Refinement of the structure of the ligand-occupied cholecystokinin receptor using a photolabile amino-terminal probe.

Affinity labeling is a powerful tool to establish spatial approximations between photolabile residues within a ligand and its receptor. Here, we have utilized a cholecystokinin (CCK) analogue with a photolabile benzoylphenylalanine (Bpa) sited in position 24, adjacent to the pharmacophoric domain of this hormone (positions 27-33). This probe was a fully efficacious agonist that bound to the CCK receptor saturably and with high affinity (K(i) = 8.9 +/- 1.1 nm). It covalently labeled the CCK receptor either within the amino terminus (between Asn(10) and Lys(37)) or within the third extracellular loop (Glu(345)), as demonstrated by proteolytic peptide mapping, deglycosylation, micropurification, and Edman degradation sequencing. Truncation of the receptor to eliminate residues 1-30 had no detrimental effect on CCK binding, stimulated signaling, or affinity labeling through a residue within the pharmacophore (Bpa(29)) but resulted in elimination of the covalent attachment of the Bpa(24) probe to the receptor. Thus, the distal amino terminus of the CCK receptor resides above the docked ligand, compressing the portion of the peptide extending beyond its pharmacophore toward the receptor core. Exposure of wild type and truncated receptor constructs to extracellular trypsin damaged the truncated construct but not the wild type receptor, suggesting that this domain also may play a protective role. Use of these additional insights into molecular approximations provided key constraints for molecular modeling of the peptide-receptor complex, supporting the counterclockwise organization of the transmembrane helical domains.

and glycoproteins can elicit similar conformational changes in the cytosolic face of their receptors (where G protein-coupling occurs) are far from clear. Our best understanding of this process relates to the smallest ligands that appear to bind within the confluence of helices within the lipid bilayer (1), where we have analogous low resolution crystal structures on which to rely (2,3). As the ligands get larger and more structurally complex, the binding domains tend to move toward the extracellular face of the membrane, with extracellular loop and amino-terminal tail domains becoming more important (1,4). These are receptor domains for which we have minimal meaningful structural data.
We have been quite interested in the molecular basis of ligand binding to the type A cholecystokinin (CCK) receptor (5)(6)(7)(8)(9)(10). This receptor is a member of the class I family of G protein-coupled receptors, along with rhodopsin and the ␤-adrenergic receptor (11). CCK occurs as a series of linear peptides, having lengths ranging from 8 to 58 residues (12). These all share their carboxyl-terminal domain, with the carboxylterminal heptapeptide-amide representing the minimal region that has full potency and efficacy for stimulating targets of this hormone. Two molecular approximations have been experimentally determined for residues within this region of CCK and residues within the ligand-binding domain of this receptor (8 -10). Extension or structural modification of the amino terminus of CCK-8 has been well tolerated, without interfering with receptor binding or activation (13,14). One provocative report has recently suggested that the naturally occurring amino-terminal extension that is present in CCK-58 can affect the conformation of the biologically active carboxyl-terminal octapeptide and can have a positive effect on the action of this hormone (15).
With these observations in mind, we initiated the current group of studies. We were particularly interested in the structural details of how an amino-terminal extension from the CCK pharmacophore might be positioned relative to the CCK receptor. This work has given us new insights into the structure and function of the amino-terminal tail of this receptor. Unlike affinity labeling through photolabile residues that are positioned within the pharmacophore of CCK that have yielded only focused sites of covalent attachment to the receptor (8,9), the present work with a photolabile benzophenone residue positioned outside of this domain has resulted in demonstration of the ability to label either of two sites in distinct regions of the CCK receptor. This suggests that this position within the receptor-bound ligand may retain substantial mobility and not be held tightly in a single position relative to the receptor.
One of the sites of covalent labeling was in a position in the amino-terminal tail of the CCK receptor that can be eliminated without having any negative impact on receptor binding or signaling. The other site of labeling was within the third extracellular loop domain. However, by truncation of the region of the receptor amino terminus in which the first contact was present, this photolabile residue no longer came in contact with or labeled the CCK receptor. These observations suggest that the amino-terminal domain of the receptor might provide a protective cover for the peptide-binding domain within the receptor. Indeed, when comparing the sensitivity of wild type and truncated receptor constructs to extracellular protease treatment, we found the latter to be much more amenable to damage by tryptic protease.
