Receptor Coupling in

Jan 11, 2006 - ... biosensor-based approach to monitor the molecular interactions between G ... Simultaneous and Label-Free Determination of Wild-Type...
42 downloads 4 Views 229KB Size
Anal. Chem. 2006, 78, 1228-1234

Surface Plasmon Resonance Study of G Protein/ Receptor Coupling in a Lipid Bilayer-Free System Konstantin E. Komolov,†,‡ Ivan I. Senin,‡ Pavel P. Philippov,‡ and Karl-Wilhelm Koch*,†

AG Biochemistry, Faculty V, IBU, University of Oldenburg, D-26111 Oldenburg, Germany, and A. N. Belozersky Institute of Physico-Chemical Biology, Moscow State University, 119992 Moscow, Russia

Surface plasmon resonance (SPR) spectroscopy is a technique to study protein-protein interactions in real time; however, application of SPR spectroscopy for investigations of membrane receptors is difficult with respect to functional and uniform immobilization of receptors on a biosensor surface. In the current study, we developed a simple, direct, biosensor-based approach to monitor the molecular interactions between G protein transducin (Gt) and rhodopsin (Rho), a prototypical G protein-coupled receptor (GPCR). Detergent-solubilized dark-adapted Rho was captured onto a biosensor surface via lectin interaction, enabling site-directed immobilization of the receptor that made its cytoplasmic surface accessible to a coupling G protein. The system resembled the natural system with respect to receptor density, binding of Gt following flash or constant light application, fast GTP-dependent dissociation of Gt from Rho, regeneration of Rho, and dependence of Gt binding on light intensity and on concentration of Gt. The apparent KD of the Gt/Rho interaction was 13.6 nM. Our results validate the use of SPR spectroscopy as a tool to study G protein activation in GPCR systems and could be extended for application to other interaction partners of GPCRs. G protein-coupled receptors (GPCR) constitute a significantly large part of mammalian genomes (1-2% in humans) and serve key functions in hormone, neurotransmitter, and sensory signaling. All members of this superfamily share the typical seventransmembrane helix topography and the property to catalyze GDP/GTP-exchange on downstream interacting heterotrimeric G proteins.1-3 The light-sensitive photopigment in rod outer segments (ROS) is rhodopsin (Rho), which belongs to the family of GPCRs and is among the best understood.1,4,5 Knowledge about Rho includes a variety of structural and functional aspects. For example, its three-dimensional structure has been solved by X-ray analysis,6 and the interaction of Rho with the G protein transducin * Corresponding author. Phone: +49-441-798-3640. Fax: +49-441-798-193640. E-mail: [email protected]. † University of Oldenburg. ‡ Moscow State University. (1) Helmreich, E. J. M.; Hofmann, K. P. Biochim. Biophys. Acta 1996, 1286, 285-322. (2) Ji, T. H.; Grossmann, M.; Ji, I. J. Biol. Chem. 1998, 273, 17299-17302. (3) Pierce, K. L.; Premont, R. T.; Lefkowitz, R. J. Mol. Cell. Biol. 2002, 3, 639650. (4) Ernst, O. P.; Bartl, F. J. ChemBioChem 2002, 3, 968-974. (5) Filipek, S.; Stenkamp, R. E.; Teller, D. C.; Palczewski, K. Annu. Rev. Physiol. 2003, 65, 851-879.

