Direct Identification of a Peptide Binding Region in the Opioid Receptor-like 1 Receptor by Photoaffinity Labeling with [Bpa10,Tyr14]Nociceptin*

The heptadecapeptide nociceptin, also known as orphanin FQ, is the endogenous agonist of the opioid receptor-like 1 (ORL1) G protein-coupled receptor. An affinity labeling approach has been implemented to probe the interactions of the neuropeptide with the receptor using the photolabile nociceptin derivative, [p-benzoyl-l-Phe10,Tyr14]nociceptin ([Bpa10,Tyr14]noc). In recombinant Chinese hamster ovary cells expressing the human ORL1 receptor, [Bpa10,Tyr14]noc binds the receptor with high affinity (K i ∼0.7 nm) and is as potent as nociceptin in the inhibition of forskolin-induced cAMP synthesis (EC50 ∼0.5 nm). UV irradiation at 365 nm of the complex formed by the ORL1 receptor and radioiodinated [Bpa10,Tyr14]noc results in the irreversible labeling of a glycoprotein of ∼65 kDa, determined by SDS-polyacrylamide gel electrophoresis. Complete digestion of the partially purified 65-kDa complex with kallikrein generates a single labeled fragment (∼6.5 kDa) that is readily cleaved by endoproteinase Glu-C to yield a labeled fragment of ∼3.2 kDa. Kallikrein treatment of the photoaffinity cross-linked Glu295 → Asp mutant receptor also yields a single labeled fragment of ∼6.5 kDa but is resistant to further cleavage by endoproteinase Glu-C. Based upon the expected proteolytic fingerprint of the labeled receptor, the photoreactive region can be identified as ORL1-(296–302; residues Thr-Ala-Val-Ala-Ile-Leu-Arg) spanning the C terminus of extracellular loop 3 and the N terminus of transmembrane helix VII. Molecular modeling of the ORL1 receptor complex with [Bpa10]noc suggests that reaction of the Bpa carbonyl group may occur with the side chain of Ile300 within the experimentally identified photoreactive region.

The heptadecapeptide nociceptin, also known as orphanin FQ, is the endogenous agonist of the opioid receptor-like 1 (ORL1) G protein-coupled receptor. An affinity labeling approach has been implemented to probe the interactions of the neuropeptide with the receptor using the photolabile nociceptin derivative, [p-benzoyl-L-Phe 10 14 ]noc results in the irreversible labeling of a glycoprotein of ϳ65 kDa, determined by SDS-polyacrylamide gel electrophoresis. Complete digestion of the partially purified 65-kDa complex with kallikrein generates a single labeled fragment (ϳ6.5 kDa) that is readily cleaved by endoproteinase Glu-C to yield a labeled fragment of ϳ3.2 kDa. Kallikrein treatment of the photoaffinity cross-linked Glu 295 3 Asp mutant receptor also yields a single labeled fragment of ϳ6.5 kDa but is resistant to further cleavage by endoproteinase Glu-C. Based upon the expected proteolytic fingerprint of the labeled receptor, the photoreactive region can be identified as ORL1-(296 -302; residues Thr-Ala-Val-Ala-Ile-Leu-Arg) spanning the C terminus of extracellular loop 3 and the N terminus of transmembrane helix VII. Molecular modeling of the ORL1 receptor complex with [Bpa 10 ]noc suggests that reaction of the Bpa carbonyl group may occur with the side chain of Ile 300 within the experimentally identified photoreactive region.
Nociceptin (Phe-Gly-Gly-Phe-Thr-Gly-Ala-Arg-Lys-Ser-Ala-Arg-Lys-Leu-Ala-Asn-Gln, noc) 1 (1), a neuropeptide also known as orphanin FQ (2), is the endogenous agonist of the opioid receptor-like 1 (ORL1) receptor, a G protein-coupled receptor whose cDNA had been previously cloned from a human brain stem library (3). Nociceptin bears structural resemblance to opioid peptides, particularly dynorphin A, the presumed natural agonist of the -opioid receptor. The peptide has been variously shown to modulate nociception, locomotion, stress and anxiety, food intake, neuroendocrine secretion, learning and memory, drug addiction, and smooth musculature tone in the cardiovascular system and respiratory, gastrointestinal, and urogenital tracts (see Refs. 4 -7 for reviews). The broad pharmacological spectrum of nociceptin suggests that ORL1 receptor agonists and/or antagonists may find several therapeutic applications, for example as analgesics or nootropic agents (7).
