Anal. Chem. 1996, 68, 2818-2825
Retention Behavior of Amino Acids and Peptides on Protoporphyrin-Silica Stationary Phases with Varying Metal Ion Centers Jie Xiao and Mark E. Meyerhoff*
Department of Chemistry, The University of Michigan, Ann Arbor, Michigan 48109
Various metalloprotoporphyrins (MProP) covalently linked to silica supports are examined as novel immobilized metal ion affinity chromatography (IMAC) stationary phases for separations of amino acids/peptides. Under reversed-phase HPLC conditions, the MProP-silicas exhibit high affinity toward L-histidine via metal-nitrogen axial ligation interactions, with an increasing degree of histidine retention highly dependent on the specific metal ion (M) in the center of the protoporphyrin (ProP) structure: Fe(III) > Ni(II) > Cu(II) > Zn (II) ≈ Cd(II). Aromatic amino acids (i.e., L-tryptophan and L-phenylalanine) are also retained on MProP columns through π-π interactions with the immobilized porphyrins, with the greatest affinity for L-tryptophan observed on CuProPsilica columns. Peptides rich in L-histidine and L-tryptophan residues are selectively retained on most of the MProP-silica phases examined; however, the addition of an organic modifier and/or lowering the pH of the mobile phase can be used independently to attenuate the π-π and metal ion-nitrogen ligation interactions, respectively. Reproducible separations of His-Phe and tryptophan releasing hormone are achieved on a FeProP-silica column even after extensive washing with 50 mM EDTA, demonstrating a fundamental advantage of the new MProPsilica over existing IMAC stationary phases, in which the metal ion is anchored weakly to the support via immobilized iminodiacetate and related ligands. The continued growth in the development of therapeutics based on naturally occurring and recombinant proteins/peptides as well as the increasing characterization of protein structures via tryptic mapping methods has created a need for better resolution in the separation of complex protein and peptide mixtures.1-8 Liquid chromatography continues to play a central role in these areas, especially in the purification, quality assurance, and pharmacokinetic studies of recombinant species. Novel stationary phases with high selectivities and long lifetimes can dramatically reduce the process time (i.e., single-step separation) (1) Regnier, F. E. Science 1987, 238, 319-323. (2) Welinder, B. S.; Kornfelt, T.; Sørensen, H. H. Anal. Chem. 1995, 67, 39A43A. (3) Compton, B. J.; Kreilgaard, L. Anal. Chem. 1995, 67, 1175A-1180A. (4) Garnick, R. L.; Solli, N. J.; Papa, P. A. Anal. Chem. 1988, 60, 2546-2557. (5) Narayanan, S. R. J. Chromatogr. 1994, 658, 237-258. (6) Knight, P. Bio/Technology. 1990, 7, 243-249. (7) Jones, K. Chromatographia 1991, 32, 469-480. (8) Scho¨neich, C.; Hu ¨ hmer, A. F. R.; Rabel, S. R.; Stobaugh, J. F.; Jois, S. D. S.; Larive, C. K.; Siahaan, T. J.; Squier, T. C.; Bigelow, D. J.; Williams, T. D. Anal. Chem. 1995, 67, 155R-181R.