1228 Analytical Chemistry, Vol. 78, No. 4, February 15, 2006

(Gt) has been elucidated in great detail.7 Therefore, the lighttriggered enzyme cascade is often called a benchmark system of G protein-coupled signaling pathways. Due to the importance of GPCRs in basic biological and applied pharmaceutical research, powerfull tools are needed to allow investigation of proteinprotein interactions and drug screening. Biosensor-based chip technology is widely applied to study biomolecular interactions, and in particular, surface plasmon resonance (SPR) spectroscopy is frequently employed.8-10 However, a prerequisite of sensor-chip-based interaction studies is the immobilization of one binding partner. This is more challenging in the case of membrane receptors, since it requires keeping a protein with large hydrophobic parts in a functional state within a rather artificial environment. Therefore, investigation of whole GPCRs by SPR spectroscopy on a routine basis has not been widely established.11-14 For example, Rho, the prototype GPCR, and its interaction with Gt has been studied by evanescent wavebased biosensor systems (mainly by SPR spectroscopy),15-20 but the results are rather heterogeneous and apparently depend on factors such as immobilization strategy, bleached or light-activated status of Rho, and presence of lipid bilayers (a closer inspection of these aspects is presented in the Discussion Section). (6) Palczewski, K.; Kumasaka, T.; Hori, T.; Behnke, C. A.; Motoshima, H.; Fox, B. A.; Le Trong, I.; Teller, D. C.; Okada, T.; Stenkamp, R. E.; Yamamoto, M.; Miyano, M. Science 2000, 289, 739-745. (7) Arshavsky, V. Y.; Lamb, T. D.; Pugh, E. N., Jr. Annu. Rev. Physiol. 2002, 64, 153-187. (8) Schuck, P. Annu. Rev. Biophys. Struct. 1997, 26, 541-566. (9) Koch, K.-W. Methods Enzymol. 2000, 315, 785-797. (10) Lee, H. J.; Yan, Y.; Marriott, G.; Corn, R. M. J. Physiol. 2005, 563.1, 61-71. (11) Salamon, Z.; Cowell, S.; Varga, E.; Yamamura, H. I.; Hruby, V. J.; Tollin, G. Biophys. J. 2000, 79, 2463-2474. (12) Rao, N. M.; Silin, V.; Ridge, K. D.; Woodward, J. T.; Plant, A. L. Anal. Biochem. 2002, 307, 117-130. (13) Neumann, L.; Wohland, T.; Whelan, R. J.; Zare, R. N.; Kobilka, B. K. ChemBioChem. 2002, 3, 993-998. (14) Stenlund, P.; Babcock, G. J.; Sodroski, J.; Myszka, D. G. Anal. Biochem. 2003, 316, 243-250. (15) Salamon, Z.; Wang, Y.; Brown, M. F.; Macleod, H. A.; Tollin, G. Biochemistry 1994, 33, 13706-13711. (16) Heyse, S.; Ernst, O. P.; Dienes, Z.; Hofmann, K. P.; Vogel, H. Biochemistry 1998, 37, 507-522. (17) Bieri, C.; Ernst, O. P.; Heyse, S.; Hofmann, K. P.; Vogel, H. Nat. Biotechnol. 1999, 17, 1105-1108. (18) Clark, W. A.; Jian, X.; Chen, L.; Northup, J. K. Biochem. J. 2001, 358, 389397. (19) Karlsson, O. P.; Lo ¨fås, S. Anal. Biochem. 2002, 300, 132-138. (20) Minic, J.; Grosclaude, J.; Aioun, J.; Persuy, M.-A.; Gorojankina, T.; Salesse, R.; Pajot-Augy, E.; Hou, Y.; Helali, S.; Jaffrezic-Renault, N.; Bessueille, F.; Errachid, A.; Gomila, G.; Ruiz, O.; Samitier, J. Biochim. Biophys. Acta 2005, 1724, 324-332. 10.1021/ac051629t CCC: $33.50

