Activation of the luteinizing hormone/choriogonadotropin hormone receptor promotes ADP ribosylation factor 6 activation in porcine ovarian follicular membranes.

Previously we demonstrated in a cell-free ovarian follicular plasma membrane model that agonist-dependent desensitization of the luteinizing hormone/choriogonadotropin receptor (LH/CG R) is GTP-dependent, mimicked by the addition of ADP-ribosylation factor (ARF) nucleotide binding site opener, which acts as a guanine nucleotide exchange factor for ARFs 1 and 6, and selectively inhibited by synthetic N-terminal ARF6 peptides. We therefore sought direct evidence that activation of the LH/CG R promotes activation of ARF1 and/or ARF6. Using a classic ARF activation assay, the cholera toxin-catalyzed ADP-ribosylation of G alpha(s), results show that LH/CG R activation stimulates an ARF protein by a brefeldin A-independent mechanism. Synthetic N-terminal inhibitory ARF6 but not ARF1 peptide blocks LH/CG R-stimulated ARF activity. LH/CG R activation also promotes the binding of a photoaffinity GTP analog to a protein that migrates on one- and two-dimensional polyacrylamide gel electrophoresis with ARF6. These results suggest that ARF6 is the predominant ARF activated by the LH/CG R. To activate ARF6, the LH/CG R does not appear to signal through the C-terminal regions of G alpha(i) or G alpha(q) or through the second or third intracellular loops or the N terminus of the cytoplasmic tail of the LH/CG R. Although exogenous recombinant ARNO promotes only a small increase in ARF6 activation in the presence of activated LH/CG R, hCG-stimulated ARF6 activation is reduced to basal levels by catalytically inactive ARF nucleotide binding-site opener. These results provide direct evidence that LH/CG R activation leads to the activation of membrane-delimited ARF6.


Previously we demonstrated in a cell-free ovarian follicular plasma membrane model that agonist-dependent desensitization of the luteinizing hormone/choriogonadotropin receptor (LH/CG R) is GTP-dependent, mimicked by the addition of ADP-ribosylation factor (ARF)
nucleotide binding site opener, which acts as a guanine nucleotide exchange factor for ARFs 1 and 6, and selectively inhibited by synthetic N-terminal ARF6 peptides. We therefore sought direct evidence that activation of the LH/CG R promotes activation of ARF1 and/or ARF6. Using a classic ARF activation assay, the cholera toxincatalyzed ADP-ribosylation of G␣ s , results show that LH/CG R activation stimulates an ARF protein by a brefeldin A-independent mechanism. Synthetic Nterminal inhibitory ARF6 but not ARF1 peptide blocks LH/CG R-stimulated ARF activity. LH/CG R activation also promotes the binding of a photoaffinity GTP analog to a protein that migrates on one-and two-dimensional polyacrylamide gel electrophoresis with ARF6. These results suggest that ARF6 is the predominant ARF activated by the LH/CG R. To activate ARF6, the LH/CG R does not appear to signal through the C-terminal regions of G␣ i or G␣ q or through the second or third intracellular loops or the N terminus of the cytoplasmic tail of the LH/CG R. Although exogenous recombinant ARNO promotes only a small increase in ARF6 activation in the presence of activated LH/CG R, hCGstimulated ARF6 activation is reduced to basal levels by catalytically inactive ARF nucleotide binding-site opener. These results provide direct evidence that LH/CG R activation leads to the activation of membranedelimited ARF6.
We have recently shown in a cell-free plasma membrane model that binding of endogenous ␤arrestin1 (Arrestin 2) to the third intracellular (3i) 1 loop of the active luteinizing hormone/ choriogonadotropin (LH/CG) receptor promotes receptor desensitization by reducing the ability of the receptor to activate the stimulatory guanine nucleotide binding protein (G s ) and resulting adenylyl cyclase (AC) (1,2). The binding of ␤arrestin1 to the active LH/CG receptor is obligatory for LH/CG receptor desensitization, and there is sufficient ␤arrestin1 present in the membranes to promote ϳ80% LH/CG receptor desensitization (1)(2)(3). The pool of membrane-delimited ␤arrestin1 is made available to the activated LH/CG receptor by one or more steps that occur in response to LH/CG receptor activation and are dependent upon GTP (3). We therefore sought to elucidate the basis for the GTP dependence of ␤arrestin1-dependent LH/CG receptor desensitization. To this end, we have shown that LH/CG receptor desensitization appears to be independent of heterotrimeric G s , G i , and G q proteins, and of the Ras, Rap, and Rac families of small G proteins, based on the inability of C-terminal peptides or antisera directed toward the C termini of the G␣ proteins, sequestration of G␤␥ (4), or clostridial toxins (3) to disrupt LH/CG receptor desensitization. Rather, LH/CG receptor desensitization appears to be dependent on activation of the small G protein ADP-ribosylation factor 6 (ARF6) (3). This conclusion is based on results showing that both ␤arrestin1 release from its membrane docking site and subsequent LH/CG receptor desensitization are inhibited by preincubation of membranes with the inhibitory N-terminal ARF6 peptide but not with the analogous ARF1 peptide. As all G protein activation is dependent on a guanine nucleotide exchange factor (GEF), we sought to identify a GEF that might be involved in ␤arrestin1-dependent LH/CG receptor desensitization. LH/CG receptor desensitization is insensitive to brefeldin A (3), a fungal metabolite, which inhibits the guanine nucleotide exchange activity of most GEFs that activate ARFs 1-5, including Gea1p, Gea2p, GNOM, Sec7p, and BIG1 and 2, but not that of the GEFs comprising the ARNO/cytohesin-1/ GRP1, EFA6, or ARF-GEP 100 subfamilies, which activate ARFs 1 and 6 (5, 6). Moreover, ␤arrestin1 release from its membrane docking site and LH/CG receptor desensitization are stimulated by the addition of recombinant ARNO, a GEF for ARFs 1 and 6 (3), and blocked by a catalytically inactive recombinant ARNO (7). These results suggest that endogenous ARNO, or an ARNO-like GEF, activates ARF1 and/or ARF6 to promote ␤ar-restin1 release and consequent LH/CG receptor desensitization. It was therefore important to ascertain directly whether LH/CG receptor activation indeed promotes activation of an ARF and if the activated ARF corresponds to ARF1 and/or ARF6.
