314
Bioconjugate Chem. 2001, 12, 314−319
Mechanistic Investigation of Smart Polymer-Protein Conjugates Tsuyoshi Shimoboji, Zhongli Ding, Patrick S. Stayton,* and Allan S. Hoffman* Department of Bioengineering, University of Washington, Seattle, Washington 98195. Received September 25, 2000; Revised Manuscript Received November 8, 2000
Many affinity separation and diagnostic applications rely upon both capture and release steps. There is thus a need for methods to enhance the reversibility of biomolecular interactions. We have previously demonstrated that stimuli-responsive polymers can be used to gate biomolecular reactions when conjugated near the active site of proteins. Here we have used a new smart polymer, N,N-dimethyl acrylamide-co-4-phenylazophenylacrylate that has allowed a mechanistic investigation of the smart polymer switches. This polymer was conjugated via a vinyl sulfone terminus to cysteine residues of genetically engineered streptavidin mutant E116C, where the polymer is conjugated close to the biotinbinding site, and streptavidin mutant S139C, where the conjugation site is distant. The biotin binding switching activity was strongly dependent on conjugation position, as the E116C conjugate displayed a large thermal response while the S139C conjugate displayed only small effects. Kinetic measurements of biotin release demonstrated that the off-rate of biotin was unperturbed and that the thermally triggered release of biotin with the E116C conjugate was due to the blocking the reassociation of biotin. The addition of free polymer to purified E116C conjugates was also shown to increase the blocking and release properties of the switch. This effect was site dependent, suggesting that the conjugated polymers were directing a physical aggregation near the binding site that effectively enhanced the switching activity. These investigations provide mechanistic insight that can be utilized to design better molecular switches for a variety of stimuli-responsive polymer-protein conjugates.
INTRODUCTION
Capture and release steps are involved in many affinity separation and diagnostic applications, and the reversibility of affinity interactions is thus an important underlying aspect of these technologies. Reversibility is intrinsically problematic, however, because the capture step typically requires high binding energy and specificity. This often makes it necessary to use relatively harsh conditions for release that can damage delicate biological components or targets. A striking example of these conflicting needs is found with streptavidin, one of the most utilized affinity reagents. Streptavidin (SA)1 displays a remarkably high affinity for the ligand biotin, and this interaction is utilized in a wide variety of biological, clinical, and diagnostic applications (1, 2). The high affinity is also a major limitation, however, for applications where the recovery of the biotinylated molecules are desirable. Many strategies for reversibly manipulating biomolecular recognition have been devised, but there is still a clear need for improvement. We have been developing a new approach that utilizes site-specific conjugation of * To whom correspondence should be addressed. (A.S.H.) Phone: (206) 543-9423. Fax: (206) 543-6124. E-mail: hoffman@ u.washington.edu. (P.S.S.) Phone: (206) 685-8148. Fax: (206) 685-8256. E-mail:
[email protected]. 1 Abbreviations: AIBN, 2,2′-azobisisobutyronitrile; AZAA, 4-phenylazophenylacrylate; DMA, N,N-dimethylacrylamide; DMAA, N,N-dimethylacrylamide-co-4-phenylazophenylacrylate copolymer; DMAA-VS, vinyl sulfone terminated DMAA; DMF, dimethylformamide; DTNB, 5,5′-thiobis (2-nitrobenzoic acid); DVS, divinyl sulfone; EDTA, ethylenediamine tetraacetic acid; LCST, lower critical solution temperature; MEO, mercaptoethanol; MW, molecular weight; SA, streptavidin; TCEP-HCl, tris(2-carboxyethyl)-phosphine hydrochloride; THF, tetrahydrofuran; VS, vinyl sulfone.
