Using Receptor Conformational Change To Detect Low Molecular

BD Technologies, Research Triangle Park, North Carolina 27709. Small molecules are difficult to directly detect using commercially available surface p...
0 downloads 0 Views 61KB Size
Anal. Chem. 2001, 73, 5732-5737

Using Receptor Conformational Change To Detect Low Molecular Weight Analytes by Surface Plasmon Resonance Jason E. Gestwicki,† Helen V. Hsieh, and J. Bruce Pitner*

BD Technologies, Research Triangle Park, North Carolina 27709

Small molecules are difficult to directly detect using commercially available surface plasmon resonance (SPR) instruments. This is because low molecular weight compounds do not have sufficient mass to cause a measurable change in refractive index. Refractive index is sensitive, however, to other properties besides the mass of the analyte. Recently the detection of substantial conformational changes for immobilized proteins using SPR has been reported. However, this property has not yet been exploited for the detection of low molecular weight ligand binding to immobilized protein receptors. Here we demonstrate that ligand-induced conformational changes can be used to monitor the binding of small molecules to immobilized maltose-binding protein and tissue transglutaminase. Ligand binding to a receptor that decreases in hydrodynamic radius yielded a net decrease in refractive index. A net positive change in refractive index was observed for a receptor that increases in hydrodynamic radius. Refractive index changes could not be explained by addition of analyte molecular mass to the surface. These SPR responses were a result of specific receptorligand interactions, as judged by the reversibility of the response and the similarities between the SPR-determined equilibrium dissociation constants and reported dissociation constants. Additionally, this technique proved to be effective at detecting specific ligands from a panel of small molecules. This SPR method required no alterations in widely used and commercially available instrumentation yet allowed direct detection of very small molecules such as calcium ions (40 Da). Use of receptor conformation to detect low molecular weight analytes has potential applications in the high-throughput screening of small molecule drug libraries and the development of biosensors.

on a surface and introduction of the receptor in the flow system. Binding is easily monitored in this arrangement due to the large change in refractive index that occurs upon binding of the high molecular weight receptor to the small, immobilized ligand. There are distinct advantages, however, in being able to measure the binding of small molecules to immobilized receptors. This arrangement would minimize the quantities of receptor used and allow high-throughput screening of small molecules introduced in the flow system. The advantages of this arrangement have been previously recognized and procedures have been explored to exploit these features.3-5 Some of these strategies have involved direct detection of the small molecules by highly sensitive SPR instruments. For example, Frostell-Karlsson et al., reported detection of immobilized human serum albumin binding to molecules as small as 138 Da.6 Another strategy is the use of a high molecular weight ligand that competes with the small molecular weight analyte for receptor binding.5 The high molecular weight competitor provides the mass necessary for generation of the SPR signal. While useful, this method requires the development of an appropriate competitor with favorable dissociation kinetics. It would be advantageous to use a method that allows the direct detection of small molecules with low molecular weights. Recent reports have suggested that alterations in the conformation of surface-immobilized proteins can generate a detectable change in SPR signal.The first SPR detection of protein conformational change was demonstrated using acid denaturation of dihydrofolate reductase (Escherichia coli).7 Likewise, altering the reduction potential of a solution causes a change in the conformation of cytochrome c and a corresponding change in signal.8 Recently, Salamon et al. reported the use of an SPR-related technique (coupled plasmon-waveguide resonance spectroscopy, CPWR) to study ligand-induced conformational changes in a G-protein coupled receptor embedded in a lipid bilayer.9 These reports suggest that changes in protein hydrodynamic volume,

Surface plasmon resonance (SPR) has found expanded use due to the availability of sensitive benchtop instruments and the expansion of strategies for studying biological or biochemical systems.1,2 Typical applications involve immobilization of the ligand

