Heterologous desensitization of the glucagon-like peptide-1 receptor by phorbol esters requires phosphorylation of the cytoplasmic tail at four different sites.

Glucagon-like peptide-1 stimulates glucose-induced insulin secretion by binding to a specific G protein-coupled receptor that activates the adenylyl cyclase pathway. We previously demonstrated that heterologous desensitization of the receptor by protein kinase C correlated with phosphorylation in a 33-amino acid-long segment of the receptor carboxyl-terminal cytoplasmic tail. Here, we determined that the in vivo sites of phosphorylation are four serine doublets present at positions 431/432, 441/442, 444/445, and 451/452. In vitro phosphorylation of fusion proteins containing mutant receptor C-tails, however, indicated that whereas serines at position 431/432 were good substrates for protein kinase C (PKC), serines 444/445 and 451/452 were poor substrates, and serines 441/442 were not substrates. In addition, serine 416 was phosphorylated on fusion protein but not in intact cells. This indicated that in vivo a different PKC isoform or a PKC-activated kinase may phosphorylate the receptor. The role of phosphorylation on receptor desensitization was assessed using receptor mutants expressed in COS cells or Chinese hamster lung fibroblasts. Mutation of any single serine doublet to alanines reduced the extent of phorbol 12-myristate 13-acetate-induced desensitization, whereas substitution of any combination of two serine doublets suppressed it. Our data thus show that the glucagon-like peptide-1 receptor can be phosphorylated in response to phorbol 12-myristate 13-acetate on four different sites within the cytoplasmic tail. Furthermore, phosphorylation of at least three sites was required for desensitization, although maximal desensitization was only achieved when all four sites were phosphorylated.

Glucagon-like peptide-1 stimulates glucose-induced insulin secretion by binding to a specific G protein-coupled receptor that activates the adenylyl cyclase pathway. We previously demonstrated that heterologous desensitization of the receptor by protein kinase C correlated with phosphorylation in a 33-amino acid-long segment of the receptor carboxyl-terminal cytoplasmic tail. Here, we determined that the in vivo sites of phosphorylation are four serine doublets present at positions 431/432, 441/442, 444/445, and 451/452. In vitro phosphorylation of fusion proteins containing mutant receptor C-tails, however, indicated that whereas serines at position 431/432 were good substrates for protein kinase C (PKC), serines 444/445 and 451/452 were poor substrates, and serines 441/442 were not substrates. In addition, serine 416 was phosphorylated on fusion protein but not in intact cells. This indicated that in vivo a different PKC isoform or a PKC-activated kinase may phosphorylate the receptor. The role of phosphorylation on receptor desensitization was assessed using receptor mutants expressed in COS cells or Chinese hamster lung fibroblasts. Mutation of any single serine doublet to alanines reduced the extent of phorbol 12-myristate 13-acetate-induced desensitization, whereas substitution of any combination of two serine doublets suppressed it. Our data thus show that the glucagon-like peptide-1 receptor can be phosphorylated in response to phorbol 12-myristate 13-acetate on four different sites within the cytoplasmic tail. Furthermore, phosphorylation of at least three sites was required for desensitization, although maximal desensitization was only achieved when all four sites were phosphorylated.
Glucagon-like peptide-1 (GLP-1) 1 is an intestinal peptidic hormone that is secreted in the blood in response to nutrient ingestion. One of its major functions is to potentiate the glucosedependent secretion of insulin by pancreatic ␤ cells (1)(2)(3). This effect requires the binding of GLP-1 to a specific G-protein coupled receptor that is linked to the activation of the adenylyl cyclase pathway. The effect of GLP-1 on insulin secretion de-pends on the presence of glucose concentrations equal to or greater than the normoglycemic value of ϳ5 mM. This hormone is thus a potentiator of the glucose signaling pathway, and its action is mediated by the production of cAMP and, probably, the activation of protein kinase A (4 -6).
We previously demonstrated that the GLP-1 receptor-mediated cAMP response could undergo both homologous (following receptor exposure to its agonist) and heterologous (following the activation of PKC) desensitization (7). Both forms of desensitization were correlated with receptor phosphorylation and internalization. PMA-induced phosphorylation and desensitization but not that induced by GLP-1 were completely suppressed in the presence of the PKC inhibitor RO-318220. Removal of the last 33 amino acids of the carboxyl-terminal cytoplasmic tail of the receptor abolished both homologous and heterologous desensitization and phosphorylation.
