EP1295120A2 - Agonist versus antagonist binding to g protein-coupled receptors - Google Patents

Abstract

A method of characterizing the biophysical properties of G protein-coupled receptors in response to binding by ligands. The clone human δopioid receptor immobilized in a solid-supported lipid bilayer was investigated by a method featuring coupled plasmon-waveguide resonance (CPWR) spectroscopy. The invention offers a highly sensitive method that directly monitors mass density, conformation, and molecular orientation changes occurring in anisotropic thin films, and allows direct determination of binding constants. Although both agonist and antagonist binding to the receptor cause increases in molecular ordering within the proteolipid membrane, only agonist binding induces an increase in thickness and molecular packing density of the membrane (10). This provides a method of discriminating between agonist and antagonist binding.

Description

AGONIST VERSUS ANTAGONIST BINDING TO G PROTEIN-COUPLED RECEPTORS

U.S. GOVERNMENT RIGHTS This invention was supported in part by grants from the National Science Foundation ( CB-9904753) , and the U.S. Public Health Service, National Institute of Drug Abuse (DA-06284) . The Government has certain rights in the invention.

CROSS-REFERENCE TO RELATED APPLICATION This application is based on U.S. Provisional Application No. 60/213,575, filed on 06/22/2000, and entitled "Plasmon Resonance Studies of Agonist/Antagonist Binding to the Human Delta-Opioid Receptor: New Structural Insights into Receptor-Ligand Interactions."

BACKGROUND OF THE INVENTION

Field of the Invention This invention pertains in general to the field of surface plasmon resonance (SPR) spectroscopy. In particular, the invention relates to a novel SPR approach involving the coupling of plasmon resonances in a thin metal film and the waveguide modes in a dielectric overcoating and the use of such coupled plasmon-waveguide resonance (CPWR) spectroscopy to study structural changes accompanying the binding of ligands to G protein-coupled receptors immobilized in a solid-supported lipid bilayer.

Description of the Related Art

Most ligands responsible for cell-cell signaling (including neurotransmitters, peptide hormones, and growth factors) bind to receptors on the surface of their target cells. Thus, deciphering the mechanisms by which cell- surface receptors and their ligands mediate signaling remains an important focus of study in biology. The majority of transmembrane signal transduction responses to hormones, neurotransmitters, phospholipids, photons, odorants, and growth factors are mediated by a superfamily (containing nearly 2,000 members and growing) of seven transmembrane helical G protein-coupled receptors (GPCR) . Activation of these receptors is assumed to require protein conformational changes, which are induced by the binding of ligands to the receptor' s extracellular domain. Subsequently, an associated G protein separates from the receptor and conveys a signal to an intracellular target, such as an enzyme or ion channel .

Current methods used to examine ligand-binding interactions with GPCRs, as well as with other membrane- bound receptors, suffer from several deficiencies. These include the use of radiolabeled ligands, which require special synthetic methodologies and present special disposal and potential toxicity problems. In some cases, ligands with fluorescent probes can be used, but the modification of the ligand by the fluorophore often leads to changes in the binding and other physical/chemical properties of the ligand. Perhaps most importantly, current binding methods, whether using radiolabeled or fluorescent-labeled ligands, provide no information regarding the changes in receptor structure that accompany ligand-receptor interactions, nor do they distinguish the different structural changes that occur for agonists and antagonists interacting with the same receptor.

Thus, there remains a need in the art for new and improved ways of characterizing the biophysical properties of ligand-GPCR interactions. As described herein, a new method employing plasmon resonance spectroscopy is utilized to characterize the binding interactions of peptide ligands with a GPCR. The information thereby obtained includes the direct determination of the thermodynamic binding constant for the non-covalent ligand-receptor interaction, and an assessment of the structural changes that accompany this interaction, all in a single highly sensitive measurement using unmodified materials.

SUMMARY OF THE INVENTION

The invention meets the aforementioned need by providing a new and improved method of characterizing the biophysical properties of GPCR-ligand interactions. In general, the inventive method utilizes a newly developed variant of the surface plasmon resonance (SPR) technique referred to as coupled plasmon-waveguide resonance (CPWR) spectroscopy (Salamon et al., Biophys . J. 73, 2791-2797, 1997; Salamon et al. Trends Biochem. Sci . 24:213-219, 1999; Salamon and Tollin, Encyclopedia of Spectroscopy and Spectrometry, Vol. 3 J.C. Lindon, et al . Eds. Academic Press, San Diego, pp. 2311-2319, 1999; Encyclopedia of Analytical Chemistry R.A. Meyers, Ed. Wiley, New York, 2000; See also U.S. Patents 5,521,702 and 5,991,488 both issued to Salamon et al . ) that allows the characterization of anisotropic membrane systems (Salamon et al., Biophys. J. 73, 2791-2797, 1997; Biophys . J. 75:1874-1885, 1998; Trends Biochem. Sci . 24:213-219, 1999), as well as other anisotropic nanostructures (Salamon and Tollin,

Encyclopedia of Spectroscopy and Spectrometry, Vol. 3 J.C. Lindon, et al . Eds. Academic Press, San Diego, pp. 2294- 2302, 1999; Encyclopedia of Analytical Chemistry R.A. Meyers, Ed. Wiley, New York, 2000) .

An object of the invention is to provide a method of characterizing the biophysical properties of G protein- coupled receptors, and their interactions with ligands, that is more rapid and direct then existing methodologies,

A second object of the invention is to provide a highly sensitive method of characterizing the biophysical properties of G protein-coupled receptors and their interactions with ligands.

A third object of the invention is to provide a method of characterizing the biophysical properties of G protein- coupled receptors, and their interactions with ligands, that does not produce toxic or radioactive waste products.

A fourth object of the invention is to provide a method of characterizing the biophysical properties of G protein- coupled receptors, and their interactions with ligands, that does not modify the physical or chemical properties of the molecules being characterized.

A fifth object is to proyide a method of distinguishing between agonist and antagonist ligands of G protein- coupled receptors.

