Biomacromolecules 2004, 5, 1637-1641
1637
Protein Transfer through Polyacrylamide Hydrogel Membranes Polymerized in Lyotropic Phases Michael J. Monteiro,*,† Geoff Hall,‡ Sarah Gee,‡ and Li Xie‡ School of Molecular and Microbial Sciences, Australian Institute of Bioengineering and Nanotechnology, University of Queensland, Brisbane QLD 4072, Australia, and Gradipore, Limited, P.O. Box 6126, Frenchs Forest NSW 2086, Sydney, Australia Received April 8, 2004; Revised Manuscript Received June 22, 2004
A way to control the average pore size in cross-linked polyacrylamide-based membranes is by altering the ratio of cross-linker to acylamide monomer. Larger pore sizes are prepared with a minimum amount of cross-linker, resulting in membranes that are mechanically weak and have short lifetimes. The aim of this study was to prepare cross-linked polyacrylamide membranes with large pore sizes and with good mechanical integrity. The methodology was to carry out the polymerization in a template, formed from the self-aggregation of surfactant. Two surfactant templates were used, and their pore size was examined with proteins of different sizes. The surfactants chosen for this study were sodium dodecyl sulfate (SDS, ionic surfactant) and TERIC BL8 (nonionic surfactant), both of which have very different aggregation properties. The data showed that at 10% and greater of TERIC BL8, a very different and open gel structure is formed, in which the pore size was significantly increased. SDS seemed to have little effect on the pore size. The data suggests that the gel structures for both surfactants up to 4% (w/v) are similar and micellular, because SDS is known to favor a micelle structure. Above 4% (w/v), TERIC BL8 then goes through a change in its lyotropic phase, thus, producing membranes of a large pore size. In conclusion, the pore size and gel structure of polyacrylamide hydrogel membranes can be significantly increased using TERIC BL8 (nonionic) surfactant. This allows large-pore-size membranes with a high cross-link density and consequently high mechanical strength to be prepared for the separation of large biomolecules. Introduction Polyacrylamide hydrogel membranes have been used to separate biomacromolecules in an electrophoretic environment.1 The Gradiflow instrument uses polyacrylamide membranes for the separation and isolation of proteins based on either size or charge in an electric field (see Figure 1 for more details of the Gradiflow instrument).2,3 This instrument has been successfully used in a variety of applications, including the separation of monoclonal antibodies,4 proteins from plasma,2 certain recombinant proteins,5 viruses,6 and to concentrate and remove infectious and noninfectious prion proteins from human biological samples. The membranes are formed through the free-radical polymerization of acrylamide (AAm) in the presence of a cross-linking agent, N,N′methylenebisacrylamide (Bis), which results in the formation of a mesoporous network.7,8 The pore size distribution can be changed by altering the total weight of monomers to the volume of the solution (%T) and mass of cross-linker to the total mass of monomers (%C). Separation of larger biomolecules requires larger pore sizes, which can be achieved by lowering %T and %C. However, the cross-link density is consequently reduced and so too is the elastic modulus. This * To whom correspondence should be addressed. E-mail: m.monteiro@ uq.edu.au. † University of Queensland. ‡ Gradipore, Limited.
Figure 1. Schematic diagram of the Gradiflow. The four chambers in the separation unit are connected to three flow circuits, each consisting of a reservoir and a circulating pump. The Gradiflow is cooled by an ice-filled stainless steel beaker placed in the buffer reservoir. Two stainless steel tube heat exchangers in the buffer reservoir cool Stream 1 and Stream 2. The electrodes are connected to an external power.
