Introduction

Top Summary Introduction Procedures Results Discussion References

GPCRs are thought to be integral membrane proteins that span the lipid bilayer seven times and give rise to an extracellular amino-terminal domain, three ECLs, three intracellular loops, and an intracellular carboxyl-terminal domain (1). Such a model is based on a limited amount of biochemical and biophysical data including a low resolution map of rhodopsin, although in the absence of high resolution crystallography, the exact structure of these receptors remains speculative (2). The structure of an apparently unrelated seven-TM protein, bacteriorhodopsin, has been determined at high resolution, and structural models of GPCRs have been created using bacteriorhodopsin coordinates (3). In the absence of crystallographic data, most investigators have therefore resorted to indirect methods such as site-directed mutagenesis and receptor chimeras to elucidate the structural and functional domains of GPCRs (4-9). Based on such studies, it has been suggested that the ligand binding site of GPCRs consists of a number of noncontiguous amino acid residues that form a binding pocket within the folded receptor (6, 7). In the case of rhodopsin and bioamine receptors, such a binding pocket lies deep within the plasma membrane and is made up exclusively of residues within the TMs. In contrast, large protein ligands such as glycoprotein hormones interact principally with the amino-terminal segment. For peptide receptors, however, the ligand binding pocket typically involves residues in the ECLs or both ECLs and TMs (7-9).

The neuropeptide SRIF interacts with a family of GPCRs with five current members termed sst receptor subtypes 1-5 (10). The receptors share a high degree of sequence identity in the TMs (55-70%) and diverge the most within the extracellular segments. Four of the five sst receptors (subtypes 1-4) display approximately equal affinity for both naturally occurring SRIF ligands, SRIF-14 and SRIF-28 (10-13). The sst₅R displays the same affinity for SRIF-14 as the other sst receptors but has a 2-30-fold higher affinity for SRIF-28 (10, 11). Conformationally restricted analogs like the octapeptide SMS (octreotide) or hexapeptide MK678 (seglitide) bind to only three of the sst receptor subtypes: 2, 3, and 5. By exploiting the differential ability of SMS to bind to sst₂R but not to sst₁R, Kaupmann et al. (14) systematically mutated the sst₁R to resemble sst₂R. Their findings suggest that the binding pocket for SMS involves residues located exclusively within TMs III-VII. Of the eight residues postulated to form the binding pocket, experimental evidence, however, exists for the involvement of only three of these: an aspartic acid in TM III, which is predicted to interact with lysine in the Phe-Trp-Lys-Thr core of both ligands, and asparagine and phenylalanine located at the outer end of TMs VI and VII, respectively (present in sst₂R but not sst₁R), necessary for binding of both ligands (14-16). A potential role of nonconserved residues, including most of the ECLs, has not been evaluated. Furthermore, the question of whether sst receptor subtypes 3-5 share a similar ligand binding pocket with sst₁R and sst₂R remains to be determined. Because SRIF-14 binds with roughly equal affinity to all five sst receptors, it is probable that the binding pocket for this ligand is identical in the five receptors. On the other hand, the ligand binding pocket of other GPCRs that recognize the same ligand has been shown to consist of some nonconserved residues, suggesting that the ligand binding pocket of one GPCR cannot always be used to predict the location of the pocket for the same ligand on a closely related member of the same subfamily (17-19). In all cases so far examined, at least a portion of the extracellular region of a GPCR is required to interact with peptide ligands like SRIF (4, 8, 9, 20). Accordingly, in the current study, we systematically examined the role of exofacial structures in ligand binding using as a model hsst₅R. With a series of hsst₅R mutants in which the entire NH₂-terminal segment has been deleted or the ECLs have been altered by CSE mutagenesis (20), we report that a major fraction of the extracellular domain of hsst₅R does not play a role in ligand binding but that mutations in the second ECL abolish high affinity ligand binding.