(3) Karlsson, R.; Roos, H.; Bruno, J.; Stolz, L. BIA J. 1997, 18-21 (Special Issue. Drug Design). (4) Karlsson, R.; Kullman-Magnusson, M.; Ha¨ma¨la¨inen, M. D.; Remaeus, A.; Andersson, K.; Borg, P.; Gyzander, E.; Deinum, J. Anal. Biochem. 2000, 278, 1-13. (5) Karlsson, R. Anal. Biochem. 1994, 221, 142-151. (6) Frostell-Karlsson, A.; Remaeus, A.; Roos, H.; Andersson, K.; Borg, P.; Ha¨ma¨la¨inen, M.; Karlsson, R. J. Med. Chem. 2000, 43, 1986-1992. (7) Sota, H.; Hasegawa, Y.; Iwakura, M. Anal. Chem. 1998, 70, 2019-2024. (8) Boussaad, S.; Pean, J.; Tao, N. J. Anal. Chem. 2000, 72, 222-226. (9) Salamon, Z.; Cowell, S.; Varga, E.; Yamamura, H. I.; Hruby, V. J.; Tollin, G. Biophys. J. 2000, 79, 2463-2474.

* Corresponding author: (e-mail) [email protected]; (fax) 919-597-6400. † Current address: Department of Biochemistry, University of Wisconsins Madison, Madison, WI 53706. (1) Green, R. J.; Frazier, R. A.; Shakesheff, K. M.; Davies, M. C.; Roberts, C. J.; Tendler, S. J. B. Biomaterials 2000, 21, 1823-1835. (2) Myszka, D. G. J. Mol. Recognit. 1999, 12, 279-284.

5732 Analytical Chemistry, Vol. 73, No. 23, December 1, 2001

10.1021/ac0105888 CCC: $20.00

© 2001 American Chemical Society Published on Web 11/02/2001

hydration state, and electronic state can sufficiently report upon the occurrence of a biochemical event at a SPR surface. We envisioned that conformational changes triggered by specific binding might allow direct screening of small molecules without the need for a high molecular weight competitor. In this paper, we demonstrate that the detection of normally SPR-transparent small molecules can be achieved by measuring the change in refractive index that accompanies receptor conformational change. We introduce the concept that SPR signals do not have to be strictly mass induced but can also be conformationally induced by the binding of small molecules to a select group of conformationally active receptor proteins. This approach allows the direct detection of molecules as small as calcium ions (40 Da). This effect was general to two receptors that undergo divergent ligand-induced conformational changes. MATERIALS AND METHODS Materials. Maltose-binding protein (MBP) was prepared from E. coli osmotic shock lysates. All sugars and tissue transglutaminase (tTG) were from Sigma (St. Louis, MO). Experiments were performed on a Biacore Upgrade (Biacore AB, Uppsala, Sweden). Receptor Immobilization. In general, the proteins MBP and tTG were coupled to a research grade CM5 chip (Biacore) using standard EDC/NHS procedures with reagents from the Biacore EDC/NHS coupling kit. Proteins were injected at flow rates of 5-10 µL/min in 10 mM sodium acetate. MBP was injected at 0.4 mg/mL at pH 4.5 for varying lengths of time to yield surfaces from 300 to 4000 resonance units (RU). tTG was injected at 1.3 mg/mL at pH 4.5 for 16 min to generate 9000-10 000 RU surfaces. HEPES-buffered saline-EP (HBS-EP: 0.01 M HEPES pH 7.4, 0.15 M NaCl, 3 mM EDTA, 0.005% surfactant P20; Biacore AB) supplemented with 5 mM CaCl2 was used as an eluent for the MBP immobilization; HBS-EP was the eluent for tTG. Ethanolamine (1.0 M, pH 8.0) was used to quench unreacted esters and was also found to be effective at removing noncovalently adsorbed protein from the chip surface. For MBP, blank control surfaces were activated with EDC/NHS and then quenched with ethanolamine as above; for tTG, unactivated surfaces were used as blank control surfaces. Binding of Small Molecules. Ligands were dissolved in the buffers indicated in the figure legends. Running buffers are also indicated in legends. When appropriate, relative responses were determined by subtracting report points collected from injections over a blank surface. Experiments were performed at 25 °C, using flow rates of 10 µL/min. Surface Regeneration. Bound ligands were removed from MBP by a 20-µL injection of 100 mM recrystallized glucose in 10 mM sodium acetate, pH 4.5. This glucose had been slowly recrystallized from water as the monohydrate one time, to remove possible maltose impurities. Although glucose itself is not a ligand for MBP, it is apparently effective at disrupting the interaction of the disaccharide maltose under flow conditions at high concentrations and low pH. Nonrecrystallized glucose did not have this effect (Figure 1) presumably due to the presence of maltose impurities in commercial glucose sources. After regeneration, the surface was allowed to stabilize for 300 s before additional ligand injections. A number of other routine SPR regeneration conditions were investigated and failed to regenerate active MBP surfaces after maltose addition (0.01-0.1% sodium dodecyl sulfate; 50 mM

