Evidence that the pertussis toxin-sensitive trimeric GTP-binding protein Gi2 is required for agonist- and store-activated Ca2+ inflow in hepatocytes.

The role of a trimeric GTP-binding protein (G-protein) in the mechanism of vasopressin-dependent Ca2+ inflow in hepatocytes was investigated using both antibodies against the carboxyl termini of trimeric G-protein α subunits, and carboxyl-terminal α-subunit synthetic peptides. An anti-Gi1−2α antibody and a Gi2α peptide (Gi2α Ile345-Phe355), but not a Gi3α peptide (Gi3α Ile344-Phe354), inhibited vasopressin- and thapsigargin-stimulated Ca2+ inflow, had no effect on vasopressin-stimulated release of Ca2+ from intracellular stores, and caused partial inhibition of thapsigargin-stimulated release of Ca2+. An anti-Gqα antibody also inhibited vasopressin-stimulated Ca2+ inflow and partially inhibited vasopressin-induced release of Ca2+ from intracellular stores. Immunofluorescence measurements showed that Gi2α is distributed throughout much of the interior of the hepatocyte as well as at the periphery of the cell. By contrast, Gq/11α was found principally at the cell periphery. It is concluded that the trimeric G-protein, Gi2, is required for store-activated Ca2+ inflow in hepatocytes and acts between the release of Ca2+ from the endoplasmic reticulum (presumably adjacent to the plasma membrane) and the receptor-activated Ca2+ channel protein(s) in the plasma membrane.

Receptor-activated calcium channels (RACCs) 1 are present in most non-excitable and in some excitable animal cells and are responsible for allowing the inflow of Ca 2ϩ to specific regions of the cytoplasmic space and the refilling of intracellular Ca 2ϩ stores (most likely a region of the endoplasmic reticulum) (1)(2)(3). For a number of cell types it has been shown that ago-nist-receptor complexes open at least two types of RACCs differing in selectivity for divalent cations (3). The mechanism(s) by which RACCs are opened is poorly understood (1)(2)(3). The hypothesis presently favored is the store-operated (capacitative) mechanism in which an increase in inositol 1,4,5-trisphosphate (InsP 3 ) and the release of Ca 2ϩ from a region of the InsP 3 -sensitive store are prerequisites for channel activation (1)(2)(3). This hypothesis is based, in part, on the observation that thapsigargin, which inhibits the endoplasmic reticulum (Ca 2ϩ ϩ Mg 2ϩ )-ATPase causing the release of Ca 2ϩ from this organelle, leads to a stimulation of Ca 2ϩ inflow (2). The results of a variety of experimental approaches have implicated the InsP 3 receptor (4), a mobile intracellular messenger (5-7), a monomeric G-protein (8,9), a trimeric G-protein (10 -14), protein phosphorylation (15), and/or elements of the cytoskeleton (16) in the mechanism that couples the release of Ca 2ϩ from the endoplasmic reticulum to activation of the plasma membrane Ca 2ϩ channels.
In hepatocytes, agonists that employ RACCs include vasopressin, adrenaline, angiotensin II, and epidermal growth factor (11)(12)(13). It has been shown previously that pretreatment of hepatocytes with pertussis toxin, or the microinjection of GDP␤S, inhibits store-operated, as well as vasopressin-stimulated, Ca 2ϩ inflow. These treatments did not affect the release Ca 2ϩ from intracellular stores. This suggested that in addition to G q/11 , a slowly ADP-ribosylated pertussis toxin-sensitive trimeric G-protein is required for activation of the hepatocyte RACC(s) (11)(12)(13). Since the two pertussis toxin-sensitive trimeric G-proteins present in hepatocytes at detectable levels are G i2 and G i3 (17,18), it was considered likely that one of these G-proteins is involved in activation of the hepatocyte RACC(s) (11,12). The aim of the present experiments was to identify the trimeric G-proteins involved in store-operated Ca 2ϩ inflow in hepatocytes. The approach employed used antibodies generated against peptides corresponding to a region of the carboxyl termini of G ␣ (19 -21) and synthetic peptides corresponding to specific regions of the carboxyl termini of G ␣ (22,23). These antibodies and peptides have been shown by others to inhibit the activation of trimeric G-proteins (19 -23). Thapsigarginstimulated Ca 2ϩ inflow is inhibited by an anti-G i1-2␣ antibody and by a G i2␣ peptide, indicating that the trimeric G-protein G i2␣ controls store-operated Ca 2ϩ inflow in hepatocytes. Immunofluorescence experiments indicate that the G i2␣ polypeptide is located in the interior of the hepatocyte and is not restricted in intracellular location to the plasma membrane.