	Experimental Procedures

Top Summary Introduction Procedures Results Discussion References

Materials. SRIF-14 was from Ayerst Laboratories (Montreal, Canada). SRIF-28 and Leu⁸-D-Trp²²,Tyr²⁵-SRIF-28 (LTT SRIF-28) was from Bachem (Marina Del Ray, CA). SMS and Tyr³-SMS were from Sandoz (Basel, Switzerland). MK678 was from Merck Frosst Laboratories (Montreal, Canada). Mouse monoclonal antibody to vimentin was purchased from Sigma Chemical (St. Louis, MO) and obtained as a gift from Dr. D. Laird (McGill University, Montreal Quebec, Canada).

Construction of hsst₅r cassette cDNA. A cassette construct cDNA was created consisting of the entire coding sequence of hsst₅r in which silent mutations had been introduced to generate unique restriction sites to facilitate the manipulation of the sequence as discrete restriction fragments (Fig. 1) (20). Mutations were introduced into a specific hsst₅R restriction fragment by PCR and subsequently used to replace the corresponding wild-type fragment in the cassette construct. The choice of restriction sites was dictated by their absence in the expression vector (see below) and by the fact that the creation of the site by mutagenesis would not alter the amino acid sequence of the receptor. Five fragments encompassing the entire hsst₅R coding sequence were generated by PCR as follows: a HindIII/KpnI fragment containing the translational initiation codon as well as a Kozak consensus sequence (21) up to nucleotide 126, a KpnI/EcoRV fragment containing nucleotides 126-410, an EcoRV/BstXI fragment containing nucleotides 410-540, a BstXI/MluI fragment containing nucleotides 540-739, and an MluI/EcoRI fragment containing nucleotides 739-1092, including the translational termination codon. All five fragments were ligated by PCR (20); a restriction map of the resultant construct is shown in Fig. 1. Sequence analysis was used to confirm the structure of the cassette construct (University Core DNA Service, University of Calgary, Calgary, Alberta, Canada). The cDNA was then cloned into the HindIII/EcoRI sites of pTEJ8, a mammalian expression vector that uses the human ubiquitin promoter to drive expression (22).

View larger version (7K): [in this window] [in a new window]   Fig. 1.   Structure of the hsst5R cassette construct. The locations of the new restriction sites introduced via silent mutations are shown with respect to TMI-VII. A Kozak consensus sequence (GCCGCCACC) was also introduced directly 5 to the translational initiation codon (ATG). Bottom, the structure is numbered in nucleotides (base pairs). The entire hsst5R cassette construct was subcloned into the unique HindIII/EcoRI sites of the polylinker region of the mammalian expression vector pTEJ8.

Generation of hsst₅R mutants. To characterize the role of the exofacial structures of hsst₅R in ligand binding, a series of mutants of the NH₂-terminal segment and ECL1-3 were constructed as follows.

NH₂-terminal deletion mutant. An NH₂-terminal hsst₅R mutant (5NT) was constructed in which the NH₂-terminal 36 residues were removed by deletion of a portion of the cassette construct. Of the predicted 39 NH₂-terminal residues, only three were retained (Fig. 2).

View larger version (28K): [in this window] [in a new window]   Fig. 2.   Putative membrane topography of hsst5R and comparison of the sequences of the hsst5R mutants with their wild-type counterparts. A, Location and borders of extracellular regions. NH2, amino terminus; HOOC, carboxyl terminus; CHO, putative glycosylation site; also depicted is the potential palmitoylation site in the intracellular carboxyl-tail, which serves as a membrane anchor. B, Alignment of the ECL, NH2-terminal deletion, and ECL3 deletion mutants with their wild-type (w.t.) counterparts reveals residues that were conserved (:) as well as the conservative exchanges. Bold, wild-type sequences are numbered according to their location in the receptor sequence (11). Thirty-five residues of the amino terminus of hsst5R were removed to create the mutant 5nt del. An A36M mutation was introduced to create the new translational initiation codon of the mutant; therefore, only 3 of 39 amino-terminal hsst5R residues remain in the mutant.