An understanding of the underlying molecular basis of how nociceptin binds and activates the ORL1 receptor is of key importance in the rational design of new pharmacophores. Structure-activity relationship studies (8 -13) have established that nociceptin residues 1-13 are sufficient both for high affinity binding to the ORL1 receptor and biological activity. Screening of a combinatorial library, however, has uncovered hexapeptides that are as potent ORL1 receptor agonists as nociceptin (14). Although these hexapeptides, Ac-Arg-Tyr-Tyr-(Arg/Lys)-(Ile/Trp)-(Arg/Lys)-NH 2 , bear little obvious sequence similarity with nociceptin, the three positively charged amino acid residues may mimic the electrostatic properties of the basic noc- (8 -13) sequence. Consistent with other observations that the second, highly negatively charged exofacial loop of the receptor is required for activation (15,16), these data suggest that the positively charged Arg 8 -Lys-Ser-Ala-Arg-Lys nociceptin core is necessary for biological activity. We have proposed a molecular model of the ORL1 receptor complex with nociceptin (17) in which noc-(8 -13) engages in multiple electrostatic interactions with the second exofacial receptor loop, and the N-terminal Phe 1 -Gly-Gly-Phe tetrapeptide, which closely resembles the Tyr 1 -Gly-Gly-Phe opioid "message" sequence, binds in a structurally conserved transmembrane cavity formed by helices III and V-VII.
Here we have sought to probe the model experimentally by covalent modification of the ORL1 receptor with a synthetic photoactivable analogue of nociceptin, [Bpa 10 ,Tyr 14 ]noc. The Bpa-based photoaffinity labeling approach (18) has been successfully employed to locate binding sites in several peptide receptors, including those for angiotensin II (19), cholecystokinin (20 -23), parathyroid hormone (24 -26), secretin (27,28), substance P (29,30), and vasopressin (31). We report the specific reaction of 125 I-labeled [Bpa 10 ,Tyr 14 ]noc with the ORL1 receptor sequence (Thr 296 -Ala-Val-Ala-Ile-Leu-Arg 302 ), comprising the C terminus of extracellular loop 3 and the N terminus of transmembrane helix VII. Molecular modeling of the [Bpa 10 ]noc⅐ORL1 receptor complex directly based on our nociceptin⅐ORL1 model complex (17) predicts the Bpa side chain to bind in a pocket formed by the second and third extracellular loops. The reactive Bpa ketone oxygen is able to make Van der Waals contact with the side chain of Ile 300 , which lies within the experimentally identified photoreactive region of the receptor. These data lend strong support in favor of the correct positioning and orientation of the nociceptin Ser 10 side chain in the original model and are thus also consistent with the proposed participation of the flanking positively charged side chains of the neuropeptide in a hydrogen bonding network with the negatively charged side chains of the second extracellular loop of the receptor (17).