has an adverse effect of mechanically weakening the membranes. It has been proposed that, to concomitantly increase both the mechanical strength and pore size, polymerizations should be carried out with high %T and %C in the presence of a
10.1021/bm049789m CCC: $27.50 © 2004 American Chemical Society Published on Web 08/18/2004
1638
Biomacromolecules, Vol. 5, No. 5, 2004
template consisting of self-assembled surfactant structures.9-12 The resulting hydrogel membrane would consist of a highly cross-linked network surrounding these self-assembled surfactant templates (e.g., spherical micelles). The surfactant is then removed by washing with water, leaving behind voids where the original self-assembled surfactant structures had once been. The hydrogel membrane would consist of a high cross-link density (of high mechanical strength) with large pores, which are a copy of the surfactant structures. However, the polymerization of a network evokes entropic restrictions because of its confined geometry and, thus, a change in surfactant structure to a different lyotropic phase. Therefore, the membrane morphology is not a direct cast of the original surfactant assembly.9,12,13 The morphology was shown to be controlled by the kinetics of the polymerization and the demixing of the gelling polymer and surfactant phases.12 The demixing process results in a more confined volume for the surfactant and, thus, a change in the lyotropic phase. The change is dependent on the type and concentration of the surfactant. It has been shown that cetyltrimethylammonium bromide changes from globular micelles to the more stable hexagonal phase at weight percents greater than 20%.14 Therefore, it is conceivable that during the polymerization the local surfactant concentration can increase with a phase change. This will be dependent upon the rate of polymerization and when the gel point is encountered. Antonietti et al.12 polymerized a range of monomers in the presence of surfactants, differing in hydrophobicities from Brij56 (hydrophilic-lipophilic balance, HLB, 12.9), Tween60 (HLB 14.9) to Brij58 (HLB 15.7). It was found that these surfactants allowed the preparation of highly ordered polymer gels that had a much greater pore size than the original mesophase structure. This supported their argument that demixing of the surfactant and polymer phases during polymerization resulted in a larger hexagonal template phase. Interestingly, the choice of surfactant had little influence on the gel morphology, which was explained by the initial phase being hexagonal prior to polymerization and maintaining this phase throughout the reaction. The aim of this work was to prepare polyacrylamide membranes with a high mechanical strength and a pore size that allows the transfer of large biomolecules. The surfactants chosen were sodium dodecyl sulfate (SDS, an anionic surfactant) and TERIC BL8 (a nonionic surfactant, HLB 14.5). TERIC BL8 was chosen because it had a similar hydrophobicity to Tween60 and a HLB value between the other two Brij surfactants used in previous studies.12 SDS was used to determine whether ionic surfactants would also alter the gel architecture. The membranes were washed with water to remove the surfactant and then tested with proteins (in their native state) of various molecular weights to determine the “effective” pore size (see Table 1). Experimantal Section Materials. AAm (99% purity) was purchased from ICN Biochemials, Inc. (U.S.A.). Bis (>98%), N-tris(hydroxyl)methyl-3-aminopropane sulfonic acid, and tris(hydroxymethyl)aminomethane were purchased from Ultrapure. Am-
Communications Table 1. List of Proteins Used To Test Pore Size of the Separation Membranesa R-lactalbumin trypsin inhibitor alcohol dehydrogenase catalase ferritin thyroglobulin a
Mw (kDa)
mg/mL
14 20 140 232 440 660
1 1 1 1 0.1 1
Note: The proteins are diluted in the running buffer.