In accordance with these and other objects, the inventive method has the unique capability of independently examining real-time changes in the structure of a GPCR, both parallel and perpendicular to the lipid membrane plane, in response to receptor-ligand interactions. The method also provides greatly enhanced sensitivity and spectral resolution compared to conventional SPR. For example, only femtomole amounts of receptor (and ligand) are needed for complete spectral determination and analysis. Furthermore, since radioactivity measurements do not need to be performed, the methodology is much more rapid and direct in the determination of critical binding parameters .

The invention thus provides a general procedure that can replace previous methods in characterizing ligand-GPCR interactions, and which at the same time can provide new information about ligand-GPCR structural transitions that are not available using prior methodologies.

The inventive method is illustrated herein via' incorporation of the human δ-opioid receptor into a preformed lipid bilayer, examination of the binding to the receptor of the highly selective ligand DPDPE (c-[D-Pen², D-Pen⁵] enkephalin; (Mosberg et al., PNAS 80:5871-5874, 1983) , demonstration of the reversal of binding using the selective antagonist naltrindol (NTI; Raynor et al., Mol . Pharmacol . 45:330-334, 1994; Korlipara et.al., J". Med. Chem. 38:1337-1343, 1995),, and evaluation of the changes in the receptor structure which accompany, these binding interactions . Significantly different structural changes are shown to be induced in the δ-opioid receptor upon binding with either DPDPE or NTI, thereby providing new insights into the structural basis of receptor function.

Various other purposes and advantages of the invention will become clear f om its description inj the specification that follows, and from the novel features particularly pointed out in the appended claims . Therefore, to the accomplishment of the objectives described above, this invention consists of the features hereinafter illustrated in the drawings, fully described in the detailed description of the preferred embodiments, and particularly pointed put in the claims. However, such drawings and description disclose only some of the various ways in which the invention may be practiced.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1: CPWR spectra obtained for a supported lipid bilayer containing 75 mol% egg phosphatidylcholine and 25 mol% phosphatidylglycerol before (curve 1) and after the addition of aliquots of human δ-opioid receptor in octyl glucoside-containing buffer into the aqueous compartment of the CPWR cell (final receptor concentration in the bulk solution for curve 2 is 4.8 nM and for curve 3 is 12.8 nM) . The buffer composition was 10 mM Tris (pH 7.3), 0.5 ttiM EDTA and 10 mM KC1. The octyl glucoside concentration in the receptor solution was 30 mM; after dilution into the sample cell the concentration ranged from 0 to 5 mM. Data obtained with p-polarized (panel A) and s-polarized (panel B) exciting light are shown. Dotted lines represent theoretical fits. The refractive index and thickness values obtained from these fits are given in Fig. 4. In all cases, CPWR spectra were obtained after equilibrium was reached (20-40 minutes) .

Figs. 2A and 2B: CPWR spectra obtained in an experiment in which addition into the aqueous compartment of agonist (DPDPE) is followed by addition of antagonist (NTI) . The cell contained a lipid membrane with the receptor incorporated (final bulk receptor concentration was 12.8 nM; spectra shown in curves 1 in panels A and B of Fig. 2A) . Curves 2 in panels A and B show the spectra obtained after addition of 79 nM DPDPE. Curve 1 in panels C and D (Fig. 2B) is the same as curve 2 in panels A and B. Curves 2 in panels C and D show the spectra obtained after addition of 0.64 nM NTI. Other conditions as in Fig. 1.

Fig. 3A and 3B: CPWR spectra obtained in experiment in which addition of antagonist is followed by addition of agonist. The cell contained a lipid membrane containing the receptor, as in Fig. 2 , (final bulk concentration was 12.8 nM; spectra shown in curves 1, panels A and B of Fig. 3A) . Curves 2 in panels A and B show the spectra obtained after addition of 0.144 nM NTI. This was followed by addition of 360 nM DPDPE (curve 2; panels C and D of Fig. 3B) . Curve 1 in panels C and D is the same as curve 2 in panels A and B. Other conditions as in Fig. l.

Fig. 4: Dependence of the relative position of the CPWR resonance minimum on the agonist (DPDPE) and antagonist (NTI) concentrations in the sample compartment of the cell, obtained using either p- (circles) or ≤r-polarization (triangles) . Resonance position displacement towards higher values represents shifts to larger, angles of incidence. Results were obtained in a continuation of the experiment described in Fig. 2. After receptor incorporation, aliquots of the agonist solution in buffer were added; this was followed by addition of aliquots of the antagonist solution. Other conditions as in Fig. 1.

Figure 5: Dependence of the relative position of the CPWR resonance minimum on the antagonist (NTI) and agonist (DPDPE) concentration in the sample compartment of the cell, obtained using either p- (circles) or s-polarization (triangles) . Resonance position displacement towards higher values represents shifts to larger angles of incidence. Results were obtained in a continuation of the experiment described in Fig. 3. After receptor incorporation, aliquots of antagonist solution in buffer were added; this was followed by addition of agonist solution. Conditions as in Fig. 1.

Figure 6: CPWR time-resolved spectra obtained with s- polarized light, using a lipid membrane containing receptor (final bulk concentration 12.8 nM) as described in Fig. 1 (curve 1) . Curves 1 and 2 were obtained 20 s and 60 s after addition of 7 nM agonist to the CPWR sample cell, respectively. Insert shows the resonance position shift as a function of time after addition of agonist.

Figure 7: Refractive index (panel A) and thickness (panel B; triangles) values of the proteolipid film as a function of the receptor concentration, obtained from theoretical fits as described in Fig.l. Data in panel A obtained using either p- (circles) or s-polarized (triangles) light (error bars due to uncertainties in curve fitting lie within symbols) . Panel B also shows the refractive index anisotropy (ϋ^; circles,) as a function of the receptor concentration. Solid lines in both panels represent nonlinear least squares fits to a hyperbolic function. These yield limiting values of both n (given in Figure) and t (6.8 nm) extrapolated to infinite concentration of the receptor, as well as an apparent binding constant K_D (given in Figure) .