The classic method used to demonstrate ARF activation is the ability of cholera toxin (CTX) to catalyze the ADP-ribosylation of G s (8). ARF functions in the reaction as a cofactor by lowering the K m for both the ADP-ribose donor NAD and G␣ s (9), stimulating the reaction 50-fold (10). We have previously reported that preincubation of ovarian follicular membranes with hCG but not with BSA promotes CTX-catalyzed ADPribosylation of especially the long form but also the short form of G␣ s , both of which are immunoprecipitated with anti-G␣ s antisera (11,12). CTX-catalyzed ADP-ribosylation of G␣ s was dependent on the concentration of hCG and increased with time of incubation in the presence but not in the absence of hCG (11). These results are consistent with our hypothesis that LH/CG receptor activation promotes activation of an ARF in follicular membranes.
In the following studies we therefore sought to determine directly whether the agonist-activated LH/CG receptor promotes activation of ARF1 and/or ARF6 in follicular membranes. Using three different ARF activation assays, results show that ARF6 is the predominant ARF activated by the LH/CG receptor.

EXPERIMENTAL PROCEDURES
Purified hCG (CR-127) and FSH (oFSH-19) were kindly provided by Dr. A. F. Parlow of the National Hormone and Pituitary Program Harbor-UCLA Medical Center (Torrance, CA). Purified deglycosylated (dg) hCG was kindly provided by Dr. Patrick Roche, Mayo Clinic, Rochester MN. The ovarian follicular membrane fraction, which was purified by sucrose gradient centrifugation and enriched in AC activity, was prepared as previously described and stored at Ϫ70°C (13). Activation of CTX was done as previously described (11).
For the ARF activation assay, membranes (100 g of protein) were preincubated with indicated additions at 4°C for 30 or 60 min and then incubated at 30°C for 20 min, unless otherwise indicated, in a final volume of 100 l containing 1 mM ATP, 15 mM thymidine, 5 mM ADP-ribose, 20 mM L-arginine-HCl, 5 mM dithiothreitol, 25 mM Tris-HCl, pH 7.5, 20 M [ 32 P]NAD (1 Ci/mmol), 50 g/ml activated CTX, and 10 g/ml BSA or hCG (11). GTP was not added unless so specified. Membranes were then washed by adding 1 ml of 10 mM Tris-HCl, pH 7.5, and 1 mM EDTA, pelleted, and resuspended in SDS-STOP (2% SDS, 50 mM Tris-HCl, pH 8.75, 10% glycerol, 5% ␤-mercaptoethanol, 2 mM EDTA) (12,14). Proteins were separated by SDS-PAGE, and then gels were stained with Coomassie Blue, dried, and then exposed to Kodak X-Omat film (Eastman Kodak, Rochester, NY). The incorporation of 32 P into G␣ s was quantitated from scanned autoradiograms with the Molecular Analyst/PC Image Analysis software program. Alternatively, membranes (ϳ30 g of membrane protein) were preincubated 30 min at 4°C with or without indicated peptides, then subjected to a 5-min AC assay, as detailed in legend to Fig. 4. The reaction was then stopped and [ 32 P]cAMP was purified as previously described (15,16). Photoaffinity labeling with the GTP analog P 3 -(4-azidoanilido)-P 1 -PЈ-GTP ([ 32 P]AAGTP (17)) and separation of membrane proteins into Triton X-100 soluble and insoluble fractions was as previously described (12). Two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) was performed (18) with indicated ampholines. The pH gradient of the tube gel was determined by placing gel slices in 0.5 ml of H 2 0 overnight and then measuring the pH. Crude pellet and supernatant fractions of porcine ovarian follicles were obtained by homogenizing tissue in 10 mM Tris-HCl, pH 7.2, 1.0 mM EDTA, with a glass-glass Dounce homogenizer (ϳ10 strokes), followed by centrifugation at 1000 ϫ g for 5 min and then at 10,000 ϫ g for 30 min. The final pellet was resuspended in the volume of the original homogenate. SDS-STOP was then added to both the final pellet and supernatant fractions, and the samples were boiled for 10 min and stored at Ϫ70°C. SDS-PAGE and Western blotting were as previously described (11,12).
Anti-caveolin was obtained from Transduction Laboratories, Lexington, KY. The following antibodies were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA): anti-G␣ i (C-10), C-terminal peptide antibody to G␣ i3 , which reacts with all G␣ i proteins; anti-G␣ q /11 (C-19), C-terminal peptide antibody specific for G␣ q and G␣ 11 ; anti-G␣ s /olf (C-18), C-terminal peptide antibody to G␣ s . Preparation and specificity of anti-ARF antibodies was as previously described (19). LH/CG receptor antibody was made in male rabbits by the Fazleabas laboratory against a synthetic peptide, conjugated to keyhole lymphocyte hemagglutinin, corresponding to amino acids 257-271 of the extracellular domain of the human LH/CG receptor. This antibody reacts on SDS-PAGE (1:5000 dilution) with a band in porcine ovarian follicular membranes at ϳ88 kDa corresponding to the LH/CG receptor and with an unidentified second band of ϳ55 kDa.