EXPERIMENTAL PROCEDURES
Reagents-All were of the highest grade available. Fmoc-p-benzoyl-L-phenylalanine was purchased from Bachem.  14 ]Noc was synthesized using standard Fmoc solid phase strategy on Wang resin. The synthetic peptide was Ͼ95% pure, as assessed by semi-preparative, reverse-phase high performance liquid chromatography (RP-HPLC), and had the expected mass, as assessed by electron-spray ionization mass spectrometry. Radioiodination of [Bpa 10 ,Tyr 14 ]noc was performed using immobilized N-chloro-benzenesulfonamide (IODO-BEADS®, Pierce) as oxidizing agent, as originally described by Markwell (32). Typically, two beads were washed in 0.5 ml of phosphate buffer (100 mM, pH 6.8), dried, and incubated for 5 min with 1 mCi of Na 125 I (NEN Life Science Products) in 0.2 ml of phosphate buffer. Ten nmol (10 l) of [Bpa 10 ,Tyr 14 ]noc was added, and the reaction was allowed to proceed for a further 5 min. The solution was removed and loaded onto an HPLC semi-preparative C 18 column. Elution with acetonitrile (25-60%, v/v; 30 ml) in 0.08% TFA was performed at a constant flow rate of 1 ml/min. Fractions of 0.5 ml were collected, and radioactivity was counted in a Cobra Auto-Gamma (Packard) counter. A single symmetrical peak of bound radioactivity, corresponding to the monoiodinated peptide derivative (specific radioactivity, ϳ2,000 Ci/mmol; 1 Ci ϭ 37 GBq) eluted just behind [Bpa 10 ,Tyr 14 ]noc, used as a reference. The radioactive fractions were pooled, evaporated under a nitrogen stream, and brought to the desired concentration with Tris-HCl buffer (50 mM, pH 7.4) supplemented with proteinase-free BSA (0.1 mg/ml).

Recombinant Cells and Crude Membrane Fraction-Recombinant
Chinese hamster ovary (CHO) cells stably expressing the human ORL1 receptor (ϳ20 pmol/mg membrane protein) were engineered and selected for and grown in culture essentially as described previously (3). Mutant ORL1-(Glu 295 3 Asp) was obtained by introduction of the appropriate point mutation in the Bluescript SK ϩ /hORL1 construct using mutated oligonucleotides (Genosys Biotech), the ExSite PCR mutagenesis kit (Stratagene) and Vent polymerase (New England Biolabs), as described previously (33). The mutant construct was subcloned in pRc/CMV (Invitrogen), and the recombinant vector was used to transfect COS-7 cells (see Ref. 3 for details). The cells were harvested, frozen at Ϫ70°C for at least 1 h, and homogenized in 50 mM Tris-HCl, pH 7.4, in a Potter-Elvehjem tissue grinder. The nuclear pellet was discarded by centrifugation at 1,000 ϫ g, and the crude membrane fraction was collected upon centrifugation at 100,000 ϫ g (3).
Ligand  14 ]noc, in 0.5 ml of 50 mM Tris-HCl buffer, pH 7.4, supplemented with proteinase-free BSA (0.1 mg/ml) to avoid tube wall adsorption of the radioligand. Nonspecific binding was determined in the presence of 1 M unlabeled nociceptin. Bound radioligand was collected by filtration on polyethyleneimine-treated glass fiber filters (GF/B; Whatman), and radioactivity was counted in a Packard model 2100TR liquid scintillation analyzer. The competition data were fit to a sigmoidal dose-response curve with variable slope parameter using the Prism program (GraphPad software). Slope factors close to unity indicated homogenous binding site populations in the membrane preparations examined. IC 50 values (the concentration of inhibitor giving half-maximal specific binding of radio-ligand) were converted to K i values using the Cheng  , and 180 l of fresh KRH was added to each tube. Intracellular accumulation of cAMP was initiated by the addition of 100 M forskolin (Sigma), 1 mM 3-isobutyl-1-methylxanthine (Sigma), 1 mM Ro20-1724 (Biomol), and the ligand(s) to be tested at the desired concentration in 20 l of KRH. The reaction was stopped after exactly 10 min at 37°C by the addition of 20 l of HCl 2.2 N and rapid mixing (Vortex). The [ 3 H]cAMP content of each tube was determined by selective batch elution on acidic alumina columns, essentially as described by Alvarez and Daniels (35). The data were fit to a sigmoidal dose-response curve with variable slope parameter using Prism.
Partial Purification of the Labeled Receptor-The labeled membrane suspension was heated at 95°C for 3 min and analyzed by SDS-gel electrophoresis in 1.5-mm 8% acrylamide gels according to Laemmli (36). The gel was then vacuum-dried on filter paper and exposed to x-ray film (Kodak X-Omat), with two intensifying screens, for 18 h at Ϫ70°C. The radioactive band was excised from the dried gel and immersed in 6 ml of 0.1% (w/v) SDS in 100 mM NH 4 HCO 3 buffer (100 mM, pH 8.0). Extraction was performed over 4 days with constant shaking at 4°C. The eluate (6 ml ϩ 10 ml buffer rinse) was filtered through 0.22-m Minisart (Sartorius) membranes, and concentrated to 0.2-0.3 ml by centrifugation in Centricon-20 tubes. Samples were then twice desalted by dilution with 15 ml of water and re-centrifugation. Recovery of radioactivity was in the range 70 -80%. Proteins were precipitated (5 h at Ϫ20°C) in 80% (v/v) acetone, dried, and stored at Ϫ80°C for further use.
