Determination of the Binding Site on the Extracellular Domain of Guanylyl Cyclase C to Heat-stable Enterotoxin*

Guanylyl cyclase C, one of the family of membrane-bound guanylyl cyclases, consists of an extracellular domain and an intracellular domain, which are connected by a single transmembrane polypeptide. The extracellular domain binds unique small polypeptides with high specificity, which include the endogenous peptide hormones, guanylin and uroguanylin, as well as an exogenous enterotoxigenic peptide, heat-stable enterotoxin, secreted by pathogenic Escherichia coli. Information on this specific binding is propagated into the intracellular domain, followed by the synthesis of cGMP, a second messenger that regulates a variety of intracellular physiological processes. This study reports the design of a photoaffinity labeled analog of heat-stable enterotoxin (biotinyl-(AC5)2-[Gly4,Pap11]STp(4–17)), which incorporates a Pap residue (p-azidophenylalanine) at position 11 and a biotin moiety at the N terminus, and the use of this analog to determine the ligand-binding region of the extracellular domain of guanylyl cyclase C. The endoproteinase Lys-C digestion of the extracellular domain, which was covalently labeled by this ligand, and mass spectrometric analyses of the digest revealed that the ligand specifically binds to the region (residue 387 to residue 393) of guanylyl cyclase C. This region is localized close to the transmembrane portion of guanylyl cyclase C on the external cellular surface. This result was further confirmed by characterization of site-directed mutants of guanylyl cyclase C in which each amino acid residue was substituted by an Ala residue instead of residues normally located in the region. This experiment provides the first direct demonstration of the ligand-binding site of guanylyl cyclase C and will contribute toward an understanding of the receptor recognition of a ligand and the modeling of the interaction of the receptor and its ligand at the molecular level.

Guanylyl cyclase C, one of the family of membranebound guanylyl cyclases, consists of an extracellular domain and an intracellular domain, which are connected by a single transmembrane polypeptide. The extracellular domain binds unique small polypeptides with high specificity, which include the endogenous peptide hormones, guanylin and uroguanylin, as well as an exogenous enterotoxigenic peptide, heat-stable enterotoxin, secreted by pathogenic Escherichia coli. Information on this specific binding is propagated into the intracellular domain, followed by the synthesis of cGMP, a second messenger that regulates a variety of intracellular physiological processes. This study reports the design of a photoaffinity labeled analog of heat-stable enterotoxin (biotinyl-(AC 5 ) 2 -[Gly 4 ,Pap 11 ]STp(4 -17)), which incorporates a Pap residue (p-azidophenylalanine) at position 11 and a biotin moiety at the N terminus, and the use of this analog to determine the ligand-binding region of the extracellular domain of guanylyl cyclase C. The endoproteinase Lys-C digestion of the extracellular domain, which was covalently labeled by this ligand, and mass spectrometric analyses of the digest revealed that the ligand specifically binds to the region (residue 387 to residue 393) of guanylyl cyclase C. This region is localized close to the transmembrane portion of guanylyl cyclase C on the external cellular surface. This result was further confirmed by characterization of site-directed mutants of guanylyl cyclase C in which each amino acid residue was substituted by an Ala residue instead of residues normally located in the region. This experiment provides the first direct demonstration of the ligand-binding site of guanylyl cyclase C and will contribute toward an understanding of the receptor recognition of a ligand and the modeling of the interaction of the receptor and its ligand at the molecular level.
Guanylyl cyclase C (GC-C), 1 a member of the growing family of membrane-bound guanylyl cyclases, catalyzes the synthesis of cGMP as a second messenger in intestinal and kidney cortex epithelial cells in response to stimulation by a ligand (1)(2)(3). It is generally thought that this process is involved in the regulation of intestinal and kidney fluids and electrolytes, via the endogenous peptide hormone, guanylin or uroguanylin (4,5). It is also noteworthy that GC-C greatly increases cGMP levels in intestinal epithelial cells, on interaction with an exogenous peptide, heat-stable enterotoxin (STa), which is produced by enteric bacteria such as enterotoxigenic Escherichia coli. This results in an efflux of watery electrolytes into the intestinal lumen, which leads to, in turn, acute diarrhea in human and domestic animals (6).
