Differential Contributions of Motilin Receptor Extracellular Domains for Peptide and Non-peptidyl Agonist Binding and Activity*

The family of G protein-coupled receptors that includes receptors for motilin, ghrelin, and growth hormone secretagogue has substantial potential importance as drug targets. Understanding of the molecular basis of hormone binding and receptor activation should provide insights that are helpful in the development of such drugs. We previously examined the unique second extracellular loop domain of the motilin receptor, identifying key epitopes in perimembranous locations at each end of this long loop (Matsuura, B., Dong, M., and Miller, L. J. (2002) J. Biol. Chem. 277, 9834–9839). Here, we have extended that work, examining the other predicted extracellular domains of the motilin receptor by using sequential deletions of segments ranging from one to six amino acid residues and site-directed alanine replacement mutagenesis approaches. Each construct was transiently expressed in COS cells, and characterized for motilin- and erythromycin-stimulated intracellular calcium responses and motilin radioligand binding. Only those receptor segments that included key Cys residues in positions 25, 30, and 111 or perimembranous regions at the ends of the amino terminus and the first and third extracellular loops disrupted motilin biological activity. Each of these Cys deletions also disrupted action of erythromycin. Alanine replacements for each of the potentially important amino acid residues in the perimembranous segments revealed that residues Gly36, Pro103, Leu109, and Phe332 were responsible for the selective negative impact on motilin biological activity, while responding normally to erythromycin. These results support the presence of functionally important disulfide bonds in the motilin receptor ectodomain and demonstrate that the structural determinants for binding and biological activity of peptide and non-peptidyl agonist ligands are distinct, with a broad extracellular perimembranous base contributing to normal motilin binding.

Understanding of the molecular basis of hormone binding to and activation of receptors provides important insights into the active conformation of such receptors, providing critical insights helpful for the design and refinement of receptor-active drugs. The family of guanine nucleotide-binding protein (G protein)-coupled receptors that includes receptors for motilin, ghrelin, and growth hormone secretagogue includes potentially important drug targets (1)(2)(3)(4). The physiological actions of these hormones make these targets compelling, with potential roles in gastrointestinal motility disorders, abnormalities of appetite and obesity, and disorders of growth and development (1)(2)(3)(4). However, this receptor family is unique in that its natural ligands include substantial structural diversity, such as the rare, yet critical, post-translational modification representing octanoylation of ghrelin (3,5,6).
As an initial effort to localize key receptor regions for motilin action, we examined the second extracellular loop domain of the motilin receptor, which is structurally divergent from the other members of this family, being much longer than that of the growth hormone secretagogue receptor (7). Of note, only the perimembranous ends of this loop that are conserved in the related receptors were shown to be functionally important for motilin binding and biological activity, with the long inserted sequence not contributing to motilin function (7). In the present work, we have extended that analysis to include all of the other predicted extracellular regions of the motilin receptor. This includes the amino-terminal tail and the predicted first and third extracellular loop regions. Each of these receptor regions was studied by analogous approaches to those that have been utilized to examine the second extracellular loop region. We have performed serial segmental deletions of the regions of current interest to identify functionally important areas and have further focused on specific residues within these regions using site-directed alanine replacement mutagenesis.
These studies have further supported the theme that agonists having distinct chemical structures, here a peptide and a non-peptidyl antibiotic, can act to stimulate receptor activation via interaction with distinct domains. The broad and diffuse base for natural motilin action is also further defined and found to be consistent with regions identified in photoaffinity labeling studies using probes with photolabile sites of covalent attachment in motilin peptide residues 1 and 5 (8,9). This work also adds insights into key structural features of this receptor, confirming the presence of a disulfide bond that is highly conserved across this superfamily (10,11) and drawing attention to another likely bond in the amino-terminal tail linking Cys 25 and Cys 30 , thus supporting its functional importance.

