Milton. T. W. Hearn*
Centre for Bioprocess Technology, Department of Biochemistry and Molecular Biology, Monash University, Clayton, Victoria 3168, Australia
The retention behavior of linear and cyclic peptides has been studied by capillary electrochromatography (CEC) with a variety of different n-alkyl silica reversed-phase sorbents and also with mixed-mode phases containing both strong cation-exchange (sulfonic acid) and n-alkyl groups bonded onto the silica surface, using eluents ranging from pH 2.0 to pH 5.0. Depending upon the amino acid sequence, electrochromatographic retention of the peptides was strongly affected by the composition of the eluent, its pH value, and the choice of sorbent packed into the capillaries. The dominant separation processes operating with these charged analytes could be modulated inter alia by the content of organic modifier, acetonitrile, in the eluent, with peptide resolution predominantly arising from electrophoretic migration processes at high acetonitrile content. As the concentration of acetonitrile was decreased, chromatographic retention processes became more pronounced. With the n-alkyl silica CEC columns used in this study, silanophilic interactions between the sorbents and the charged peptides could be suppressed by increasing the molarity of the buffer and by adjusting the pH of the eluent to lower values. On the other hand, electrostatic interactions between basic peptides and the surface of strong cationexchanger, mixed-mode materials can be suppressed at low pH values by using higher ionic strength conditions in the eluent. Different selectivity behavior was achieved with desmopressin and the other peptides with Spherisorb C18/SCX and Hypersil mixed-mode materials when an identical eluent composition of 60% (v/v) acetonitrile with 7.6 mM triethylammonium phosphate, pH 3.0, was used. These findings confirm that the surface charge density of the sorbent fulfills an important role in the modulation of peptide selectivity in CEC. These studies also confirm that the dependency of the logarithm of the CEC retention coefficients, i.e., log Kcec, of a peptide separated with n-octadecyl silica sorbents under CEC conditions, on the volume fraction, ψ, of the organic solvent modifier, acetonitrile, within the range of 0.20 e ψ e 0.60, can be approximated by a linear relationship. Moreover, these 4924 Analytical Chemistry, Vol. 73, No. 20, October 15, 2001
studies show that the selectivity differences of peptides separated by CEC with nonpolar sorbents in packed capillary systems can be discussed in terms of semiempirical dependencies that link peptide retention behavior with their molecular descriptor properties, e.g., their hydrophobicity, surface charge anisotropy, surface area, molecular mass and intrinsic charge, and thus to their corresponding linear free energy relationships. Capillary electrochromatography (CEC) is rapidly emerging as a powerful analytical separation technique, with several different instrumental formats and CEC capillary systems now available. As an advanced nanoseparation technology, CEC represents a hybrid of two well-known orthogonal separation methods, namely, high-performance capillary electrophoresis (HPCE), where analytes are separated according to their relative electrophoretic mobility, and high-performance liquid chromatography (HPLC), where resolution of the analytes is achieved through specific interactions with ligands immobilized onto the surface of a chromatographic sorbent. Currently, most CEC analyses are performed with “in-house”-built equipment or modified commercial HPCE instruments. Usually packed CEC capillaries of 50100-µm i.d. containing n-alky silica particles are used for analytical separations, but open-tubular (OT-CEC) fused-silica capillaries with the inner wall etched to increase the surface area and chemically modified to increase interactions with the solutes have also been employed. To date, CEC of neutral as well as charged biomolecules has however been somewhat disappointing, partly due to difficulties in controlling the electroosmotic flow, poor column reproducibility, and lack of a rigorous theoretical framework to interpret and predict retention and peak-broadening behavior. The current investigation specifically addresses these issues. With both packed and open tubular CEC systems, the eluent is transported through the column by electroosmotic flow (EOF), which originates from the migration of solvated ions when an electric field is applied across the solid-liquid interfaces within the capillary. In the case of packed CEC columns, the main * To whom correspondence should be sent. E-mail: med.monash.edu.au. Fax: Int + 61 + 3 + 9905 5882. 10.1021/ac0013352 CCC: $20.00
milton.hearn@
© 2001 American Chemical Society Published on Web 09/18/2001