2818 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
and thereby reduce the overall production cost of given therapeutic peptides/proteins. In addition, new stationary phases can provide orthogonality to conventional stationary phases (i.e., C18 and CN), which is required for two-dimensional chromatographic purifications of peptides/proteins from complex matrices.9-12 One of the more interesting chromatographic techniques for peptide/protein separations is immobilized metal ion affinity chromatography (IMAC),13-22 introduced by Porath et al. in 1975.23 This method relies on the interaction of certain amino acids (e.g., L-histidine, L-cysteine, and L-tryptophan) within the peptide/protein structure with certain metal cations (i.e., Ni(II), Zn(II), Cu(II)) immobilized via ligands (e.g., iminodiacetate (IDA)) covalently tethered to the stationary phase. For example, it has been demonstrated that a single-step purification of a recombinant protein (e.g., D-xylose isomerase) from the lysed Escherichia coli. cell media can be achieved on a Cu(II)-loaded IMAC column.24 Because of its simplicity and unique selectivity, IMAC has attracted considerable interest from both academic and industrial researchers.22 However, metal ion leaching from the current immobilized chelators employed for IMAC has kept IMAC from becoming a commercially attractive technology for routine biopolymer purification and analytical quality control.16,25-27 Indeed, such leaching can lead to unacceptable contamination of the “purified” peptide/protein with metal ions. Further, continued metal ion leaching (enhanced by the presence of common peptide/protein stabilizers such as EDTA or certain amine buffers (e.g., Tris17) (9) Giddings, J. C. Anal. Chem. 1984, 56, 1258A-1270A. (10) Giddings, J. C. J. Chromatogr. Sci. 1990, 50, 1-27. (11) Bushy, M. M.; Jorgenson, J. W. Anal. Chem. 1990, 62, 161-167. (12) Berkowitz, S. A. Adv. Chromatogr. 1989, 29, 175-192. (13) (a)Chicz, R. M.; Regnier, F. E. Anal. Chem. 1989, 61, 1742. (b)Arnold, F. H. Biotechnology 1991, 9, 151-156. (c) Todd, R. J.; Johnson, R. D.; Arnold, F. H. J. Chromatogr. 1994, 662, 13-26. (14) Yip, T. T.; Nakagawa, Y.; Porath, J. Anal. Biochem. 1989, 183, 159. (15) Belew, M.; Yip, T. T.; Andersson, L.; Ehrnstro¨m, E. Anal. Biochem. 1988, 168, 75. (16) Belew, M.; Porath, J. J. Chromatogr. 1990, 516, 333-354. (17) Belew, M.; Yip, T. T.; Andersson, L.; Ehrnstro¨m, E. Anal. Biochem. 1987, 164, 457-465. (18) Figueroa, A.; Corradini, C.; Feibush, B.; Karger, B. L. J. Chromatogr. 1986, 371, 335. (19) El Rassi, Z.; Horva´th, C. J. Chromatogr. 1986, 359, 241. (20) (a)Anspach, F. B. J. Chromatogr. 1994, 672, 35-49. (b) Anspach, F. B. J. Chromatogr. 1994, 676, 249-266. (21) Belew, M.; Yip, T. T.; Andersson, L.; Porath, J. J. Chromatogr. 1987, 403, 197. (22) Cramer, S. M.; Jayaraman, G. Curr. Opin. Biotechnol. 1993, 4, 217. (23) Porath, J.; Carlsson, J.; Olsson, I.; Belfrage, G. Nature 1975, 258, 598. (24) Mrabet, N. T. Biochemistry 1992, 31, 2690-2702. (25) Belew, M.; Yip, T. T.; Andersson, L.; Ehrnstro ¨m, R. Anal. Biochem. 1987, 164, 457. (26) Muszynska, G.; Zhao, Y. J.; Porath, J. J. Inorg. Biochem. 1986, 26, 127. (27) Andersson, L.; Sulkowski, E.; Porath, J. J. Chromatogr. 1987, 421, 141. S0003-2700(96)00052-2 CCC: $12.00
© 1996 American Chemical Society