EXPERIMENTAL PROCEDURES
Materials-A rabbit polyclonal anti-G i1-2␣ antibody, which recognizes G t␣ , G i1␣ , and G i2␣ , was raised against the peptide KENLKD-* This work was supported by a grant from the National Health and Medical Research Council of Australia. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 (Ile 345 -Phe 355 ), Ile-Lys-Asn-Asn-Leu-Lys-Asp-Cys-Gly-Leu-Phe; peptide G i3␣ (Ile 344 -Phe 354 ), Ile-Lys-Asn-Asn-Leu-Lys-Glu-Cys-Gly-Leu-Tyr; FITC, fluorescein isothiocyanate; GDP␤S, guanyl-5Ј-yl thiophosphate; GTP␥S, guanosine 5Ј-3-O-(thio)triphosphate. CGLF, a region of the carboxyl amino acid sequence for G t␣ , and a rabbit polyclonal anti-G q-11␣ antibody, which recognizes G q␣ and G 11␣ , was raised against the peptide QLNLKEYNLV, a common region of the carboxyl amino acid sequence for G q␣ and G 11␣ . The antibodies were prepared using peptides linked to keyhole limpet hemocyanin, affinitypurified as described previously (24), and routinely stored at Ϫ70°C in Tris-glycine buffer, pH 8.3. The abilities of the antibodies to recognize G i2␣ (anti-G i1-2␣ ) and G q/11␣ , respectively, in extracts of hepatocytes were confirmed by Western blot analysis. The anti-G i1-2␣ antibody detected a single band with an apparent molecular mass of 42 kDa and the anti-G q-11 antibody a single band with an apparent molecular mass of 43 kDa.
The peptides against which antibodies were raised, and the peptides IKNNLKDCGLF (G i2␣ Ile 345 -Phe 355 (peptide G i2␣ )) and IKNNLKECGLY (G i3␣ Ile 344 -Phe 354 (peptide G i3␣ )), corresponding to the carboxyl termini of G i2␣ and G i3␣ , respectively, were synthesized (as free COOH) by the Merrifield solid-phase synthesis procedure using an Applied Biosystems 430A synthesizer and were analyzed by quantitative amino acid analysis and mass spectrometry, as described previously (25). Solutions of peptides G i2␣ and G i3␣ were prepared fresh each day by dissolving the peptide in a solution of 10 mM fura-2 in 125 mM KCl to give a concentration of 12 mM peptide in the microinjection pipette tip (estimated intracellular concentration 160 M).
Fluorescein isothiocyanate (FITC)-conjugated rabbit IgG was obtained from Sigma-Aldrich, Castle Hill, New South Wales, Australia; and indocarbocyanine (Cy3) from Jackson, West Grove, PA. All other reagents were obtained from sources described previously (11).
