Synthesis and Characterization of New Zirconia-Based Polymeric

Anal. Chem. 1998, 70, 1934-1942

Synthesis and Characterization of New Zirconia-Based Polymeric Cation-Exchange Stationary Phases for High-Performance Liquid Chromatography of Proteins Yue Hu and Peter W. Carr*

Department of Chemistry, University of Minnesota, Kolthoff and Smith Halls, 207 Pleasant Street SE, Minneapolis, Minnesota 55455

Ion-exchange chromatography is a major method used for large-scale protein separations. New zirconia-based polymeric cation-exchange HPLC stationary phases have been developed for protein separations. Two routes were employed for the synthesis. In one method, polyethyleneimine (PEI) was adsorbed onto porous zirconia particles and cross-linked with 1,4-butanediol diglycidyl ether (BUDGE). Succinic anhydride was then reacted with the remaining primary and secondary amine groups on PEI to afford anionic functionalities. The second method utilizes poly(acrylic acid) anhydride as both the crosslinker and the stationary phase. The resulting stationary phases act to separate proteins by a weak cation-exchange mechanism with a slight contribution to retention from hydrophobic interactions. In the presence of 20 mM phosphate buffer, Lewis acid/base interactions between the zirconia support and the proteins, which can significantly broaden the peaks, are sufficiently suppressed. The effects of ionic strength, mobile phase pH, and salt type are discussed. Protein mass recovery and loading capacity for protein separations on these phases have been evaluated. These weak cation-exchange stationary phases exhibit good stability under normal separation conditions for months and are stable in alkaline solution up to pH 10. In contrast to zirconia supports modified with small anionic species, these new phases have no limitation on the type of salt used as the eluent, and they exhibit unique selectivities. Therefore, they offer interesting alternatives for protein separations. To our knowledge, this work represents the first successful example of protein separations using porous zirconia-based polymeric phases under normal chromatographic conditions, which will definitely help make zirconia-based supports more useful for bioseparation. Ion exchange is the most common mode of chromatography for separating proteins, including antibodies and other large biomolecules.1-4 It is also frequently used in the analysis of * To whom correspondence should be addressed ([email protected]). (1) Regnier, F. E. Science 1983, 222, 245-252. (2) Thompson, J. A. Biochromatography 1986, 1, 16-20.

1934 Analytical Chemistry, Vol. 70, No. 9, May 1, 1998

drugs.5,6 However, because of the generally lower separation efficiency of ion exchange, it is gradually being replaced by reversed-phase chromatography.7,8 Recently, highly efficient chromatographic separations of basic drugs have been achieved using ion-exchange phases, thereby attracting renewed interest to this area.5,9 Currently, the commercially available ion-exchange stationary phases include silica-based silane bonded phases, polymeric phases, and inorganic substrates coated with polymers.10 Silica-based silane bonded phases are not stable in alkaline solutions, and silica itself is slightly soluble, even in neutral (pH 7) solutions, especially in the presence of phosphate, a common buffer used in biochemistry.11,12 Although great progress has been made in the development of highly cross-linked polymeric materials, such as styrene/divinylbenzene copolymer, polymethacrylate, and polyvinyl resins, they generally have low mechanical stability and tend to swell in solutions containing organic solvents and at high ionic strength.13 In the case of inorganic substrate-based polymeric phases, polymer coatings are deposited in the pores and only to some extent on the surface of the inorganic sorbent. Polymer coatings are “immobilized” by cross-linking with or without bonding to the inorganic sorbent. Such a composite phase can combine the rich chemistry of polymer coatings with the mechanical stability of inorganic sorbents and has seen a considerable increase in popularity in the past few years.14-17 This approach is especially suitable for synthesizing stationary phases (3) Zon, G.; Thompson, J. A. Biochromatography 1986, 1, 22-31. (4) Mhatre, R.; Nashabeh, W.; Schmalzing, D.; Yao, X.; Fuchs, M.; Whitney, D. J. Chromatogr. A 1995, 707, 225-231. (5) Croes, K.; McCarthy, P. T.; Flanagan, R. J. J. Chromatogr. A 1995, 693, 289-306. (6) Knox, J. H.; Jurand, J. J. Chromatogr. 1973, 87, 95-108. (7) Mant, C. T.; Hodges, R. S. HPLC of Peptides and Proteins: Separation, Analysis and Conformation; CRC Press: Boca Raton, FL, 1991. (8) McNair, H. LC-GC Int. 1992, 5, 10. (9) Law, B.; Appleby, J. R. G. J. Chromatogr. A 1996, 725, 335-341. (10) Weiss, J. Ion Chromatgoraphy, 2nd ed.; VCH Verlagsgesellschaft mbH: Weinheim, 1995. (11) Kirkland, J. J.; Glajch, J. L.; Kohler, J. J. Chromatogr. 1987, 384, 81-96. (12) Unger, K. K. Porous Silica; Elsevier: Amsterdam, 1979. (13) Arshady, R. J. Chromatogr. 1991, 586, 199-219. (14) Kennedy, L. A.; Kopaciewicz, W.; Regnier, F. E. J. Chromatogr. 1986, 359, 73-84. (15) Mu ¨ ller, W. J. Chromatogr. 1990, 510, 133-140. (16) Alpert, A. J.; Andrews, P. C. J. Chromatogr. 1988, 443, 85-96. (17) McNeff, C.; Carr, P. W. J. Chromatogr. 1995, 67, 3886-3892. S0003-2700(97)01024-X CCC: $15.00

