Two Cytoplasmic Loops of the Glucagon Receptor Are Required to Elevate cAMP or Intracellular Calcium*

The glucagon receptor is a member of a distinct class of G protein-coupled receptors (GPCRs) sharing little amino acid sequence homology with the larger rhodopsin-like GPCR family. To identify the components of the glucagon receptor necessary for G-protein coupling, we replaced sequentially all or part of each intracellular loop (i1, i2, and i3) and the C-terminal tail of the glucagon receptor with the 11 amino acids comprising the first intracellular loop of the D4 dopamine receptor. When expressed in transiently transfected COS-1 cells, the mutant receptors fell into two different groups with respect to hormone-mediated signaling. The first group included the loop i1 mutants, which bound glucagon and signaled normally. The second group comprised the loop i2 and i3 chimeras, which caused no detectable adenylyl cyclase activation in COS-1 cells. However, when expressed in HEK 293T cells, the loop i2 or i3 chimeras caused very small glucagon-mediated increases in cAMP levels and intracellular calcium concentrations, with EC50values nearly 100-fold higher than those measured for wild-type receptor. Replacement of both loops i2 and i3 simultaneously was required to completely abolish G protein signaling as measured by both cAMP accumulation and calcium flux assays. These results show that the i2 and i3 loops play a role in glucagon receptor signaling, consistent with recent models for the mechanism of activation of G proteins by rhodopsin-like GPCRs.

The glucagon receptor is a member of a distinct class of G protein-coupled receptors (GPCRs) sharing little amino acid sequence homology with the larger rhodopsin-like GPCR family. To identify the components of the glucagon receptor necessary for G-protein coupling, we replaced sequentially all or part of each intracellular loop (i1, i2, and i3) and the C-terminal tail of the glucagon receptor with the 11 amino acids comprising the first intracellular loop of the D4 dopamine receptor. When expressed in transiently transfected COS-1 cells, the mutant receptors fell into two different groups with respect to hormone-mediated signaling. The first group included the loop i1 mutants, which bound glucagon and signaled normally. The second group comprised the loop i2 and i3 chimeras, which caused no detectable adenylyl cyclase activation in COS-1 cells. However, when expressed in HEK 293T cells, the loop i2 or i3 chimeras caused very small glucagon-mediated increases in cAMP levels and intracellular calcium concentrations, with EC 50 values nearly 100-fold higher than those measured for wild-type receptor. Replacement of both loops i2 and i3 simultaneously was required to completely abolish G protein signaling as measured by both cAMP accumulation and calcium flux assays. These results show that the i2 and i3 loops play a role in glucagon receptor signaling, consistent with recent models for the mechanism of activation of G proteins by rhodopsin-like GPCRs.
The peptide hormone glucagon plays a pivotal role in the maintenance of metabolic homeostasis. To achieve its intracellular effects, glucagon must bind to a glycoprotein receptor that spans the plasma membrane. The glucagon receptor belongs to the superfamily of heptahelical, transmembrane GPCRs 1 that makes up approximately 1-2% of the total number of genes in the human genome (1).
Characterization of GPCRs within a particular receptor class has been facilitated by the fact that many structural motifs and functional domains have been conserved. For example, eight highly conserved amino acid residues within the transmembrane domains of GPCRs within the rhodopsin-like class, which includes visual opsins, biogenic amine receptors, and various chemokines, were identified (2,3). These residues permit precise alignments of receptor primary structures so that results of structure-function studies can be generalized in many cases. For example, the location of the ligand-binding domain (4) and the mechanism of agonist-induced receptor activation (5) are probably conserved for all GPCRs within the rhodopsin-like class. Rhodopsin and the ␤-adrenergic receptor have been widely studied as prototypical GPCRs (6,7).
However, the glucagon receptor belongs to a separate class of related GPCRs that includes receptors for glucagon-like peptide-1, secretin, vasoactive intestinal peptide, calcitonin, calcitonin gene-related peptide, PTH/PTH-related peptide, growth hormone-releasing hormone, pituitary adenylyl cyclase-activating peptide, glucose-dependent insulinotropic peptide, and corticotrophin-releasing factor (8 -10). They share almost no conserved sequences with the larger rhodopsin-like class of GPCRs. Even the familiar, rigorously conserved eight residues and other motifs such as the C-terminal palmitoylation sites are absent. Thus, few studies of rhodopsin or biogenic amine receptors provide applicable structural-functional information about the ligand-binding domain or the molecular mechanism underlying the G protein coupling of the glucagon receptor and other receptors in its unique GPCR subfamily. In previous studies (11,12) of the glucagon receptor, we identified general regions involved in ligand binding and cell surface expression, focusing on the extracellular and transmembrane domains. Our current work attempts to identify regions of the intracellular domain of the glucagon receptor involved in G protein coupling and signal transduction. Furthermore, we test the hypothesis that two cytoplasmic loops (loops i2 and i3) are required together to obtain the full effect of glucagon-induced adenylyl cyclase activation and mobilization of intracellular calcium.
