Effects of Solution Properties on Solute and Permeate Flux in Bovine

May 15, 1994 - Russell G. Nel, S. F. Oppenheim, and V. G. J. Rodgers*. Department of Chemicaland Biochemical Engineering, The University of Iowa, 125 ...
0 downloads 0 Views 436KB Size
Biotechnol. frog. 1994, 10, 539-542

539

NOTES Effects of Solution Properties on Solute and Permeate Flux in Bovine Serum Albumin-IgG Ultrafiltration Russell G. Nel, S. F. Oppenheim, and V. G. J. Rodgers* Department of Chemical and Biochemical Engineering, The University of Iowa, 125 C-B, Iowa City, Iowa 52242-1219

It has been observed that bovine serum albumin (BSA, 69 000 Da) exhibits high rejection in a pH 7.4,0.15 M NaCl solution containing low concentrations of immunoy-globulins (16,155 000 Da). Although it is apparent that this phenomenon is likely due to the complex interactions of the proteins as well as to simple steric hindrance, it is not clear to what extent these resistances contribute to the solute flux loss. This study investigates how variations in IgG concentration, solution pH, and ionic strength can affect the solute and the permeate fluxes for the ultrafiltration of BSA through 100 000 MWCO cellulosic membranes in a batch cell. The results showed that, unlike cases performed at pH 7.4 in 0.15 M NaC1, the presence of IgG may increase the transport of BSA under certain conditions. This study does show the potential value and insights that will result from further binary protein ultrafiltration research.

Introduction A significant amount of attention has focused on the membrane separation of bovine serum albumin and IgG from both a basic research and biomedical interest point of view (e.g., Blatt et al., 1970; Porter, 1979; Rodgers and Sparks, 1991). It has been observed that bovine serum albumin (BSA, 69 000 Da) exhibits high rejection in a physiological solution of pH 7.4,0.15 M NaCl containing low concentrations of immuno-y-globulins (IgG, 155 000 Da). During the ultrafiltration process with 100 000 MWCO and larger pore size membranes, the BSA permeate concentration was found to decrease due to the introduction of IgG (Blatt et al., 1970;Porter, 1979;Miller et al., 1993). Although it is apparent that this phenomenon is likely due to the complex interactions of the proteins as well as to simple steric hindrance, it is not clear to what extent these resistances contribute to the solute flux loss. Since solute-solute and solute-membrane interactions are dependent on solution properties, if simple steric hindrance, caused by the presence of the larger macromolecule IgG, was the dominant factor, then it would not be expected that solution property variations would significantly alter this observation. Because of the inherent complexity on the subject, very little research has investigated the effects of solution properties on multiple solute ultrafiltration (Ingham et al., 1980). In particular, the effect of solution property variations on solutions containing both IgG and BSA has not been studied in detail. This study investigates variations in IgG concentration, solution pH, and ionic strength that can affect the solute and the permeate fluxes for the ultrafiltration of BSA through 100000 MWCO cellulosic membranes in a batch cell. In order to reduce pore plugging, transmembrane pressure pulsing was employed (Rodgers and Sparks, 1991; Miller et al., 1993).

* Author to whom correspondence should be addressed. 8756-7938/94/3010-0539$04.50/0

Experimental Procedures Materials and Methods. The proteins used in this study were bovine serum albumin (BSA, fraction V powder, purity of 96-99% by gel electrophoresis, Sigma Chemical Co., St. Louis, MO, No. A-4503, Lot No. 60H0018) and immuno-y-globulin (IgG, fraction V powder, purity of approximately 99% by electrophoresis, Sigma Chemical Co., No. G-7516, Lot No. 107F-935611. These are the same protein fractions used in the study of Miller et al. (19931, which used physiological solution properties in the transmembrane pressure pulsed separation of these two species. All solutions were prepared in a phosphate-buffered saline solution (powder, Sigma, No. 1000-31, with 0.02 g/lOO mL sodium azide (powder, Sigma, No. S-2002, Lot No. 29F-0534) added as a preservative. The p H s of the solutions were adjusted to the desired values by the addition of small amounts of 0.1 M NaOH or HCl. Protein solutions were carefully prepared using ultrafiltered deionized water with a resistivity of 18.3 M k m . The problem of bacterial contamination was minimized by storing all prepared solutions at 10 "C. All solutions were used within 4 days after preparation. Determination of Isoelectric Points. The nondenatured pZ values of BSA and IgG were determined by isoelectric focusing electrophoresis using a Hoefer SE 250 vertical electrophoresis cell. The cooling temperature was maintained constant at 10 "C. The techniques of Robertson et al. (1987) and Coulson (1981) were employed. The BSA was run on a 5% acrylamide matrix. IgG was examined on both 5% acrylamide and 1% isoelectric focusing agarose gels. Several runs were performed for each protein. After each run, the pH gradients of the gels were examined using a surface pH probe to ensure linearity. The pZ values were determined from linear interpolation of known samples (Pharmacia LKB Biotechnology); pH ranges were between 2.5 and

