INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

DIABETES REMAINS A MAJOR PUBLIC health concern, with a high frequency and many associated outgrowing genetic anomalies. It is characterized by the loss of postprandial insulin secretion by -cells in response to glucose and nutrients. Desensitization of pancreatic -cells leads to progressive reduction of the insulin -cell population and chronic hyperglycemia (19). Most of the mechanisms behind this alteration in signal-secretion coupling and insulin status are unclear. Hyperglycemia impairs -cell function in type 2 diabetes through multiple dysfunctions along the cAMP pathway, insulin secretion, and biosynthesis (35). Alterations in G protein-coupled receptor signaling constitute the potential targets for pancreatic -cell desensitization in diabetes (13, 38). In -cells, the glucose-dependent insulinotropic polypeptide (GIP) and its structurally related peptides, truncated glucagon-like polypeptide 1 (tGLP-1) and pituitary adenylyl cyclase (AC)-activating polypeptide, are critical activators of the cAMP pathway and potentiate glucose-induced insulin release, insulin gene transcription, and -cell growth (13).

The membrane-bound ACs are critical effectors of the Gs family of G-protein subunits acting through cAMP production and protein kinase A (PKA) activation. Recent studies demonstrated that the cAMP pathway is also connected with several other downstream signaling cascades, including cyclic nucleotide-gated ion channels and Epacs (exchange protein directly activated by cAMP), which are guanine-nucleotide-exchange factors (cAMP-dependent GEFs) for the small G protein Rap1 (12, 29).

Golf, a member of the Gs family originally discovered in the olfactory neuroepithelium and striatum, was subsequently identified in peripheral tissues, including testis, spleen, lung, and heart (16, 23, 37, 46). Recently, we identified Golf in pancreatic -cells by in situ hybridization, immunocytochemistry, and electron microscopy (2). In Langerhans islets, Golf is located inside the insulin-secretory granules and -cell plasma membrane, suggesting that it is involved in granule migration, processing, and exocytosis. Golf has an 88% amino acid homology with Gs, as it comprises three conserved functional domains involved in guanosine triphosphate (GTP)-binding affinity, GTPase activity, receptor-dependent GTP binding, and GTP-induced conformational change (22). To date, the molecular partners of the Golf-signaling cascade in -cells, including upstream activation by serpentine receptors and Golf downstream transduction elements, are still unknown. Due to the sequence divergences between Golf and Gs (23, 27), we postulated that these two G-protein subunits receive distinct upstream extracellular signals and exert a differential functional activation on the AC isoforms downstream.

Nine AC isoenzymes, AC-I to -IX, and two splice variants of AC-VIII have been identified in mammals. Most of these isoenzymes are present in Langerhans islet -cells and are overexpressed in diabetic animal models (21, 28, 40). Recently, a new mammalian soluble AC (sAC) isoenzyme insensitive to forskolin (FK) was described in testis (8).

The present study was therefore conducted to identify the AC isoforms preferentially activated by Golf in transient transfection assays and the cellular functions controlled by this G protein in mouse insulin-secreting TC-3 cells. For this purpose, we first cotransfected the human embryonic kidney cell line HEK-293T using a constitutively active GTPase-deficient form of Golf (GolfQ214L) and of Gs (GsR201C), respectively, designated below as AGolf and AGs, together with expression vectors encoding each of the individual AC isotypes (AC-I to -VIII) or sAC. Second, we established the TC-3 cell lines stably transfected with wild-type (wt)Golf or AGolf to explore the potential role of Golf on the insulin status and cell survival, which constitute two major impacts on the functional -cell abnormalities in type 2 diabetes. Thus we measured insulin accumulation, secretion, and the expression of the prohormone convertases (PC) 1 and 2 in TC-3-wtGolf and -AGolf-transfected cells (39). We showed 1) that the G protein Golf preferentially activates AC-I and AC-VIII in transiently transfected HEK-293T cells, 2) that overexpression of wt- or AGolf weakened cAMP and insulin responses to the G protein-coupled receptors tGLP-1 in TC-3 cells, and 3) that AGolf protected -cells from the apoptosis induced by serum withdrawal.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cell culture and human insulinomas. Human embryonic kidney HEK-293T epithelial cells were maintained in DMEM containing 25 mM glucose (GIBCO-BRL, Life Technologies) supplemented with 10% fetal bovine serum (FBS; Boehringer Mannheim), L-glutamine, and antibiotics. The human collecting duct (HCD) cell line immortalized by SV40 virus, a generous gift from Dr. P. Ronco (Institut National de la Santé et de la Recherche Médicale U489), was cultured in DMEM consisting of Ham's F-12 medium (vol:vol) containing 2% FBS, 5 µg/ml transferrin, 50 nM sodium selenite, 2 mM L-glutamine, 0.5 nM dexamethasone, 5 µg/ml insulin, and 20 mM HEPES (36). Pancreatic TC-3 cells derived from transgenic mice expressing SV40-T antigen in Langerhans -cells were a generous gift from Dr. Shimon Efrat (Albert Einstein College of Medicine, Bronx, NY). TC-3 cells were grown in DMEM containing 5 mM glucose supplemented with 12.5% horse serum, 2.5% FBS (Biowest, Paris, France), L-glutamine, and antibiotics (10). Human insulinomas were well-differentiated endocrine tumors by microscopic examination and according to the detection of insulin by immunohistochemistry.

