Anal. Chem. 2002, 74, 5641-5649
Gradient Chromatofocusing. Versatile pH Gradient Separation of Proteins in Ion-Exchange HPLC: Characterization Studies Lian Shan† and David J. Anderson*
Department of Chemistry, Cleveland State University, Cleveland, Ohio 44115
A new chromatofocusing technique called gradient chromatofocusing is characterized. Gradient chromatofocusing generates linear pH gradients on anion-exchange columns with inexpensive low molecular mass buffer components via HPLC gradient mixing. Gradient chromatofocusing results are compared with that of conventional chromatofocusing in the chromatography of several proteins on a Mono P column, including β-lactoglobulin A and B, ovalbumin, BSA, and conalbumin. Gradient chromatofocusing shows superior performance, with resolution increases greater than 3-fold being realized for the entire protein mixture and up to 25-fold for a particular protein pair. This performance superiority arises from inherent advantages in the gradient chromatofocusing technique in optimizing conditions pertinent to separation, including buffer concentration and pH gradient slope. These resolution gains arise from both increases in separation factor and decreases in peak width achieved with the pH gradient chromatofocusing technique through the manipulation of buffer concentration and the pH gradient profile. Gradient chromatofocusing is also compared with conventional NaCl gradient ion-exchange chromatography using the same Mono P column, demonstrating 3-fold resolution gains, resulting from a 3-fold decrease in peak width. The present work demonstrates the significantly improved performance that gradient chromatofocusing has in protein separations compared to other ionexchange chromatographic techniques. Mechanisms for the various effects are discussed. Conventional chromatofocusing is an ion-exchange chromatography technique developed by Sluyterman and co-workers1-5 in the late 1970s, which can produce a pH gradient on an ionexchange column and separate proteins based on their different * Corresponding author. E-mail:
[email protected]. Fax: 216-687-9298. † Current address: Department of Cell Biology, Lerner Research Institute, The Cleveland Clinic Foundation, 9500 Euclid Ave., Cleveland, OH 44196. (1) Sluyterman, L. A. AE.; Wijdenes, J. In Electrofocusing and Isotachophoresis [Proceedings of the International Symposium, August 2-4, 1976, Hamburg, Germany]; Radola, B. J., Graesslin D., Eds.; Walter de Gruyter: Berlin, 1977; pp 463-466. (2) Sluyterman, L. A. AE.; Elgersma, O. J. Chromatogr. 1978, 150, 17-30. (3) Sluyterman, L. A. AE.; Wijdenes J. J. Chromatogr. 1978, 150, 31-44. (4) Sluyterman, L. A. AE.; Wijdenes, J. J. Chromatogr. 1981, 206, 429-440. (5) Sluyterman, L. A. AE.; Wijdenes, J. J. Chromatogr. 1981, 206, 441-447. 10.1021/ac020169q CCC: $22.00 Published on Web 10/08/2002
© 2002 American Chemical Society
charge characteristics. There are many limitations to the conventional chromatofocusing technique, as discussed below. One limitation of the conventional technique is the characteristics of the mobile-phase buffer. The technique requires polymeric ampholyte buffers, which have limitations of expense, poor chromatographic reproducibility because of variability in the physical and chemical properties,6,7 and association with some proteins leading to additional complications.8,9 To overcome these limitations, conventional chromatofocusing techniques using mobile phases employing low molecular mass buffer species have been tried.10-14 The quality of the pH gradient generated using these low molecular mass buffer components has been generally poor compared to that generated by polymeric ampholyte buffers, giving nonlinear gradients, cascade steps,10,13 spikes,4 and protein elution plateaus.4,11 The second limitation of the conventional chromatofocusing technique concerns the concentration of the buffer in the mobile phase. Relatively low buffer concentrations are required in the conventional chromatofocusing technique to generate pH gradient slopes that are not too steep, to obtain reasonable resolution.2,3,10,15 Hearn and Lyttle10 found a buffer concentration range of 2.5-5 mM Buffalyte WR 3-10 on a DEAE cellulose column produced suitable pH gradients for conventional chromatofocusing analysis. A high concentration of buffer in the mobile phase produces too steep a pH gradient slope, as the mobile-phase buffering capacity overwhelms the buffering capacity of the column. Thus, possible optimization gains in chromatofocusing separations using high buffer concentrations in the mobile phase cannot be studied/ realized in the conventional chromatofocusing technique. The third limitation in conventional chromatofocusing is its inflexibility in controlling pH gradient slope. The only way to control the pH gradient slope for a given column in conventional (6) Hutchens, T. W. In Protein Purification; Janson J., Ryden, L., Eds.; VCH: New York, 1989; Chapter 5. (7) Bates, R. C.; Frey, D. D. J. Chromatogr., A 1998, 814, 43-54. (8) Scott, J. H.; Kelner, K. L.; Pollard, H. B. Anal. Biochem. 1985, 149, 163165. (9) Rodkey, L. S.; Hirata, A. Protides Biol. Fluids 1988, 34, 745-748. (10) Hearn, M. T. W.; Lyttle, D. J.J. Chromatogr. 1981, 218, 483-495. (11) Wagner, G.; Regnier, F. E. Anal. Biochem. 1982, 126, 37-43. (12) Hutchens, T. W.; Li, C. M.; Besch, P. K. Protides Biol. Fluids 1986, 34, 765-768. (13) Hutchens, T. W.; Li, C. M.; Besch, P. K. J. Chromatogr. 1986, 359, 157168. (14) Hutchens, T. W.; Li, C. M.; Besch, P. K. J. Chromatogr. 1986, 359, 169179. (15) Hjerte´n, S.; Li. J.-P. J. Chromatogr. 1989, 475, 167-175.
