Concentration of large biomolecules with hydrogels

transcript

Chemical En@neering Science, Vol. 47, No. 1, pp. 314, 1992. coo9-22509/92 s5.00’+ 0.m

Printed in Great Britain. g199tPqsmooaRasPk

CONCENTRATION OF LARGE BIOMOLECULES WITH HYDROGELS

EBRAHIM VASHEGHANI-FARAHANI,t DAVID G. COOPER, JUAN H. VERA and MARTIN E. WEBER’

Department of Chemical Engineering, McGill University, Montreal, Quebec, Canada

(Received fir publication 16 May 1991)

Abstract-The swelling and exclusion behavior of crosslinked hydrogek were detemincd in aqueous solutions of electrolytes and in aqueous solutions of proteins and enzymes. The gels were ionic copolymers of a&amide, ionic copolymers of N-isopropylacrylamide (NIPA) and homopolymers of NIPA. 30th anionic (weak acid and strong acid) and cationic (strong base) monomers were used. Crosslinked polymer gels, whose swelling is sensitive to the temperature or concentration of the surrounding medium, can be exploited to concentrate dilute aquwus solutions of macromolecular solutes including proteins and enzymes, with no adverse effect on the activity of the enzyme. Exclusion was primarily by size and net charge although some proteins adsorbed onto the surfaces of hydrogels having the same charge. The variable with the largest effect on the size exclusion behavior of hydrogels was the monomer concentration at gel formation.

INTRODUCTION

Separation processes play an important role in the chemical industry. The common separation processes, however, are rarely suitable for the separation of delicate biological products, which are typically pro- duced in dilute broths containing several contam- inants. One promising technique involves the use of hydrophilic polymer gels as extraction solvents. The process consists of an equilibrium step followed by a regeneration step. Gel particles of low water content are added to the dilute aqueous solution of biological molecules. The gel swells by sorbing water and molecules of low molecular weight while large molecules (and ionic solutes if the gel is a polyelectrolyte) are excluded. The swollen gel is regenerated by collapsing it through exposure to different pH, temperature and/or salt concentration. The collapsed gel is separated from the released solution, the extract, and re- used. The first step of this process, i.e. concentrating dilute aqueous solutions of macromolecules, was ori- ginally reported by Flodin et al. (1960), but its applica- tion was limited due to the lack of a simple, low cost method for regeneration of the gel. The second step of this process was investigated by Cussler and co- workers (Cussler et al., 1984; Gehrke et al., 1986; Freitas and Cussler. 1987; Trank et al., 1989) after the discovery of phase transitions in polymeric gels by Du&k and Patterson (1968), and Tanaka (1978).

The candidate gels for this extraction process should exhibit the following behavior:

(a) swelling by an order of magnitude in dilute aqueous solutions containing salts found in fer- mentation broths.

‘Present address: Department of Chemical Engineering, University of Tarbiat, Modarres, P.O. Box 14155-4838. Tehran, I&.

*Author to whom correspondence should be addressed.

(b) nearly complete exclusion of large molecules. (c) little effect on the activity of biological mole-

cules. (d) nearly complete collapse upon exposure to mild

conditions during regeneration. (e) stability over repeated cycles of swelling and

collapse.

To meet these criteria, synthetic gels must be hydrophilic and they must be crosslinked sufficiently to maintain integrity but still alIow considerable volume change. Crosslinked gels of acrylamide and its deriv- atives exhibit most of the desirable properties.

GEL PREPARATION AND EXPERIMRNT AL PROCEDURE

Copolymers of acrylamide and sodium acrylate (anionic gel) or 3-(methacrylamido) propyltrimethylammonium chloride (cationic gel) were prepared. Gels were also prepared as homopolymers of N-isopropylacrylamide (NJPA), copolymers of NIPA and sodium acrylate or 2-acrylamido-2-methyl- 1 -propanesulfonic acid (R-SO;H+) or sodium vinylsulfonatc. The hydrogels were prepared by free radical solution polymerization in distilled water (PH H 5.7), under a nitrogen atmosphere. at 23°C. Monomers and crosslinking agent (N,N’-methylenebisacrylamide) were added to sufficient nitrogen-sparged distilled water to make 50 ml of solution after all materials were added. This solution was flushed with nitrogen until the materials dissolved completely. The initiator (ammonium persulfate) and accelerator [sodium metabisulfite or N,N,N’,N’-tetraethyhnethylenediamine (cationic gel)] were added and the solution was transferred to a large test tube which contained a number of smaller glass tubes of 2.4 mm i-d. The solution was flushed with nitrogen for S to 10 minutes and then scaled. After 24 hours the gels were forced from the glass tubes and cut into cylinders having lengths roughly equal to their diameter. The original mass, M,, of each

32 EBRAHIM VASHEDHANI-FARAHAN et al.

piece was determined and then it was dialyzed for 48 hours against a large excess of distilled water to remove minute quantities of impurities and unreacted monomers or oligomers trapped in the network.

The gel composition is specified in terms of grams of total monomers per 100 cm3 of solution at preparation (% T), weight percentage of crosslinker (%C) and mole percentage of ionizable monomer (% I). These quantities are defined by:

% T = mass of all monomers (g) x loo volume of solution (cm3) (1)

%C = mass of crosslinking agent

mass of all monomers x 100 (2)

Q/o I = moles of ionizable monomer

moles of all monomers x 100. (3)

In swelling experiments, the degree of swelling was measured by immersing a piece of dialyzed gel (original mass, M,, of approximately 0.01 g) or collapsed gel in 2 1 of a solution of known concentration, pH and temperature. After equilibration (1248 h), the pH was measured, and the swollen gel was removed from the solution and weighed. The swelling ratio was defined as the ratio of swollen mass, M, to the original mass, M,, or to the mass after regeneration, M,.