stimuli-responsive polymers near the protein-binding site. These stimuli-responsive polymers undergo a large change in size and physical properties in response to external stimuli such as temperature, pH, or specific wavelengths of light. We have previously demonstrated that the change in physical properties can be used as a temperature- or pH-dependent activity switch to control ligand capture and release (3-5). We report here an investigation into the mechanism of these smart polymerprotein conjugates that utilizes the new end-reactive polymer N,N-dimethyl acrylamide-co-4-phenylazophenylacrylate (DMAA). The use of this polymer and a control streptavidin mutant has allowed a determination of the role of conjugation site and free polymer, along with the mechanism of biotin release. EXPERIMENTAL PROCEDURES
Materials. 4-Hydroxyazobenzene (Aldrich, Milwaukee, WI) was recrystallized from ethanol and water, and dried in vacuo. N,N-Dimethyl acrylamide (DMA) (Fluka, Milwaukee, WI), diethyl ether, acryloyl chloride, triethylamine, and 2-mercaptoethanol (MEO) (Aldrich, Milwaukee, WI) were purified by distillation under reduced pressure. 2,2′-azoisobutyronitrile (AIBN) (J. T. Baker, Phillipsburg, NJ) was recrystallized from methanol. Ethanol, dimethylformamide (DMF), tetrahydrofuran (THF), methylene chloride, potassium tert-butoxide, divinyl sulfone (DVS), 5,5′-thiobis (2-nitrobenzoic acid) (DTNB), ethylenediamine tetraacetic acid (EDTA) (Aldrich, Milwaukee, WI), tris(2-carboxy ethyl)-phosphine hydrochloride (TCEP-HCl) (Pierce, Rockford, IL), iminobiotin beads, cysteine hydrochloride (Pierce, Rockford, IL), Microcon-30, Centriprep-10 (Millipore, Bedford, MA), and [3H]biotin (d-[8,9-3H(N)]biotin, Dupont, Wilmington, DE) were used as received. All other reagents were of
10.1021/bc000107b CCC: $20.00 © 2001 American Chemical Society Published on Web 02/20/2001
Smart Polymer-Streptavidin
Bioconjugate Chem., Vol. 12, No. 2, 2001 315
Figure 1. Synthetic scheme for DMAA and DMAA-VS.
analytical grade. Restriction enzymes (MscI, MluI, and HindIII) and T4 DNA Ligase were purchased from New England Biolabs (Beverly, MA). Qiaex II kit for extracting DNA was purchased from Qiagen (Santa Clarita, CA). pET21a vector was obtained from Novagen (Madison, WI). Azobenzene Monomer Synthesis. We synthesized an azobenzene-containing monomer, 4-phenylazophenylacrylate (AZAA). 4-Hydroxyazobenzene (0.2 mol) and triethylamine (0.26 mol) were dissolved in diethyl ether (200 mL). Acryloyl chloride (0.24 mol) dissolved in 80 mL of diethyl ether was added dropwise at 0 °C under nitrogen atmosphere with stirring. The reaction mixture was allowed to come to room temperature and stirred for 4 h. The reaction solution was filtered to remove triethylammonium chloride, washed with water and evaporated. The product was recrystallized from ethanol-water mixture and dried in vacuo. The structures were confirmed by 1H NMR (d-DMSO). AZAAm: δ 5.7-6.7 (m, 3H), 7.4-7.8 (m, 5H), 7.8-8.2 (m, 4H). AZAA: δ 6.16.7 (m, 3H), 7.3-7.7 (m, 5H), 7.8-8.2 (m, 4H). Temperature Responsive Polymer Preparation. We synthesized a temperature responsive polymer with hydroxyl termini, DMAA (DMA-co-AZAA copolymer), by free-radical polymerization in DMF at 60 °C for 20 h, using MEO as a chain transfer regent and AIBN as an initiator (monomer concentration ) 2 mol/L, monomer/ MEO/AIBN ) 100/1/0.2 molar ratio) (Figure 1). The feed ratios of AZAA was 4.0 mol %. The product was purified by precipitation into diethyl ether three times and dried in vacuo. The content of AZAA incorporated in the copolymers was determined by 1H NMR (Spectrospin and Bruker, dpx200), comparing the ratio of aromatic and aliphatic hydrogens. The molecular weight (MW) of the polymer was determined by gel permeation chromatography