Analytical Chemistry, Vol. 74, No. 21, November 1, 2002 5641
chromatofocusing is to change the mobile-phase buffer concentration, with lower mobile-phase buffer concentrations giving shallower pH gradients. However, linear pH gradients are difficult to generate using very low buffer concentrations, as it has been reported that low-concentration polymeric ampholyte buffers (diluted more than 1:10) give irregular and irreproducible pH gradients.11,16 Frey et al.17-19 have been able to generate smooth linear and concave pH gradients using step changes between buffers containing positively or neutrally charged low molecular mass amine buffer components on anion-exchange columns. However, anomalies were noted in the pH range 5.5-6.0, limiting the procedure to pH ranges of 9.5-6.0 and 5.5-4.0.17 In addition, this technique was limited to lower concentration buffers (2-5 mM) in order to generate reasonable pH gradient slopes. The above limitations have been overcome by the gradient chromatofocusing technique developed by our group,20-22 which utilizes common buffer species having low molecular mass and a gradient pump system to gradually change the proportion of application to elution buffers. The use of common low molecular mass buffers addresses all the limitations of the polyampholyte buffers. The use of a gradient pump system allows a gradual pH change to be delivered to the inlet of the column, as opposed to conventional chromatofocusing in which there is a precipitous change of 1-3 pH units introduced to the column. The pH gradient slopes can be readily controlled in gradient chromatofocusing by changing the gradient program. The outlet pH gradient (at the column outlet) will roughly mirror the inlet pH gradient (at the column inlet), although it will be shifted in time due to the buffering action of the column. This also makes possible the use of a high concentration of buffers in the mobile phase, since this high concentration of buffers can be delivered to the column with a gradual change in pH (instead of a step change as in the conventional chromatofocusing technique), which does not overwhelm the buffering capacity of the column. The present work directly compares the results of gradient chromatofocusing with conventional chromatofocusing employing a Mono P anion-exchange column. The advantages of gradient chromatofocusing over conventional chromatofocusing are demonstrated, as significant optimization in gradient chromatofocusing is attained by increasing buffer concentration and controlling pH gradient slope, parameters that cannot be manipulated in the conventional chromatofocusing technique. EXPERIMENTAL SECTION Materials. Conalbumin (from chicken egg white, Catalog No. C-0755), bovine serum albumin (BSA; Catalog No. A-7906), ovalbumin (from chicken egg, Catalog No. A-2512), β-lactoglobulin A (from bovine milk, Catalog No. L-7880), β-lactoglobulin B (from bovine milk, Catalog No. L-8005), bis-tris propane (>99%, Catalog No. B-9410), piperazine (Catalog No. P3896), bis-tris (>98%, (16) Li, C. M.; Hutchens, T. In Methods in Molecular Biology, Practical Protein Chromatography; Kenny, A., Fowell, S., Eds.; Humana Press: Towata, NJ, 1992; , Vol. 11, Chapter 15. (17) Bates, R. C.; Kang, X.; Frey, D. D. J. Chromatogr., A 2000, 890, 25-36. (18) Kang, X.; Bates, R. C.; Frey, D. D. J. Chromatogr., A 2000, 890, 37-43. (19) Kang, X.; Frey, D. D. Anal. Chem. 2002, 74, 1038-1045. (20) Liu, Y.; Anderson, D. J. J. Chromatogr., A 1997, 762, 207-217. (21) Liu, Y.; Anderson, D. J. J. Chromatogr., A 1997, 762, 47-54. (22) Shan, L.; Anderson, D. J. J. Chromatogr., A 2001, 909, 191-205.