within the injected sample preparation) results in irreproducible retention behavior as the amount of stationary phase (metal ion) decreases after each subsequent sample injection. This limits the potential applications of IMAC for analytical purposes. Although time-consuming preconditioning procedures can be used to recharge the columns with metal ions to achieve more reproducible results, such efforts make the technique less attractive in comparison to classical reversed-phase LC on C18 or other columns. Further, in order to suppress interference from ionic interactions with unmetalated chelator sites (IDA), a very high salt concentration (i.e., 1 M NaCl) in the mobile phase is often required, making the method even more problematic.28 In recent years, several new chelators (e.g., nitrilotriacetic acid (NTA)34 and tris(carboxymethyl)ethelene diamine (TED)35 ) have been developed for IMAC in an effort to overcome the current limitations. Although IMAC stationary phases based on soft gel matrices,13-16,22-27 e.g., agarose (Sepharose) or cross-linked dextran (Sephadex), have been examined in most of these studies, several researchers have designed IMAC columns based on silica supports in an effort to develop high-performance IMAC systems.18-20 Recently, we have found that porphyrin-silica stationary phases (metalated and unmetalated) offer unique selectivities for the separation of aromatic carboxylates and sulfonates, polycyclic aromatic hydrocarbons (PAHs), and fullerenes.29-32 Further, in preliminary studies,29 it was demonstrated that Zn(II) tetraphenylporphyrin-silica selectively retains imidazoles and L-histidine via strong metal ion-nitrogen ligation reactions. This observation suggests that metalloporphyrin-based stationary phases may be useful in IMAC type separations of peptides/proteins. Toward this end, we now examine the retention behavior, under HPLC conditions, of amino acids and small peptides on columns packed with various metalloprotoporphyrin (MProP)-silicas (where M ) Fe(III), Cu(II), Zn(II), Cd(II), and Ni(II)). It will be shown that (i) metalloporphyrin-silica phases exhibit high affinity toward L-histidine, with Fe(III)ProP-silica yielding the greatest specific retention of L-histidine; (ii) both metalated and unmetalated ProPsilica exhibit strong π-π interactions with L-tryptophan and L-phenylalanine, with Cu(II)ProP-silica yielding the strongest interaction with L-tryptophan; (iii) through the use of organic modifiers and/or changes in the mobile phase pH, either retention mechanism can be attenuated in order to optimize a given polypeptide separation; and (iv) ProP-silica complexes with given metal ions are strong enough to withstand extensive washings with EDTA without change in solute retention times. EXPERIMENTAL SECTION Reagents. Protoporphyrin IX was purchased from Midcentury Chemicals (Posen, IL). Hyperprep 8-µm silica gel (Shandon HPLC) with a surface area of 90 m2/g and a pore size of 300 Å was obtained from Keystone Scientific (Bellefonte, PA). Aminopropyltriethoxylsilane was obtained from Hu¨ls (Piscataway, NJ). (28) Wong, J. W.; Albright, R. L.; Wang, N. L. Sep. Purif. Methods 1991, 20, 49-106. (29) Kibbey, C. E.; Meyerhoff, M. E. Anal. Chem. 1993, 65, 2189. (30) Kibbey, C. E.; Meyerhoff, M. E. J. Chromatogr. 1993, 641, 49. (31) Xiao, J.; Meyerhoff, M. E. J. Chromatgr. 1995, 715, 19-29. (32) Xiao, J.; Savina, M. R.; Martin, G. B.; Francis, A. H.; Meyerhoff, M. E. J. Am. Chem. Soc. 1994, 116, 9341-9342. (33) Nomura, A.; Yamada, J.; Tsunoda, K. Anal. Sci. 1987, 3, 209-212. (34) Hochuli, E.; Dobeli, H.; Schacher, A. J. Chromatogr. 1987, 411, 177. (35) Porath, J.; Olin, B. Biochemistry 1983, 22, 1621.