We constructed a series of chimeras between the rat glucagon receptor and the human D4 dopamine receptor (Fig. 1). Each region of the glucagon receptor to be studied was replaced with all or part of the 11 amino acids comprising the first intracellular loop of the D4 dopamine receptor. This sequence was chosen because: (i) it shared no homology with any part of the intracellular portion of the glucagon receptor; (ii) the periodicity of charged residues did not match that of any intracellular region of the glucagon receptor; and (iii) it was not likely to have a role in signal transduction (13). The mutant receptor genes were expressed in cell culture and evaluated for cell surface expression, ligand binding, and the ability to elevate cellular second messengers cAMP and calcium. We report that a tripeptide sequence in loop i1 and the membrane-proximal portion of the C-terminal tail are required for proper intracellular processing and cell surface expression. Furthermore, we show that loop i2 and loop i3 are each involved in coupling to G s . However, only simultaneous replacement of both loops eliminated completely glucagon-induced signal transduction as measured by elevation of intracellular cAMP and calcium levels. This finding is consistent with the mechanism of receptor-G protein coupling proposed for the rhodopsin-like class of GPCRs (14 -16). Namely, at least two intracellular loops of the glucagon receptor (loops i2 and i3) act synergistically to bind G s and catalyze guanine-nucleotide exchange.

EXPERIMENTAL PROCEDURES
Construction of the Rat Glucagon Receptor Mutants-The construction of the synthetic rat glucagon receptor gene (GenBank TM /EMBL Data Bank accession number U14012) was previously reported (11,17). Site-directed mutagenesis of the gene was performed using a combination of restriction fragment replacement "cassette mutagenesis" (18) and PCR-based mutagenesis in a modified pGEM-2 (Invitrogen) cloning vector (17). All mutants were screened by restriction endonuclease analysis and positive clones were confirmed using Taq FS dye terminator cycle fluorescence-based sequencing on a Perkin Elmer/Applied Biosystems Model 377A DNA Stretch Sequencer (Perkin Elmer). The receptor genes were then subcloned into the eukaryotic expression vector pcDNA3 (Invitrogen). The mutant receptors are shown schematically in Figs. 1 and 2.
The four loop i1 receptor chimeras (G1D1, G1D1n, G1D1i, and G1D1c) and the point mutant H178R, were prepared by replacing a 76-base pair BssHII-BstEII restriction fragment with synthetic duplexes that encoded the desired amino acid residues. In G1D1, the first intracellular loop of the glucagon receptor (residues Lys 169 -Ile 177 ) was replaced with the undecapeptide TERALQTPTNS, which corresponds to the first intracellular loop of the human D4 dopamine receptor (2,19). In G1D1n, the tripeptide Lys 169 -His 171 was replaced with TERA. In G1D1i, the tripeptide Cys 172 -Arg 174 was replaced with LQT. In G1D1c, the tripeptide Asn 175 -Ile 177 was replaced with PTNS. The successive replacements corresponded to the N-terminal, intermediate, and Cterminal portions of loop i1, respectively. The loop i2 chimera, G2D1, was prepared by restriction fragment replacement. An 82-base pair SpeI-NsiI restriction fragment was replaced with a synthetic duplex in which base pairs coding for amino acid residues Ser 252 -Glu 261 of the glucagon receptor were substituted with a sequence coding for the first intracellular loop of the D4 dopamine receptor (TERALQTPTNS). The loop i3 chimera, G3D1, was prepared by first introducing a BglII restriction site into the synthetic gene in the pGEM-2 cloning vector using PCR (17). The 101-base pair PacI-BglII restriction fragment was then replaced with a synthetic duplex in which the sequence coding for amino acid residues Leu 334 -Ala 349 was replaced with a sequence coding for the first intracellular loop of the D4 dopamine receptor (TER-ALQTPTNS). The double-loop chimera, G23D1, was prepared by combining the appropriate segments of G2D1 and G3D1 at a PacI site. The C terminus chimera, GCD1, was prepared by replacement of a 63-base pair AlwNI-BspEI restriction fragment with a synthetic duplex in which codons for amino acid residues Glu 407 -Arg 415 were substituted with a sequence encoding the first intracellular loop of the D4 dopamine receptor (TERALQTPTNS) followed by two successive stop codons. The construction of truncation mutants T7a and T7b was described previously (11).