5642
Analytical Chemistry, Vol. 74, No. 21, November 1, 2002
Catalog No. B-7535), iminodiacetic acid (Catalog No. I-0774), Na2HPO4 (>99%, Catalog No. S-7907), and NaCl (> 99%, Catalog No. S-7653) were from Sigma Chemical Co. (St. Louis, MO). Glacial acetic acid (99.7%, Catalog No. 3-9507) was from J. T. Baker (Phillipsburg, NJ), lactic acid (75%, Catalog No. 02634) was from Merck (Whitehouse Station, NJ), chloroacetic acid (>99%, Catalog No. 40,292-3), pyridine (99.9+%, HPLC grade, Catalog No. 27,040-7), and 2-methylpiperazine (95%, Catalog No. M7,240-4) were from Aldrich (Milwaukee, WI), and H3PO4 (Catalog No. A242212) and HCl (Catalog No. A144-212) were from Fisher Scientific (Pittsburgh, PA). Polybuffer 74 (Catalog No. 17-0713-01) was from Amersham Biosciences (Piscataway, NJ).The protein sample solutions were prepared with a 20.0 mM NaH2PO4-Na2HPO4, pH 7.00, buffer solution (prepared by adding concentrated H3PO4 to 20.0 mM Na2HPO4) and aliquots stored frozen at -20 °C until use. All solutions were aqueous, prepared with HPLC grade water. Chromatographic Setup. The same HPLC system and procedures were used for the conventional and the gradient chromatofocusing experiments, as well as the salt gradient runs, as described below. The HPLC system consisted of a System Gold 127 solvent module gradient system from Beckman Instruments (Fullerton, CA), a SP4270 integrator from Thermo Finnigan (San Jose, CA), a variable-wavelength detector from Dionex (Sunnyvale, CA), and an Epson Equity I+ personal computer. A Rheodyne model 7125 injection valve from Rainin (Woburn, MA) was used, employing a 500-µL injection loop. A 5-cm Mono P HR 5/5 column (Catalog No. 17-0611-01) and a prefilter (Catalog No. 19-5084-010), placed between the injection valve and the Mono P column for protecting the column, were purchased from Amersham Biosciences. The volume from the gradient valve to the column was ∼2.8 mL (including the 500-µL injection loop). The void volume of the mono P column was 0.4 mL (provided by the manufacturer). The flow rate of the mobile phase was 1.0 mL/min. Absorbance was monitored at 280 nm. The column was equilibrated with the application buffer prior to the start of each run (usually 30 min) until the pH of the column eluant was the same as that of the applied application buffer. Gradient Chromatofocusing Runs. Buffer A (high pH application buffer) consisted of bis-tris propane and piperazine, adjusted to pH 7.60 with concentrated HCl. Buffer B (low-pH elution buffer) consisted of pyridine, acetic acid, lactic acid, and chloroacetic acid (no pH adjustment). Different concentrations of application and elution buffers were used, with the concentration of each buffer component being the same in both the 100% application and 100% elution buffer for a particular run, being either 6.25, 12.5, 25.0, 50.0, 75.0, or 100 mM. That is, for the 6.25 mM runs, buffer A was 6.25 mM bis-tris propane and 6.25 mM piperazine and buffer B was 6.25 mM pyridine, 6.25 mM acetic acid, 6.25 mM lactic acid, and 6.25 mM chloroacetic acid. For the 12.5 mM runs, all the buffer components in buffer A and buffer B were 12.5 mM, and so forth. The pH gradients in the gradient chromatofocusing runs were produced by gradient programs found by trial and error, by measuring the eluent pH and empirically modifying the program until the desired pH gradient was obtained (see Supporting Information for programs used). Conventional Chromatofocusing. The manufacturer’s pH 5-4 protocol (Amersham Biosciences) was used: buffer A (highpH application buffer) was 25 mM methylpiperazine hydrochlo-
Figure 1. Titration of Mono P column with 0.001 M HCl at 1.00 mL/min flow rate. The column was conditioned with 0.01 M NaOH, 0.01 M HCl, and 0.01 M NaOH, as given in the Experimental Section.