ECL1-3 conservative segment exchange mutants. The technique of CSE mutagenesis was used to systematically evaluate the role of the three ECLs of hsst₅R in ligand binding (20). In CSE mutagenesis, large stretches of residues are replaced by chemically similar residues, with the exception of residues with no chemically similar counterparts (proline, glycine, and cysteine), which are not altered (Fig. 2). The ECL1 of hsst₅R consists of 13 residues that include a conserved cysteine residue found in the ECL1 of most GPCRs as well as the highly conserved WPXG motif (23). Because these residues may play major structural roles in the receptor (20), they were not altered in the CSE mutant for ECL1 (mut1) (Fig. 2B). The ECL2 region with 19 residues was divided into two portions, and two distinct mutants were constructed. Six of the 9 NH₂-terminal residues were conservatively altered to create mutant mut2N, whereas 7 of the 10 COOH-terminal residues were mutated to create mut2C (Fig. 2B). In the case of ECL3, 7 of the 9 residues were conservatively mutated to create mutant mut3 (Fig. 2). CSE mutagenesis has proved to be valuable in identifying key regions involved in the ligand binding domain of receptors such as the AT₁ receptor (20). Its advantage lies in the ability to mutate long stretches of amino acids without grossly altering receptor structure. Limitations of the technique include the inability to substitute residues proline, glycine, and cysteine or the WPXF motif in ECL1 as well as the generation of false-negative results because the chemical nature rather than the exact residue may be important in determining its ligand binding affinity.

ECL3 deletion mutant. An hsst₅R mutant (ECL3) was constructed in which four amino acid residues were removed from ECL3 to mimic the size of the loop in hsst₁R and hsst₄R (Fig. 2).

Binding assays. CHO-K1 cells expressing wild-type and mutant hsst₅R constructs were cultured in D-75 flasks to 70% confluency in Ham's F-12 medium containing 10% fetal calf serum and 400 µg/ml G418. The cells were washed and harvested by centrifugation, and membranes were prepared by homogenization. Binding studies were carried out for 30 min at 30° with 20-40 µg of membrane protein and ¹²⁵I-LTT SRIF-28 in 50 mM HEPES-KOH buffer, pH 7.5, containing 5 mM Mg²⁺, 0.02% BSA, 200 kallikrein inhibitor units/ml aprotinin, 0.02 µg/ml phenylmethylsulfonyl fluoride, and 0.02 µg/ml bacitracin (11, 12, 27). Incubations were terminated by the addition of 1 ml of ice-cold HEPES-KOH containing 0.2% BSA, rapid centrifugation, and washing. Radioactivity associated with membrane pellets was quantified in an LKB -counter (LKB-Wallac, Turku, Finland) with a counting efficiency of 69% and background of <50 cpm Specific binding is defined as the difference between counts bound in the absence and presence of 100 nM SRIF-28. Saturation binding experiments were performed with membranes using increasing concentrations of ¹²⁵I-LTT SRIF-28 (2-2000 pM) under equilibrium binding conditions as described previously (11, 12, 27). Competition analyses were carried out through incubation of membranes with ¹²⁵I-LTT SRIF-28 (~60 pM) and increasing concentrations of SRIF-14, SRIF-28, and SMS. Binding data were analyzed with INPLOT 4.03 (GraphPAD Software, San Diego, CA). As reported previously, the affinity of LTT SRIF-28 for sst₅ is 3-fold greater than that of SRIF-28 (12).

Generation of hsst₅R-specific antibodies. Peptide [³H]GLFPASTPSKK (G5nt) containing the extracellular amino-terminal segments 4-11 of the deduced sequence of hsst₅R with an [³H]Gly addition at the NH₂ terminus (to monitor coupling to protein) and a Lys-Lys addition at the COOH terminus was synthesized by solid phase and covalently conjugated to BSA with glutaraldehyde. Peptide YLFPASTPSKK (Y5nt) was synthesized for radioiodination. The G5nt-BSA conjugate was emulsified with complete Freund's adjuvant (2 mg/ml) and injected subcutaneously into 3-kg New Zealand White rabbits. Booster injections were administered every 4 weeks. The rabbits were bled from the central ear artery before immunization (preimmune serum) and 2 weeks after each boost. Antibody production was monitored by immunoprecipitation of ¹²⁵I-Y5nt. Serum harvested at 7 months from one of the immunized rabbits was selected for immunocytochemistry.