Characterization of Mutant Glucagon Receptors-The characterization of the antipeptide, antireceptor antibody DK-12 was reported previously (11,12). Glucagon receptor and mutant receptor genes were expressed in COS-1 and HEK 293T cells by transient transfection using LipofectAMINE (Life Technologies, Inc.). Preparation of membranes from transfected COS-1 cells was carried out as described previously (11). The treatment of expressed glucagon receptor and mutants with glycosidases was described previously (11). Immunoblot analyses of expressed glucagon receptor and receptor mutants were carried out essentially as described (11,17). Competitive binding studies with 125 I-glucagon were performed as described (17). Adenylyl cyclase activity was determined by measuring intracellular cAMP levels as a function of hormone concentration as reported (11). Assays of glucagon-dependent increases in intracellular calcium concentration in HEK 293T cells were performed as follows. Cells were washed once with phosphate-buffered saline (137 mM NaCl, 8.0 mM Na 2 HPO 4 , 2.7 mM KCl, 1.5 mM KH 2 PO 4 ) 36 -48 h after transfection and then each 100-mm plate was incubated in 5 ml of phosphate-buffered saline for 15-20 min at 37°C. The cells were collected either by tituration or scraping and added to 10 ml of Dulbecco's modified Eagle's medium supplemented with 8% newborn calf serum, 2% fetal bovine serum in Petri dishes at 37°C for 2-3 h. The cells were collected in 17 ϫ 120-mm conical tubes, pelleted by spinning at 2,000 rpm for 2 min, washed twice in EBSS-H buffer (125 mM NaCl, 5 mM KCl, 1 mM MgSO 4 , 1 mM NaH 2 PO 4 , 26 mM HEPES, 5.6 mM glucose, 0.1% (w/v) bovine serum albumin, 2 mM CaCl 2 ), and resuspended in 1 ml of EBSS-H at a density of 0.5-1.0 ϫ 10 6 cells/ml. The fluorophore Fluo-3/AM (Molecular Probes, Inc.) was added to a final concentration of 2 M and the cells were incubated for 60 -75 min at room temperature. The cells were then pelleted and resuspended in 1 ml of EBSS-H, washed again, and incubated in EBSS-H for 30 -45 min. Just prior to the assay, the cells were pelleted and transferred to cuvettes at a density of 0.25-0.5 ϫ 10 6 cells/ml. Fluorescence due to intracellular calcium was measured at room temperature in a Hitachi F-2000 fluorescence spectrophotometer using excitation and emission wavelengths of 505 and 525 nm, respectively. Ligand binding, cAMP, and calcium studies were performed at least three times on independent samples to verify reproducibility. For graphical representation of an individual experiment, the data at each concentration of glucagon were fit to a 4-parameter logistic function of the form: 50 and EC 50 values were calculated from the inflection point (c) of the best-fit curve (Sigmaplot, Jandel Scientific Software).

Preparation and Expression of Mutant Glucagon Receptor
Genes-The cytoplasmic surface of the glucagon receptor consists of three loops and a C-terminal tail (Fig. 1). Five mutant receptors were constructed in which the first intracellular loop of the human D4 dopamine receptor was substituted for one or more portions of the cytoplasmic domain of the rat glucagon receptor. G1D1, G2D1, G3D1 signify the mutant receptors in which the first, second, or third intracellular loops of the glucagon receptor were replaced, respectively ( Fig. 1). In G23D1, both the second and third intracellular loops were replaced. In GCD1, the entire C-terminal tail was replaced. Additional mutants were constructed to study the role of specific segments of intracellular loop i1 of the glucagon receptor (Fig. 2). Successive tripeptides from the N-terminal, intermediate, and Cterminal portions of the loop were replaced with the analogous portions from loop i1 of the D4 receptor. Since loop i1 of the D4 receptor is one amino acid longer than loop i1 of the glucagon receptor, three of the four loop i1 chimeras are longer than the original receptor (Fig. 2). The primary structures of all mutants are shown schematically in Figs. 1 and 2. The construction, expression, and characterization of the truncation mutants T7a and T7b were described previously (11). Mutants T7a and T7b have 70 and 42 amino acid residues, respectively, deleted from the C terminus of the glucagon receptor. All mutant receptor genes were expressed in COS-1 cells by transient transfection.
Characterization of the Expressed Glucagon Receptor Mutants with Glycosidases-Glycosidase analysis was employed to determine if alterations of the cytoplasmic loops affected the ability of the mutant receptors to reach the cell surface. Total cellular membranes were isolated by sucrose-gradient centrifugation. Immunoblot analysis of the membrane preparations using the DK-12 antibody and the effects of glycosidase treatment are shown in Fig. 3. All mutant receptors immunoreacted with the DK-12 antibody. This behavior was expected since the DK-12 epitope resides on the extracellular surface of the receptor, which was not altered by any of the mutations. The enzyme N-glycosidase F cleaves all asparagine-linked carbohydrates, leaving an aspartic acid in place of the original asparagine. The enzyme Endoglycosidase H (Endo H) cleaves only high mannose carbohydrates to yield asparagine-linked N-acetylglucosamine, but cannot remove N-linked complex carbohydrates. The glucagon receptor, like other glycoproteins, is modified by high-mannose carbohydrate moieties when originally inserted into the endoplasmic reticulum, and is therefore sensitive to both enzymes. However, as the receptor is transported through the Golgi apparatus to the cell surface, its glycosylation undergoes progressive modification to the complex carbohydrate form, losing its sensitivity to Endo H, but retaining its susceptibility to N-glycosidase F cleavage (20,21). We used this dif-ference in endoglycosidase sensitivity to demonstrate the degree of cell surface expression of the receptor mutants, as reported for a different set of glucagon receptor mutants in a previous study (11). Fig. 3A shows the results of an experiment to test the effect of endoglycosidase treatment on the three loop chimeras (G1D1, G2D1, and G3D1) and the C-terminal tail chimera (GCD1). Fig. 3B shows the effect on the double-loop replacement mutant G23D1. The loop i1 mutants (G1D1n, G1D1i, and G1D1c) are presented in Fig. 3C. The pattern of glycosidase sensitivity differed among the mutant receptors, but fell into two distinct groups. The first group consisted of mutant receptors that were sensitive to both N-glycosidase F and Endo H, indicating that they were not expressed on the cell surface. This group included mutant receptors G1D1 and GCD1 (Fig.  3A), and G1D1c (Fig. 3C). Mutants G1D1 and G1D1c had apparent molecular weights equal to that of the wild-type receptor, which indicated that a mutant receptor of the expected size was in fact being translated and inserted into the cell membranes. The apparent molecular weight of GCD1 was slightly less than that of the wild-type receptor as expected. The second group consisted of mutant receptors that were sensitive to N-glycosidase F but not Endo H, indicating that they were processed and transported to the cell surface. This group included G2D1, G3D1, G23D1, G1D1n, and G1D1i. The mutant receptor H178R, with a single amino acid change in loop i1, also fell into this category. All six of these mutant receptors were expressed on the cell surface. They also had the same apparent molecular weights as the wild-type receptor as judged by their electrophoretic mobilities.