© 1998 American Chemical Society Published on Web 03/19/1998

based on metal oxide substrates including alumina,18 titania, and zirconia,17 since siloxane bonds to these materials are quite unstable.19,20 Zirconia is a new type of inorganic support which has been extensively studied in this laboratory during the past decade.21-25 Zirconia is mechanically and thermally stable.19,26 It shows little change in pore volume and surface area at temperatures exceeding 800 °C.27 More importantly, it is chemically stable at extremes of pH, even at temperatures of 100 °C, and has been used from pH 1 to 14.21,28 Such chemical resistance is advantageous when corrosive eluents are used or when the column needs to be sanitized.29 Successful depyrogenation of the column using alkaline alcohol media has been reported.30 This is especially attractive for bioseparations where column fouling is a common problem.31 In these respects, zirconia can be an attractive alternative to traditional silica substrates. However, zirconia’s surface is chemically heterogeneous32 and lacks strong bonding chemistry.33 There are many different types of functional sites on the surface of metal oxides; in particular, there is a high density of troublesome Lewis acid sites. The surface of zirconia has a preponderance of sites which can be classified as very hard Lewis acids. The hard Lewis acid sites will interact with hard Lewis base groups (COO-, PO43-, etc.) on solutes such as proteins, leading to badly broadened peaks or irreversible adsorption.34 These sites can be modified by exposing them to strong Lewis bases such as phosphate or fluoride to produce cation-exchange phases. Quite successful protein separations have been performed on these phases.25,35 However, the eluting salt must be restricted to those which do not remove the modifying Lewis bases from the sorbent surface. Moreover, the modification of zirconia is not permanent and can be sensitive to chromatographic conditions and column history. To produce a more robust cation-exchange phase from zirconia, the stationary phases should be permanently fixed in place, and solute access to the Lewis acid sites should be blocked. Since Zr-O-Si bonds are much more labile than SiO-Si bonds, the silanization process cannot be used to link stationary phases to the surface.33 One alternative is to deposit a polymer layer on the surface or in the pore of the particles. So (18) Heinemann, G.; Ko¨hler, J.; Schomburg, G. Chromatographia 1987, 23, 435441. (19) Tru ¨ dinger, U.; Mu ¨ ller, G.; Unger, K. K. J. Chromatogr. 1990, 535, 111125. (20) Schomburg, G. LC-GC 1988, 6, 36-50. (21) Rigney, M. P.; Weber, T. P.; Carr, P. W. J. Chromatogr. 1989, 484, 27391. (22) Weber, T. P.; Carr, P. W.; Funkenbusch, E. F. J. Chromatogr. 1990, 519, 31-52. (23) McNeff, C.; Zhao, Q. H.; Carr, P. W. J. Chromatogr. A 1994, 684, 201-11. (24) Sun, L.; Carr, P. W. Anal. Chem. 1995, 67, 2517-23. (25) Blackwell, J. A.; Carr, P. W. J. Chromatogr. 1991, 549, 59-75. (26) Rigney, M. P.; Funkenbusch, E. F.; Carr, P. W. J. Chromatogr. 1990, 499, 291-304. (27) Weber, T. P.; Jackson, P. T.; Carr, P. W. Anal. Chem. 1995, 67, 30423050. (28) Kawahara, M.; Nakamura, H.; Nakajima, T. J. Chromatogr. 1990, 515, 149158. (29) Glavanovich, M. H.; Carr, P. W. Anal. Chem. 1994, 66, 2584-2589. (30) McNeff, C.; Flickinger, M. C.; Carr, P. W. Anal. Biochem., submitted. (31) Weary, M.; Pearson, F. Biochem. Pharmacol. 1988, 1, 22-29. (32) Nawrocki, J.; Rigney, M. P.; McCormick, A.; Carr, P. W. J. Chromatogr. A 1993, 657, 229-282. (33) Rigney, M. P. Ph.D. Thesis, University of Minnesota, Minneapolis, MN, 1989. (34) Sun, L.; Carr, P. W. J. Chromatogr. 1994, 658, 465-473. (35) Schafer, W. A.; Carr, P. W. J. Chromatgr. 1991, 587, 149-160.

far, polybutadiene21 and polyethyleneimine23 have been successfully used on zirconia to give reversed and anion-exchange phases, respectively. In this study, a new type of cation-exchange stationary phase has been synthesized by depositing, cross-linking, and functionalizing polyethyleneimine on the surface of zirconia particles following a procedure that emulates a method developed by Kopaciewicz and Regnier for silica.36 In this work, the phases were chemically and physically characterized. The chromatographic performance was studied by measuring column efficiency and examining the separation of small basic compounds and proteins. Protein chromatography was studied in detail. The effects of ionic strength, salt type and concentration, and pH were assessed. Protein recovery, column loading capacity, and phase stability were also investigated. EXPERIMENTAL SECTION Reagents. All reagents were obtained from commercial sources and, unless noted otherwise, were reagent grade or better. Polyethyleneimine (MW 1800) was obtained from Polysciences, Inc. (Warrington, PA). Poly(acrylic acid) (MW 2000), succinic anhydride, 1,4-butanediol diglycidyl ether, N,N-diisopropylethylamine, 4-(dimethylamino)pyridine, m-xylylenediamine, and 2,6diaminopyridine were obtained from Aldrich (Milwaukee, WI). N,N-Dimethylformamide, potassium hydroxide, and 50% sodium hydroxide solution were obtained from Fisher Scientific (Fair Lawn, NJ). Concentrated hydrochloric acid, pyridine, potassium sulfate, potassium chloride, and potassium fluoride were obtained from EM Science (Gibbstown, NJ). Potassium phosphate dibasic, potassium phoshate monobasic, and ammonium phosphate dibasic were obtained from J.T. Baker Chemical (Phillipsburg, NJ). Potassium phosphate dibasic was obtained from Mallinckrodt (Paris, KY). House deionized water was further treated by a Barnsted Nanopure deionizing system with an organic-free cartridge and a 0.2-µm filter, and it was finally boiled to remove carbon dioxide before use. Myoglobin (equine skeletal muscle), lysozyme (chicken egg white), R-chymotrypsin (bovine pancreas), ribonuclease A (bovine pancreas), ribonuclease B (bovine pancreas), and cytochrome c (type VI from horse heart) were obtained from Sigma and were used without further purification. Preparation of Zirconia-Based Polymeric Cation-Exchange Stationary Phase. Microparticulate zirconia particles (batch PICA-7) were synthesized by the polymerization-induced colloid aggregation method.37 Prior to use, the particles were washed in 0.5 M nitric acid and then 0.5 M sodium hydroxide for 8 h, respectively. The bare zirconia particles are relatively monodisperse, with an average size around 2.5 µm in diameter (by SEM), and have a pore size of 250 Å (based on the 4V/A method), a specific surface area of 26 m2/g, and a specific pore volume of 0.16 mL/g by nitrogen sorptometry. Succinylated Polyethyleneimine (PEI-SUCC) Coating. This material was synthesized by a modification of the method developed by Kopaciewicz and Regnier for use on silica.36 Zirconia (3.5 g) was dried in a vacuum oven at 120 °C overnight and then (36) Kopaciewicz, W.; Regnier, F. E. J. Chromatogr. 1986, 358, 107-117. (37) Sun, L.; Annen, M.; Lorenzaro, F.; Carr, P. W.; McCormick, A. V. J. Colloid Interface Sci. 1994, 163, 464-473.