All other chemicals and synthetic peptides were obtained from sources previously described (1,2,4,11). Results were analyzed using Student's t test (20). Final concentrations are indicated for all reactions.

RESULTS
Immunoreactive ARFs are detected in porcine ovarian follicular membranes-We first sought to determine which ARFs are present in our plasma membrane model. We have concentrated on ARFs 1 and 6, based on our evidence that both ␤arrestin1 release from its membrane docking site and LH/CG receptor desensitization are activated by ARNO, which activates ARFs 1 and 6 (21)(22)(23)(24), and inhibited by catalytically inactive E156K ARNO as well as the inhibitory N-terminal ARF6 but not ARF1 peptides (3). Both ARFs 1 and 6 are generally believed to be localized in an inactive, GDP-bound state in a cytosolic fraction or (for ARF6) in a subpopulation of endosomes and to translocate to Golgi membranes for ARF1 or to the plasma membrane for ARF6 upon binding GTP (5). However, there are also reports that ARF1, like ARF6, can be recruited to the plasma membrane (25,26). There is also evidence in some cells, including rat granulosa cells, that a sizable fraction of the total ARF6 is constitutively associated with the plasma membrane (3,19,27,28), even when cells are homogenized in a magnesium-containing buffer (3,29).
We first determined if immunoreactive ARFs 1 and 6 are detectable in purified porcine follicular membranes. Results (Fig. 1A, lane 1) show that the pan-ARF antibody 1D9, which reacts well with all ARFs but less so with ARF4 (19), detects a single band of ϳ21 kDa in purified porcine follicular membranes. Use of specific ARF6 and ARF1 antisera (19) shows that this 1D9-reactive band contains both ARFs 6 and 1 (Fig.  1A, lanes 2 and 3). ARFs 2-5 can also be detected in these purified membranes of ϳ21 kDa, using specific antisera (19) (not shown). Because these results do not allow conclusions regarding which of the ARFs predominate in purified follicular membranes, we performed 2D-PAGE (with pH 3-10 ampholines) of the purified membrane fraction and probed the resulting blot with both the pan-ARF antibody 1D9-and ARF1specific antibody. The ARF1-specific antibody detects a single dot of ϳ21 kDa with a pI of ϳ7.1 (not shown). Results with 1D9 ( Fig. 1B) indicate the presence not only of ARF6 with a pI of ϳ8.0 and ARF1 with a pI of ϳ7.1 (28,30) but also of the additional more acidic ARFs 2-5 (pI Ͻ 7.0) (31,32). Quantitation of the amount of ARF6 in this membrane fraction by Western blotting with the ARF6-specific antibody, using the signal generated by recombinant ARF6 as the standard, yielded a concentration of ϳ8 g of ARF6 protein/mg of membrane protein (Fig. 1C). The abundance of ARF6 in this ovarian tissue is consistent with the earlier report of very high ARF6 expression in the human ovary (19). Results in Fig. 1D show that ARF6 is detectable, using the ARF6-specific antibody, in the 10,000 ϫ g pellet and not in the supernatant fractions of porcine ovarian follicles in equivalent levels throughout antral follicular development. Taken together, these results show that, although ARFs 1 and 6 as well as other ARFs are detected in the membrane fraction, ARF6 is a relatively abundant protein and is present through antral follicular development.
hCG Stimulates the Binding of the GTP Photoaffinity Analog [ 32 P]AAGTP to ARF6 -We next sought to determine which plasma membrane-localized ARF(s) is activated as a consequence of LH/CG receptor activation. One technique to assess the activation of any G protein is its agonist-dependent binding of GTP. Release of GDP from G proteins is rate-limiting and stimulated by GEFs like the G protein-coupled receptors (GPCRs) for heterotrimeric G proteins or specific GEFs for the many small G proteins (33,34). Hormone-dependent binding of GTP or its photoaffinity analog [ 32 P]AAGTP provides a method to detect G protein activation. Initial experiments showed that hCG stimulates binding of the photoaffinity GTP analog [ 32 P]AAGTP to one or more proteins of ϳ21 kDa in porcine follicular membranes ( Fig. 2A). However, because the porcine follicular membrane fraction likely contains many small G proteins such as Ras (11) and the ARFs, we first sought to obtain a membrane fraction enriched in ARF6 to ascertain whether hCG promotes activation of ARF6. We determined that upon extraction of membrane proteins with 1% Triton X-100, ARF6 as well as the Triton-insoluble marker protein caveolin (35) remain in the Triton-insoluble fraction whereas G␣ s (12) and the LH/CG receptor are localized to a Tritonsoluble fraction (Fig. 2B). Neither ARF1 (Fig. 2C) nor ARFs 3, 4, or 5 (not shown) is detectable in the Triton-insoluble fraction. Consistent with this result, 2D-PAGE (with 70% pH 8 -10, 30% pH 3-10 ampholines) of the Triton-insoluble fraction followed by Western blotting with pan-ARF antibody 1D9 (Fig. 2D) reveals that the immunoreactive ARF proteins in this fraction are basic, with pI values of ϳ8.0 consistent with the pI of ARF6 (28,30) and not with that of ARF1. The more acidic ARFs 2-5 (31), which should migrate to the acidic end of the isoelectric focusing (IEF) gel (marked by the " ٚ "), were not detected. These results indicate that, among the ARFs, only ARF6 segregates into the Triton-insoluble membrane fraction. We next determined whether hCG promotes binding of the photoaffinity GTP analog [ 32 P]AAGTP to ARF6. Follicular membranes were incubated for 10 min with [ 32 P]AAGTP in the presence of BSA or hCG, UV-irradiated to covalently bind the GTP analog to protein, then extracted with 1% Triton X-100. Results (Fig. 2E) show that hCG increased binding of [ 32 P]AAGTP to a protein retained in the Triton-insoluble membrane fraction ϳ21 kDa, which corresponds to the molecular weight of ARF6. We next sought to determine whether the 21-kDa protein, which binds [ 32 P]AAGTP in the total membrane fractions, corresponds to ARF6. 2D-PAGE of the total membrane fraction incubated with hCG and [ 32 P]AAGTP shows that only a single protein of ϳ21 kDa with a pI of ϳ8.0 covalently binds [ 32 P]AAGTP (Fig. 2F, upper panel) and that this protein comigrates with immunoreactive ARF6 (lower panel). No labeling of the more acidic ARFs, which should migrate toward the acidic end of the IEF gel (marked by the "v"), was detected. These results therefore suggest that ARF6 is indeed activated to bind [ 32 P]AAGTP in response to LH/CG receptor activation.