EXPERIMENTAL PROCEDURES
Materials-Motilin, erythromycin, and protease-free cell dissociation buffer were from Sigma-Aldrich. Eukaryotic expression vector, pcDNA3.1(Ϫ), was from Invitrogen. The human motilin receptor cDNA was kindly provided by Dr. A. D. Howard of Merck Research Laboratories (1). Enzymes used for receptor mutagenesis were purchased from Roche Applied Science or Stratagene (La Jolla, CA). Fura-2/acetoxymethyl ester (Fura-2/AM) 3 and Alexa 488 -conjugated goat anti-mouse IgG were from Molecular Probes (Eugene, OR). The mouse 12CA5 monoclonal antibody against the hemagglutinin (HA) epitope was from Roche Applied Science.
Receptor Constructs-Motilin receptor constructs were prepared that included segmental deletions of one to six amino acid residues and site-directed alanine replacement mutations in regions predicted to represent the amino-terminal tail and the first and third extracellular loops. These are illustrated in Fig. 1. Constructs were prepared using an oligonucleotide-directed approach. Polymerase chain reactions (PCR) were performed with Pfu Turbo TM DNA polymerase, running 18 cycles of 95°C for 30 s, 65°C for 1 min, and 68°C for 14 min. Products of PCR and restriction enzyme digestions were separated on 1% agarose gels and purified using the Qiagen reagent (Valencia, CA). Receptor constructs were subcloned into the eukaryotic expression vector, pcDNA3.1(Ϫ). Sequences of each of the receptor constructs were confirmed by direct DNA sequencing using an ABI Prism (Foster City, CA) DNA sequencer.
Additionally, a series of HA-tagged constructs were also prepared for immunostaining studies; these included HA-tagged wild-type, alanine site mutants (G36A, P103A, L109A, and P332A), and cysteine deletion mutants (⌬25C, ⌬30C and ⌬111C). They were prepared using the same strategy as described above, by placing the HA sequence (YPYDVP-DYA) at the amino terminus between residues Met 1 and Gly 2 of the wild type or mutant receptors.
Receptor Expression-Receptor constructs were expressed transiently in COS-1 cells (American Type Culture Collection). In brief, 0.5 ϫ 10 6 cells plated on tissue culture plasticware were transfected with 3 g of DNA using a modification of the DEAE-dextran method. Cells were maintained in Dulbecco's modified Eagle's medium supplemented with 5% fetal bovine serum. 72 h after transfection, cells were harvested with cell dissociation medium and used in biological activity and/or radioligand binding assays.
Biological Activity Assays-The ability of each motilin receptor construct to stimulate an intracellular signal in response to motilin and erythromycin was studied using a well established assay for intracellular calcium in Fura-2/AM-loaded transfected COS cells. In this assay, 2.0 ϫ 10 6 receptor-bearing cells were loaded with 5 M Fura-2/AM in Krebs-Ringer-HEPES medium (KRH; 25 mM HEPES, pH 7.4, 104 mM NaCl, 5 mM KCl, 1.2 mM MgSO 4 , 1 mM KH 2 PO 4 , 2 mMCaCl 2 ) containing 0.2% bovine serum albumin and 0.01% soybean trypsin inhibitor. Cells were incubated for 20 min at 37°C. Fluorescence was quantified in a LS55 luminescence spectrometer (PerkinElmer Life Sciences). Excitation was performed at both 340 and 380 nm, with emission determined at 520 nm. Calcium concentrations were calculated from the ratios of these values as described by Grynkiewicz et al. (12). The peak intracellular calcium concentrations achieved were utilized to determine the agonist concentration-dependence of the biological responses. All assays were repeated at least four times.
Receptor Binding Assays-Radioligand binding assays utilized various mutant motilin receptor-bearing cells with a constant amount of radioligand (3-5 pM [ 125 I-Ile 13 ]-motilin) and varied concentrations of unlabeled motilin (ranging from 0 to 1 M) in KRH medium containing 0.2% bovine serum albumin and 0.01% soybean trypsin inhibitor. Incubations were carried out for 60 min at 25°C. The binding assays were performed in 24-well tissue culture plates. Nonspecific binding was determined in the presence of 1 M motilin and represented less than 20% of total binding. All assays were repeated at least four times.