To investigate gel selectivity, ionized copolymer gels of acrylamide and ionic and nonionic N-isopropylacrylamide (NIPA) gels were used. The effects of monomer concentration, crosslinking ratio, network and solute charge, pH and ionic strength of the solution on the exclusion efficiency of hydrogels were studied. These experiments were similar to the equilibrium swelling experiments. Dried polymer particles with known weights (Md = 0.2-2 g) were brought into contact with a solution of known volume, pH and concentration. The initial volume of the solution was approximately twice the volume of the swollen gel at equilibrium. The swollen gel was separated from the retentate (or raffinate) and weighed. The volume and concentration of retentate were also determined.

The solutes used for this study included: polyethylene glycols of different molecular weights (Aldrich and Polyscience), &galactosidase (Aldrich), dextran, dextran sulfate, ovalbumin, bovine serum albumin, /I-lactoglobulin, cytochrome c, a-amylase and lipase, which were purchased from Sigma Chemical Company.

Polyethylene glycol (PEG), with a stoichiometric formula of H(OCH&H&OH, is available com- mercially in a range of molecular weights. The concentration of polyethylene glycol solutions was determined by refractometry.

An industrial grade dextran (MW = 19,500) and a laboratory grade (MW = 9400) as well as dextran sulfate (sodium salt) with an average molecular weight of approximately 8000 were used. The concentrations of dextran and dextran sulfate were measured using a Polax-D polarimeter manufactured by Atago.

Ovalbumin has a molecular weight of approximately 45,000 and an isoelectric pH of 4.7. The ovalbumin used here was Grade V. Bovine serum albumin has a molecular weight of approximately 66,000 with an isoelectric point of 4.8. Sigma Fraction V powder with 9699% protein was used. b-lactoglobulin (from bovine milk) was a 3 x crystallized and lyophilii protein with a molecular weight of approximately 37,000. This protein contains subunits A and B, each with a molecular weight of 18,400. Cytochrome c (Sigma Type V-A) from bovine heart had a molecular weight of 12,330. Its concentration was determined by speetrophotometry at 550 nm.

The a-amylase was a 4 x crystallid and lyophilized powder from Bacillus species (Sigma Type II-A), with an approximate molecular weight of SO,OO& 55,000. The enzyme activity was measured using the Somogyi (1960) method with a Sigma Diagnostic Kit (Procedure No. 700). The lipases used in this study were from wheat germ (Sigma Type I) and Cundida cyhdracea (Sigma Type VII). Lipase from wheat germ was a lyophilized powder containing 95% protein. It contained acid phosphates and its molecular weight is not available. Candida cylindracea is available as an impure preparation. Tomizuka et al. (1966) determined the amino acid composition of the puri- fied enzyme and estimated a molecular weight of 100,00&120,000. The enzymatic activity of the lipase from Candida cylindracea was measured using a Sigma procedure (Sigma Diagnostic Kit, No. 800) which is based on the method of Tietz and Fiereck (1966). #l-Galactosidase consists of four subunits forming an active tetramer. Goldberg (1969) reported a molecular weight of 595,000 for this enzyme. Erickson (1970) demonstrated discrepancies in former studies concerning the molecular weight and sugges- ted a subunit weight of 91,000. Kalnins et al. (1983), using DNA sequencing techniques, reported that /I- galactosidase consists of 1023 amino acid residues, resulting in a protein with a molecular weight of 116,353 per subunit. B-Galactosidase activity was as- sayed as described by Miller (1972).

Spectrophotometry at 280nm was used for the determination of the concentration of the proteins (enzymes). This method is accurate for concentrations of proteins up to 1 g/l. Additional experimental details are given elsewhere (Vasheghani-Farahani, 1990).

EXPERIMENTAL RESUL’IS

Swelling behavior The success of the proposed separation process is

heavily dependent on the regeneration step because there must be a large volume change from the collapsed (regenerated) state to the swollen state. After regeneration the gel may be larger or smaller than it was at preparation.

The swelling behavior as a function of temperature of NIPA gels at pH 7 in distilled water and in 1 M sodium chloride solutions is shown in Fig 1. The mass of polymer in the gel at preparation was about

Concentration of large biomolecules with hydr&els

20 30 a 00 eo

Temperature( “C)

Fig. 1. Effect of temperature on the swelling behavior of Fig. 2. Time variation of swelling of gels in lo-) M NIPA gels at pH 7: (-) distilled water, (- - - -) 1 M Ca(NO,), solution at pH 7, after regeneration in dlffer- NaCl,(A)NIPA,[(O)5%Z,(O)lO%Z,(0)15%ZjNIPA ent media: [(I) 5%Z, (0) lO%Zl acrylamlde +

+ Na-acrylate, (I) NIPA + R-SO,H (lO%Z). Na-acrylate, (e) NIPA. (0) NIPA + Na-acrylate (5% I).

8% of M,; therefore, the mass of water absorbed per unit mass of dry polymer was approximately 12 times larger than the swelling ratio, M/M,. Ionic gels which contained more than 5% ionizable monomer did not collapse even at 50°C. The nonionic gel collapsed by a factor of ten from 23 to 35°C in agreement with Freitas and Cussler’s results (1987).