Figure 1. Representative SPR sensorgram demonstrating change (in RU) upon injection of maltose in running buffer (HBS-EP with 5 mM CaCl2) on a MBP surface and requirements for regeneration of that response. (A) The full sensorgram, which shows the bulk refractive index changes. (B) An enlargement of the area of interest. Injections are as follows: (1) 1 mg/mL maltose; (2) repeat injection of 1 mg/mL maltose, demonstrating saturation; (3) 100 mM glucose; (4) 100 mM recrystallized glucose in pH 7.0 running buffer; and (57) 100 mM recrystallized glucose in 10 mM sodium acetate pH 4.5, showing progressive restoration of signal. (C) An overlay of injections of 1 mg/mL maltose in running buffer over the MBP surface (solid line) and a blank control surface (dashed line).

sodium hydroxide; 1 M sodium chloride; 6 M guanidinium hydrochloride; 6 M urea; 0.5 mM sodium acetate, pH 4.5). Calcium was removed from tTG by injection of 50 mM Tris-HCl, pH 7.6, with 5 mM EDTA. Denaturation of Transglutaminase. An active transglutaminase surface (∼10 000 RU) was denatured by continuous exposure to 6 M guanidine hydrochloride, overnight at 25 °C. A solution of 100 mM CaCl2 in 45 mM Tris-HCl, pH 2.3, was applied to the tTG surface and a control surface to test surface activity. Partial refolding of the protein was performed by continuous exposure to 50 mM Tris-HCl, pH 7.6, overnight at 25 °C. Data Analysis. (1) Equilibrium Dissociation Constant. Concentration curves were fit according to the equation

R ) Rinf + (R0 - Rinf)/(1 + x/Kd)

(1)

where R is the resonance signal, Rinf is the signal at infinite solution ligand, and R0 is the signal at zero ligand concentration. (2) Calculation of Theoretical Response. The theoretical maximum signal, Rmax, was determined using an equation derived by Karlsson et al.3 The equation is

Rmax ) (mwanalyte/mwligand)n(A)Rimmobilized Analytical Chemistry, Vol. 73, No. 23, December 1, 2001