ride, pH 5.7, and buffer B (low-pH elution buffer) was 1:10 diluted Polybuffer 74-HCl, pH 4.0. Two other manufacturer protocols were also tried (pH 5.5-4.5 and pH 7-4)- however, the pH 5-4 mobilephase system gave the best separation (See Supporting Information). The pH gradients were produced by a step change from 100% A to 100% B. Other Procedures. The pH of 1.0- and 2.0-min fractions was determined for conventional and gradient chromatofocusing techniques, respectively. Resolution (Rs), plate number (N), and separation factor (R) were calculated as reported previously.22 Solvent peaks were present in the chromatograms due to the presence of pyridine in the mobile phase. These solvent peaks were corrected for by subtracting the baseline of a blank run (without sample injection) from that of the sample runs. The titration procedure for the Mono P column was as follows (flow rate was 1.00 mL/min): (1) a 0.01 M NaOH solution was pumped through the column until the pH of the column effluent reached ∼12, at which point the pumping of 0.01 M NaOH was continued for an additional 30 min; (2) the column was then washed with a 0.01 M HCl solution until the pH of the column effluent reached ∼2, at which point the pumping of 0.01 M HCl was continued for an additional 30 min; (3) the column was then washed with 0.01 M NaOH solution again until the column effluent reached ∼12. The mobile phase was then step changed to a 0.001 M HCl solution for 120 min. Fractions of the inlet and outlet mobile phases were collected every 2 min and the pH was measured. RESULTS AND DISCUSSION Buffering pH Range of the Mono P Column. The principal buffering range of the Mono P column was determined to be 10.59, with only a relatively small amount of buffering capacity at pHs below 9 (see Figure 1). Linear pH gradients can still be produced below pH 9 using this column, as was done in the present work. This fact demonstrates that the generation of linear pH gradients in chromatofocusing does not require strong inherent buffering capacity of the column in the pH range of the gradient. In fact, others7,23-29 have shown that adsorbed buffer components from (23) Murel, A.; Vilde, S.; Pank, M.; Shevchuk, I.; Kirret, O. J. Chromatogr. 1985, 347, 325-334 (24) Murel, A.; Vilde, S.; Pank, M.; Shevchuk, I.; Kirret, O. J. Chromatogr. 1986, 362, 101-112. (25) Helfferich, F. G.; Bennett, B. J. React. Polym., Ion Exch., Sorbents 1984, 3, 51-66. (26) Helfferich, F. G.; Bennett, B. J. Solvent Extr. Ion Exch. 1984, 2, 1151-84. (27) Frey, D. D. Biotechnol. Prog. 1996, 12, 65-72.
the mobile phase are a central aspect of the mechanism by which the column participates in the generation of linear pH gradients. Hearn and Lyttle10 proposed a displacement mechanism for conventional chromatofocusing, which is the same as the mechanism of ampholyte displacement chromatography.30 This mechanism proposes the establishment of a pH gradient within a column via a gradient distribution of different buffer components within the ion-exchange column, with the proportion of strongly to weakly bound buffer components (more ionized to less ionized components) steadily decreasing down the length of the column. Buffer components move down the column, with the successively stronger affinity components (greater degree of ionization) being retained and moving down the column at successively lower velocities than the comparatively weaker affinity buffer components (lower degree of ionization). The stronger affinity buffer components effectively displace in succession weaker affinity components. Thus, for anion-exchange chromatography, the pH of the column is continuously lowered as successively more acidic components are eluted from the column, producing a continuous decreasing pH gradient. The results of Figure 1 at least bring up the possibility of other factors (besides column buffering capacity) playing a role in pH gradient generation, such as the aforementioned ionic displacement. However, an ionic displacement mechanism cannot explain pH gradient generation at the beginning of the gradient (at high pHs) in the present work, as the amine buffer components of bistris propane, piperazine, and pyridine are either positively or neutrally charged and thus would not be expected to be retained on the column. Frey et al. have derived equations for nonabsorbed buffer species in conventional chromatofocusing and used them in a computer model to predict experimentally generated linear or concave pH gradients.17,19 The mathematical model assumes a mechanism in which the ion-exchanger buffers a pH change of the mobile phase to produce the pH gradients. These authors have experimentally shown that pH gradients can be generated on a Mono P column using a step change in buffers consisting exclusively of amine components,17,18 indicating some adsorptive/ buffering role of the Mono P column. Figure 1 does show some, albeit small, buffering capacity of the Mono P column at pHs