Anhydrous dimethylformamide, used for the immobilization of protoporphyrin IX, and HPLC grade dimethylformamide, used in the metalation of protoporphyrin-silica, were products of Aldrich (Milwaukee, WI). HPLC grade acetonitrile was from Mallincrodt (Paris, KY). 1,1′-Carbonyldiimidazole, used to link the protoporphyrin to the solid phase, was from Fluka (Ronkonkoma, NY). The metal salts, FeCl3‚6H2O from Fisher Scientific (Fair Lawn, NJ), CuCl2‚2H2O and NiCl2‚6H2O from Mallinckrodt (Paris, KY), CdCl2‚2H2O from J. T. Baker Chemical (Phillipsburg, NJ), and ZnCl2 from Aldrich (Milwaukee, WI) were used. All amino acids and peptides examined in this work were purchased from Sigma (St. Louis, MO), except His-Gly-Gly-Phe-Gly, which was synthesized at the Peptide Synthesis Core Facility in the Medical School at the University of Michigan. Mobile phase buffers were prepared with analytical grade NaH2PO4‚H2O (Mallinckrodt, Paris, KY) and doubly deionized water. The pH of the phosphate buffer was adjusted to the desired value by adding concentrated NaOH or HCl solution. All mobile phases were premixed and then filtered through a 0.45-µm polypropylene membrane from Alltech Associates (Deerfield, IL). Preparation of the Protoporphyrin-Silica Stationary Phase. Silica gel was dried thoroughly in a vacuum oven (1 Torr) at 80 °C overnight. Aminopropyl-silica was then synthesized by reaction of the dried silica and (aminopropyl)triethoxysilane in toluene.33 Protoporphyrin IX was covalently attached onto the aminopropylderivatized silica gel through formation of an amide bond, as shown in Figure 1. Five grams of ProP was dissolved in 50 mL of anhydrous DMF. Two- and four-tenths grams (2.2 equiv) of 1,1′-carbonyldiimidazole was then used to activate both carboxylic acid groups of protoporphyrin IX. This solution was sonicated for about 10 min and then shaken for an additional 20 min. When outgassing ceased, 10 g of dried aminopropyl-silica was added, and the entire solution was sonicated for an additional 30 min. Using an electrical shaker, the reaction slurry was then gently shaken for 12 h in the dark. A fused-glass funnel was employed to filter the silica gel. The ProP-silica was thoroughly washed with DMF and acetone until the filtrate was clear and colorless. Physically adsorbed porphyrin was washed from the silica surface with a solution of 10% (v/v) glacial acetic acid in acetone, followed by thorough rinsing with pure acetone. The reddish dark silica gel was then allowed to dry overnight at 110 °C. To deactivate the residual amine sites on the surface of silica gel (end-capping), 10 g of the porphyrin-derivatized silica gel was refluxed in 50 mL of acetic anhydride for 2 h. After thorough washing with acetone and deionized water, the dark ProP-silica gel product was dried overnight at 110 °C. Metalation of the ProP-Silica. The protoporphyrin-silica stationary phases were metalated with Zn(II), Cu(II), Fe(III), Ni(II), and Cd(II) by the DMF method in 2-g batches as reported previously.29 In short, to prepare Zn(II)ProP-silica, ∼1.5 g of ZnCl2 was dissolved in 50 mL of hot DMF. This was followed by the addition of 2 g of H2ProP-silica gel, and the slurry was then refluxed for about 4 h. After the slurry cooled to room temperature, the silica gel was filtered and washed sequentially with DMF, acetone, water, 20% (v/v) glacial acetic acid in acetone, 20% (v/v) glacial acetic acid in water, water, and acetone until the filtrate was colorless. To fully deactivate the remaining amine groups, each MProP-silica phase was reacted with acetic anhydride as described previously.29 The resulting darkened silica stationary phase was then dried overnight at 110 °C. Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
2819
Figure 1. Reaction sequence used to prepare the immobilized protoporphyrin-silica stationary phases in these studies.
Table 1. Elemental Analysis and Surface Coverages of Bonded Silicasa
bonded silica aminopropyl-silica (a)b H2ProP-silica Zn(II)ProP-Silica Cd(II)ProP-silica Cu(II)ProP-silica aminopropyl-silica (b)b Fe(III)ProP-silicac Ni(II)ProP-silica
elemental analysis (%) carbon nitrogen 2.15 12.06 9.85 9.88 5.94 1.76 9.80 15.86
0.68 2.10 1.87 1.86 1.87 0.84 1.85 2.48
surface coverage (µmol/m2) 5.63 3.29 2.73 2.72 2.72 7.03 2.25 3.98
a Pore size 300 Å; specific surface area 90 m2/g. b Two separate lots of aminopropyl silica (a and b) used to prepare various ProP-silica phases. c Longer metalation reaction time was used for Fe(III) ProPsilica (8 h) compared to those for all other metalloprotoporphyrin phases reported (4 h).
