BIOTECHNOLOGY AND BIOENGINEERING
VOL. XVIII (1976)
Purijication of Glucose Isomerase by A f i n i t y Chromatography For some years purification of enzymes by affinity chromatography has been held to be the ultimate process, to be used whenever possible in the laboratory and even to be tested on a commercial scale. Since its reduction to practice by Cuatrecasas and Wilchek,' it has been perfected by Cuatrecasas and co-workers2 and used successfully on a variety of different kinds of systems. Affinity chromatography was recently reviewed by May and Zaborskya and Nishikawa4 and is covered in detail in the volume edited by Jakoby and W i l ~ h e k . ~ To date no reports have been issued from industrial laboratories announcing its successful use. On the contrary, in the only report from commerce so far, Hsu et a1.6 opined on the basis of their work that it was likely to be too expensive for the purification of a desired enzyme, but would be very useful in removal of trace impurities such as proteases. The object of the work reported here was to develop an affinity chromatography system for the purification of glucose isomerase. This enzyme, which catalyzes the reversible conversion between glucose and fructose, is being widely used commercially in the starch industry. Standard inhibition studies were run on several possible candidates for the glucose isomerase reaction using fructose as a substrate. The best inhibitor studied was xylitol which had a value of K , = 5 X 10-3M for the production of glucose from fructose a t 60°C, pH 6.8 in 0.01M succinate in the presence of cobalt and magnesium.' This K , is not as small as is recommented in the literature2 but we chose it anyway for preliminary work. Sepharose 4B was activated with CNBr followed by successive reactions with l,6-hexanediamine and succinic anhydride according to the method of Cuatrecasas and co-workers.2 Finally, xylitol was added using dicyclohexylcarbodiimide (DCC) as a condensing agent to form a final affinity adsorbent. The derivatized Sepharose was then placed in a glass column 0.6 cm i.d. X 5 cm and the temperature of the column was maintained a t 2°C with ice water. Before each run, the column was flushed thoroughly with 2M NaC1, distilled water, and, finally, the buffer solution to be used in the run. The flow rate used in irrigation and elution was 0.76 ml/min, which is about 0.6 column volumes per minute or a superficial velocity of 3 cm/min. In each run a quantity of sample was added which was usually just less than the amount that caused activity to appear in the effluent in a run in which activity is retained. Each 1.5 ml of effluent was analyzed for protein and enzyme activity. Figure 1 shows the chromatogram obtained. Some protein showing no activity washed through immediately while the bulk of the protein was retained. All of the protein that bound to the column was then flushed off with NaCl, which is a rapid and complete means of cleaning the column, although nonspecific. The specific activity of fraction number 23 was 1.5 times that of the original enzyme preparation. The enzyme could also be rapidly eluted from the column with 0.4M succinate a t pH 7.0. 1639 @ 1976 by John Wiley & Sons, Inc.
1640
BIOTECHNOLOGY AND BIOENGINEERING VOL. XVIII (1976)
- 0.16 -€%- 2 .E
0.8
0.12
0.6
0.08
0.4
fl .E 2- 0.04
0.2
c g 0
2-
.;
!.:O D n
a 0.00
4
8
12 Effluent
16 20 24 (Column Volumes 1
28
0.o
Fig. 1 . Chromat.ogram with concentrated feed. 0.05M,pH 7.4 succinate buffer. 24"C, column was 0.5 X 4 cm. (0-0) Protein; (0-0) enzyme activity. Xylitol was also used to elute the enzyme retained in the column but the rate of elution was so slow that the method was considered impractical. This very slow rate of elution was probably caused by the very low equilibrium free enzyme concentrat,ions next to the solid adsorbent, which will keep the overall rate of desorption very low. As another test of specificity of the affinity adsorbent, a synthetic mixture of bovine serum albumin and the enzyme preparation was applied to the affinity column. The result as shown in Figure 2 shows that bovine serum albumin went, right, through the column without adsorbing and the enzyme
Effluent
Column Volumes )
Fig. 2. Separation of bovine serum albumin from glucose isomerase. 0.05M, pH 7.3 succinate buffer, 0"C, column was 0.5 X 5 cm. Feed was 50 mi enzyme Protein; solution (400 mg protein) and 2.5 mg bovine serum albumin. (0-0) (0-0) enzyme activity.