Analytical Chemistry, Vol. 70, No. 9, May 1, 1998

1935

suspended in 15 mL of methanol solution containing 1% (w/v) PEI-18 by applying vacuum sonication for 5 min. The slurry was kept at room temperature for 2 h. The solution was then vacuum filtered. Without a washing step, the adsorbed PEI-18 was then cross-linked with 10 mL of a 5% (w/v) methanolic BUDGE solution at room temperature for 24 h and then heated at 80 °C for 1 h. Excess solution was removed by vacuum filtration, and the particles were washed with methanol. The particles were dried at 100 °C under vacuum for 2 h. The dry zirconia particles were then suspended in 10 mL of dry dimethylformamide containing 0.5 mL of diisopropylethylamine (DIEA) and 0.5 mL of succinic anhydride. The reaction was conducted at 64 °C for 24 h. The particles were collected on a medium-porosity sintered glass funnel and successively washed with DMF, methanol, water, and methanol. After drying, the particles were further cleaned by drypacking into a 250-mm × 4.6-mm-i.d. column and flushing with 300 mL of 0.5 M KCl solution. Polyacrylated Polyethyleneimine (PEI-PAA) Coatings. PEI-18 was adsorbed onto zirconia as in the above synthesis. The coated zirconia was then dried under vacuum at 100 °C for 2 h. A 3.5-g portion of the dry material was suspended in 10 mL of DMF containing 0.5 mL of dry DIEA and either 0.1 or 1.0 g of poly(acrylic acid) anhydride (PAA). The reaction and the subsequent cleanup steps were similar to those of the PEI-SUCC synthesis. Poly(acrylic acid) anhydride was synthesized by heating polyacylic acid under vacuum at 180 °C for 4 h. The zirconia particles were packed by the upward stirred slurry method at 5000 psi. 2-Propanol was used as both the slurry solvent and the pushing solvent. Stainless steel (316) column blanks with dimensions of 50 mm × 4.6 mm id. and 2-µm stainless steel frits were obtained from Alltech (Alltech Associates Inc., Deerfield, IL). Physical Characterization of Polymer-Coated Zirconia Particles. Elemental analysis was done by Microanalysis, Inc. Nitrogen adsorption measurements were carried out using a Micrometrics ASAP 2000 sorptometer (Micrometrics Instrument Corp., Norcross, GA). The data were processed according to both the BET and BJH methods.38,39 Samples were heated at 100 °C under vacuum for 4 h to remove any adsorbed gases. Chromatographic Conditions. Chromatographic studies were carried out on a Hewlett-Packard 1090 liquid chromatograph with an autosampler, a temperature controller, and a diode array detector (Hewlett-Packard S.A., Wilmington, DE). Data were processed using Hewlett-Packard Chemstation software. All buffer solutions were filtered using Millipore (type HA) 0.45-µm membrane filters prior to use. Samples were prepared in the mobile phases. Small basic solutes were prepared as 1-2 mg/mL solution, and the injection volumes were usually 2 µL. Protein samples were typically prepared as 0.5% (w/v) solution in the corresponding mobile phase. Plate numbers were calculated according to eq 1.

N ) 5.54(tR/w1/2)2

(1)

The column dead volumes were determined using acetone as a dead volume marker. (38) Brunauer, S.; Emmett, P. H.; Teller, E. J. Am. Chem. Soc. 1938, 60, 309. (39) Barrett, E. P.; Joyner, L. G.; Halender, P. P. J. Am. Chem. Soc. 1951, 73, 373-380.


Protein Recovery Studies. The protein mass recovery was measured according to a previously reported method.40 Two 5.0cm × 0.46-cm-i.d. columns packed with the same material were used. One column (column 2) was connected and disconnected during the test, while the other column (column 1) remained in the system. The mass recovery was calculated as shown in eq 2.

% mass recovery ) area(col 1 + col 2)/area(col 1) × 100 (2)

The proteins were eluted isocratically, and the ionic strength of the mobile phase was adjusted to ensure sufficient retention (k′ > 1) so that any disturbance at the dead time was not a probem in peak integration. RESULTS AND DISCUSSION Synthesis and Physicochemical Characterization of the Stationary Phases. The zirconia-based polymeric cationexchange stationary phases were synthesized by two routes. In one method, polyethyleneimine (PEI) was first adsorbed onto porous zirconia particles and cross-linked with 1,4-butanediol diglycidyl ether (BUDGE). Succinic anhydride was then reacted with the remaining primary and secondary amine groups on PEI to generate the anionic functionalities that serve as the cationexchange sites. In the second method, poly(acrylic acid) anhydride (PAA) was used to replace both the BUDGE and the succinic anhydride. PAA acts to simultaneously cross-link the adsorbed PEI and provide the cation-exchange functional groups. Preliminary experiments showed that PEI concentrations above 1% (w/v) and adsorption times longer than 2 h did not lead to an increase in the quantity of adsorbed PEI. Therefore, these conditions were used throughout this work. The amount of BUDGE used was based on previous reports.23 No further optimization of the coating procedure was conducted. The polymer loadings and pore structures of the stationary phases were characterized by elemental analysis and nitrogen adsorption analysis. The results are shown in Table 1. In general, about 13 µmol of PEI repeating units are immobilized per square meter of zirconia surface. This is comparable to the value reported previously.23 Apparently, PAA is as effective a cross-linker as BUDGE, based on the amount of PEI retained after washing. The lower loading of PEI on the PEI-PAA1 phase is due to the smaller quantity of PAA used. Based on the calculation from elemental analysis, about 60% of the amine groups of immobilized PEI are derivatized with succinic anhydride. The rest can be accounted for by the existence of unreactive tertiary amines and unaccessible amines. This result is very close to the 70% value reported by Kopaciewicz and Regnier.36 The PEI-SUCC phase, has a much higher carbon content than does the PEI-PAA phase, mainly due to the extra carbon added by cross-linking with BUDGE. Polymer coating resulted in decreases in surface area, pore volume, and average pore diameter, as demonstrated by nitrogen adsorption analysis (Table 1). As expected, the higher polymer loadings gave bigger decreases in surface area and pore volume; the changes in pore diameter do not follow this trend. The PEISUCC phase showed only a slight decrease in pore diameter, (40) Yang, Y. B.; Harrison, K.; Kindsvater, J. J. Chromatogr. A 1996, 723, 1-10.