UV-visible spectroscopy was used to characterize each preparation of MProP-silica. About 100 mg of each MProP-silica was treated with a solution of 50% NaOH solution and methanol (1:1). After 2 h, the slurry was filtered through a microfilter, and the UV-visible spectrum of the filtrate was obtained using a Shimadzu UV160 spectrophotometer. Column Packing. After being sieved through a 20-µm screen, ∼1.4-g amounts of each of six stationary phases (ProP-silica, Zn(II)ProP-silica, Fe(III)ProP-silica, Ni(II)ProP-silica, Cd(II)ProPsilica, and Cu(II)ProP-silica), prepared as described above, were suspended in a solution of 20 mL of methanol containing 10% (v/ v) glycerol. The slurry was thoroughly sonicated for 30 min and then transferred into the 20-mL slurry chamber of an Alltech highpressure liquid chromatography column packing system. The slurry was packed into 100-mm × 4.6-mm-i.d. stainless steel columns (Alltech Associates, Deerfield, IL) by the down-fill slurry method, using 100% methanol under a packing pressure of 6000 psi. After a total of 200 mL of solvent had eluted from the column, the column was removed from the packing instrument, and appropriate end fittings were attached. Instrumentation. The columns packed with the various MProP-silicas were evaluated using an HPLC system consisting 2820
Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
of a Spectra Physics (San Jose, CA) SP 8700 solvent delivery system, a Spectra Physics SP 4290 computing integrator, a Kratos (Ramsey, NJ) Spectroflow 773 variable-wavelength UV-VIS detector, and a Rheodyne (Cotati, CA) Model 7010 injection valve equipped with a 20-µL sample loop. Elemental analysis of the various ProP-silica gels was carried out by a service group within the Department of Chemistry. A Shimadzu UV160 spectrophotometer (Columbia, MD), controlled by PC 160 plus Personal Spectroscopy software (Shimadzu) on a IBM 486 computer, was used to record the UV-visible spectra of various protoporphyrin solutions. RESULTS AND DISCUSSION In previous work,29-32 we have shown that the covalently bound tetraphenylporphyrin (TPP)-silica stationary phases exhibit interesting shape selectivity in the HPLC separation of PAHs and fullerenes.30,31 Incorporation of various metal ions (in the 3+ or 4+ oxidation state) in the center of the immobilized TPP structure has also enabled such phases to be useful in anion exchange chromatography.29 In contemplating the use of metalloporphyrinsilicas as stationary phases for peptide separations, as proposed herein, it was envisioned that the four phenyl rings of TPP would render this stationary phase too hydrophobic for practical peptide separations; hence, a stationary phase consisting of immobilized protoporphyrin IX on silica was prepared to pursue the present studies. As shown in Figure 1, protoporphyrin IX possesses only four methyl and two ethylene groups, yielding a somewhat less hydrophobic phase (compared to the immobilized TPP), and more importantly, less steric hindrance for anticipated metal-nitrogen ligation interactions with L-histidine and other amino acids. Further, using the method outlined in Figure 1, it has been found that protoporphyrin IX can be immobilized onto aminopropyl-silica gel at much higher surface coverages (see Table 1) than reported previously29 for TPP (i.e., 0.2-0.4 µmol/m2). The surface coverages of porphyrin bonded to silica reported in Table 1 were calculated according to the increase in nitrogen content of each ProP-silica preparation (as determined by elemental analysis) after each step of the immobilization sequence, in accordance with the method described in ref 29. It should be noted that surface
coverages for the metalated ProP-silicas tend to be somewhat less than those of the corresponding unmetalated phases. This appears to be due to the generation of hydrochloric acid at the surface of the silica during the metalation procedure, creating a local environment that can hydrolyze either the amide bond linking ProP to the aminopropyl silica or the Si-O-Si linkages coupling the aminopropyl groups to the solid support. In addition, since both carboxylic acid groups of protoporphyrin IX are activated (see Experimental Section), it is possible that two products (A and B in Figure 1) are formed during the immobilization reaction. The free carboxyl group on