Figure 8: Average change in thickness (circles) and refractive index anisotropy values (triangles) for a proteolipid membrane containing opioid receptor (bulk concentration 12.8 nM) as a function of the agonist and antagonist concentration. Results obtained from the experiment in Fig. 4 by theoretical fits to the experimental spectra (see Figs. 1 and 7). Insert shows the square of the average refractive index of the proteolipid membrane as a function of agonist concentration. Solid lines represent nonlinear least squares fits to a hyperbolic function from which the binding constant values (K_D) were obtained. For purposes of clarity, error bars (corresponding to curve fitting errors) for circles and triangles shown only for one of the two curves in the main panel .

Figure 9: Average change in thickness (circles) and refractive index anisotropy values (triangles) as a function of antagonist and agonist concentration. Results obtained from the experiment in Fig. 5 by theoretical fits to the experimental spectra. Other details as in Fig. 8.

Figure 10: Schematic representation of changes in conformation (evaluated by refractive index anisotropy and membrane thickness values) and mass distribution (evaluated by membrane thickness and average refractive index values) of lipid and receptor molecules during interaction of the receptor with either agonist or antagonist molecules. For clarity, only four of the transmembrane helices and the extramembrane loops of the receptor are shown. Structural transitions occurring upon adding agonist subsequent to antagonist addition, and antagonist subsequent to agonist addition are also shown. See text for further description. DESCRIPTION OF THE PREFERRED EMBODIMENTS OF THE INVENTION

The invention generally involves the application of coupled plasmon wavelength resonance technology in a method of characterizing the biophysical properties of membrane-bound G protein-coupled receptors. To illustrate the invention, the human δ-opioid receptor was incorporated into a pre-formed lipid bilayer and processed as described hereinafter. However, this illustration is not intended to limit the method of the invention to a particular GPCR.

Purification of the Receptor .

The human brain δ-opioid receptor, accession number U07882 (Knapp et al., Life Sci . 54 :PL463-PL469, 1994), mediates analgesic responses to endogenous enkephalins, as well as to a variety of synthetic agonists . A fully functional receptor, labeled at the C-terminus with a myc epitope (Gimpl et al . , Eur. J. Biochem. 237:768-777, 1996) and His tag (Grisshammer and Tucker, Biochem. J. 317:891-899, 1996) was prepared by inserting the DNA of the human δ- opioid receptor, which was modified by incapacitating the stop codon of the receptor, into the pcDNA3 vector containing the myc/E±s tag (Invitrogen) . The entire vector was verified by DNA sequencing, and stably transfected into a CHO cell line using DEAE-Dextran (Promega) . The transfected clones were selected using G418 as an antibiotic. These were grown to confluency in Hamm' s F12 medium with 10% fetal bovine serum containing penicillin (lOOU/mL) and streptomycin (100 μg/mL) in a humidified C0₂ atmosphere at 37°C. Related experiments characterizing the modified receptor have been carried out (See Okuara et al . , Eur. J. Pharmacol . 387 :R11 -R13, 2000).

After harvesting the cells and washing several times, they were suspended in Tris-Cl buffer at pH=7.4 and centrifuged at 42,000 rp (160,000 X g) at 4°C for 30 minutes. The buffer was decanted and the membranes were solubilized by homogenization in a solution containing 25 mM Hepes, 0.5 M KC1, 30 mM octylglucoside and protease inhibitors designed to be used with metal chelating columns (Sigma) , buffered at pH = 7.4. After homogenization the solution was centrifuged at 42,000 rpm again for 60 minutes to remove cell debris .

The receptor was purified on a TALON™ Co*² metal chelating column (Clontech) with gentle rocking for 48 hours at 12°C and eluted with 25 mM Hepes, 0.5 M KC1, 30 mM octyglucoside and 100 mM imidazole buffered at pH = 7.4. Although the binding can be carried out in 24 hours, this experiment was allowed to go for 48 hours in order to maximize binding of the receptor to the TALON column. The column and receptor homogenate were kept at 12°C in order to minimize any possible denaturation of the receptor due to heat or protease, which may still be present in the system. The concentration of receptor in the purified sample was determined in a binding assay using a radioactive ligand (Okuara et al . , Eur. J^". Pharmacol . 387:R11-R13, 2000) .

The agonist (DPDPE) used in this work was synthesized in Dr. Victor Hruby' s laboratory (Mosberger et al . , PNAS

80:5871-5874, 1983) and the antagonist (NTI) was obtained from RBI Labs.

Formation of Solid-Supported Lipid Bilayers Self-assembled solid-supported lipid membranes were prepared according to the method used for formation of freely suspended lipid bilayers (Mueller et al., Nature 194:979-980, 1962). This involves spreading a small amount of lipid solution across an orifice in a Teflon sheet that separates the thin dielectric film (Si0₂) from the aqueous phase (Salamon et al., Trends Biochem. Sci . 24:213-219, 1999; See also U.S. Patents 5,521,702 and 5,991,488 both issued to Salamon et al . ) . The hydrophilic surface of hydrated Si0₂ attracts the polar groups of the lipid molecules, thus inducing an initial orientation of the lipid molecules, with the hydrocarbon chains pointing toward the droplet of excess lipid solution. The next steps of bilayer formation, induced by adding aqueous buffer into the sample compartment of the CPWR cell, involve a thinning process and formation of a plateau- Gibbs border of lipid solution that anchors the membrane to the Teflon spacer. In the present experiments, the lipid films were formed from a solution containing 5 mg/mL egg phosphatidylcholine (PC) and l-palmitoyl-2-oleoyl-sn- glycero-3- [phospho-rac- (1-glycerol) (sodium salt) (POPG) (75:25 mol/mol) in squalene/butanol/methanol (0.05:9.5:0.5, v/v) . The lipids were purchased, from

Avanti Polar Lipids Inc. (Birmingham, AL) . All experiments were carried out at ambient temperature using 10 mM Tris buffer containing 0.5 mM EDTA and 10 mM KC1, pH 7.3, in the 2 mL sample cell .

