AC Research
Anal. Chem. 1998, 70, 2019-2024
Accelerated Articles
Detection of Conformational Changes in an Immobilized Protein Using Surface Plasmon Resonance Hiroyuki Sota* and Yukio Hasegawa
Department of Research and Development, Amersham Pharmacia Biotech K.K., 3-25-1, Hyakunincho, Shinjuku-ku, Tokyo 169-0073, Japan Masahiro Iwakura
National Institute of Bioscience and Human Technology, 1-1 Higashi, Tsukuba, Ibaraki 305-0046, Japan
Utilizing surface plasmon resonance (SPR), we have developed novel methodology for the detection of conformational change(s) in immobilized proteins. A genetically altered E. coli dihydrofolate reductase (DHFR-ASC) was attached to a carboxymethyldextran matrix layer covering the sensor surface of an SPR biosensor through a disulfide linkage at the engineered protein’s C-terminus. The DHFR-ASC-immobilized surface exhibited a larger response to acid treatment than reference surfaces lacking immobilized proteins. The SPR signal of the tethered protein and the molar ellipticity of DHFR-ASC in solution responded similarly to pH changes, consistent with the interpretation that changes in the SPR signal reflect conformational changes occurring during acid denaturation. A pH shift observed between the SPR signal and ellipticity changes may reflect a difference between surface and bulk pH. The tethered protein sensor surface was stable to repeated acid treatment using solutions in the pH range of 0.12-7.80 and yielded reproducible measurements. This is the first demonstration of detection of conformational changes in an immobilized protein using an SPR biosensor. This technique has potential for developing novel sensors and/or switching devices in response to protein conformational changes.
Recent sophisticated methodologies for analyzing protein structure, e.g., circular dichroism,1 small-angle X-ray scattering,2 * To whom correspondence should be addressed: (fax) +81-3-5331-9361; (email)
[email protected]. S0003-2700(97)01366-8 CCC: $15.00 Published on Web 04/17/1998
© 1998 American Chemical Society
and fluorescence energy transfer,3 have enabled the detailed characterization of protein folding and unfolding, a process increasingly relevant as the fields of protein engineering and proteome research expand.4 However, most current methodologies were developed for measuring the conformational behavior of proteins dispersed uniformly in solution and are unsuitable for the detailed characterization of proteins distributed asymmetrically, such as immobilized proteins. Nevertheless, the folding behavior of such proteins is important biologically (e.g., membraneassociated proteins) as well as in engineered systems utilizing immobilized proteins. Surface plasmon resonance (SPR) is an optical phenomenon5 that has been exploited for monitoring the behavior of immobilized proteins, although heretofore only with respect to their binding of solute molecules. SPR is a function of the effective refractive index integrated along an electromagnetic field that decays with distance from a thin metal layer (evanescent wave), and even minute changes can be detected quantitatively with a suitable apparatus.6-8 The apparent refractive index is determined in turn by the mass and dielectric properties (polarity magnitude and distribution with respect to the metal surface) of the substances (1) Evans, P. A.; Radford, S. E. Curr. Opin. Struct. Biol. 1994, 4, 100-106. (2) Eliezer, D.; Jennings P. A.; Wright, P. E.; Doniach, S.; Hodgson, K. O.; Tsuruta, H. Science 1995, 270, 487-488. (3) Rischel, C.; Poulsen F. M. FEBS Lett. 1995, 374, 105-109. (4) Dill, K. A.; Fersht, A. R. Curr. Opin. Struct. Biol. 1996, 6, 1-2. (5) Kretschmann, E.; Raether, H. Z. Naturforsch. 1968, 23, 2135-2136. (6) Nylander, C.; Liedberg, B.; Lind, T. Sens. Actuators 1982, 3, 79-88. (7) Terrettaz, S.; Stora, T.; Duschl, C.; Vogel, H. Langmuir 1993, 9, 13611369. (8) Kooyman, R. P. H.; Kolkman, H.; Van Gent, J.; Greve, J. Anal. Chim. Acta 1988, 213, 35-45.