Immunofluorescent labeling and confocal microscopy for sstR expression. Cell surface expression of mutant sst receptor protein was determined by immunocytochemistry of live unfixed cells. CHO-K1 cells stably transfected with wild-type hsst₅R or hsst₅r mutants were cultured to ~70% confluency, washed twice in PBS, and incubated with hsst₅r primary antibody (1:200) in serum-free culture medium for 8-10 hr at 4° (28). Cultures were then rinsed twice in PBS and fixed with either methanol acetone (4:1) or freshly prepared paraformaldehyde (4% in 0.1% PBS) for 20 min. After three washes in 50 mM Tris·HCl and 1.5% NaCl, pH 7.4, cells were incubated for 90 min at 20° with rhodamine-conjugated goat anti-rabbit secondary antibody (1:100). Cells were finally washed thrice in 50 mM Tris·HCl and 1.5% NaCl, pH 7.4, and mounted with Immunofluor for confocal microscopy. Preimmune serum, antigen-absorbed antibody, and nontransfected CHO-K1 cells were used as controls. To monitor the integrity of the plasma membrane, unfixed cells were immunostained with an antibody to the intermediate filament protein vimentin (1:200), followed by incubation with goat anti-mouse fluorescein isothiocyanate-conjugated IgG (1:100). In addition, both sstR and vimentin immunoreactivities were localized in CHO-K1 cells permeabilized with 0.2% Triton X-100 for 5 min at 20° and processed under identical conditions. All fluorescent images were visualized on a Zeiss LSM 410 inverted confocal microscope equipped with an argon/krypton laser. Rhodamine signal was imaged on a photomultiplier after passage through FT510, FT560, and FT590 filter sets. Images were obtained as single optical sections taken through the middle of the cells and averaged over 32 scans/frame. All images were collected and handled identically. They were archived on a Bernoulli multidisc and printed on a Kodak XL58300 high resolution (300 dpi) color printer.

	Results

Top Summary Introduction Procedures Results Discussion References

Pharmacological characterization of wild-type and mutant forms of hsst₅R. Total and nonspecific binding of ¹²⁵I-LTT SRIF-28 to membranes prepared from stable CHO-K1 cells transfected with cDNAs for the hsst₅R cassette construct and the hsst₅R mutants is shown in Fig. 3. Except for the two ECL2 mutants (mut2N, mut2C) wild-type hsst₅R and all other mutants displayed specific binding of ¹²⁵I-LTT SRIF-28 ligand. Comparable binding was obtained with a second ligand, ¹²⁵I-Tyr3 SMS, which also failed to bind to the ECL2 mutants. Saturation analysis of the hsst₅R revealed high levels of expression of membrane receptors with a B_max value of 119 ± 24 fmol/mg of protein and a K_d value of 0.335 ± 0.075 nM (Table 1). These values are consistent with our previous characterization of hsst₅R (9). The receptor expressed from the hsst₅R cassette construct bound SRIF-28  SRIF-14 > SMS (Table 2).

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Binding of 125I-LTT SRIF-28 to membranes (40 µg of total protein) in the absence (total binding) or presence (nonspecific binding) of 100 nm SRIF-28 in wild-type and mutant sst5R-expressing cells. Typical data are representative of duplicate measurements. See text for experimental details.

NH₂-terminal deletion mutant. Saturation binding analysis with membranes prepared from CHO-K1 cells expressing the 5NT mutant revealed high levels of expression of the mutant receptor (290 fmol/mg of protein), which exhibited high affinity binding of ¹²⁵I-LTT SRIF-28 ligand comparable to that of the wild-type receptor (Table 1). The binding affinities of the mutant receptor for SRIF-14, SRIF-28, and SMS obtained by displacement analysis revealed only a minor (<2-fold) change compared with wild-type (Fig. 4 and Table 2), indicating that the amino-terminal domain of hsst₅R is not critical for high affinity binding of these ligands (29).