© 2006 American Chemical Society Published on Web 01/11/2006

In the present work, we intend to establish a robust method to analyze Rho/Gt coupling which should fulfill the following requirements: nonactivated Rho should be immobilized and be activated by direct illumination on the chip surface; density of receptor on the surface must be variable; the system should work in a lipid bilayer-free and lipid bilayer-containing environment; and the system should be fully reversible, allowing repeated immobilization and analytic cycles. EXPERIMENTAL SECTION Materials. CM5 and L1 sensor chips were from Biacore AB. Concanavalin A (ConA), methyl R-D-mannopyranoside, 3-cholaminopropyldimethyl-ammonio-1-propane sulfonate (CHAPS) were from Serva; dodecyl-β-D-maltoside, 9-cis-retinal, and GTP were from Sigma. PD-10 columns were from Amersham Biosciences. All other reagents were obtained from Merck, Fluka, and Serva and were at least analytical grade. Preparation of Washed ROS Membranes and Gt. Bovine ROS were prepared from fresh bovine retinas and stored at -80 °C as described previously.21 Hypotonically stripped disk membranes were prepared from ROS by repetitive washes with lowsalt buffer at 4 °C as described.22 All steps that required handling of nonactivated Rho were performed under dim red light. The Rho (42 kDa) concentration in the washed membranes was determined from its absorption spectrum using 498 ) 42000 M-1 cm-1.5 Gt was isolated from ROS by selective extraction steps involving elution with GTP as described.22 Two consecutive PD10 column chromatography steps were used to remove excess nonbound GTP and to exchange a sample of Gt into the SPR running solution (50 mM 3-(N-morpholino)propanesulfonic acid (MOPS), pH 7.5; 150 mM NaCl; 3 mM MgSO4; 10 µM CaCl2; 10 µM MnCl2). Purified Gt was analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis, as described.23 ConA-purified Rho samples were analyzed in the same way and showed a homogeneous preparation. The sample of Gt was frozen in liquid nitrogen and stored at -80 °C until use. Gt concentration was determined by a Bradford assay24 using bovine serum albumin as the standard. Immobilization of ConA and Rho. All steps involving monitoring of SPR signals were performed with a Biacore 1000; the main principle of operation has been described before.8,9,14,18,19 To capture Rho on the dextran surface of a CM5 sensor chip (Biacore), ConA was first immobilized by amine linkage to carboxy-activated dextran.18 The dextran matrix of the sensor chip was activated by 35 µL of 50 mM N-hydroxysuccinimide and 200 mM N-ethyl-N′-[(dimethylamino)propyl]carbodiimide at a flow rate of 5 µL/min, followed by a 7-min injection of 0.1 mg/mL ConA (diluted in 100 mM sodium acetate, pH 5.0). Any remaining reactive carboxy groups were deactivated using a 7-min pulse of 1 M ethanolamine hydrochloride, pH 8.5. Prior to immobilization, 50 µM Rho in washed ROS membranes was solubilized with 25 mM CHAPS in 10 mM MOPS, pH 7.5, at 4 °C for 30 min (gentle shacking). Nonsolubilized material was sedimented for 10 min at 14000g (4 °C), and the supernatant (diluted 5-fold with a solution (21) Koch, K.-W.; Lambrecht, H.-G.; Haberecht, M.; Redburn, D.; Schmidt, H. H. H. W. EMBO J. 1994, 13, 3312-3320. (22) Ku ¨ hn, H. Curr. Top. Membr. Transp. 1981, 15, 171-201. (23) Laemmli, U. K. Nature 1970, 227, 680-685. (24) Bradford, M. M. Anal. Biochem. 1976, 72, 248-254.