Analytical Chemistry, Vol. 70, No. 10, May 15, 1998 2019
present. With commercially available devices, the SPR signal can be monitored in real time, enabling direct inference of kinetic parameters as well as the equilibrium parameters conventionally measured in studies of biospecific interactions. To date, biochemical studies employing SPR have measured ligand binding, i.e., changes primarily in mass, on the sensor surface.9-11 However, SPR is also capable of detecting changes in the dielectric properties of substances on or near the sensor surface.12-14 Since protein folding states will affect their dielectric properties, it should be feasible to utilize SPR for probing changes in the folding state per se. In turn, the ability to monitor such conformational changes in immobilized proteins could enable the development of novel sensors and/or switching devices. To demonstrate SPR’s potential in this regard, acid denaturation of matrix-bound Escherichia coli dihydrofolate reductase (DHFR, EC 1.5.1.3) was analyzed using a Biacore apparatus. This is the first real-time analysis of the conformational behavior of an immobilized protein. EXPERIMENTAL SECTION Construction and Preparation of Engineered Protein. A genetically constructed cysteine-free mutant DHFR (DHFR-AS)15 was previously engineered to have an extra cysteine residue attached to its C-terminal end (DHFR-ASC).16 Another two derivatives of mutant DHFR-ASC were constructed by cassette mutagenesis according to a previous report.17 These mutants, which are designated DHFR-ASG2C and DHFR-ASG4C, have two and four extra glycine residues added penultimately to the C-terminus, respectively (Figure 1). The proteins produced by the transformants were purified to a homogeneous state, and their terminal thiol groups were fully reduced prior to the immobilization as previously described.18 Apparatus and Reagents. Biacore and sensor chip CM5, which is an exchangeable sensor surface consisting of a glass support covered by a thin gold layer, where the carboxymethyldextran matrix is covalently bound, were from Biacore AB. 4-(2Hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) was purchased from Dojindo Laboratories (Kumamoto, Japan). NHydroxysuccinimide (NHS), N-ethyl-N′-[3-(dimethylamino)propyl]carbodiimide (EDC), 2-(2-pyridinyldithio)ethaneamine (PDEA), ethanolamine hydrochloride, and surfactant P20 were obtained from Biacore AB and were prepared prior to use. Site-Specific Immobilization of Mutant Protein onto the Sensor Surface. Site-specific immobilization of mutant DHFRs was achieved by automatically injecting the reagents into a flow cell constructed on the sensor surface covered with a carboxy(9) Granzow, R.; Reed, R. Bio/Technology 1992, 10, 390-393. (10) Szabo, A.; Stolz, L.; Granzow, R. Curr. Opin. Struct. Biol. 1995, 5, 699705. (11) Shinohara, Y.; Sota, H.; Gotoh, M.; Hasebe, M.; Tosu, M.; Nakao, J.; Hasegawa, Y.; Shiga, M. Anal. Chem. 1996, 68, 2573-2579. (12) Liedberg, B.; Lundstro ¨m, I.; Stenberg, E. Sens. Actuators B 1993, 11, 6372. (13) Salamon, Z.; Wang, Y.; Brown, M. F.; Macleod, H. A.; Tollin, G. Biochemistry 1994, 33, 13706-13711. (14) Chao, N.-M.; Chu, K. C.; Shen, Y. R. Mol. Cryst. Liq. Cryst. 1981, 67, 261275. (15) Iwakura, M.; Jones, B. E.; Luo, J.; Matthews, C. R. J. Biochem. (Tokyo) 1995, 117, 480-488. (16) Vigmond, S. J.; Iwakura, M.; Mizutani, F.; Katsura, T. Langmuir 1994, 10, 2860-2862. (17) Iwakura, M.; Honda, S. J. Biochem. (Tokyo) 1996, 118, 414-420. (18) Iwakura, M.; Kokubu, T. J. Biochem. (Tokyo) 1993, 114, 339-343.