Immunofluorescence Microscopy-For morphological assessment of receptor expression on the cell surface, COS cells transiently transfected with HA-tagged wild type or mutant motilin receptors were replated to grow on coverslips for 48 h. Cells were washed with PBS and fixed in 2% paraformaldehyde for 30 min. After being washed once with PBS and twice with 1% normal goat serum in PBS, cells were incubated with mouse monoclonal anti-HA antibody FIGURE 1. Illustration of the human motilin receptor constructs used in this report. Shown is a schematic diagram of the primary sequence and a possible membrane topology (based on hydrophobicity) of the motilin receptor, along with the design of the sequential deletions (in numbered brackets) and Ala replacement constructs (in dark circles with white lettering).
(1:500) for 1 h. Cells were incubated with Alexa 488 -conjugated goat anti-mouse IgG (1:200) for 1 h after three washes with 1% normal goat serum in PBS. Coverslips were then washed three times with PBS, mounted on slides, and examined with a Zeiss Axiovert 200M inverted microscope (Carl Zeiss, Oberkochen, Germany) equipped for epifluorescence. All of the above procedures were performed at room temperature.
Statistical Analysis-Biological activity curves and binding curves were analyzed and plotted using the nonlinear regression analysis program in the Prism software package (GraphPad Software, San Diego, CA). Binding kinetics were determined by analysis with the LIGAND program of Munson and Rodbard (13). All observations were expressed as means Ϯ S.E. Data were analyzed using the two-tailed two-way analysis of variance followed by the Newman-Keul test for multiple comparisons. A value of p Ͻ 0.05 was considered significant.

Mutagenesis of the Amino Terminus of the Motilin Receptor-In this
series of studies, we deleted sequential segments ranging in length from one to four amino acid residues from the predicted amino-terminal tail region of the motilin receptor. These started immediately after the initiator methionine residue and were extended through Gly 36 (Fig. 1). Constructs were transiently expressed in COS cells and were characterized functionally (quantitative analysis in Table 1). As shown in Fig. 2A, deletion of only three segments in this region resulted in marked reduc-tion in motilin-stimulated intracellular calcium responses. These included deletion of residues 33 through 36, adjacent to the first transmembrane segment, and deletion of each of the two cysteine residues in this region, Cys 25 and Cys 30 .
To identify potentially important residues in the region between Phe 33 and Gly 36 , each of the relevant four residues was mutated to Ala. Fig. 2B shows that in F33A, P34A, and L35A mutants the intracellular calcium responses to motilin were similar to that in wildtype motilin receptor (quantitative analysis in Table 1). In contrast, the mutation of G36A of the predicted carboxyl-terminal perimembranous end of this extracellular domain resulted in a decreased intracellular calcium response to motilin. The maximal calcium response to motilin for this mutant was only 25% of that for the wild-type motilin receptor.
Shown in Fig. 2, C and D, are the motilin competition-binding curves for each of the deletion (Fig. 2C) and Ala replacement site mutants (Fig.  2D) (quantitative analysis in Table 1). These curves correlated nicely with the activity assays for these constructs (Fig. 2, A and D), with the most dramatic reduction in binding observed for deletion mutants (⌬Cys25, ⌬Cys30, and ⌬33-36) and the G36A site mutant.
Mutagenesis of the First Extracellular Loop of the Motilin Receptor-In this region, segments ranging in length from one to six amino acid residues were deleted (Fig. 1). The constructs, representing ⌬103-104, ⌬105-110, and ⌬111C were transiently expressed in COS cells and functionally characterized (quantitative analysis in Table 1). As shown in Fig. 3A, all of the three mutants decreased the intracellular calcium response relative to that in the wild-type receptor (Fig. 3A). Consistent with this finding, none of these mutants exhibited detectable binding (Fig. 3C). These data are consistent with photoaffinity labeling data using a photolabile motilin analogue that incorporates a benzoylphenylalanine in position 1 (8).