The insights coming from the molecular approximations with a photolabile residue sited in position 24 of a CCK-like ligand, when combined with our previous ligand binding and crosslinking data (8 -10), also provided us with the opportunity to propose a distinct topological model for CCK binding to its receptor, having a counterclockwise helix bundle topology. This refined molecular model of the ligand-receptor complex is quite distinct from models recently proposed in the literature based on less direct receptor mutagenesis studies (16 -18). The current model is fully consistent with all existing experimental data and will continue to spawn experimentally testable predictions as it is further refined.
Receptor Preparations-The Chinese hamster ovary cell line that was previously prepared to express the rat type A CCK receptor (CHO-CCKR) (22) was used as a source of wild type receptor for this work. Cells were grown as a monolayer in flasks containing Ham's F-12 medium supplemented with 5% fetal Clone-2 (HyClone Laboratories, Logan, UT) in a humidified environment containing 5% carbon dioxide. Cells were harvested with protease-free cell dissociation medium or mechanically for all described uses. Enriched plasma membranes were prepared by suspending these cells in 0.3 M sucrose containing 0.01% soybean trypsin inhibitor and 1 mM phenylmethylsulfonyl fluoride and sonicating in a Sonifier cell disrupter (Plainview, NY) at setting 7 for 10 s. The sucrose concentration was then adjusted to 1.3 M, placed in the bottom of the tube, and overlayered with 0.3 M sucrose prior to centrifugation at 225,000 ϫ g for 1 h. The membrane band at the sucrose interface was then harvested, diluted with iced-cold water, and pelleted by centrifugation at 225,000 ϫ g for 30 min. Membranes were then resuspended in Krebs-Ringers/HEPES (KRH) medium containing 25 mM HEPES, pH 7.4, 104 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 1.2 mM MgSO 4 , 1 mM KH 2 PO 4 , 0.01% soybean trypsin inhibitor, and 1 mM phenylmethylsulfonyl fluoride and were stored at Ϫ80°C until ready for use.
A CCK receptor construct in which the Val residue at position 342 in the third extracellular loop of the receptor is replaced by Met (V342M) was previously produced and characterized to bind and signal like wild type CCK receptor (9). This cell line was cultured as described above for the CHO-CCKR cell line.
An amino-terminally truncated CCK receptor construct was also prepared. This represented the elimination of receptor residues 1-30. This was accomplished using oligonucleotide-directed mutagenesis, with the wild type rat CCK receptor construct used as template (23). The sequence of the construct was confirmed by direct DNA sequencing (24). A CHO-K1 cell line stably expressing this construct was produced using the same methodology previously utilized (22).
Functional Characterization of the CCK Receptor Probe-The Bpa 24 analogue of CCK was characterized to determine its ability to bind to and to activate the CCK receptor. Binding to the CCK receptor was characterized in a standard competition-binding assay, using conditions that have been previously established (22). This utilized enriched plasma membranes prepared from the CHO-CCKR cells as source of receptor (5-10 g per tube) and the CCK-like radioligand, 125 I-D-Tyr-Gly-[Nle 28,31 ]CCK-(26 -33) (1-2 pM) (14). Incubations were carried out at 25°C for 60 min, conditions adequate to achieve steady-state binding, in the absence and presence of varied concentrations of the Bpa 24 analogue. Rapid separation of bound from free radioligand was accomplished with a Skatron cell harvester (Molecular Devices, Sunnyvale, CA), using receptor-binding filter mats. Bound radioactivity was quantified with a gamma spectrometer. Data were graphed using Prism software (GraphPad software, San Diego, CA) and were analyzed using the nonlinear least squares curve fitting program of Munson and Rodbard (25), LIGAND.
The agonist activity of the probe was studied using a well characterized assay for stimulation of intracellular calcium accumulation in CHO-CCKR cells (22). In this assay, ϳ2 million receptor-bearing cells were loaded with 5 M Fura-2AM in Ham's F-12 medium for 30 min at 37°C. They were then washed and stimulated with varied concentrations of the Bpa 24 analogue at 37°C, with fluorescence quantified in a PerkinElmer Life Sciences LS50B luminescence spectrometer. Excitation was performed at 340 and 380 nm, and emissions were determined at 520 nm, with calcium concentration calculated from the ratios (26). The peak intracellular calcium transient was utilized to determine the agonist concentration dependence of this biological response.