(2) 5733

where mwanalyte is the molecular weight of the analyte in the flow system, mwligand is the molecular weight of the immobilized protein, n is the number of analyte-binding sites on the protein, A is the fraction of active sites on the immobilized protein, and Rimmobilized is the RU of protein on the surface. For MBP-maltose, we assume n ) 1; for transglutaminase-calcium, we assume n ) 6.10 We also assume 100% active protein on the surface. The molecular weights used are as follows: MBP (40 600), tTG (77 124), maltose (360), and calcium (40). RESULTS MBP Surface: Negative Signal. In initial experiments, we immobilized maltose-binding protein, a bacterial periplasmic binding protein, to a carboxymethyl dextran surface. Application of maltose to the MBP surface caused an initial increase in SPR signal, due to changes in the bulk refractive index (Figure 1A). However, the baseline remained below the initial value following the injection (Figure 1B). This decrease in baseline was reversed only by injection of high concentrations (100 mM) of recrystallized glucose in 10 mM sodium acetate at pH 4.5. Nonrecrystallized glucose was ineffective for regeneration (even the highest purity commercial samples have 0.1-0.2% maltodextrin impurities, which may interact with MBP at glucose concentrations of 1 mM or above). Therefore, the observed drop in the immobilized MBP response was reversible, signifying that this was not simply due to mass loss from the surface. These surprising initial observations encouraged us to further explore this phenomenon. Maltose solution was injected over MBP surfaces and blank control surfaces (Figure 1C). Subtraction of this signal from that of the control resulted in a reproducible net decrease in signal. A positive change in signal would have been expected for mass buildup of maltose on the immobilized MBP (see Discussion). Because the observed drop in refractive index could also be a result of nonspecific matrix effects, we conducted a series of experiments designed to determine the requirements for the response. MBP Surface Density. Maltose at saturating concentrations (1 mg/mL) was injected over MBP surfaces of varying protein densities. The change in response was measured relative to a blank surface. The relative response was determined by collecting points in the dissociation phase, which was found to yield the most reproducible data. The decrease in signal was dependent on the amount of MBP immobilized (Figure 2). As this effect scales with protein concentration, it is unlikely to be a nonspecific matrix effect. MBP-Maltose Equilibrium Dissociation Constant. A series of maltose solutions varying in concentration was injected over an MBP surface (∼2800 RU). We again observed negative changes in refractive index upon maltose binding. The negative response was dependent on maltose concentration (Figure 3). The equilibrium dissociation constant for MBP-maltose was determined to be 0.3 µM, using eq 1. This value is similar to the reported dissociation constant for maltose binding to MBP in solution (∼1 µM).11 These results suggest that the observed drop in RU is a result of a specific receptor-ligand interaction. (10) Mariani, P.; Carsughi, F.; Spinozzi, F.; Romanzetti, S.; Meier, G.; Casadio, R.; Bergamini, C. M. Biophys. J. 2000, 78, 3240-3251. (11) Thomson, J.; Liu, Y.; Sturtevant, J. M.; Quiocho, F. A. Biophys. Chem. 1998, 70, 101-108.

5734 Analytical Chemistry, Vol. 73, No. 23, December 1, 2001

Figure 2. Dependence of negative RU response on the amount of immobilized MBP. Variable-volume injections were performed to immobilize MBP. The amount of immobilized MBP was determined by determination of the change in RU after EDC/NHS injection, MBP injection, and ethanolamine quenching of unreacted esters. Maltose (1 mg/mL) in water was injected over the surfaces. All surfaces were regenerated with recrystallized glucose. The relative response was determined by subtracting a report point 100 s postinjection from a similar point on a control sensorgram. Each point is the average of at least two injections. Injections were 20 µL at a flow rate of 10 µL/ min (2-min contact time). Running buffer was HBS-EP.

Figure 3. Dependence of negative RU change on the concentration of maltose injected. Maltose solutions were in water at the indicated concentrations. Injections were performed on a surface bearing 2800 RU MBP and a blank control lane. Injections were repeated 3-5 times. Injections were 20 µL at a flow rate of 10 µL/min (2-min contact time). Report points (9) were collected at 100 s postinjection and subtracted from similar points from a blank control. Running buffer was HBS-EP with 5 mM CaCl2. The equilibrium dissociation constant for the MBP-maltose interaction was 0.3 µM. The inset shows the data for the MBP (2) and control (b) surfaces prior to subtraction.

MBP Specificity. We envisioned that this technique may be a useful way to screen small molecules for potential ligands. To test this hypothesis, a series of saccharides were injected over an MBP surface (∼2800 RU). Some of the sugars in this series are known to bind to MBP; the remainder do not interact with the protein.11,12 Substantial negative SPR responses (10-20 RU) were observed only for those ligands previously shown to bind to MBP (Table 1). At the tested concentration of glucose (10 mM), (12) Adler, J.; Hazelbauer, G. L.; Dahl, M. M. J. Bacteriol. 1973, 115, 824-847.