(GPC, Water, Styragel HR3 and 4) in THF, using polystyrene as standards. The hydroxyl terminus of DMAA was converted to a vinyl sulfone (VS) group for conjugation to the protein SH sites (Figure 1). DMAA was dissolved in 20 mL of methylenechloride with 0.03 g of potassium tert-butoxide, and 100 µL of divinyl sulfone (DVS) (DVS/OH ) 10/1 molar ratio). The solution was stirred for 12 h at room temperature under nitrogen atmosphere. The polymer was precipitated and then washed in diethyl ether and dried in vacuo. The amount of active VS in VS-terminated DMAA (DMAA-VS) was determined by the reaction with cysteine. Since cysteine was detected colorimetrically by the reaction with 5,5′-thiobis (2-nitrobenzoic acid) (DTNB), the amount of VS was calculated by the consumed cysteine in the solution. Briefly, 0.2 mL of polymer solution (10 mg/mL in 0.1 M PB, pH 8.0) was mixed with 2 mL of cysteine solution (1.5 mM cysteine hydrochloride in 0.1 M PB, pH 8.0, with 10 mM EDTA), and reacted for 4 h. A total of 0.2 mL of the solution was mixed with DTNB solution (0.2 mg of DTNB in 2 mL of 0.1 M PB, pH 8.0) and incubated for 15 min. Then, the absorbance at 412 nm was measured to calculate the amount of cysteine reacted with DTNB. The LCST of the polymer was determined as the temperature at 10% of the maximum absorbance at 600 nm. The polymer concentration was 2 mg/mL in doubledistilled water or 100 mM phosphate buffer (PB) pH 7.4. Genetic Engineering of Streptavidin. The streptavidin mutant E116C and S139C were constructed by sitedirected, cassette mutagenesis, using a synthetic “core” streptavidin gene previously designed and constructed for protein expression in Escherichia coli (5). The construction and characterization of the E116C (5) and S139A (6) have been previously described. Analysis of the
316 Bioconjugate Chem., Vol. 12, No. 2, 2001
streptavidin mutants to confirm the correct production included SDS-PAGE, mass spectrometry, and biotin binding studies. Conjugation of Polymer to Protein. Conjugation between streptavidin (E116C, S139C) and the polymer (DMAA-VS) was carried out in 100 mM sodium tetraborate buffer, pH 8.0, containing 50 mM sodium chloride and 5 mM EDTA. Cysteine residues in E116C or S139C form disulfide bonds during storage. To reduce disulfide bonds to sulfhydryl groups, TCEP at 50-fold molar excess to the mutants was added into the solution, which was then rotated at room temperature for 20 min. Then, DMAA-VS was added in 50-fold molar excess to the mutants to obtain high conjugation efficiency, and reacted for 1 h at room temperature, followed by the reaction for 24 h at 4 °C. The conjugates were separated from the unconjugated mutants by thermally induced precipitation, i.e., centrifuging the mixture at 15 000 rpm (31000g) for 15 min at 50 °C with 4 wt % of ammonium sulfate to enhance the precipitation. The unconjugated mutants were retained in the supernatant. The thermal-induced precipitation was repeated three times. Further purification was carried out by using an iminobiotin column to investigate the biotin-binding properties of the isolated conjugates from free polymers. Iminobiotin has been used for purifying streptavidin and avidin widely owing to its reversible binding to these proteins at different pHs. The conjugates were bound to iminobiotin on the beads in the binding buffer (50 mM sodium carbonate, 500 mM sodium chloride, pH 11) at room temperature, and the free polymers were washed off by the same buffer. Then, the bound conjugates were eluted in the elution buffer (100 mM acetic acid, pH 2.6) at room temperature. The purified conjugates were condensed by a Centriprep-10 (MW cut off: 10 000) to 30-300 µg/mL. Biotin-Binding Assay. [3H]Biotin was employed to quantitate biotin binding. A total of 10 µg of the conjugate was suspended in 0.9 mL of 100 mM of sodium phosphate buffer, pH 7.4, containing 50 mM of sodium chloride and 5 mM of EDTA. A mixture of [3H]biotin and unlabeled biotin (1% of [3H]biotin) was added to the conjugate solution (total volume was 1 mL) after incubation at 52 °C for 1 h. The ratio of total biotin/biotin binding sites was 1.5 (molar ratio). The solution was incubated at a certain temperature, and a 100 µL aliquot was taken at the end of each incubation. The unbound biotin was separated from the bound biotin by ultrafiltration of the aliquot using a Microcon-30 (MW cutoff: 30 000) at the incubation temperature of the solution. A 30 µL aliquot of the filtrate that contained unbound biotin was counted in a β-counter (LS7000 Liquid scintillation system, Beckman Instruments, Inc. Fullerton, CA), to determine the unbound biotin. The biotin binding could be quantitated by the decrease of free biotin in the solution. RESULTS AND DISCUSSION