Table 1. Physical and Chemical Characterization of Zirconia-Based Polymeric Cation-Exchange Stationary Phases stationary phase

% Ca

% Na

PEI unitsb (µmol/m2)

SUCC/PAA unitsc (µmol/m2)

surface aread (m2/g)

pore volumed (cm3/g)

pore diameterd,g (Å)

bare zirconia PEI-SUCC PEI-PAA1e PEI-PAA2f

3.51 2.12 2.51

0.49 0.42 0.48

13.7 11.7 13.4

8.0 15.2 18.3

25.6 19.2 25.7 24.7

0.160 0.111 0.132 0.127

250 232 206 206

a Percent by weight amount of carbon and nitrogen on zirconia obtained by elemental analysis. b PEI repeating units absorbed on zirconia is calculated on the basis of the nitrogen analysis. c Immobilized succinic anhydride molecules or poly(acrylic acid) repeating units are calculated on the basis of the analysis of carbon and nitrogen. d Results obtained from nitrogen adsorption analysis. e 0.1 g of PAA was used for every 3.5 g of zirconia; see Experimental Section for details. f 1.0 g of PAA was used for every 3.5 g of zirconia; see Experimental Section for details. g Pore diameters were calculated as 4(pore volume/surface area).

Figure 2. Chromatography of small basic solutes using the PEISUCC phase. Mobile phase: 20 mM KH2PO4, pH 5.5. Other conditions: 1 mL/min; 30 °C; detection at 254 nm. Samples: a, pyridine; b, 4-(dimethylamino)pyridine; c, m-xylylenediamine; d, 2,6diaminopyridine. Figure 1. Nitrogen sorptometry analysis of stationary phases. (A) Adsorption direction pore volume plot; (B) adsorption direction pore area plot. (s) Bare zirconia, (- - -) PEI-SUCC phase, (-‚-) PEIPAA1 phase, and (‚‚‚) PEI-PAA2.

although it had the highest polymer loading. The area-based pore size distribution (based on the adsorption direction in the BET analysis)39 showed that the PEI-SUCC phase has a smaller contribution to the surface area from pores smaller than 40 Å (see Figure 1). This suggests that some small pores are preferentially filled by polymer, which, in turn, could cause a bigger decrease in surface area than in pore volume. We attribute the apparently small change in the average pore diameter based on the 4V/A method to this.41 The greatest challenge in depositing polymers on the surface of porous particles is to achieve a uniform coating and avoid blocking access to the pores.32 As reported for polybutadiene (PBD)-coated zirconia, PBD has a very low affinity for zirconia, and PBD molecules in the absence of solvent aggregate to form patches.41 When a high polymer loading of PBD (>5% (w/w)) is used, severe pore blockage results. We expect that the more hydrophilic and ionic polyethyleneimine will have a much higher affinity for zirconia than does nonpolar PBD, because PEI can interact with zirconia by dipolar interaction, hydrogen bonding, and even Coulombic interactions (between (41) Reeder, D. H.; Li, J.; Carr, P. W.; Flickinger, M. C.; McCormick, V. J. Chromatogr. A 1997, 760, 71-79.

-NH2+ and ZrO- groups). Furthermore, PEI is coated by adsorption, not by solvent evaporation. Therefore, PEI coatings should be more homogeneous and thinner. Overall, the shapes of the pore size distribution curves (see Figure 1) of the three phases resemble that of the bare zirconia substrate, suggesting that uniform coating has been achieved, although we have no strong data to support our view. Chromatography of Small Basic Solute. Small basic solutes were separated on the PEI-SUCC phase simply to provide a test of its chromatographic performance prior to detailed studies with proteins. Good peak shapes and column efficiencies were obtained (see Figure 2), considering that ion-exchange stationary phases usually show lower column efficiency than do reversed phases. Solutes with a higher positive charge, such as 2,6diaminopyridine (k′ ) 7.44, N ) 2510) and m-xylylenediamine (k′ ) 4.21, N ) 2760) were more retained than solutes with lower charge, such as pyridine (k′ ) 0.50, N ) 4050). The more hydrophobic, singly charged 4-(dimethylamino)pyridine (k′ ) 1.82, N ) 4200) was less retained than the more hydrophilic, doubly charged 2,6-diaminopyridine. These results suggest that hydrophobic contributions to the retention are not so strong as to cause a singly charged cation to be more retained than a doubly charged cation. Clearly, cation exchange is the dominant mechanism of retention on the PEI-SUCC phase. The low ionic strengths required for elution suggest that these phases may be useful for ion chromatography. Analytical Chemistry, Vol. 70, No. 9, May 1, 1998

1937

Figure 3. Plots of the reduced plate height against the reduced velocity using the PEI-SUCC phase. See Figure 2 for conditions. The solid symbols are experimental data, and solid lines are fits to the Knox equation. Symbols: b, pyridine; 9, 4-(dimethylamino)pyridine; 2, m-xylylenediamine; and 1, 2,6-diaminopyridine.