Table 1. Panel of Sugars Tested as Ligands for MBP analyte

known ligand capabilitya

SPR responseb

control (no sugar) maltose (2.8 mM) D-galactose (10 mM) maltotetraose (1.5 mM) maltotriose (10 mM) lactose (10 mM) D-glucose (10 mM)

+ + + -

-0.1 ( 4.3 -21 ( 1.9 -0.5 ( 1.6 -15.2 ( 0.3 -10.3 ( 0.8 -3.5 ( 1.6 -7.8 ( 3.7

a See refs 11 and 12. b Defined as reproducible negative change in RU compared to injection over a control surface. Experiments are the averages of 2-7 runs. HBS-EP with 5 mM CaCl2 was the running buffer; the sugars were injected in water.

Figure 5. Calcium concentration dependence of a tTG surface. Sodium chloride or calcium chloride was injected over a surface bearing 9000 RU tTG or a blank control lane. Results represent RU after subtraction of the blank control lane. In two experiments, calcium chloride (9) and sodium chloride (b) samples were in Tris buffer, which was also the running buffer. The calcium chloride-dependent signal was also examined in the presence of sodium chloride. Calcium chloride concentration was varied in Tris buffer with 150 mM NaCl ([) or in Tris buffer with sufficient NaCl to maintain a constant ionic strength equivalent to 150 mM NaCl (2). For these latter experiments, the running buffer was Tris buffer with 150 mM NaCl. The Kd for tTGcalcium in Tris buffer was 2.9 mM for this particular set of data, as fit by eq 1. Table 2. Denaturation of the tTG Surfacea Figure 4. Representative sensorgrams during injection of 25 mM CaCl2 in 50 mM Tris-HCl, pH 7.6, over a both surface bearing 9000 RU tTG and a blank control surface. Running buffer was 50 mM TrisHCl, pH 7.6 (Tris buffer).

some signal was observed, but it may be attributed to maltose impurities. This experiment indicates that MBP-ligand specificity is maintained after immobilization. tTG. To explore the generality and specificity of this effect, we immobilized tissue transglutaminase, an enzyme involved in posttranslational protein modification and apoptosis, on a carboxymethyl dextran surface. This allosteric protein was selected on the basis of its ability to undergo significant conformational changes upon binding to calcium ions.13 To measure the response to small-molecule ligands, calcium solutions were injected over blank surfaces and tTG-bearing surfaces. The response was measured by using report points near the end of the injections (-10 s). Interestingly, calcium binding to tTG induced a positive change in refractive index (Figure 4). Sodium chloride induced a much smaller response from the tTG-bearing surface than calcium chloride, suggesting that the interaction was specific (Figure 5). In the presence of either constant amounts of sodium chloride (150 mM), or sufficient sodium chloride to maintain a constant ionic strength, the presence of calcium again caused a significant increase in the response. The Kd for tTG-calcium in Tris buffer was 3.7 ( 0.8 mM as calculated by eq 1 (four binding curves (13) Casadio, R.; Polverini, E.; Mariani, P.; Spinozzi, F.; Carsughi, F.; Fontana, A.; Polverino de Laureto, P.; Matteucci, G.; Bergamini, C. M. Eur. J. Biochem. 1999, 262, 672-679.

tTG surface

response to 100 mM calcium chloride (RU)

before denaturation postdenaturation after refolding

1300 60 400

a Calcium chloride (100 mM) was applied to both a tTG surface (10 000 RU) and a blank control surface prior to denaturation (50 mM Tris-HCl, pH 7.6), immediately postdenaturation (45 mM Tris-HCl, pH 2.3), and after partial refolding had occurred (50 mM Tris-HCl, pH 7.6). Results represent RU changes for the tTG surface after subtraction of the control lane.