hCG Stimulates ARF6 Activity-The ability of a receptor agonist to stimulate CTX-catalyzed ADP-ribosylation of G␣ proteins has been used as evidence that an agonist-activated receptor signals to one or more G proteins (36 -38). This technique is based on the observation that G␣ subunits serve as optimal substrates for CTX-catalyzed ADP-ribosylation when the guanine nucleotide-binding pocket of the G␣ subunit of the G protein heterotrimer becomes devoid of nucleotide through release of bound GDP (36,38). This obligatory GDP release normally occurs in response to agonist-dependent receptor activation. Alternatively, in the absence of receptor activation, GDP release from the G␣ GDP ␤␥ heterotrimer can be stimulated by addition of exogenous GTP, resulting in the generation of a Immunoreactive bands below 21 kDa on the ARF6-specific blot (lane 2) are believed to represent proteolytic breakdown products of ARF6. In B, proteins in follicular membranes (75 g of membrane protein) were separated by 2D-PAGE, as described under "Experimental Procedures." Ampholines consisted of 100% pH 3-10. An aliquot of total membrane fraction not subjected to IEF (Total) was loaded onto SDS-PAGE gel, as indicated. The symbol "ٙ" marks the edges of the tube IEF gel. Following SDS-PAGE, proteins were transferred to Immobilon membrane and blot was probed with pan-ARF antibody 1D9. In C, the amount of ARF6 protein in purified follicular membranes was evaluated by Western blotting with ARF6-specific antibody based on the signal generated by indicated amounts of recombinant (r) ARF6 protein. Signal was detected with 1 ng of rARF6 with longer exposure of the blot. In D, follicles measuring 1-2 mm in diameter (small, S), 3-5 mm (medium, M), and 6 -10 mm (large, L) were dissected from fresh porcine ovaries obtained from the slaughterhouse (12). Follicles were homogenized and separated by centrifugation at 10,000 ϫ g into pellet and supernatant fractions, as described under "Experimental Procedures." The pellets were resuspended in the same volume as the original homogenate, and proteins denatured. Equal protein concentrations (ϳ60 g) of supernatant fractions were loaded into gel wells; equal volumes of corresponding pellet fractions were loaded into gel wells. Following SDS-PAGE and transfer to Hybond membranes, the blot was probed with ARF6specific antisera (19). transient pool of G␣ Ϫ ␤␥ substrate for CTX-catalyzed ADPribosylation (38). For G␣ s , the ADP-ribose derived from NAD is covalently attached to arginine 201 in the long form and to arginine 187 in the short form of G␣ s (39). The ADP-ribosylation of G␣ s Ϫ ␤␥, followed by subsequent binding of GTP and resulting heterotrimer dissociation, promotes an inhibition of the G␣ s GTPase activity, resulting in elevated AC activities (40). However, G␣ subunits become very poor substrates for CTX-catalyzed ADP-ribosylation upon binding of GTP to G␣ s Ϫ ␤␥ and dissociation of G␣ s GTP from ␤␥ subunits and receptor (41).
The ability of CTX to stimulate the ADP-ribosylation of G␣ s is also the classic method to demonstrate ARF activation, because ARF in its active, GTP-bound form is an obligatory cofactor in this reaction (8). The first 13 amino acids of the ARFs are required to bind G␣ s (10), whereas CTX binds to a more C-terminal region (42). In previous studies designed to determine which G proteins were activated downstream of the LH/CG receptor, we reported that hCG stimulates the time-dependent ADP-ribosylation of both the long and short forms of G␣ s (11). Consistent with our earlier report, results in Fig. 3 show that the ADP-ribosylation of the short form of G␣ s is always stronger than that of the long form of G␣ s and, as will be seen in subsequent figures, is often less dependent on LH/CG receptor activation. The ADP-ribosylated G␣ sL also often resolves into a doublet (Ref. 11 and Fig. 3), the basis for which is not known. Because hCG-stimulated ADP-ribosylation of G␣ s constitutes direct evidence that hCG activates an ARF present in the follicular membrane preparation (8), in the following studies we determined whether ARF6 is required for hCG to stimulate the CTX-catalyzed ADP-ribosylation of G␣ s .