A could potentially interfere with amino acid/peptide separations via electrostatic interactions. However, subsequent studies of the retention behavior of simple amino acids on the various ProP-silica stationary phases (see below) indicate that such columns display relatively little cation exchange behavior, suggesting that B is likely the major form of the immobilized porphyrin structure. Metalation of the ProP-silica phase was confirmed by changes of the Soret and Q bands in the UV-visible spectrum of the porphyrin before and after metalation.36 Such experiments were conducted by hydrolyzing the amide bond linking ProP to the aminopropyl-silica support in MeOH/NaOH, as previously described for TPP-silica phases,29 and comparing the λmax values for the resulting supernatants to those of authentic standards of the various MProP species. In separate experiments, concentrated H2SO4 was used instead of MeOH/NaOH to cleave ProP from the silica. After 2 h of equilibration, the UV-visible spectra of the resulting supernatant solution for each porphyrin-silica yielded the identical UV-visible spectra corresponding to protonated protoporphyrin IX (H2ProP),36 indicating that metalloporphyrins were cleaved and simultaneously demetalated. These experiments further confirmed that metalation of the ProP-silica stationary phase does indeed take place when the reflux method described in the Experimental Section is used. Columns (10 cm) packed with the various MProP-silicas generally exhibited only modest efficiencies (e.g., ∼2000 plates/column), as determined using CS2 as an unretained solute. While these efficiencies are poorer than those of typical commercial reversed-phase HPLC columns, they are similar to those of IMAC-silica phases reported previously in the literature18,19 and thus were viewed as sufficient to assess the general chromatographic behavior and novel chemical selectivities of these new stationary phases. Retention of Amino Acids on MProP-Silica Phases. A test group of nine amino acids was selected to be representative of the various classes of the 20 natural amino acids: i.e., L-glycine (Gly) and L-leucine (Leu) as hydrophobic amino acids; L-glutamate (Glu), L-lysine (Lys), L-serine (Ser), and L-cysteine (Cys) as charged/polar amino acids, and L-histidine (His), L-phenylalanine (Phe), and L-tryptophan (Trp) as aromatic amino acids. The separation of these test solutes on columns packed with the various ProP-silicas was monitored at 214 nm. Figure 2a summarizes the absolute capacity factors (k′) measured on each of the six columns for each test amino acid using a 50 mM phosphate buffer, pH 7.0, as the mobile phase. Figure 2b shows the same data reported as the column selectivity (R) for each solute relative to Leu. By normalizing the selectivity with respect to a single test solute, variations in capacity factors due to differences in (36) Falk, J. E. Porphyrins and Metalloporphyrins: Their General, Physical and Coordination Chemistry and Laboratory Methods; Elservier: Amsterdam, 1964; Chapter 6.
Figure 2. (a) Capacity factors (k′) and (b) selectivity factors (R) over Leu of model amino acids on the H2ProP (0), FeProP (]), CuProP (O), ZnProP (4), NiProP (9), and CdProP ([) silica stationary phases: (I) all amino acids and (II) without tryptophan. Mobile phase, 100% 50 mM phosphate buffer, pH 7.0; flow rate, 1 mL/min; temperature, ambient; detection, UV at 214 nm (0.100 AUFS) (capacity/selectivity factors reported are averages of three measurements, with typical RSD of His ≈ Lys > Leu > Glu on H2ProP, ZnProP, and CuProP columns and Trp . His > Phe > Lys > Leu > Glu on FeProP column) differ from those found on conventional reversed-phase HPLC columns. For example, it had been reported that on phenyl columns,38 these same six solutes elute in the following order: His > Phe > Trp > Leu ≈ Lys > Glu, whereas on an ODS (C18) column38 the order is Trp > Phe > His > Lys > Leu > Glu, and on an IMAC column16 loaded with Cu(II) the order is His > Trp > Phe > Leu ≈ Lys > Glu. More detailed studies on the effect of organic modifier (acetonitrile) in the mobile phase (pH 7.0 phosphate buffer) on the retention of Trp, His, and Phe on the CuProP and FeProP (38) Lundanes E.; Greibrokk, T. J. Chromatogr. 1978, 149, 241-254. (39) Molna´r, L.; Horva´th, C. J. Chromatogr. 1977, 1492, 623-640.