Amiya et al. (1987) reported that at 42°C ionized NIPA gels containing more than 4.7% sodium acrylate were still swollen, but gels with fewer ionizable groups were collapsed. Hirotsu et al. (1987) found that the transition temperature of ionic NIPA gels increased as the ionic concentration was increased, but all volumes were essentially the same in the shrunken state. Their ionic gel having 10% ionizable monomer underwent a discontinuous volume change around 42°C but the gel having the highest ionic concentration (18.8% I) did not shrink up to 80°C. On the other hand, a discontinuous volume phase transition was observed around 60°C for spherical NIPA gels containing 18.8% sodium acrylate, prepared by inverse suspension polymerization (Matsuo and Tanaka, 1988). These discrepancies about the phase transition behavior of ionic NIPA gels are probably associated with the pH of the external solution which was not reported in most studies.

Addition of salt to the external solution had a profound effect on the swelling of ionic NIPA gels. Copolymers of NIPA and sodium acrylate (5 and lO%Z), which would not collapse in distilled water below 50°C collapsed in 1 M NaCl solution at 23°C. Similar behavior was observed for the cationic copolymer gels of NIPA and 3-(methacrylamido) propyltrimethylammonium chloride (Beltran et al., 1990). These results agree with those of Ohmine and Tanaka (1982), RiEka and Tanaka (1984) and Vasheghani- Farahani et al. (1990) for the effect of salt on the swelling behavior of polyekctrolyte gels. Figure 2

Regeneration conditions indicated in parentheses.

shows time-dependent swelling of ionic copolymer gels of acrylamide and ionic and nonionic NIPA gels in 10m3 M Ca(NO,), solutions at pH 7. The ordinate is the ratio of the swollen maas at time t, M,, to the mass of dry polymer, M,. These gels were regenerated at different conditions. Ionic copolymer gels of acrylamide and sodium acrylate were regenerated at pH 4 (5%1) and pH 2 (10%1) where the diameters of the collapsed gels were about 1.3 and 1.1 times larger than the diameter of gel at preparation, d, = 2.4 mm, respectively. Nonionic NIPA gel was regenerated at 35°C in distilled water of pH 5.7 while ionic NIPA gel (5%2) was regenerated at pH 2. The collapsed diameters of both ionic and nonionic NIPA gels were about 0.54d0. Copolymer gels of acrylamide exhibited higher equilibrium swelling ratios than NIPA gels and the equilibrium swelling ratio increased as the amount of ionizable monomer increased in accordance with the thermodynamic treatment of ionic gel swelling (Vasheghani-Farahani et aZ_, 1990). These expcri- mental results are replotted in Fig. 3 as M,/M,, where M, is the regenerated (collapsed) mass of the gel. NIPA gels exhibited larger volume changes from the collapsed state upon swelling than acrylamide gels, although the latter exhibited larger equilibrium swelling ratios, M/M,,. Although the rate of water uptake for copolymer gels of acrylamide was faster than that of NIPA gels (Vasheghani-Farahani, 1990), they were not easily regenerated. For the water removal process the nonionic NIPA gel, which re- quires mild conditions for collapse (35°C in distilled water), is more promising.

Exclusion behavior The exclusion behavior of copolymer gels of acryl-

amide, and homopolymer and copolymer gels of

34 EBRAHIM VAsnEGIiANI-FARAIiAm et al.

. L 12

Time, hrs

Fig. 3. Time variation of swelling of gels in lo-” M Ca(NO,), solution at pH 7, after regeneration in different media: [(I ) 5% I, ( 0) lO%g acrylamide + Na-acrylate, ( 0 ) NIPA, ( 0 ) NIPA + Na--aerylate (5% I). Regeneration

conditions indicated in parentheses.

N-isopropylacrylamide (NIPA) was studied for different crosslinking ratios, monomer concentrations, per- centages of ionizable monomer as well as the solute size and charge. The selectivity results are reported in terms of the distribution coefficient, K,, defined as the ratio of the solute concentration in the gel phase, c, to that in the retentate or raffinate, C,, i.e.

Kd = @2,. (4)

A mass balance on the solute gives

vrc, = vxcx + lfc (5)

where V, and Cp are the feed volume and concentration, respectively; and V, and V are the retentate volume and the swollen gel volume, respectively. The distribution coefficient was calculated from experimental concentration measurements by combining eqs (4) and (5) to give

& = (V,C, - v,c,)/(vc,).~ The fraction of solute which is excluded, q, is given by

vRcR 1

1+ (KdV/VR)’ Complete exclusion occurs for K, = 0 and complete removal from solution for K, + a . In addition to & the fraction of solute excluded depends on the volume of gel as well as on the volume of solution.

Effect of solute size. Size selective separation depends upon the relationship between the molecular sixe and shape of the solute and the pore size of the hydrogel. The size of an uncharged solute like polyethylene glycol (PEG) is characterized by its molecular weight. Figure 4 shows the effect of PEG molecular weight on the swelling and exclusion behavior of a weak acid copolymer gel of NIPA and sodium acryl-

Fig. 4. Effect of PEG molecular weight on the swelling and exclusion behavior of ionic NIPA gels: (0) Na-acrylate copolymer gel (10%1), (0) R-SOsH copolymer gel (5%1).

ate (7.95%T, 1.67%C, 10%1) and a strong acid copolymer gel of NIPA and 2-acrylamido-2-methyl-l- propanesulfonic acid (R-SO; H+) (8.43% T, 1.58% C, 5% I). The swelling ratio is defined as the ratio of swollen gel mass, M, to dry mass of polymer, M,. Swelling ratios and the distribution coefficients were slightly higher for the weak acid gel which contained more ionizable groups. The distribution coefficient, K,, decreased as the PEG molecular weight increased, thus indicating that the gels sorbed low molecular weight solutes (MW = 800) but excluded higher molecular weight solutes (MW = 18,500). The distribution coefticient of low molecular weight PEG approaches unity thus indicating equal concentrations of the solute in both gel phase and external solution. Sinoe the feed concentration, Cx, of PEG was 25 g/l in each ease, the higher exclusion of solute (lower K,) for higher molecular weight PEG resulted in an external solution (retentate) of higher concentration at equilibrium which, in turn, lowered the swelling ratio of gels at higher molecular weight. This is inconsistent with the thermodynamic requirements of gel swelling.