GC-C is a single subunit protein molecule (1050 residues) with a unique structure consisting of an N-terminal extracellular domain (ECD, ϳ407 residues), which is responsible for ligand binding, and a C-terminal intracellular domain (ϳ619 residues), for the catalysis of cGMP synthesis, which are connected through a single transmembrane polypeptide (ϳ24 residues) (1,7,8). The C-terminal intracellular domain is comprised of two functional regions, a kinase homology region and a guanylyl cyclase catalytic region. The N-terminal domain is structurally specific for a ligand such as guanylin and uroguanylin, and the resulting information is then transferred to the cytoplasmic intracellular domain via the transmembrane polypeptide, resulting in the stimulation of the guanylyl cyclase catalytic region in the intracellular domain. The molecular mechanism for recognizing the structure of a ligand and leading to signal transduction by GC-C currently is unknown, but one plausible scenario is that GC-C induces a change in conformation or topology between each subunit when ligand binding occurs, resulting in the removal of a negative regulatory effect of the kinase homology region on the activation of the guanylyl cyclase catalytic region and, thus, creating a favorable environment for the synthesis of cGMP (9). GC-C exists as an oligomer in cultured cells irrespective of the presence or absence of a ligand (9,10). On the contrary, the ECD molecule forms an oligomer in a ligand-dependent manner (11). Thus, the relationship of the oligomeric state of GC-C to the binding to a ligand remains unclear.
Studies of the site-directed mutagenesis of porcine GC-C revealed two regions in the ECD that are sensitive to point mutations (12); one is a highly conserved region from residue 91 to residue 155 (numbers denote the positions of amino acid residues relative to the N terminus of porcine GC-C) in the amino acid sequences of GC-Cs determined thus far, and the other is in the sequence from residue 274 to residue 409 which is close to the transmembrane portion. A mutation in the region from residue 347 to residue 401 in the latter, which is posi-tioned near the transmembrane portion, causes a significant reduction in both ligand binding activity and guanylyl cyclase catalytic activity. These results suggest that the region on the ECD for interacting with a ligand is focused on or located within these regions, the mutation of which strongly affects the biological properties of GC-C.
We describe herein the identification of the specific region on the ECD of porcine GC-C bound to a ligand. For defining the region on the ECD that confronts a ligand, we designed and synthesized a photoaffinity labeling analog of STa (biotinyl-(AC 5 ) 2 -[Gly 4 ,Pap 11 ]STp(4 -17)) ( Fig. 1) with a Pap residue (pazido-L-phenylalanine) at position 11 and a biotin moiety at the N terminus. This analog of STa would be expected to undergo UV-induced binding to the ECD, and a peptide fragment bound to this analog could then be salvaged from the enzymatic digest of the photoaffinity labeled ECD by using an avidin-immobilized matrix. MALDI-TOF mass spectrometry was used to map a peptide on ECD for biotinyl-(AC 5 ) 2 -[Gly 4 ,Pap 11 ]STp(4 -17), which has been specifically limited to the amino acid sequence from residue 387 to residue 393, close to the transmembrane portion. This finding was confirmed by site-directed mutagenesis because substitution by an Ala residue for Thr-389, Phe-390, or Trp-392 in the sequence, respectively, greatly affects binding ability to the ligand, as well as cyclase catalytic activity of GC-C. These results provide novel insights into the elucidation of the recognition and activation mechanism of GC-C by a ligand.

EXPERIMENTAL PROCEDURES
Materials-The reagents used for peptide synthesis were purchased from the Peptide Institute Inc. (Minoh, Japan) and Nacalai Tesque, Inc. (Kyoto, Japan). T4 DNA ligase and restriction enzymes were from Takara Shuzo Co. (Kyoto, Japan) and New England Biolabs, Inc. (Beverly, MA), respectively. Na 125 I (carrier free) was purchased from NEN Life Science Products and used for the iodination of STp (4 -17). Other reagents and solvents were purchased from Sigma and Katayama Chemicals Inc. (Osaka, Japan), all of which were reagent grade.
