Interaction of [fluorescein-Trp25]glucagon with the human glucagon receptor expressed in Drosophila Schneider 2 cells.

The human glucagon receptor was expressed at high density in Drosophila Schneider 2 (S2) cells. Following selection with G418 and induction with CuSO4, the cells expressed the receptor at a level of 250 pmol/mg of membrane protein. The glucagon receptor was functionally coupled to increases in cyclic AMP in S2 cells. Protein immunoblotting with anti-peptide antibodies revealed the expressed receptor to have an apparent molecular mass of 48 kDa, consistent with low levels of glycosylation in this insect cell system. Binding of [fluorescein-Trp25]glucagon to S2 cells expressing the glucagon receptor was monitored as an increase in fluorescence anisotropy along with an increase in fluorescence intensity. Anisotropy data suggest that the mobility of the fluorescein is restricted when the ligand is bound to the receptor. Kinetic analysis indicates that the binding of glucagon to its receptor proceeds via a bimolecular interaction, with a forward rate constant that is several orders of magnitude slower than diffusion-controlled. These data would be consistent with a conformational change upon the binding of agonist to the receptor. The combination of [fluorescein-Trp25]glucagon with the S2 cell expression system should be useful for analyzing glucagon receptor structure and function.

Glucagon is a 29-amino acid peptide produced by proteolytic cleavage of the proglucagon gene product in the A cells of the pancreas. The peptide acts at the liver to increase the rate of gluconeogenesis and glycogenolysis, in this regard serving as the major counterregulatory hormone of insulin (for review, see Ref. 1). Glucagon binds to specific receptors on the surface of hepatocytes to stimulate increases in cyclic AMP, inositol phosphate, and intracellular calcium. The rat and human glucagon receptors have been cloned (2)(3)(4) and shown to contain seven putative transmembrane domains characteristic of the G protein-coupled family of receptors (5). The glucagon receptor shares significant sequence homology with the subfamily of G protein-coupled receptors that includes receptors for glucagonlike peptide-1 and parathyroid hormone (for review, see Ref. 6). This subclass of G protein-coupled receptors bears little sequence homology to the well characterized ␤-adrenergic/rhodopsin subfamily, and relatively little is known about the molecular interactions of ligands with receptors in this class.
Glucagon itself has been the subject of a wide variety of physical studies such as x-ray crystallography (7), NMR (8), and circular dichroism (9). However, these studies have been performed in various lipid or detergent solutions, which affect the conformation of the peptide. In addition, it has been observed that the secondary structure of glucagon is dependent on the concentration of peptide used in the experiment (9). For these reasons, the structure determined by these techniques may not reflect the physiological state of glucagon as it interacts with its receptor.
As is the case for most G protein-coupled receptors, biophysical and structural characterization of the glucagon receptor has been hampered by an inability to produce sufficient quantities of receptor protein. In the present study, we have expressed the glucagon receptor at high density using the Drosophila Schneider 2 (S2) cell system (10). [fluorescein-Trp 25 ]Glucagon (11) has been used as a tool for obtaining kinetic and structural information on the human glucagon receptor. The high levels of expression of human glucagon receptor obtained in S2 cells have allowed us to directly monitor changes in the fluorescence properties of this ligand as it binds to the receptor. This system has proven useful in understanding the environment of the ligand binding site of the human glucagon receptor and in monitoring conformational changes in both the receptor and the ligand during the binding interaction.

EXPERIMENTAL PROCEDURES
Materials-Synthetic human glucagon was purchased from Sigma or Peninsula Laboratories or synthesized on an Applied Biosystems 432A peptide synthesizer.  ]Glucagon was prepared from bovine/ porcine glucagon (Sigma) as described previously (12) and then reacted with 5- This plasmid construct is subsequently referred to as pRm-hGlu.
