Comparison of the Adsorption Characteristics of Expanded Bed Adsorbent with Conventional Chromatographic Adsorbent

Transcription

1 Korean J. Chem. Eng., 18(3), (2001) Comarison of the Adsortion Characteristics of Exanded Bed Adsorbent with Conventional Chromatograhic Adsorbent Hong-Bo Hu, Shan-Jing Yao, Zi-Qiang Zhu and Byung-Ki Hur * Deartment of Chemical and Biochemical Engineering, Zhejiang University, Hangzhou, , P. R. China *Deartment of Biological Engineering, Inha University, Inchon , Korea (Received 28 December 2000 acceted 20 March 2001) Abstract The static and kinetic adsortion characteristics of Streamline DEAE and DEAE-Seharose FF were studied under various oerating conditions. The adsortion isotherms for the two tyes of adsorbents were obtained and found to fit well to a Langmuir-tye exression. The adsortion kinetics of Streamline DEAE at different concentrations, temeratures, and viscosities were studied and a mathematical model including article size distribution was develoed to describe the adsortion erformance of Streamline DEAE. Comaring the utake curves of Streamline DEAE with DEAE-Seharose FF, it could be concluded that Streamline DEAE achieves equilibrium faster to get equilibrium than DEAE-Seharose FF, indicating that Streamline DEAE could be used in higher flow rate systems. Key words: Adsortion, Exanded Bed, Kinetics, Model INTRODUCTION To whom corresondence should be addressed. biosys@inha.ac.kr Exanded bed adsortion is one tye of integrated technology now being considered for the urification of desired roducts directly from cell susensions, crude cell lysates, and refolded mixtures. Exanded bed adsortion combines classification rocessing, such as microfiltration and centrifugation, and artial urification rocessing into a one-unit oeration, roviding roduct isolation while minimizing the number of rocessing oerations required for roduct recovery [Chase, 1994]. Therefore, the use of exanded bed adsortion in the searation rocesses of biotechnology has received considerable attention in the ast ten years and a wide range of alications have been roosed [Chase and Dreager, 1984; Chase, 1994; Clemmitt and Chase, 2000; Hjorth, 1997; Hu et al., 1999; Mullick and Flickinger, 1999; Owen and Chase; 1999]. However, most of the studies have been focused on the alication of this technique, with little attemt to investigate the fundamental adsortion roerties. Streamline adsorbents are a set of adsorbents secially designed for the exanded bed adsortion rocedure by Amersham Pharmacia Biotech [Hjorth, 1997]. Comared with the conventional chromatograhic matrices, they have high-density inert cores that allow them to achieve the high throughut required in industrial alications of adsortion chromatograhy. Moreover, they have a comaratively wide article size distribution that reduces the back-mixing in the column. Most of the alication studies on exanded bed adsortion used this tye of adsorbent for the searation rocess [Chase and Dreager, 1984; Chase, 1994; Clemmitt and Chase, 2000; Hjorth, 1997; Hu et al., 1999]. However, the fundamental adsortion research for the adsorbent was scarce. Although there have been a large number of mathematical models concerning rotein adsortion [Chase, 1984; Graham and Fook, 1982; Horstmann and Chase, 1989; Johnston and Hearn, 1990; Johnston et al., 1991; Skidmore et al., 1990; Tsou and Graham, 1985] and some reorts about the effect of the adsorbent size on the bioroduct searation [Choi, 1990], the effect of surface ressure on the adsortive behavior of a globular rotein [Cho et al., 1999] and the effect of overflow arameter on the frequency resonse of a continuous flow adsorber [Park et al., 2000], there have been no reorts until now concerning the adsortion model that includes both size distribution and heterogeneous structure. Most researchers have assumed Streamline adsorbents as the conventional adsorbent in order to use a traditional adsortion model [Chang and Chase, 1996; Karau et al., 1997; Wright et al., 1999]. Unlike traditional adsorbents, the article size distribution was designed urosely for the Streamline adsorbent. Therefore, a realistic adsortion model concerning a Streamline adsorbent should take into account the article size distribution. In this aer, Streamline DEAE and DEAE Seharose FF were chosen as the model adsorbents and Bovine serum albumin (BSA) as the model rotein. The static and kinetic adsortion erformance of the two adsorbents was studied under various conditions. Moreover, a heterogeneous adsortion model including article size distribution was develoed in order to describe the kinetic adsortion erformance of Streamline DEAE. MATERIAL AND METHODS Streamline DEAE and DEAE-Seharose FF were urchased from Amersham Pharmacia Biotech. The article size distribution data of Streamline DEAE was determined by Coulter LS-230 laser size analyser. Bovine serum albumin (BSA) (fraction V) was urchased from Sino-American Biotech. All other chemicals used were of analytical grade and were obtained from commercial sources. 