Dynamic Study. To further characterize the polymeric stationary phase, dynamic studies were performed using the above-mentioned small basic solutes. Plate heights were measured at different linear velocities. Reduced plate heights were then plotted against the reduced velocity, and the curves were fitted to the Knox equation (see Figure 3). The fitting results are given in Table 2. Diffusion coefficients were calculated using the modified Wilke-Chang equation.42 In general, the A terms are much bigger than desired (g1) and also vary considerably with k′. Furthermore, we see a big variation in hmin with k′. These results indicate that the column packing is not good. So far, the packing of 2.5-µm zirconia particles is still a challenge.43 The C term of the Knox equation generally reflects the mass-transfer properties both in the stagnant mobile phase and in the stationary phase. The smaller the C term, the faster is the overall intraparticle mass-transfer process.44 The C terms for the PEI-SUCC phase are bigger than those of ODS silica phases. Cross-linked “fuzzy” structures of polymeric phases are expected to hinder mass transfer in the stagnant mobile phases more and cause slower diffusion in the stationary phases than do monomeric ODS chains.43 However, the C terms for the PEI-SUCC phase are generally less than 0.2. This is comparable to and even less than that of PBD-coated zirconia.45 We point out, as reported by Berthod,46,47 that it is very difficult to correctly assess the theoretical plate height for non-Gaussian peaks. This may cause uncertainty in A, B, and C terms. A larger A term also makes the C term less reliable. Nonetheless, the efficiencies exhibited here are quite good for ion-exchange phases, although they are not as good as those observed in RPLC. Chromatography of Basic Proteins. As pointed out above, one of the major applications of ion-exchange chromatography is the analysis and separation of biomolecules such as peptides, proteins, and nucleic acids. As a major goal of this work, the (42) Li, J.; Carr, P. W. Anal. Chem. 1997, 69, 2530-2536. (43) Li, J.; Carr, P. W. Anal. Chem. 1997, 69, 2193-2201. (44) Grushka, E.; Snyder, L. R.; Knox, J. H. J. Chromatogr. Sci. 1975, 13, 2537. (45) Hu, Y., results unpublished. (46) Berthod, A. J. Liq. Chromatogr. 1989, 12, 1187-1201. (47) Berthod, A. J. Liq. Chromatogr. 1989, 12, 1169-1185.


suitability of these novel zirconia-based polymeric cation-exchange phases for protein chromatography was studied in detail. A set of model basic proteins was separated on the three polymeric phases (see Figure 4). The elution orders are the same for all three phases, implying that the retention mechanism is the same on all three. Interestingly, both PEI-PAA phases are much more retentive toward lysozyme than is the PEI-SUCC phase. This cannot be explained simply by invoking a higher ionexchange capacity for the PEI-PAA phases, as this will lead to increased retention for all proteins. One possible explanation is that the PAA chains are not fixed on the support surface, but, instead, they are extended out, like tentacles. This may favor interaction between the stationary phase and specific solutes such as lysozyme. Another explanation is that the PEI-PAA phases are more hydrophobic than is the PEI-SUCC phase, which thereby increases the retention of the more hydrophobic lysozyme. However, the results from the study of the hydrophobicity of these phases favor the former explanation (vide infra). Due to its potential strong interactions with certain proteins, e.g., lysozyme, the PEI-PAA phases are less attractive as cationexchange stationary phases for protein chromatography than is the PEI-SUCC phase, although they may be useful for immobilzing proteins for enzyme reactors or other applications. One of the most important features of these new phases is that they show selectivities very different from those of cation exchangers based on zirconia modified with phosphate groups or with adsorbed fluoride. Table 3 summarizes the elution orders of some proteins on chemically different zirconia-based cationexchange phases. On the polymeric phases, lysozyme was eluted last and ribonuclease was eluted before R-chymotrypsin, while on phosphate- or fluoride-modified zirconia the opposite is observed. On the other hand, ethylenediamine-N,N′-tetra(methylphosphonic acid) (EDTPA)-modified zirconia shows selectivity similar to that of the polymeric phases, except that the elution order of lysozyme and cytochrome c is switched. This may suggest that the chemical properties of EDTPA-modified zirconia more closely resemble those of the polymer-coated zirconia. Interestingly, the elution order of the proteins tested using the polymeric phases coincides with their isoelectric points (see Table 4). The higher pI values the proteins have, the more retained they are. In contrast, the work of Lenhoff shows that the average charge density and not the pI controls retention.48 The selectivity of the polymeric phases relative to the totally inorganic phases (Zr-P, Zr-F) may be due to a combination of electrostatic interactions and hydrophobic interactions (vide infra). The hydrophobicity of the PEI-SUCC and PEI-PAA phases was studied using alkylphenones as probe solutes. The free energy of transfer of a methylene unit from a purely aqueous mobile phase to the stationary phase was determined to be about -1.0 kJ/mol for PEI-SUCC phase, -0.86 kJ/mol for PEI-PAA1 phase, and -0.91 kJ/mol for PEI-PAA2 phase (see Figure 5). These values are about half of that observed for an ODS phase (-2.4 kJ/mol) in a 100% aqueous eluent.17 Without doubt, the hydrophobicity of the stationary phase must play an important role in the retention of proteins on this phase. It is noteworthy that ∆G°CH2 values of the PEI-SUCC and PEI-PAA phases are very close to that reported for the BUDGE cross-linked PEI phase (48) Yoon, B. J.; Lenhoff, A. M. J. Phys. Chem. 1992, 96, 3130.