Photoaffinity Labeling of the CCK Receptor-Photoaffinity labeling of the CCK receptor was performed using 75-100 pM radioiodinated CCK analogue probes and 50 -100 g of membrane. This was incubated at 25°C for 60 min in KRH medium in the absence and presence of competing unlabeled peptide. Nonspecific binding was determined in the presence of 1 M CCK. Photolysis was performed in a Rayonet model RP-100 apparatus equipped with 3500-Å lamps at 4°C for 30 min. Bound and free radioligand were separated by centrifugation. Membrane proteins were then solubilized with 1% Nonidet P-40 in KRH medium, prior to wheat germ agglutinin-agarose affinity chromatography and subsequent SDS-polyacrylamide gel electrophoresis using the conditions described by Laemmli (27). After resolution on SDSpolyacrylamide gel electrophoresis, the affinity labeled receptor was visualized by autoradiography, eluted, and lyophilized.
Identification of the Domains of CCK Receptor Labeling-Affinity labeled, lectin-purified receptor and relevant receptor fragments were deglycosylated with endoglycosidase F using conditions we described previously (19).
The initial level of identification of the domains of labeling was performed using CNBr cleavage. This highly efficient cleavage method theoretically results in 17 receptor fragments having a broad range of expected masses (see Fig. 4). As we described previously (8), this was applied to the affinity labeled, gel-purified native and deglycosylated CCK receptor in 70% formic acid. The resultant fragments were then resolved on 10% NuPAGE gels (Novex, San Diego, CA) with MES running buffer and were visualized by autoradiography.
Each of the gel-purified, labeled CNBr-cleaved fragments was further digested with endoproteinase Lys-C and SAP. Lys-C (200 g/ml) digestion was performed at 37°C for 24 h in 50 mM Tris-HCl, pH 8.0, 10 mM EDTA, and 0.1% SDS. SAP (500 g/ml) digestion was performed in 0.1 M phosphate buffer, pH 7.8, containing 0.1% SDS at 25°C for 24 h. The products of digestion were also resolved on 10% NuPAGE gels.
Identification of the Sites of CCK Receptor Labeling-The gel-purified receptor fragments were desalted using a Dispo-Biodialyzer (The Nest Group Inc., Southborough, MA) at 25°C for 48 h prior to HPLC purification. For this, the samples were injected onto a PerkinElmer Life Sciences/Brownlee microbore C-18 column (number 219855), at a flow rate of 5 l/min. The system was run isocratically at 5% solution B for 50 min, followed by a linear gradient up to 90% solution B over 320 min (solution A, 0.1% trifluoroacetic acid, and solution B, 0.085% trifluoroacetic acid in acetonitrile). A 280 nm was monitored with UV absorbance detector. Eluate from the microbore column was collected directly onto a polyvinylidene difluoride membrane, and autoradiography was used to detect the position of the radioactive product. This position on the filter was excised and subjected to Edman degradation sequencing using an Applied Biosystems automated instrument.
The purified products of CNBr digestion of the CCK receptor that were labeled with the Bpa 24 analogue were also manually sequenced using Edman degradation chemistry, quantifying radioactivity released in each cycle. For this, purified fragments were coupled to N-(2-aminoethyl-1)-3-aminopropyl glass beads through the sulfhydryl side chain of a Cys residue. This was accomplished by derivatizing the amino groups on the beads with m-maleimidobenzoyl-N-hydroxysuccinimide ester at pH 7.0, quenching remaining amino reactivity with Tris, and then adding a purified, labeled receptor fragment. Cycles of Edman degradation were repeated manually in a manner that has been described previously in detail (8).
Characterization of Amino-terminally Truncated CCK Receptor Construct-Existing literature (28) has suggested that truncation of the first 37 residues of the amino-terminal tail of the CCK receptor does not affect CCK binding or biological action at the CCK receptor. We performed binding and biological activity studies with the cell line expressing our truncated receptor construct (eliminating residues 1-30), using the methodology described above. This construct was also used as a target of affinity labeling using the Bpa 24 and Bpa 29 probes described above.
Protease Resistance of CCK Receptor Constructs-Truncated and wild type CCK receptor constructs were studied to determine their abilities to resist attack by proteases from outside the cell. Trypsin was used since there are seven tryptic cleavage sites within this receptor that are theoretically exposed to the extracellular milieu. Intact receptor-bearing cells were lifted from their culture dishes using proteasefree cell dissociation medium, washed in KRH medium, and suspended in this medium in the presence or absence of 0.05% trypsin. This was allowed to react for 20 min at 37°C. At that time, the cells were washed with KRH medium and standard CCK radioligand-binding assays were performed (as described above).