Phosphorylation of G-coupled receptors by multiple protein kinases has been shown to be involved in both heterologous and homologous desensitization. For instance, for the ␤ 2 -adrenergic receptor, heterologous desensitization can be induced by activation of protein kinase A or protein kinase C, an effect that requires the presence of a protein kinase A/PKC consensus site in the third intracellular loop of the receptor but not that present in the C-tail (8 -11). On the other hand, homologous desensitization results from receptor phosphorylation by Gcoupled receptor-specific kinases (12) that form a family of structurally related isoforms (13). This phosphorylation takes place in the serine/threonine-rich C-tail of the receptor (11) and induces association of the receptor with ␤-arrestins (14 -17). This interaction prevents further activation of G-proteins following agonist binding to the receptor.
GLP-1 and phorbol ester-induced desensitization and phosphorylation of the GLP-1 receptor are additive, and both are suppressed by deletion of the last 33 amino acids of the receptor C-tail (7). This suggests that this receptor portion contains the amino acids that can be phosphorylated in both homologous and heterologous desensitization. These results are in contrast with those obtained with the ␤ 2 -adrenergic receptor for which heterologous desensitization (protein kinase A-and PKC-mediated) requires phosphorylation in the third cytoplasmic loop, whereas ␤-adrenergic receptor kinases phosphorylate the Ctail of the receptor. In addition, phosphorylation of the C-tail by ␤-adrenergic receptor kinases, but not phosphorylation of the third cytoplasmic loop by protein kinase A or PKC, induces arrestin binding to the receptor (9), indicating different mechanisms by which phosphorylation causes desensitization. With the GLP-1 receptor both forms of desensitization correlate with phosphorylation of the same, relatively short, region of the receptor, suggesting that homologous and heterologous desensitization may be mediated by similar mechanisms.
To better understand the molecular mechanisms of heterol-ogous desensitization we have identified the sites of phosphorylation of the GLP-1 receptor following activation of protein kinase C in intact cells. We demonstrate that four serine doublets located in the last 33 amino acids of the receptor become phosphorylated upon activation of PKC. Furthermore, we show that phosphorylation of at least three sites is required for desensitization of the receptor, which becomes maximal when all the four sites are phosphorylated.

MATERIALS AND METHODS
Cells and Cell Culture-COS cells and Chinese hamster lung (CHL) fibroblasts were cultured as described (5). Clone 5 cell is a CHL fibroblast stably transfected with the rat GLP-1 receptor cDNA (5,18). Transformation of COS cells and generation of stable CHL transformants were performed as described earlier (5), except that when the cDNAs were inserted in the pcDNA-3 vector (see below), co-transformation of the fibroblasts with the pWLneo plasmid was not required.
Mutagenesis-The different GLP-1 receptor mutants used in this study are described in Fig. 2. The deletion mutant ⌬CT431 was described previously (7). The other mutations were generated by polymerase chain reaction amplification as described (19), and each mutant was verified by DNA sequencing. The mutant GLP-1 receptor cDNAs were subcloned in the pcDNA-3 vector (Invitrogen, Leek, The Netherlands) and in the pmlMTIi vector (7). The cDNAs subcloned in the pcDNA-3 and pmlMTIi vectors are under the control of the cytomegalovirus and metallothionein promoters, respectively.
Binding Studies-Determination of dissociation constants (K d ) was derived from Scatchard plot analysis (20). The number of receptors expressed per cell were derived either from Scatchard plot analysis or from saturation binding experiments performed as described (18), except that 250 -500 pM iodinated GLP-1 mixed with 3-5 nM unlabeled GLP-1 was used and that all incubation and washing buffers contained 0.01% polyoxyethylene-sorbitan monolaurate (Tween 20) (P-1379, Sigma).
Desensitization of the GLP-1-induced cAMP Production-Desensitization was assessed, as described earlier (7), by comparing the doseresponse curves of cAMP production as a function of increasing GLP-1 concentrations for control and PMA-or thrombin-treated cells.