Binding of 125 I-Glucagon to Transfected Cell Membrane The intracellular portions of the receptor subject to mutagenesis are in a dashed box. The first, second, and third intracellular loops, and the first 10 amino acids of the C-terminal tail, are represented, respectively, by i1, i2, i3, and Ct, and comprise the amino acids shown below. The numbering is based on the sequence of the rat glucagon receptor (38). Chimeric mutants G1D1, G2D1, G3D1, and GCD1, in the solid boxes, were constructed by substituting the native sequence at the positions indicated with the 11 amino acids of the loop i1 of the human D4 dopamine receptor as depicted in bold (19). Loop i1 of the D4 dopamine receptor, D4 i1, is shown in the solid box on the top left. Mutant G23D1 had replacements of both loops i2 and i3. Mutant receptors T7a and T7b were truncated in the C-terminal tail. Their construction was described previously (11). Preparations-The results of competitive ligand-binding experiments are presented in Fig. 4. Increasing concentrations of glucagon were used to compete with 125 I-glucagon for binding to COS-1 cell membranes containing expressed glucagon receptor or mutant receptors. The cytoplasmic loop mutations were not expected to affect directly the ligand-binding domain of the glucagon receptor, which consists primarily of the N-terminal tail, first extracellular loop, and the transmembrane helices (11,12,22). Each of the mutant glucagon receptors that reached the cell surface as judged by endoglycosidase sensitivity analysis bound glucagon with essentially normal affinity (Fig. 4A). The data points from experiments with these receptors were well approximated by a 4-parameter logistic function characteristic of ligand-receptor equilibrium binding (Fig. 4A). These data were used to determine the concentrations of unlabeled glucagon required to displace 50% of receptor-bound 125 Iglucagon (IC 50 values) ( Table I (11,(23)(24)(25). As expected, the mutant receptors with immature glycosylation (G1D1c, G1D1, and GCD1) did not bind glucagon with measurable affinity (Fig.  4B).
Glucagon-dependent Stimulation of Adenylyl Cyclase in Transfected COS-1 Cells-The mutant glucagon receptors were assayed for the ability to mediate the activation of adenylyl cyclase (Fig. 5). The amount of glucagon-dependent cAMP accumulation in COS-1 cells expressing the glucagon receptor or receptor mutants reflects the level of adenylyl cyclase activation and thereby indicates the efficiency of receptor coupling to G s . Of the six mutant receptors that reached the cell surface and bound glucagon (Fig. 4A), only three (G1D1n, G1D1i, and H178R) mediated a glucagon-dependent increase in cAMP concentration (Fig. 5A). The basal cAMP levels in COS cells expressing these receptors were similar to cells in which the wild-type glucagon receptor was expressed (data not shown). This result indicates that none of the mutant receptors exhibited constitutive activity. The cAMP levels obtained upon glucagon treatment of the cells expressing the mutant receptors were consistently 50 -100% of the maximum change seen in cells expressing wild-type receptor.
In the assay for adenylyl cyclase activation, the loop i2 and i3 chimeras (G2D1, G3D1, and G23D1) did not increase cAMP concentrations above basal levels (Fig. 5B). Each of these three receptor mutants was expressed on the cell surface and bound glucagon as judged by endoglycosidase analysis and hormone competition-binding experiments, respectively. Thus, these three receptor mutants were defective in glucagon-dependent G protein coupling. The extent of the G protein-coupling defect of these mutants was evaluated in more detail as described below. As expected, the receptor mutants that failed to reach the cell surface and bind glucagon also failed to mediate an increase in cellular cAMP levels (Fig. 5B). The effective glucagon concentration at 50% maximal stimulation of adenylyl cyclase (EC 50 value) for each of the mutant receptors is presented in Table I. Receptor mutants G1D1n, G1D1i, and H178R displayed EC 50 values nearly identical to that of the wild-type receptor (4.8 nM).