Solute size alone does not determine selectivity; the shape of the solute molecule is also important. An extended molecule like polyethylene glycol (MW = 18,500) was almost completely excluded by a

strong acid NIPA gel (K, = 0.01 in Fig. 4) whereas a globular protein (enzyme) like lipase (MW Y 10’) partitioned with a distribution coefficient of about 0.2 as shown in Table 3 and discussed later.

Eflect of monomer concentration and crosslinking ratio. The structure of a synthetic gel is determined by the chemical (structure) properties of its components, the concentrations of the reactants and the conditions

Concentration of large biomolecules with hydrogels 35

Fig. 5. Effect of monomer concentration on the swelling and exclusion behavior of NIPA gels: (0 ) NIPA, (0 ) NIPA + R-SO,H (2.5%1), solute PEG (MW = 3400,

C, = 25 g/l)_

during polymerization. Gels may be considered to be swollen networks through which small molecules move freely while large molecules are excluded. The system of crosslinks acts as a physical barrier for molecules of certain sixes and shapes.

The exclusion behavior of a gel changes if the pore size of the network is changed. Gels with smaller pore sizes can be prepared by increasing the concentration of the monomers or of the crosslinking agent at gd formation. To study the effect of monomer concentration and crosslinking ratio on the selectivity of hydrogels, N-isopropylacrylamide gel and its ionic counterparts with different concentrations of monomers and crosslinking ratios were prepared. Poly- ethylene glycol (MW = 3400) was used as the test solute. The experimental results for the effect of monomer concentration, (% T), on the swelling and exclusion behavior of homopolymer gels of NIPA (1.64%C) and copolymer gels of NIPA and R-SO;H+ (1.61%C, 2.5%1) are summarized in Fig. 5. Figure 6 shows the effect of crosslinking ratio, (%C), on the swelling and exclusion behavior of copolymer gels of NIPA and R-SO;H+ (2.5%1) having a total monomer content of approximately 8.3 % T.

The extent of swelling and the distribution coefficient, K.,, decreased with increasing monomer concentration and with increasing amounts of crosslinker.

The value of K,, decreased with increasing crosslinker or monomer concentration, but it approached an asymptote at approximately S%C or 16%T. These results reflect the structural properties of the network. When a network is formed at low concentration, ring formation (cyclixation) is favored. Ilavsk$ and Hrouz

Fig. 6. EIfect of crosslinker on the swelling and bxclusion behavior of the copolymer gel of NIPA and RSO,H (2.5%

I): solute PEG (MW = 3400, C, = 25 g/l).

(1983) reported that the crosslinking efficiency increased with increasing monomer content at gel formation mainly due to decreased cyclixation at network formation. Vasiliev et al. (1985) found that both chemical crosslinking and physical crosslinking (chain entanglements) increased with an increase in monomer concentration at gel formation.

Increasing the percentage of crosslinker increases the numher of crosslinks in the gel network with a consequent decrease in pore size and swelling. Baselga et al. (1987) studied the effbct of crosslinking agent on the properties of polyacrylamide gels and found that the swelling of hydrogels decrease d with increasing percentage of crosslinker and reached a plateau at 7%C. Baselga et al. (1989) concluded that network defects increased with further increase of the amount of crosslinking agent making the crosslinking efficiency progressively lower. Tanaka et al. (1988) investigated the influence of monomer concentration and crosslinking ratio on the chemical gel&on dynamics of acrylamide in water. They observed spatial heterogeneity of the network chains at high acrylamide and crosslinker concentrations. Heterogeneous networks are not transparent in the swollen state (Tanaka et al., 1988). Turbidity was observed here for copolymer and homopolymer gels of N-isopropylacrylamide at monomer concentrations above 12% T or percentage crosslinker above 5 % C.

Since size selective separation depends on the relationship between solute size and pore size, any variable which affects pore size will affect exclusion behavior. The ionizable monomer .content of the gel alSects the degree of swelling, and hence exclusion. Nonionic solutes like PEG (MW = 3400) will be less

EBRAI-IIM VASHEGHANX-FARAMNI et al.

Fig. 7. Effect of ionizable monomer on the swelling and exclusion behavior of the copolymer gel of NIPA and

RSO,H: solute PEG (MW = 3400, Cp = 25 g/l).

excluded as the swelling of the network increases. Figure 7 shows the swelling and exclusion behavior of copolymer gels of NIPA and R-SO; H+ as a function of the percentage of ionizable monomer. The swelling ratio, M/M,, and the distribution coefficient, K,,, increased with increasing content of ionizable monomer.