Table 2. Dynamic Study of the PEI-SUCC Phasea solute

k′

Ab

Bb

Cb

νoptimumc

hminc

pyridine 4-(dimethylamino)pyridine m-xylenyldiamine 2,6-diaminopyridine

0.50 1.82 4.21 7.44

1.85 ( 0.12 1.82 ( 0.09 2.58 ( 0.10 3.35 ( 0.25

4.06 ( 0.17 5.41 ( 0.17 6.12 ( 0.22 8.94 ( 0.71

0.17 ( 0.02 0.08 ( 0.01 0.16 ( 0.01 0.00 ( 0.04

4.5 4.0 3.3 3.7

4.4 4.4 6.3 8.0

a Measurements were conducted in 20 mM KH PO aqueous mobile phase at 30 °C. The flow rate was varied from 0.1 to 4 mL/min. The 2 4 detection wavelength was set at 254 nm. b Determined by fitting the plot of reduced plate height (h) vs reduced velocity (ν) to the Knox equation, 1/3 h ) Aν + B/ν + Cν. Diffusion coefficients estimated: 1.09 × 10-5 cm2/s for pyridine, 8.27 × 10-6 cm2/s for 4-(dimethylamino)pyridine, 7.91 × 10-6 cm2/s for m-xylenyldiamine, and 9.05 × 10-6 cm2/s for 2,6-diaminopyridine. c From experimental results.

Figure 4. Protein separations using the PEI-SUCC and PEI-PAA phases. Experimental conditions: 20 to 500 mM K2HPO4 (pH 7) gradient in 20 min; 1 mL/min; 30 °C; detection at 220 nm. (A) PEISUCC phase, (B) PEI-PAA1 phase, and (C) PEI-PAA2 phase. Samples: 1, myoglobin; 2, ribonuclease B; 3, ribonuclease A; 4, cytochrome c; and 5, lysozyme.

(∆G°CH2, PEI ) -0.90 kJ/mol).17 This indicates that further modification of the PEI phase with either succinic anhydride or poly(acrylic acid) anhydride has little effect on the hydrophobicity. Study of the Effect of Salt. The influence of the counterion concentration on the retentions of proteins was studied first under isocratic elution conditions. Figure 6 shows the variation in protein retention as the concentration of potassium chloride was varied. Increasing the amount of potassium chloride in the mobile phases decreases the retention time of all proteins. This is typical ion-exchange behavior. We note that the rate of decrease in k′ slows down slightly at the higher salt concentrations. This further suggests that hydrophobic interaction may contribute to retention. The slope of the plot of the logarithm of the capacity factors for one protein against the logarithm of the reciprocal displacing salt concentration is considered by some to be a measure of the number of electrostatic interactions between the protein and the ion-exchange support, and it is named the Z value.49,50 Proteins tested on the PEI-SUCC phase have Z values between 1.5 and (49) Kopaciewicz, W.; Rounds, M. A.; Fausnaugh, J.; Regnier, F. E. J. Chromatogr. 1983, 266, 3-21.

2.5 (see Table 4). Lysozyme has a Z value of 1.76, which fell in the range of 1.67-2.92 reported for fluoride-modified zirconia phase.25 However, no positive correlation between Z values and retention can be seen. The Z values of these proteins should be obtained using different displacing salts at different pH values to make a more thorough comparison. The effect of different counterions was also investigated. Under the same gradient elution conditions, three counterions, K+, NH4+, and Na+, all showed roughly similar retention behavior for all proteins tested (Table 5). The eluting power of K+ is slightly greater than that of NH4+. In the case of Na+, ribonuclease A and cytochrome c were more retained, while R-chymotrypsin and lysozyme were less retained than when K+ and NH4+ were used. In genaral, the counterions with the higher valence, smaller solvated volume, and greater polarizability have stronger dispacing power.10,51 To a crude approximation, the elution strength of the above three cations should follow the order of K+ > NH4+ > Na+.51 The anomalous behavior of Na+ compared to K+ or NH4+ is suspected to be related to both the different ion pairing power of the counterions with proteins and the amphoteric nature of proteins. If solutes are simply positively charged, the counterions with the stronger elution strength will elute them from cation-exchange phases earlier. However, proteins are amphoteric at pH 7, and their negatively charged sites will be repelled by the negative charges of the stationary phase, leading to shorter retention time. The counterion with stronger ion paring power can better ion pair with these negatively charged sites and weaken such repulsion, leading to longer retention time. One of the important features of zirconia as a chromatographic support is that it has strong hard Lewis acid sites which bind tightly to hard Lewis bases. While this feature can be used to advantage, as with the use of phosphate- or fluoride-modified zirconia, it is generally harmful for protein chromatography, as it will cause excessive retention, low recovery, and extremely poor column efficiency.34 To overcome this problem, strong Lewis bases such as phosphate or fluoride are used in high concentrations in the eluting buffer for protein separation on phosphateand fluoride-modified zirconia.25,35 It is, therefore, of interest to investigate the influence of the anion of the eluting salts on protein separations with these novel polymer-coated zirconia cation exchangers. Four potassium salts with HPO42-, F-, SO42-, and Cl- as counterions, respectively, were used to elute proteins under similar conditions with the same gradient of potassium ions. (50) Rounds, M. A.; Regnier, F. E. J. Chromatogr. 1984, 283, 37-45. (51) Walton, H. F. Ion Exchange Chromatography; Halsted Press Division of John Wiley and Sons: Stroudsburg, PA, 1976.

Analytical Chemistry, Vol. 70, No. 9, May 1, 1998

1939

Table 3. Elution Orders of Proteins on Different Zirconia-Based Cation-Exchange Stationary Phases stationary phases coateda

PEI/SUCC EDTPA modifiedb phosphate modifiedc fluoride modifiedd

elution orders of proteins myoglobin (1) < ribonuclease B (2) < ribonuclease A (3) < R-chymotrypsin (4) < cytochrome c (5) < lysozyme (6) myoglobin (1) < ribonuclease B (2) < ribonuclease A (3) < R-chymotrypsin (4) < lysozyme (6) < cytochrome c (5) myoglobin (1) < lysozyme (6) < R-chymotrypsin (4) < ribonuclease A (3) < cytochrome c (5) lysozyme (6) < ribonuclease B (2) < R-chymotrypsin (4) < ribonuclease A (3) < myoglobin (1) < cytochrome c (5)

a 20 mM to 0.212 M K HPO (pH 7) gradient in 20 min, 1 mL/min. b 50 mM to 0.5 M K HPO (pH 7) gradient in 30 min, 0.5 mL/min (from ref 2 4 2 4 54). c 50 mM to 0.5 M K2HPO4 (pH 7) gradient in 30 min, 0.5 mL/min (from ref 35). d A, 0.1 M NaF + 20 mM MES (pH 5.5); B, A + 0.5 M Na2SO4; 0% to 100% B gradient in 30 min, 1 mL/min (from ref 25).