Synthetic peptides corresponding to the N terminus of the ARFs inhibit such ARF activities as intra-Golgi transport, the accumulation of coated vesicles and buds from Golgi preparations, phospholipase D (PLD) activation, and catecholamine secretion (30,(43)(44)(45)(46). We have shown that a pool of ␤arrestin1 is docked at the plasma membrane at a site distinct from the LH/CG receptor and that, upon LH/CG receptor activation, ␤arrestin1 binds to the LH/CG receptor resulting in an uncoupling of receptor and G s (1-3). ␤arrestin1 release from its membrane docking site can also be stimulated by 100 M GTP, and this GTP-dependent ␤arrestin1 release can be inhibited by Results for each antibody are from separate experiments, and each are representative of three experiments. In C, follicular membranes (150 g) were extracted with 1% Triton X-100 as in B, and 100% of the Triton-soluble and -insoluble fractions was loaded onto the gel for SDS-PAGE. The blot was probed with ARF1-specific antibody. In D, proteins in the Triton X-100-insoluble fractions were separated by 2D-PAGE, as described under "Experimental Procedures." Ampholines consisted of 70% pH 8 -10, and 30% pH 3-10. An aliquot of total membrane fraction not subjected to IEF was loaded onto SDS-PAGE gel, as indicated. The symbol " ٚ " marks the edges of the IEF tube gel. Following SDS-PAGE, proteins were transferred to Immobilon membrane and blot probed with pan-ARF antibody 1D9. In E, membranes (80 g of membrane protein) were incubated in an incubation medium consisting of 1 mM ATP, 5 mM MgCl 2 , 0.4 mM EDTA, 1 mM EGTA, 10 M GDP, and 25 mM BTP, pH 7.2, 25 mM creatine phosphate, and 0.2 mg/ml creatine phosphokinase, in the presence of 0.5 M [ 32 P]AAGTP and 10 g/ml BSA or hCG at 30°C for 10 min (12); membranes were washed, resuspended in incubation medium and UV-irradiated 3 min at 4C (12); membrane proteins were then solubilized in 1% Triton X-100 (12). Triton-insoluble proteins were heat-denatured and subjected to SDS-PAGE. Results are representative of two separate experiments. In F, membranes were incubated as in E but only in the presence of hCG, and following UV irradiation total membrane fraction was pelleted, heatdenatured, and subjected to 2D-PAGE with 70% pH 8 -10 and 30% pH 3-10 ampholines; proteins transferred to Immobilon membranes were then subjected first to autoradiography (upper panel) then to Western blotting using pan-ARF antibody 1D9 (lower panel). Results are representative of two experiments.

FIG. 2. ARF6 fractionates into the Triton X-100 insoluble membrane fraction and binds [ 32 P]AAGTP in response to LH/CG receptor activation.
In A, follicular membranes (30 g of membrane protein) were incubated in the presence of 1 mM AMP-PNP, 3 mM MgCl 2 , 0.5 mM EDTA, 1 mM EGTA, and 25 mM bis-Tris propane (BTP), pH 7.2, 10 nM [ 32 P]AAGTP, and 10 g/ml BSA or hCG at 30°C for 10 min; membranes were washed, resuspended in incubation medium without GTP or hormone, and UV-irradiated 3 min at 4°C (12). Corresponding Coomassie blue-stained membrane proteins are shown below the autoradiogram. Results are representative of three experiments. In B, follicular membranes (300 g of membrane protein) were incubated with 10 g/ml BSA or hCG in an incubation medium (IM) consisting of 1 mM ATP, 5 mM MgCl 2 , 0.4 mM EDTA, 1 mM EGTA, 10 M GTP, and 25 mM BTP, pH 7.2, for 40 min at 30°C. Membrane proteins were then pelleted, resuspended in buffer containing 1% Triton X-100, and stirred at 4°C for 1 h (12), then separated into a Triton-soluble and -insoluble fractions by centrifugation at 105,000 ϫ g for 60 min. Pellet and supernatant fractions were then heat-denatured. Blots were probed with ARF6-specific, caveolin, and LH/CG receptor antibodies, as indicated. 100% of the Triton-insoluble and 33% of the Triton-soluble fraction was loaded onto the gel for SDS-PAGE for ARF6 and caveolin blots (12); 100% of both fractions was loaded for the LH/CG receptor blot shown. the addition of inhibitory synthetic myristoylated (Myr)-and non-Myr-(2-13)ARF6 peptides (3). We therefore sought to determine whether the ARF6 N-terminal peptide inhibited ARFdependent CTX-stimulated AC activity. When membranes were incubated with 100 M GTP, CTX raised AC activities over levels seen without CTX (H 2 O), as a consequence of the ARF-dependent ADP-ribosylation of G␣ s and resulting inhibition of the GTPase activity of G␣ s (Fig. 4). Addition of Myr-ARF6 N-terminal peptide promoted a concentration-dependent reduction of CTX-stimulated AC activities whereas Myr-ARF1 N-terminal peptides were ineffective (Fig. 4).
We also previously showed that, by preventing ␤arrestin1 release from its membrane docking site, ARF6 N-terminal peptides inhibit LH/CG receptor desensitization (3). We therefore sought to determine whether the ARF6 N-terminal peptide also inhibited hCG-stimulated CTX-catalyzed ADP-ribosylation of G␣ s . Myr-ARF6 N-terminal peptide reduced hCG-stimulated CTX-catalyzed ADP-ribosylation of G␣ s in a concentration-dependent manner whereas the corresponding Myr-ARF1 peptide was ineffective (Fig. 5). Taken together, the results of both ARF activation assays suggest that ARF6 and not ARF1 is the predominant cofactor in the ADP-ribosylation of G␣ s in porcine follicular membranes and that LH/CG receptor activation leads to the activation primarily of ARF6.