Cross-linking Experiments-The ability of the mutated GLP-1 receptors to be immunoprecipitated by carboxyl terminus-specific anti-GLP-1 receptor antibodies (antibody 9(2)) (18) was measured as follows. COS cells were transfected with plasmids encoding the wild-type or mutant GLP-1 receptors and cultured in 6-well plates. The GLP-1 receptors expressed at the surface of these cells were then cross-linked with iodinated GLP-1, as described (18), except that the bifunctional crosslinker was ethylene glycol bis(succinimidylsuccinate) (21565; Pierce Europe B.V., Oud Beijerland, The Netherlands) and that 10% ␤-mercaptoethanol replaced iodoacetamide in the sample buffer. The crosslinked receptors were visualized following SDS/polyacrylamide gel electrophoresis and autoradiography, and the bands corresponding to the cross-linked receptors were analyzed by densitometry scanning, as described (18). The extent of immunoprecipitation (I) of the wild-type and mutant GLP-1 receptors was calculated as follow from the densitometry scanning measures (D 1 in arbitrary units) corrected for the cellular content of the culture wells (P 1 in g) and for the cell surface receptor expression (S 1 in cpm, derived from saturation binding experiments): Phosphorylation Experiments on Intact Cells-PMA-induced phosphorylation of the wild-type or mutant GLP-1 receptors expressed by transiently transfected COS cells was assessed as described earlier (7). When phosphorylation was induced by dioctanoyl-sn-glycerol (Sigma), this compound was added at a concentration of 30 M for 10 min. The bands corresponding to the phosphorylated GLP-1 receptors were analyzed by densitometry scanning. The extent of phosphorylation (F) of the wild-type and mutant GLP-1 receptors was calculated from the difference between the densitometry measures obtained from PMAtreated (D PMA , arbitrary units) and untreated (D control , arbitrary units) cells corrected for the cellular content of the culture wells in which the assay was performed (P 2 in g), for the cell surface receptor expression (S 2 in cpm, measured as described for the S 1 parameter (see above)) and for the ability of the receptors to be immunoprecipitated (I, as defined above): Phosphoamino Acid Analysis-Phosphoamino acid analysis was performed by thin layer chromatography using a HTLE 7000 electrophore- The immunoprecipitated proteins were detected following gel electrophoresis and autoradiography. The receptor was then extracted from the gel and hydrolyzed in 6 N HCl. The resulting amino acids were separated by two-dimensional electrophoresis, and radioactive residues were detected by autoradiography. The migration of unlabeled threonine, serine, and tyrosine residues is indicated by dotted circles. Only serine residues are phosphorylated following PMA stimulation.

FIG. 2. Amino acid sequence of the different mutants of the GLP-1 receptor used in this study.
The last 52 amino acids of the rat GLP-1 receptor (residues 412-463) are shown at the top of the figure (one-letter code). The sequence of the different mutants is presented below the wild-type sequence (hyphens indicate the same residue as the wild-type sequence). Residues 1-411 of the mutant receptors are identical to those found in the wild-type receptor sequence. Residues marked by an asterisk are putative PKC phosphorylation sites.