S2 cells were maintained and induced to express recombinant protein essentially as described by Bunch et al. (10). S2 cells (provided by Dr. L. S. B. Goldstein, University of Arizona) were maintained at 27°C in Schneider media supplemented with 10% heat-inactivated fetal calf serum, two mM glutamine, and 50 g/ml gentamycin. S2 cells (1 ϫ 10 7 cells) were cotransfected with 1 g of pUChsneo (gift of Dr. Herman Steller) (13) and 20 g of pRm-hGlu using the CaPO 4 precipitation method. pUChsneo encodes a gene for G418 resistance whose expression is driven by the Drosophila heat shock promoter. Twenty-four hours after transfection, G418 was added to a final concentration of 300 g/ml. Cells were split approximately every 5-7 days. A stable population of G418-resistant cells was obtained in 3-4 weeks. S2 cells (1-2 ϫ 10 6 cells/ml) were induced to express receptor by the addition of 1 mM CuSO 4 .
Fluorescence Flow Cytometry and Cell Sorting-Cells were analyzed for single cell fluorescence on a FACScan/Vax flow cytometeter (Benton Dickinson Immunocytometry Systems, San Jose, CA). The cells were sorted based on fluorescein intensity using a FACStar PLUS cell sorterconsort/Vax (FACS, Benton Dickinson) following standard procedures.
Fluorescence Spectroscopy-Fluorescence measurements were performed on an SLM 48000 spectrofluorometer at 20°C. Excitation was achieved with an argon laser, where the 488-nm line was selected by a sharp bandpass filter and attenuated with a 0.3-1.0 optical density neutral density filter. In order to obtain a correction factor to compensate for differences between photomultipliers, the sample was excited with both vertical and horizontal light. This was achieved by first depolarizing the laser light and then selecting either vertical or horizontal light with a polarizer. Emission light was selected with a 530-nm bandpass filter (Corion) followed by either a parallel or perpendicular polarizer. Using the T format, parallel and perpendicular emissions were collected simultaneously. Cuvettes were filled with 1.5 ml of buffer and were continuously stirred. Background and correction factors were recorded prior to the relevant portion of the data collection for timebased acquisition. All signals were recorded in the ratio mode using rhodamine as the standard. The background fluorescence (S2 cell membranes in the absence of [fluorescein-Trp 25 ]glucagon) was subtracted from each signal, and then the correction factor was applied to the vertical intensity. The data were then analyzed as follows.
Data Analysis-The fraction of bound and free ligand were determined by Equation 3 (17) r obs ϭ r bound ͑fraction ligand bound͒ ϩ r free ͑fraction ligand free͒ (Eq. 3) where r obs is the observed anisotropy of a mixture of free ligand and ligand bound to the receptor. Equation 3 assumes that there is no quantum yield change upon ligand binding. In order to evaluate the fraction bound by anisotropy with a quantum yield change, the following modification was used (17) fraction bound ϭ ͑r obs Ϫ r free ͒/͑͑r bound Ϫ r obs ͒Q ϩ r obs Ϫ r free ͒ (Eq. 4) where Q is the relative quantum yield change upon binding. Association time courses were fit to Equation 5 using the Marquardt's algorithm as described by Press et al. (18).
where r(t) is the anisotropy at time t, and the initial value is assumed to be the anisotropy of the free ligand. The fitted values A is the amplitude and is the relaxation time for the exponential increase. Dissociation time courses were fit as follows where A 1 , A 2 , 1 , 2 are the amplitudes and relaxation times for two kinetic phases. The same equation was used for fitting intensity data (I(t)). For radioligand and functional assays, IC 50 (the concentration of ligand displacing 50% of the labeled ligand from the binding site) and EC 50 (the half-maximal effective concentration of ligand in a functional assay) values were calculated using the Prism program (Graphpad Software). B max values (the concentration of ligand bound at saturation) were determined from the fitted competition curves as described previously (19).

Expression of the Human Glucagon Receptor in S2 Cells-
The human glucagon receptor cDNA was subcloned into vector pRmHa3 and cotransfected with pUChsneo into S2 cells. pRmHa3 contains a Drosophila metallothionein promoter that is tightly regulated in S2 cells (10). After 3 weeks of growth in 0.3 g/ml G418, cells were induced to express glucagon receptor by incubation with 1 mM CuSO 4 . After 3 days of induction, the cells bound [ 125 I]glucagon with a B max of 110 pmol/mg of membrane protein, whereas mock-transfected S2 cells displayed no specific [ 125 I]glucagon binding.