Polymer Characterization and Protein Conjugation. The DMAA copolymer terminated with a hydroxyl group was synthesized by free radical copolymerization of DMA and AZAA in DMF using MEO as a chain transfer regent (Figure 1). GPC and 1H NMR analyses showed that the DMAA had a weight average molecular weight (MW) of 18 kDa and an AZAA content of 5.9 mol %. The LCST of DMAA was measured by the absorbance change at 600 nm and determined to be 37 °C. The hydroxyl terminus of DMAA was reacted with DVS to
Shimoboji et al.
Figure 2. Three-dimensional model of the streptavidin displaying the relative locations of position 116 and the C-terminus. The X-ray crystal structure is defined to the highlighted position 134, with the 139 position three residues away. The biotinbinding site and Trp 120 are also shown for orientation.
make DMAA-VS (Figure 1). DMAA-VS contained 97.5 molar percentage of reactive VS as determined with the colorimetric DTNB assay. The polymer was conjugated to SA site specifically by the reaction of the vinyl sulfone group of the polymer with the sulfhydryl groups of the protein. The conjugate was separated from the unreacted SA by thermally induced precipitation of the polymer. The polymer-protein conjugates could be subsequently purified from free polymer by affinity chromatography. The number of polymer molecules conjugated to a SA was calculated from spectral analysis of the protein and polymer absorption spectra as 2.03 for the E116C-DMAA conjugate and 2.11 for the S139C-DMAA conjugate. This suggests that one polymer chain is conjugated to each of the two streptavidin binding faces, with steric repulsion likely preventing the conjugation of a second polymer at each face. Conjugation Site Dependence of Activity Switch. Each subunit of E116C has a unique cysteine residue, which replaces the glutamic acid at position 116 (Figure 2). This conjugation site was previously chosen to be as close as possible to Trp120, one of the energetically most important biotin binding contacts (7). To test the sitedependence of conjugation for the activity switch, we have characterized the S139C mutant, which has a unique cysteine, engineered at the C-terminus of streptavidin (Figure 2). The streptavidin structure is crystallographically determined to residue 134, where the distance to the biotin binding site is approximately 2.4 nm. The relative change in biotin binding capacity for the E116C conjugate is shown in Figure 3. The thermal response for the purified S139C conjugate was also determined, and compared both to protein alone and physical mixtures of SA and hydroxyl terminated polymers as controls. The E116C conjugate displayed a striking thermally dependent biotin binding capacity that translated into biotin blocking and releasing activities. Moving the conjugation site away from the biotin binding site had a large effect on the switching ability, as the S139C conjugate did not display a thermal dependence above the controls of protein alone and protein/polymer physical mixtures. These results demonstrate that the switching activity is strongly dependent on the location
Smart Polymer-Streptavidin
Bioconjugate Chem., Vol. 12, No. 2, 2001 317
Figure 3. The effect of thermal cycling on the biotin-binding capacity for the E116C-DMAA conjugate, the S139C-DMAA conjugate, physical mixtures of E116C or S139C and DMAA, and E116C and S139C alone in 100 mM sodium phosphate buffer at pH 7.4. The biotin-binding capacity was determined assuming 100% binding at 4 °C after 1 h incubation. For the conjugates, the biotin-binding capacity was plotted as the percentage of conjugated sites.