sis system (C. B. S. Scientific Company, Del Mar, CA) as per the manufacturer's protocol and Cooper et al. (21). Fig. 2 and residues 412-430 of mutant ⌬CT431 were fused to the glutathione S-transferase protein encoded by plasmid pGEX-1 (22). This was accomplished by polymerase chain reaction amplification of the cDNAs encoding the wild-type and the mutant GLP-1 receptors using the antisense nucleotide SP6/EcoRI (GCGAGC-GAATTCATTTAGGTGACACTATAG), containing an EcoRI site, and the sense oligonucleotide EcoRI-1250 (CAGATGGAATTCCGGAA-GAGCTGGGAG) or 416EcoRI (CAGATGGAATTCCGGAAGGCCTGG-GAGCGCT). Both sense oligonucleotides contain an EcoRI site. In addition, oligonucleotide 416EcoRI encodes an alanine instead of a serine at position 416 and thus will generate mutation S416A when used (see Fig. 2). The amplified fragments were digested with EcoRI and subcloned into the EcoRI site of the pGEX-1 vector. Each construction was verified by sequencing. The expression and purification of the fusion proteins were performed with a slightly modified version of Smith and Johnson's method (22). Overnight cultures of Escherichia coli transformed with parental or recombinant pGEX-1 plasmid were diluted 1/10 in 100 ml of LB medium (10 g/liter NaCl, 10 g/liter Tryptone; 5 g/liter yeast extract, pH 7.4) and grown for 1 h at 37°C before adding isopropyl-␤-D-thiogalactopyranoside to 0.1 mM. After a further 4 -5 h of growth, bacteria were pelleted and resuspended in 10 ml of sonication buffer (50 mM Tris, pH 8, 50 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride; 20 g/ml DNase I, 1% Triton X-100) and incubated 10 min at 4°C on a rotating platform. The suspension was then sonicated (1 min, position 7 using the sonicator B-12 (Branson sonic power company, Danbury, Connecticut)), and pelleted at 11,200 ϫ g for 10 min at 4°C. The supernatant was incubated twice with 100 l of 50% glutathione-agarose beads (G-4510, Sigma, Buchs, Switzerland) for 15 min at 4°C on a rotating platform. The beads were then pooled and washed twice with 10 ml NETN (0.5% Nonidet P-40, 20 mM Tris, pH 8, 100 mM NaCl, 1 mM EDTA). Fusion proteins were eluted with 3 ϫ 5 min washes with 100 l of elution buffer (20 mM reduced glutathione, 100 mM Tris, pH 8, 120 mM NaCl) and kept at Ϫ20°C before use. In vitro PKC-induced phosphorylation of the fusion proteins was performed as described (23), except that 2-3 g of fusion proteins and 540 nM [␥-32 P]ATP were used. RESULTS PMA-induced activation of PKC can increase the phosphorylation of the GLP-1 receptor but not that of a truncated mutant lacking the last carboxyl-terminal 33 amino acids. To determine which amino acids of the C-tail are phosphorylated following PMA stimulation, fibroblasts expressing the wildtype receptor were labeled with [ 32 P]orthophosphate and stimulated with PMA, and the receptor was purified by immunoprecipitation and gel electrophoresis (Fig. 1, left). Phosphoamino acid analysis of the purified receptor revealed that only phosphoserines were present (Fig. 1, right).

In Vitro Phosphorylation of Fusion Proteins Containing Wild-type or Mutated Carboxyl-terminal GLP-1 Receptor Sequences-Residues 412-463 of the wild-type GLP-1 receptors and of the last 6 mutant GLP-1 receptors described in
The last 33 carboxyl-terminal amino acids of the GLP-1 receptor contain 10 serine residues, 8 of which are present as serine doublets (Fig. 2). To determine whether the residues phosphorylated in response to PMA may be found within the 4 serine doublets, we have generated a series of receptor mutants in which one or more of the serine doublets were replaced with alanine residues (see Fig. 2). The mutant receptors were then expressed transiently in COS cells and tested for their ability to be phosphorylated in response to PMA. Fig. 3A shows that mutation of all the 4 serine doublets led to the suppression of receptor phosphorylation. In Table I, we present the quantitative determination of the extent of phosphorylation of receptor mutants with substitution to alanines of either single, pairs, or the four serine doublets. These quantitations take into account the efficiency of immunoprecipitation and the cell surface expression of the transfected mutant receptors. These data show that mutation of each serine doublet reduces the extent of phosphorylation by ϳ50% except for the S431A/S432A (Fig. 2, 431AA) mutant for which reduction is ϳ25%. When any pair of serine doublets were substituted with alanines there was a 70 -80% reduction in the extent of phosphorylation. Finally, elimination of all 4 serine doublets, either by truncation of the receptor C-tail (⌬CT431) or by substitution of serines with alanine residues, prevented phosphorylation of the receptor.
These data suggest that each serine doublet is phosphorylated following activation of PKC by phorbol esters. To further explore this point, we generated mutants in which only one doublet of serines was left. Fig. 3B shows that each doublet could indeed be phosphorylated following activation of PKC although to a variable extent, as assessed by the quantitative analysis presented at the bottom of Fig. 3B.