Photoaffinity Labeling of ECD6H with Biotinyl-(AC 5 ) 2 -[Gly 4 , Pap 11 ]STp(4 -17)-The expression and purification of ECD6H (ECD with a hexahistidine sequence attached to the C terminus) from Sf21 insect cells were carried out according to methods described in an earlier report (11). That is the recombinant baculovirus carrying the cDNA of ECD6H was infected into Sf21 insect cells (5 ϫ 10 7 , 80% confluent state) in a 175-cm 2 flask for 1 h. The infected cells were cultured for an additional 2 days in spinner flasks (5 ϫ 10 5 cells/ml). The culture supernatant (1200 ml) was treated batchwise with 10-ml bed volumes of concanavalin A-immobilized agarose. The concanavalin A-agarose was filtered and washed with buffer A (50 mM Tris (pH 7.5), 100 mM NaCl, 1 mM CaCl 2 , and 1 mM MgCl 2 ) (50 ml). The adsorbed protein was eluted with buffer A (50 ml) containing 500 mM ␣-methyl-D-mannoside. The non-adsorbed fraction was treated repeatedly by the same procedure, and the eluted fractions were pooled. The eluate was concentrated to 20 ml using an Ultrafiltration Cell (Amicon Inc., MA). The solution contained about 800 pmol of the purified ECD6H. The resulting purified ECD6H was mixed with biotinyl-(AC 5 ) 2 -[Gly 4 ,Pap 11 ]STp(4 -17) (100 nmol) and stored at 37°C in the dark for 1 h. The solution was then exposed to UV irradiation (302 nm, model UVM-57, UVP Inc., CA) for 30 min on ice. After the photoaffinity labeling, the labeled protein was separated from excess ligand and extraneous materials by using Ni 2ϩ -chelating affinity chromatography according to the procedure described in Ref. 11. Tris was added into the reaction mixture as a scavenger to inhibit nonspecific cross-linking reaction as well as serve as a buffer for the reaction solution (16,17). These purification procedures yielded about 260 pmol of the labeled ECD6H, based on the starting crude ECD6H (800 pmol), as judged from the ligand-binding capacity of the ECD6H treated without photoaffinity labeling (control experiments).
Digestion of Photoaffinity Labeled ECD6H-The solution that contained the photoaffinity labeled ECD6H, described above, was directly diluted with 1 ml of buffer (100 mM Tris (pH 9.0) and 8 M urea) and incubated at 4°C for 2 h. The solution was then diluted with 100 mM Tris buffer (pH 9.0) (5 ml) containing 0.01% SDS and concentrated to 0.5 ml on a Centricon-10 (Amicon Inc.). Peptide N-glycosidase F (0.4 units) was added to the solution for removal of carbohydrate chains at the N-linked glycosylation sites and allowed to incubate at 37°C for 2 h. The resulting solution was then incubated with endoproteinase Lys-C (3 g) at 37°C for 16 h.
Isolation of a Photoaffinity Labeled Peptide-The above digested ECD6H solution was mixed with avidin-immobilized agarose (50-l bed volume) and incubated at room temperature for 30 min. The avidinimmobilized agarose was collected on an Ultrafree-MC 0.1 m filter unit (Millipore Inc.). The agarose was washed twice with 100 mM Tris (pH 9.0) (300 l) containing 100 mM NaCl and then with sufficient 0.05% trifluoroacetic acid to remove salts. The peptides were removed from the agarose by heating in a mixture of 0.05% trifluoroacetic acid and 50% acetonitrile at 100°C for 15 min. The supernatant, which contained the peptides, was concentrated in vacuo and subjected to mass measurements with an Voyager Elite-XL MALDI-TOF mass spectrometer (Japan PerSpective Biosystems, Inc., Tokyo, Japan) using ␣-cyano-4-hydroxycinnamic acid as a matrix.