Figure 8 is a compact presentation of the data from Figs 5-7 for the effect of monomer concentration, crosslinker and mol% of ionizable monomer on the distribution coefficient of PEG (MW = 3400) in NIPA gels. Figure 8 also includes data for NIPA gels in equilibrium with a solution of PEG (C, = 25 g/l) in 0.038 M NaCl. The abscissa in the figure is the volume

Fig. 8. Distribution coefficient of PEG (MW = 34.00) as a function of polymer volume fraction for NIPA gels: [( l ) nonionic gel, ( 0 ) ionic gel] effect of monomer conocntra- tion, ( a ) efkct of crosslinker, C( Cl ) distilled water, (-I )

0.038 hi NaClJ effect of ionizable monomer.

fraction of polymer in the swollen state, u. This volume fraction was calculated from the swelling ratio, M/M,,, by

where p# and pp are the gel aad polymer density, respectively. The values of p. and pp determined experimentally were 1.04 + 0.04 and 1.16 f 0.04, re- spectivel y.

The PEG distribution coefficient decreased as the polymer volume fraction increased. Highly swollen gels exhibited decreased exclusion. As u + 0 the distribution coefficient approaches unity, i.e. the solute concentration is the same inside and outside the gel. Up to a polymer volume fraction of about 0.04 all data lie on a single curve. For higher volume fractions the data for the nonionic NIPA gels (from Fig. 5) lie below the data for the ionic gels. The effect of monomer concentration at gel formation on the exclusion behavior of hydrogels was more pronounced for nonionic MPA gels. These results agree with those of Abe et al. (1990) who studied the effect of crosslinker and monomer concentration at gel formation on the structural properties of thermosensitive poly N-acryloyl- pyrrolidine. The porosity and the pore volume decreased with an increasing percentage of crosslinker, but the mean pore radius changed only slightly. How- ever, with increasing monomer concentration, large pores disappeared and the pore sixes in the network became smaller and more uniform. The results in Fig. 8 show that increasing the monomer concentration was the most effective way to lower Kd and thus improve exclusion for nonionic gels.

E@ct of charge. The exclusion behavior of a polyelectrolyte gel is highly dependent on its interaction with charged solutes. An ionic solute like sodium pentachlorophenolate was almost completely excluded by a pH-sensitive ionic hydrogel (Gehrke et al., 1986) whereas its exclusion was negligible for the nonionic NIPA gel (Freitas and Cussler, 1987). Be- cause of the fixed negative charges on the polymer backbone, the negatively charged pentachlorophenolate ions were excluded.

The effect of solute charge on the swelling ratio and the distribution coefficient of copolymer gels of N- isopropylacrylamide with diierent amounts of ionizable strong acid monomer (R-SO; H ‘) is shown in Fig. 9. This figure shows the swelling ratio and the distribution coefficient in dextran and dextran sulfate solutions_ Figure 9 shows both the effect of size and the effect of charge. To isolate the effect of charge on exclusion behavior for the two solutes of comparable size, dextran (MW = 9400) and dextran sulfate (MW = 8000), NaCl was added to the feed dextran solu-

tion. The data for dextran in 0.5 M NaCl solution (MW = 19,500, 9400) show that swelling increased with ionizable monomer content, but that the value of K, changed only slightly as the percentage of ionizable monomer increased for this nonionic solute. The

Concentration of large biomolecules with hydrogels 37

polymer gel with 10%X Donnan equilibrium was responsible for the enhanced exclusion of the negatively charged solute by the copolymer gel whose backbone carries the same charge.

Fig. 9. Effect of charge on the swelling and exclusion behavior of NIPA gels: (a) dextran (MW = 9400) in 0.5 M NaCl, (I) dextran (MW = 19,500) in 0.5 M NaCi, (0) dextran sulfate (MW = 8000) ia 0.5 M N&l, (0) dextran

sulfate (MW = 8000) in distilled water.

dextran with the higher molecular weight was more completely excluded, i.e. it had a lower value of K,,. In 0.5 M NaCl dextran solution the swelling ratio was essentially equal to that in distilled water/dextran sulfate solution over the complete range of ionizable monomer content. Although identical swelling ratios were obtained, the value of & was lower for dextran sulfate when the gel contained ionizable monomer. For a nonionic gel (O%Z) the value of K4 was lower for dextran than for dextran sulfate because the molecular weight of dextran was higher. The value of K, for dextran sulfate was essentially constant between 2.5 and lO.O%Z. This unexpected behavior was caused by the increase in swelling as the content of ionizable monomer increased, i.e. the increasing pore size resulting from the larger swelling counteracted the charge effect. To demonstrate the effect of charge more clearly, exclusion tests were performed using dextran sulfate in 0.5 M NaCl. In this solution the swelling ratio was much more nearly constant than it was in distilled water. In 0.5 M NaCl the distribution coefficient for dextran sulfate decreased from 0.37 for nonionic NIPA gel to almost zero for an ionic co-

Exclusion of proteins. Among the important characteristics of proteins are the pendant chemical groups such as amine (-NH,) and carboxylic acid (-COOH) groups. By altering the pH, the net charge on the protein can be changed. At the isoelectric point, p1, a protein has an equal number of positive and negative charges and its net charge is zero. Ion exchange chromatography and hydrophobic interaction chromatography are protein purification techniques which arc dcscribcd in terms of the net charge on the protein and its hydrophobicity. Only recently has the role of heterogeneity on the protein surface been considered (Regnier, 1987; Ruckenstein and Lesins, 1988). If ionic or hydrophobic groups are distributed nonuniformly on the protein surface, interactions may occur between a sorbent and these ionic or hydrophobic re- gions. Consequently, if there is a nonuniform charge distribution on the protein surface, it is not necessary for the net charge of the protein to be opposite to that of the sorbent, but rather the occurrence of an op- positely charged patch on the protein surface is sufi- cient to cause adsorption. Lesins and Ruckenstein (1988) demonstrated that proteins can adsorb to surfaces of like charge. This “patch-controlled” interaction is dependent upon the characteristics of the surrounding medium (i.e. pH, ionic strength) and the adsorbent.