Surprisingly, among the 4 serine doublets, only serines 431/ 432 are in a putative PKC phosphorylation site (see Fig. 2). However, in a few cases, PKC may phosphorylate residues that are not in consensus phosphorylation sites. To determine whether all serine doublets of the cytoplasmic tail could be phosphorylated by PKC, we generated fusion proteins consisting of glutathione S-transferase and the C-tail of the receptor from amino acids 412 to 463 and containing different replacements of serines to alanines. These were then used in an in vitro PKC phosphorylation assay, and the level of phosphorylation was determined. Fig. 4 shows that the glutathione Stransferase protein was not phosphorylated by PKC (lane 1), whereas the fusion protein containing the wild-type receptor C-tail sequence was phosphorylated (lane 2). When all four serine doublets were substituted with alanines, the fusion protein was still phosphorylated, although to a much lower extent as compared with the wild-type protein (lane 5). A similar situation was observed with the truncated C tail of ⌬CT431 (lane 3). Serine 416, which is in a consensus PKC site (see Fig.  2), was responsible for this phosphorylation as demonstrated by the abolition of the phosphorylation when this residue was mutated to alanine together with all four serine doublets (lane 4). Thus, serine 416 is an in vitro PKC phosphorylation site but is not phosphorylated in intact cells in response to PMA (see Table I, ⌬CT431 and the quadruple mutant). To assess the ability of individual serine doublets to be phosphorylated by purified PKC, we have thus mutated serine 416 and combinations of 3 serine doublets to alanine residues so that the resulting fusion proteins bear only one potential site of phosphorylation. As expected, serines 431/432 were very efficiently phosphorylated by PKC (Fig. 4, lane 9). In contrast, the other serine doublets were not (441/442) or only poorly (444/445 and 451/452) phosphorylated by PKC (lanes 5-7). Thus, the 4 serine doublets are phosphorylated in response to PMA in intact cells, whereas only serines 431/432 and 416 are efficiently phosphorylated in vitro by purified PKC.
To understand the role of PKC-dependent phosphorylation in receptor desensitization, we expressed the receptor mutants in different cellular systems: transiently transfected COS cells that express ϳ10 6 receptors/cell, CHL fibroblasts expressing the GLP-1 receptor under the control of the CMV promoter with surface expression of ϳ10 5 receptors/cell ("high expressor" fibroblasts), and CHL fibroblasts expressing the receptor under the control of the metallothionein promoter leading to a surface expression of 1,000 -10,000 receptors/cell ("low expressor" fibroblasts; see Table II). Fig. 5 shows that in all three cell types expressing the wild-type receptor, the receptor-mediated cAMP response is desensitized by PMA. As expected for cells expressing high levels of receptors (24), desensitization in COS cells and in high expressor fibroblasts results in the increase of the    b Dose-response curves such as those presented in figure 5, middle panels were analyzed with the KaleidaGraph v2.1 software (Abelbeck Software) and fitted with yϭm1/(1ϩ(m2/x))ϩm3 where y ϭ cAMP production; x ϭ GLP-1 concentration; m1 ϭ estimated maximal increase in cAMP production; m2 ϭ estimated EC50 of the curve; m3 ϭ estimated basal response. The parameter m1 was defined as the maximal agonist stimulation (Vmax). The results are expressed as the maximal GLP-1induced cAMP production in treated cells relative to untreated cells average of two experiments Ϯ half range of the two experiments. EC 50 . On the other hand, as expected for cells expressing low levels of receptors, the desensitization in low expressor fibroblasts results mainly in the reduction in the maximal GLP-1mediated cAMP response. For ⌬CT431 receptor mutant expressed in all three cell types, no desensitization could be induced by PMA (Fig. 5, lower panels). These findings indicate that PMA-induced desensitization of the GLP-1 receptor could be measured both in permanent and transient expression systems. Thus, we further characterized the desensitization properties of the phosphorylation site mutants expressed in fibroblasts (low expressors) and in COS cells.