CPWR Spectroscopy

Details of the procedures for CPWR measurement and data analysis have been described elsewhere (Salamon et al . , Biophys. J. 73, 2791-2797, 1997; Trends Biochem. Sci . 24:213-219, 1999; Salamon and Tollin, Encyclopedia of

Spectroscopy and Spectrometry, Vol. 3 J.C. Lindon, et al. Eds. Academic Press, San Diego, pp. 2311-2319, 1999; See also U.S. Patent 5,991,488 issued to Salamon et al.). The method is based upon the resonant excitation by polarized light from a CW He-Ne laser (λ = 632.8 nm) , passing through a glass prism under total internal reflection conditions, of collective electronic oscillations (plasmons) in a thin metal film (Ag) deposited on the external surface of the prism which is overcoated with a dielectric layer (Si0₂) . The resonant excitation of plasmons generates an evanescent electromagnetic field localized at the outer surface of the dielectric film, which can be used to probe the optical properties of molecules immobilized on this surface (for further details see refs . Salamon et al., Biophys. J. 73, 2791-2797,, .1997; Salamon et al. Trends Biochem. Sci . 24:213-219, 1999; Salamon and Tollin, Encyclopedia of Spectroscopy and Spectrometry, Vol. 3 J.C. Lindon, et al . Eds. Academic Press, San Diego, pp. 2311- 2319, 1999; Encyclopedia of Analytical Chemistry R.A. Meyers, Ed. Wiley, New York, 2000) . Resonance is achieved either by varying the incident-light wavelength (λ) , at a fixed incident angle (α) , or by varying α at a fixed λ (in the present experiments the latter protocol was used) . Because the resonance coupling generates electromagnetic waves at the expense of incident light energy, the intensity of totally reflected light is diminished. The reflected light intensity as a function of either λ or α results in a CPWR resonance spectrum.. The resonance can be excited with light polarized either parallel (p) or perpendicular (s) to the incident plane, resulting in two well separated spectra (Salamon et al . , Biophys. J. 73:2791-2797, 1997), thereby allowing characterization of the molecular organization of anisotropic systems such as biomembranes containing integral proteins (Salamon et al . , Trends Biochem. Sci . 24:213-219, 1999 and references cited therein). Under the experimental conditions employed in this work, the optical parameters obtained with p-polarization refer to the perpendicular direction, and with s-polarization to the parallel direction, relative to the bilayer membrane surface .

CPWR spectra can be described by three parameters: α (or λ) ; the spectral width; and the resonance depth. These depend on the refractive index (n) , the extinction coefficient (k) and the thickness (t) of the plasmon- generating and emerging media, the latter including a thin film deposited on the silica surface (i.e. a proteolipid membrane in the present case) in contact with an aqueous solution. Thin-film electromagnetic theory based on Maxwell' s equations provides an analytical relationship between the spectral parameters and the optical properties of these media. This allows evaluation of n, k and t uniquely for the three media (i.e., the plasmon-generating medium, the proteolipid membrane and the aqueous buffer solution) , by non-linear least-squares fitting of the theoretical spectrum to the experimental one (for details see ref . Salamon et al . , Trends Biochem. Sci . 24:213-219, i999; Biophys . J. 78:1400-1412, 2000; Salamon and Tollin, Encyclopedia of Spectroscopy and Spectrometry, vol . 3. J.C. Lindon et al. Eds. Academic Press, San Diego, pp. 2294-2302, 1999) . Inasmuch as the excitation wavelength (632.8 nm) is far removed from the absorption bands of the lipids, protein and ligands used in this work, a k value other than zero reflects a decrease of reflected light intensity due only to scattering resulting from imperfections in the proteolipid film.

It is important to point out that for an anisotropic thin film, such as the proteolipid membrane in the present , work, the thickness (t) represents an average molecular length perpendicular to the plane of the film, and will be

I' independent of light polarization. In contrast, the values of the refractive index (n) will be very much dependent on the polarization of the excitation light. Furthermore, for uniaxial anisotropic structures in which the optical axis is parallel to the p-polarization direction, the n^ value will always be larger than n^. This is a consequence of the fact that the measured refractive index of a material is determined by the polarizability of the individual molecules. The latter property describes the ability of a molecule to interact with an external electromagnetic field, and in general is anisotropic with respect to the molecular frame. In the simplified case in which the molecular shape is rod-like (e.g., the phospholipid molecules used in this work), one can assign two different values to the polarizability: the larger one, longitudinal and the smaller one, transverse.

If in addition to the anisotropy in molecular shape and polarizability, the system which contains these molecules is ordered such that the long axis of the molecules are parallel, this results in long range order usually described by the order parameter S. In this situation the values of the polarizability, averaged over the. whole system and measured either parallel or perpendicular to the direction of the long axis of the molecules, will be different (i.e., the parallel value will be larger than the perpendicular one) . These conditions create an optically anisotropic system, with the optical axis perpendicular to the plane of the proteolipid membrane, and the values of the refractive index measured with two polarizations of light (i.e. parallel, n^, and perpendicular, n_a to the optical axis) will describe this optical anisotropy (A as follows: , = (n - n ) / (n^ + 2) (1)

In this equation ^ is the average value of the refractive index, and for a uniaxial system is given by: n²-_^ = 1/3 (n + 2n\)

(2)

In summary, the anisotropy in the refractive index reflects both the anisotropy in the molecular polarizability and the degree of long range order of molecules within the system, and therefore can be used as a tool to analyze structural organization (i.e., molecular orientation) . This is particularly important in the context of the present work in which structural alterations of a proteolipid membrane, consisting of a single lipid bilayer with inserted receptor molecules, caused by ligand binding have been monitored by changes in the refractive index anisotropy.

Furthermore, as can be seen from the Lorentz-Lorenz relation, the average value of the refractive index is also directly related to the mass density (Born and Wolf, Principles of Optics, Permamom Press, New York, 1965; Cuypers et al . , J. Biol . Chem. 258:2426-2431, 1983).