Table 4. Z Values for Selected Proteins on a PEI-SUCC Phasea protein

pIb

k′ c

Zd

ribonuclease B ribonuclease A R-chymotrypsin cytochrome c lysozyme

9.3 9.3 9.8 10.2 11.0

0.22 0.59 1.24 5.25 54.45

1.81 1.78 1.69 2.21 1.76

a See Figure 5 for conditions. b From refs 55 and 56. c Retentions in 136 mM KCl. d Z values obtained by fitting the curves in Figure 5 to the equation log k′ ) 2Z log(1/[KCl]) + log Kz (see ref 49).

Figure 6. Concentration effect of the displacing ion on protein elution using the PEI-SUCC column. Experimental conditions: isocratic elution with 20 mM K2HPO4 + variable concentration of KCl; 1 mL/min; 30 °C; detection at 220 nm. The solid symbols are experimental data, and the solid lines are linear regression fits. Symbols: b, lysozyme; 9, cytochrome c; 1, R-chymotrypsin; [, ribonuclease A; and 2, ribonuclease B. Table 5. Effect of Cation Type on Protein Separations with a PEI-SUCC Phasea retention time (min) protein Figure 5. Retention of the homologue series of alkylphenones versus carbon number of alkyl chain on the PEI-SUCC and PEIPAA columns. Experimental conditions: 20 mM KH2PO4, pH 7; 0.7 mL/min; 30 °C; detection at 245 nm. Symbols: b, PEI-SUCC column; 9, PEI-PAA1 column; and 2, PEI-PAA2 column.

These four anions are listed in order of decreasing Lewis basicity.52 If Lewis base/acid interactions contribute significantly to the retention of proteins on polymer-coated zirconia, a trend of increasing retention should be seen as the eluting salt is changed in the order K2HPO4, KF, K2SO4, and KCl. The anion component of the eluting salts has only a small effect on both the retention times and peak shapes (see Table 6). Only in the case of lysozyme when KF was used was a significant increase in retention time observed. This is most likely attributed to the individual protein. It is safe to conclude that, in the presence of 20 mM K2HPO4 buffer, there are only minimal effects of the Lewis acid sites of zirconia on protein elution. This suggests that polymer coating and the 20 mM K2HPO4 buffer effectively shield the surface Lewis acid sites of zirconia. Therefore, zirconia-based polymeric cation(52) Blackwell, J. A.; Carr, P. W. Anal. Chem. 1992, 64, 863-873.



K2HPO4

(NH4)2HPO4

Na2HPO4

3.64 4.58 6.58 8.03 15.49

3.93vb

3.83Vc 5.16v 6.35V 11.28v 14.59V

4.90v 6.98v 9.19v 16.59v

a Mobile phases: A, 20 mM K HPO (pH 7); B, A + 0.5 M of the 2 4 indicated salt. Experimental conditions: gradient elution from 0% to 80% B in 40 min; 1 mL/min; 30 °C; 220 nm. b v indicates an increase in retention time compared to the previous column. c V indicates a decrease in retention time compared to the previous column.

exchange phases can be used with no additional limitation on the choice of eluting salts. This is a significant advantage compared to the zirconia modified with small anions. Study of pH Effects. Since the pKa values of carboxylic groups on the cation-exchange phases are around 4.5, these phases should be used at pH above 5.5 in order to fully utilize the ion-exchange capacity.36 The retention behavior of five model proteins was studied between pH 5.5 and 8.5. A monotonic decrease in retention with increasing pH values of the mobile phase was observed for all proteins studied (see Figure 7). Similar results were reported on a similarly modified silica support.36 The higher pH of the mobile phase may deprotonate the proteins.

Table 6. Anion Effect on Protein Separations with PEI-SUCC Phasea

Table 7. Protein Recovery Study on PEI-SUCC Column

retention time (min) protein

K2HPO4

K2SO4

KCl

KF


3.64 4.58 6.58 8.03 15.49

3.71 4.78 6.26 8.23 14.35

4.12 5.30 6.91 8.28 14.63

3.71 4.79 7.05 8.07 19.18

proteins

mobile phase

% recoverya,b


44 mM K2HOP4, pH 7 44 mM K2HOP4, pH 7 68 mM K2HOP4, pH 7 92 mM K2HOP4, pH 7 160 mM K2HOP4, pH 7

99.4 ( 4.1 96.5 ( 0.3 99.5 ( 2.3 93.9 ( 3.9 95.9 ( 1.0

a See Experimental Section for calculation. b Experimental conditions: 1 mL/min, 30 °C, detection at 220 nm.

a Mobile phases: A, 20 mM K HPO (pH 7); B, 0.5 M K HPO (pH 2 4 2 4 7), or A + 0.5 M K2SO4, or A + 1 M KCl, or A + 1 M KF. Experimental conditions: gradient elution from 0% to 80% B in 40 min; 1 mL/min; 30 °C; 220 nm.

Figure 7. pH effect on protein retentions on the PEI-SUCC column. Experimental conditions: gradient elution in 20 min from 20 mM to 0.5 M K2HPO4 at variable pH adjusted by KOH; 1 mL/min; 30 °C; detection at 220 nm. Symbols: b, ribonuclease B; 9, ribonuclease A; 2, R-chymotrypsin; 1, cytochrome c; and [, lysozyme.