Plasmid DNA preparations. Mammalian expression vector encoding wtGolf was constructed by a 2.7-kbp NotI-SalI fragment excision from pBluescript SK⁺, corresponding to whole rat Golf cDNA, and inserted in a frame into the corresponding sites of the pCI-neo vector (Promega). To generate the hemagglutinin (HA)-tagged and GTPase-deficient mutant of Golf encoding the constitutively activated form of this G protein (HA-AGolf), a first generation of DNA fragments of 3'-EcoRI/Bsu36I and Bsu36I/BglII-5' was amplified by PCR. With the use of these fragments as a template, a second generation of 3'-EcoRI/BglII-5' fragments was amplified, thus introducing a point mutation in Golf by replacing glutamine 214 with leucine (GolfQ214L), which is involved in constitutive GTP binding to Golf and in the inhibition of intrinsic GTPase activity. The primer sequences are available on request. The PCR-amplified 1.15-kbp DNA fragment was inserted in a frame into the corresponding sites of pCEFL, a modified pcDNA3 expression vector containing the elongation factor 1 promoter that drives the expression of an inframe NH₂-terminal tag of nine amino acids of the HA epitope. All constructs were verified by DNA sequencing and expression in HEK-293T cells. Sequences were also identified by BLAST analysis of the GenBank database. The cDNA encoding the constitutively activated form of the G-protein subunit GsR201C (HA-AGs) was cloned in the expression vector pCEFL. sAC (8) was a generous gift from Dr. L. Levin (Weill Medical College, Cornell University, NY). The membrane-bound AC isoforms AC-I, -II, and -V in pXMD1 (14, 24, 43) were a generous gift from Dr. T. Pfeuffer (Zentrum für Molekulare Biologie, Heidelberg, Germany); AC-III, -IV, -VI, and -VII in pXMD1 (3, 4, 18, 25, 44) were gifts from Drs. R. R. Reed, A. Gilman (University of Texas, Austin, TX), and P. Watson (Weis Center, Geisinger Clinic, Danville, PA); and AC-VI and -VIII in pCMV-neo (9, 25) were gifts from Dr. Krupinski (Bristol-Myers Squibb, Princeton, NJ). The pCMV--gal vector expressing -galactosidase was purchased from Clontech Laboratories (Palo Alto, CA).

Transient cotransfection assays. HEK-293T cells were plated at the density of 75 × 10³ cells per 6-well petri dish and cultured for 24 h. Cells were transiently transfected using LipofectAMINE Plus reagent, according to the manufacturer's protocol (GIBCO-BRL). Briefly, HEK-293T cells were transfected with the AGolf (1 µg) or AGs (0.2 µg) expression vectors, either alone or combined with one of the AC expression vectors (0.2 µg), together with 0.1 µg pcDNA3--gal. To transfect constant amounts of DNA, samples were supplemented with the control vector pCEFL. Transfection efficiency of 40-50% was optimized and determined by both X-GAL staining and -galactosidase activity using the Luminescence Galacto-Star kit (Tropix, Bedford, MA). After incubation at 37°C for 6 h in a humidified atmosphere of 5% CO₂ and 95% air, the remaining DNA-liposome complexes were removed. HEK-293T cells were then cultured for 24, 48, or 72 h in the standard medium and challenged for the cAMP assay described below.

Stable transfection of TC-3 cells. Approximately 2 × 10⁶ TC-3 cells in 100-mm-diameter petri dishes were stably transfected by the LipofectAMINE Plus reagent with the pCI-neo-wtGolf or pCEFL-AGolf expression plasmid. Transfection efficiencies were 10-12%, as determined by -galactosidase staining. After 3 days, transfected cells were selected for 3 wk of culture in 400 µg/ml G-418 (Geneticin, GIBCO-BRL, Life Technologies). Resistant colonies were either ring-cloned as individual colonies of TC-3-wtGolf cells or pooled as TC-3 cells expressing constitutively activated AGolf (TC-3-AGolf pools). Cells were subjected to analysis of ectopic overexpression of the Golf transgenes by immunoblot analysis or RT-PCR and further tested for cAMP generation, insulin status, and cell survival.