1. Adsortion Equilibrium Isotherms Flasks of 25 ml total volume were filled with 0.5 ml resin and 14.5 ml BSA solution in a 20 mm sodium hoshate buffer (H 7.0). After the 20 h incubation, the rotein content in the suernatant was determined by UV adsortion (280 nm). The amount of 357

2 358 H.-B. Hu et al. rotein adsorbed was calculated from the difference between the initial and final concentrations. 2. Adsortion Kinetics Adsortion kinetics was measured via rotein absorbency at 280 nm by using the aaratus described by Horstmann and Chase [Horstmann and Chase, 1989]. The adsorbent was equilibrated in the required buffer, and made to a 1:1 (v/v) susension. Twenty ml of buffer was circulated in the system by means of a eristaltic um. The initial otical density of the buffer was recorded as the blank samle, and then a small volume of a concentrated solution of BSA (8 mg/ml) in buffer was added. The total volume was ket at 50 ml adjusted by buffer. After the otical density reached equilibrium, the value was recorded and 2 ml of the adsorbent susension was added. The otical density of the solution decreased with time as BSA was adsorbed to the adsorbent until the density became constant. The equilibrium concentration was determined by alying the constant otical density to the correlation curve between otical densities and BSA concentrations. A series of such exeriments was carried out at different initial concentrations of BSA, temeratures and solution viscosities. Adding a quantity of glycerol modified the solution viscosity. 3. A Model for Adsortion Kinetics In order to roose the adsortion kinetics model of Streamline DEAE, the assumtions roosed by Hostmann and Chase were used [Horstmann and Chase, 1989]. Comared to the conventional chromatograhic adsorbent, an additional assumtion was made due to the secific roerties of Streamline DEAE. The adsorbent article was regarded as sherical with an inert core. Moreover, the thickness of the agarose shell was constant, and the size distribution of the adsorbent was a function of that of the inert core. For diffusion within the adsorbent article, the oint concentration of a solution was given by Eq. (1). The second-order reaction model develoed by Chase [Chase, 1984] was used to describe the equilibrium adsortion behavior exressed by Eq. (2). As described above, Eq. (2) became Eq. (3) at equilibrium, which is the form of the Langmuir equation. Where q m is maximum caacity and K d dissociation constant. According to this assumtion [Horstmann and Chase, 1989], the binding rate was instantaneous comared to the rotein transfer rate as given by Eqs. (4) and (5). and C ε i = ε D 2 C i C i ( r 2 r r 1 ε ) q i q i = k 1 C i ( q m q i ) k 1 q i q q i = m C i K d + C i q i q = C i i C i q i q = m K d C i ( K d + C i ) 2 (1) (2) (3) (4) (5) Substituting Eq. (5) into Eq. (4) gave q i q = m K C d i ( K d + C i ) 2 Substituting Eq. (6) into Eq. (1) gave 1 ε 1 + q m K d C i = ε ( K d + C i ) 2 D 2 C i C i r 2 r r The rate of mass transfer through the external film related the bulk liquid concentration in the ore liquid is exressed by Eq. (8) at the surface of the article. C i k = f ( C r r = D Rj ε L C i ) r = Rj The rate becomes zero at the surface of the inert core of the article. C i = 0 r r = r0, j Based on the data of article size distribution, the adsorbents were divided into several fractions, and the adsorbent diameter within each secific fraction was assumed to be uniform. In a stirred tank system, the rate of change of bulk concentration was given by Eq. (10). dc L = dt (6) (7) (8) (9) (10) Eqs. (7) to (10) were the general forms of the new model. When r 0,j =0, and j=1, the model was changed to the traditional liquid film and ore diffusion model, so the new model was named the modified liquid film and ore diffusion model (MFPDM). The model was solved with Matlab 5.0 by the orthogonal