The desensitization of the GLP-1-mediated cAMP production was tested in low expressor fibroblasts for each mutant receptor using two different clones to limit the effect of clonal variations. Table II shows for each clone the cell surface receptor expression and the remaining maximal cAMP production induced by GLP-1 (V max ) after PMA preincubation. In wild typeexpressing cells, PMA induced a ϳ60% reduction in the V max as compared with untreated cells. The reduction was less marked (30 -50%) in cells expressing receptors with single serine doublet substitution. When mutant receptors with two substituted doublets were studied, desensitization could no longer be observed, as was the case for the ⌬CT431 mutant that lacked all 4 serine doublets. Fig. 6 shows representative experiments performed with transiently transfected COS cells to determine the dose-response curves for PMA-induced desensitization of the wild-type receptor (top panel), the receptor with mutation of residues 431/432 to alanines (middle panel), and a double mutant (431AA/451AA, lower panel). Desensitization of the wild-type receptor is measured as a shift in the EC 50 of 2.4 Ϯ 0.2-fold (mean Ϯ S.E., n ϭ 3). When single mutants were tested, the shift was still about 2-fold (mutant S431A, 2.3-fold (Fig. 6); S444A, 1.11-fold; S451A, 2.4-fold). With mutants carrying two doublet substitutions, the shift in EC 50 was reduced to about 1.3-fold (mutant S431A/S451A, 1.2-fold (Fig. 6); range 1.2-to 1.4-fold for four other different double mutants tested). These results indicated that at least three phosphorylation sites must be present in order to induce a measurable desensitization of the receptor and that further mutation of any other serine doublet led to receptor that could not be desensitized. These results are in agreement with the experiments performed in fibroblasts.
To determine whether physiological activators of protein kinase C were able to induce receptor phosphorylation and desensitization, we performed two kinds of experiments. First, we transiently transfected COS cells with the GLP-1 receptor and exposed the [ 32 P]orthophosphate labeled cells to dioctanoyl-snglycerol for 10 min. Following immunoprecipitation, the receptor was analyzed by gel electrophoresis as described above. Fig.  7 shows that dioctanoyl-sn-glycerol induced wild-type receptor phosphorylation and that phosphorylation was abolished when all the identified PKC phosphorylation sites were mutated to alanines. Secondly, we showed that activation of the thrombin receptor, which is coupled to activation of the phospholipase C pathway (5), led to a marked desensitization of the GLP-1-dependent cAMP response in GLP-1 receptor-transfected fibroblasts (Fig. 8). Together these data demonstrate that activation of PKC in intact cells with a cell-permeable diacylglycerol or by an agonist of a phospholipase C-coupled receptor, reproduces the effects observed with PMA. Here, using a set of receptor mutants expressed in COS cells, we have identified the sites of the GLP-1 receptor that are phosphorylated following activation of PKC. We determined that four serine doublets located in the C-tail of the receptor at positions 431/432, 441/442, 444/445, and 451/452 were phosphorylated in intact cells following PMA treatment of the cells. This was inferred from two sets of observations. First, mutation to alanines of any single serine doublet markedly reduced the extent of phosphorylation, indicating that each doublet was phosphorylated. Second, each of the receptor mutants with a single serine doublet left could be phosphorylated after activation of PKC. Interestingly, the level of phosphorylation was not strictly correlated with the number of phosphorylation sites present, because removal of a single serine doublet led to a reduction in phosphate incorporation by about 50%. This there-fore suggests some cooperativity in the phosphorylation process, as reported previously for several other proteins (25). Among the four different phosphorylation sites only one, serines 431/432, is a classical phosphorylation site for PKC. To determine whether the identified phosphorylation sites are actually PKC substrates, we performed in vitro phosphorylation reactions on fusion proteins containing the different mutated forms of the receptor. These experiments indicated that the in vitro phosphorylation sites were different from those identified in intact cells. Specifically, serine 416, which is in a PKC consensus site, is a good in vitro substrate for bovine brain PKC but is not phosphorylated in intact cells. Serines 431/432, which are also in a consensus PKC site, are also strongly phosphorylated in vitro, whereas serines 444/445 and 451/452 are poor PKC substrates, and serines 441/442 are not substrates. All these sites are, however, phosphorylated almost equally well in intact cells as determined with the mutants having only one serine doublet left. These observations thus indicate that the phosphorylation sites in vitro and in intact cells are at least in part different. This may result from the existence of different isoforms of PKC in COS cells as compared with the bovine brain kinase preparation. Alternatively, the conformation of the native receptor prevents phosphorylation of consensus sites by PKC such as serine 416. Still another possibility is that PKC may activate other kinases for which the receptor is a substrate. In that respect, it has recently been demonstrated that the ␤-adrenergic receptor kinase-1 can be activated by phosphorylation by protein kinase C (26). However, for ␤-adrenergic receptor kinase-1 to phosphorylate the receptor, the receptor has to be in agonist-bound form. In our conditions, receptor phosphorylation after treatment with phorbol esters is observed independently of agonist binding, suggesting that if another kinase is activated by PKC it may be different from ␤-adrenergic receptor kinase-1.