The exclusion behavior of an anionic copolymer gel of acrylamide and sodium acrylate (5.29% T, 2S%C, lO%Z) and a cationic copolymer gel of acrylamide and 3-(methacrylamido) propyltrimethylammonium chloride (6.2%T, 2.15%C, lO%Z) is summarized in Table 1 in terms of the.distribution coefficients. This table also shows the swelling ratios for each gel. For the nonionic solute PEG both anionic and cationic gels gave similar distribution coefficients. The higher molecular weight PEG was excIuded more completely. For proteins the anionic and cationic gels gave markedly different values of K,. At pH 5.8, which is above the isoelectric point (PI), both proteins carry a net negative charge. The interaction between the positively charged network and the negatively charged proteins is controlled by the global charge of the proteins. These results imply that the cationic gel

Table 1. Exclusion behavior of anionic and cationic copolymer gels of acrylamide

Anionic gel Cationic gel

Solute MW MfM, K, MIMd Kd

PEG’ PEG’

1:= 380 0.85 246 0.82

45:OcO 172 0.30 120 028

Ovalbumin (PI = 4.6)’ 113 0.07 115 0.73 Bovine serum albumin @I = 4.8)’ 66,000 157 0.82 113 0.73

tCP = 25 g/l, pH = 6.1. $CF = 1 g/l, pH = 5.8.

38 EBMHIM VWI~~IX~~-FA~AHAN~ et al.

Table 2. Exclusion behavior of copolymer gel of NIPA and R-SO;H+ (8.43% T, 1.58% C, 5 % I) for protein concentration’

Solute MW PH ; &

1 L./ii* = 5 L./d0 = 0.5

&Lactoglobulin 37,ooo 3:; 150 0.19 0.19 (pI=I=z;)

45,000 Sf

132 52 0.66 0.01 0.01 3.22

(PI = 4.6) 53 5.0 7.2 Bovine serum albumin 66,000 11 105 0.05 0.07

(PI = 4.8) 5.8 127 0.82 1.85 3.8 60 3.82 8.30

‘Cr = 0.25 8/r.

is not suitable to concentrate protein solutions, although it showed desirable swelling characteristics (Vasheghani-Farahani et al., 1990).

Ovalbumin was more excluded, i.e. had a lower K,, than bovine serum albumin, even though the latter has a higher molecular weight. One factor which may contribute to this behavior is the extended shape of ovalbumin which has dimensions of 33 x 33 x 96A (Haurowitx, 1963). In spite of the high molecular weight of bovine albumin and its net negative charge at pH 5.8, its distribution coefficient was near unity for the anionic gel. This result suggests that nonuniform charge distribution on the surface of the protein controls the behavior and the protein adsorbs onto the outer surface of the gel particles.

To test this hypothesis anionic NIPA gel particles were equilibrated with protein solutions at different pH values. The gel particles were cylinders with two different length-to-diameter ratios, L,/d, = 5 and LJd, = 0.5. Experiments were conducted in duplic- ate: one set used the low LJd,, particles and the other used the high L,/d, particles. The same mass of gel was used in each case. The surface area of the particles having Lo/d, = 0.5 was approximately twice that of the particles having Lo/d, = 5; hence, if surface adsorption were important, the calculated value of K, would be larger for Lo/d,, = 0.5. The results of the experiments are shown in Table 2. At pH 5.8 the distribution coefficient. of B-lactoglobulin was the same for short or long gel particles, indicating that the net charge of the protein was responsible for the interaction between the solute and the ionic network. At pH 2.9, however, the net charge of the protein was positive and it adsorbed onto the surface of the negatively charged hydrogel giving distribution coefficients above those at pH 5.8. In addition, the value of K, was larger for the low length-to-diameter ratio particles. Similar behavior was exhibited by ovalbumin. For bovine serum albumin at pH 11, well above the isoelectric point, the solute was almost completely excluded, but there was some small surface adsorption which might be due to hydrophobic interactions. At pH 5.8, still above p1, the distribution coefficient of bovine serum albumin was larger for the shorter particles, indicating that the nonuniform charge distribution on the surface of the protein

(patch-controlled mechanism) was important. At pH 3.8 the net charge of the bovine serum albumin was positive, there was strong adsorption by the negatively charged network and K1 increased with decreasing L, Id,,.

The effect of ionizable monomer was examined for concentration of cytochrome c (MW = 12,330) with NIPA gels. At pH 5.5, which is well below the isoelectric point of cytochrome c (PI = 9.3), the positively charged solute exhibited high afhnity towards anionic and nonionic NIPA gels. The distribution coefficient increased with increasing the amount of ionizable monomer, ranging from 4.2 for nonionic NIPA gel to 7.3 for a copolymer gel of NIPA and R-SO;H+ (8.6%T, 1.31%C, 7S%Z). The value of K, = 4.2 for the nonionic gel indicates that in addition to the electrostatic interactions, hydrophobic effects were involved between cytochrome c and the gel network.