of the conjugation site, consistent with the need for the polymer to be near the binding site. Mechanism of Biotin Releasing Activity. To investigate whether the off-rate of biotin was increased when the polymer was collapsed, the time dependence of biotin release was characterized (Figure 4). The unconjugated E116C exhibits the same biotin off-rate as wild-type streptavidin, and the increase in off-rate at 52 °C for unconjugated E116C is not large enough to significantly reduce the amount of bound biotin at equilibrium. This can be observed in Figure 4B, where there is no significant change in the percentage of bound biotin for the control E116C as the temperature is shifted from 4 to 52 °C. When biotin was first added to the DMAA-E116C conjugate at 52 °C, binding was largely blocked by the collapsed polymer. The temperature was then cycled to 4 °C where the polymer was hydrated, and this opened the blocked binding sites. After recycling back to 52 °C, the time course of biotin release was quantitated. From these data, the first-order dissociation rate constant for biotin was determined to be 1.6 × 10-4 s-1 for the E116C-DMAA conjugate. The off-rate is thus very close to the 3.0 × 10-4 s-1 constant of wild-type SA at 52 °C. This finding suggests that the collapse of DMAA is not inducing an intrinsic increase in the biotin off-rate, but is blocking the reassociation of biotin. It is interesting that the biotin can exit through the collapsed polymer from the binding pocket, but not re-enter from bulk solution. Role of Free Polymer in Switching Activity. Iminobiotin affinity chromatography was used to purify the E116C-DMAA and S139C-DMAA streptavidin conjugates. This purification allowed the study of the role of free polymer in the blocking and release switching activity. The thermal cycling response for binding and release of purified conjugates is compared in Figure 5 to conjugates where free polymer was also present. Although the purified conjugates showed reversible thermal switching for biotin binding, they exhibited smaller responses than when free polymer was added. The responses increased in magnitude as more of the free polymer was added.
Figure 4. Kinetics of biotin release after cycling to 52 °C from 4 °C. (A) E116C-DMAA conjugate and (B) E116C control. Assays were performed in 100 mM sodium phosphate buffer, pH 7.4. The biotin-binding capacity is determined assuming 100% binding at 4 °C after 1 h incubation.
318 Bioconjugate Chem., Vol. 12, No. 2, 2001
Shimoboji et al.
Figure 5. Effect of adding free polymer on the biotin-binding capacity at 52 °C for E116C-DMAA and S139C-DMAA in 100 mM sodium phosphate buffer, pH 7.4. The unpurified samples were those isolated by thermal precipitation from the conjugation reaction. The biotin-binding capacity was determined assuming 100% binding at 4 °C after 1 h incubation. For the conjugates, the biotinbinding capacity was plotted as the percentage of conjugated sites. The molar ratios of free polymer/conjugate in the unpurified states were calculated as 1430 for E116C-DMAA and 1150 for S139C-DMAA conjugates by use of the conjugation conversions.