Glucagon-dependent Elevation of Intracellular Calcium Concentrations in Transiently Transfected HEK 293T Cells-To further characterize the difference in signaling among the mutant glucagon receptors, we compared their ability to stimulate increases in [Ca 2ϩ ] i in HEK 293T cells. Transfected cells were loaded with the calcium-sensitive dye Fluo-3/AM, which is converted to Fluo-3 by intracellular esterases. The response to changes in [Ca 2ϩ ] i was measured in arbitrary units of fluorescence at 525 nm. Fig. 6 shows that both carbachol binding to endogenous muscarinic receptors and glucagon binding to transiently transfected glucagon receptors led to increases in [Ca 2ϩ ] i . The profiles of the traces representing the change in [Ca 2ϩ ] i as a function of time were similar in character for both carbachol and glucagon. The responses to both glucagon and carbachol were only moderately reduced by the absence of extracellular calcium (data not shown), indicating that the increase in [Ca 2ϩ ] i came predominantly from intracellular storage sites. After drug addition, the [Ca 2ϩ ] i levels increased rapidly, reaching a peak after approximately 10 s. The subsequent decrease in [Ca 2ϩ ] i followed a pseudo first-order exponential decay and reached a new baseline 50 -100 s after the drugs originally were added. As the concentration of an added drug was decreased, a dose-dependent decrease in fluorescence peak height was observed, with an associated increase in both the time-to-peak and the half-life of the exponential decay of the [Ca 2ϩ ] i signal. Accurate dose-response data were collected by measuring the effect of various concentrations of glucagon on the [Ca 2ϩ ] i fluorescence signal. Fig. 7 shows the quantification of the capability of each mutant glucagon receptor to mediate a ligand-dependent elevation of [Ca 2ϩ ] i . For a positive control that could be used to standardize the response, 1 mM carbachol was added prior to glucagon, as demonstrated in Fig. 6. The height of each fluorescence peak for glucagon was then divided by the carbachol peak. This ratio was calculated over a range of serial dilutions of glucagon, permitting the generation of a dose-response curve showing the relative increase in [Ca 2ϩ ] i versus glucagon concentration. EC 50 values were determined from the dose-response curves in Fig. 7 and are presented in Table I. The calcium flux assay is about 10-fold more sensitive than the cAMP accumulation assay as judged by the EC 50 values measured for the glucagon receptor, 4.8 nM versus 370 pM, respectively (Table I).
The glucagon receptor mutants T7a and T7b, which were truncated at different points of their C termini, were shown previously to couple to G s with essentially wild-type efficacy (11). These mutant receptors also mediated glucagon-dependent calcium fluxes (Fig. 7A). All mutant receptors that were capable of binding glucagon and elevating intracellular cAMP in COS-1 cells (G1D1n, G1D1i, and H178R) also increased intracellular calcium levels with EC 50 values similar to that of the wild-type receptor (Fig. 7B, Table I). As expected, mutant receptors that did not reach the cell surface and bind glucagon (G1D1c, G1D1, and GCD1) produced no change in calcium levels in response to glucagon (data not shown).
As described above, three mutant receptors (G2D1, G3D1, and G23D1) bound glucagon, but did not cause an elevation of cAMP levels in COS-1 cells. However, for mutant receptors G2D1 and G3D1, a dose-dependent increase in [Ca 2ϩ ] i was observed when assayed in the more sensitive calcium flux assay (Fig. 7C). The dose-response curves for G2D1 and G3D1 The concentration of unlabeled glucagon required to displace 50% of receptor bound 125 I-glucagon. The average of three (G1D1n, G1D1i, H178R, and G2D1), four (G3D1), or seven (GR) independent competition-binding curve determinations in which each point was measured in duplicate is given.
b The effective glucagon concentration at 50% stimulation of adenylyl cyclase. The average of three (G1D1n, G1D1i, and H178R) or six (GR) independent cAMP accumulation determinations in which each point was measured in duplicate is given.
c The effective glucagon concentration at 50% stimulation of intracellular calcium flux. The average of three (H178R, G2D1, T7a, and T7b), four (G1D1n, G1D1i, and G3D1), or 16 (GR) independent calcium flux experiments is given. d Significantly different from the EC 50 value measured for wild-type, p Ͻ 0.05 by Student's t test.

FIG. 4. Competitive displacement assay of 125 I-glucagon binding to COS-1 cell membranes expressing native or chimeric mutant glucagon receptor genes.
COS-1 cells were transiently transfected with native rat glucagon receptor gene (GR) or chimeric mutants. Cellular membranes were isolated and incubated with 125 I-glucagon and the indicated concentrations of unlabeled glucagon as described under "Experimental Procedures." Data are presented as percentage of total binding of the radiolabeled hormone at the lowest glucagon concentration versus the log of glucagon concentration. Each symbol represents the mean of duplicate determinations and was curve-fitted based on a single ligand-binding site model as described under "Experimental Procedures." Data shown are from independent experiments and are representative of at least three determinations. A, the concentration of unlabeled glucagon required to displace 50% of receptor-bound 125 I-glucagon (IC 50 value) was calculated for each mutant (G1D1n, G1D1i, H178R, G2D1, G3D1, and G23D1) and is shown in Table I. B, membranes from COS-1 cells transiently transfected with the glucagon receptor gene (GR), chimeric mutants (G1D1c, G1D1, and GCD1), or control vector (pcDNA3) were incubated with radiolabeled glucagon and increasing concentrations of unlabeled glucagon. Data shown are representative of at least two independent determinations. Each of the three mutations resulted in the complete inability of the mutant receptor to bind glucagon.
were shifted to the right and gave EC 50 values of 13.8 and 13.0 nM, respectively, indicating that the mutant receptors were defective in signaling (Fig. 7C, Table I). Mutant receptor G23D1, which combined both the loop alterations of G2D1 and G3D1, bound glucagon but did not cause a measurable change in [Ca 2ϩ ] i even at nearly 3 M glucagon concentration (Fig. 7C).