of 50 mM MES, pH 6.0; 1 mM CaCl2; and 1 mM MnCl2) was used for immobilization. Solubilization and purification of Rho in dodecyl-β-D-maltoside resulted in a less stable preparation, and SPR signals were less reliable. Rho (4-400 µg/mL) was injected over the ConA-modified surface (see below) at a flow rate of 2 µL/min for 25 min. After the injection, the system was washed at high flow rate with the SPR running solution until a stable baseline was restored (30 min). This washing step works like a flow dialysis procedure and ensures the removal of the detergent CHAPS (cmc ) 6.5 mM) from the sensor chip surface; however, hydrophobic parts of the transmembrane receptor rhodopsin will still be attached to some lipid or detergent molecules to maintain its functional integrity. All subsequent injections were performed at a flow rate of 5 µL/min or otherwise as indicated. Regeneration of Rho-Coated Surfaces. A critical feature of SPR analysis is the reliable reuse of coated surfaces for repeated injections of analytes at different concentrations. Once Rho is illuminated, it needs to be regenerated on the chip surface or to be replaced by fresh dark adapted Rho. Although regeneration of Rho by 9-cis-retinal worked well, it was never complete and, therefore, less suitable than replacement of activated Rho by nonactivated. Therefore, we tried to regenerate the surface before starting a new cycle of Rho immobilization, Gt application, and illumination. Successful regeneration of the surface was achieved with a sequence of two different solutions, 100 mM methyl R-Dmannopyranoside supplemented with 30 mM n-octyl-β-glucoside, and 6 M urea. ConA on the sensor chip surface withstood these treatments and was able to bind Rho to a similar extent. Thus, this method was well-suited to analyze repeated analyte injections, each with a completely new coated surface of dark adapted Rho. If Rho was kept dark-adapted on the chip, complete removal by methyl R-D-mannopyranoside without urea was sufficient. SPR Measurements and Data Evaluation. For measurements of Gt binding to Rho, the BIAcore 1000 system was modified as to allow a laser light illumination of the flow cell by an external device. For this purpose, an optical fiber from Diode Pumped Nd: YAG solid-state laser (model LL-11 qps; Laser-Compact Co. Ltd.) was used to activate Rho by illumination (λ ) 532 nm) of the surface of the sensor chip. Instrument temperature was set to 25 °C. All operations were performed under dim red light. Binding of Gt to immobilized Rho was performed as follows: 0.0175-0.7 µM of Gt was injected over the Rho surface for 7 min. Rho was activated by laser light when dark (or nonspecific) binding of Gt was complete (usually 1-2 min after Gt injection). Photoexcitation of immobilized Rho triggered the binding of Gt. After the end of Gt injection, the flowcell was washed for 5 min with SPR running solution, and then 200 µM GTP was injected for 5 min in the SPR running solution. In some experiments, the laser light was attenuated by neutral density filters of 5 or 10% transmission. After every cycle of illumination and Gt binding, the surface of ConA was regenerated as described above. Calculation of receptor densities and of amount of analyte binding was done by using the Biacore standard relation of 1 RU ) 1 pg of protein mm-2. Association and dissociation rate constants were calculated by nonlinear fitting of the primary sensorgram data using BIAevaluation 3.1 software. Scatchard analysis of sensorgrams was performed as described.9 Bulk refractive index changes that originated from a change in running solution were subtracted to Analytical Chemistry, Vol. 78, No. 4, February 15, 2006

1229

Figure 1. Immobilization of Rho on a sensor chip surface and subsequent interaction with Gt. Covalent linkage of ConA to dextran of a sensor chip formed a surface for the receptor deposition. Dark-adapted solubilized Rho was captured to a ConA coated sensor chip surface by its N-terminally located carbohydrates. This procedure ensures uniform immobilization of the receptor. Illumination of sensor chip by laser light activates the immobilized receptor and triggers binding of the heterotrimeric G protein Gt. Injection of GTP led to fast dissociation of Gt from the complex with Rho. Any increase or decrease of mass at the sensor surface is monitored as a change in resonance units (RU) and recorded as a sensorgram.

facilitate recognition of protein binding and dissociation steps or removed by the Biaevaluation cut function. Random occurring spikes were also eliminated by the Biaevaluation cut function. RESULTS Strategy of Rho Immobilization and Real-Time Detection of Gt Activation. To monitor the binding and activation of G protein Gt by bovine Rho, we developed a procedure for the immobilization of the functional receptor on a biosensor surface. We chose to capture native Rho via noncovalent attachment of its N-terminally linked carbohydrate moiety to the lectin ConA,18 because it is expected from the specificity of the lectincarbohydrate interaction that Rho is uniformly oriented on the chip surface. Figure 1 illustrates the main aspects of the strategy that we followed: (1) ConA was covalently linked to the carboxyactivated dextran surface of a CM5 chip, and then Rho molecules were deposited on the sensor chip surface in one uniform orientation that allowed the cytoplasmic domain of the receptor to have access to a coupling G protein. (2) The receptor was not chemically modified, which could, in principle, impair G protein interaction. (3) Nonactivated, that is, dark-adapted, Rho was immobilized, which allowed subsequent activation by light. Rho was isolated under dark conditions by detergent treatment of hypotonically stripped ROS disk membranes, and the resulting extract was used for immobilization on the ConA surface. Interaction of Rho with ConA on the chip surface was robust and resulted in a stable baseline after washing away any unbound molecules of Rho, lipids, or detergent (not shown). We tested the assumption of a uniform ConA-Rho interaction by applying a washing step with methyl R-D-mannopyranoside to dark-adapted immobilized Rho. The carbohydrate methyl R-D-mannopyranoside is routinely used as a competing reagent to release glycoproteins from ConA columns during lectin affinity chromatography; therefore, it should remove any specifically bound Rho from the ConA surface. By this test, we were able to remove nearly 100% of the immobilized Rho; the effect was reproducible for different ConA surfaces. This indicated that Rho was uniformly oriented on the sensor chip 1230 Analytical Chemistry, Vol. 78, No. 4, February 15, 2006