COMMUNICATIONS TO T H E EDITOR
1641
activity appears to have been retained as usual. The specific activity of fraction number 23 was over nine times that of the synthetic mixture applied to the column. Thus, reasonable increases in purity could be obtained with dilute feeds to the column. This result may be far from a demonstration that affinity is strong enough to outweigh ionic attraction for the bovine serum albumin. Bovine serum albumin has a net negative charge, there are positive charges on the derivatized Sepharose, and given a large enough residence time, bovine serum albumin would probably adsorb to the column. What the test. shows is that the affinity is strong enough to outweigh the ionic att,raction with t,his small residence time. This kinetic result may be worthwhile using in practical applications, keeping contact time short so that. nonspecific interactions are minimized. Several control column runs were also made using either the Sepharose with shorter spacer arm or intermediate Sepharose derivatives as adsorbent materials. When xylitol was attached directly to Sepharose via hexanoic acid and run in a column, the leakage of the enzyme activity was pronounced. This may be due to a decrease in hydrophobicity7 of t,he affinity adsorbent and/or an increased steric hindrance for the enzyme in adsorbing to the matrix. When hexanoic acid alone was attached to Sepharose and run as a control column, the enzyme activity was not. retained a t all. This is consistent with the idea that the positively charged group resulting from CNBr activation, namely the isourea linkage, was counterbalanced by the negatively charged carboxyl group and the hydrophobic force alone could not retain the enzyme. But the situation became quite different. for a hexanediamine-succinic anhydride derivatized Sepharose, which ought to be t,he same as the former with respect to charge effect, where a portion of the activity was retained. Since there is no net charge in the adsorbent, the above result. was caused by the increased hydrophobicity of the adsorbent. Sepharose derivatiaed with hexanediamine was also tested. The column was run a t pH 7, which is the same as other runs, and a t this pH this adsorbent is expected to have two positive charges per link. In this case, most of the protein was retained in the column just like the final affinity adsorbent but the enzyme activity could not be eluted by flushing the column with NaCI, although some protein having no enzyme act.ivity appeared in the effluent. HofsteeQhas shown that hydrophobic groups will adsorb many proteins, and that 3M NaCl often will not elute the protein from the column. Thus, both the hydrophobic ligand and the double charge effect seem to be responsible for this phenomenon. From the control experiments, it is apparent that the final affinity adsorbent would exert not only the specific interaction but also ionic and hydrophobic bindings with the enzyme. When there is a good ligand for a specific enzyme to be separated, it would be better to eliminate forces other than biospecificity to minimize nonspecific bindings in preparing an affinity adsorbent. On the contrary, if only a weak ligand is available, as in our case, it appears useful to reinforce t,he biospecificity with other forces. The ninefold purification obtained wit,h the dilute glucose isomerasebovine serum albumin feed appears to have resulted from combined biospecific, ionic, and hydrophobic forces. The balance of these three forces is important ; increasing the hydrophobicity of the adsorbent would eventually cause adsorption of bovine serum albumin? and decrease the separation. Although the weak biospecific forces alone are not sufficient to purify the glucose isomerase, they appear to be necessary in the purification.
1642
BIOTECHNOLOGY AND BIOENGINEERING VOL. XVIII (1976)
Large purification factors will only be achieved when the initial concentration of desired enzyme is very low. Economic considerations show that the expensive adsorbent must be used efficiently to have an economically competitive process. This would require an extremely fast adsorption-desorption cycle for concentrated enzyme solutions. Thus, in agreement with Hsu et a1.,6 it appears that economic use of affinity chromatography will result when very low concentrations of the enzyme are either being recovered as the desired product or removed as a contaminant. References 1. P. Cuatrecasas and hl. Wilchek, Biochem. Riophys. Res. Commun., 33, 235 (1968). 2 . P. Cuatrecasas and C. B. Anfinsen, in Methods i n Enzymology, vol. 22, W. B. Jakoby, Ed., Academic Press, New York, 1971, pp. 345-378. 3. S. W. May and 0. R. Zaborsky, Separ. Purific. Methods, 3, 1 (1974). 4. A. H. Nishikawa, Chem. Technol., 5 , 564 (197.5). 5 . W. B. Jakoby and M. Wilchek, Eds., in Methods i n Enzymology, vol. 34, Academic Press, New York, 1974. 6. L. H. Hsu, R . M. Flora, and H. It. Bungay, paper presented a t ACS Symposium on Enzyme and Antibody Engineering, Purdue University, January 23, 1974. 7. Y. H. Lee, M.S. Thesis, Purdue University, Indiana 1974. 8. S. Shaltiel and 2.Er-el, Proc. Nal. Acad. Sci. U S A , 70, 778 (1973). 9. B. H. J. Hofstee, Prep. Riochem., 5 , 7 (197.5).