0 1994 American Chemical Society and American Institute of Chemical Engineers

Biotechnol. frog., 1994,Vol. 10, No. 5

540 Table 1. Study Results results

vH 5.8 5.8 5.8 5.8 5.8 5.8 8.5 8.5 8.5 8.5 8.5 8.5

solution properties [NaClI (M) % IeG 0.01 0.00 0.07 0.01 0.01 0.15 0.15 0.00 0.15 0.07 0.15 0.15 0.01 0.00 0.01 0.07 0.01 0.15 0.15 0.00 0.15 0.07 0.15 0.15 I

permeate flux ( x lo4)at 3000 s 0.69f 0.06 0.44f 0.05 0.44f 0.06 0.33 f 0.08 0.23f 0.08 0.42f 0.05 0.03f 0.01 0.33f 0.09 0.59f 0.09 0.45 f 0.07 0.28f 0.20 0.53f 0.21

6.5 and between 5.0 and 10.5. Coomassie Brilliant Blue R-250 was used to visualize the gels. The results for BSA showed a PIrange between 4.3 and 4.5 with no noticeable impurities. Two distinct bands were formed. These bands have been observed by others (Sluyterman and Wijdenes, 1978) and can be attributed to the structural shifts around the PIof BSA (Peters, 1985). IgG showed numerous bands on both types of gels. The agarose gel showed bands between pH 5.2 and 10.0 while the acrylamide gel showed pH bands between 5.2 and 9.2, with a n error in pH of fO.l. The most prominent bands for IgG appeared in the agarose gel at pH values of 5.8 and 8.5. Since these are the most dominant isoelectric points for the IgG used, they were selected as the pH values used in this investigation. Protein Analysis. BSA concentrations were determined spectrophotometrically using a bromcresol green (BCG, Sigma Chemicals) reagent. A spectrophotometer (Model UV-1201, Shimadzu, Japan) was used to determine the sample absorbance. This resulted in an error in BSA concentration of f0.007 w t %. IgG concentrations were determined, using a total globulin determination reagent (Sigma Diagnostics, Lot No. 72H6112). The uncertainty in IgG concentration from this technique was determined to be f0.006 wt %. Experimental Apparatus. For all of the experimental work, cellulosic membranes with 100 000 MWCO (Amicon Corp., Lexington, MA) were used for analysis. These membranes were soaked in distilled water for 30 min, to remove glycerin, before usage. Hydraulic permeability tests were performed before and after each experimental run. The transmembrane pressure pulsed batch cell apparatus is summarized elsewhere (Miller et al., 1993). Transient permeate flux, species flux, and solute permeate concentration were determined for the ultrafiltration of binary protein solutions under the conditions of negative transmembrane pressure pulsing. Hydraulic permeability studies were also performed before and &r each run to evaluate the temporal physical characteristics of the membrane. All binary solutions were prepared with 1 w t % BSA. Each run was performed twice with different membranes. The runs were randomly selected in the sequence. The results were averaged, and the error used is based on the deviations of the two values from their average.

Results and Discussion Hydraulic Permeability. For fresh membranes, the hydraulic permeability was determined to be (1.31 f 0.102) x cm s-l kPa-l, using distilled water. All hydraulic permeability tests were performed with distilled water, except those noted below. The difference

% BSA

vermeate a t 3000 s 0.55 f 0.06 0.69 f 0.02 0.54f 0.08 0.53 f 0.02 0.46 f 0.02 0.56f 0.01 0.52f 0.01 0.33f 0.08 0.74f 0.05 0.54f 0.02 0.60f 0.02 0.61 f 0.02

postoperative hydraulic permability (cm s-l Wa-l x lo4) 0.236f 0.019 0.175f 0.074 0.081f 0.016 0.103f 0.031 0.221f 0.063 0.183f 0.063 0.163 f 0.154 0.133f 0.032