The Cys 111 residue that was deleted in the ⌬111 construct can be aligned with a Cys residue that is conserved throughout the superfamily of G protein-coupled receptors and that contributes to the formation of the highly conserved disulfide bond linking the first and second extracellular loops of these receptors. The Cys 111 deletion mutant had a profoundly negative impact (Fig. 3, A and C).
To identify other important residues in this extracellular loop region, we mutated each of eight residues between Pro 103 and Leu 110 to Ala (Fig.  3B) and characterized them functionally in COS cells (quantitative analysis in Table 1). The mutation of P103A within the predicted aminoterminal perimembranous end of this loop and the mutation of L109A within the predicted carboxyl-terminal perimembranous end of this loop resulted in loss of the intracellular calcium responses to motilin. The maximal calcium response to motilin for these mutants was only 25% of that for the wild-type motilin receptor. Once again, motilin radioligand binding data correlated with the biological activity studies (Fig. 3D).

Mutagenesis of the Third Extracellular Loop of the Motilin
Receptor-This is a small region, consisting of only five residues. Therefore, only two deletion constructs (⌬328 -330 and ⌬331-332) were prepared (Fig. 1). They were expressed in COS cells and characterized for biological activity and binding studies (quantitative analysis in Table 1). As shown in Fig. 4A, of the two constructs tested, only the carboxyl-terminal deletion of this loop (⌬331-332) decreased the intracellular calcium response (Fig. 4A) and binding (Fig. 4C). To further explore the importance of individual residues, we mutated Tyr 331 and Phe 332 to Ala. The mutation of F332A within the predicted carboxyl-terminal end of this loop resulted in loss of intracellular calcium response to motilin (Fig. 4B) and motilin binding (Fig. 4D). This is consistent with our most recent identification of this residue as the site of covalent labeling with a photolabile motilin analogue incorporating a benzoylphenylalanine in position 5 (9).
Evidence of Surface Expression of the Functionally Impaired Mutant Motilin Receptor Constructs-To examine cell surface expression of the functionally impaired motilin receptor constructs as identified above, each of the relevant HA-tagged constructs was transfected into COS cells and immunostained with anti-HA monoclonal antibody. As shown in Fig. 5, each mutant receptor construct was shown to be expressed on the cell surface in density similar to the wild-type receptor. No fluores-cence was observed in negative control cells that had been transfected with an empty expression vector or in cells transfected with wild-type receptor but stained with the second antibody only. It should be noted that the HA-tagged motilin receptor had motilin-stimulated biological activity response similar to the wild-type receptor (data not shown).
Biological Responses to Erythromycin-In contrast to the markedly reduced biological responses to motilin for each of the four key alanine site mutants (G36A, P103A, L109A, F332A), these mutants exhibited normal intracellular calcium responses to the non-peptidyl agonist, erythromycin (Fig. 6A, Table 1). Maximal responses were not different for these constructs from values for the wild-type motilin receptor. These support the normal biosynthesis and delivery to the cell surface for each of these important constructs while at the same time supporting differential molecular determinants for peptide and non-peptidyl agonist action.
The constructs that represented the deletion of Cys residues in positions 25, 30, and 111 resulted in decreases in intracellular calcium responses to the natural peptide agonist, motilin, and also resulted in decreased responses to the non-peptidyl agonist, erythromycin (Fig. 6B, Table 1). We previously reported that deletion of Cys 235 resulted in reduced intracellular calcium responses to both motilin and erythromycin (7). These findings support the importance of the conserved disulfide bond  linking Cys 111 with Cys 235 in maintaining the receptor architecture important for responses to both peptide and non-peptidyl agonist action.