surface, which means that the oligosaccharide chains of the amino terminus of Rho interacted with ConA, thereby leaving the cytoplasmic part accessible to analytes in the running buffer. Consequently, we were able to monitor Rho/Gt interaction, as shown in the sensorgram of Figure 2. Isolated nucleotide-free Gt (0.7 µM) was rinsed over the sensor chip surface containing immobilized Rho for 2 min in the dark. Under these conditions, only a small positive change in RU was observed that corresponded to either nonspecific interaction of Gt with the sensor chip matrix or to binding of Gt to activated Rho that was already present on the chip surface (see small change in RU at the beginning of the sensorgram in Figure 2). Significant binding of Gt was triggered by continuous green laser light (λ ) 532 nm), which was delivered to the sensor chip surface by an optical fiber.

Figure 2. Real-time detection of Gt association and dissociation to Rho by SPR spectroscopy. Injection of Gt onto the Rho-coated sensor chip surface resulted in a small increase of RU (see main text). Specific light-induced binding of Gt was triggered by continuous laser light at time t0 ) 0. The sensorgram was recorded with 0.7 µM Gt in the running buffer (black bar). Slow dissociation was initiated by flushing with running buffer only (open bar). Complete and fast dissociation was triggered by injection of 200 µM GTP (hatched bar). Surface density of Rho was 3.0 × 1010 molecules/mm2.

Table 1. Analysis of Gt/Rho* Interaction in the Absence of Nucleotidesa Gt/Rho* ka,

M-1

s-1

(1.59 ( 0.63) ×

kd, s-1 104

(1.9 ( 0.72) ×

KD, nM 10-4

13.6 ( 6.2

a Association and dissociation phases of the primary sensorgram data were analyzed by nonlinear fitting using BIAevaluation 3.1 software (nine independent experiments) to yield ka, kd, and KD.

The association of Gt and Rho was monitored as a positive binding signal of more than 300 RU (Figure 2). In the absence of Gt, the SPR signal was not changed upon illumination (data not shown) which confirms that the binding signal corresponds exclusively to Gt association. Stop of Gt application resulted in a slow, continuous decrease of the resonance signal representing the dissociation of nucleotide-free Gt from Rho (open bar in Figure 2). Finally, rapid and complete dissociation of Gt from Rho was achieved by injection of 200 µM GTP (Figure 2, hatched bar). The dissociation rate constant of nucleotide-free Gt from Rho was (1.9 ( 0.72) × 10-4 s-1 (analysis of 9 experiments as shown in Figure 2), and that of GTP-bound Gt was 0.8 ( 0.15 s-1 (4 determinations). With the association rate constant of (1.59 ( 0.63) × 104 M-1 s-1, the resulting KD of the Rho/Gt complex in the absence of nucleotides was 13.6 ( 6.2 nM (Table 1). These results showed that the essential features of Rho/Gt coupling1,7 are preserved in our experimental approach: light triggers conformational changes in Rho that lead to binding of Gt. The complex of Gt and activated Rho shows a very low dissociation rate in the absence of nucleotides, but it rapidly dissociates in the presence of GTP. In addition, the results showed that the system works also in a lipid bilayer-free environment and therefore allowed monitoring of Gt association with Rho without interference of intermediate membrane-binding steps (see also Discussion Section). Dependence of Gt Binding on the Density of the Receptor. A critical parameter in SPR studies is the surface density of the ligand.8 In general, we observed that the amount of immobilized Rho on ConA-coated surfaces was very reproducible if the same ConA surface was used for several cylces of interaction analysis. However, different surface densities of Rho were obtained if we varied the density of immobilized ConA. Alternatively, contact time of Rho with ConA or concentration of Rho during injection was altered. In all tested cases, the immobilization signal never exceeded 6000 RU, corresponding to a receptor density of 8.6 × 1010 Rho molecules/mm2. Figure 3 shows binding of Gt at different densities of immobilized Rho. We varied the Rho density on a chip by over 1 order of magnitude ((0.54-8.05) × 1010 Rho/mm2) and subsequently injected a saturating concentration of Gt (0.7 µM, see below). Binding of Gt was triggered by continuous illumination allowing maximal interaction of Gt with Rho (Figure 3A). The amount of bound Gt increased with the amount of immobilized Rho (Figure 3A). In all cases, the signal reached saturation, indicating that the maximal binding capacity for a certain Rho density was attained. The relation between the amplitude of the maximum Gt binding and density of immobilized Rho was linear (Figure 3B), suggesting that the receptor was immobilized on the