Western blot analysis. For immunoblotting, cultured cells were homogenized for 30 min at 4°C in lysis buffer containing 20 mM HEPES buffer at pH 7.5, 10 mM EGTA at pH 8.0, 2.5 mM MgCl₂, 40 mM -glycerophosphate, 2 mM sodium orthovanadate, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 1% Triton X-100, and 2% leupeptin and aprotinin as phosphatase and protease inhibitors. Proteins were resolved by 12.5% SDS-PAGE and transferred to polyvinylidene difluoride membranes (Amersham Pharmacia Biotech, Little Chalfont, UK). The membranes were blocked with 10% defatted milk and probed for 4 h at room temperature with one of the following antibodies: the monoclonal antibody (mAb) 3F10 (1:500) that recognizes the HA peptide sequence YPYDVPDYA of the human hemaglutinin protein (Hoffmann-La Roche, Basel, Switzerland); the Golf polyclonal antibody (pAb) K19 (1:500) (Santa Cruz Biotechnology, Santa Cruz, CA); the mAb AC-15 specific for -actin (1:4,000) (Sigma-Aldrich, Saint-Quentin Fallavier, France); and the pAbs 2B6 and 4BF (39) directed against the protein convertases PC-1 and PC-2 (1:1,000), a generous gift from Dr. P. Kitabgi (Centre National de la Recherche Scientifique, Valbonne, France). After being blocked, membranes were washed in PBS containing 0.1% Tween-20 and probed for 90 min with a secondary antibody (1:2,000) consisting of horseradish peroxidase (HP)-linked goat anti-mouse IgGs pAb (Santa Cruz Biotechnologies) or HP-linked donkey anti-rabbit IgG (Amersham). Immunoblots were revealed by enhanced chemiluminescence (ECL) Plus Western detection from Amersham Pharmacia Biotech.

RT-PCR. Total RNA was extracted by the TRIzol reagent (GIBCO-BRL, Life Technologies) from human tissues (insulinomas and their adjacent pancreatic tissues and testis) and cultured cell lines (HEK-293T, HCD, and TC-3 cells). RT-PCR was performed using the Superscript detection kit (GIBCO-BRL, Life Technologies), according to the manufacturer's protocol. The experiment comprised one cycle of 30 min at 55°C and one of 2 min at 94°C followed by 40 cycles of 30 s at 94°C, 30 s at 55°C, and 1 min at 72°C, with a final extension time of 10 min at 72°C. For amplification of the human and rat Golf cDNAs, human insulin, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNAs, we used the following sense and antisense oligonucleotides (Genset, Ivry, France): hGolf, 5'-CAGCGTCAGCTTGGTTGACTA-3' and 5'-GTAATGTTTGCCGTCACCGGT-3'; rGolf, 5'-ATGGGGTGTTTGGGCAACAGC-3' and 5'-ACTCACGCTGTCAATCCTTTC-3'; human insulin, 5'-TGCATCAGAAGAGGCCATCAAGCA-3' and 5'-GAGAGATGGAATAAAGCCCTTGAAC-3'; and the human and mouse GAPDH, 5'-ATCACCATCTTCCAGGAGCG-3' and 5'-CAAGAAGGTGGTGAAGCAGG-3'. The primers amplified the 526-bp fragment specific for the hGolf transcript, the 519-bp rGolf amplicon, and the 444- and 574-bp products corresponding to the insulin and GAPDH messengers, respectively. PCR products were then separated by electrophoresis in 1.5% agarose gel and detected under ultraviolet light. DNA sequencing (Genset) confirmed the specificity of the amplified products for human and rat Golf.

cAMP assay. For measurement of cAMP generation in transiently transfected HEK-293T cells (24-72 h posttransfection), the culture medium was replaced by 1 ml DMEM containing 1% BSA, 20 mM HEPES, and 1 mM isobutylmethylxantine (IBMX), a phosphodiesterase inhibitor. After 10 min of preincubation at 37°C, cells were further incubated for 1 h either alone or in the presence of the AC activators, 10 µM FK (Sigma-Aldrich), or 5 mM MnCl₂. Parental TC-3 cells stably transfected with wtGolf and AGolf (3 × 10⁵ cells) were cultured for 4 days before cAMP assay. The assay was performed in Krebs-Ringer bicarbonate (KRB) buffer containing 129 mM NaCl, 5 mM NaHCO₃, 4.8 mM KCl, 1.2 mM KH₂PO₄, 2.5 mM CaCl₂, 1.2 mM MgSO₄, 5 mM glucose, 0.1% BSA, and 10 mM HEPES (pH 7.4). -Cells were incubated in 3 ml KRB buffer for 1 h at 37°C, alone or in the presence of 0.5 mM IBMX, or 0.1 µM tGLP-1 or both (Sigma-Aldrich). Aliquots of the supernatant (200 µl) were then removed for the insulin assay. HEK-293T or TC-3 cells were lysed by the addition of ice-cold perchloric acid (1.1 M final concentration). cAMP content was measured by RIA (Amersham Pharmacia Biotechnology) using the succinylated ¹²⁵I-cAMP RIA competition method (7). Data were expressed as picomoles of cAMP produced per milligram protein for HEK-293T cells or per 10⁶ cells for TC-3 cells.