Thus, from the thickness of the proteolipid film and the average value of the refractive index one can calculate the surface mass density (or molecular packing density) , i.e., mass per unit surface area (or number of moles per unit surface area.

In the present experiments, the plasmon-generating device was calibrated by measuring the CPWR spectra obtained from a bare silica surface in contact with aqueous buffer with both p- and s-polarized light, and then fitting these with theoretical curves. The goal of such a calibration is to obtain the optical parameters of the silica layer (i.e., refractive indices, extinction coefficients and thickness) used in these experiments. This provides an input set of data used in analyzing the resonance spectra obtained with proteolipid membranes deposited on the silica surface. Thus, the resonance spectra obtained after a single lipid bilayer membrane was created on the hydrophilic surface of silica were fit using these data, yielding the optical parameters (JO.,, n_a , and t) for the lipid bilayer. These allowed the calculation of the refractive index anisotropy and the surface mass density (i.e., molecular packing density) of the bilayer. After incorporation of the receptor molecules into the lipid membrane, the resulting CPWR spectra allowed the characterization the, structural consequences of receptor incorporation. Finally, addition into the aqueous sample compartment of the CPWR cell of either agonist or antagonist again resulted in changes of the CPWR spectra, which reflected structural alterations in the proteolipid membrane caused by the, receptor-ligand interaction.

Incorporation of the δ-Opioid Receptor into a Pre-formed Lipid Bilayer

Receptor molecules were incorporated into a preformed lipid membrane deposited on the hydrophilic surface of the silica film by adding small aliquots of a concentrated solution of the human δ-opioid receptor splubilized in 30 mM octyl glucoside to the aqueous compartment of the CPWR cell, thereby diluting the detergent to a final concentration below its critical micelle concentration (25 mM) (Salamon et al., Biochemistry 33 : 13706-13711, 1994; Biophy. J. 71:283-294, 1996) . This resulted in spontaneous transfer of the receptor from the micelle to the lipid bilayer. The overall orientation of the receptor in the bilayer is not known. However, as will be shown below, ligand binding to the incorporated receptor occurs efficiently, so that one can presume that at least 50% of the receptors are bound with the liganding site facing the aqueous buffer.

Fig. 1 shows typical CPWR spectra, obtained with either p- (panel A) or s-polarized exciting light (panel B) , for a solid-supported lipid membrane prior to (curve 1) and after two additions of detergent-solubilized receptor to the aqueous compartment of the sample cell (curves 2 and 3) . As noted previously with other integral membrane proteins including rhodopsin (Salamon et al . , Biochemistry 33:13706-13711, 1994; Biophys . J. 71:283-294, 1996), protein incorporation into the bilayer influences all three parameters of the resonance spectrum, i.e., angular position, depth and spectral half-width. Such changes are due both to mass density changes and to structural alterations of the proteolipid membrane (reflected in changes in refractive index and thickness) . These will be considered further below.

Binding of Agonist (DPDPE) and Antagonist (NTI) to Incorporated Receptor

This section describes the primary spectral data obtained upon adding DPDPE and NTI to the previously incorporated receptor. As will be demonstrated, these data clearly reveal different patterns of receptor-agonist and receptor-antagonist interaction, thereby providing a method of distinguishing agonist and antagonist binding.

When aliquots of either DPDPE or NTI solutions are added to the sample cell after receptor incorporation into a pre-formed bilayer, significant changes in the position, width and depth of the CPWR resonance curve occur. These spectral changes reflect the binding of these molecules to the proteolipid membrane. Control experiments involving addition of comparable amounts of these ligands to a CPWR cell containing a preformed bilayer in the absence of receptor produced no measurable effects on the CPWR spectra (data not shown) , indicating that non-specific binding to the membrane is not detected in these experiments. Thus, the spectral changes observed when the receptor is present must reflect receptor-ligand interactions .

In order to illustrate these changes, examples of resonance spectra obtained with both p- a d s-polarized exciting light are shown in Figs. 2A and 2B and 3A and 3B. Figs. 2A and 2B show the results of an experiment in which agonist is added to the receptor-containing CPWR cell first, followed by antagonist addition, and Figs. 3A and 3B an experiment in which antagonist is added first followed by agonist. As can clearly be seen, the effects of these two ligands on the resonance spectra are easily measurable and quite different. Although all three spectral parameters (i.e., position, width and depth) are significantly altered by both ligands, appreciable differences are seen in the amplitude and direction of the resonance shifts. Thus, DPDPE causes much larger changes in both p- and s-polarized spectra (compare panels A and B of Fig. 2A with panels A and B of Fig. 3A) than are induced by NTI . In addition, DPDPE shift.s both resonances to larger incident angle values (see Figs. 2A and 3B) , although the change in the p-polarized signal is quite small (see Fig. 5) . In contrast, NTI moves the p- polarized resonance to larger (see panel C of Fig. 2B and panel A of Fig. 3A) , and the s-polarized resonance to smaller, incident angles (see Figs. 2B and 3A) .

To further illustrate these differences, plots of the resonance position shifts as a function of added concentration of the two,ligands are shown in Figs. 4 and 5. These data also illustrate the fact that adding antagonist after agonist does not simply reverse the changes generated by agonist binding (see Fig. 4) . In contrast, agonist added after antagonist is bound is able to reverse the changes caused by the receptor-antagonist interaction (as can be clearly seen in the s-polarized component in Fig. 5) .

It is essential to emphasize that these spectral changes saturate within concentration ranges (0-40 nM for DPDPE; see Fig. 4, and 0-0.1 nM for NTI; see Fig. 5) that are consistent with literature data for the binding characteristics of the ligands (see discussion below) . Thus, it is very unlikely that such high binding affinities result from non-specific receptor-ligand interactions. Furthermore, the results presented in Figs. 4 and 5 also clearly indicate that, although the direction of the shifts remains the same regardless of which ligand is added first, the concentration ranges in which the resonance shifts occur depend on the sequence of addition (compare Figs. 4 and 5). Thus, in the experiments in which agonist is added first, the antagonist concentration range is significantly higher than that for the opposite case. The same observation applies to the agonist when the antagonist is added first .