Site-directed Mutagenesis of Porcine GC-C-The preparation of the constructs of the recombinant GC-C and its mutant proteins was carried out according to procedures reported previously (12), with minor modifications. These recombinant proteins were transiently expressed in 293T human embryonic kidney cells, which were grown in Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% fetal bovine serum, using the Superfect transfection reagent according to the manufacturer's specifications (Qiagen). Briefly, the 1.0-kilobase pair cDNA fragment between the KpnI and HindIII sites was isolated from pCG-STaR. The cDNA fragment was subcloned into pBluescript and used as a template for the PCR reaction as follows. The first PCR fragment was prepared by using a universal primer (M13-20) and sense primers for the mutation (CCGGTGGATAAGG*C*CCCCACATTCATC for S387A; GTGGATAAGAGCG*CCACATTCATCTGG for P388A; GATAAGAGC-CCCG*CC*TTCATCTGGAAG for T389A; AAGAGCCCCACAG*C*CA-TCTGGAAGAAC for F390A; AGCCCCACATTCG*C*CTGGAAGAAC-CAC for I391A; CCCACATTCGCCG*C*GAAGAACCACAAA for W392A: asterisks indicate position of the mutated nucleotides), and the second PCR fragment was prepared by another universal primer (M13 reverse) and the antisense primers. The third PCR fragment was constructed with the universal primers (M13-20 and M13 reverse) using the first and second PCR fragments as templates. The third PCR fragment was digested with KpnI and HindIII in order to subclone to the XbaI/HindIII sites of pBluescript SK(Ϫ) together with the cDNA fragment between the XbaI and KpnI sites. These fragments were individually inserted into the pCG vector. The vectors were ligated with a 1.7-kilobase pair cDNA fragment that encodes for the C-terminal portion of porcine GC-C with HindIII sites at both ends. The DNA sequences of the vectors constructed in this study were confirmed by an Applied Biosystems 373A DNA Sequencing System using an ABI PRISM Dye terminator cycle sequencing kit (Perkin-Elmer).
Assay of cGMP Production of the Mutant Proteins of GC-C-cGMP production of the mutant proteins of GC-C, expressed on 293T cells, was assayed as described previously (8). Specifically, the 293 cell membranes, which expressed the mutant proteins, were mixed with STp (4 -17) and incubated at 37°C for 10 min. The reaction was terminated by the addition of 10% trichloroacetic acid, and the cells were then rapidly frozen at Ϫ80°C and thawed at room temperature. The cell debris was removed by centrifugation at 15,000 rpm (18,500 ϫ g) for 15 min. The resulting supernatant was extracted three times with water-saturated ether. The cGMP concentration was determined by a radioimmunoassay kit according to the manufacturer's specifications (YAMASA, Chiba, Japan). Other related assays have been described in an earlier paper (18).

Synthesis of Biotinyl-(AC 5 ) 2 -[Gly 4 ,Pap 11 ]STp(4 -17)-A novel
STa analog (biotinyl-(AC 5 ) 2 -[Gly 4 ,Pap 11 ]STp(4 -17), Fig. 1) was synthesized as a probe for mining a peptide fragment that encompasses the ligand-binding site on the ECD of GC-C. The synthetic ligand contained two functional residues as follows: one was a photosensitive amino acid (Pap) with azido group, which is easily converted to nitrene by radiation with UV light (300 nm) and covalently anchored to electron-rich groups such as N-H, O-H, etc. on the receptor molecule (17); and the other was a biotinyl moiety, which noncovalently binds to avidin with an extremely high affinity (K d , 10 Ϫ15 M) and is widely used as a carrier in purification of proteins using avidin-based affinity chromatography. This synthetic compound was confirmed not only to have the same conformation as that of STp(4 -17) by comparison of their CD spectra but also to be activated by UV radiation, showing it to be useful as a ligand for ECD. In addition, this ligand was confirmed to be efficiently adsorbed on avidinimmobilized agarose and easily removed from the avidin-agarose by heating at 100°C for 10 min (data not shown).