2020 Analytical Chemistry, Vol. 70, No. 10, May 15, 1998
Figure 1. Schematic drawing of the engineered DHFRs used in this study. Replacement of the wild-type residues Cys85 and Cys152 with Ala and Ser, respectively was confirmed to have no significant effect on protein stability. Each mutant DHFR has an extra cysteine residue at its C-terminal end. This figure also represents the manner of the tethering of the designed proteins onto the matrix on the sensor surface.
methyldextran (CM-dex) matrix layer of ∼100-nm thickness,19 according to the thiol-coupling method using an NHS/EDC mixture (activation time, 2 min) and a reactive bifunctional reagent, PDEA.20 Between injections, the sensor surface was continuously washed with 10 mM HEPES, pH 7.4, containing 150 mM NaCl and 0.005% surfactant P20. As references, an L-cysteineimmobilized sensor surface (Cys surface) and an ethanolamineimmobilized sensor surface (Etam surface) were constructed by injection of 50 mM L-cysteine at pH 4.0 instead of the protein and by injection of 1.0 M ethanolamine hydrochloride at pH 8.5 instead of PDEA, respectively. The amount of immobilized molecules was determined from the final increment in the resonance signal.21 The resonance signal obtained from Biacore is expressed in an arbitrary resonance unit (RU). A response of 1 RU corresponds to a ∼10-4 deg shift in the angle of minimum reflected light intensity and to a surface protein concentration of ∼1 pg/mm2. Acid Denaturation. Acid solutions were prepared by titration of 10 mM potassium phosphate (pH 7.8) with concentrated HCl solution to the desired pH. Acid treatment was started by pulse injection of the acid solution (50 µL, 10 µL/min) onto the sensor surface in the flow cell, which was held at 25.0 °C and was equilibrated with running buffer, 10 mM potassium phosphate at pH 7.8, prior to the injection (10 µL/min). The injection was terminated by replacement with running buffer. The value of the equilibrium resonance signal of the acid treatment was obtained (19) Lo ¨fa˚s, S.; Johnsson, B. J. Chem. Soc. Chem. Commun. 1990, 1526-1528. (20) Lo ¨fa˚s, S.; Johnsson, B.; Edstro¨m, A° .; Hansson, A.; Lindquist, G.; Hillgren, R.-M. M.; Stigh, L. Biosens. Bioelectron. 1995, 10, 813-822. (21) Stenberg, E.; Persson, B.; Roos, H.; Urbaniczky, C. J. Colloid Interface Sci. 1991, 143, 513-526.
Figure 2. Typical SPR time course monitoring the site-specific immobilization of DHFR-ASC onto a carboxymethyldextran matrix, subsequent repeated exposure to protein denaturants, and, finally, protein detachment. Reagent injections are denoted numerically as follows: (1) 100 mM NHS/400 mM EDC; (2) 80 mM PDEA at pH 8.5; (3) 10 µg/mL DHFR-ASC at pH 4.0; (4) 50 mM L-cysteine/1.0 M NaCl at pH 4.0; (5) 8.0 M urea at pH 4.0; (6) 6.0 M Gdn‚HCl at pH 4.0; (7) 100 mM DTT/6.0 M Gdn‚HCl at pH 8.5. The thick bars indicate the duration of injections. The amount of DHFR-ASC immobilized in this experiment resulted in a 3234.1 RU increase above baseline. During reagent injections 5-7, the resonance signal exceeded the range of the instrument.