Preliminary time-resolved measurements of, the CPWR spectra following ligand addition demonstrate quite different kinetic properties depending on which ligand is interacting with the receptor. Fig. 6 shows an example of such a time-dependent spectral sequence obtained with DPDPE using s-polarized light. There are two significant features of these spectral changes which distinguish the receptor-agonist interaction from that of the receptor-antagonist interaction. First, agonist addition results in a very slow (on the order of minutes) time course of spectral changes, whereas antagonist addition results in spectral changes that occur faster than the- resolution time (about 10s) of the present experiments.

Second, the kinetic properties of the spectral alterations observed with the agonist are quite complicated, involving negative shifts followed by positive shifts in an overall multiphasic process (which we have not characterized in full detail) . Such results indicate a complex process of receptor-ligand interaction. It is important to note that a similar complex pattern of spectral changes is observed with the p-polarized component (data not shown) .

The above-noted differences between agonist and antagonist binding properties cannot be explained simply by differences in either the adsorbed mass of the ligand or its rate of diffusion to the receptor, inasmuch as these ligands have very comparable molecular masses (i.e., 648 for DPDPE and 414 for NTI) . Furthermore, preliminary experiments using another highly selective δ-opioid agonist, deltorphin II (Tyr-D-Ala-Phe-Glu-Val-Val-Gly-NH₂; data not shown) , reveal a similar kinetic pattern as observed with DPDPE.

It is also important to note that the present data with the δ-opioid receptor have striking parallels to recent studies with the β₂ adrenergic receptor, in which fluorescence spectroscopy, was used to delineate structural changes associated with receptor-ligand interaction (Gether et al., EMBO J. 16:6737-6747, 1997). In these experiments the time course of fluorescence clearly demonstrated that the kinetics of the receptor-agonist interaction are very comparable to those observed in the present study (Fig. 6, insert) , showing slow (on the order of minutes) multiphasic kinetics, whereas the receptor- antagonist interaction is much faster and simpler.

Although it is clear that further time-resolved studies are necessary to fully understand the complexity of the receptor-agonist interaction process, it is possible to conclude from the present data that the interaction of the δ-opioid receptor with agonist or antagonist generates different structural states of the proteolipid membrane, whose properties depend on the sequence of ligand addition. In order to provide a quantitative description of such states it is necessary to analyze the spectral changes in more detail, taking into account alterations of all three spectral parameters (i.e., resonance position, depth and width) . Such an analysis (see next section) yields the optical parameters of the system, which can be used in a quantitative characterization of the receptor- ligand binding processes .

Characterization of the Receptor-containing Lipid Membrane Quantitative analysis of the plasmon resonance spectra obtained during receptor incorporation can be accomplished by fitting theoretical curves to the experimental spectra (see Fig. 1) . Fig. 7 shows plots of the optical parameters obtained from .such a procedure (n [panel A] ; t and [panel B] ; see Methods section for parameter definitions) as a function of added receptor. The solid lines are single hyperbolic curves fitted to the data points. These results indicate the following: 1) The process of receptor incorporation is satisfactorily fit by a simple Langmuir isotherm; 2) The low value of the apparent insertion constant (« 14 nM) argues for a quite high efficiency of incorporation; 3) The extrapolated thickness value (6.8 nm) describes the dimension of the incorporated protein molecule perpendicular to the membrane plane (i.e., the distance between the external loops plus bound water on both sides of the membrane) . Extrapolation of the refractive index curves to infinite receptor concentration (Fig. 7A) results in values (n_Ei_and Ai that characterize a monolayer of densely packed receptor molecules . From these one can calculate an average value of the refractive index (from equation [2] above) , and then mass density or surface concentration- (Salamon et al . , Trends Biochem. Sci . 24:213-219, 1999). From the latter value and the molecular weight of the receptor (using M_r = 60 kD) , the surface area occupied by one receptor molecule can be evaluated as S_rβ0 = 1200±100 A².

It is important to note that this value for the opioid receptor is in very good agreement with reported values for rhodopsin obtained with several techniques: SPR (1260 A²; Salamon et al . , Biophys . J. 71:283-294, 1996), electron cryomicroscopy (about 1000 A²; Schertler et al . , 1993; Unger and Schertler, Biophys . J. 68:1776-1786, 1995), and a rhodopsin Langmuir-Blodgett film by x-ray scattering (about 1100 A²; Maxia et al . , Biophys . J. 69:1440-1446, 1995) ; 4) The increase in the proteolipid membrane anisotropy occurring during the process of receptor incorporation (shown in Fig. 7, panel B) , clearly reflects a corresponding increase of the average long-range molecular order in the membrane resulting from receptor- jLipid interactions. _l

Characterization of the Receptor-Ligand Interactions In general, CPWR spectral changes obtained with an optically anisotropic thin proteolipid membrane (i.e., changes in position, depth and width) are the result of both mass density (molecular packing density) and structural alterations occurring within the system. The mass density changes are directly reflected by the average value of the refractive index changes (see equation [2] ) , whereas structural alterations will influence the refractive index anisotropy due to changes in ^"the orientational order of molecules within the membrane. The latter quantity can be measured by changes in the refractive index values obtained with p- and s-polarized light (see equation [1] , and further discussion below) . Distinguishing between these two types of changes is especially important in the case of receptors, in which the receptor-ligand interactions are thought to result in structural alterations.

This distinction can be accomplished according to the invention by fitting theoretical resonance curves to the experimental CPWR spectra. Using the structural parameters obtained for the receptor-containing proteolipid membrane (described in the preceding section) , theoretical resonance spectra are fitted to the experimental curves obtained in both agonist-antagonist and antagonist-agonist experiments (see Figs. 2 - 5). The results, expressed as changes in refractive index anisotropy, A-,, and proteolipid membrane thickness, as a function of added ligand are shown in Figs. 8 and 9, respectively.