Binding The photoaffinity labeled ECD6H was treated with peptide N-glycosidase F for removal of carbohydrate moieties and then digested with endoproteinase Lys-C. The digest was passed through a column of the avidin-immobilized matrix to adsorb the photoaffinity labeled peptide fragments, which were then removed from the avidin-immobilized matrix by heating. The eluate from the avidin matrix was analyzed by direct MALDI-TOF mass spectrometry. Fig. 3 shows a typical mass spectrum of the eluate from the avidin-immobilized matrix. The mass values of two major signals, observed at m/z ϭ 878. 6 Fig. 3 conform only to peptides that encompass residues 387-393, bound to biotinyl-(AC 5 ) 2 -[Gly 4 ,Pap 11 ]STp(4 -17) but did not fit those of any other sequences (Fig. 4). These results are also supported by the observation that the signals (Fig. 3) were not detected in the digest of ECD6H that was treated in the absence of biotinyl-(AC 5 ) 2 -[Gly 4 ,Pap 11 ]STp(4 -17) by the same procedure as that described above (control experiment) (not shown).
Site-directed Mutational Analysis of the Ligand-binding Site-To investigate the issue of whether the ligand-binding region is truly localized within the amino acid sequence (resi-  (4 -17) (B). Six Cys residues are intramolecularly linked by disulfide linkages between Cys 5 and Cys 10 , Cys 6 and Cys 14 , and Cys 9 and Cys 17 . (4 -17) or STp (4 -17). Each data point represents the mean values of triplicate data sets. due 387 to residue 393) in ECD, it was mutagenized using site-directed mutagenesis to give six mutant proteins (Fig. 5), in which each of the amino acid residues in the sequence (residue 387 to residue 392) were individually substituted by an Ala residue by employing a recombinant GC-C expression system using 293T human embryonic kidney cells. The expression of the mutant proteins was confirmed by Western blotting and staining with an anti-GC-C antibody raised against a synthetic peptide covering the C-terminal region (residue 1036 to residue 1050) of porcine GC-C (20), as shown in Fig. 5A. The recombinant GC-C gave two protein bands on SDS-PAGE (lane 1 in Fig. 5A), in which the heterogeneity of molecular weight of GC-C was likely caused by differences in the extent of N-linked glycosylation, as has been reported previously (20). The mutant proteins exhibited two protein bands on SDS-PAGE, which were stained by an antibody detection reagent, similar to the wild type of GC-C (lanes 2-7), suggesting that all the mutant proteins were expressed in the same quantity as that of the wild type GC-C.
The response of the guanylyl cyclase activation of the mutant proteins for exposure to STp(4 -17) was compared with that observed for the wild type GC-C by assaying the cGMP concentration in the 293T cells which express either these mutant proteins or the wild type GC-C (Fig. 5C). The formation of cGMP for each of the mutant proteins was correlated to the binding ability to STa, even though all the mutant proteins, when stimulated by STa, exhibited cGMP formation to some degree. The mutant protein (S387A) gave rise to cGMP at a level similar to that of the wild type GC-C. On the contrary, the mutant proteins (P388A and I391A) showed 6.0 and 24% cGMP formation, and the other three mutant proteins (T389A, F390A, and W392A) showed only 0.5-1%, compared with the wild type GC-C, respectively. This again indicates that the mutant proteins (T389A, F390A, and W392A) have practically no ability to bind to STa (1 ϫ 10 Ϫ5 M). Collectively, these experiments demonstrate that the substitution by an Ala residue for amino acid residues Thr-389, Phe-390, and Trp-392 in GC-C abolishes their ability to bind to a ligand and, hence, the cyclase catalytic activity of GC-C.