by averaging the signals in a time range from 450 to 475 s using the software BIAevaluation (Biacore AB). This software was also used for nonlinear least-squares fitting of the kinetic data to obtain the apparent relaxation times.15 Circular Dichroism Measurement. The circular dichroism (CD) spectra of the mutant DHFRs were measured using an Aviv model 62A DS spectrometer (Aviv Associates, Inc., Lakewood, NJ) at 25.0 °C. RESULTS AND DISCUSSION Immobilization of Mutant DHFR onto the Sensor Surface. Taking advantage of its unique C-terminal cysteine residue, DHFRASC was immobilized onto the CM-dex layer with a resulting increase of a few thousand resonance units. This change in resonance corresponds to ∼2-3 ng of attached protein/mm2. A typical resonance signal time course for DHFR-ASC immobilization is shown in Figure 2. After protein attachment, the DHFR-ASCimmobilized surface (DHFR surface) was exposed to repeated pulses of 6.0 M guanidine hydrochloride (Gdn‚HCl) at pH 4.0. Following each pulse, the resonance signal recovered to its preexposure level. Moreover, treatment with 100 mM dithiothreitol (DTT) in the presence of 6.0 M Gdn‚HCl, pH 8.5, caused the resonance signal to revert to a value very near that of the sensor surface before protein attachment (Figure 2). These results are consistent with the intended immobilization of DHFRASC through a site-specific disulfide linkage. Retention of enzymatic activity has been confirmed in DHFR similarly attached to a biopolymer resin.18 Furthermore, reversible denaturation of immobilized DHFR through temperature, urea, Gdn‚HCl, and acid treatment has been demonstrated by enzymatic assay (Iwakura, unpublished results). While the Gdn‚HCl pulses described above could be expected to denature the protein, the resonance response to Gdn‚HCl itself was off-scale so that any contribution by protein
Figure 3. Typical resonance signal time courses of the DHFR (1), Cys (2), Etam (3), and CM (4) surfaces in response to injections of acidic solutions at pH 1.01 (A), pH 3.14 (B), and pH 5.81 (C). The amount of immobilized protein on the DHFR surface corresponded to 1857.2 RU. The thick bar indicates the duration of injection.
denaturation to the SPR signal could not be assessed. Likewise, thermal denaturation studies were beyond the scope of the Biacore instrument. Reliable measurements at any given temperature require long equilibration periods, making accurate monitoring of rapid thermal transitions impossible. Furthermore, the maximum temperature setting is 37 °C, well below the 50 °C and higher midpoints for thermal unfolding of wild-type DHFR22 and its Cysfree derivatives.23 In view of these limitations we confined subsequent experiments in this study to acid-induced denaturation. Acid Denaturation of Immobilized DHFR-ASC. Figure 3 shows typical resonance signal time courses for the DHFR surface in response to acidic pulses. The DHFR surface responded immediately to the injection of acidic solution, but the resonance signal continued to climb slightly for a few minutes until reaching an equilibrium value. This value was significantly higher than those obtained for the reference surfaces, i.e., an L-cysteineimmobilized sensor surface (Cys surface), an ethanolamineimmobilized sensor surface (Etam surface), and the original CMdex-covered surface (CM surface) as shown in Figure 3 and summarized in Figure 4. This difference in response increased in proportion to the amount of DHFR-ASC immobilized (data not shown). Although the three reference surfaces had different electrostatic properties, their responses to altered pH were substantially similar, increasing with decreasing pH in correspondence to the HCl concentration of the injected solution. This result indicates that the resonance signal responses of these surfaces are attributable to the bulk effects of these solutions on the refractive index rather than to their effects on surface charge. (22) Ohmae, E.; Kurumiya, T.; Makino, S.; Gekko, K. J. Biochem. (Tokyo) 1996, 120, 946-953. (23) Luo, J.; Iwakura, M.; Matthews, C. R. Biochemistry 1995, 34, 10669-10675.
Analytical Chemistry, Vol. 70, No. 10, May 15, 1998
2021
Figure 4. (A) pH dependence of the equilibrium resonance signals of the DHFR and reference surfaces. The DHFR surface is the same as in Figure 3. (B) pH dependence of the corrected resonance signal, calculated by subtracting the equilibrium resonance signal of the Etam surface from that of the DHFR surface. (C) pH dependence of the molar ellipticity (222 nm) of DHFR-ASC in solution.