contribution to the EOF is due to the packed particles themselves, with only a small effect arising from the capillary wall, due to the greater surface area of the packed particles compared to the inner capillary wall.1 With etched OT capillaries, a similar effect arises due to the greatly enhanced surface area (typically by a factor of ∼1000) that is achieved by the etching process. Under most operating pH values (i.e., at pH >3), the silanol groups on the capillary wall or particle surfaces are partially or fully ionized. When an electric field is applied over either type of capillary system, cations in the electrolyte are arranged as an ionic double layer due to this negatively charged surface. Cations closest to the sorbent surface are constrained in a rigid plane, the so-called Stern layer, and will not move, whereas solvated cations further away from the sorbent surface can freely migrate, generating a more diffuse ionic atmosphere. Solvated cations in this diffuse layer move toward the cathode and drag the inner bulk liquid along, with the EOF exhibiting a characteristic plug flow profile.2-4 In CEC with reversed-phase (RP) sorbents, neutral compounds are separated according to their ability to partition between the n-alkyl chains on the silica surface and the eluent. When n-alkyl silica sorbents are used with charged analytes, such as peptides, CEC resolution can, in principle, be achieved from the combined effects of electrophoretic migration and chromatographic retention.5,6 In this latter case, the intrinsic electrophoretic velocity of a peptide is superimposed with the electroosmotic velocity of the system. Effects due to hydrophobic as well as silanophilic interactions will arise due to the nature of the chemically modified surfaces of the sorbent in the packed capillaries or the modified surface of the capillary wall with OT-CEC systems. The EOF is very sensitive to changes in the charged density of the sorbent surface. In turn, the charged density depends on the physical characteristics of the buffer (electrolyte) used for the separation with respect to its pH, ionic strength, temperature, and content of organic modifier as well as the magnitude of the electric field strength applied to the packed or open tubular capillary.7 Thus, depending on the composition of the eluent, the CEC migration of peptides is anticipated to reflect the interplay of both separation processes, with the dominance of one contribution over the other achieved by, for example, changing the content of organic modifier in the eluent. Similarly, the participation of electrostatic interactions between a peptide and the sorbent can be decreased or suppressed by increasing the molarity of the electrolyte, which at the same time will affect the magnitude of the EOF and thus the selectivity. Counterbalancing these effects with buffers of low pH values, i.e., at pH 0) of peptides separated with n-alkyl silica sorbents under CEC conditions, and the volume fraction, ι, of the organic solvent modifier, can be approximated36 to a linear dependency, particular when the experimental regime favors charge suppression. Empirically, the dependency of log κcec on ψ then takes the form
log κcec ) A′′ - B′′ψ
(12)
where A′′ is the log κcec value when ψ f 0 and B′′ is a complex function related to the dependency of µa on ψ for a particular field strength. From eqs 10-12, the dependency of κcec on ψ, molecular mass, M, and calculated peptide charge, qcalc (calculated according to the Henderson-Hasselbach equation31 using pKa values empirically derived42 for the N-terminal, C-terminal and amino acid sidechain functional groups of the individual amino acid residues within the peptide sequence) can be expressed as (41) Eisenberg, D. Annu. Rev. Biochem. 1998, 53, 595-623. (42) Rickard, E. C.; Strohl, M. M.; Neilson, R. G. Anal. Biochem. 1991, 197, 197-209.
κcec )
[Akoe(b-S)ψ - Eµe] [Aebψ + Eµe]
[
)
]
F1qcalc/Fz a koe-X′′(M) ψAebψ - E 1/3 F2M + F3M2/3I1/2 (13) F1qcalc/Fz Aebψ + E F2M1/3 + F3M2/3I1/2
[
]
where X′′ is a constant derived from plots of S versus (M)j from the experimental data of the peptides when the separation is dominated by reversed-phase chromatographic processes; F1 ) X(RRs); F2 ) 6πγ(f/fo)(4πN/3υ)-1/3; γ is the surface tension of the eluent; (f/fo) is the frictional ratio; υ is the partial specific volume; F3 ) 6πγ(21/2Ne)(orRT)-1/3(f/fo)2(4πN/2υ)-1/3; and Fz is the charge proportionality factor (i.e., the ratio qcalc/qobs). By substitution of the appropriate terms, the dependency of κcec on peptide molecular mass, effective charge, temperature, and ionic strength over the operational range of ψ values takes the form
κcec ) [koe-X′′(M) ψAebψ - E[(X(RRs)qcalc/Fz)/ a
(6πγ(f/fo)(4πN/3υ)-1/3M1/3 + 6πγ(x2Ne)(orRT)-1/3(f/fo)2(4πN/2υ)-1/3M2/3I1/2)]]/ [Aebψ + E[(X(RRs)qcalc/Fz)/ (6πγ(f/fo)(4πN/3υ)-1/3M1/3 + 6πγ(x2Ne)(orRT)-1/3(f/fo)2(4πN/2υ)-1/3M2/3I1/2)]] (14) Since for small peptides the value of X′′ and a are ∼0.540,43 and assuming that υ and (f/fo) remain constant over the range of the organic solvent concentrations that are employed as the usual operational range, then for a defined ψ value (such as ψ ) 0.4), the dependency of κcec on M at a fixed temperature, T, and at an ionic strength, I, can be approximated to
[
X(RRs)qcalc/Fz Aebψ - E 1/3 ∃1M + ∃2(r)-1/3M2/3I1/2 qcalc/Fz Aebψ + E 1/3 ∃1M + ∃2(r)-1/3M2/3I1/2 (15)
koe-0.5(M)
κcec ≈
0.5ψ
where the constant terms ∃1 and ∃2 corresponds to F2 ) 6πγ(f/fo)(4πN/3υ)-1/3 and to (F3)()1/3 ) 6πγ(21/2Ne)(oRT)-1/3(f/fo)2(4πN/2υ)-1/3 respectively. Thus, within the operational range -1.0 e κcec e ∞, where the organic solvent content can in principle be varied between 0.01 e ψ e 0.99, the κcec values will incorporate retention contributions with silica-based reversed-phase sorbents due to peptide-nonpolar ligand interaction, peptide-solvent interactions, peptide-silanophilic interactions, ionization and charge-dependent effects, and mass and intrinsic hydrophobicity properties as well as effects arising from changes in the EOF that are mediated by the nature and concentration of the organic (43) Snyder, L. R.; Dolan, J. W.; Lommen, D. C. In High performance liquid chromatography of peptides and proteins; Mant, C. T., Hodges, R. S., Eds.; CRC Press: Boca Raton, FL, 1991; pp 725-733.