Molecular Modeling-Three-dimensional models were generated for the type A CCK receptor using the low resolution rhodopsin structure (3) as a template. Both clockwise and counterclockwise seven-helix bundles were generated, and extracellular and cytosolic loops were added as described previously (29). The ligand, D-Tyr-Gly-[Nle 28,31 ,Bpa 29 ]CCK- (26 -33), was docked manually in the receptor models, using available photoaffinity labeling and other ligand binding data as constraints to facilitate exact placement and orientation. The solution phase conformation reported for CCK-(26 -33) (30) was initially adopted for the ligand, although this conformation was permitted to relax in subsequent energy minimization calculations. All manual three-dimensional model building was performed with the interactive molecular graphics program, PSSHOW (31). Docked ligand-receptor complexes were refined with limited energy minimization and low temperature molecular dynamics calculations using the AMBER 5.0 suite of programs (32).

Characterization of the Amino-terminal CCK Receptor
Probe-The synthetic Bpa 24 analogue of CCK was purified to homogeneity ( Fig. 1) and was analyzed by mass spectrometry to ensure its identity. This was oxidatively radioiodinated and purified by reversed-phase HPLC to yield a specific radioactivity of 2000 Ci/mmol. The probe bound to the CCK receptorbearing membranes from CHO-CCKR cells saturably, specifically, and with high affinity (K i ϭ 8.9 Ϯ 1.1 nM) (Fig. 2). Nonspecific binding, as determined in the presence of 1 M CCK, represented less than 10% of total binding. This CCK analogue stimulated an increase in intracellular calcium concentrations in Fura-2AM-loaded CHO-CCKR cells in a concentration-dependent manner (Fig. 2). It was similarly efficacious to natural CCK and was quite potent (EC 50 ϭ 0.5 Ϯ 0.04 nM).
Photoaffinity Labeling of the CCK Receptor-The photolabile CCK probe covalently labeled the CCK receptor, with the product migrating at M r ϭ 85,000 -95,000 on a 10% SDS-polyacrylamide gel (Fig. 3). The labeling of this band was inhibited in a concentration-dependent manner by competition with unlabeled CCK, and labeling was absent in nonreceptor bearing CHO cell membranes. As previously reported for the CCK receptor, deglycosylation with endoglycosidase F shifted the migration of the labeled band to M r ϭ 42,000, the position of migration of the receptor core protein. Fig. 3 also shows the densitometric quantitation of the receptor labeling in the presence of increasing concentrations of CCK, supporting the expected saturability and high affinity interaction between ligand and receptor.
Identification of the Domains of Receptor Labeling-CNBr cleavage of the affinity labeled CCK receptor was used as a first indication of the receptor domains being covalently labeled by the Bpa 24 analogue of CCK. Fig. 4 shows a typical autoradiograph of a NuPAGE gel used to separate the products of CNBr cleavage of the affinity labeled native and deglycosylated CCK receptor. Two distinct mixtures of molecular weight markers were used, due to substantial differences in the migration of the lower mass markers (migration positions noted). This cleavage of the intact receptor reproducibly generated labeled fragments migrating in the range of M r ϭ 4,000 -7,000 (the smaller fragment) and M r ϭ 25,000 (the larger fragment). In addition, as seen in this figure, larger labeled bands representing products of partial digestion were also sometimes observed. The smaller fragment was always predominant, representing 75 Ϯ 2% of the labeling.
The cleavage of the deglycosylated CCK receptor always produced only two labeled bands, migrating at M r ϭ 4,000 -7,000 and M r ϭ 8,500 (Fig. 4). The former was shown to come from the band migrating in the same position from the digestion of the native receptor, thus representing a nonglycosylated fragment. The latter was shown to come from deglycosylation of the glycosylated larger products of CNBr digestion of the intact receptor.
The specific covalent labeling of multiple bands was different from our previous experience with affinity labeling this receptor through photolabile residues sited within the pharmacophoric domain of the ligand in which only a single site of labeling was observed (8,9). Fig. 4 also shows the structure of the rat CCK receptor with theoretical sites of cleavage by CNBr and the characteristics of the expected fragments. The fragment extending from residue 10 to 72 best matches the M r ϭ 25,000 fragment, as supported by its shift in migration after deglycosylation and the sum of the masses of that fragment and the covalently attached probe (1,506 Da). Fragments 7, 9, 11, and 14 were all potential candidates for representing the smaller labeled fragment, based on topology, apparent size, and absence of glycosylation.