N-Glycosidase F Digestion of the Partially Purified Labeled Receptor-The dry acetone precipitate (see above) was dissolved in 25 l of sodium phosphate buffer (50 mM, pH 7.5) containing 1% (v/v) Nonidet P-40, and 500 units of peptide:N-glycosidase F was added. Digestion was allowed to proceed for 5 h at 37°C, at which point a 25-l aliquot of 2ϫ concentrated Laemmli (sample) buffer was added, and the sample was analyzed by SDS-PAGE in 1.5-mm 8% acrylamide gels.
Molecular Modeling of the ORL1 Receptor Complex with [Bpa 10 ]Noc-The non-covalent complex formed by the receptor with [Bpa 10 ]noc was built from the coordinates of our model of the nociceptin⅐ORL1 receptor complex (17). Initial inspection of possible 1 side chain rotamers (see Fig. 1A) revealed torsion angles of Ϫ60°or ϩ60°to result in severe steric clashes with the receptor, and in the latter case with the peptide main chain also. The Bpa 1 side chain torsion angle was therefore set to 180°, compatible with the main chain conformational class at position 10 (see Ref. 17). The complex was then energy-minimized to a final convergence gradient of 0.1 kcal mol Ϫ1 Å Ϫ1 . Some manual intervention and limited reconstruction of the second and third extracellular loops was required to avoid trapping of the complex in high energy local minima. Interactive graphics manipulations were performed using the SYBYL (version 6.5) software package (Tripos Inc.). Energy minimization was carried out using the SYBYL implementation of the Powell torsional gradient algorithm and the AMBER 1986 all-atom force field of Weiner et al. (38), extended as described below to accommodate Bpa. The electrostatic model comprised a distancedependent dielectric constant with a non-bonded cut-off of 8 Å and an ⑀ value of 4. A total of six additional harmonic constraints of 1 kcal mol Ϫ1 were applied to prevent buckling of the Bpa ring systems in response to external forces.
AMBER Force Field Parameters for p-benzoyl-L-phenylalanine-All the Bpa residue atoms can be represented by the AMBER atom types defined in the Weiner et al. (38) force field (see Fig. 1B). Most parameters in the potential function could therefore be abstracted directly from the all-atom protein set. Assignment of the remaining terms was assisted by ab initio quantum mechanical calculations on the model compound, 4-methyl-benzophenone (Fig. 1C) using GAUSSIAN 94 (39). In order for the sum of the three equilibrium bond angle ( 0 ) values used to describe the angular geometry around the ketone carbon to be 360°, 0 for CAϪCϪO was set to 120°, with an associated force constant (K ) of 85 kcal mol Ϫ1 radian Ϫ2 , equal to that for the CAϪCϪCA bond angle. The average value for the two CAϪCϪO angles in the HF/6-31G* geometry-optimized 4-methyl-benzophenone structure was 119.7 Ϯ 0.1°, and the CAϪCϪCA bond angle was 120.6°. An improper dihedral angle term for the CAϪCAϪCAϪC atom quartet was also added to the potential function with V n /2 ϭ 2 kcal mol Ϫ1 , ␥ ϭ 180°, and multiplicity, n ϭ 2. Partial charges for the Bpa residue are presented in Fig. 1B. They were obtained by mapping of charges calculated for 4-methylbenzophenone onto the side chain benzophenone moiety, and re-distribution of the excess charge required for neutrality onto the C ␤ , H ␤1 , and H ␤2 Bpa bridge atoms according to the (C ϩ H): H charge ratio in the 4-methyl group of the model compound. The 4-methyl-benzophenone point charge set was calculated by two-stage restrained electrostatic potential (RESP) fitting (40). The fit to the 6-31G* quantum mechanical potential was performed using default weights of 0.0005 in the first stage and 0.001 in the second stage in which conformationally interchangeable atoms are equivalenced. The resultant partial charges were then scaled by a factor of 0.91 for a more balanced integration with the Weiner et al. (38) charges, derived for the most part using the STO-3G basis set. Conjugate gradient energy minimization of the HF/6-31G* 4-methyl-benzophenone structure to a convergence gradient of Ͻ0.001 kcal mol Ϫ1 Å Ϫ1 was performed as a check on the assigned Bpa force field parameter values. This resulted in a root mean square distance of 0.065 Å for the 14 heavy atoms in the model compound. The angle formed between the two ring systems in the molecular mechanics energyminimized structure, which may be calculated as the sum of 3 and 4 (see Fig. 1A), was 52.8°as compared with 57.3°( 3 ϭ Ϫ152.9°; 4 ϭ Ϫ149.8°) for the HF/6-31G*-optimized geometry.