DISCUSSION
The membrane-bound guanylyl cyclases represent a single transmembrane type receptor with distinct ligand specificities and are found in both mammals and lower eukaryotes (2,3). The extracellular domains of these receptor proteins show sequence diversity, compatible with their ability to recognize and discriminate a diversity of ligand structures. On the contrary, the intracellular domains, in particular the guanylyl cyclase catalytic regions, have high sequence homology among the receptors, consistent with their unique function in the synthesis of cGMP. This suggests that the receptors have a similar structural topology and regulate diverse physiological processes through ligand-specific synthesis of intracellular cGMP as an intracellular message. The identification of the binding sites of ligands to the receptors on the external surfaces of membranes will lead to a better understanding of the molecular basis of ligand binding and receptor activation.
GC-C serves as a receptor protein for the polypeptide ligands: guanylin (4) and uroguanylin (5) for endogenous ligands as well as heat-stable enterotoxin (STa) (21) for an exogenous ligand produced by enteric bacteria, such as enterotoxigenic E. coli. Our previous experiments (15,22,23) demonstrated that [Pap 11 ]STp(4 -17) substituted by a Pap residue in place of Asn 11 , located near the key amino acid residue Ala 13 in the context of binding to GC-C, is an efficient probe for photoaffinity cross-linking of STa to GC-C. In addition, we demonstrated that the ECD of GC-C, which is expressed in a system consisting of insect cells and a recombinant baculovirus, retains a ligand-binding ability that is similar to that of GC-C and could be prepared as a soluble, homogeneous protein (11). These data allowed us to examine the interaction of ECD with a ligand at the molecular level.
In this work, we focused on the determination of the specific region of the ECD that is involved in binding to STa. For this purpose, we used the following four-step strategy: covalent binding of an STp analog with a photoaffinity functional group to ECD, digestion of the ECD which is covalently coupled with a photoaffinity labeled STp analog, isolation of the photoaffinity labeled fragment from the digest by affinity chromatography, and identification of the labeled fragment by mass spectrometry. An STp analog (biotinyl-(AC 5 ) 2 -[Gly 4 ,Pap 11 ]STp(4 -17)) not only covalently linked to the ECD but permitted the isolation of a peptide fragment from the digest of the photoaffinity labeled ECD by affinity chromatography, a technique that takes advantage of the non-covalent binding between biotin and avidin. In general, the binding of a photoaffinity labeled ligand to its receptor protein appears to proceed with considerable low efficiency, as has been shown in other cases (24). Therefore, an expedient technique is needed, in order to detect the photoaffinity labeled receptor protein or its fragment from a photoaffinity labeled reaction mixture. In the present case, the photoaffinity labeled peptide fragment was successfully recovered by the attachment of a biotin moiety at the N terminus of the ligand from the digest of the photoaffinity labeled ECD6H but in only a tiny amount. Mass spectrometry identified the peptide fragment, which encompasses the amino acid sequence from residues 387 to 393 (SPTFIWK) (Fig. 4), positioned near the transmembrane portion of GC-C, in the digest of the ECD6H photoaffinity labeled with biotinyl-(AC 5 ) 2 -[Gly 4 ,Pap 11 ]STp (4 -17). Moreover, the peptide fragments that are bound to the photoaffinity ligand with the sequence from residue 388 to residue 391 (PTFI) and that from residue 390 to residue 393 (FIWK) were observed by mass spectrometry (Fig.  3). These findings strongly suggest that the ligand binds to the amino acids, Phe or Ile, at positions 390 and 391, respectively, which are common in the three peptide fragments observed by mass spectrometry. The result was again supported by sitedirected mutagenesis of the amino acid residues in the binding region, which show that the substitution of an Ala residue at positions Pro-388, Thr-389, Phe-390, or Trp-392 caused the complete loss of binding ability to the ligand (Fig. 5).