Y. H. LEE P. C. WANKAT A. H. EMERY School of Chemical Engineering Purdue University West Lafayette, Indiana 47907 Accepted for Publicat,ion June 29, 1976
VOL. XVIII (1976)
Purijication of Glucose Isomerase by A f i n i t y Chromatography For some years purification of enzymes by affinity chromatography has been held to be the ultimate process, to be used whenever possible in the laboratory and even to be tested on a commercial scale. Since its reduction to practice by Cuatrecasas and Wilchek,' it has been perfected by Cuatrecasas and co-workers2 and used successfully on a variety of different kinds of systems. Affinity chromatography was recently reviewed by May and Zaborskya and Nishikawa4 and is covered in detail in the volume edited by Jakoby and W i l ~ h e k . ~ To date no reports have been issued from industrial laboratories announcing its successful use. On the contrary, in the only report from commerce so far, Hsu et a1.6 opined on the basis of their work that it was likely to be too expensive for the purification of a desired enzyme, but would be very useful in removal of trace impurities such as proteases. The object of the work reported here was to develop an affinity chromatography system for the purification of glucose isomerase. This enzyme, which catalyzes the reversible conversion between glucose and fructose, is being widely used commercially in the starch industry. Standard inhibition studies were run on several possible candidates for the glucose isomerase reaction using fructose as a substrate. The best inhibitor studied was xylitol which had a value of K , = 5 X 10-3M for the production of glucose from fructose a t 60°C, pH 6.8 in 0.01M succinate in the presence of cobalt and magnesium.' This K , is not as small as is recommented in the literature2 but we chose it anyway for preliminary work. Sepharose 4B was activated with CNBr followed by successive reactions with l,6-hexanediamine and succinic anhydride according to the method of Cuatrecasas and co-workers.2 Finally, xylitol was added using dicyclohexylcarbodiimide (DCC) as a condensing agent to form a final affinity adsorbent. The derivatized Sepharose was then placed in a glass column 0.6 cm i.d. X 5 cm and the temperature of the column was maintained a t 2°C with ice water. Before each run, the column was flushed thoroughly with 2M NaC1, distilled water, and, finally, the buffer solution to be used in the run. The flow rate used in irrigation and elution was 0.76 ml/min, which is about 0.6 column volumes per minute or a superficial velocity of 3 cm/min. In each run a quantity of sample was added which was usually just less than the amount that caused activity to appear in the effluent in a run in which activity is retained. Each 1.5 ml of effluent was analyzed for protein and enzyme activity. Figure 1 shows the chromatogram obtained. Some protein showing no activity washed through immediately while the bulk of the protein was retained. All of the protein that bound to the column was then flushed off with NaCl, which is a rapid and complete means of cleaning the column, although nonspecific. The specific activity of fraction number 23 was 1.5 times that of the original enzyme preparation. The enzyme could also be rapidly eluted from the column with 0.4M succinate a t pH 7.0. 1639 @ 1976 by John Wiley & Sons, Inc.
1640
BIOTECHNOLOGY AND BIOENGINEERING VOL. XVIII (1976)
- 0.16 -€%- 2 .E
0.8
0.12
0.6
0.08
0.4
fl .E 2- 0.04
0.2
c g 0
2-
.;
!.:O D n
a 0.00
4
8
12 Effluent
16 20 24 (Column Volumes 1
28
0.o
Fig. 1 . Chromat.ogram with concentrated feed. 0.05M,pH 7.4 succinate buffer. 24"C, column was 0.5 X 4 cm. (0-0) Protein; (0-0) enzyme activity. Xylitol was also used to elute the enzyme retained in the column but the rate of elution was so slow that the method was considered impractical. This very slow rate of elution was probably caused by the very low equilibrium free enzyme concentrat,ions next to the solid adsorbent, which will keep the overall rate of desorption very low. As another test of specificity of the affinity adsorbent, a synthetic mixture of bovine serum albumin and the enzyme preparation was applied to the affinity column. The result as shown in Figure 2 shows that bovine serum albumin went, right, through the column without adsorbing and the enzyme
Effluent
Column Volumes )
Fig. 2. Separation of bovine serum albumin from glucose isomerase. 0.05M, pH 7.3 succinate buffer, 0"C, column was 0.5 X 5 cm. Feed was 50 mi enzyme Protein; solution (400 mg protein) and 2.5 mg bovine serum albumin. (0-0) (0-0) enzyme activity.