RESULTS
Photolabile Nociceptin Derivatives-The choice of [Bpa 10 , Tyr 14 ]noc as a photoaffinity probe of the ORL1 receptor was guided by observations that (i) various substitutions at position 10 in nociceptin, such as alanine (8,9) or nitrobenzoxadiazolyl-2,3-Ldiaminopropionic acid, 2 do not significantly affect either binding or biological activity, and (ii) [Tyr 14 ]noc and the mono-iodo-Tyr 14 analogue bind and activate the ORL1 receptor with essentially the same affinity and biological potency as nociceptin (2). Fig. 2A shows that [Bpa 10 ,Tyr 14 ]noc binds the ORL1 receptor with high affinity (K i ϳ0.7 nM). [Bpa 10 ,Tyr 14 ]Noc is also a potent agonist of the ORL1 receptor with an EC 50 value of ϳ0.5 nM for the inhibition of forskolin-induced accumulation of cAMP in intact recombinant CHO[hORL1] cells (Fig. 2B). Likewise, binding of 125 I-labeled [Bpa 10 ,Tyr 14 ]noc to CHO[hORL1] cell membrane preparations displays a hyperbolic concentration dependence, with a K d of 0.15 nM (Fig. 2, C and D).
Photoaffinity Labeling of the ORL1 Receptor-UV irradiation of the pre-equilibrated mixture of radioiodinated [Bpa 10 ,Tyr 14 ]noc and CHO[hORL1] cell membranes, followed by solubilization with SDS under reducing conditions and gel electrophoresis, yields a radioactive macromolecular species that migrates as a diffuse band centered at ϳ65 kDa (Fig. 3). This band is absent when photoactivation is carried out either in the presence of an excess of nociceptin or with membrane preparations from wild-type CHO cells (data not shown), indicating specific irreversible modification of the receptor. Treatment of the photolabeled ORL1 receptor with N-glycosidase F results in a discrete radioactive macromolecular species of ϳ40 kDa (Fig. 3), in good agreement with the value calculated from the amino acid sequence of the ORL1 receptor. The receptor thus appears to be heavily glycosylated in CHO[hORL1] cells, consistent with the presence of three potential N-linked glycosylation sites (Asn-Xaa-Ser) in the N-terminal domain (3).
To identify the covalently modified receptor region, the 65-kDa labeled species was recovered and treated with the specific endoproteinases, tissue kallikrein and endoproteinase Glu-C, 2 J.-L. Butour, unpublished data. both separately and in series. Since nociceptin does not possess cleavage sites for these enzymes, digestion of the cross-linked receptor leads to labeled fragments with molecular mass contributions of ϳ2.1 kDa from the ligand.