performed on two surfaces). This value is similar to the reported Kd of calcium for this protein (0.2-3 mM).14 tTG Control: Denatured Surface. Previous work15 has shown that a high concentration (6 M) guanidine hydrochloride is required to denature tTG. In addition, tTG can be refolded by dilution back into buffer. To demonstrate that the observed tTGCa signal requires a properly folded protein, a tTG surface (∼10 000 RU immobilized) was denatured and then refolded. Application of high-concentration calcium chloride prior to denaturation demonstrated that the surface was active, while little response was observed immediately after the denaturation step (Table 2). Partial activity (30%) was recovered upon refolding. DISCUSSION We report here the development of a new assay for SPRmediated detection of small molecule-receptor interactions. This (14) Mottahedeh, J.; Marsh, R. J. Biol. Chem. 1998, 273, 29888-29895. (15) Di Venere, A.; Rossi, A.; De Matteis, F.; Rosato, N.; Finazzi Agro´, A.; Mei, G. J. Biol. Chem. 2000, 275, 3915-3921.

Analytical Chemistry, Vol. 73, No. 23, December 1, 2001

5735

method differs from other approaches in that it is based on monitoring the conformational changes of surface-immobilized receptors that accompany small-molecule binding. This method was originally discovered during injections of maltose over a surface bearing MBP. The mass of maltose (360 Da) should have been sufficient to cause a positive change in refractive index of up to 25 RU, as predicted by an equation used in standard Biacore applications (eq 2). A positive signal of this size has in fact been observed for maltose binding to a high-density maltose-specific antibody SPR surface.16 Maltose injection on an MBP surface, however, resulted in a reproducible negative change in refractive index (-20 RU). It was apparent from these results that there was an additional component of the MBP-maltose interaction that contributed to the SPR signal. Additional experiments suggested that this response was the result of a specific receptor-ligand interaction. For example, the negative refractive index change was dependent on the amount of MBP on the surface and the concentration of maltose in solution. In addition, the Kd measured by SPR was similar to the Kd for the MBP-maltose interaction, and only known ligands for MBP caused a significant negative signal. One prior SPR study of MBP has described immobilization of MBP to a Ni2+-nitrilotriacetic (NTA) SPR chip through histidine tags, but a response to maltose was not described.17 An interesting consequence of this negative signal was the need for regeneration. Among a number of conditions tested, only recrystallized glucose in acetate restored the original signal and maltose-binding activity. Based on the known kinetic association rate18 for MBP-maltose of 107 M-1 s-1 and given an MBP surface density of ∼2800 RU, it would appear the system is probably mass transport limited. Therefore, buffer flow alone is insufficient for timely surface regeneration. Maltose binding to MBP is known to induce a conformational change, and detailed information is available from numerous X-ray crystal structures. This change is a hinge-twist between the two domains of MBP that causes a net decrease in hydrodynamic radius.19 We propose that the negative change in RU we observed is a result of this ligand-induced conformational change. Additionally, we propose that the sign of the refractive index change is a function of the net change in hydrodynamic radius that occurs upon ligand binding. Consistent with these hypotheses, a positive change in RU was observed for calcium binding to tTG. tTG is allosterically regulated by calcium and undergoes a positive change in hydrodynamic radius as a result of an approximately 15° rotation.10,13,15 We suggest that receptors that decrease hydrodynamic radius yield negative changes in RU and receptors that undergo a positive change in hydrodynamic radius yield a net positive change in RU. The positive change in refractive index we report for calcium binding to tTG cannot be explained by addition of mass at the surface. For calcium binding to a 9000 RU protein surface, the expected change in RU due to mass alone is only +28 RU. In (16) Ohlson, S.; Jungar, C.; Standh, M.; Mendenius, C.-F. Trends Biotechnol. 2000, 18, 49-52. (17) Nieba, L.; Nieba-Axmann, S.; Persson, A.; Ha¨ma¨la¨inen, M.; Edebratt, F.; Hansson, A.; Lidholm, J.; Magnusson, M.; Karlsson, A. F.; Pluckthun, A. Anal. Biochem. 1997, 252, 217-228. (18) Miller, D. M.; Olson, J. S.; Pflugrath, J. W.; Quiocho, F. A. J. Biol. Chem. 1983, 258, 13665-13672. (19) Sharff, A. J.; Rodseth, L. E.; Spurlino, J. C.; Quiocho, F. A. Biochemistry 1992, 31, 10657-10663.