We thus discount charge effects and conclude that the increment for the DHFR surface reflects changes specific to the resonance contribution of DHFR-ASC itself. Following the acidic pulse, the resonance signal of the reference surfaces returned instantaneously to the preinjection level. In contrast, the signal from the DHFR surface returned to its initial level more gradually (Figure 3). We analyzed the kinetics of these delay curves following pH pulses in the range of 0.12-3.14 (Figure 5). A nonlinear three-exponential least-squares fit was carried out for the delay curve from t ) 490 s through t ) 2000 s; the initial 8 s was omitted from the analysis due to undesired buffer mixing at the pulse interface, complicating curve interpretation. The fitting yielded apparent relaxation times for the three phases, t1 (212 ( 20 s; mean ( SD), t2 (44 ( 6.7 s) and t3 (7.7 ( 0.8 s). These values were comparable to those for the three slowest phases (∼110, ∼40, and ∼1.5 s, respectively) of refolding for the cysteine-free mutant DHFR-AS free in solution.15 We interpret this delay to reflect the refolding process of the tethered DHFR-ASC molecules. Extension of the fitted curve back to the end of the acidic pulse makes it clear that components of the least-squares fit are missing; the theoretical curves for both pH levels achieve only ∼70% of the measured values. This result 2022 Analytical Chemistry, Vol. 70, No. 10, May 15, 1998
Figure 5. Typical corrected resonance signal responses of the DHFR surface to acid solutions at pH 1.01 (A) and pH 3.14 (B). The thick bar indicates the injection duration for the acidic solution, terminated by replacement with running buffer at t ) 482 s. A threeexponential least-squares fit was carried out for the delay curve from t ) 490 s through t ) 2000 s; the initial 8 s was omitted from this analysis due to undesired buffer mixing at the pulse interface, complicating curve interpretation. Curve fitting yielded t1, t2, and t3 values of 199, 37, and 7.5 s, respectively, for the pH 1.01 pulse, and 191, 34, and 6.2 s for the pH 3.14 pulse. The fitted and measured curves are shown together in the expanded insets with the arrow in each indicating the switch from acid solution to running buffer. Residual plots of the data used in fitting are shown at the top of each panel to indicate noise levels.
is also consistent with the report cited above which suggested two additional faster phases for refolding.15 pH Dependence of the Equilibrium Resonance Signal. The equilibrium resonance signal of the DHFR surface was corrected by subtracting the signal of the Etam surface. The pH dependence of this value and the pH-dependent ellipticity (222 nm) of DHFR-ASC free in solution were determined (Figure 4B and C, respectively). The pH-dependent ellipticity curve was typical of a type IA protein as recently described by Fink et al.,24 proceeding from a native state to an acid-unfolded form (“UA state”) and finally to a compact molten globule-like conformation (“A state”) with decreasing pH. A similar pH-dependent ellipticity was also observed by Ohmae et al.22 for wild-type DHFR. As shown in Figure 4B, SPR measurements of the tethered DHFRASC also showed an apparent three-state transition, although the pH values of the transition points are shifted toward milder pH. At face value, this shift suggested some strain predisposing the protein to denaturation. Attempting to ease any structural strain that might have arisen through the tight association of the protein with the polymer matrix, we tested proteins with a lengthened tether in the form of glycine residues added penultimately to the C-terminus. Oligomeric glycine is known to exhibit a flexible conformation, and four glycine residues aligned linearly have a (24) Fink, A. L.; Calciano, L. J.; Goto, Y.; Kurotsu, T.; Palleros, D. R. Biochemistry 1994, 33, 12504-12511.
Figure 6. Comparison of pH responses of three engineered DHFR proteins to gauge the impact of tether length on protein stability. Three DHFR proteins, differing in the number of glycine residues added penultimately to the C-terminus, were subjected to a range of pH, and the behavior of the tethered or free molecules was monitored by SPR (A) or molar ellipticity (B), respectively. The resonance signal (A) was corrected as in Figures 4B and 5 and then normalized for the amount of immobilized protein (1822.1, 1976.4, and 1741.8 RU for DHFR-ASC, DHFR-ASG2C, and DHFR-ASG4C, respectively).