Insulin assay. After incubation of TC-3 cells for 1 h at 37°C, insulin secretion was measured in the 200-µl KRB buffer aliquot used for cAMP determination. Insulin secretion was also measured in parental TC-3 cells and their stably transfected counterparts TC-3-wtGolf and TC-3-AGolf cells after 24 h of culture (10). Supernatants were then saved for insulin determination, and cells were lysed in ethanol acid for measurement of their total insulin content. Media and cell extracts were centrifuged, and the supernatants were stored at 20°C. Samples were then assayed by RIA using a commercially available radioimmunoassay kit (Insulin-CT-RIA, CIS Bio International, Gif-sur-Yvette, France). Data were expressed as milliunits of insulin per 10⁶ cells.

Apoptosis assay. TC-3 parental cells and stably transfected TC-3-AGolf cells were plated at the density of 75 × 10⁴ cells in petri dishes (60-mm diameter) in the presence or absence of serum. For fluorescence-activated cell sorter (FACS) analysis, floating cells were combined with trypsinized adherent cells and fixed in 70% ethanol. The fixed cells were stained with 50 µg/ml propidium iodide (Sigma- Aldrich) in PBS containing 0.5 mg/ml RNAse (Hoffmann-La Roche) and assayed as previously described (20). Cell apoptosis was analyzed by a FACS Calibur apparatus (Becton Dickinson) equipped with an argon laser tuned to 488 nm. About 10,000 cells were recorded per assay.

Statistical analyses. Data are means ± SE for the number of experiments performed in duplicate or triplicate. The significance of the differences between experimental values was assessed by the unpaired Student's t-test, and a P value <0.05 was considered to be statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Golf is expressed in human normal pancreas, insulinomas, testis, and kidney epithelial cells. We tested various protocols to obtain effective transfection of the insulin-secreting TC-3 cell line with the expression vectors encoding AGolf, AGs, and ACs. Because TC-3 cells were hardly transfectable in both stable and transient transfection assays, pilot experiments were performed with the human embryonic kidney HEK-293T epithelial cells, which are susceptible to high expression of transgenes after transient transfection by the LipofectAMINE method. We therefore used RT-PCR to confirm the expression of the Golf transcripts in human pancreas, insulinomas, and testis, and to investigate the possible expression of the Golf gene in human kidney epithelial cells (Fig. 1). The predicted size of the 526-bp Golf product was characterized in three human insulinomas (Fig. 1A, lanes 1-3). As expected, the Golf transcripts were much more abundant in the three insulinomas, compared with their corresponding nontumorous pancreatic tissues (control tissues), because only specimen 2 was positive (Fig. 1B, lanes 1-3). Our data therefore suggest that Golf can play a regulatory role during the transformation of the pancreatic -cells, including deregulation of insulin biosynthesis and secretion. The Golf transcript was also detected in a human testis as positive control, in human embryonic kidney HEK-293T cells, and adult immortalized kidney epithelial HCD cells (Fig. 1C). To confirm the accuracy of these results, the Golf PCR products were sequenced and analyzed in the GenBank database and were 100% identical to the corresponding human Golf sequence. As previously shown in our laboratory, Golf was specifically expressed in insulin granules of pancreatic -cells. To confirm this observation, we also amplified the 444-bp fragment corresponding to the human insulin transcript in all three of the above insulinomas and normal pancreatic tissues (Fig. 1B). As internal RT-PCR control, the 574-bp fragment of human GAPDH was detected in all human tissue samples.

View larger version (46K): [in this window] [in a new window]   Fig. 1.   Golf expression in human insulinomas, pancreas, and kidney epithelial cells. Total RNAs were extracted from 3 human insulinomas (A), their paired control pancreatic tissues (lanes 1-3; B), human testis (C), human embryonic kidney cells HEK-293T, and the SV40-immortalized human collecting duct (HCD) cell line. RT-PCR was performed as described in MATERIALS AND METHODS. In all cases, the specificity of the amplified bands was confirmed by DNA sequencing. Three independent batches of RNA gave the same results. Human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) is shown in bottom panel of all 3 sections, normalizing the loading samples.

Transient expression of ectopic AGolf and AGs enhances cAMP generation in human embryonic kidney HEK-293T cells. Because HEK-293T cells express the Golf gene, we used this homologous system to study the possible effect of this G protein on the endogenous ACs (Fig. 2). We therefore generated a eukaryotic expression vector for AGolf, as described in MATERIALS AND METHODS. The recombinant AGolf vector was engineered so as to have an in-frame HA-epitope-tag at the NH₂-terminal end of this protein. To determine the optimal level of ectopic AGolf expression and function, the kinetics of its expression were monitored for 24, 48, and 72 h, using transient transfection of HEK-293T cells with 1 µg AGolf plasmid (Fig. 2A). Maximal accumulation of the HA-AGolf protein (46 kDa) was observed 48 h after transfection, and similar kinetics were observed for HA-AGs (52 kDa), as shown in Fig. 2A. We used the same filter for -actin immunoblot, which acted as protein-loading control. HEK-293T cells were transfected by different amounts of expression vectors for AGolf (1 µg) and AGs (0.2 µg) to identify the kinetics of their maximal expression (48 h). Lower amounts of the AGolf expression vector (0.1 to 0.5 µg) resulted in lower signals, using the Western blot and cAMP assays. The observed higher expression of AGs vs. AGolf subunit (1.7-fold at 24 and 72 h; 2.4-fold at 48 h) was likely related to the differential stability of the corresponding messengers and proteins, because the HA antibody has similar affinity for this epitope in the HA-tagged AGs and AGolf proteins. The aim of this study was to identify the AC isotype preferentially activated by Golf. In our experiment, the AGs protein also constitutes an internal control of the functional activity of a given AC isotype introduced, compared with FK.