Endoproteinase Glu-C Digestion of the Photolabeled hORL1 Receptor-The preferred cleavage site of endoproteinase Glu-C (EC 3.4.21.19) is the scissile bond C-terminal to a glutamate residue. A maximum of 12 possible radioactive species with molecular masses of 2.4, 3.1, Ͼ21.8 (glycosylated), 2.4, 2.2, 2.4, 2.6, 7.8, 6.7, 5.3, 6.6, and 2.7 kDa (see Fig. 5A) can therefore be expected to result from enzymatic digestion of the (reduced) photolabeled ORL1 receptor. Again it is reasonable to assume that the two modified receptor fragments, ORL1-(325-364) (6.6 kDa) and ORL1-(365-370) (2.7 kDa), derived from the C-terminal domain will not be observed. Fig. 5B shows that endoproteinase Glu-C digestion of the (reduced) photolabeled hORL1 receptor, although incomplete, generates at least two radioactive species. The smaller species has an estimated value of ϳ5.5 kDa and could therefore correspond to the ORL1-(296 -324)-modified fragment that has a mass of 5.3 kDa. This fragment embraces the C-terminal part of exofacial loop 3 and transmembrane helix VII of the ORL1 receptor. The larger of the two minor radioactive species has an approximate molecular mass of 10 kDa and could correspond to partial digestion products containing the receptor elements ORL1-(254 -324) (9.9 kDa) or ORL1-(296 -364) (9.8 kDa).
The presumed kallikrein digestion product of the photolabeled ORL1 receptor, ORL1-(260 -302), contains a glutamate residue at position 295. This therefore affords the opportunity of further cleavage by endoproteinase Glu-C to generate one of two possible radioactive modified fragments, with molecular mass values of 5.5 (ORL1-(260 -295)) or 3.1 kDa (ORL1-(296 -302)) (Fig. 6A). Endoproteinase Glu-C digestion of the kallikrein fragment is complete and produces a single radioactive species of ϳ3.2 kDa (Fig. 6B). This result strongly suggests that the photoreactive region in the receptor is indeed hORL1-(296 -302). This short stretch of seven amino acid residues spans the C-terminal section of the third exofacial loop and the N-terminal part of transmembrane helix VII.
Photoaffinity Labeling and Endoproteinase Treatment of the Glu 295 3 Asp Mutant ORL1 Receptor-In order to obtain further evidence that the 6.5-kDa kallikrein fragment (see above) does correspond to the chemically cross-linked ORL1-(260 -302) peptide, the Glu 295 3 Asp mutant ORL1 receptor was constructed and transiently expressed in COS cells. This mutation was expected to result in a kallikrein digestion product that cannot be further cleaved by endoproteinase Glu-C. The mutant receptor binds nociceptin and [Bpa 10 ,Tyr 14 ]noc with the same high affinity (K i ϳ0.4 nM) as the wild-type receptor (data not shown). Photoaffinity labeling with radioiodinated [Bpa 10 ,Tyr 14 ]noc also produces a 65-kDa radioactive species (Fig. 7A). Kallikrein treatment of the partially purified labeled Glu 295 3 Asp mutant receptor generates a single radioactive species with the same mass of ϳ6.5 kDa (Fig. 7B) as that obtained by kallikrein digestion of the wild-type receptor. However, the modified fragment from the mutant is now insensitive to endoproteinase Glu-C treatment, as shown in Fig. 7B. 10 ]Noc-An energy-minimized molecular model of the noncovalent complex formed between [Bpa 10 ]noc and the ORL1 receptor has been constructed using our earlier model of the nociceptin⅐ORL1 receptor complex (17) as a template. The experi-mentally identified photoreactive region (Thr 296 -Arg 302 ) and the Bpa 10 side chain are highlighted in Fig. 8. Setting of the 1 torsion angle in the trans conformation permits the accommodation of the bulky Bpa side chain in a pocket bounded by the second (e2) and third (e3) extracellular loops. This accords with the observed tolerance to serine replacement by Bpa at position 10 in nociceptin, which did not significantly affect either binding or biological activity. Benzophenones are known to react preferentially with C-H bonds (18). Receptor residues with CϪH n atoms falling within a 5-Å radius of the reactive ketone group of the Bpa 10 side chain are documented in Table I. These include Gln 286 at the interface of helix VI and the N terminus of e3, Val 290 in e3, and Ile 300 which lies within the identified photoreactive region. The shortest inter-atomic distance is that between the Bpa 10 ketone oxygen and the C ␦1 atom of Ile 300 , which make Van der Waals contact. The model is thus consistent with the photoaffinity labeling results and suggests that reaction could occur with a side chain methyl of Ile 300 .