monium persulfate (>98%), ammonium metabisulfite (>98%), and SDS (>98%) were purchased from Aldrich Fine Chemicals (Australia). Terric BL8 surfactant was purchased from Huntsman Australia (>97%). All protein used in this work were purchased from Sigma (>98%, U.S.A.). Polyacrylamide Membrane Synthesis. The membranes were synthesized using AAm, Bis, ammounium persulfate, and either SDS or TERRIC BL8 in water and purged with nitrogen for 40 min to remove excess oxygen. The other redox couple (ammonium metabisulfite) was added, and the reaction mixture was immediately cast between two glass plates containing a PET support membrane under a nitrogen atmosphere and left to polymerize for a further 3 h. The membranes were then washed thoroughly with deionized water to remove excess surfactant. Residual surfactant was removed during the Gradiflow run under a current in the absence of protein. This should remove most of the SDS but should have little effect in removing the TERRIC BL8. Gradiflow Instrument. A description of the Gradiflow instrument is given in Figure 1. The membrane synthesized in this work was used as the separation membrane with a small pore size membrane on either side to stop protein transfer into the back buffer stream. A solution of tris[hydroxymethyl]aminomethane and N-tris[hydroxymethyl]methyl-3-aminopropane sulfonic acid at pH 8 was circulated through the streams at a flow rate of 20 mL/min. Stream temperatures during the run are generally between 8 and 14 °C. The voltage was set to 250 V, and the current was noted at the beginning and at the end of the run. A precise amount of protein (in general 1 mg/mL) was introduced into stream 1 (total volume 10 mL). Determination of Young’s Modulus (E). The Young’s modulus was determined from eqs 1 and 2. The volume fraction V2 was determined from swelling experiments of the gel with water. The gel was dried in a vacuum oven overnight or until a constant weight was reached. The weighted dried gel was immersed in deionized water for 24 h, then patted with filter paper to remove surface water, and weighed. Results and Discussion Membrane synthesis was based on previous recipes to give a range of pore sizes.7,15,16 The smallest pore size membranes were prepared with 35%T and 8.99%C, to the largest pore size prepared using 15.04%T and 2.82%C. Therefore, different proteins were used to quantify the pore size of each set of membranes. The polyacrylamide membranes were prepared by polymerizing AAm, Bis, and water at room temperature with a
Biomacromolecules, Vol. 5, No. 5, 2004 1639
Communications Table 2. Membranes Prepared with 35.6%T and 8.99%C and Varying Amounts of Surfactant, TERIC BL8 or SDS
Table 3. Membranes Prepared with 16.08%T and 5.17%C and Varying Amounts of Surfactant, TERIC BL8 or SDS
% transfer (30 min) % surfactant
lactalbumin 14 kDa
trypsin inhibitor 20 kDa
0.02% TERIC 0.4% TERIC 4% TERIC 10% TERIC 50% TERIC
15 16 26 100 100 18 24 23 23
0.02% SDS 0.4% SDS 4% SDS 10% SDS
ferritin 440 kDa
% transfer (30 min) thyroglobulin 660 kDa
% surfactant
11 18 33 80 91
0 0 0 37 25
0.02% TERIC 0.4% TERIC 4% TERIC 10% TERIC
28 23 25 28
0 0 0 0
redox couple, consisting of sodium persulfate and sodium metabisulfite, in the presence of varying amounts of surfactant under an inert atmosphere and using poly(ethylene terephthalate) nonwoven paper as a support. The redox initiators were chosen to exclude the partitioning of either sodium persulfate or sodium metabisulfite compounds into the surfactant hydrophobic phase. Initial experiments carried out used the redox couple, sodium persulfate, and N,N,N′,N′tetramethylethylenediamine (TEMED), in which TEMED partitioned to a large extent into the surfactant hydrophobic phase, resulting in little or no polymerization. Therefore, TEMED was substituted with the water-soluble sodium metabisulfite. These membranes were then placed in the Gradiflow, and their transfer efficiency was tested with a range of proteins. It should be noted that the hydrodynamic volume and structural conformation of proteins may change in the electrophoretic medium (i.e., pH and ionic strength) used in the Gradiflow; therefore, the Mw’s given in Table 1 are used solely as a guide. The buffer used was tris(hydroxy)aminomethane and N-tris(hydroxy)methyl-3-aminopropane sulfonic acid, which was formulated to give a pH of 8.5. All the proteins in Table 1 were selected because they have pI’s (isoelectric points) less than 6 and are, thus, negatively charged at pH 8.5. When the electric field is applied, the negatively charged proteins will move across the membrane from the starting stream (Stream 1) to the collection stream (Stream 2) where the positive electrode is located. The percent transfer was calculated from the amount of protein in Stream 2 after 30 min detected using a UV-vis spectrometer at 280 nm. In all experiments, recovery was