In an attempt to select a subpopulation of cells with higher expression, cells induced to express the glucagon receptor were incubated with [fluorescein-Trp 25 ]glucagon and sorted by FACS. The 5% of the cells displaying the highest levels of binding were collected and expanded. Upon induction with 1 mM CuSO 4 , cells displayed a time-dependent increase in [ 125 I]glucagon binding with maximal levels of 250 pmol/mg of protein achieved at 3-4 days after induction (data not shown). This represents about a 2-fold increase in receptor expression levels following FACS. After 4 months of continuous culture in the absence of CuSO 4 , no significant loss of receptor expression has been observed.
Immunoblotting of the Human Glucagon Receptor Expressed in S2 Cells-Antisera from rabbits injected with either peptide LX1 or LX2 recognized a protein with an apparent molecular mass of 48 kDa in transfected S2 cells induced with CuSO 4 (Fig. 1, lanes C and F). This protein was absent in nontransfected or transfected but uninduced S2 cells (Fig. 1). The M r ϭ 48 kDa observed in this system differs from that measured by affinity labeling of the glucagon receptor in rat liver membranes (M r ϭ63 kDa, Ref. 20 25 ]glucagon demonstrated that the affinity of the fluorescein-labeled glucagon was similar to that of unmodified glucagon (Fig. 2), as previously reported by Heithier et al. (11). When expressed in S2 cells, the human glucagon receptor was functionally coupled to increases in cAMP. Incubation of S2 cells expressing the receptor with glucagon or [fluorescein-Trp 25 ]glucagon led to a 30-fold increase in cAMP levels, with an EC 50 of 1.6 nM for glucagon and 1.9 nM for [fluorescein-Trp 25 ]glucagon (data not shown). [fluorescein-Trp 25 ]Glucagon thus functions as an agonist in the S2 cells, with an efficacy and potency similar to that of unlabeled glucagon.
[fluorescein-Trp 25 ]Glucagon binding to membrane preparations from S2 cells expressing the human glucagon receptor was detected by monitoring changes in fluorescence anisotropy and intensity. Fig. 3A shows the time-dependent increase in anisotropy that was observed following the addition of [fluorescein-Trp 25 ]glucagon to these membranes and the time-dependent reversal of this increase upon displacement with one M unlabeled glucagon. The increase in anisotropy was not observed with S2 membranes from nontransfected cells (Fig. 3C) and was blocked by preincubation with unlabeled glucagon (Fig. 3A). An increase in fluorescence intensity was also observed upon binding of [fluorescein-Trp 25 ]glucagon to its receptor, which was reversed by the addition of unlabeled glucagon (Fig. 3B). Nonspecific interactions were examined by adding [fluorescein-Trp 25 ]glucagon to membranes prepared from nontransfected S2 cells not expressing the glucagon receptor (Fig.  3C). The intensity transformation displayed an initial decrease followed by a gradual return to a base-line level that was not affected by the subsequent addition of unlabeled glucagon. The initial decrease was observed frequently, but not always, and is assumed to result from a mixing artifact and/or absorption to the membranes of [fluorescein-Trp 25 ]glucagon. Because anisotropy measurements were less sensitive to nonspecific binding or mixing artifacts than were the intensity changes (Fig. 3C), changes in anisotropy were used as a readout to examine the kinetic parameters of the interaction of [fluorescein-Trp 25 ]glucagon with the human glucagon receptor.