As noted above, binding of the two ligands drives the proteolipid membrane system into distinct states characterized by different spectral characteristics (see Figs. 2 - 6) . Based on the results in Figs. 8 and 9, one can conclude that agonist binding (either before or after antagonist binding) causes conformational changes in the receptor molecule which result in net increases of both anisotropy and mass density of the proteolipid system. Mass density increases arje shown by the increased values of both n^ (see insert in Fig. 8) and t (preliminary experiments using another δ-opioid receptor agonist, deltorphin II, showed very similar changes in mass density, A and t upon binding to the receptor; data not shown) .

JEn contrast, antagonist binding to the receptor induces only anisotropy changes in the system (i.e., there are no measurable changes in either n^ or t values) . These conclusions are consistent with the data given in Figs. 2 - 5 and are especially well illustrated by the resonance position shifts shown in ,Fig. 4, in which the agonist induces unidirectional (i.e. both p and s-components shift in the same direction) whereas the antagonist induces bidirectional resonance position shifts. Unidirectional shifts of both spectral components in the agonist case is clear evidence of an increase in both and values (i.e., an increase of the average refractive index value; see equation [2] ) , which occurs as a result of a mass density increase.

In contrast, addition of antagonist either before (Fig. 5) or after (Fig. 4) agonist addition does not result in mass density changes. In the latter case, all the spectral changes are related to structural alterations . Due to the fact that both ligands have comparable molecular masses, these results must be a consequence of addition of lipid mass to the bilayer caused by the structural changes of the receptor upon interaction with the agonist (for further discussion see below) . It is also important to note that the conformational state of the receptor induced by the antagonist has a much higher refractive index anisotropy than that produced by the agonist. This is clearly shown in both types of experiments (see Figs. 8 and 9) . Thus, when the agonist is added before the antagonist, the latter ligand increases the anisotropy to almost double the value produced by the agonist. In contrast, when the agonist is added after the antagonist, the value of A, is decreased to a level comparable with the increase produced by the agonist alone. In general, changes in refractive index anisotropy are produced by alterations in the molecular ordering with respect to the bilayer normal . In the present invention, this must be a consequence of conformational changes in the receptor molecules accompanying ligand binding, i.e., changes in position and orientation of the transmembrane helices involving tilting and rotational movements, as well as movements occurring in the extramembrane loops . Changes in the acyl chai ordering of the lipid molecules induced by these protein structural alterations may also contribute.

In summary, the lack of measurable alterations in mass density or membrane thickness upon antagonist binding clearly implies a critical difference in the conformational changes induced by such binding compared with those induced by the agonist. This distinction is also reflected in the fact that the state of the proteolipid membrane created by the addition of antagonist prior to agonist is different from that created when antagonist is added after agonist addition. These differences arise because the agonist is able to generate structural alterations perpendicular to the plane of the membrane changing its thickness, whereas the antagonist cannot do so. Thus, the antagonist produces two sub-states depending upon whether it is interacting with the unliganded receptor, or with a receptor that has agonist bound to it and therefore has changed its dimensions relative to the membrane normal. Although these two sub-states are characterized by similar optical anisotropies, they have different dimensions and mass density. Since NTI is a pure delta receptor antagonist with no reported partial agonist biological activities (Wild et al . , PNAS 91:12018- 12021, 1994) it is reasonable to conclude that both of these sub-states are inactive in signal transduction.

While it is possible that the short lived state represents the receptor state that could lead to negative intrinsic activity (Costa et al . , Mol . Pharmacol . 41:290-297, 1992) if it were long lived enough, a more likely possibility is that the two sub-states represent non-equilibrium steady states (Kenakin, Drugs 40:666-687, 1990) that are accessible to the receptor, with one of these being only transiently observed when NTI interacts with the delta opioid receptor. In order to obtain further insights into these states, structural changes in the lipid and protein components must be separately determined for both agonist and antagonist binding, and under a wide range of ratios of agonist to antagonist . This can be done using chromophore-labelled lipids, and such experiments are underway.

Thermodynamic values for the individual ligand dissociation constants can easily be evaluated from the hyperbolic fits to the anisotropy changes presented in Figs. 8 and 9. The results are given in Table 1. It is evident that these dissociation constants strongly depend on whether the agonist is present when the antagonist is added, and vice versa. Thus, the presence of the other ligand causes an appreciable shift of K_D to higher values. This observation is especially significant in the present system in which the antagonist has a much higher binding affinity than the agonist (by 2-3 orders of magnitude) . Despite this, when NTI is added after DPDPE, its dissociation constant increases significantly (about 4- fold) . This finding cannot be simply explained by competition between these two ligands. This constitutes another indication that different conformational states are induced by these ligands, which are characterized by different binding constants for the other ligand.

The binding constants determined here for, DPDPE and NTI are similar to those reported in the literature using a variety of δ-opioid receptor membrane preparations and a variety of radiolabeled competitive ligands. For DPDPE, some typical values reported in the literature include 3.3-5.2 nM in several rat, brain membrane preparations (Akiyama et al . , PNAS 82:2543-2547, 1985), 1.2 nM for the receptor cloned into the NG-108-15 cell line (Akiyama et al., PNAS 82:2543-2547, 1985), and 85 nM for the receptor cloned into the CHO cell , line (unpublished data). Thus, the K_D values of 10-40 nm reported here are consistent with those expected for a fully functional receptor. Likewise, the K_D values previously reported (Wild, et al . , PNAS 91:12018-12021, 1994) for NTI (0.9 nM in NG-108-15 cloned receptors, 0.13 nM in mouse brain membranes and 0.15 nM in mouse spinal chord preparations) are consistent with the values of 0.02-0.10 nM reported here.