View larger version (18K): [in this window] [in a new window]   Fig. 4.   Competitive binding analyses of hsst5R CSE and amino-terminal deletion mutants. Competitive dose-dependent inhibition of 125I-LTT SRIF-28 binding to hsst5R and hsst5R mutants by SRIF-114, SRIF-28, and SMS. Membranes were prepared from cells stably expressing the wild-type hsst5R cassette construct () or 5nt mutant (×), mut1 (*), mut2N (), mut2C (), mut3 (), and ECL3 mutants (+) were incubated with 60 pM125I-LTT SRIF-28 and the indicated concentrations of SRIF-14 (top), SRIF-28 (middle), or SMS (bottom) under equilibrium conditions (see Experimental Procedures). Error bars were omitted for clarity (n = 3). Total and nonspecific binding of each construct are shown separately in Fig. 3.

ECL1. Saturation analysis revealed that the mutant receptor mut1 retained a high level of membrane expression (B_max = 144 fmol/mg of protein) as well as high affinity for ¹²⁵I-LTT SRIF-28 (Table 1). The affinity of mut1 was further assessed by competitive binding analysis and showed a slight (<-2.5-fold) decrease in affinity for SRIF-14, SRIF-28, and SMS, indicating that ECL1 similarly is not critical for high affinity agonist binding.

ECL3. Proper expression and targeting of the ECL3 CSE mutant mut3 were similarly suggested by saturation binding analysis that showed a B_max value of 106 fmol/mg of protein and a 3-fold lower binding affinity (K_d) than the wild-type receptor (Table 1). Competitive binding analysis revealed that the affinity of mut3 for SRIF-14 and SMS decreased 4-fold, whereas the affinity for SRIF-28 decreased 8-fold (Fig. 4 and Table 2). This suggests that ECL3 also is not critically important for high affinity ligand binding. In addition to a role of specific amino acid residues, the size of the ECLs could be important in the formation of the binding pocket. For example, the subtypes hsst₁R and hsst₄R, which fail to bind conformationally restricted SRIF analogs, display noticeably shorter ECL3 segments, prompting speculation that a smaller ECL3 could sterically hinder ligand access to the proposed binding pocket in TMs 6 and 7 (24). To test this hypothesis, we constructed the ECL3 mutant. This mutant (ECL3) was readily expressed at the cell membrane, where it displayed a binding affinity similar to that of the wild-type receptor (Table 1). This mutant displayed a 12-fold decrease in affinity for SRIF-14 and SRIF-28 and an 8-fold decrease for SMS (Table 2 and Fig. 4). These changes are relatively small and indicate that the size of ECL3 cannot account for the reported >1000-fold lower affinity for short SRIF analogs such as SMS displayed by the two sstR subtypes (hsst₁R and hsst₄R) that feature relatively short ECL3 (23).

ECL2. No specific ¹²⁵I-LTT SRIF-28 binding could be detected in CHO-K1 cells stably transfected with mut2N and mut2C (Fig. 3 and Table 1). Because it is possible that such dramatic loss of binding is due to improper targeting of the mutant receptors to the plasma membrane, expression of receptor protein was analyzed immunocytochemically in live, unfixed cells. With the NH₂-terminally directed anti-peptide hsst₅R primary antibody, CHO-K1 cells expressing wild-type hsst₅R showed rhodamine immunofluorescence localized to the cell surface (Fig. 5A). The two ECL2 mutants also exhibited surface immunofluorescence with this antibody, indicating that the mutant receptors are properly localized to the plasma membrane (Fig. 5, B and C). Permeabilization of CHO-K1 cells expressing wild-type or the two mutant receptors resulted in immunoreactivity of cytosolic structures (Fig. 5, D-F). No specific immunofluorescence was detected in nontransfected unfixed CHO-K1 cells or in transfected cells probed with preimmune serum or antigen-absorbed primary antibody (Fig. 5G). To exclude hsst₅r immunostaining of cytosolic structures beneath the membrane as a result of permeabilization of the plasma membrane during incubation with the primary antibody, parallel immunocytochemistry was carried out with antibody to vimentin, an intracellular protein, and showed no surface or cytoplasmic labeling (Fig. 5H). When the cells were permeabilized with 0.2% Triton X-100 before application of vimentin antibody, however, there was intense cytoplasmic staining (Fig. 5I).