We tried to identify the mechanism by which the mutants G2D1, G3D1, but not G23D1, increased [Ca 2ϩ ] i . All three mutants failed to increase cAMP in COS-1 cells under the conditions of the assay (Fig. 6). However, mutant receptors G2D1 and G3D1 both induced a very small but statistically significant increase in cAMP when assayed in HEK 293T cells (Fig. 8). In contrast, mutant receptor G23D1 had no affect on cAMP level in either COS-1 or HEK 293T cells (Fig. 8). Since activation of adenylyl cyclase stimulates increases in [Ca 2ϩ ] i in several cell types (26,27), our results do not rule out that G2D1 and G3D1 elevated [Ca 2ϩ ] i via receptor-mediated increases in cAMP.

DISCUSSION
To determine the components of the intracellular domain of the glucagon receptor required for proper biosynthesis and signal transduction, we prepared a series of eight glucagon/D4 dopamine receptor chimeras. A schematic representation of the mutant receptors is shown in Fig. 1. The N-and C-terminal domains and the seven putative transmembrane helices (TM1-TM7) with intervening extracellular (e1-e3) and intracellular loops (i1-i3) are based on previous models of GPCRs (28 -30). For GPCRs in the rhodopsin-like family, also called Class I GPCRs, the majority of studies indicates that G protein activation is regulated by amino acids in loop i2 and the N-and C-terminal regions of loop i3. The membrane-proximal portion of the C terminus also may form part of the receptor-G protein interface (31,32). In contrast, much less is known about the  (G1D1n, G1D1i, or H178R). Each symbol represents the mean of duplicate determinations and is plotted as percentage of total cAMP accumulation mediated by wild-type receptor (GR) at the highest ligand concentration versus the log of glucagon concentration. Data shown are from independent experiments and are representative of at least three determinations. The effective concentrations at 50% maximal stimulation of adenylyl cyclase (EC 50 values) for cells expressing the glucagon receptor or the chimeric mutants were determined as described under "Experimental Procedures" and are reported in Table I. B, COS-1 cells were transfected with vector containing the synthetic glucagon receptor gene (GR), chimeric mutant receptors (G1D1c, G1D1, G2D1, G3D1, GCD1, and G23D1), or control vector (pcDNA3). The increase in intracellular cAMP levels when cells were incubated with increasing concentrations of glucagon was determined as above. Each symbol represents the mean of duplicate determinations and is plotted as the percentage of total cAMP accumulation mediated by wild-type receptor (GR) at the highest ligand concentration versus the log of glucagon concentration. Data shown are from independent experiments and are representative of at least two determinations. None of the six mutant receptors was able to increase cAMP above basal levels in the presence of glucagon.
role of these intracellular regions in the glucagon-like receptor family, also called Class II GPCRs, in G protein coupling and signal transduction. The Class II peptide hormone receptors have little sequence homology with receptors in the Class I family. In terms of the intracellular domains, neither loop size nor primary structure is conserved between the two receptor classes. The differences between the structures of the cytoplasmic domains of the Class I and Class II GPCRs are discussed below as related to surface expression, ligand binding, and signal transduction of the glucagon receptor and receptor mutants presented in this report.
The First Intracellular Loop-The high conservation of residues within each superfamily suggests that the first intracellular loop might play a structural role, as seen with the ␤ 2adrenergic receptor (33). Structural models, mutagenesis, and peptide studies on many receptors in the rhodopsin-like family have indicated that loop i1 does not have a direct role in G protein coupling (30,32). Class II receptors may be different, using loop i1 for structure and G protein coupling. Although highly conserved, several receptor subtypes have an alternatively spliced version in which 10 -30 amino acids are inserted into loop i1. These modifications in the human calcitonin and corticotrophin-releasing factor receptors affect G protein coupling (34,35). Like other receptors in its immediate family, the glucagon receptor appears to exist as only one isoform (36), so the observations regarding splice variants presented above may not apply directly. Moreover, a loop insertion may alter the tertiary structure of other intracellular loops to affect signaling. In the present study, we substituted residues from the loop i1 of the D4 dopamine receptor, a distantly related Class I GPCR, for residues of loop i1 of the glucagon receptor. The mutant receptors fell into one of two categories. Mutant receptors G1D1c and G1D1 were translated and inserted into the endoplasmic reticulum membrane but were not expressed on the cell surface. These receptors did not bind glucagon or mediate signaling. Mutant receptors G1D1n, G1D1i, and H178R were expressed normally on the cell surface. These mutant receptors bound glucagon and coupled to G proteins to elevate cAMP levels and cause calcium flux normally. These observations support the suggestion that the high homology in loop i1 reflects an essential role for these residues in protein folding and surface expression, but not for direct receptor-G protein coupling (37). In particular, the analysis of the loop i1 sequences of the mutant receptors (Fig. 2) suggests that the FIG. 7. Dose-response curves for the glucagon-mediated calcium flux in HEK 293T cells transiently transfected with native or chimeric mutant glucagon receptor genes. HEK 293T cells were transiently transfected with vectors containing the synthetic glucagon receptor gene (GR) or chimeric mutant receptors. Each symbol represents the ratio of the height of the peak for glucagon in a given experiment to that of the peak for a fixed concentration of carbachol (1 mM) in the same experiment. The data for each receptor were normalized, with 100% activity defined as the glucagon peak/carbachol peak ratio measured at the highest concentration of glucagon (ϳ3 M). Each trace is representative of at least three different sets of experiments. The effective concentrations at 50% maximal stimulation of calcium flux (EC 50 values) for cells expressing the glucagon receptor or the chimeric mutants were determined as described under "Experimental Procedures" and are reported in Table I. A, C-terminal tail mutants T7a and T7b. B, loop i1 mutants G1D1n, G1D1i, and H178R. C, loop i2 and i3 mutants G2D1, G3D1, and G23D1. The chimeric receptors G2D1 and G3D1 were able to mediate a small but statistically significant increase in cAMP concentrations in response to glucagon. The chimeric receptor G23D1 did not increase cAMP. The inset shows the comparison of mutants G2D1, G3D1, and G23D1 over a smaller range. The * indicates p Ͻ 0.05, and ** indicates p Ͻ 0.005. tripeptide Asn-Tyr-Ile in the loop i1 of the glucagon receptor is necessary for cell surface expression.