The identity of the M r ϭ 25,000 fragment was further de- fined by enzymatic cleavage of the purified fragment. The receptor fragment extending from residues 10 to 72 contains sites of N-linked glycosylation and two lysine residues that can be cleaved by endoproteinase Lys-C (Fig. 5). One Lys is very close to the carboxyl terminus of this fragment, and the other is positioned effectively to separate a nonglycosylated fragment FIG. 2. Binding and biological activity of the photoreactive Bpa 24 analogue of CCK. The probe, desamino-Tyr-Gly-Bpa-Gly-[Nle 28,31 ]CCK- (26 -33), displaced CCK radioligand binding to CCK receptor-bearing membranes from CHO-CCKR cells (left) and stimulated an increase in intracellular calcium in Fura-2AM-loaded CHO-CCKR cells (right) in a concentration-dependent manner. Data are expressed as means Ϯ S.E. of values from three independent experiments. Inset is a typical intracellular calcium transient response to 100 nM peptide over time in these cells.

FIG. 3. Photoaffinity labeling of the CCK receptor with desamino-Tyr-Gly-Bpa-Gly-[Nle 28,31 ]CCK-(26 -33).
Shown on the left is a typical autoradiograph of an SDS-polyacrylamide gel used to separate products of the labeling of receptor-bearing membranes in the absence and presence of increasing amounts of competing unlabeled CCK. The labeled band migrated at the expected position of M r ϭ 85,000 -95,000 that shifted to M r ϭ 42,000 after deglycosylation with endoglycosidase F. Absence of significant labeling of nonreceptor-bearing CHO cell membranes is also shown. Shown on the right is the densitometric analysis of the receptor labeling in the presence of increasing concentrations of competing CCK in four independent experiments (means Ϯ S.E.).

FIG. 4. CNBr cleavage of the photoaffinity labeled CCK receptor.
The left panel represents a diagram of the rat CCK receptor structure that includes theoretical sites of cleavage by CNBr, with the masses noted for the generated fragments. The first 15 residues of the amino terminus of the rat receptor are shown in parentheses, since this region is not present in the type A CCK receptor cloned from human or other species. The numbering scheme used, therefore, reflects the more highly conserved domains. A typical autoradiograph of a NuPAGE gel used to separate the products of CNBr cleavage of affinity labeled native and deglycosylated CCK receptor is shown on the right. CNBr digestion of affinity labeled CCK receptor molecules generated two distinct fragments, migrating at apparent positions of M r ϭ 4,000 -7,000 and M r ϭ 25,000, the former being predominant. Fragment 4 best matches the M r ϭ 25,000 fragment, as evidenced by its migration at apparent M r ϭ 8,500 after deglycosylation. Fragments 7, 9, 11, and 14 are potential candidates for the M r ϭ 4,000 -7,000 fragment, based on apparent size and absence of glycosylation. Two mixtures of molecular weight markers, Multimark standards (Invitrogen, Carlsbad, CA) and Kaleidoscope polypeptide standards (Bio-Rad), were used to indicate better the apparent mass of the smaller fragment. with a mass of 3,771 Da from a glycosylated fragment with core protein of 3,045 Da. The gel-purified M r ϭ 25,000 product of CNBr cleavage was further digested with endoproteinase Lys-C and deglycosylated with endoglycosidase F. Fig. 5 shows a typical autoradiograph of a NuPAGE gel used to separate the products of these treatments. Lys-C cleavage of this fragment yielded a band migrating at approximate M r ϭ 21,500 that shifted to approximate M r ϭ 5,000 after deglycosylation. Considering the mass of the covalently attached probe (1,506 Da), the amino-terminal part of the amino-terminal receptor fragment, extending from residues 10 to 37, best fit the data. This region of the CCK receptor is known not to be critical for CCK binding, based on the ability to truncate the first 37 residues of the amino-terminal tail of the receptor, while maintaining normal function (28). For this reason, no further specific localization of this site of labeling was pursued.