solvent.29,30 Since at low ionic strength values, i.e., I < 0.05, such as employed in the present investigations, X(RRs) ≈ 1.0, then eq 15 can be rewritten as
[
qcalc/Fz Aebψ - E 1/3 ∃1M + ∃2(r)-1/3M2/3I1/2 qcalc/Fz Aebψ + E 1/3 ∃1M + ∃2(r)-1/3M2/3I1/2 (16)
koe-0.5(M)
κcec ≈
0.5ψ
Several predictions can thus be made from eqs 10-16 about the CEC retention behavior of small peptides when such conditions prevail. First, for peptides separated with n-alkyl silica CEC sorbents over narrow ranges of organic solvent content, the plots of log κcec versus ψ are expected to follow near-linear dependencies provided the ionic strength of the CEC buffer is relatively low, i.e., I < 0.05. Second, with n-alkyl silica CEC sorbents, the magnitude of the log κcec term will be dependent on the charge, molecular weight, and surface area characteristics of the peptide and specifically dominated by the effective charge and intrinsic hydrophobicity. This dependency is expected, from linear free energy arguments, to be manifested (to a first approximation) as a linear relationship between log κcec and ([θ(ω)]mom(Mn)) over a range of ψ values. Third, for any set of peptides a buffer pH condition can always be selected so that the qobs value for one of peptide in the set approaches or becomes zero, and thus, the CEC migration of this peptide will be totally dominated by chromatographic retention effects. The migration characteristics of this selected peptide as the solvent content is changed thus enables differences in peptide ionization, buffer ionic strength, solventinduced polarizability changes, and changes in the effective hydrophobic moment to be ascertained. A codocil also arises from these considerations. Thus, for negatively charged peptides, i.e., the B-peptide, 12, at pH 4.8 (Table 1), in the CEC mode the electrophoretic migration will tend to counterbalance the chromatographic retention while with positively charged peptides these migration effects will synergistically reinforce each other. Monitoring the Performance and Lifetime of the CEC Columns. Because various investigators have previously reported difficulties with the reproducibility and reuse of packed and open tubular CEC capillaries, the performance of the Hypersil CEC n-octyl silica and n-octadecyl-packed capillaries used in these studies were monitored frequently with a set of three n-alkylbenzenes, namely, ethylbenzene, n-butylbenzene, and n-pentylbenzene.28-30,44 The test mixture was analyzed with an eluent composed of 25 mM Tris-HCl, pH 8.0/acetonitrile, 1:4 (v/v). The temperature of the cassette was set to 20 °C, and both the inlet and the outlet capillary ends were pressurized at 10 bar. The performance of the packed capillaries was tested with the n-alkylbenzenes before a peptide study was initiated and also between experiments, when the pH or the molarity of the buffer was changed, or if the CEC system was not used for several days to confirm the reproducibility of the retention behavior of the peptides from time to time. The detection window and the frits of the CEC columns are extremely delicate areas and therefore easily (44) Walhagen, K.; Unger, K. K.; Hearn, M. T. W. J. Chromatogr., A 2000, 894, 35-43.
Analytical Chemistry, Vol. 73, No. 20, October 15, 2001
4929
Table 1. Amino Acid Sequences, Molecular Weight, Mr, and Calculated pI Values for the Linear and the Cyclic Peptides 1-12 effective charge code
peptide name
abbrev
1 2 3 4 5 6 7
[Met5]-enkephalin [Leu5]-enkephalin angiotensin II triptorelin desmopressin
GY VYV Met-E Leu-E AII Tript Desm
8
carbetocin
9
sequencea
Mr
pI
pH 2.1
pH 4.8
H-Gly-Tyr-OH H-Val-Tyr-Val-OH H-Tyr-Gly-Gly-Phe-Met-OH H-Tyr-Gly-Gly-Phe-Leu-OH H-Asp-Arg-Val-Tyr-Ile-His-Pro-Phe-OH