View larger version (41K): [in this window] [in a new window]   Fig. 2.   Kinetics of the transient expression of ectopic AGolf and AGs proteins in human embryonic HEK-293T kidney cells: Western blot analysis and cAMP production. Cultured HEK-293T cells (75 × 103 cells/dish) were transiently transfected with the pCEFL plasmid vector, alone, or recombined with the cDNA encoding the hemagglutinin (HA) epitope-tagged AGolf or AGs subunits, together with the pCMV--gal expression vector, using the LipofectAMINE Plus reagent. A: at the times indicated, cell lysates were collected and assayed for the expression of HA-tagged AGolf and AGs by immunoblot. The accompanying -actin immunoblot of the same filter served as protein-loading control. B: in parallel experiments, total cell lysates and conditioned media underwent cAMP extraction and measurement by radioimmunoassay, as described in MATERIALS AND METHODS. cAMP generation was expressed as the percentage of control cAMP activation measured in HEK-293T cells transfected with AGolf or AGs, and it was compared with the cotransfections performed with the empty control vector pCEFL. Data are means ± SD of 4 independent experiments performed in duplicate and are normalized for -galactosidase activity and protein content.

Selective activation of AC isoforms by AGolf and AGs. With the use of transient cotransfection assays, we compared the sensitivity of eight distinct AC isoforms to AGolf and AGs in HEK-293T cells (Fig. 3). First, we attempted to establish whether or not the ectopic AC isoforms introduced were functionally competent for cAMP generation in our experimental model (Fig. 3A). For this purpose, HEK-293T cells were transiently transfected by a given AC isotype (ACI-VIII) or the pCEFL control vector and challenged for 1 h by the AC activator FK. The amount of cAMP generated by a given ectopic AC isoform was reduced by its corresponding intrinsic activity measured in the absence of FK and indexed by the same cAMP values as those obtained with the endogenous AC, i.e., FK (ectopic AC)  ectopic AC, vs. the values obtained with endogenous AC: FK (pCEFL)  pCEFL. As shown in Fig. 3A, transfection of HEK-293T cells by the ACI-VII isoforms induced an increase in cAMP generation 1.5- to 3-fold greater than that measured with endogenous AC. In the case of AC-VIII, FK-stimulated cAMP levels rose 10-fold more than with endogenous AC, suggesting that this isoform is more sensitive to effective stimulation by FK. In parallel experiments, we compared the ability of AGolf or AGs to mimic the effect of FK on the ectopic ACI-VIII (Fig. 3A). We first observed that AGolf induced 0.9- to 27-fold increases in the activity of the ectopic AC isoforms introduced by transient transfection into HEK-293T cells, the largest increases being those for AC-I and -VIII (15- and 27-fold, respectively). In comparison, AGs enhanced the activity of all ectopic AC isoforms considered ~45- to 60-fold. Accordingly, AGolf was ~40% as effective as AGs in stimulating ectopic AC-I and -VIII; in the presence of ectopic AC-I, total cAMP levels generated were increased from 42 ± 6 to 245 ± 23 and 487 ± 63 pmol/mg protein by AGolf and AGs, respectively. Similarly, cAMP generation in HEK-293T cells transfected by the AC-VIII expression vector was 197 ± 15 pmol/mg protein in control cells and was further increased to 1,709 ± 380 and 4,399 ± 340 pmol/mg protein by AGolf and AGs, respectively.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Differential activation of ectopic adenylyl cyclase (AC) isoforms by AGolf and AGs in HEK-293T cells. A: cAMP generated by the AC isoforms (0.2 µg plasmid cDNA) was measured in HEK-293T cells and expressed as the fold elevation of forskolin (FK)-induced activation by each ectopic AC minus its basal activity vs. the corresponding values obtained from endogenous AC in HEK-293T cells submitted to mock transfections by the control pCEFL vector. B: relative activation of ectopic AC isoforms I to VIII by AGolf vs. AGs. HEK-293T cells were transfected with 0.2 µg plasmid cDNA encoding 1 of the AC isoforms considered, alone, or combined with pCEFL-AGolf or pCEFL-AGs, and 0.1 µg pCMV--gal plasmid for transfection efficiency normalization. Results are expressed as the percentage of ectopic AC activation by AGolf, minus endogenous AC activation by AGolf, over the corresponding values obtained by AGs activation of ectopic and endogenous ACs. Data are means ± SD of 4 independent experiments performed in duplicate.