Microinjection of Fluorescent Dyes, Antibodies, and Peptides to Hepatocytes and Measurement of Ca 2ϩ Inflow-The isolation of hepatocytes, attachment of hepatocytes to coverslips coated with collagen, the microinjection of fura-2, antibodies, and peptides, and measurement of the fluorescence of single hepatocytes loaded with fura-2 were conducted as described previously (11). The dilution factor for the microinjection of reagents to hepatocytes, determined previously (12), was approximately 75-fold. Antibodies were concentrated in a buffer composed of 27 mM K 2 HPO 4 , 8 mM Na 2 HPO 4 , and 26 mM KH 2 PO 4 (adjusted to pH 7.3 by addition of KOH) (phosphate buffer) (26) using a Centricon-10 concentrator (Amicon Inc., Beverly, MA) at 4,000 ϫ g for 30 min, stored at 4°C, and used within 1-8 days. The final concentration of antibody in the microinjection pipette tip ranged from 2.5 to 3.0 mg/ml (estimated intracellular concentration 30 -40 g/ml). Antibodies were co-injected with fura-2 (10 mM in the pipette tip, estimated intracellular concentration 130 M) at the concentrations indicated in the figure legends. In control experiments, the G i1-2␣ antibody (2.2-3.0 mg/ml in the pipette tip) was mixed with blocking peptide KENLKDCGLF (1.7 mM in pipette tip) and co-injected with fura-2. Peptides G i2␣ (Ile 345 -Phe 355 ) and G i3␣ (Ile 344 -Phe 354 ) were dissolved in phosphate buffer, mixed with fura-2 to give 12 mM peptide and 10 mM fura-2 in the micropipette tip, and introduced to the cytoplasmic space of hepatocytes by microinjection. The fluorescence of FITC-conjugated rabbit IgG injected into hepatocytes was measured using excitation and emission wavelengths of 490 and 540 nm, respectively.
Immunofluorescence-Freshly isolated hepatocytes (10 6 cells/50-mm plastic Petri dish) were cultured for 24 h in William's Medium E, fixed with 4% (w/v) paraformaldehyde, and permeabilized using Triton X-100 as described by Lièvremont et al. (27). After washing twice with 0.05% (w/v) Tween in PBS (Tween-PBS), the cells were incubated overnight with a given affinity-purified rabbit anti-G ␣ antibody (50 g/ml) or with a given anti-G ␣ antibody which had been mixed with a molar excess of appropriate blocking peptide. The cells were washed five to six times with Tween-PBS to remove primary antibody and then incubated with donkey anti-rabbit IgG antibody conjugated to indocarbocyanine (Cy3) (1 in 200). Immunofluorescence was detected using a Nikon inverted microscope, a ϫ 60 or ϫ 100 oil immersion objective lens and a Bio-Rad MRC 1000 scanning confocal imaging system incorporating a kryptonargon laser.

RESULTS
Hepatocytes loaded with anti-G i1-2␣ antibody exhibited a substantial inhibition of vasopressin-stimulated Ca 2ϩ inflow compared with control hepatocytes (Fig. 1a, solid line, cf. Fig.  1b). As reported previously (8,11,12,19,26), there was some heterogeneity in the responses given by individual hepatocytes. Details of the total number of cells tested and the numbers of cells yielding a given type of response are set out in Table I. In the majority of cells tested the ability of vasopressin to release Ca 2ϩ from intracellular stores was not affected by microinjec-tion of the antibody ( Table I). Pretreatment of the anti-G i1-2␣ antibody with the peptide against which the antibody was raised (the blocking peptide) prevented the inhibition of vasopressin-stimulated Ca 2ϩ inflow (Fig. 1a, broken line; Table I).
The anti-G i1-2␣ antibody also inhibited thapsigargin-stimulated Ca 2ϩ inflow and caused some inhibition of thapsigarginstimulated release of Ca 2ϩ from intracellular stores (Fig. 1c, Table I). No inhibition of vasopressin-stimulated Ca 2ϩ inflow was observed in cells loaded with an anti-G i2␣ antibody raised against a peptide which corresponds to a region near the amino terminus of G i2␣ (results not shown).
The effects of the anti-G i1-2␣ antibody were compared with those of an antibody against G q/11␣ . This inhibited vasopressinstimulated Ca 2ϩ inflow in most cells tested and caused partial inhibition of the vasopressin-stimulated release of Ca 2ϩ from intracellular stores (Fig. 1d, Table I), as shown previously for hepatocytes by Yang et al. (19). The possibility that the microinjected anti-G q-11␣ antibody was incompletely distributed in the cytoplasmic space was tested by microinjecting FITC-conjugated rabbit IgG to hepatocytes. The fluorescence signal diffused evenly in recipient cells within 5 min following the microinjection of the antibody, indicating that antibody microinjected to an hepatocyte is distributed throughout the cell within 5 min following its microinjection (results not shown).