Our findings provide strong evidence that phorbol esterinduced phosphorylation is strictly related to desensitization of the GLP-1 receptor expressed both in fibroblasts and COS cells. This was shown, in fibroblasts, by the fact that removal of any pair of serine doublets created receptors that were totally resistant to the desensitization, whereas mutation of any one of the four serine doublets led to a receptor that was still desensitized, although to a reduced extent as compared with the wild-type receptor. Because the receptor mutants lacking one and two serine doublets are phosphorylated to ϳ50 and ϳ25%, respectively, of the wild-type receptor, it appears that a threshold of phosphorylation comprised between 25 and 50% of the maximal phosphorylation is required to mediate the desensitization induced by PMA.
In COS cells, the results were similar to those obtained in fibroblasts. Deletion of any pair of serine doublets led to receptors that were no longer desensitized. Mutation of a single phosphorylation sites led to a small reduction in the shift of EC 50 , although it is difficult, with the small variations in EC 50 obtained, to have very significant values between those of the wild-type receptor and those of the desensitization-resistant mutants. However, these data confirm that maximal desensitization is obtained only when all the four serine doublets can be phosphorylated and that phosphorylation on two sites is not sufficient for desensitization to be observed.
Importantly, we demonstrated that physiological activators of protein kinase C such as the diacylglycerol dioctanoyl-snglycerol or thrombin can induce phosphorylation and desensitization of the receptor to a similar extent as observed with PMA. Our present data using PMA as an activator of PKC are thus relevant to more physiological situations in which a GLP-1 receptor-expressing cells is subjected to activation of PKC by agonists of phospholipase C-coupled receptors.
We previously reported that agonist-induced desensitization of the receptor also involves phosphorylation of the receptor on the same last 33-amino acid segment of the receptor C-tail. Furthermore, PKC and agonist-induced desensitization and phosphorylation were additive, and the phosphorylation induced by PMA was blocked by the RO-318220 inhibitor, whereas that induced by agonist binding was not impaired. This suggests that the phosphorylation sites involved in homologous versus heterologous desensitization are different. In the present work we have not determined which individual serine of the different doublets were phosphorylated in response to PKC activation. It is possible that different residues on each doublet are phosphorylated in homologous and heterologous desensitization. The C-tail of the receptor also contains two more serine residues at position 461 and 463, which could potentially be phosphorylated. It will thus be important to determine which serine residues are phosphorylated by the agonist-activated kinase and whether this kinase is one of the already characterized G-coupled receptor-specific kinases (27).
There are a number of hormones and neurotransmitters, such as acetylcholine, that can activate PKC in pancreatic ␤ cells and thus stimulate insulin secretion. We have previously suggested that these secretagogues induce desensitization of the GLP-1 receptor in order to prevent overstimulation of insulin secretion and thus hypoglycemia that could occur in the combined presence of these secretagogues and GLP-1. However, our data indicate that at least three serine doublets must be phosphorylated to induce some desensitization but that maximal desensitization is achieved only when all four doublets are phosphorylated. What could be the biological significance of these observations? Glucose-induced secretion is a very tightly regulated mechanism, and insulin oversecretion should not occur. Thus when an agonist only moderately activates the phospholipase C pathway and, consequently, mildly stimulates insulin secretion, no desensitization of the GLP-1 receptor is needed and partial phosphorylation of the receptor leaves the receptor fully active. Only in conditions of strong stimulation of the PLC pathway will the receptor be fully phosphorylated and desensitized to prevent overstimulation of insulin secretion by increasing GLP-1 plasma levels. This finely regulated phosphorylation and desensitization of the GLP-1 receptor may thus have evolved to tightly control insulin secretion by a variety of hormones that are key in the complex control of glucose homeostasis.