Molecular Modeling of ORL1 Receptor Complex with [Bpa
Modeling of the [Bpa 10 ]noc⅐ORL1 complex was achieved with minimal structural disturbance to the starting configurations of either the receptor or ligand. The overall root mean square distance over (277) C ␣ positions in the receptor transmembrane domain was 0.34 Å. Structural change was largely confined to the extracellular loops, and the maximum displacement of any individual C ␣ atom was 1.4 Å, recorded for Ile 204 in e2. The average changes in the previously reported and torsion angle values for nociceptin (17) were 6.1°and 8.5°, respectively. Together with the results of the photoaffinity labeling study, these observations suggest not only that the nociceptin Ser 10 side chain is appropriately orientated in the original nociceptin⅐ORL1 model complex but also that the ligand itself is sunk to a correct depth within the transmembrane bundle. DISCUSSION The present study has made use of the photolabile nociceptin derivative [Bpa 10  duced accumulation of cAMP in recombinant cells expressing the receptor. On this basis it is reasonable to suppose that the photoaffinity ligand adopts a similar binding mode to that taken up by nociceptin. Molecular modeling of the non-covalent complex formed by [Bpa 10 ]noc with the ORL1 receptor indicates that the Bpa side chain, which is significantly bulkier than the Ser 10 side chain in nociceptin, can be accommodated in a pocket bounded by the second and third extracellular loops. The existence and apparent availability of this pocket to large side chains can further explain the observed tolerance to nitrobenzoxadiazolyl-2,3-L-diaminopropionic acid at position 10 in nociceptin. 2 That other small amino acids such as alanine (8,9) can also be accepted at this position suggests that the Ser 10 side chain contributes little to the binding specificity of nociceptin.
The proposed binding mode of [Bpa 10 ]noc with the ORL1 receptor shown in Fig. 8 is entirely consistent with the photoaffinity labeling results. UV irradiation at 365 nm of the complex formed by the ORL1 receptor and radioiodinated [Bpa 10 ,Tyr 14 ]noc results in the irreversible labeling of a glycoprotein of ϳ65 kDa, with approximately 40% of the mass contributed by oligosaccharides. Complete digestion of the partially purified 65-kDa complex by kallikrein generates a single chemically modified fragment (ϳ6.5 kDa). This can be further cleaved by endoproteinase Glu-C to yield a labeled fragment of ϳ3.2 kDa. Based upon the expected proteolytic footprints of the cross-linked receptor (see Figs. 4A, 5A, and 6A), the [Bpa 10 ,Tyr 14 ]noc-reactive region can be deduced to be ORL1-(296 -302). This short segment (Thr 296 -Ala-Val-Ala-Ile-Leu-Arg 302 ) straddles the C terminus of the third extracellular loop and the N terminus of transmembrane helix VII.
Our reasoning is based on the following: (i) the close correspondence of the molecular mass of the kallikrein fragment (ϳ6.5 kDa), determined by SDS-PAGE (Fig. 4), with the calculated mass of the cross-linked ORL1-(260 -302) fragment (6.7 kDa); and (ii) the presence of the Thr 296 -Arg 302 sequence in three cross-linked fragments (ORL1-(296 -324)), ORL1-(254 -324), and ORL1-(296 -364)) compatible with the SDS-PAGE footprint resulting from the partial digestion of the photoaffinity labeled receptor by endoproteinase Glu-C (Fig. 5). However, two other glutamate-containing candidate peptide substrates of endoproteinase Glu-C with molecular masses similar to that determined experimentally may, in principle, result from kallikrein digestion of the photoaffinity labeled receptor. One of these, ORL1-(341-370) (5.4 kDa), can be effectively ruled out since it derives from the C-terminal receptor domain. The presence of the second fragment, ORL1-(303-331), as a potential kallikrein digestion product cannot be immediately discounted in this way. However, further digestion of this cross-linked fragment, which also has a calculated molecular mass of 5.4 kDa, by endoproteinase Glu-C could also yield a 3.1-kDa modified fragment (ORL1-(325-331)) that does derive from a cytoplasmic structural element (i4), assumed to be inaccessible to the site-directed probe. In order to confirm our identification of the photoreactive region as ORL1-(296 -302), we have constructed the Glu 295 3 Asp mutant receptor. Kallikrein treatment of the photoaffinity cross-linked mutant receptor also results in a single labeled fragment (ϳ6.5 kDa) but which can no longer be further cleaved by endoproteinase Glu-C (cf. Figs. 6 and 7).