View larger version (22K): [in this window] [in a new window]   Fig. 4.   Differential activation of ectopic soluble AC (sAC) by AGolf and AGs in HEK-293T cells. cAMP production was measured in HEK-293T cells transiently transfected with 0.2 µg of the pBK-CMV vector carrying the full-length sequence of sAC, either alone or combined with pCEFL-AGolf or pCEFL-AGs, and 0.1 µg pCMV--gal plasmid for transfection efficiency normalization. The activity of endogenous AC and ectopic sAC in HEK-293T cells was also measured with or without 5 mM MnCl2. Results are expressed as percent of the maximal cAMP production induced by AGs + sAC. Data are means ± SD of 4 independent experiments performed in duplicate.

Establishment and characterization of TC-3 cell lines stably transfected with wt- or constitutively activated AGolf. TC-3 cells were stably transfected with wtGolf or constitutively activated AGolf using the LipofectAMINE Plus reagent. After selection by G-418, individual colonies were ring-cloned or pooled for further analysis. As shown in Fig. 5A, we isolated by immunoblot seven stable TC-3 clones overexpressing the wtGolf protein (46-kDa band). This overexpression was confirmed by RT-PCR in clones 3, 8, 9, and 14 of TC-3 cells stably transfected with wtGolf, which expressed the expected size of the 519-bp product corresponding to the Golf transcript (Fig. 5B). We also identified, by RT-PCR, four pools of colonies that were positive for the ectopic overexpression of AGolf (pools 1-4). We therefore decided to extend our studies of cAMP generation and insulin status in stably transfected TC-3 cells overexpressing wtGolf (clones 9 and 14) or AGolf (pool 2).

View larger version (59K): [in this window] [in a new window]   Fig. 5.   Ectopic overexpression of wild type (wt) and AGolf by imunoblotting and RT-PCR in stably transfected TC-3 cells. A: -cells transfected with wtGolf were selected by the neomycin analog G-418 (Geneticin), and resistant colonies were amplified. Equal amounts of protein (200 µg) were separated by SDS-PAGE and subjected to immunoblotting using either the polyclonal antibody K-19 directed against Golf or the monoclonal antibody AC-15 specific for -actin as loading control. Stable overexpression of wtGolf (B) and AGolf (C) revealed by RT-PCR in neomycin-resistant TC-3 cells; the transcript of the Golf transgene was clearly detected as the expected 519-bp product and was normalized by the 574-bp product of GAPDH. Data are representative of 3 experiments.

cAMP generation and insulin secretion in stably transfected TC-3-wtGolf or -AGolf cells. As shown in Fig. 6A, cAMP levels in parental TC-3 cells rose from the basal level of 14.3 ± 1.9 pmol/10⁶ cells to 28.7 ± 5.2 and 41.2 ± 7.2 pmol/10⁶ cells after the addition of the phosphodiesterase inhibitor IBMX (0.5 mM) or the insulinotropic hormone tGLP-1 (0.1 µM). Combined addition of IBMX and tGLP-1 resulted in a synergistic cAMP response consisting of a 12.5-fold increase over basal levels. Similarly, ectopic overexpression of wtGolf exerted cooperative action on the cAMP generation induced by tGLP-1, alone or combined with IBMX, which was not observed when TC-3-AGolf cells were challenged with these two effectors. In agreement with the results for cAMP, Fig. 6B shows that combined addition of IBMX and tGLP-1 enhanced insulin secretion in parental TC-3 cells from the basal level of 1.7 ± 0.3 to 4.4 ± 1 mU/10⁶ cells. In contrast, the response of insulin was abolished in both TC-3-wtGolf and -AGolf cells when they were challenged with IBMX, tGLP-1, or both. Less insulin was always secreted in TC-3 cells expressing wtGolf than in parental and TC-3-AGolf cells. Thus long-term activation of the cAMP pathway by wt- and AGolf affected the insulin secretion capacity of TC-3 cells in response to activation by the serpentine tGLP-1 receptor. To explore this hypothesis, we next examined the insulin status (insulin secretion and content) of parental and stably transfected TC-3-wtGolf and -AGolf cells cultured for 24 h in standard medium.

View larger version (33K): [in this window] [in a new window]   Fig. 6.   cAMP generation (A) and insulin secretion (B) in TC-3 cells stably transfected with wt- and AGolf. Parental and transfected TC-3 cells were incubated for 1 h at 37°C, in the absence (Control) or presence of the phosphodiesterase inhibitor isobutylmethylxantine (IBMX; 0.2 mM), truncated glucagon-like polypeptide 1 (tGLP-1; 0.1 µM), or both. Cells then underwent RIA for cAMP or insulin. Data are means ± SD of 3 independent experiments performed in triplicate.