length (∼15 Å) nearly equivalent to the radius of gyration of native DHFR (∼18 Å; Iwakura et al., unpublished result). Therefore, we tested two derivatives of DHFR-ASC, with either two or four added glycine residues. Figure 6 shows that DHFR-ASG2C and DHFR-ASG4C yielded pH-dependent resonance curves indistinguishable from DHFR-ASC, suggesting that such steric effects between the protein and the polymer matrix are irrelevant. No other transition phases were observed for either the free DHFR by far-UV CD measurements or for the immobilized DHFR by SPR measurements. The pH shift in transition points for the free vs immobilized DHFR appears to be due to immobilization itself. With respect to how immobilization might have this effect, the following two factors may be considered: (1) structural strains resulting from the very tight tethering of the protein to the matrix and (2) a special environment created in the attachment matrix that might differ from that of the bulk solution. Since increasing the tether length had no discernible effect, we rule out the first possibility. The second factor, a special environment, remains. The actual pH of an unstirred layer occurring on a negatively charged surface has been shown to be lower than the apparent bulk pH,25,26 and this effect is consistent with the pH shift we observed for the denaturation transition points inferred from the SPR signal. Johnsson et al.27 reported 30-40% of the carboxyl (25) Szundi, I.; Stoeckenius, W. Biophys. J. 1989, 56, 369-383. (26) Schasfoort, R. B. M.; Bergveld, P.; Kooyman, R. P. H.; Greve, J. Anal. Chim. Acta 1990, 238, 323-329. (27) Johnsson, B.; Lo ¨fa˚s, S.; Lindquist, G. Anal. Biochem. 1991, 198, 268-277.
groups in the CM-dex layer to be activated with a 7-min pulse of the NHS/EDC mixture. Under the conditions we employed, the matrix should have been less activated and, therefore, could be expected to carry an even greater negative charge. In turn, this charge is expected to have resulted in a lower pH in the immediate vicinity of the matrix. Do the changes in the DHFR-attributable signal as the pH is lowered merely reflect changes in the net charge of the protein? Such an explanation is not consistent with the pattern of change we observed. DHFR has a pI of 5.3. Below this point the net charge of the protein should increase continuously with decreasing pH, irrespective of any conformational changes. In contrast, the DHFR-attributable resonance signal first increases and then decreases again as the pH is lowered further. Furthermore, the Cys, Etam, and CM surfaces all showed very similar values as pH varied despite quite different surface charges. In fact, the similar signal responses of these varied surfaces indicate that the CM matrix was not even significantly deformable in response to changes in surface charge. On the basis of these considerations, we conclude that the SPR signal from the DHFR surface reflects acid-induced conformational changes in the tethered protein. The direction of the SPR signal change upon protein denaturation may seem counterintuitive. Normally increases in the SPR signal reflect the attachment of proteins or the binding of ligands, both transactions that increase the mass density at the sensor surface with a consequent increase in the dielectric character. In the changes studied here, a compact globular protein tethered to a matrix on the sensor surface denatured, presumably decreasing the mass density very near the surface by allowing portions of the molecules to extend further out into the buffer. Viewed from this simple perspective, it would seem that the SPR signal should decrease. However, this view ignores the importance of water, more specifically, how the water molecule’s interactions with the protein affect their dielectric properties and consequently the apparent refractive index. Refractive index is wavelength dependent, and Biacore instruments utilize a visible wavelength (760 nm). While the specific refractive index increment of a given polymer in this wavelength region is determined primarily by its molarity, conformation also plays a role.28 The increment for native globular proteins is nearly uniform for a wide range of proteins independent of amino acid composition, since their densities fall within a narrow distribution.28 However, the increments for their unfolded forms deviate from those for the native ones.29 Globular proteins form a complex with the surrounding water molecules, with the amount bound inversely proportional to protein compactness. By reducing the water molecules’ random orientation and making them act as condensed extensions of itself, a protein molecule increases the surrounding water’s total dielectric contribution. In fact, Kashpur et al. obtained a relatively high bound water content for an acidunfolded protein compared with its native state; acid unfolding of protein resulted in a significant (∼50%) increase in bound water content.30 In this regard, it is also worth noting that the density of the hydrated (bound) water is greater than that of the bulk (28) Seferis, J. C. Polymer Handbook, 3rd ed.; Wiley-Interscience: New York, 1989; pp 451-461. (29) Derechin, M.; Jordan, B. E. Mol. Biol. Rep. 1975, 2, 107-112. (30) Kashpur, V. A.; Maleev, V. Y.; Shchegoleva, T. Y. Mol. Biol. 1976, 10, 462-469.