LH/CG Receptor Stimulation of the ADP-ribosylation of G␣ s : Potential Role for ARNO, a Guanine Nucleotide Exchange Factor for ARF6 -Brefeldin
A fails to inhibit the GTP exchange activity of ARNO, cytohesin-1, GRP-1, ARF-GEP 100 , and EFA6 GEFs for ARF1 and/or ARF6 but does inhibit the guanine nucleotide exchange activity of most other GEFs that activate ARFs 1-5 (5, 6). We therefore determined whether hCG-stimulated ARF6 activation was inhibited by brefeldin A. Results in Fig. 6 show that neither the basal nor hCG-stimulated CTXcatalyzed ADP-ribosylation of the short or long forms of G␣ s is inhibited by brefeldin A. Thus, the LH/CG receptor activates ARF6 through a brefeldin A-insensitive ARF GEF present in follicular membranes to trigger CTX-catalyzed ADP-ribosylation of G␣ s .
Because our follicular membrane model contains relatively high levels of endogenous ARNO (ϳ1.5 g/mg of membrane protein (7)), we sought to determine whether catalytically inactive ARNO, containing a mutation at E156, could function in a dominant negative manner to inhibit the ability of the LH/CG receptor to promote the ADP-ribosylation of G␣ s . As shown in Fig. 7, E156K ARNO significantly reduced the ability of hCG to activate ARF6 to promote the ADP-ribosylation of G␣ s . These results suggest that the guanine nucleotide exchange activity of endogenous ARNO or an ARNO-like GEF is obligatory for the LH/CG receptor to activate ARF6.
We next sought to determine whether addition of exogenous ARNO to the already high levels of ARNO present in follicular membranes further enhances the ADP-ribosylation of G␣ s . Addition of exogenous recombinant ARNO did not increase the CTX-catalyzed ADP-ribosylation of G␣ s in the absence of receptor agonist (Fig. 8A, lanes 1, 3, 5, and 7; Fig. 8B, lanes 1 and  5). However, in the presence of hCG, exogenous recombinant ARNO promoted a small increase in ARF activation (Fig. 8A,  lane 4 versus 2). Addition of 4 mM MgCl 2 to the reaction mix was found to diminish the ability of hCG to stimulate the ADP-ribosylation of G␣ s (Fig. 8A, lanes 5 and 6). The addition of exogenous recombinant ARNO to a reaction mix containing 4 mM MgCl 2 restored hCG-stimulated CTX-catalyzed ADP-ribosylation of both the long and short forms of G␣ s (Fig. 8A, lanes 5 and 6 versus 7 and 8; note reduced protein load in lanes 7 and 8). Activation of the FSH receptor also stimulated the ADPribosylation of G␣ s (Fig. 8B, lanes 1 and 3), and this response is also slightly enhanced by the addition of exogenous ARNO (lanes 3 and 4). Aluminum fluoride (AlF Ϫ ) promoted only a very modest activation of the ADP-ribosylation of G␣ s (Fig. 8C), and this response was unaffected by the addition of exogenous ARNO. The inability of AlF Ϫ to promote robust CTX-catalyzed ADP-ribosylation of G␣ s is consistent with an earlier report (47) and likely results from dissociation of G␣ s GDPALF Ϫ from ␤␥ subunits based on the ability of AlF Ϫ to mimic the terminal phosphate on GTP to promote dissociation of GDP-bound heterotrimeric G proteins (33).
These results show that in the presence of already high levels of endogenous ARNO, exogenous recombinant ARNO promotes only a small further increase in hCG-stimulated CTX-catalyzed ADP-ribosylation of G␣ s , but under experimental conditions where endogenous ARNO is either inactive or unavailable, then exogenous ARNO promotes a robust increase in hCG-stimulated CTX-catalyzed ADP-ribosylation of G␣ s probably by enhancing ARF6 activity. The inability of exogenous ARNO to promote CTX-catalyzed ADP-ribosylation of G␣ s in the absence of LH/CG receptor activation is most likely attributable to the absence of substrate, i.e. the presence of G s in the G␣ GDP ␤␥ conformation. Additionally, these results might suggest that the active conformation of the LH/CG receptor is required to activate ARNO or that receptor activation releases ARF6, potentially from the receptor, to be activated by available ARNO.