Ej4ect of soZuute concentration. It is of practical im- portance to show to what extent dilute aqueous solutions of biological products can be concentrated effi- ciently. The effect of solute concentration on the distribution coefficient of ovalbumin is shown in Fig. 10 for a copolymer gel of NIPA and R-SO; H + (8.6%T, 1.31%C, 7.5%1). At low concentrations (i.e. CR d 1 g/l) ovalbumin was completely excluded; however, the distribution coe&ient increased, i.e. solute exclusion decreased with increasing ovalbumin concentration even though the swelling ratio decreased. This result implies that dilute solutions can be concentrated in repeated cycles of water sorption, but that exclusion efficiency decreases as the number of cycles increases. The increased distribution coefh- cient is attributed to the entrainment of small amounts of the retentate between gel particles due to the increased viscosity of more concentrated solutions. Increasing the gel bead size and washing the swollen gel should reduce the effect of solute entrainment.

Enzyme concentration. Table 3 gives the swelling and exclusion behavior of nonionic NIPA gel (8.09% T, 1.64%C) and an ionic copolymer gel of NIPA and sodium vinylsulfonate (8.15% T, 1.63 %C, 5%Z) in solutions of different enzymes in the presence of salts

Concentration of large biomolecules with hydrogels

Table 3. Exclusion behavior of NIPA gels for concentration of 0.25 g/l enzyme solutions at 6°C

Solute MW

NIPA NIPA + Na-vinyl- gel sulfonate gel

M/M, & M/M, &

a-Amylaset a-Amylases

Lipase’ (Candida cyliadracea)

Lipase’ (wheat germ)

p-Galactosidase+ ,+Galaetosidase*

u 52,500 20 0.21 57 0.40 51 52,500 17.6 0.17 18.6 0.20

_ 100,000 19.2 0.02 37 020

Unknown 19 0.27 22.5 0.28

466,140 24 0.0 70 0.0 466,140 17.3 0.0 19.7 0.0

‘Distilled water at pH 5.8. *9.3 x IO-z M KH,PO, + 3.9 x lo-* M MgSO,. ‘5.6x lo-* M KH5P0, + 2.3 x lo-* M MgSO,.

Fig. 10. Effect of solute (ovalbamin) concentration on the swelling and exclusion behavior of copoIymer gel of NIPA and R-SO;H+ (7.5%1) at pH 5.8: (0) swelling, (0) ex-

clusion.

which are frequently encountered in buffered enzyme solutions. The fl-galactosidase was completely excluded by both ionic and nonionic gels. Lipase (wheat germ) was only partially excluded by both ionic and nonionic gels. Lipase (Candidu cylindracea) was excluded almost completely by the nonionic gel, but it was partially (K,, = 0.2) excluded by the ionic gel. a-Amylase was more excluded by the nonionic gel, but complete exclusion was not achieved with either gel.

A successful exclusion process does not affect the activity of the enzyme upon concentration. To test the effect of the gel exclusion process on the activity of enzymes, the retentate concentration was determined and then this solution was diluted to yield a solution with a concentration equal to that of the feed. The activities of the diluted retcntate and the feed were measured. The activity of a-amylase was about 720 (Somogyi units/mg solid) in both feed and retentate. The lipase activity in both feed and ranate was about 12.4 (Sigma-Tietz units/mg solid). The enzymatic activity of the j?-gaJactosidase was 105

results Jndkate that the concentration process using NIPA gels had no adverse effect on the activity of these enzymes.

CONCLUSION

Crosslinked polymer gels, whose swelling is sensitive to the temperature or concentration of the surrounding medium, can be exploited to concentrate cliiute aqueous solutions of macromolecular solutes including proteins (cnzymcs) in a gel extraction process. The efficiency of solute exclusion, based on the size of the solute and the pore size of the network, increased as the polymer volume fraction increased. The variable with the largest effect on the size exclusion behavior of a gel was the monomer concentration at gel formation.

Simple ionic solutes were excluded more effectively by polyelectrolyte gels due to Donnan equilibrium. The interaction between biological products and polyelectrolyte gels was not determined solely by the net charge of the solute. The nonuniform charge distribution on the surface of biological macromolecules plus hydrophobic effects also play an important role. The concentration of enzyme solutions using equilibration with NIPA gels did not decrease the activity per unit mass of enzyme.

Acknowledgement-This work was supported by the Natural Sciences and Engineering Research Council of Canada.

c, Cx % c d dc. %Z Kd

L LO M

NOTATION

solute concentration in the gel phase feed conantration retentate concentration weight percentage of crosslinker diameter of gel particle diameter of gel particlo at preparation mole percentage of ionizable monomer distribution coefficient length of gel particle length of gel particle at preparation swollen gel mass

(units/mg solid) in both feed and the retentate. These Md dry mass of polymer

40 EsRAHn.4 VASHEGH

M, mass of gel after regeneration

M, swollen gel mass at time t MO original mass of gel at preparation

MW molecular weight % T grams of total monomers per 100 cm3 of

solution at preparation V gel volume

vF feed volume

vR retentate volume V polymer volume fraction

Greek letters

v fraction of solute excluded

Plr gel density

PP polymer density

REFERENCES

Abe. T., Egava, H. and Nit& A., 1990, Synthesis and charac- terization of thermosensitive polymeric beads. J. uppl. PO@. Sci. 40, 1223-1235.

Amiya, T., Hirokawa, Y., Hirose. Y., Li, Y. and Tanaka, T., 1987, Reentrant phase transition of N-isopropylacrylamide gels in mixed solvents. J. them. Pkys. 86, 2375-2379.