5736

Analytical Chemistry, Vol. 73, No. 23, December 1, 2001

contrast, we observed a signal that was approximately +1000 RU. Therefore, using a conformationally active receptor (tTG) and measuring calcium-induced conformational change by SPR increased the signal intensity 36-fold. Unlike calcium, maltose is of sufficient molecular mass (360 Da) to be observed directly by SPR. However, we observed negative changes in refractive index for maltose binding to MBP. This suggests that, for MBP, the negative SPR signal is larger than the positive response that would be caused by accumulation of maltose mass. In this case, therefore, conformational change is likely the dominant determinant of the net change in refractive index. The net effect of these negative and positive responses could account for the relatively small change in RU observed for the MBP-maltose interaction (5-30 RU) compared to that observed for tTG (1000 RU). A consideration that is raised by these results is that conformationally induced SPR may be a component of other SPR experiments. This component of the sensorgram may be minor in many cases, particularly when the protein conformational changes are minimal. In other examples, however, this component may be large. The conformational changes in MBP and tTG would be considered to be large. Correspondingly, there is a significant contribution from conformational change to their SPR signals. The mechanism for these signals is not clear, but it may involve refractive index changes through disruption of the CMD matrix. Alternatively, the change in protein hydrodynamic volume may disrupt protein association with surrounding water molecules, altering dielectric properties of the protein-water matrix.7 A more detailed analysis of the relationship between changes in hydrodynamic radius and the contribution from conformational change to the SPR signal is needed to define this parameter. There are significant limits on the applicability of this technique. The primary issue is that only receptors that undergo conformational change20 upon binding to small molecules are expected to yield a response. The type of conformational change (hinge-twist or rotation), however, was not problematic. MBP and tTG undergo divergent conformational changes, yet both yielded measurable SPR responses. Finally, the use of conformational change as a detection method limits detection to those small molecules that are able to induce the change. For some receptor antagonists, for example, ligand binding might occur but fail to yield a response due to a lack of accompanying conformational change. However, this aspect of our approach could be a potentially powerful advantage when the goal is to discover small molecules that activate conformationally responsive receptors. Only those small molecules that induce the change would be detected. In other traditional binding assays, such as radioactive ligand-binding experiments, all small molecules that bind would be detected regardless of their ability to activate conformational changes. Specific interactions could be more easily studied. With MBP, for example, nonspecific binding events could be readily distinguished by a positive RU signal, as opposed to a negative specific signal. Thus, we feel that the nature of conformationally induced SPR relegates the technique to select applications, but it may have distinct advantages within those applications. (20) Gerstein, M.; Lesk, A. M.; Chothia, C. Biochemistry 1994, 33, 6739-6749.

We have demonstrated that low molecular weight analytes can be detected by conformationally induced SPR on a commercially available instrument. We were able to directly detect small molecules with lower mass than previously detectable by SPR and without the requirement for high molecular weight competitors. These results reveal this technique as a method to investigate the conformational changes in receptors and the mechanism of action of agonists, partial agonists, and antagonists. Additionally, conformationally induced SPR expands the possible strategies for the screening of drug libraries and may have potential application in the development of biosensors.

ACKNOWLEDGMENT We thank Dr. David J. Wright (BDT) for supplying critical reagents and Christopher W. Cairo (UWsMadison) for helpful conversations. J.E.G. acknowledges the NIH Biotechnology Training Program (T32GM08349) for support.

Received for review May 29, 2001. Accepted September 20, 2001. AC0105888

Analytical Chemistry, Vol. 73, No. 23, December 1, 2001

5737