collocation method [Finlayson, 1980]. The values of D and k f were obtained by the regression of exerimental data. If the rotein-binding rate is assumed to be the rate-limiting ste, a model named as kinetic constant model can be obtained by integration of Eq. (2) [Johnston and Hearn, 1990; Skidmore et al., 1990], and is given by Eq. (11). where 3ν j k f ( C R j V L C i ) r = R j j ( b + a) 1 ex 2aν k Ct () = C ν V 1 t V b + a b a 1 ex 2aν k V 1 t C a 2 = b 2 0 V ν qm 1 b = -- C V + q 2 ν m + K V d ν (11) (12) (13) The model was used to simulate the exerimental results for comarison. RESULTS AND DISCUSSION 1. Particle Size Distribution of Streamline DEAE May, 2001

3 Adsortion Characteristics of Exanded Bed Adsorbent 359 Fig. 1. The article size distribution data of Streamline DEAE measured by Coulter LS-230 laser size analyser. Table 1. The static adsortion data at varied temeratures (H 7.0) Temerature ( o C) STREAMLINE DEAE DEAE-Seharose FF q m (mg/ml) K d (mg/ml) q m (mg/ml) K d (mg/ml) Fig. 2. The adsortion rofile for BSA by the modified film and ore diffusion model at varied initial concentrations. Fig. 3. The adsortion rofile for BSA by the modified film and ore diffusion model at varied viscosities. Fig. 1 shows the article size distribution of Streamline DEAE measured by Coulter Laser L-230. The grah shows that the article size distribution was similar to the logarithmic Gaussian distribution and the largest volume of solid was given by articles of diameter of aroximately 200 µm. The corresonding mean diameter of the article was µm, which was in good agreement with the result rovided by Amersham Pharmacia Biotech. 2. Equilibrium Adsortion Isotherm Equilibrium adsortion isotherm exeriments of Streamline DEAE and DEAE-Seharose FF were conducted at four different temeratures ranging from 15 o C to 35 o C. The equilibrium adsortion erformances of the two tyes of adsorbents could be described by the Langmuir tye equation and the adsortion arameters are summarized in Table 1. It can be concluded from Table 1 that the disassociate constant K d and maximum adsortion caacity q m of Streamline DEAE are similar to those of DEAE-Seharose FF. The reason might be that both of the functional grous of the adsorbents are diethylaminoethyl, which results in a similar binding force and selectivity to BSA. Moreover, though the article diameter of Streamline DEAE is larger than that of DEAE-Seharose FF ( µm), the inert core in Streamline DEAE decreases the content of agarose, leading to a similar maximum adsortion caacity. Furthermore, the value of K d decreased with the increase in temerature while the q m increased, indicating that the adsortion erformance on BSA of the two adsorbents is imroved with the increase of temerature. 3. Adsortion Kinetics 3-1. Effect of Various Oerating Conditions on the Kinetic Adsortion Performance Fig. 2 shows how the utake curves varied with the change of initial concentration. The effect of rotein concentration on the adsortion kinetics is difficult to observe directly and is illustrated by the kinetic model, which is discussed in the following section. Fig. 3 shows the effect of solution viscosity on the kinetic adsortion erformance. As shown in the grahs, the utake curves become shallower and require more time to reach equilibrium as the solution viscosity increases. However, as reorted by Chase and Dreager [Chase and Dreager, 1992], the maximum adsortion caacity was not affected by the change of viscosity because the final concentration in the solution was the same under the different conditions. Fig. 4 shows the effect of temerature on the kinetic adsortion erformance. Based on the exerimental data of equilibrium adsortion isotherms, it can be anticiated that the utake curves will become steeer with the increase in temerature. Moreover, the viscosity of the solution decreases and the movement of the rotein molecules increases with the increase in temerature, which imroves the transfer erformance of the rotein. The exerimental data suorts this assumtion The Simulation Results of the Two Models Two tyes of models, a kinetic constant model and a modified Korean J. Chem. Eng.(Vol. 18, No. 3)