In our earlier work (12), the site-directed mutational analysis of the extracellular domain of porcine GC-C revealed that amino acid replacement (residues 347-348, 363-365, 373-375, and 399 -401) in a region (residue 347-401), which is close to the transmembrane region and surrounds the peptide fragment found in this study, causes a strong reduction in both binding activity to STa and guanylyl cyclase catalytic activity. In addition, the mutation of the Asn-379 residue to Ala, which is positioned at an N-glycosylation consensus site, lost the ability to contain an oligosaccharide at this residue and resulted in a strong reduction in binding ability of GC-C to STa (18). These previous results provide support for the conclusion that the region of GC-C involved in the binding to a ligand is located from Ser-387 to Lys-393, very near the transmembrane domain. It is also interesting to note that the amino acid sequence of the ligand-binding region that was found in porcine GC-C in this experiment is highly homologous to those not only of mammalian species but also of an amphibian (Xenopus laevis, African clawed frog) and a fish (Oryzias latipes, Japanese medaka), although this homology is not found in the amino acid sequences of the entire extracellular domains (1,7,8,25) [2][3][4][5] (Fig. 4).
The prediction of the secondary structure of ECD, as examined according to the method of Chou and Fasman (26), suggests that ECD is composed of two subdomains as follows: an ␣/␤-rich region (residue 1 to residue 331) and a ␤-rich region (residue 332 to residue 401), as described in Fig. 4. The ␣/␤ region consists of alternating ␣-helixes and ␤-strands and contains eight Cys residues, probably linked by intramolecular disulfide bridges, suggesting that this subdomain, which consists of 75% of the amino acid residues of the extracellular domain, functions as a determinant of the basic architecture of the entire molecule. In contrast, the ␤-rich region has a low propensity for ␣-helix in contrast to high propensities for a ␤-strand and turn and is comprised of hydrophilic amino acid residues, including a Cys residue which is not found in other species and perhaps is not involved in a disulfide linkage, suggesting that this subdomain forms a flexible conformation. The x-ray crystallography of the complex of human growth hormone and its receptor protein suggests that the ligandbinding region on the receptor protein is largely comprised of turn structures (27). The same may be inferred in the case of GC-C, because the sequence from Ser-387 to Pro-388 present in 2 R. T. MacFarland, DDBJ/GenBank TM /EBI accession number D49837. 3   the ligand-binding region is assumed to form a turn structure, and in particular, Pro-388 could represent an element of a turn structure. Indeed Pro-388 was identified as a significant amino acid residue, as evidenced by site-directed mutational analysis of GC-C (Fig. 5). Furthermore, the data herein are in agreement with the conclusions that the region around this site points to a high content of hydrophilic amino acid residues, suggesting that this region is located on the surface of the molecule, thus facilitating interaction with a ligand. In addition, we recently found that the ␤-rich region has a binding ability to STa, when the interaction of a recombinant protein, comprising the ␤-rich region, with STp(4 -17) was examined. 6 This experiment also provides evidence that the ␤-rich region near the transmembrane domain plays a role in ligand binding.
In this study, we identified a region of GC-C that is involved in ligand binding and that is critical for signaling via the ligand to the intracellular domain of GC-C, which is located near the transmembrane portion on the external surface of the cellular membrane. On the basis of the present finding, together with previously reported data, we hypothesize that the binding of a ligand to the extracellular domain induces a clustering (or oligomerization) of the extracellular domain and, in turn, modulates the topological relationship between each of the ligandbinding regions of ECD, as has been recently proposed in the case of the epidermal growth factor binding protein (28,29). This process may facilitate a change of the conformation of the intracellular cytoplasmic domain, resulting in the activation of the catalytic domain. It would then be interesting to see whether a principle for the transmission of a ligand signal from an extracellular domain into an intracellular domain exists in GCs, perhaps in a manner analogous to GC-C. The extracellular domain of GC-C activates the intracellular domain in the same manner as in other GCs, when generating cGMP, a common cytoplasmic second messenger, because the intracellular domain of GC-C is highly homologous to other GCs in terms of primary structure, although the extracellular domain of GC-C has a quite different amino acid sequence from those of other GCs (2). In any event, the determination of the binding site of GC-C for interacting with its ligand will provide new insights into our current understanding of the recognition and activation mechanism of GC-C by a ligand. A three-dimensional structural analysis of the extracellular domain and its complex with a ligand will be required, in order to answer the issue of how the extracellular domain propagates a signal to the intracellular cyclase catalytic domain on binding with a ligand.