Analytical Chemistry, Vol. 70, No. 10, May 15, 1998
2023
(free) water.31 Aside from the contribution of water, the protein’s unfolded state itself may increase polarity. In the compact globular state, close juxtaposition of opposite changes would be favored, whereas unfolding could be expected to increase separation between charged residues and hence their dielectric contributions. Our observation of an increase in RU upon acid unfolding of DHFR is consistent with both these considerations. While we favor the interpretation that the RU increase directly reflects protein unfolding, we cannot completely exclude the possibility that protein conformational changes induce deformation of the dextran matrix and that the detected signal results from such deformation. If the RU increase reflects protein unfolding directly, the refractive index of the protein/hydrated water complex free in solution should also increase upon acid denaturation. Unfortunately, protein aggregation at concentrations above 10 µg/mL and the sensitivity limitations of available instruments have precluded such measurements. Regardless of which explanation is valid, the SPR signal would still ultimately reflect protein conformational changes and, therefore, be useful in monitoring the folding state. A conclusive understanding of how protein conformational changes contribute to the SPR signal must await further studies. Feasibility and Potential Application. Throughout all the denaturation experiments, the signal was reversible and the immobilized DHFR-ASC was stable to repeated exposure to solutions in a pH range from 0.12 to 7.80. Attachment of protein at even higher densities than shown here sometimes led to nonlinear effects during the unfolding/refolding cycles which we attributed to steric constraints between neighboring bound protein molecules (data not shown). However, as the results shown here indicate, attachment of moderately abundant amounts of the protein through a unique residue created a robust system capable (31) Wei, Y.-Z.; Kumbharkhane, A. C.; Sadeghi, M.; Sage, J. T.; Tian, W. D.; Champion, P. M.; Sridhar, S.; McDonald, M. J. J. Phys. Chem. 1994, 98, 6644-6651.
2024 Analytical Chemistry, Vol. 70, No. 10, May 15, 1998
of undergoing repeated denaturation/renaturation cycles without noticeable impairment. Due to the limited range of the Biacore instrument, we did not examine temperature-, urea-, or Gdn‚HClinduced denaturation, but this is not an inherent limitation of the SPR detection method itself. Development of a modified SPR sensor specifically designed for protein folding studies would enable the utilization of other conventional denaturation procedures. The methodology reported here establishes a basis for probing the structural characteristics of immobilized proteins and their relationship to the matrix, thus solving a central problem which has resisted investigation since immobilized enzymes were first developed in the 1960s. Furthermore, this detection technique could also provide a basis for developing novel sensors and/ or switching devices based upon specific protein conformational changes. The stability and reproducibility we observed suggest that this system is well-suited for the realization of an intelligent biological device. ACKNOWLEDGMENT The authors express their gratitude to Dr. S. Tuzi of the Himeji Institute of Technology for his helpful advice as well as to Dr. R. F. Whittier of Amersham Pharmacia Biotech K.K. for reading and criticizing the manuscript. This work was performed under the management of the Research Association for Biotechnology as part of the R&D Project on Basic Technology for Future Industries supported by NEDO (New Energy and Industrial Technology Development Organization) in the Ministry of International Trade and Industry. A part of the progress of this work was presented at the 1997 Miami Nature Biotechnology Winter Symposium held at Fort Lauderdale, FL, on February 3, 1997.
Received for review December 19, 1997. Accepted March 21, 1998. AC9713666