4 360 H.-B. Hu et al. Fig. 4. The adsortion rofile for BSA by the modified film and ore diffusion model at varied temeratures. film and ore diffusion model (MFPDM), were used to simulate the utake curves. For MFPDM, the adsorbent was divided into four sections. The average diameter for the four sections was µm, µm, µm and µm, resectively, and the volume ercentage was 24.7%, 24.4%, 24.2%, and 26.6%, resectively. The value of liquid film mass transfer coefficient, k f, estimated by the correlation used by Horstman and Chase [Horstmann and Chase, 1989] was not suitable for streamline DEAE due to its articular structure. Therefore, in this aer, the value of k f was obtained by the regression of exerimental data. The thickness of the agarose shell was assumed to be 60 µm in this study. The simulation results are listed in Table 2. The kinetic constant arameter k 1, the effective ore diffusivity D, and the liquid film transfer coefficient, k f, changed under various conditions. The reason the values of all three arameters were deendent on the oerating conditions was that all the arameters were lumed constants because of a simlification of the models. The deendency of the three arameters on the oerating conditions was reorted by Johnston and Hearn [Johnston and Hearn, 1990] and Horstmann and Chase [Horstmann and Chase, 1989]. The comarisons of relative errors of simulation obtained by the modified film and ore diffusion model with those of the kinetic constant model are shown in Figs. 5, 6 and 7. As shown in these Fig. 5. Comarison of calculated concentration of rotein with exerimental data at varied initial concentrations. Fig. 6. Comarison of calculated concentration of rotein with exerimental data at varied viscosities. figures, the relative errors of the modified film and ore diffusion model (±5%) are smaller than those of the kinetic constant model (±10%), indicating that the modified film and ore diffusion model Table 2. The arameters obtained by the regression of the two models Oeration conditions D ( 10 7 cm 2 /s) k f ( 10 6 cm/s) k 1 ( 10 3 ml/mgs) Initial concentration (mg/ml) Comosition of glycerol (w/w) 10% glycerol/buffer % glycerol/buffer % glycerol/buffer Temerature ( o C) May, 2001

5 Adsortion Characteristics of Exanded Bed Adsorbent 361 CONCLUSIONS Fig. 7. Comarison of calculated concentration of rotein with exerimental data at varied temeratures. The aim of these investigations was to show the adsortion erformance of Streamline adsorbent and to develo a mathematical model to describe the secific roerties of this kind of adsorbent. The equilibrium adsortion isotherm and adsortion kinetics of Streamline DEAE and DEAE-Seharose FF were studied under different oerating conditions. The adsortion isotherms for the two tyes of adsorbents were found to fit well to the Langmuir tye exression and the values of two arameters, K d and q m, were very similar between the two tyes of adsorbents as shown in Table 1. The kinetic adsortion characteristics of Streamline DEAE at different concentrations, temeratures, and viscosities were investigated and a mathematical model including article size distribution was develoed in order to describe the adsortion erformance. The simulation results indicated that the model could better describe the kinetic adsortion erformance. Further work is being undertaken on the adsortion erformance in exanded beds and the model develoed in this aer will be used to describe the adsortion erformance in exanded beds. NOMENCLATURE Fig. 8. Adsortion rofiles for BSA by DEAE Seharose FF and STREAMLINE DEAE. can be more accurate in describing the kinetic adsortion erformance of Streamline DEAE Comarison of Kinetic Adsortion Performance of Streamline DEAE and DEAE-Seharose FF Fig. 8 shows the kinetic adsortion curves of Streamline DEAE and DEAE-Seharose FF. As shown in the figure, the utake curves of both of the two adsorbents are the same at the initial adsortion stage, although the difference becomes larger and larger as the adsortion time elases. It can be exlained as follows. The transfer resistance is rimarily focused on the side of the liquid film at the initial stage and the liquid film transfer resistance does not deend on the article size or the structure of adsorbent, so there is no difference in the utake curves at the beginning of the adsortion stage. However, at the final stage of adsortion, the main transfer resistance governing the adsortion erformance is ore diffusion; the diffusion distance in the Streamline DEAE is shorter than that in DEAE-Seharose FF due to the inert core in Streamline DEAE. Therefore, Streamline DEAE is faster in reaching equilibrium than DEAE-Seharose FF. This result indicates that Streamline DEAE can be used in high-flow-rate