Examination of the molecular model of the [Bpa 10 ]noc⅐ORL1 complex indicates that reaction of Bpa with the experimentally identified photoreactive region could occur with a side chain methyl of Ile 300 (see Fig. 8). Modeling of the complex, which was based on our previously reported model of the nociceptin⅐ORL1 receptor complex (17), was achieved with minimal structural disturbance to either the receptor or ligand. This together with the results of the photoaffinity labeling study suggests that the nociceptin Ser 10 side chain is appropriately orientated in the original model and that the ligand itself is sunk to a correct depth within the transmembrane bundle. The orientation of the Ser 10 side chain toward the third rather than the second extracellular ORL1 receptor loop facilitates the proposed participation of the flanking neuropeptide Arg 8 -Lys 9 and Arg 12 -Lys 13 side chain pairs in a hydrogen bonding network with acidic side chains of the second extracellular receptor loop (17).
Evidence that interactions of the nociceptin core (Arg 8 -Lys 9 -Ser 10 -Ala 11 -Arg 12 -Lys 13 ) with the second extracellular receptor loop are an absolute requirement for receptor activation derives in the main from studies of -opioid-ORL1 chimeric receptor systems (15,16). In particular, it was demonstrated that high affinity nociceptin binding in the absence of the native second extracellular loop leads only to sluggish receptor activation. A site-directed mutagenesis study has been performed with the intention of disrupting the putative electrostatic interactions of the negatively charged second extracellular ORL1 receptor loop with the positively charged noc (8 -13) core (42). Four residues (Glu 194 -Asp-Glu-Glu 197 ) in the N-terminal sec-  tion of the receptor loop were replaced by their (nominal) structural equivalents (Arg 213 -Gln-Gly-Ser 216 ) in the -opioid receptor. The binding affinity of the mutant ORL1 receptor for nociceptin was decreased by 10-fold, as compared with an observed disproportionate 40-fold loss in biological potency. These data imply that binding interactions involving the second extracellular loop are necessary to stabilize the receptor in the active form but do not contribute significantly to intrinsic cognate agonist binding per se. Indeed, in many cases of biological relevance, desolvation effects may be sufficiently great that electrostatic interactions between oppositely charged groups at solvent-exposed surface positions actually hinder binding (43).
We have recently identified Gln 286 as a crucial switch residue at the membrane interface, involved in the preferential stabilization of the active form of the receptor, again most probably hydrogen bond-mediated (33). This residue is predicted by the model to lie close to nociceptin residue positions 5 and 6, responsible for the selective preference of the hORL1 receptor for nociceptin over dynorphin A (12). The largest free energy ligand binding contributions might therefore be expected to arise from non-polar interactions of noc (1-4) (Phe 1 -Gly-Gly-Phe) within the transmembrane region. Such interactions may also provide an important means of overcoming large anticipated entropy losses upon binding of nociceptin to the receptor. These considerations first led us to propose that the binding of the N-terminal tetra-peptide may serve a mainly locatory function (17). This view receives further support from the identification of a hairpin peptide antagonist of the ORL1 receptor that is also an opioid receptor agonist (44), and which would therefore be expected to bind within the transmembrane region.
In conclusion, the results of the photoaffinity labeling study of the ORL1 receptor provide an anchor point for the interaction with position 10 of nociceptin. The compatibility of the model of the nociceptin⅐ORL1 receptor complex with the sitedirected chemical modification data provides a molecular basis for the identification of interactions that contribute to nociceptin binding specificity by stabilizing the active form of the receptor.