in the postoperative hydraulic permeability can provide information about the species adsorbed on the membrane and how they affect solvent transport. Table 1 shows the values postoperative hydraulic permeability for the binary protein solutions, as determined from solute-free flux through 100 000 MWCO membranes. As can be observed, there is a reduction in the membrane hydraulic permeability after protein ultrafiltration of between 76% and 93%. The trend shows that the dominant contribution to potential change is associated with salt concentration. An increase in the salt concentration from 0.01 to 0.15 M decreased the average permeability by 51.7%. Runs with the highest salt concentrations generally resulted in lower hydraulic permeabilities &r the ultrafiltration run. The observation is in agreement with the work of Cheryan and Merin (19801, which indicated that salt tends to increase the binding of proteins to the membrane. The error in the postoperative hydraulic permeability can be associated with the variability of the membrane pore structure. Two additional runs were performed for the cases with feed solutions with a pH of 8.5 and an IgG concentration of 0.07%. The postoperative hydraulic permeability was performed in the same manner for the two different salt conditions, except that the deionized water at neutral pH was replaced with a protein-free phosphate buffer solution and adjusted to the corresponding pH and salt concentration. The resulting hydraulic permeabilities were 0.294 f 0.104 and 0.121 f 0.033 for the low- and high-salt cases, respectively. These observations were within the error limits of the runs performed with deionized water for the cases. This implies that the binding of the protein was not significantly altered by the pH adjustment of the permeate solution. Permeate Flux. Transient permeate flux was determined for both the single-solute and binary-solute studies. Table 1 summarizes the average permeate flux for all cases after 3000 s of operation. In nearly all of the cases studied, the presence of IgG had little effect on decreasing the permeate flux, and often the presence of IgG resulted in an increase in the permeate flux at the solution properties studied. This is most apparent when the solution properties are at a pH of 8.5 and an ionic strength of 0.01 M (Figure 1). It appears that the solution properties have the most dramatic effect when BSA is alone (single-solute permeate flux case). In the presence of IgG, the permeate flux is altered by solution properties, but generally, the fluxes were all within the same range over time. Since these runs are performed with transmembrane pressure pulsing in a batch cell, the dominant flux reduction is due to concentration polarization and adsorption, with the effect of simple pore plugging minimized. It may be argued that since the pH

Biotechnol. Prog., 1994,Vol. 10, No. 5

541

I

5 ,

O.@ 0.7

,-. u

2s 3 1

T

-. 0

e

1000

2000

3000

4000

5000

0 . 2 4 . L 0 1000

Ulaafilnation Time (sec)

Figure 1. Transient permeate flux comparing single-solute and

binary-solute studies for pH 8.5 and ionic strength 0.01 M NaC1. 0 represents the single-solute case, 0 represents the case with 0.07%IgG, and 0 represents the case with 0.15%IgG. The error bars indicate experimental errors based on the standard deviation of repeated runs.

values used in this study are at the isoelectric points of IgG, the multiple factors associated with the permeate resistance offset one another at these pH values. The increase in permeate flux due to the presence of IgG in solution at its isoelectric points may be associated with the increased concentration of the lower osmotic pressure species (IgG) in the polarization boundary layer. This may also explain why this apparent anomaly did not appear at the physiological pH of 7.4 (Miller et al., 1993). Further quantification of this theory would require the calculation of the overall osmotic pressure of the solution at the membrane wall as a function of both species concentrations. At the high concentrations associated with the boundary layer at the wall, this is not a summation of the independent species' osmotic pressures, but rather a nonlinear function of the concentrations. This research is currently being performed and is the subject of a future paper. I t should also be noted that since the flux is relatively low in the batch cell ultrafiltration study when compared with the cross-flow studies (Blatt et al., 1970; Porter, 19791, the additional resistance associated with solute adsorption, which is usually small for cross-flow, is also a significant factor in overall permeate flux reduction. Permeate BSA Concentration. At a pH of 5.8, the BSA permeate concentration for the single-solute cases showed a higher permeate concentration at an ionic strength of 0.01 M NaCl than a t 0.15 M NaC1. This observation was reversed slightly for cases a t pH 8.5. A higher permeate flux was also observed for this case. This implies that BSA adsorption at the membrane surface is significantly reduced for these solution properties. The postoperative hydraulic permeabilities indicate that this is also true for the binary protein cases. Surprisingly, IgG did not always reduce the BSA concentration in the permeate, and in some cases BSA concentration increased as the presence of IgG in the feed was increased. In the cases where the pH was 5.8 and the ionic strength was 0.15 M NaCl (Figure 2), and in the cases where the pH was 8.5 and the ionic strength was 0.01 M NaCl (Figure 31, the introduction of 0.07% IgG caused a decrease in BSA permeate concentration. However, for these same solution properties, when the IgG concentration was increased to 0.15%, the BSA permeate concentration increased. For the case with pH 8.5 and ionic strength 0.01 M, the narrow error in the

-

I

.