Identification of the smaller labeled fragment (M r ϭ 4,000 -7,000) was more of a challenge, since migration on a gel may not precisely reflect mass (particularly for small peptides, as demonstrated by the differential migration of the molecular weight standards in the two mixtures). Focus on the absolute mass of a single standard marker in a previous report (21) resulted in consideration of too few candidate fragments. Of the potential candidates for the current labeling (fragments 7, 9, 11, and 14), fragments 7, 9, and 14 contain Lys, Asp, or Glu residues that can be cleaved by the specific proteases, Lys-C or SAP. Therefore, these enzymatic reactions were first utilized to gain preliminary insight into the identity of this fragment. Fig.  6 shows a diagram of these CNBr fragments, with theoretical cleavage sites by these proteases. As hoped, both Lys-C and SAP treatment resulted in shifts in the migration of the labeled band on a gel (Fig. 6). Only fragments 7 and 14 contain residues that can be cleaved by both of these enzymes, suggesting that they were the most suitable candidates to represent this smaller labeled fragment. This fragment was definitively identified as CNBr fragment 14 by direct Edman degradation sequencing, in which the first 11 cycles matched the expected sequence (Pro-Ile-Phe-Ser-Ala-Asn-Ala-Trp-Arg-Ala-Tyr).
The site of covalent attachment of this probe to this fragment was further characterized by radiochemical sequencing of the labeled products of CNBr cleavage of wild type and V342M receptor constructs. Although no radioactivity above background was observed through 15 cycles for the labeled CNBr fragment of the wild type receptor, a peak was consistently observed in cycle 3 for this fragment of the V342M construct in four independent experiments (Fig. 7). This corresponds with covalent labeling of Glu 345 in the third extracellular loop of the receptor.
Characterization and Affinity Labeling of the Amino-terminally Truncated CCK Receptor Construct- Fig. 8 shows that the amino-terminally truncated CCK receptor construct recognizes and responds to CCK in a similar manner to the wild type receptor. This confirms the high affinity binding of CCK previously reported for another truncated CCK receptor construct in which residues 1-37 were eliminated (28). It extends the previous observations by adding insight into normal signaling by such a construct. Fig. 9 shows the attempt to affinity label the truncated CCK receptor construct with the Bpa 24 analogue of CCK. Of note, despite high affinity binding, this probe did not covalently label this receptor construct. Even prolonged exposure of the gel for autoradiography failed to demonstrate a labeled receptor band. This means that both sites of covalent attachment were impacted by the absence of the receptor amino-terminal domain. As a control, another analogue of CCK that incorporated the FIG. 5. Proteolysis and deglycosylation of the purified M r ‫؍‬ 25,000 CNBr fragment. Shown on the left is a diagram of fragment 4 with endoproteinase Lys-C cleavage sites. Shown on the right is a typical autoradiograph of a NuPAGE gel used to separate the products of digestion and deglycosylation of this fragment. Lys-C digestion yielded a band migrating at approximate M r ϭ 21,500 that shifted to apparent M r ϭ 5,000 after deglycosylation. This localized one site of labeling to the region between Asn 10 and Lys 37 , based on the mass of this fragment and that of the covalently attached probe (1,506 Da) and its glycosylation status.
FIG. 6. Proteolysis of the purified M r ‫؍‬ 4,000 -7,000 CNBr fragment. Shown on the left is a diagram of fragments 7, 9, 11, and 14 with sites of cleavage by endoproteinase Lys-C and SAP. Shown on the right is a typical autoradiograph of a NuPAGE gel used to separate the products of proteolysis of the labeled fragment. Treatment with both Lys-C and SAP resulted in shifts in the electrophoretic migration of the labeled band, suggesting that fragments 7 and 14 are the most suitable candidates to represent this smaller labeled fragment. same photolabile residue within the pharmacophoric domain was also used in an analogous series of affinity labeling experiments. This probe, the Bpa 29 analogue of CCK (9), efficiently labeled this receptor construct.
Protease Resistance of CCK Receptor Constructs-Exposure to trypsin of cells expressing similar densities of wild type and truncated CCK receptor constructs that had similar CCK-binding capacities resulted in substantial differences. Despite having multiple sites in the extracellular domain for digestion with this protease, only the truncated construct was negatively affected by this treatment, with its binding markedly inhibited (Fig. 10). The glycosylated amino terminus was able to provide a protective cover for the peptide-binding domain of the fully intact wild type CCK receptor.
Molecular Modeling-Based on results from our earlier photoaffinity labeling studies (8 -10), we initially positioned the peptide ligand such that a photolabile residue (Bpa) at position 29 would be well situated to form covalent cross-links with receptor residues His 347 and Leu 348 in the third extracellular loop, whereas a photolabile residue at position 33 would form cross-links with Trp 39 in the amino terminus of the receptor. Since structure-activity relationship data for CCK peptides indicate that residues positioned as an extension from the amino terminus of the pharmacophoric domain (residues Tyr 24 and Gly 25 ) have no impact on receptor binding (5, 20, 33), we oriented the peptide ligand so that its amino terminus made no specific contacts with the receptor but was instead exposed to solvent. Given that the amino terminus and extracellular loops of the receptor likely possess some conformational flexibility, then it is quite plausible that a photoaffinity label substituted in the position of Tyr 24 (shown in Fig. 11) would alkylate the amino terminus and possibly the third extracellular loop of the receptor.