Determination of the Anisotropy and Fluorescence Intensity of Bound [fluorescein-Trp 25 ]Glucagon-The observed anisotropy is a combination of signals from the bound and free ligand. The anisotropy of the bound ligand alone could be measured if receptor were present in excess or if free ligand were removed from the solution. Adding excess receptor was not practical under these experimental conditions. However, the signal from free ligand could rapidly be removed by adding anti-fluorescein antibody, which quenched up to 95% of the fluorescein fluorescence upon binding (21,22). As shown in Fig. 4A, the addition of anti-fluorescein antibody to the [fluorescein-Trp 25 ]glucagonreceptor complex resulted in a rapid quenching of the fluorescence of the free ligand, while that of the bound ligand was protected from the antibody. As the ligand dissociated from the receptor, its fluorescence was quenched by the anti-fluorescein antibody, as observed by a slow decrease in intensity after the rapid quenching phase. While the intensity was decreasing during this period, the anisotropy remained stable (Fig. 4B). This anisotropy signal originated from the bound ligand and was determined to be 0.281 Ϯ 0.001 (n ϭ 2). Once the anisotropy of the bound ligand was determined, Equation 3 could be used to estimate the fraction of ligand bound to the receptor. Applying Equation 3 to the data in Fig. 4 (r obs ϭ 0.176, r free ϭ 0.096 (data not shown), r bound ϭ 0.281) indicated that 43% of the ligand is bound under these conditions, consistent with the amount of fluorescence protected from quenching by the antibody (Fig. 4A).

Dissociation Kinetics of [fluorescein-Trp 25 ]Glucagon from Its
FIG. 1. Immunoblot analysis of membranes prepared from S2 cells. Membranes were prepared from S2 cells, and 10 g of protein were loaded/lane on a 10% SDS-polyacrylamide gel. Following electrophoresis, proteins were transferred to nitrocellulose, and Western blotting was performed as detailed under "Experimental Procedures" with antisera to LX1 (lanes A-C) or LX2 (lanes D-F) at a 1:8000 dilution. Lanes A and D, nontransfected cells; lanes B and E, cells transfected with pRm-hGlu; lanes C and F, cells transfected with pRm-hGlu and incubated with 1 mM CuSO 4 for 3 days. Blot was exposed to Hyperfilm-ECL for 30 s.

Receptor-The dissociation time course of [fluorescein-
Trp 25 ]glucagon was measured in the absence of Gpp(NH)p using low concentrations (2-10 nM) of the agonist. Under these conditions, two distinct relaxation phases could be observed (Fig. 5, Table I), suggestive of agonist binding to two affinity states of the receptor. The slowly dissociating phase corresponds to the high affinity guanine nucleotide sensitive site noted in the competition binding experiments shown in Fig. 2, whereas the fast phase corresponds to the low affinity state of the receptor. These experiments, in which approximately 10 nM ligand and receptor were used, indicate that most of the glucagon receptor in the S2 cells is in the low affinity uncoupled state, as would be expected for a system in which the receptor is significantly overexpressed and G protein may be limiting. In contrast, at the submaximal concentration of [ 125 I]glucagon used in the binding assays, most of the radioligand would bind to the high affinity state of the receptor, so that the high affinity state would be significantly overrepresented in the competition binding experiments (Fig. 2).
For more detailed kinetic analysis, Gpp(NH)p was included in the incubation to simplify the kinetics by converting all of the receptor to the low affinity state and to reduce the time needed for the reaction to come to completion. The addition of excess glucagon caused a decrease in anisotropy and intensity as the [fluorescein-Trp 25 ]glucagon dissociated from the receptor. The intensity decrease was about 9% (4.02 to 3.65) and was . The data were fit to a single or two-phase exponential decay using Equation 6. The fitted parameters for a two-phase fit were amplitude 1 ϭ 74.5% with a rate of 4.7 ϫ 10 Ϫ3 s Ϫ1 ; amplitude 2 ϭ 25.5% with a rate of 6.6 ϫ 10 Ϫ4 s Ϫ1 ; and an end point of 0.113. For a single phase fit, the data were fit with a rate of 3.4 ϫ 10 Ϫ3 s Ϫ1 and an end point of 0.120. Both fits are presented in the figure; however, the two-phase fit is obscured by the experimental data points. Residuals for the one and two phase analyses are shown in the insets. minimally affected by changes in pH (data not shown). Using the anisotropy of the bound ligand (0.281) determined from Fig.  4, the fraction of ligand bound in Fig. 6 was determined to be 30% from Equation 3. Knowing the fraction of ligand bound and the intensity change upon dissociation, the quantum yield increase of [fluorescein-Trp 25 ]glucagon upon binding to the receptor was determined to be 1.34-fold. Equation 3 must be applied with some caution in this situation since the intensity of the ligand is not constant during the binding reaction. The kinetics of the dissociation of [fluorescein-Trp 25 ]glucagon when monitored by intensity changes were similar, but not identical, to the kinetics when monitored by changes in anisotropy (Fig. 6). This difference has been postulated to arise from an optical effect on the anisotropy resulting from changes in fluorescence intensity. The relationship between intensity and anisotropy can be described by the following equations (23) where x i is the fraction of the ligand bound or free, q i eff is the relative increase in intensity of that component and I(t) and r(t) are the observed intensity and anisotropy, respectively. Using these equations, a global analysis was performed in which the anisotropy and intensity data were analyzed together (23, 24) (Fig. 6B). The q eff for the free ligand was 1.00, and the fitted value for q eff for the bound ligand was 1.42, which agreed fairly well with the value of 1.34 determined using Equation 3. The global fit indicated that 23.3% of the ligand was bound under these conditions, compared with 30% bound ligand calculated using Equation 3. The fitted value for the dissociation rate determined by global analysis (6.6 ϫ 10 Ϫ3 s Ϫ1 ) was similar to that determined by a local fit of the anisotropy data (5.8 ϫ 10 Ϫ3 s Ϫ1 ). Because the magnitude of the effect of the intensity change on the anisotropy measurement was small and the intensity data were less precise than the anisotropy data, subsequent analysis used only the local fit of the anisotropy data for determination of rate constants.
Association Kinetics of [fluorescein-Trp 25 ]Glucagon with its Receptor-Association rates for [fluorescein-Trp 25 ]glucagon were measured in a range of 5-150 nM ligand in the presence of 100 M Gpp(NH)p (Fig. 7). Gpp(NH)p was included to simplify the analysis by converting the receptor to a single population of binding sites. Pilot experiments showed that Gpp(NH)p had minimal effects on the association rate of 10 nM [fluorescein-Trp 25 ]glucagon with the receptor, lowering the total binding by 18% and increasing the rate of binding by about 20% (data not shown). The association rate data were adequately fit by a single relaxation time using Equation 5 and could be evaluated by a reversible one-step binding mechanism where k 1 and k Ϫ1 are the forward and reverse rate constants, respectively. The inverse relaxation time of ligand association ( Ϫ1 ) would be defined as A plot of Ϫ1 (also defined as the observed rate, k obs ) versus L should yield a straight line with the slope equivalent to k 1 and the intercept equivalent to k -1 (25). As shown in Fig. 7, inset, the data fit well to this model, with k 1 ϭ 7.9 Ϯ 1.1 ϫ 10 4 M Ϫ1 s Ϫ1 and k Ϫ1 ϭ 5.5 Ϯ 0.01 ϫ 10 Ϫ3 s Ϫ1 (n ϭ 2). The dissociation rate in the presence of Gpp(NH)p measured directly (5.9 Ϯ 0.5 ϫ 10 Ϫ3 s Ϫ1 n ϭ 4; Fig. 6; Table I) was similar to the k Ϫ1 value determined in this experiment, thus supporting the above model. The model was further supported by the agreement of the K d derived from this association experiment (k Ϫ1 /k 1 ϭ 69 Ϯ 10 nM, n ϭ 2) with the thermodynamic K d derived from the Unlabeled glucagon was added after an additional 10 min (time ϭ 0) and the dissociation time course was measured. A, anisotropy and intensity were fit separately to a single exponential decay using Equation 6. A rate of 5.8 ϫ 10 Ϫ3 s Ϫ1 was obtained from the anisotropy transformation and 6.8 ϫ 10 Ϫ3 s Ϫ1 from the intensity transformation. B, global analysis using Equations 7 and 8. Each data set was weighted by its standard deviation. The standard deviation was estimated by a linear regression analysis of the plateau regions before the addition of glucagon. The bound anisotropy (0.281, determined from Fig. 4) was constrained as a constant during the fit. The calculated values were 23.3% of the ligand bound with a relative fluorescence increase of 1.42. The free anisotropy (or end point) was calculated to be 0.109, and the fitted dissociation rate was 6.6 ϫ 10 Ϫ3 s Ϫ1 . Residuals (fit value Ϫ experimental value) are shown in the insets.  Fig. 2).