Figure 3. Binding of Gt to Rho at different receptor densities. (A) Sensorgrams of light-triggered binding of Gt to Rho that was immobilized on a sensor chip at different densities (as indicated on the right). Concentration of Gt was 0.7 µM in all cases. (B) Plot of the amplitude of Gt binding as a function of immobilized Rho density (shown is one representative data set from a total of four independent experiments).

sensor surface with the same topology and the same accessibility to Gt over the whole range of receptor densities. This assumption is supported by our result (above) that immobilization of Rho occurred with a uniform topology. The stoichiometric ratio between bound Gt and immobilized Rho was between 1/15 and 1/12 and independent of receptor density. Thus, only 6-9% of all hypothetical binding places were occupied or, in other words, a maximum of 6-9% of all immobilized Rho molecules became light-activated. The initial slopes of the sensorgrams in Figure 3 expressed in the change of RU per second directly correspond to molecules of Gt that bind to illuminated Rho (1 RU × s-1 corresponds to the binding of 1 pg of Gt/ s). This initial slope increased with increasing Rho density on the chip surface, but it was nearly constant if the slope was normalized to the number of Rho molecules, yielding a mean value of (6.3 ( 1.0) × 10-4 Gt/Rho × s. Thus, one Gt molecule encounters 1 Rho/s out of approximately 1600, irrespective of the receptor density. Activation of Gt by Surface-Immobilized Rho Depends on Light. The rate of Gt binding to the surface was directly dependent on light intensity. Light intensity was attenuated by neutral density filters of known cutoff, leading to a 10- or 200-fold reduction of light intensity (see the bar in Figure 4). As it is apparent from Figure 4, the increase in light intensity led to elevation of Gt association, indicating direct dependence of Gt binding rate on the amount of Rho activation on a sensor chip surface. Analytical Chemistry, Vol. 78, No. 4, February 15, 2006

1231

Figure 4. Kinetics of Gt binding at different light intensities. Immobilized Rho was illuminated by laser light of different intensities. Neutral density filters were used to attenuate laser light intensity to 10 and 0.5% of maximal intensity (as indicated in the upper bar). The sensorgram was recorded with 0.7 µM Gt and a surface density of Rho of 3.64 × 1010 molecules/mm2.

Figure 6. Affinity of the Gt/Rho complex. Overlay of sensorgrams of Gt binding to Rho recorded with varying concentrations of Gt (0.0175-0.7 µM). Binding of Gt was triggered by a 2-s light pulse at time t0 ) 0. Surface density of Rho was 2.88 × 1010 molecules/mm2 in all cases.

Figure 5. Titration of Gt binding by light. (A) Light titration sensorgram of Gt binding (0.7 µM) to immobilized Rho (4.08 × 1010 molecules/mm2), which was illuminated by laser light of different durations (1, 2, 4, 6, 8, or 10 s). (B) Plot of the amplitude of Gt binding as a function of accumulating total time of illumination. The punctuated curve indicates the deviation from the linear relationship (analysis of two data sets).

over this period of illumination time. At longer illumination levels, the system finally reached saturation, as can be seen in a decrease of the initial slope (last flash of 10 s duration in Figure 5A) and a deviation from the linear slope above 20 s of total illumination time in Figure 5B. Multiple Gt Binding on the Same Surface of Immobilized Rho. In the light titration experiment in Figure 5, we showed that at a low-light activating level (