Effect of Heterotrimeric G Protein C-terminal Peptides and Antisera and of Synthetic Peptides Corresponding to Selected
Regions of the LH/CG Receptor on the Ability of LH/CG Receptor Activation to Activate ARF6 -Finally, we sought to determine whether the ability of the LH/CG receptor to activate the membrane-delimited ARF6 requires selected regions of the LH/CG receptor or the C-terminal regions of G␣ s , G␣ i , or G␣ q . Synthetic C-terminal G␣ peptides have been shown to compete specifically with G␣ proteins for binding to a receptor and therefore to inhibit downstream events (48 -52). G␣ C-terminal peptide-directed antisera have also been shown to inhibit receptor coupling to specific G␣ proteins (53)(54)(55)(56)(57)(58). We have previously shown that G␣ s -(354 -372) inhibits (ϳ50%) the ability of the LH/CG receptor to activate Gs and AC (4) based on the ability of this peptide to reduce G␣ s signaling to AC (50), and that LH/CG receptor 3i and 3iTM6 peptides selectively ablate LH/CG receptor desensitization based on their ability to compete with endogenous receptor for ␤arrestin1 (1). We therefore determined whether we could block the ability of the LH/CG receptor to activate ARF6 to stimulate the ADP-ribosylation of G␣ s by preincubating membranes with antibodies directed to the C-terminal domains of G␣ s , G␣ i , or G␣ q , or with synthetic peptides directed to the C termini of G␣ s , G␣ i , or G␣ q , or with synthetic peptides directed toward selected intracellular loops of the LH/CG receptor. As seen in Fig. 9, A-C, none of these reagents blocks the ability of hCG to activate ARF6 to stimulate the ADP-ribosylation of G␣ s . However, the LH/CG receptor antagonist dghCG does not stimulate the ADP-ribosylation of G␣ s (Fig. 9C, lane 3). These results suggest that the ability of the LH/CG receptor to activate ARF6 is dependent on LH/CG receptor activation but independent of the C-terminal regions of G␣ i and G␣ q and of the 2i and 3i loops and the N terminus of the cytoplasmic tail (4i) of the LH/CG receptor or that it involves regions on the effector that are not sensitive to these reagents. The availability of substrate for ADP-ribosylation (i.e. G␣ s Ϫ ␤␥) even in the presence of G␣ s C-terminal peptide, which reduces signaling to G s /AC by ϳ50% (4), likely shows that G s is not limiting in our membranes. Results (Fig. 9C) also show that ARF6 activation by the LH/CG receptor is not inhibited by wortmannin, a phosphatidylinositol 3-kinase (PI3K) inhibitor. DISCUSSION We have obtained direct evidence that agonist-dependent activation of the LH/CG receptor promotes the activation of a membrane-delimited ARF. The predominant ARF in follicular membranes that is activated upon engagement of the LH/CG receptor is ARF6, but we cannot rigorously exclude actions of other ARF isoforms. Under experimental conditions where receptor activation leads to the binding of a photoaffinity GTP analog to activated G proteins (12), we detect binding of GTP to a ϳ21-kDa protein, which exhibits a basic pI consistent with the pI of ARF6 and not with that of the other ARFs (31,32). hCGstimulated GTP binding to the 21-kDa protein is still detected when we segregate ARF6 from the other ARFs following hCGstimulated ARF activation by removing proteins soluble in Triton X-100. ARF6 is abundant in purified follicular membranes, present at a concentration of ϳ8 g/mg of membrane protein, and based on 2D-PAGE followed by Western blotting with the pan-ARF antibody, predominates over ARF1 in this membrane model. Moreover, ARF activation stimulated by hCG or high concentrations of GTP (100 M) is blocked by the inhibitory synthetic N-terminal Myr-ARF6 peptide and not by the corresponding Myr-ARF1 peptide. Our conclusion that agonist-dependent LH/CG receptor activation leads to activation of ARF6 is consistent with our earlier report (3), which showed that an inhibitory N-terminal ARF6 but not ARF1 peptide also prevents GTP-stimulated ␤arrestin1 release from its membrane docking site and LH/CG receptor desensitization. We have shown that ARF6 is localized to the Triton X-100insoluble fraction in membranes treated either with BSA or with hCG to promote receptor activation. In contrast to ARF6, the majority of G␣ s (12) and LH/CG receptors is restricted to the Triton X-100-soluble membrane fraction. The basis for the localization of ARF6 to the Triton X-100-insoluble fraction is not known and requires additional studies. However, based on evidence of an association between ARF6 and cortical actin (59 -62), the Triton-insolubility of ARF6 might reflect its association with actin or other cytoskeletal proteins. Our results also indicate that, as in many (19,27,28) but certainly not all cell models (27,29,59,(62)(63)(64), ARF6 in ovarian follicular cells appears to be constitutively associated with the plasma membrane both in its inactive and active conformation. Studies in rat ovarian granulosa cells also show that the majority of ARF6 is detected by Western blotting in the membrane/pellet fraction of cells and is not redistributed in response to LH/CG receptor activation (3), even in a magnesium-containing buffer that can dislodge ARF6 from the pellet fraction in other cells (29).
We designed a series of experiments to begin to determine how the LH/CG receptor promotes ARF6 activation. We first determined whether the active conformation of the LH/CG receptor signals through ARNO to activate ARF6. ARNO has been shown to catalyze the exchange of GDP for GTP on ARFs 1 and 6 (22,23). Based on our evidence that follicular membranes contain ϳ1.5 g of ARNO/mg of membrane protein (7), we determined whether the approximate tripling of the endogenous level of ARNO, by adding ϳ0.25 g of recombinant ARNO to 100 g of membrane protein, promoted a further increase in ARF6 activation, assessed as CTX-stimulated ADPribosylation of G␣ s . Our results showed that the addition of exogenous ARNO to follicular membranes in the presence of the active LH/CG receptor promoted only a minimal increase in ARF6 activation, consistent with the notion that levels of endogenous ARNO are sufficient to support ARF6 activation. Under experimental conditions where endogenous ARNO was FIG. 8. hCG-stimulated ARF activation is partially mimicked by ARNO. In A, membranes (100 g of membrane protein in 25 l) were preincubated 1 h at 4°C with 10 l water or recombinant ARNO at a final concentration of 50 nM and then subjected to ARF activation assay (Fig.  3), in the absence or presence of 4 mM MgCl 2 . In B and C, membranes were preincubated with water or 50 nM ARNO, as in A, then subjected to the ARF activation assay in the presence of 10 g/ml BSA, hCG, FSH, or 10 M AlF Ϫ (10 M AlCl 3 ϩ 10 mM NaF), as indicated. Results are representative of three separate experiments.