COMMUNICATIONS TO T H E EDITOR
1641
activity appears to have been retained as usual. The specific activity of fraction number 23 was over nine times that of the synthetic mixture applied to the column. Thus, reasonable increases in purity could be obtained with dilute feeds to the column. This result may be far from a demonstration that affinity is strong enough to outweigh ionic attraction for the bovine serum albumin. Bovine serum albumin has a net negative charge, there are positive charges on the derivatized Sepharose, and given a large enough residence time, bovine serum albumin would probably adsorb to the column. What the test. shows is that the affinity is strong enough to outweigh the ionic att,raction with t,his small residence time. This kinetic result may be worthwhile using in practical applications, keeping contact time short so that. nonspecific interactions are minimized. Several control column runs were also made using either the Sepharose with shorter spacer arm or intermediate Sepharose derivatives as adsorbent materials. When xylitol was attached directly to Sepharose via hexanoic acid and run in a column, the leakage of the enzyme activity was pronounced. This may be due to a decrease in hydrophobicity7 of t,he affinity adsorbent and/or an increased steric hindrance for the enzyme in adsorbing to the matrix. When hexanoic acid alone was attached to Sepharose and run as a control column, the enzyme activity was not. retained a t all. This is consistent with the idea that the positively charged group resulting from CNBr activation, namely the isourea linkage, was counterbalanced by the negatively charged carboxyl group and the hydrophobic force alone could not retain the enzyme. But the situation became quite different. for a hexanediamine-succinic anhydride derivatized Sepharose, which ought to be t,he same as the former with respect to charge effect, where a portion of the activity was retained. Since there is no net charge in the adsorbent, the above result. was caused by the increased hydrophobicity of the adsorbent. Sepharose derivatiaed with hexanediamine was also tested. The column was run a t pH 7, which is the same as other runs, and a t this pH this adsorbent is expected to have two positive charges per link. In this case, most of the protein was retained in the column just like the final affinity adsorbent but the enzyme activity could not be eluted by flushing the column with NaCI, although some protein having no enzyme act.ivity appeared in the effluent. HofsteeQhas shown that hydrophobic groups will adsorb many proteins, and that 3M NaCl often will not elute the protein from the column. Thus, both the hydrophobic ligand and the double charge effect seem to be responsible for this phenomenon. From the control experiments, it is apparent that the final affinity adsorbent would exert not only the specific interaction but also ionic and hydrophobic bindings with the enzyme. When there is a good ligand for a specific enzyme to be separated, it would be better to eliminate forces other than biospecificity to minimize nonspecific bindings in preparing an affinity adsorbent. On the contrary, if only a weak ligand is available, as in our case, it appears useful to reinforce t,he biospecificity with other forces. The ninefold purification obtained wit,h the dilute glucose isomerasebovine serum albumin feed appears to have resulted from combined biospecific, ionic, and hydrophobic forces. The balance of these three forces is important ; increasing the hydrophobicity of the adsorbent would eventually cause adsorption of bovine serum albumin? and decrease the separation. Although the weak biospecific forces alone are not sufficient to purify the glucose isomerase, they appear to be necessary in the purification.
1642
BIOTECHNOLOGY AND BIOENGINEERING VOL. XVIII (1976)
Large purification factors will only be achieved when the initial concentration of desired enzyme is very low. Economic considerations show that the expensive adsorbent must be used efficiently to have an economically competitive process. This would require an extremely fast adsorption-desorption cycle for concentrated enzyme solutions. Thus, in agreement with Hsu et a1.,6 it appears that economic use of affinity chromatography will result when very low concentrations of the enzyme are either being recovered as the desired product or removed as a contaminant. References 1. P. Cuatrecasas and hl. Wilchek, Biochem. Riophys. Res. Commun., 33, 235 (1968). 2 . P. Cuatrecasas and C. B. Anfinsen, in Methods i n Enzymology, vol. 22, W. B. Jakoby, Ed., Academic Press, New York, 1971, pp. 345-378. 3. S. W. May and 0. R. Zaborsky, Separ. Purific. Methods, 3, 1 (1974). 4. A. H. Nishikawa, Chem. Technol., 5 , 564 (197.5). 5 . W. B. Jakoby and M. Wilchek, Eds., in Methods i n Enzymology, vol. 34, Academic Press, New York, 1974. 6. L. H. Hsu, R . M. Flora, and H. It. Bungay, paper presented a t ACS Symposium on Enzyme and Antibody Engineering, Purdue University, January 23, 1974. 7. Y. H. Lee, M.S. Thesis, Purdue University, Indiana 1974. 8. S. Shaltiel and 2.Er-el, Proc. Nal. Acad. Sci. U S A , 70, 778 (1973). 9. B. H. J. Hofstee, Prep. Riochem., 5 , 7 (197.5).
Y. H. LEE P. C. WANKAT A. H. EMERY School of Chemical Engineering Purdue University West Lafayette, Indiana 47907 Accepted for Publicat,ion June 29, 1976