View larger version (90K): [in this window] [in a new window]   Fig. 5.   Immunocytochemical analysis of hsst5R and the ECL2 CSE mutants of hsst5R. Confocal immunohistochemical localization of wild-type and ECL2 mutant hsst5Rs (mut2N, mut2C) in stably transfected CHO-K1 cells. Nonpermeabilized cells (A-C) and Triton X-100-permeabilized cells (D-F) were labeled with rabbit anti-hsst5R primary antibody and rhodamine-conjugated goat anti-rabbit secondary antibody. The integrity of the plasma membrane during the immunohistochemical procedure was monitored in nonpermeabilized and Triton X-100-permeabilized CHO-K1 cells expressing wild-type hsst5R, each immunostained with vimentin antibody under identical conditions as for sst5R localization (H and I). A, Representative unfixed cell expressing wild-type hsst5R displays strong immunofluorescence over the cell surface. mut2N (B) and mut2C (C) in unfixed CHO-K1 cells also display surface immunofluorescence, indicating plasma membrane expression of the mutant receptors. Permeabilization of CHO-K1 cells expressing wild-type (D) mut2N (E), and mut2C (F) results in immunostaining of cytosolic structures. Specificity of sst5R fluorescence was validated using nontransfected unfixed CHO-K1 cells (G) or with preimmune serum and antigen-absorbed antibody (not shown). Specific vimentin immunoreactivity was undetectable in nonpermeabilized CHO-K1 cells expressing wild-type hsst5R (H) but readily detectable in permeabilized cells (I). Scale bar, 10 µM.

	Discussion

Top Summary Introduction Procedures Results Discussion References

Using a cassette construct of hsst₅R cDNA for the efficient generation of conservative segment exchange and deletion mutants, we mapped the entire exofacial domain of the receptor for potential interaction with SRIF-14, SRIF-28, and SMS ligands. Saturation analysis with ¹²⁵I-LTT SRIF-28 revealed that mutants of the NH₂-terminal segment of ECL1 and ECL3 were properly expressed and retained high affinity binding. Deletion of the NH₂-terminal domain produced a slight (<2-fold) decrease in binding potency (K_i) for SRIF-14, SRIF-28, and SMS compared with the wild-type construct. hsst₅R thus joins other GPCRs, such as TRH and glucagon receptors, whose NH₂-terminal domain does not influence ligand binding (29, 30). Because the NH₂-terminal segment contains two putative N-linked glycosylation sites, our results further suggest that glycosylation of hsst₅R is not a determinant of high affinity ligand binding. This is in contrast to solubilized rat brain sstRs, which have been reported to lose high affinity agonist binding on deglycosylation (31). Binding of SRIF-14, SRIF-28, and SMS to the CSE mutants of ECL1 and ECL3 was also comparable with that of the wild-type receptor, suggesting that no crucial interaction occurs between the three peptides and residues in ECL1 and ECL3. In contrast, radioligand binding to the two ECL2 mutants was completely abolished. Because the changes introduced are conservative, we cannot rule out the possibility that the altered residues can still interact with the ligand. In this case, the location of the residue would be the determinant of ligand interaction. Using specific antibodies directed against the hsst₅R NH₂-terminal segment, the mutant ECL2 receptors were shown by immunocytochemistry to be correctly targeted to the cell surface. Immunocytochemistry using live unfixed cells ensured the integrity of the plasma membrane and established that the hsst₅r immunofluorescence was due to membrane and not cytosolic receptors. Thus, failure of the mutant proteins to bind agonist could be explained by a model in which the altered residues represent crucial points of interaction between receptor and ligand. Alternatively, the mutations in ECL2 might have allosterically disrupted the ligand binding pocket. To distinguish between these two possibilities, nonpeptide antagonists, which, in many instances, bind to a region of the receptor separate from that involved in agonist peptide binding, have provided a valuable tool in mutational mapping studies of several GPCR systems (4, 7, 20, 32). For example, CSE mutants of ECL3 of the AT₁ receptor display differential loss of binding for the agonist angiotensin II while retaining full binding for the nonpeptide antagonist L-158,809, suggesting that the mutation directly affected the ligand binding cavity without altering overall receptor structure (20). Similar analysis of sstR ECL2 mutants of this study, however, could not be undertaken because of the current lack of sstR nonpeptide antagonists. In view of increasing evidence that different agonists may occupy different binding pockets in the same receptor, another approach is to test the interaction of various agonists with the mutant receptor (9, 17, 20). In the case of sstRs, mutational analyses so far have suggested that the binding pocket for SMS involves residues in TMs exclusively (14). Accordingly, we used ¹²⁵I-Tyr3 SMS as an alternative ligand to ¹²⁵I-LTT SRIF-28 to detect binding by the two ECL2 mutants. Neither of the two radioligands bound to these mutants, implying that residues in ECL2 form part of the ligand binding pocket for both ligands or mutations in ECL2 caused a structural change in the receptor with secondary loss of binding. Accumulating evidence now suggests an important role of ECL2 in ligand/receptor interactions with the exterior portions of a number of GPCRs (9). Further studies using point mutations of ECL2 are in progress and should help to clarify which, if any, individual residue in ECL2 is required in ligand binding of sstRs.