Base&a, I., Hernandez-Fuentes, I., Pierola, I. F. and Llorente, M. A., 1987, Elastic properties of highly crosslinked polyacrylamide gels. Macromolecules 28. 3060-3065.

Base@, J., Llorente, M. A., Hern&ndez-Fuentes, I. and Pierola, I. F., 1989, Network defects in polyacrylamide gels. Eru. Polym J. 25.471-47s.

Beltran, S., Hooper, H. H., Blanch, H. W_ and Prausnitz, J. M., 1990, Swelling equilibria for ionized temperature- sensitive gels in water and in aqueous salt solutions. .I. them. Phys. 92,2061-2066.

Cussler, E. L., Stokar, M. R. and Varberg, J. E., 1984, Gels as size selective extraction solvents. A.2.Ch.E. J. 30,578-582.

Dtiek, K. and Patterson, D., 1968. Transition in swollen polymer networks induced by intramolecular condensa- tion. J. Polym. Sci. A26, 1209-1216.

Erickson, R. P., 1970, Molecular weight of Escherichia coli j3- galactosidase in concentrated solutions of guanidine hy- drochloride. Biochem. J. 120.255-261.

Flodin, P., Gelotte B. and Porath J., 1960, A method for concentrating macromolecules. Nature lgg493-494.

Freitas, R. F. S. and Cussler, E. L., 1987, Temperature sensitive gels as extraction solvents. Cheat. Engng Sci. 42, 97-103.

Gehrke, S. H., Andrews, G. P. and Cussler, E. L., 1986, Chemical aspects of gel extraction. C&m. Engng Sci. 41, 2153-2160.

Goldberg, M. E., 1969, Tertiary structure of Escherichia coli B-D-gsactosidaae. .I. mol. B&L 46, 441446.

Haurowitz, F., 1963, Size and shape of protein molecules, in The Chemistry and Function of Proteins, pp. 62-Y 1. Aca- demic Press, New York.

ANI-FARAHANI et al.

Hirotsu, S.. Hirokawa. Y. and Tanaka, T., 1987, Volume- phase transitions of ionized N-isopropylacrylamide gels, J. c&n. Phys. 87, 1392-1395.

Ilavskjr, M. and Hrouz, J., 1983, Phase transition in swollen gels. 5. Effect of the amount of diluent at network formation on the collapse and tnechanical behaviour of polyacrylamide networks. Polym. Bull. 9, 159-166.

Kalnins, A., Otto, K., Rather, U. and Miiller-Hill, B., 1983, Sequence of the la&Z gene of Escherichia coli. EMS0 J. 2, 593-597.

Lesins, V. and Ruckenstein, E.. 1988. Patch controlled attractive electrostatic interactions between similarly charged proteins and adsorbents. Colloid Polym. Sci. 266, 1187-1190.

Matsuo, E. S. and Tanaka, T.. 1988, Kinetics of diswn- tinuous volume-phase transition of gels. J. &em. Phys. QQ, 16951703.

Miller, J. H., 1972, Experiment 57, Purification of /I-gaiac- tosidase, in Experiments in Molecular Genetics, pp. 398404. Cold Sminn Harbor Laboratorv. New York.

Ohmine, I. and Tanaka, T., 1982, Salt effedts on the phase transition of ionic gels. J. cheat. Phys. 77, 5725-5729.

Regnier, F. E., 1987, The role of protein structure in chromatographic behavior. Science 23& 319-323.

RiEka, f and Tanaka, T., 1984, Swelling of ionic gels: quant- itative performance of the Donnan theory. Macromole- cules 17,2916-2921.

Ruckenstein, E. and Lesins, V., 1988, Classification of liquid chromatographic methods based on the interaction forces: the niche of potential barrier chromatography, in Down- stream Processes: Equipment and Techniques (Edited by A. Mizrahi). pp. 241-314. Alan R. Liss, New York.

Somogyi, M., 1960, Modifications of two methods for the assay of amylase. Clin. Chem. 6.23-35.

Tanaka, H., Fukumori, K. and Nishi, T., 1988, Study of chemical gelation dynamics of aorylamide in water by real-time pulsed nuclear magnetic resonance measure- ment. .I. them. Phys. %9, 3363-3372.

Tanaka T.. 1978. Collause of aels and the critical endpoint. Phys.. Rev. Let;. 40, 820-823:

Tietz. N. W. and Fiereck, E. A., 1966. A specific method for serum lipase determination. Clin. &him.-Acta 13, 352-358.

Tomizuka, N., Ota, Y. and Yamada, K., 1966, Studies on liuase from Cundti cvliudracea Part II. Amino acid composition, carbohydrate component and some physical moperties. Aar. biol. Chem. 30, 109%1096.

T&r&, S. J., Johnson, D. W. and Cussler, E. L., 1989, Isolated soy protein production using temperature-sensitive gels. -J. Fd Technol. 43161 78-83. . _,

Vasheghani-Farahani, E., 1990, Swelling and exclusion behavior of bydrozels. Ph.D. thesis, McGill University, Montreal, Quetn&, Canada.

Vasheghani-Farahani, E.. Vera, I. H.. Cooper, D. G. and Weber. M. E.. 1990. Swellinn of ionic zels in electrolyte solutions. Ind: En& Chem. Res. 29, 55-+560. _

Vasiliev, V. G., Roaovina, L. 2. and Slonimsky, G. L., 1985, Dependence of properties of swollen and- dry polymer networks on the conditions of their formation in solution. Polymer 26, 1667-1676.

Concentration of large biomolecules with hydrogels

Documents