searation rocesses. C(t) : liquid hase concentration at the time t [mg/ml] C 0 : initial liquid hase concentration [mg/ml] C L : liquid hase concentration [mg/ml] C i : oint concentration of liquid inside article [mg/ml] D : effective article diffusion coefficient [cm 2 /s] k 1 : forward rate constant for surface reaction [ml/mg s] k 1 : reverse rate constant for surface reaction constant [s 1 ] K d : disassociation constant [mg/ml] k f : liquid film mass transfer coefficient [cm/s] q i : oint concentration of solute in the absorbent [mg/ml] q m : maximal rotein concentration in the absorbent [mg/ml] R j : article radius of the fraction j [cm] r : radial co-ordinate [cm] r 0j : article radius of inert core of the fraction j [cm] t : time [s] V : volume of liquid [ml] Greek Letters δ : thickness of agarose shell [cm] ε : article orosity ν j : volume of adsorbent of the fraction j [ml] τ : dimensionless time=d t/d 2 REFERENCES Chang, Y. K. and Chase, H. A., Develoment of Oerating Conditions for Protein Purification Using Exanded Bed Techniques: The Effect of the Degree of Bed Exansion on Adsortion Performance, Biotechnol. Bioeng., 49, 512 (1996). Chase, H. A., Prediction of the Performance of Prearative Affinity Chromatograhy, J. Chromatogr., 297, 179 (1984). Chase, H. A. and Dreager, M. N., Affinity Purification of Proteins Using Exanded Bed, J. Chromatogr., 597, 129 (1992). Chase, H. A., Purification of Proteins by Adsortion Chromatograhy Korean J. Chem. Eng.(Vol. 18, No. 3)

6 362 H.-B. Hu et al. in Exanded Beds, Trends in Biotechnol., 12, 296 (1994). Cho, D. C. and Cornec, M., Adsortive Behavior of a Monoglyceride Monolayer Sread on the Aqueous Surface, Korean J. Chem. Eng., 16, 371 (1999). Choi, J. W., Bioroduct Adsortion in Immobilized Adsorbent: Local Thermodynamic Equilibrium Model, Korean J. Chem. Eng., 7, 269 (1990) Clemmitt, R. H. and Chase, H. A., Facilitated Downstream Processing of a Histidine-Tagged Protein from Unclarified E. coli Homogenates Using Immobilized Metal Affinity Exanded Bed Adsortion, Biotechnol. Bioeng., 67, 206 (2000). Finlayson, B. A., Nonlinear Analysis in Chemical Engineering, Mc- Graw-Hill Inc., New York (1980). Graham, E. E. and Fook, C. H., Rate of Protein Adsortion and Desortion on Cellulosic Ion Exchangers, AIChE J., 28, 245 (1982). Hjorth, R., Exanded Bed Adsortion in Industrial Biorocessing: Recent Develoments, Trends in Biotechnol., 15, 230 (1997). Horstmann, B. J. and Chase, H. A., Modelling the Affinity Adsortion of Immunoglobulin G to Protein A Immobilised to Agarose Matrices, Chem. Eng. Res. Des., 67, 243 (1989). Hu, H. B., Yao, S. J. and Zhu, Z. Q., Exanded Bed Adsortion Chromatograhy and its Alication in Biosearation, Chem. Eng. (in Chinese), 27, 37 (1999). Johnston, A. and Hearn, M. T. W., High-Performance Liquid Chromatograhy of Amino Acids, Petides and Proteins CIII. Mass Transfer Resistances in Ion Exchange and Dye-affinity Chromatograhy of Proteins, J. Chromatogr., 512, 101 (1990). Johnston, A., Mao, Q. M. and Hearn, M. T. W., High-Performance Liquid Chromatograhy of Amino Acids, Petides and Proteins CXII. Analysis of Oerating Parameters Affecting the Breakthrough Curves in Fixed-Bed Chromatograhy of Proteins Using Several Mathematical Models, J. Chromatogr., 548, 127 ( 1991). Karau, A., Benken, C., Thömmes, J. and Kula, M. R., The Influence of Particle Size Distribution and Oerating Conditions on the Adsortion Performance in Fluidized Beds, Biotechnol. Bioeng., 55, 54 (1997). Mullick, A. and Flickinger, M., Exanded Bed Adsortion of Human Serum Albumin from Very Dense Saccharomyces Cerevesiae Susensions on Fluoride-Modified Zirconia, Biotechnol. Bioeng., 65, 282 (1999). Owen, R. O. and Chase, H. A., Modeling of the Continuous-Current Exanded Bed Adsorber for the Purification of Proteins, Chem. Eng. Sci., 54, 3765 (1999). Park, I. S., Kwak, C. and Hwang, Y. G., Frequency Resonse of Continuous Flow Adsorber for Multicomonent System, Korean J. Chem. Eng., 17, 704 (2000). Skidmore, G. L., Horstmann, B. J. and Chase, H. A., Modelling Single-Comonent Protein Adsortion to the Cation Exchanger S Seharose FF, J. Chromatogr., 498, 113 (1990). Tsou, H. S. and Graham, E. E., Prediction of Adsortion and Desortion of Protein on Dextran Based Ion-Exchange Resin, AIChE J., 31, 1959 (1985). Wright, P. R., Muzzio, F. J. and Glasser, B. J., Effect of Resin Characteristics on Fluidized Bed Adsortion of Proteins, Biotechnol. Prog., 15, 932 (1999). May, 2001