Earlier work on the putative residues critical for ligand binding has been reported by two groups who provided a tentative model of the sstR ligand binding crevice (14, 24). Kaupmann et al. (14) found that mutating the sst₁R residues Q291N in TM-VI and S305F in TM-VII to the corresponding residues of sst₂R increased the affinity of sst₁R for SMS by 1000-fold (14). Mutation of the conserved aspartic acid residue in TM-III of sst₁R, sst₂R, and sst₃R also abolished ligand binding, but it is not known whether this is due to direct involvement of the residue in the ligand binding pocket or a secondary alteration of the receptor structure (14-16). Using these three identified residues, the known structure of the SMS ligand, and the proposed three-dimensional model of GPCRs (3), Kaupmann et al. (14) proposed that the ligand binding pocket of sst₁R is lined by residues within TMIII-VII (14). The results of Strnad and Hadcock (16) are also consistent with the notion that short SRIF analogs bind to residues within the TMs (14). Using chimeric mouse sst₁R and sst₂R, Fitzpatrick and Vandlen (24) reported that substitution of the segment between ECL2 and ECL3 in sst₁R with the comparable region in sst₂R conferred high affinity binding to MK678, providing evidence that determinants involved in recognizing this ligand reside in the portion of sst₂R connecting ECL2 and ECL3. In addition, replacement of ECL3 of sst₂R with the corresponding portion of sst₁R resulted in a 3000-fold decrease in affinity for MK678 without affecting SRIF-28 binding. These results can be explained by direct involvement of the ECL3 of sst₂R in MK678 binding or by the relatively short ECL3 of sst₁R sterically hindering access of MK678 to the ligand binding pocket. Such a role of ECLs in ligand exclusion has been proposed for other GPCRs (31). In this study, truncation of the ECL3 of hsst₅R to mimic the size of this loop in sst₁R produced only a modest 3-fold decrease in affinity for SMS. Furthermore, in the Fitzpatrick and Vandlen study (24), the putative ECL3 segment of sst₁R grafted onto sst₂R consisted not only of ECL3 but also the upper helices of TMVI and TMVII, which contain residues N and FDFV, which are critical for binding. Thus, neither the size of ECL3 nor individual residues seem to be determinants of SRIF binding or of the low affinity of sst₁R for the short SRIF analogs. It should be noted that of all the residues identified by mutagenesis as being important in recognizing SMS and MK678, none have been shown to be critical for binding the natural ligands SRIF-14 and SRIF-28 (14, 24). Furthermore, the assumption by Kaupmann et al. (14) that sst₂R shares a common ligand binding pocket with the other sstRs may not be generally applicable because there are several examples suggesting that closely related GPCRs feature different sets of epitopes for binding of a common ligand (18-20). For instance, some residues of the AT₁ receptor that have been shown to be involved in ligand binding are not conserved in the AT₂ receptor even though the two receptors bind the same ligand (20). Overall, in combining our results with those obtained by Kaupmann et al. (14), the model that emerges suggests a binding pocket for SRIF ligand lined by residues within TMIII-VII with a potential contribution by ECL2. Such a model is consistent with other peptide binding GPCRs, such as neurokinin₁, AT₂, gonadotropin-releasing hormone, and luteinizing hormone-releasing hormone, that interact with residues in both ECLs and TMs (8, 20, 34).