Insulin status and expression of the proinsulin convertases PC-1 and PC-2 in parental and stably transfected TC-3-wtGolf and -AGolf cells. The inability of -cells to increase insulin secretion, processing, and biosynthesis coincided with the onset of hyperglycemia. As shown in Fig. 7A, stable overexpression of wt- and AGolf in TC-3 cells reduced constitutive insulin secretion by 50-75%, whereas their corresponding insulin content was reduced by only 25-50%. Insulin content was 13.1 ± 1.4, 19.9 ± 0.3 (P < 0.009), and 26.1 ± 2.1 mU/10⁶ cells (P < 0.007) in wt-, AGolf, and parental TC-3 cells, respectively. However, the overstimulation of the cAMP pathway (Fig. 6A) and the decreased insulin secretory capacity observed here in TC-3-wtGolf and -AGolf cells (Figs. 6B and 7A) were not due to any change in the expression of the proinsulin convertases PC-1 and PC-2 (Fig. 7B), which are located together with Golf in the insulin secretory granules (33).

View larger version (29K): [in this window] [in a new window]   Fig. 7.   Insulin secretion, insulin content, and expression of the PC-1/ PC-2 convertases in TC-3 cells stably transfected by wtGolf or AGolf. A: insulin secretion and content (mU/106 cells) were measured in parental and TC-3 cells stably transfected with wtGolf and AGolf and cultured for 24 h. Data are means ± SD of 3 independent experiments performed in triplicate. B: Western blot analysis of the endoproteases PC-1 (66 and 87 kDa: lanes 1-3) and PC-2 (66 and 75 kDa: lanes 4-5) in TC-3 cells, before (parental cells: lanes 1 and 4) and after ectopic expression in these cells of wtGolf (lanes 2 and 5) or AGolf (lanes 3 and 6). Data were normalized by -actin staining (43 kDa) and are representative of 2 other experiments.

Overexpression of Golf induced TC-3 cell survival. As hyperglycemia occurs when a relative insulin deficiency appears, -cell secretory dysfunction is a key element in type 2 diabetes. Up till now, insulin secretion deficiency has been attributed to pancreatic -cell "exhaustion" processes. Recent data suggest that apoptotic mechanisms explain insulin deficiency through a reduction in the absolute number of pancreatic -cells (31, 32). Although the mechanisms responsible for inducing -cell apoptosis are largely unknown, signaling through the G protein-coupled receptors linked to the cAMP-dependent signaling pathway has been suspected to be involved in cell survival (45). We therefore used propidium iodide staining and FACS analysis to examine the role of AGolf in the survival of cultured -cells after serum withdrawal as an apoptotic inducer (Fig. 8). Accordingly, parental TC-3 and TC-3-AGolf were deprived of serum for 24 and 48 h, and the apoptotic response was assayed. The results were compared with those for serum-treated cells. AGolf promoted the survival of cultured mouse -cells deprived of serum for 24 h, and this protective effect was potentiated at 48 h. As demonstrated in Fig. 8, the number of cultured parental TC-3 cells that underwent apoptosis in the absence of serum rose significantly from 11.6% at 24 h to 18.6% at 48 h, whereas the average proportion of serum-treated cells undergoing apoptosis during this period was only 8%. In the presence and absence of serum, AGolf protected TC-3 cells from apoptosis. Only 3% of TC-3-AGolf cells underwent apoptosis under all conditions tested. We observed that stable overexpression of wtGolf was associated with 5% of apoptotic cells in TC-3-wtGolf cells that were serum starved for 24 h (data not shown). This finding is consistent with the observation that wtGolf also induced long-term activation of the cAMP pathway, leading to alterations in the insulin secretion capacity.

View larger version (22K): [in this window] [in a new window]   Fig. 8.   Effect of ectopic expression of AGolf in TC-3 cells on the apoptosis induced by serum withdrawal. Parental and TC-3 cells stably transfected by AGolf were plated in 60-mm-diameter dishes (0.75 × 106 cells), with or without serum. After 24 (left) or 48 h (right), cultured TC-3 cells were stained with propidium iodide, and DNA content was quantified by fluorescence-activated cell sorter analysis (see MATERIALS AND METHODS). Data are means ± SD of 3 independent experiments performed in duplicate.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The initial postulate of this study was that Golf exerts preferential action on specific AC isoforms, according to its structural differences with Gs (23, 27). We also investigated the possible role of Golf on insulin status, using the mouse secreting TC-3 cell line in culture, because we found that Golf was present in insulin secretory granules of pancreatic -cells (2). Several studies by others and us (17, 21, 28) demonstrated that all of the membrane-bound ACI-VIII isoforms are present in -cells. In pilot experiments, we showed that the diterpene FK and constitutively activated GsR201C markedly simulated the enzyme activity of the ectopic ACI-VIII isoforms when these were transiently introduced into transfected HEK-293T epithelial cells. AC activation by FK and Gs is initiated by dimerization of the C1 and C2 catalytic core present in these cAMP-producing enzymes (42).