Of particular significance is the histidine at the junction between loop i1 and TM2. This histidine is absolutely conserved across all members of the glucagon receptor subfamily, but substitution of the histidine by arginine has not yielded consistent results in in vitro experiments. In the PTH/PTHrelated peptide receptor, the His-Arg replacement mutant constitutively activated G s but lost G q coupling (38). Whereas the vasoactive intestinal peptide receptor with the His-Arg replacement also demonstrated constitutive G s activation (39), the same mutation in the glucagon-like peptide-1, glucose-dependent insulinotropic peptide, and calcitonin receptors did not increase receptor-G protein coupling (40 -42). A previous report in which H178R was constructed in the glucagon receptor showed that there was a dose-dependent constitutive activation based on receptor DNA and surface expression levels (43). In the present study, the H178R point mutation of the glucagon receptor did not have a significant effect on ligand binding or G s coupling, and basal cAMP levels were not elevated. These results suggest that the histidine at the junction between loop i1 and TM2 may be important for some structural component of receptor-G protein coupling that depends considerably on environment such as cell type and receptor number.
Second and Third Intracellular Loops-Loops i2 and i3 are the principal domains of GPCRs in the rhodopsin-like family that interact with G proteins. However, the diversity of sequence and loop size, even among related receptors, has made it difficult to identify a specific set of sequences that dictates the coupling profile of a given receptor. The initial studies identified loop i2, the N-terminal region of loop i3, and particularly the C-terminal region of loop i3 as being important for coupling to all three G protein classes, G s , G i/o , and G q (44). Structurally, the N-and C-terminal regions of loop i3 from G s -, G i/o -, and G q -coupled receptors have a periodicity of positively charged and hydrophobic residues typical of amphipathic ␣-helices (45). The precise regions necessary for certain receptor functions have been mapped for particular receptors. Based on primary structure alignments and site-directed mutagenesis studies, roles for specific amino acids necessary for G protein coupling were posited. An amphipathic portion of loop i2 acts as a conformational switch that enables a receptor to interact with G proteins after ligand binding (46). At loop i3, TM helices 5 and 6 continue into the cytoplasm to form a surface for G protein interaction. The hydrophobic amino acids of the N-and C-terminal regions of loop i3 form a pocket to recruit G proteins selectively, and basic amino acids help activate nucleotide exchange (47). Other studies using chimeric mutants of related but functionally different receptors indicated that loop i2 and the N-and C-terminal regions of loop i3 were interacting with the five C-terminal amino acids of G protein ␣-subunits to mediate signal transduction (48). Crystallographic studies support these conclusions and suggest that the positively charged residues in loops i2 and i3 might interact with the C-terminal peptide of the G protein ␣-subunit and possibly with the negative electrostatic surface of blade 7 of the ␤-propeller structure of the ␤-subunit (49).
Extrapolating this model of receptor-G protein interaction from the rhodopsin-like class to the other classes is not straightforward. The Class II receptors lack several important motifs found in the Class I receptors that are involved in G protein coupling. The glucagon receptor has no (Glu/Asp)-Arg-Tyr sequence and no positively charged residues in loop i2. In addition, the periodicity of charges in the N-terminal domain of loop i3 is similar to that of a ␤-sheet rather than to that of an ␣-helix (50). The specificity of G protein coupling in the gluca-gon receptor family appears to be dictated in part by loop i2 and the N-and C-terminal regions of loop i3 (51)(52)(53)(54)(55), which is similar to the case in the Class I receptors. However, none of the studies of Class II receptors to date identify a set of residues necessary for specific G protein interaction, as opposed to residues that might play a general structural role in maintaining a surface for G protein coupling.