Obviously, a less positively charged protein will be less retained on a cation exchanger. In contrast to the complicated pH effect for the phosphate-modified zirconia, such a montonic effect of the pH may be advantageous for method development with the PEISUCC phase. Protein Recovery and Loading Capacity. Protein recovery on the PEI-SUCC phase was studied as described in the Experimental Section. The mobile phases were chosen to give reasonable retention times and, therefore, reasonable peak widths for the proteins tested. This will ensure accurate peak integration. The PEI-SUCC phase gave good recovery ratios for all proteins tested, with the average recovery being higher than >95% (see Table 7), indicating few strong binding sites for proteins. This means not only that there is less chance of column fouling in analytical chromatography but also that good recovery is obviously an important requirement for preparative chromatography. The dynamic loading capacity of the PEI-SUCC phase was investigated by monitoring the column performance with increasing sampling size. When the sample quantities are below the loading capacity of the column, the retention time and the peak width will remain constant. Overloading the column will cause a decrease in retention time and an increase in peak width.25 Lysozyme was chosen in this study, and the results are shown in Figure 8. A 4 orders of magnitude increase in sample quantity

Figure 8. Study of loading capacity of the PEI-SUCC phase. Experimental conditions: isocratic elution in 20 mM K2HPO4 + 0.2 M KCl mobile phase at pH 7; 1 mL/min; 30 °C; detection at 254 nm (detection at 220 nm for very small amount of lysozyme). Symbols: 2, capacity factor; b, peak width.

had little effect on the capacity factor of lysozyme. On the other hand, the peak width is somewhat more sensitive to sample load. The dynamic loading capacity of the PEI-SUCC column was estimated to be about 240 µg/mL of blank column volume, based on the amount of sample which caused a 10% increase in peak width. This is 3 times higher than that reported for fluoridemodified zirconia.25 We should point out that this is a very conservative measure of loading capacity compared to the frontal analysis. Columns with higher loading capacity can handle larger sample quantity without compromising column performance. The bigger peak widths at extremely low sampling quantity were probably due to the larger integration error of very small peaks. Phase Stability. Inorganic supports such as silica, alumina, and zirconia have excellent mechanical stability compared to polymeric supports. Polymer-coated inorganic supports can combine the rich chemistry available through polymer coating and the mechanical, chemical, and thermal stability of inorganic supports to afford new types of stationary phases. Since there is no direct bonding chemistry between zirconia and polyethyleneimine, the immobilization of polymer mainly depends on the crosslinking step. High pressures and high flow rates might dislodge the polymer coating and cause back pressure surges if the crosslinking process is not successful. PEI-SUCC phases were tested under different flow rates, and the column back pressures were monitored. A reasonably linear relationship between the back pressure and the flow rates was observed, as shown in Figure 9, which indicates the high mechanical stability of the phase. Analytical Chemistry, Vol. 70, No. 9, May 1, 1998

1941

pH 12 corrosive conditions.53 This demonstrates the reasonable chemical stability of amide bonds of polylysine. However, due to the amide bonding in the PEI-SUCC phase, pH 10 may represent the high limit that should be used on this phase. Nontheless, it is superior to the currently widely used silica-based bonded cationexchange phases. When used in a buffer with pH above 7, the silica bonded phase will start to bleed.

Figure 9. Plot of column back pressure against the flow rate. Mobile phase: water; 30 °C. 2, backpressure change with decreasing flow rate; 9, backpressure change with increasing flow rate.

Figure 10. Phase stability of the PEI-SUCC phase in alkaline solution. Experimental conditions: column was flushed in 20 mM K2HPO4 mobile phase of pH 10 at 1 mL/min at 30 °C. The column was tested in 92 mM K2HPO4 mobile phase of pH 6.46 after being equilibrated for 20 min. Sample: lysozyme.

Since zirconia support is stable between pH 1 and 14, the chemical stability of the new phase is determined by the bonding chemistry of the polymer coatings. The PEI-SUCC phase was studied for more than 6 months. Good chromatographic performance was maintained throughout. To more rigorously test the phase stability, the PEI-SUCC phase was subject to continuous flushing of 20 mM K2HPO4 aqueous solution adjusted to pH 10. The integrity of the stationary phase was monitored through the retention time of lysozyme. As shown in Figure 10, flushing the column with 4000 column volumes of mobile phase did not alter the retention time of lysozyme. Therefore, the PEI-SUCC phase is chemically stable to at least pH 10. This result is supported by earlier studies on PEI-coated zirconia, which showed it to be stable to pH 10.23 Even polylysine-coated zirconia can resist corrosive


CONCLUSIONS A new type of weak cation-exchange stationary phase has been produced by immobilizing polyethyleneimine on the surface of zirconia and functionalizing it with carboxylate groups. This material utilizes the versatile chemistry of polymer coating and excellent mechanical and chemical stability of the zirconia. Both PEI-SUCC and PEI-PAA phases can be used for the separation proteins. PEI-PAA phases show particularly strong retention of lysozyme. The PEI-SUCC phase demonstrates high pressure resistance and chemical stability up to at least pH 10. In this respect, it is superior to both silica support and pure polymeric supports. The new material also shows good column efficiency for both small organic solutes and proteins. The polymer coating apparently shields to a considerable extent the strong protein binding sites on the base zirconia. Therefore, large amounts of strong Lewis base additives in eluting buffers are no longer required. This will definitely expand the applicability of this new phase compared to zirconia modified with small anions. The retention mechanism is predominantly Coulombic interactions with a minor contribution from hydrophobic interactions. Such a combination affords a unique selectivity for protein separation on this phase. Counterion concentrations and mobile phase pHs show monotonic effects on all proteins tested. Such predictable retention behavior is in contrast to that on small anion-modified zirconia, which will facilitate method development. The material also shows high loading capacity and no sign of irreversible binding of proteins, both of which are desirable proterties. Overall, this phase is a good candidate for chromatography of both small molecules and biomoleules. ACKNOWLEDGMENT The authors acknowledge the financial support by grant GM 45988-07 from the National Institutes of Health.

Received for review September 16, 1997. January 28, 1998.

Accepted

AC9710240 (53) McNeff, C. V. Ph.D. Thesis, University of Minnesota, Minneapolis, MN, 1996. (54) Clausen, A. M.; Carr, P. W. Anal. Chem. 1998, 70, 378-385. (55) Kadoya, T.; Ogawa, T.; Kuwahara, H.; Okuyama, T. J. Liq. Chromatogr. 1988, 11, 2951-2967. (56) Malamud, D.; Drysdale, W. Anal. Biochem. 1978, 86, 620-647.

Synthesis and Characterization of New Zirconia-Based Polymeric

Recommend Documents