This effect was clarified by characterizing the S139CDMAA streptavidin conjugate. The purified S139CDMAA conjugate did not display any switching capability, and addition of free polymer led to only a small increase in activity. Taken together with the E116CDMAA results, this suggests that the free polymer is likely aggregating with the conjugated polymer in the collapsed state. When directed to a site near the binding pocket, as with the E116C mutant, this DMAA aggregate more effectively blocks biotin association. But the aggregation effect is insignificant if the free polymer is directed toward a conjugation site that is too distant from the binding site. These findings confirm that the MW and type of smart polymer will likely provide important avenues to engineer the molecular switch activity. For example, larger and/or grafted polymers could provide better barriers near the binding site.
physical aggregation of the free polymer at the defined position near the biotin-binding site. Taken together, these studies elucidate the importance of conjugation site and suggest that the molecular weight and type of stimuli-responsive polymer will be essential to optimize the switching activity. ACKNOWLEDGMENT
We would like to thank Dr. J. Milton Harris of Shearwater Polymers for his advice on the addition of the polymer end hydroxyl groups to vinyl sulfone groups of DVS. We also wish to thank C. Lackey, C. J. Long, and Dr. R. To for the construction and expression of the SA mutants. The NIH (Grant 53771), UW Office of Technology Transfer, the Washington Research Foundation, and the Washington Technology Center are gratefully acknowledged for their support of this project.
CONCLUSIONS
We have utilized a new thermally responsive polymer to investigate the mechanism of the smart polymer activity switches. The use of the DMAA polymer allowed an accurate determination of the conjugation ratio. The site-selectivity of the switch was first investigated with the S139C streptavidin mutant that contains a unique cysteine at the C-terminus. The purified S139C conjugate, where the DMAA is distant from the binding site, did not display switching activity. The previously characterized E116C streptavidin mutant did display switching activity when the DMAA was conjugated at this cysteine near the biotin-binding site. Because the release of biotin was shown to proceed with the same kinetics as the biotin off-rate for wild-type streptavidin, the mechanism for release was shown to be the blocking of biotin reassociation by the collapsed DMAA. The magnitude of the switching activity could be significantly increased by the addition of free polymer, but only if the conjugated polymer was near the binding site. This suggests that the conjugated polymer is directing a
LITERATURE CITED (1) Weber, P. C., Ohlendore, D. H., Wendoloski, J. J., and Salemme, F. R. (1989) Structural origins of high-affinity biotin binding to streptavidin. Science 243, 85-88. (2) Wilchek, M., and Bayer, E. A. (1990) Avidin-biotin technology. Methods in Enzymology, Vol. 184, pp 746-784, Academic Press, San Diego. (3) Bulmus, E. V., Ding, Z., Long, C. J., Stayton, P. S., and Hoffman, A. S. (2000) A pH- and temperature-sensitive copolymer-streptavidin site-specific conjugate for pH-controlled binding and triggered release of biotin. Bioconjugate Chem. 11, 78-83. (4) Stayton, P. S., Shimoboji, T., Long, C., Chilkoti, A., Chen, G., Harris, J. M., and Hoffman, A. S. (1995) Control of protein-ligand recognition using a stimuli-responsive polymer. Nature 378, 472-474. (5) Ding, Z. L., Long, C., Hayashi, Y., Bulmus, E. V., Hoffman, A. S., and Stayton, P. S. (1999) Temperature control of biotin binding and release with a streptavidin-poly(N-isopropylacrylamide) site-specific conjugate. Bioconjugate Chem. 10, 395-400.
Smart Polymer-Streptavidin (6) Wilbur, D. S., Stayton, P. S., To, R., Buhler, K. R., Klunb, L. A., Hamlin, D. K., Stray, J. E., and Vesselle, R. L. (1998) Streptavidin in antibody pretargetting. Comparison of a recombinant streptavidin with two streptavidin mutant proteins and commercially available streptavidin proteins. Bioconjugate Chem. 9, 100-107.
Bioconjugate Chem., Vol. 12, No. 2, 2001 319 (7) Chilkoti, A., Tan, P. H., and Stayton, P. S. (1995) Sitedirected mutagenesis studies of the high-affinity streptavidin-biotin complex: contributions of tryptophan residues 79, 108, and 120. Proc. Natl. Acad. Sci. U.S.A. 92, 1754-1758.
BC000107B