DISCUSSION
The molecular basis of ligand binding to a receptor is dependent on the structural and physicochemical characteristics of both molecules. For the superfamily of G protein-coupled receptors, the heptahelical structure and confluence of these helices in the lipid bilayer are thought to be largely conserved. However, the loop and tail regions are quite varied and likely provide the diversity of themes allowing the binding of structurally diverse natural ligands. This diversity is reflected in different modes of binding for distinct groups of ligands. It is noteworthy that the receptors with structures that are most closely related to each other often have natural ligands that are also structurally related and that bind with similar mechanisms. The motilin receptor is noteworthy because it is structurally similar to the receptors for growth hormone releasing factor and ghrelin. The latter represents the first peptide hormone found to have an octanoylation modification, which is critical for binding and biological activity. Presumably this leads to directing ghrelin to a binding domain at or within the lipid bilayer.
Current insights into the mechanism of binding of motilin have come from receptor mutagenesis and photoaffinity labeling studies (7)(8)(9). The only existing mutagenesis studies have carefully examined the predicted second extracellular loop region of the motilin receptor (7). That region was targeted for study because it is quite long in the motilin receptor and much shorter in the ghrelin receptor. The rationale was that this might be critical for the binding of a peptide but less important for the binding of the octanoylated ligand. It turned out that the only regions of importance for peptide binding were the residues at the membrane interface at each end of the long loop (Val 179 , Leu 245 and Arg 246 ) (7). The current report examines all of the other predicted extracellular regions of the motilin receptor following similar experimental strategies of segment deletion and alanine replacement mutagenesis. Here, too, the regions of importance for motilin binding and action were the residues at the membrane interface at the ends of the amino-terminal tail and extracellular loop domains (Gly 36 , Pro 103 , Leu 109 , and Phe 332 ). Although all of these regions were critical for the binding and action of motilin, their mutation had no effect on the action of erythromycin, a non-peptidyl agonist acting at this receptor (7).
Of note, this mutagenesis approach also revealed the functional importance of two cysteine residues (Cys 25 and Cys 30 ) within the amino-terminal tail domain that are likely involved in an intradomain disulfide bond within this receptor. Such a bond was not previously suspected to exist. It is particularly interesting that this bond was shown to have functional significance for both motilin and erythromycin action.
Although these data are of great interest and suggest the presence of a platform at the interface between plasma membrane and extracellular domains for natural peptide motilin binding to its receptor, these results are indirect and require complementation. Such loss-of-function studies can be explained by allosteric effects rather than as representing a site of direct ligand interaction with the receptor.
These results are nicely supported by the results of photoaffinity labeling studies in which ligand probes were used that had photolabile benzoylphenylalanine sites of covalent attachment in positions one and five (8,9). The former probe labeled motilin receptor regions between residues 99 and 129, including the first extracellular loop domain, and between residues 130 and 185, including the beginning of the second extracellular loop (8). The latter probe was shown to specifically label Phe 332 in the third extracellular loop domain (9).
The other interesting and important insight coming from the current work is the clear establishment of distinct mechanisms of binding and action of peptide and non-peptidyl agonists acting at the motilin receptor. Erythromycin is a non-peptidyl motilin receptor agonist that is used therapeutically as a prokinetic agent in motility disorders (14,15). Disruption of motilin binding and biological activity by mutagenesis of the perimembranous residues in the amino terminus and each of the extracellular loops of this receptor had no negative impact on the action of erythromycin. It is likely that such ligands will bind to intramembranous regions of this receptor. The erythromycin data also represent an important positive control for the normal folding and architecture of the mutant motilin receptors that disrupted motilin binding and action.
In conclusion, we have now demonstrated that residues at each of the predicted perimembranous regions of the amino-terminal tail and extracellular loop domains of the motilin receptor contribute to natural peptide ligand binding and action and are not important for non-peptidyl erythromycin action at this receptor. Additionally, we report evidence for a functionally critical disulfide bond within the amino-terminal tail of this receptor.