Structural Basis of Receptor Function The CPWR results presented herein demonstrate the formation of several conformational states of the proteolipid membrane as a consequence of receptor-agonist and receptor-antagonist interactions. In the case of agonist binding, the slow multiphasic kinetics clearly indicate that there are a number of intermediate conformational states involved in the formation of the final activated state, as has been suggested by Gether and obilka, J. Biol . Che . 258:2426-2431, 1998. It is not clear at present whether this final state involves an equilibrium mixture of different conformational forms of the receptor, or preferential formation of one particular receptor structure (Kenakin, Trends Pharmacol . Sci . 16:232-238, 1995) .

In either case, the present method has shown that the receptor-agonist conformation produces an elongation of the receptor molecule (increase in t) , as well as an overall increase in the degree of orientational order of molecules within the membrane (increase in refractive index anisotropy, A . It is reasonable to expect this process to be relatively, slow because it also involves alterations in the lipid phase of the membrane in response to receptor elongation. Based on models for opioid receptor structural changes upon activation (Pogozeva et al., Biophys . J. 75:612-634, 1998; Knapp et al . , FASEB J. 9:516-525, 1995; Gether and Kobilka, J^". Biol . Chem. 273:17979-17982, 1998) derived from studies of rhodopsin (Farrens et al., Science 274:768-770, 1996), bacteriorhodopsin (Luecke et al., Science 286:255-260, 1999) , and the β-adrenergic receptor (Gether and Kobilka, J. Biol . Chem. 273:17979-17982, 1998), we suggest that the elongation process involves tilting and rotation of one or more of the transmembrane helices resulting in vertical movements of the extramembrane loops, and is accompanied by movement of lipid molecules that cause an increase in the positive curvature of the lipid surface.

The increase in curvature also requires the movement of lipid molecules from the plateau-Gibbs border to the bilayer phase, which increases the overall surface mass density of the proteolipid membrane. The anisotropy changes can be ascribed predominantly to orientation changes of the transmembrane helices that influence the ordering of the hydrocarbon chains of lipid molecules, without a significant contribution from the extracellular loops or lipid mass redistribution. In contrast, the binding of antagonist results only in an increase in the refractive index anisotropy, which implies localized alterations occurring within the receptor molecule that are restricted to transmembrane helix and lipid hydrocarbon chain orientation. The schematic model shown in Fig. 10 represents an attempt to visualize the structural consequences of δ-opioid receptor interaction with either agonist or antagonist based on these observations. Such a multi-state model allows a simple explanation of the well-known fact that competitive antagonists, although they occupy the same binding site in the receptor as agonists, do not transduce signals across the proteolipid membrane.

A more complete understanding of the molecular mechanisms of receptor-ligand interactions will require more detailed information about the structural changes induced in the receptor by different classes of ligands. In particular, further time-resolved studies are needed to characterize the sequence of conformational changes associated with the intermediate states that follow ligand binding. It will also be important to increase knowledge of the effects of lipid membrane structure, salt concentration, pH, other ligands such as allosteric effectors, and other proteins (e.g. G-proteins, kinases, etc.) on the formation of the liganded states of the receptor.

The present method has shown that CPWR spectroscopy provides a new and powerful experimental tool for such investigations, for GPCRs, as well as other membrane-bound receptors, enzymes, ion channels. In addition, the method herein described should be readily adaptable to high throughput screening, in view of the minute amounts of receptor and ligand needed for a complete dose-response binding assay and for evaluation of receptor structural changes .

Table 1: Dissociation constant values obtained for DPDPE and NTI obtained from experiments described in Figures 8 and 9.

DPDPE ... NTI

K_D (nM) K_D (nM).

DPDPE added first 10.0 ± 0.4* 0.10 ± 0.01** NTI added first 40.0 ± 0.4*** 0.020 ± 0.005**** *No NTI present.

**0btained in the presence of 80 nM DPDPE. ***Obtained in the presence of 0.14 nM NTI. ****No DPDPE present.

As would be understood by those skilled in the art, any number of functional equivalents may exist in lieu of the preferred embodiment described above. Thus, as will be apparent to those skilled in the art, changes in the details and materials that have been described may be within the principles and scope of the invention illustrated herein and defined in the appended claims.

Accordingly, while the present inventive method has been shown and described in what is believed to be the most practical and preferred embodiment, it is recognized that departures can be made therefrom within the scope of the invention, which is therefore not to be limited to the details disclosed herein but is to be accorde the full scope of the claims so as to embrace any and all equivalent products.

Claims

We claim:

1. A method for distinguishing agonist and antagonist ligands of G protein-coupled receptors, comprising the steps of: (a) incorporating a G protein-coupled receptor into a solid-supported, preformed lipid membrane,

(b) incubating the receptor of step (a) with a ligand, thereby forming a ligand/receptor complex,

(c) determining a coupled plasmon- aveguide resonance spectrum of the ligand/receptor complex of step (b) , and

(d) characterizing a biophysical property of the ligand/receptor complex of step (b) based on the coupled plasmon-waveguide resonance spectrum of step (c) .

2. The method of claim 1, wherein the biophysical property of the ligand/receptor complex of step (d) comprises a measurement of mass density.

3. The method of claim l, wherein the biophysical property of the ligand/receptor complex of step (d) comprises a measurement of conformation.

4. The method of claim 1, wherein the biophysical property of the ligand/receptor complex of step (d) comprises a measurement of molecular orientation changes.

5. The method of claim 1, wherein the biophysical property of the ligand/receptor complex of step (d) comprises a measurement of a binding constant.

6. The method of claim 1, wherein the G protein-coupled receptor is the human delta-opioid receptor.

7. A method for distinguishing agonist and antagonist ligands of G protein-coupled receptors, comprising the steps of: (a) incorporating a G protein-coupled receptor into a solid-supported, preformed lipid membrane,

(b) incubating the receptor of step (a) with a ligand, thereby forming a ligand/receptor complex, (c) determining a coupled plasmon-waveguide resonance spectrum of the ligand/receptor complex of step (b) , and (d) characterizing a thickness and molecular packing density of the membrane of step (a) based on the coupled plasmon-waveguide resonance spectrum of step (c) .

8. The method of claim 7, wherein the G protein-coupled receptor is the human delta-opioid receptor.