FIG. 9. hCG-stimulated CTX-catalyzed ADP-ribosylation of G␣ s is not inhibited by antibodies directed to the C terminus of G␣ s , G␣ i , or G␣ q ; by synthetic peptides corresponding to the C terminus of these G␣ proteins; or by synthetic peptides corresponding to selected regions of the LH/CG receptor. In A, membranes (100 g of membrane protein in 27 l) were preincubated 15 min at room temperature, then 1 h at 4°C with 10 l of indicated antibodies or with water. Membranes were then subjected to ARF activation assay as described in Fig. 3. In B, membranes (in 17 l) were preincubated 15 min at room temperature followed by 1 h at 4°C with 10 l of synthetic peptides directed to Cterminal regions of indicated G␣ proteins at a final concentration in the ARF activation assay of 100 M or with water and then subjected to ARF activation assay (Fig. 3). In C, membranes (in 17 l) were preincubated 1 h at 4°C with 10 l of synthetic peptides directed to indicated regions of the LH/CG receptor at a final concentration of 15 M in the ARF activation assay, with water, or with 10 l of wortmannin at a final concentration of 100 nM in the ARF activation assay and then subjected to ARF activation assay (Fig. 3), in the presence of BSA, hCG, or the LH/CG receptor antagonist deglycosylated (dg) hCG at 10 g/ml. Results for all panels are representative of at least two separate experiments. either inactive or unavailable and hCG did not activate ARF6 to stimulate CTX-catalyzed ADP-ribosylation of G␣ sL , addition of recombinant ARNO restored ARF6 activation in the presence of hCG to levels seen in the absence of added MgCl 2 . Our results showing that the addition of ϳ8-fold molar excess of recombinant catalytically inactive ARNO abolished hCG-stimulated ARF6 activation to stimulate CTX-catalyzed ADP-ribosylation of G␣ s support our conclusion that endogenous ARNO is obligatory for ARF6 activation. However, we cannot rule out the possibility that catalytically inactive ARNO is acting to sequester ARF6 and thus indirectly inhibiting ARF6 activation or that another brefeldin A-insensitive GEF promotes ARF6 activation in response to LH/CG receptor activation. Either ARNO or another ARNO-like GEF is activated in response to LH/CG receptor activation and consequently activates available ARF6, or ARF6 might be potentially bound to the inactive receptor and, upon engagement of the receptor, is freed to be acted upon by available ARNO, or aspects of both scenarios might apply, such that receptor activation leads to both ARF release from the receptor and ARNO activation. None of these alternatives can be excluded by the present results. However, our earlier result showing that addition of exogenous recombinant ARNO in the absence of LH/CG receptor activation promotes LH/CG receptor desensitization (7) suggests that LH/CG receptor activation increases the availability of ARNO rather than promotes activation of ARNO. In the absence of hCG, exogenous ARNO was ineffective in activating ARF6 to stimulate CTX-catalyzed ADP-ribosylation of G␣ s , likely because of a lack of substrate for ADP-ribosylation.
We also determined whether the activated LH/CG receptor directs ARF6 activation via its second or third intracellular loops or the N terminus of its cytoplasmic tail, or via the C-terminal regions of G␣ i or G␣ q . However, reagents established to test each of these regions of the LH/CG receptor and G␣ proteins yielded negative results. Although these results suggest that these regions of the G␣ proteins and LH/CG receptor do not participate in ARF6 activation, participation of other regions of the LH/CG receptor cannot be excluded and indeed are expected.
It is interesting that FSH also activates an ARF in porcine follicular membranes. We have linked ARF6 activation by the activated LH/CG receptor to LH/CG receptor desensitization (3). We do not yet know if this ARF6-dependent pathway, which promotes release of the ␤arrestin1 obligatory for LH/CG receptor desensitization, is unique to the LH/CG receptor or applies more universally to other G protein-coupled receptors (GPCRs). The ability of FSH to activate an ARF, however, is consistent with the possibility that GPCRs other than the LH/CG receptor can promote receptor desensitization via an ARF. The ␤-adrenergic receptor upon agonist but not antagonist binding has also been shown to stimulate CTX-catalyzed ADP-ribosylation of G␣ s (38), i.e. to activate an ARF. Moreover, there is recent evidence that overexpression of an ARF6 GTPase activating protein GIT1 inhibits ␤-adrenergic receptor internalization, consistent with the notion that the ␤-adrenergic receptor activates an ARF, such as ARF6 (65)(66)(67). ARF activation also occurs in response to activation of other GPCRs, including the m3 muscarinic acetylcholine (68), fMet-Leu-Phe (69), GnRH, H1 histamine, and B2 bradykinin (70) receptors and in response to activation of receptor tyrosine kinases like the insulin (25) and epidermal growth factor (71) receptors. At least in some of these cell models, receptor-stimulated ARF activation leads to activation of PLD 2 (70,72,73). ARF activa-tion in a number of these models involves the recruitment of the ARF and/or its GEF via phosphatidylinositol 3,4,5-trisphosphate produced by activation of PI3K and is thus inhibited by the PI3K inhibitor wortmannin (71,73,74). We have shown that wortmannin does not inhibit LH/CG receptor-stimulated ARF activation or LH/CG receptor desensitization in our plasma membrane model (7), consistent with our evidence that neither ARNO nor ARF6 needs to be recruited. Additional studies are needed to determine whether ARF activation in response to FSH and ␤-adrenergic receptor agonists leads to receptor desensitization, PLD activation, or activation of other downstream effectors.
In conclusion, we have shown that agonist-dependent LH/CG receptor activation promotes activation of membrane-delimited ARF. Our results point to ARF6 as the predominant ARF activated downstream of the LH/CG receptor and to an obligatory role for ARNO or a related GEF in ARF6 activation. Additional studies are required to elucidate how the active LH/CG receptor promotes the activation of ARF6 in ovarian follicular membranes.