greater than 90%, in which 90% of the proteins were either in Stream 1 or in Stream 2 and 10% were bound or captured by the membrane. For our purpose, this is an acceptable value. The first set of membranes synthesized was based on 35.6%T and 8.99%C (i.e., 4.584 M AAm, 0.254 M Bis), 0.183 M sodium persulfate, 0.183 M sodium metabisulfite, water with varying the percentage weight to volume (% w/v) of TERIC BL8 or SDS. These membranes should have the smallest pore size compared to the other two sets of membranes prepared in this work. As Table 2 shows, at the lowest %w/v (0.02% TERIC BL8) transfer decreased with increasing the size of the protein, suggesting that for this recipe, the pore size is quite small. As the amount of TERIC
0.02% SDS 0.4% SDS 4% SDS
alcohol dehydrogenase catalase ferritin thyroglobulin 140 kDa 232 kDa 440 kDa 660 kDa 70 58 90 89 100 60
27 19 51 78
16 17 38 77
12 13 28 81
90 40 50
70 40 40
BL8 is increased to 4%, there seems to be little or no change in the percent transfer for the three proteins tested. However, above 4% the rate of transfer is drastically increased from 26 to 100% for the smallest protein (lactalbumin) and from 33 to 91% for trypsin inhibitor, and even for the large thyroglobulin transfer increased from 0 to 25%. In the case of SDS, there seems to be little or no change in transfer for all the proteins. This suggests that SDS has little or no influence on the gel structure for this particular set of membranes. The second set of membranes synthesized were based on 16.08%T and 5.17%C (i.e., 2.145 M AAm, 0.067 M Bis), with all other conditions remaining the same as above. The recipe will result in a larger pore size than that shown above. Table 3 shows that decreasing %T and %C increased the pore size in comparison to the first set of membranes. Transfer increased for thyroglobulin from 0 to 12% at 0.02% TERIC BL8. In addition, proteins larger than lactalbumin and trypsin inhibitor, that is, alcohol dehydrogenase and catalase, showed transfers of 70 and 27%, respectively. Upon increasing the amount of TERIC BL8, alcohol dehydrogenase showed a small increase in transfer, whereas catalase transfer increased from 19 (0.4% TERIC) to 51 (at 4% TERIC BL8) to 78% (at 10% TERIC BL8). For both ferritin and thyroglobulin, the transfer increased drastically for TERIC BL8 concentrations greater than 4%, which is similar to what was found for the first set of membranes. Membranes using 0.02% SDS showed high transfer rates that decreased from 100% for alcohol dehydrogenase down to 70% for the larger thyroglobulin. An increase of SDS to 0.4% resulted in an apparent smaller pore size. The transfer of alcohol dehydrogenase decreased to 60%, while both ferritin and thyroglobulin decreased to 40%. There seemed to be no further decrease at higher SDS amounts. The data suggests that the gel microstructure, resulting in a smaller pore size, is affected by increasing the amount of SDS from 0.02 to 0.4% but does not change further upon addition of a much greater amount of SDS (4%). This suggests that the SDS micellular phase does not change and favors its micellular structure. The third set of membranes made were based on 15.04%T and 2.82%C (i.e., 2.056 M AAm, 0.034 M Bis), with all other conditions remaining the same as above. Table 4 shows that this set of membranes shows only a small increase in transfer of proteins compared to the second set with 16.08%T and 5.17%C. The transfer was relatively constant until 10% TERIC BL8, which showed for all proteins an increase from
1640
Biomacromolecules, Vol. 5, No. 5, 2004
Communications
Table 4. Membranes Prepared with 15.04%T and 2.82%C and Varying Amounts of Surfactant, TERIC BL8 or SDS % transfer (30 min) % surfactant 0.02% TERIC 0.4% TERIC 4% TERIC 10% TERIC 50% TERIC
alcohol dehydrogenase catalase ferritin thyroglobulin 140 kDa 232 kDa 440 kDa 660 kDa 33 31 31 74
2 24 11 73
20 20 16 82
Figure 2. Effect of increased percent TERIC BL8 on the Young’s modulus of poly(acrylamide-bisacrylamide) cross-linked membranes.
below 31% to above 73%. Using SDS, similar to the other sets of membranes, resulted in no changes in the percent transfer (not shown in Table 4). The mechanical strength of the membranes can be semiempirically quantified by using the Flory-Rehner equation that relates the swelling ratio of cross-linked networks to the cross-link density, n.