DISCUSSION
Biophysical and structural characterization of the glucagon receptor, like that of other G protein-coupled receptors, has been hampered by the lack of availability of sufficient quantities of active receptor protein. Using a Drosophila S2 system, we have isolated a polyclonal cell population that expresses the human glucagon receptor to a level of 250 pmol/mg as measured by [ 125 I]glucagon binding. Using this system, the interaction of the agonist [fluorescein-Trp 25 ]glucagon with its receptor could be monitored via an increase in fluorescence anisotropy and intensity during the binding reaction. The anisotropy was more stable and less sensitive to nonspecific binding than the quantum yield increase and was therefore used as the primary means to monitor the binding of [fluorescein-Trp 25 ]glucagon to the receptor.
The anisotropy of the ligand bound to the receptor was examined by using anti-fluorescein antibody to quench the free ligand in solution. The bound anisotropy was determined to be 0.281 Ϯ 0.001, indicating that the fluorescein moiety is relatively immobile when anchored in the binding pocket of the glucagon receptor. For comparison, the anisotropy of fluorescein in [fluorescein-Lys 3 ] substance P bound to the NK1 neurokinin receptor was only 0.17 (22) and that for fluorescein labeled epidermal growth factor bound to the epidermal growth factor receptor was determined to be 0.18 (26).
It is apparent from the equilibrium binding titrations (Fig. 2) and the dissociation rate studies (Fig. 5; Table I) that [fluorescein-Trp 25 ]glucagon binds with two classes of receptor binding sites and that the conversion from high to low affinity is stimulated by Gpp(NH)p. These results indicate that the expressed glucagon receptor is coupled to G protein(s) in the S2 cells, consistent with the ability of glucagon to stimulate cAMP accumulation in these cells. In order to simplify the analysis of association rates, the receptor was converted to a single class of binding sites by preincubation with Gpp(NH)p. Gpp(NH)p had only a minimal effect on the observed association rate for [fluorescein-Trp 25 ]glucagon. Monophasic association kinetics were observed in both the presence and absence of the guanine nucleotide, with only a small (Ͻ20%) increase in the association rate in the presence of Gpp(NH)p. Monophasic association kinetics were also observed in radioligand binding studies by Horwitz et al. (27) where association of [ 125 I-Tyr 10 ]glucagon was described by a single relaxation time.
A kinetic analysis of [fluorescein-Trp 25 ]glucagon binding was performed in the presence of 100 M Gpp(NH)p. The data were consistent with the model described in Equations 9 and 10, implying a simple bimolecular reaction between the ligand and the receptor. However, the association rate constant k 1 (7.9 ϫ 10 4 M Ϫ1 s Ϫ1 ) was much slower than that expected for a diffusion-controlled reaction, suggesting a more complex mechanism involving a slow conformational change in either the receptor or the ligand. Further biophysical studies will be required to address this possibility.
The increase in fluorescence intensity observed upon binding [fluorescein-Trp 25 ]glucagon to the receptor is atypical, as the association of fluorescein with a protein is often accompanied by a decrease in intensity. These data suggest that the fluorescence of [fluorescein-Trp 25 ]glucagon is quenched in solution by some internal quenching mechanism and that quenching is relieved upon receptor binding, perhaps reflecting the conformational change implied by the slow association kinetics. Thus, the receptor may actively "hold" the fluorescein away from the rest of the ligand, which would also be consistent with the high anisotropy determined for the bound ligand.
Previous structural studies of glucagon in solution have been performed at high (nonphysiological) concentrations of glucagon, or have used detergents or lipids to model receptor binding. The fluorescence analysis in the present study was performed at much lower ligand concentrations and gives direct information about the conformation of glucagon bound to its receptor. The combination of the S2 expression system with [fluorescein-Trp 25 ]glucagon should allow further biophysical analysis of this receptor ligand interaction.  Table I.