The experimental results reported here are completely consistent with our models for the CCK peptide-receptor complex. The specific photoaffinity labeling patterns observed in this study (i.e. labeling of Glu 345 in the third extracellular loop plus residue(s) in the amino terminus), together with results from previous ligand binding and photoaffinity labeling studies, allow us for the first time to propose a specific helix bundle topology for the CCK receptor. If we assume that the bound conformation of the CCK peptide ligand is not dramatically different from the solution conformation for CCK-(26 -33) (30), then a counterclockwise helix bundle arrangement for the CCK receptor best accommodates the available ligand binding data and the full set of covalent cross-links we observe in our photoaffinity labeling experiments. DISCUSSION G protein-coupled receptors can provide pockets for binding small ligands within the confluence of their seven intramembranous helical segments and for binding large glycoprotein ligands in specialized structural domains configured within large amino-terminal domains (1,4). These receptors can also bind and be activated by peptide ligands, such as CCK. The emerging themes for such binding suggest critical contributions by extracellular loop and amino-terminal tail domains that are near the external face of the membrane, as well as including examples of portions of ligands dipping down into the confluence of helices.
Our current understanding of the molecular basis of CCK binding is based on receptor mutagenesis and affinity labeling studies. Both support the importance for peptide binding of regions just outside of the first transmembrane segment and in the extracellular loop domains (8,9,16). There has, however, been substantial controversy related to the details of CCK peptide docking to these domains. Very distinct models of the peptide-occupied receptor have been proposed (8,16,17). Nonpeptidyl ligand binding to this receptor seems to occur deeper in the membrane within the confluence of helices (34). FIG. 7. Edman degradation sequencing of the affinity labeled V342M mutant CCK receptor fragment. The affinity labeled V342M mutant CCK receptor was cleaved with CNBr and purified to homogeneity before being subjected to manual Edman degradation sequencing. Cycle three, corresponding to Glu 345 , was significantly above background (p Ͻ 0.05). Data represent means Ϯ S.E. of values from four independent experiments. a.a., amino acid. Established structure-activity relationships for CCK have localized the pharmacophoric domain to the carboxyl-terminal heptapeptide-amide (12,13). Almost every residue within this domain makes an important contribution to binding and activity. In contrast, almost any modification to the amino terminus of CCK-8 that has been attempted has been well tolerated, without modifying receptor binding or signaling. It was this feature that encouraged our positioning of the amino terminus of CCK-8 above the ligand-binding domain of the CCK receptor in our evolving model, such that an extension would not make contact with the regions of this receptor that are known to be important for function (8,10). It is noteworthy that another molecular model has placed the amino terminus of CCK much closer to the membrane and directed toward the other side of the helical bundle (16,17). Such a model is inconsistent with the residue-residue approximations that have been directly established by photoaffinity labeling studies (8,10).
To date, there has been the successful photoaffinity labeling of distinct spatially approximated residues within the CCK receptor through two positions within the pharmacophoric do-main of CCK (8 -10). This has been accomplished with two different photoreactive moieties in position 33 of CCK (8,10) and with a benzophenone moiety in position 29 of CCK (9). The position 33 photoprobes covalently labeled receptor residue Trp 39 just above the first transmembrane segment. The position 29 photoprobe established a covalent bond with receptor residues His 347 and Leu 348 just above the seventh transmembrane segment. It is noteworthy that both of these represented focused contacts with a single receptor domain, as might be expected from the high affinity tight interaction between native agonist ligand and the ligand-binding domain of the receptor.
In contrast, in the present work, the Bpa residue in position 24 of CCK established covalent bonds to either of two distinct domains of the CCK receptor. This probe was fully characterized as a high affinity ligand that had full efficacy relative to natural CCK. While one of the covalently labeled receptor domains was within a region that is known to be important, the third extracellular loop, it is notable that this contact was lost when the amino terminus of the receptor was shortened by truncation. The second labeled domain was a portion of the amino-terminal tail of the CCK receptor that can be eliminated by truncation, without any detrimental effects on CCK binding or agonist-stimulated signaling. This contact is therefore not critical for ligand affinity or function.