Figure 3. Separation of six amino acids on (a) H2ProP, (b) ZnProP, (c) CuProP, and (d) FeProP-silica stationary phases. Solutes: (1) Glu, (2) Leu, (3) Lys, (4) His, (5) Phe and (6) Trp. Mobile phase, (A) 50 mM phosphate buffer, pH 7.0, and (B) acetonitrile; gradient, 0% B from 0 to 2 min, 0-10% B from 2 to 7 min, and 10-40% B from 7 to 20 min; flow rate, 1 mL/min; temperature, ambient; detection, UV at 214 nm (0.100 AUFS).
phases are shown in Figure 4. Increasing the acetonitrile content dramatically reduces the retention time of amino acids on both columns. Indeed, the retention of Trp and His on the CuProP, ZnProP, and H2ProP columns, which originates primarily from π-π interactions, decreases to nearly zero when 25% acetonitrile is added to the mobile phase. However, the capacity factors of Trp and His on the FeProP-silica phase are between 4 and 5 under the same conditions, indicating that the metal ligation interaction still provides some degree of chemical selectivity in the presence of the organic modifier. Based on these observations, it appears that the metal ligation interaction is more pronounced on the FeProP-silica phase compared to all the other ProP-silica phases examined. Lowering the pH of the mobile phase (from pH 7.0 to 2.5) dramatically reduces the retention of His on the FeProP-silica phase (from k′ ) 14.3 to 0). At low pH, the imidazole nitrogen is protonated, eliminating retention via a metal ligation reaction with the nitrogen. Similar effects are observed for Trp and Lys. Since organic modifier dramatically decreases the π-π interaction and the low pH weakens the metal ligation interactions, a simple method exists to tune these two retention mechanism to the desirable combination by selecting appropriate pH and organic modifier content of the mobile phase (see below). Retention of Small Peptides on Metalloporphyrin Stationary Phases. Since preliminary studies with amino acids dem-
Figure 4. Effect of acetonitrile content in the mobile phase on capacity factors of Trp (O), His (4), and Phe (0). Column: (a) CuProPand (b) FeProP-silica. Mobile phase, 50 mM phosphate buffer, pH 7.0, with given percent of acetonitrile; flow rate, 1 mL/min; temperature, ambient; detection, UV at 214 nm (0.100 AUFS).
onstrated that acetonitrile can significantly reduce amino acid retention (see above), a mobile phase (pH 7.0) containing 25% acetonitrile was used to elute all test peptides in a reasonable time while still allowing a practical evaluation of the different retention behaviors of the peptides on the various ProP-silica columns. The capacity factors for a group of 10 test peptides on H2ProP-, CuProP-, ZnProP-, and FeProP-silica columns are shown in Figure 5. For the most part, the results of these studies are consistent with the observed retention of individual amino acids on the same stationary phases (see above). For example, Cys-Gly elutes in the dead volume on all the porphyrin columns, while all peptides containing His and Trp are preferentially retained on the FeProPsilica column, presumably through metal ligation interactions. Other porphyrin columns exhibit only minimal retention of these peptides, except for Trp-Trp on the CuProP column. Naturally, even with 25% acetonitrile in the mobile phase, π-π interactions still contribute to some extent to the retention of the test peptides, and this explains why peptides containing both His and Phe are retained longer than those with His alone. The Trp-Trp dipeptide is retained the longest on the FeProP- and CuProP-silica phases of any of the peptides tested. Taken together, these results suggest that there is a enhanced cumulative binding affinity for small peptides containing multiple amino acids that individually exhibit strong interactions with a given MProP-silica phase. Since the metal ligation interaction in aqueous solution is pH sensitive, the retention of the small peptides can be modulated by changing the pH of the mobile phase. When the pH of the Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
2823
Figure 5. Capacity factors of small peptides on H2ProP (0), NiProP (4), CuProP (O), and FeProPCl (]) silica stationary phases. Mobile phase, 75% 50 mM phosphate buffer, pH 7.0, and 25% acetonitrile; flow rate, 1 mL/min; temperature, ambient; detection, UV at 214 nm (0.100 AUFS); (capacity factors reported are averages of three measurements with typical RSD of