[
-[ln(1 - V2) + V2 + χV22] ) V1n V21/3 -
]
V2 2
(1)
where V2 is the volume fraction of polymer swollen mass, V1 is the molar volume of the solvent, and χ1 is the FloryHuggins polymer-solvent interaction parameter. For polymers to dissolve readily in a solvent, χ1 must be less than 0.5. The values of χ1 can be calculated a priori from group contributions to the cohesive energy density. However, this value of 6 is less than satisfactory because it is much greater than that for a good solvent. Therefore, in our calculations we used a value of 0.39, which is taken for a well-known cross-linked system [i.e., poly(butadiene-styrene)/toluene]. Although the values of n are not quantitatively accurate, they allow us to qualitatively compare the cross-link density between poly(acrylamide-bisacrylamide) membranes in a water environment. The quantity n can be used to predict the Young’s modulus (E) of the membrane from the equation: E ) 3nRT
(2)
where R is the gas constant and T is temperature. Figure 2 shows the effect of %T and %C and concentration of TERIC BL8 on the relative Young’s modulus of the three sets of poly(acrylamide-bisacrylamide) membranes. A small increase in the %C from 2.82 to 5.17% and keeping %T relatively constant shows only a slight increase in the modulus. It is
only when the %T and %C are nearly doubled did the modulus increase by a factor of 4. In most cases, such high %T and %C would result in formation of membranes with small pore sizes, allowing low transfer rates of lactalbumin and trypsin inhibitor through the membrane (Table 2). It can also be shown that an increase in the concentration of TERIC surfactant had little or no effect on the cross-link network formation but did have a large effect on the pore size. For example, the membrane (35.6%T and 8.99%C) polymerized in 10% TERRIC and with the highest modulus allowed similar transfer rates of thyroglobulin in comparison to a mechanically weaker membrane (i.e., 15.04%T and 2.82%C) polymerized in 0.02% TERIC. The data presented showed that at 10% and greater of TERIC BL8 a very different and open gel structure is formed, in which the pore size is significantly increased. SDS seemed to have little effect on the pore size. For example, the comparison between the TERIC BL8 and SDS experiments in the first set of membranes (34.8%T and 8.97%C) showed that protein transfer was similar for both surfactants until 10%w/v is used. The data suggests that the gel structures for both surfactants up to 4%w/v are similar and micellular, because SDS is known to favor a micelle structure. Above 4%w/v, TERIC BL8 then goes through a change in its lyotropic phase, assumed to be hexagonal on the basis of Antonietti’s data for similar nonionic surfactants,12 and, thus, producing membranes of large pore size. In conclusion, the pore size and gel structure of polyacrylamide hydrogel membranes can be significantly increased using TERIC BL8 (nonionic) surfactant. This allows large pore size membranes with a high cross-link density and consequently high mechanical strength to be prepared for the separation of large biomolecules. Acknowledgment. The authors wish to thank Dr. Steve Botto and Dr. Hari Nair from Gradipore for their support. References and Notes (1) Chiari, M.; Nesi, M.; Roncada, P.; Righetti, P. G. Electrophoresis 1994, 15, 953-959. (2) Horvath, Z. S.; Corthals, G. L.; Wrigley, C. W.; Margolis, J. Electrophoresis 1994, 15, 968-971. (3) Rylatt, D. B.; Napoli, M.; Ogle, D.; Gilbert, A.; Lim, S.; Nair, C. H. J. Chromatogr., A 1999, 865, 145-153. (4) Lim, S.; Manusu, H. P.; Gooley, A. A.; Williams, K. L.; Rylatt, D. B. J. Chromatogr., A 1998, 827, 329-335. (5) Corthals, G. L.; Molloy, M. P.; Herbert, B. R.; Williams, K. L.; Gooley, A. A. Electrophoresis 1997, 18, 317-323. (6) Gilbert, A.; Evtushenko, M.; Nair, H. Ann. N.Y. Acad Sci 2001, 936, 625-629. (7) Righetti, P. G. J. Chromatogr. A 1995, 689, 3-17. (8) Patras, G.; Qiao, G. G.; Solomon, D. H. Macromolecules 2001, 34, 6396-6401. (9) Holtzscherer, C.; Wittmann, J. C.; Guillon, D.; Candau, F. Polymer 1990, 31, 1978. (10) Friberg, S. E.; Yu, B.; Ahmed, A. U.; Campbell, G. A. Colloids Surf. 1993, 69, 239. (11) Go¨ltner, C. G.; Antonietti, M. AdV. Mater. 1997, 9, 431. (12) Antonietti, M.; Caruso, R. A.; Go¨ltner, C. G.; Weissenberger, M. C. Macromolecules 1999, 32, 1383. (13) Antonietti, M.; Hentze, H. P. Colloid Polym. Sci. 1996, 274, 696. (14) Antonietti, M.; Go¨ltner, C. G.; Hentze, H. P. Langmuir 1998, 14, 2670.
Communications (15) Solomon, D. H.; Qiao, G. G.; Patras, G. Process for microgel preparation. PCT Int. Appl., WO 9831739, 1999 (The University of Melbourne, Australia). (16) Solomon, D. H.; Qiao, G. G.; Caulfield, M. J.; Atchison, S. Membranes and method of manufacture thereof. PCT Int. Appl.,
Biomacromolecules, Vol. 5, No. 5, 2004 1641 WO 20010820, 2002 (Gradipore, Limited, Australia). Appl.; (Gradipore Limited, Australia).
BM049789M