We substituted an entire intracellular loop from another GPCR into loop i2 and loop i3 of the glucagon receptor to evaluate the contribution of these regions to signaling by the glucagon receptor. The three loop i2 and i3 mutants in the present study (G2D1, G3D1, and G23D1) represent chimeric glucagon receptors in which the second, third, or both loops were replaced with the 11 amino acids of the D4 dopamine receptor loop i1. Deletions, especially ones that remove most of an intracellular loop, may disrupt the structural environment of residues distant from the original site and prevent firm conclusions about the role of particular regions in G protein coupling. Chimeras and point mutations address the role of replaced residues without the same risk of major structural perturbations. All three mutant receptors were expressed normally and bound glucagon with an affinity comparable to that of wild-type (Fig. 4A). In COS-1 cells, all three receptors failed to increase cAMP in response to glucagon, indicating that both loop i2 and loop i3 are involved in glucagon receptor coupling to G s (Fig. 5B). In HEK 293T cells, both G2D1 and G3D1 induced small glucagon-dependent increases in cAMP (Fig. 8). Both mutant receptors were also able to induce glucagon-dependent calcium flux, although the EC 50 values for the dose-response were increased by about 35-fold (Fig. 7, Table I). Mutant receptor G23D1 bound glucagon normally (IC 50 ϭ 72 nM), but was completely inactive in terms of G protein coupling as judged by adenylyl cyclase stimulation or calcium flux. This difference in receptor-G s coupling detected in COS-1 and HEK 293T cells is not unexpected. Such behavior typically is displayed by GPCRs expressed heterologously in these model systems (56). These results show that loop i2 and loop i3 are each involved in coupling to G s . Only simultaneous replacement of both loops in mutant G23D1 completely eliminated glucagon-induced signal transduction as measured by elevation of intracellular cAMP and calcium levels. The G23D1 mutant also did not couple to G i or G o to elevate intracellular calcium levels in HEK 293T cells despite the presence of two cytoplasmic loops of the dopamine D4 receptor. Mutant receptor chimeras with the entire putative cytoplasmic domain of the dopamine D4 receptor and the extracellular and membrane-embedded domains of the glucagon receptor did not reach the cell surface. Studies with additional chimeric receptor constructs are ongoing.
Taken together, these results are consistent with the mechanism of receptor-G protein coupling proposed for the rhodopsin-like class of GPCRs (14 -16). Namely, in the glucagon receptor, as in Class I GPCRs, intracellular loops i2 and i3 act synergistically to bind G s and catalyze guanine-nucleotide exchange. Compared with loops i2 and i3, the role of the Cterminal tail in G protein coupling is less clear and consistent across the GPCR superfamily. While both the ␤ 2 -adrenergic and angiotensin II receptors, type I GPCRs, use the N-terminal segment of the cytoplasmic tail for G protein coupling (31,57), the gonadotropin-releasing hormone activates G q without any tail whatsoever (58). Within the glucagon-like receptor family, a series of successive truncations of the PTH/PTH-related peptide receptor cytoplasmic tail showed that this domain was not necessary for coupling to either G s or G q (59,60).
We previously showed that glucagon receptor truncation mutants T7a and T7b bound ligand and increased cAMP comparably to wild-type receptors (11). The present study demon-strated that these two mutants also increased [Ca 2ϩ ] i after glucagon addition, indicating that the distal 70 amino acids of the glucagon receptor C-terminal tail do not contribute significantly to ligand binding or G protein coupling. These findings were corroborated by the characterization of a set of more extensive truncations of the C-terminal tail of the human glucagon receptor. Truncation before five conserved amino acids, Leu-Arg-Arg-Trp-Arg, at the N-terminal portion of the tail prevented cell surface expression (61). The present study supports these findings since mutant receptor GCD1, in which these five amino acids were replaced with residues from the D4 dopamine loop i1, also failed to reach the cell surface (Fig. 3A). These observations suggest that GCD1 is not expressed on the cell surface because some subset of five specific residues, most likely including the absolutely conserved Trp, is necessary for post-translational processing or trafficking within the cell.
Mechanism for Glucagon-dependent Calcium Release-All members of the glucagon-like class of GPCRs couple to G s to elevate intracellular cAMP concentrations (8 -10). Many of these receptors, including the rat glucagon receptor, also have been shown to elevate intracellular concentrations of calcium, inositol phosphates, or both. For the PTH/PTHrP and calcitonin receptors, the combination of cAMP and inositol 1,4,5trisphosphate elevations was attributed to receptor coupling to both G s and G q (62). Several reports on the glucagon receptor have demonstrated hormone-induced increases in intracellular cAMP, inositol phosphates (63), and free calcium concentrations (63,64). It has been well established that the glucagon receptor couples to G s to generate cAMP (65). The pathway leading from the activated receptor to inositol phosphate production and calcium increases is less clear. The existence of two different glucagon receptors was proposed to explain the different signal transduction pathways that were activated by glucagon in hepatocytes (66). Despite extensive efforts to identify glucagon receptor variants, only one form of the receptor without any other splice variants has been reported (36,67).
When glucagon mediates increases in both inositol 1,4,5trisphosphate and [Ca 2ϩ ] i , it suggests that the glucagon receptor behaves like the other members of its subfamily and couples to both G s and G q . However, in several systems, hormone binding to the glucagon receptor increases cAMP and [Ca 2ϩ ] i , but not inositol phosphates. In HEK 293 cells stably expressing the glucagon receptor, a calcium flux occurred after treatment with isoproterenol, glucagon, forskolin, and cAMP analogues, proving that G s -mediated activation of adenylyl cyclase was sufficient to increase [Ca 2ϩ ] i (27). Since the calcium flux occurred without a significant change in inositol phosphate concentration, G q involvement was ruled out. For systems with this signal transduction profile, it has been proposed that the glucagon receptor couples only to G s to increase cAMP, which stimulates PKA to elevate [Ca 2ϩ ] i (26,27). Similar conclusions have been made for glucagon-like peptide-1 receptor coupling in insulinoma cells (68). In the present study, mutant receptors G2D1 and G3D1 increased both cAMP and [Ca 2ϩ ] i in HEK 293T cells, but mutant G23D1 failed to affect either (Figs. 7 and 8). The linkage of cAMP and [Ca 2ϩ ] i suggests that the glucagon receptor mediates both signal transduction pathways by coupling only to G s .