These spatial approximations, however, provide the basis to postulate that the docked peptide is tucked under the protective cover of the amino terminus of the CCK receptor. This function was further supported by the demonstration that this domain protected the receptor from extracellular proteolytic attack. Eliminating this region of the receptor by truncation had no detrimental effect on ligand binding affinity or agoniststimulated signaling, but resulted in a more labile receptor that was much more sensitive to proteolysis. The glycosylation of the amino-terminal domain of the CCK receptor is probably most responsible for this resistance to proteolysis. This is a recognized function of the glycosylation of membrane proteins (35,36). Another established function for membrane protein glycosylation relates to assisting in solubility and folding during biosynthesis that is also likely relevant to the CCK receptor.
Although position 24 in the CCK ligands is external to the established pharmacophore, cross-linking data generated by FIG. 9. Affinity labeling of the truncated CCK receptor with Bpa 24 and Bpa 29 probes. Shown is a typical autoradiograph of an SDS-polyacrylamide gel used to separate products of the labeling of the truncated receptor-bearing membranes using Bpa 24 and Bpa 29 probes in the absence and presence of competing unlabeled CCK (1 M) (representative of three similar experiments). The Bpa 24 probe failed to label covalently the truncated receptor, whereas the Bpa 29 probe efficiently labeled this preparation.
FIG. 10. Protease resistance of CCK receptor constructs. Shown are CCK competition-binding curves for CHO cell lines expressing wild type (WT) and truncated CCK receptor constructs that had been treated with trypsin. Cells were exposed to trypsin digestion for 20 min at 37°C, followed by extensive washing and standard CCK radioligand binding assay. The binding capacity of the truncated receptor was significantly negatively impacted by trypsin treatment, whereas the wild type receptor was resistant to this treatment. Values are expressed as the means Ϯ S.E. of data from three independent experiments. substitution of a photolabile residue at this position provides extremely useful constraint data for three-dimensional model building studies. The results reported here provide clear evidence that our earlier decision to position the amino terminus of the ligand facing away from receptor binding site (8, 10) was appropriate. The photoaffinity labeling data for the ligand amino terminus, combined with ligand binding studies and cross-linking data for positions within the peptide pharmacophore, also enable us for the first time to propose a distinct topological model for the peptide ligand-CCK receptor complex. Based on the full set of photoaffinity labeling and ligand binding data now available, it appears likely that the CCK receptor possesses a counterclockwise helix bundle topology. More definitive biophysical data will be needed to ultimately verify this prediction.
With this refinement in our working model, the tyrosinesulfate residue in position 27 of CCK that has been shown to be so important in structure-activity studies (37) is now in direct contact with Arg 197 . These residues, therefore, represent potential partners for charge-charge interaction. Indeed, Arg 197 is one of only three basic residues in the extracellular domain of the CCK receptor that has been reported to have a marked negative impact on CCK binding when replaced with an Ala residue (38).
Thus, using an analogue of CCK with a photolabile Bpa moiety in position 24, we have learned much about the binding domain for this hormone within its receptor. While the pharmacophoric domain of the peptide is held in constant approximation with the regions of the CCK receptor close to the membrane, the amino-terminal extension from the pharmacophore likely moves further away from the bilayer and is overlaid by a region of the receptor amino terminus that has no direct effect on binding affinity or on biological activity. Instead, this region of the receptor serves a protective function over the more critical domain below. The distinct positions of covalent labeling with this probe also provide additional detail to refine the molecular model of the docked peptide agonist. This evolving model now best supports a counterclockwise helical bundle topology for the receptor and residue approximations that are fully consistent with all existing experimental data.
Addendum-Since submission and review of this manuscript, a preliminary crystal structure for bovine rhodopsin has appeared 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., and Miyano, M. (2000) Science 289, 739 -745 that exhibits a counterclockwise helix bundle topology. Since the CCK receptor is also a member of the rhodopsin ␤-adrenergic receptor family, it is quite likely that the helix bundle topology will be similar.  (26 -33), bound to the CCK receptor, as seen from the side and top. The CCK receptor backbone is shown in cyan, and the peptide ligand backbone is highlighted in yellow. Key peptide residue side chains are displayed and labeled, including Tyr 24 , Tyr 27 (SO 4 ), Bpa 29 , and Phe 33 . Sites of receptor photoaffinity labeling are displayed in red, and a putative interaction between ligand residue, Tyr 27 (SO 4 ), and receptor residue, Arg 197 , is illustrated in white.