In the present study, we found that Golf stimulates the AC-I and -VIII isoforms preferentially. These two ACs are present in pancreatic islet -cells, and their levels of expression are strongly elevated in the Goto-Kakizaki (GK) rat, a widely accepted genetic rodent model of human type 2 diabetes (21). Thus Golf preferentially activates two AC isoforms that are overexpressed in the -cells of diabetic animals. Most importantly, the Ca²⁺/calmodulin signaling pathways that are critical mediators of insulin granule exocytosis also regulate AC-I and -VIII isoforms. The results of previous studies suggested that the Golf subunit has a functional role in AC-III isoform activation, because these two signaling proteins are coexpressed in olfactory epithelium, testis, pancreatic -cells, and GK rat islets (1, 4, 11, 17, 23, 46). Similarly, odorant receptors were suggested to activate AC-III via GTP-bound Golf (41), and two point mutations in the promoter region of the AC-III gene have been associated with overexpression of the AC-III isoform in diabetic GK rats (1). However, we demonstrated here that Golf exerts no functional interaction with the AC-III isoform in transiently transfected HEK-293T cells, indicating that such correlation may not have biological significance. The preferential action of Golf in activating both AC-I and -VIII might be due to the remarkable sequence homology between these two AC isotypes in the catalytic core. In contrast, the corresponding region of AC-III exhibits divergences between two amino acids A-1014 and N-1029, which might preclude functional interactions between Golf and AC-III, vs. AC-I and AC-VIII. Similarly, Gs and Golf also exhibited two differences between amino acids S352T and R356K at interacting AC sites. In addition, upstream signals emitted by the activated serpentine receptors can also greatly affect the relative efficiencies of Golf and Gs in interacting with and stimulating differentially the eight membrane-bound AC isoforms. These enzymes are key elements in the transmission of signals from various insulin secretagogues such as tGLP-1 and GIP, which are known to induce cAMP production, insulin biosynthesis, and secretion (13). Similar to AC-III, sAC was not affected here by Golf, whereas the divalent cation Mn²⁺ enhanced sAC-mediated cAMP generation, as previously described (8).

In the second part of this study, we demonstrated that the overexpression and constitutive activation of Golf in mouse TC-3 cells are associated with a decrease in insulin content and secretion. We noted with interest that wtGolf consistently elevated cAMP levels in TC-3 cells when these cells were challenged by tGLP-1 and IBMX. This observation suggests that constitutive activation of the cAMP-generating system by Golf overexpression induced the desensitization of the insulin release controlled by serpentine receptors. Similarly, we recently demonstrated that chronic stimulation of HIT-T15 -cells by glucose attenuates insulin release by inducing a nonfunctional state of both Golf and Gs (35). Furthermore, defects in the autophosphorylation and catalytic activity of cytosolic nucleoside diphosphokinase, through which GDP is converted to GTP on G proteins, were found in GK rats (34). This defective conversion is also complementary to the newly discovered mechanism by which insulin regulates its own secretion. Specifically, it has been demonstrated that, after insulin receptor knockout, mice manifest severe progressive glucose intolerance due to the loss of the glucose-stimulated acute phase of insulin release, which in -cells is identical to Golf overexpression (26). This insulin impairment in our studies is not due to the modification of insulin-processing machinery, because the expression levels of the PC-1 and PC-2 convertases remained unchanged in Golf-transfected cells. cAMP is known to activate both PKA and Epacs/cAMP-GEFs in -cells, leading to the activation of Rap1, a small GTPase involved in cell proliferation, differentiation, and morphogenesis (12, 29). In this connection, we previously showed that Golf is expressed in human pancreatic islets in 23-wk-old fetuses (15). Therefore, Golf may be involved in the signals governing endocrine pancreatic development, -cell growth, differentiation, and survival. The coordination of these cellular activities governs the proper dynamic organization of islet morphogenesis and the functional restoration of the -cell mass in diabetes.

Pancreatic -cells are responsible for maintaining the body's nutrient levels within a very narrow range of variations, to compensate and maintain normoglycemia. During postnatal development and in adults, low levels of -cell replication and apoptosis are offset by a slowly increasing -cell mass. In type 2 diabetes, glucotoxicity and lipotoxicity due to a sedentary life and overnutrition contribute to the functional impairment of the endocrine pancreas. Recent findings suggest that cell death could explain insulin deficiency through a reduction in the absolute number of pancreatic -cells as a result of different insults induced by glucose, fatty acid, and nitric oxide production (31, 32). Here, we demonstrated that AGolf provides remarkable protection against apoptosis in insulin-secreting TC-3 cells cultured with or without serum. This demonstration is supported by the results of recent studies indicating that cAMP reduces the apoptosis induced by cycloheximide in baby hamster kidney fibroblasts (45). Similarly, activated PKA was found to mediate neuronal survival through the phosphorylation and inactivation of the glycogen synthase kinase 3- (30).

On the basis of the present findings, molecular alterations and the deregulated overexpression of Golf can be considered as mechanisms potentially involved in the -cell dysfunctions found in diabetic patients. In this connection, two intronic polymorphisms have been identified in the human Golf gene that might cause aberrant splicing of Golf mRNA (6). In conclusion, the present results demonstrated that 1) Golf directly and preferentially activates AC-I and -VIII, two cAMP-generating enzymes expressed in -cells and deregulated in diabetic animals, and that 2) the Golf subunit is involved in regulating insulin status and survival in pancreatic -cells.

Perspectives