I

2000

,

3000

I

,

,

5000

4000

Ultrafduation Time (sec)

Figure 2. Transient BSA permeate concentration comparing

single-solute and binary-solute studies for pH 5.8 and ionic strength 0.15 M NaC1. 0 represents the single-solute case, 0 represents the case with 0.07% IgG, and represents the case with 0.15% IgG. The error bars indicate experimental error based on the standard deviation of repeated runs.

".. , 0

.

I

,

1000

I

2000

I

.

.

3000

I

4000

.

-

5

Uluafilwtion Time (sec)

Figure 3. Transient BSA permeate concentration comparing single-solute and binary-solute studies for pH 8.5 and ionic strength 0.01 M NaC1. 0 represents the single-solute case, A represents the case with 0.07%IgG, and 0 represents the case with 0.15%IgG. The error bars indicate experimental errors based on the standard deviation of repeated runs.

duplicate runs, coupled with the relatively large differences in the flux values, indicates the insensitivity of this result to potential membrane pore structure variations within the same membrane lot. Since the permeate flux for this case was also higher when the IgG concentration was 0.15% (Figure 11, these observations clearly indicate that the presence of IgG does not cause a reduction in BSA flux in general. This also shows that simple steric hindrance is not, in general, the cause of BSA reduction in the presence of IgG.

Conclusion This study investigated the effect of nonphysiological solution properties on the transport of solvent and BSA in BSA-IgG ultrafiltration. The results showed that, unlike cases performed at pH 7.4 and 0.15 M NaC1, the presence of IgG may increase the transport of BSA under certain conditions. The complexity of the solution properties in this study, primarily because of the multicomponent nature of the BSA and IgG proteins used, and the lack of interaction information for these solutes or their combined osmotic pressure make further quantitative analysis impossible at this time. However, this study

Biotechnol. Prog., 1994, Vol. 10, No. 5

does show the potential value and insights that will result from further binary protein ultrafiltration research.

Acknowledgment The authors gratefully acknowledge the support of the NIH (Biomedical Seed Grant), GE Foundation, and the Byproducts for Biotechnology Consortium, USDA (Grant No. A16-20-03). The authors also thank Amicon Corporation for the generous donation of membranes.

Literature Cited Blatt, W. F.; Dravid, A.; Michaels, A. S.; Nelson, L. Solute Polarization and Cake Formation in Membrane Ultrafiltration: Causes, Consequences, and Control Techniques. Membrane Science a n d Technology; Flinn, J. E., Ed.; Plenum Press: New York, 1970; p 47. Cheryan, M.; Merin, U. A Study of the Fouling Phenomena During Ultrafiltration of Cottage Cheese Whey. In Ultrafiltration Membranes a n d Applications; Cooper, A. R., Ed.; Plenum Press: New York, 1980; p 610. Coulson, S. E. Enhancement of Complex Patterns in Isoelectric Focusing with Improved Isogel Performance. Electrophoresis 1981,81,229.

Ingham, K. C.; Busby, T. F.; Sahlestrom, Y.; Castino, F. Separation of Macromolecules by Ultrafiltration: Influence of Protein Adsorption, Protein-Protein Interaction, and Concentration Polarization. Polymer Science a n d Technology: Vol. 13,Ultrafiltration Membranes and Applications; Cooper, A. R., Ed.; Plenum Press: New York, 1980; p 141. Miller, K.; Weitzel, S.; Rodgers, V. G. J. Reduction of Membrane Fouling in the Presence of High Polarization Resistance. J . Membr. Sci. 1993,76, 77. Peters, T. Serum Albumin. Adu. Protein Chem. 1985,37,161. Porter, M. C. Membrane Filtration. In Handbook of Separation Techniques for Chemical Engineers; Schweitzer, P. A., Ed.; McGraw-Hill: New York, 1979; Section 2.1. Robertson, E. F.; Dannelly, H. K.; Malloy, P. J.; Reeves, H. C. Rapid Isoelectric Focusing in a Vertical Polyacrylamide Minigel System. Anal. Biochem. 1987,167, 290. Rodgers, V. G.J.; Sparks, R. E. Reduction of Membrane Fouling in Binary Protein Mixture Ultrafiltration. AIChE J . 1991, 37 (lo), 1517. Sluyterman, L. A.; Wijdenes, J. Chromatofocusing Isoelectric Focusing on Ion-Exchange Columns: 11. Experimental Verification. J . Chromatogr. 1978,150,31. Accepted March 7, 1994.@ @

Abstract published in Advance ACS Abstracts, May 15, 1994.