The microinjection of peptide G i2␣ (Ile 345 -Phe 355 ) inhibited vasopressin-stimulated Ca 2ϩ inflow in almost all cells tested but had no effect on the ability of vasopressin to release Ca 2ϩ from intracellular stores (Fig. 2a, Table I). By contrast, microinjection of peptide G i3␣ (Ile 344 -Phe 354 ), at the same concentration as that employed for peptide G i2␣ , caused no inhibition of either Ca 2ϩ inflow or Ca 2ϩ release from intracellular stores induced by vasopressin (Fig. 2a, Table I). Neither peptide caused an activation of Ca 2ϩ inflow in the absence of vasopressin or thapsigargin (results not shown). Thapsigargin-stimulated Ca 2ϩ inflow was also inhibited by peptide G i2␣ . Complete inhibition was observed in 55% of the cells tested (Fig. 2b, cf.  Fig. 2c, Table I). Peptide G i3␣ caused little or no inhibition of thapsigargin-stimulated Ca 2ϩ inflow (Fig. 2b, cf. Fig. 2c, Table  I). However, both peptides caused some inhibition of thapsigargin-induced release of Ca 2ϩ from intracellular stores (Fig. 2b,  cf. Fig. 2c, Table I). In 7 out of 11 cells (peptide G i2␣ ) and in 3 out of the 4 cells (peptide G i3␣ ) the inhibition of the thapsigargin-induced release of Ca 2ϩ from intracellular stores by the peptide was associated with an inhibition of thapsigargin-induced Ca 2ϩ inflow. This suggests there may be some correlation between the effects of the peptides on the release of Ca 2ϩ from intracellular stores and their effects on Ca 2ϩ inflow.
Since the results of the experiments conducted with anti-G i2␣ antibodies and site-specific G i␣ peptides suggest that G i2␣ is required for the activation of Ca 2ϩ inflow, the intracellular location of G i2␣ was investigated by immunofluorescence, using anti-G i1-2␣ antibody as the primary antibody and anti-rabbit IgG antibody coupled to the fluorescent dye Cy3. In most cells, immunofluorescence, which was dependent on anti-G i1-2␣ antibody, was found to be distributed throughout the cytoplasmic space as well as in most parts of the cell periphery (Fig. 3a). This distribution is seen more clearly at higher magnification (Fig. 3b). The fluorescence signal given by anti-G i1-2␣ antibody was not observed when the anti-G i1-2␣ antibody was omitted or when this antibody was pretreated with the blocking peptide (results not shown). A pattern of immunofluorescence similar to that given by anti-G i1-2␣ antibody was observed when an anti-G i3␣ antibody was employed as the primary antibody (results not shown). In contrast to the results obtained with the anti-G i1-2␣ antibody, when anti-G q-11␣ antibody was employed, the fluorescence signal was largely confined to the periphery of the cell, adjacent to the plasma membrane (Figs. 3, c and d). The fluorescence given by anti-G q-11␣ antibody was abolished when anti-G q-11␣ antibody was omitted or when this antibody was pretreated with blocking peptide (results not shown). The cells labeled with the anti-G q-11␣ antibody exhibited a much more defined location of immunofluorescence at the cell periphery than that exhibited by cells labeled with the anti-G i1-2␣ antibody (Fig. 3c, cf. Fig. 3a). These results indicate that while G i2␣ is distributed in the plasma membrane and in various    (Table I). c, inhibition by anti-G i1-2␣ antibody of thapsigargin-stimulated Ca 2ϩ inflow. The solid (anti-G i1-2␣ antibody present) and broken (no antibody) traces are representative of the results obtained for 1 of 8 cells (10 cells tested) and 1 of 8 cells (9 cells tested), respectively (Table I). d, inhibition by anti-G q-11␣ antibody of vasopressin-stimulated Ca 2ϩ inflow. The solid (anti-G q-11␣ antibody present) and broken (control, no antibody present) traces are represenative of the results obtained for 1 of 5 cells (7 cells tested), and for 1 of 3 cells (3 cells tested), respectively ( Table I). regions of the cytoplasmic space, G q-11␣ is located predominantly at the plasma membrane. DISCUSSION Previous studies with hepatocytes which utilized pertussis toxin, GTP␥S, and GDP␤S have shown that vasopressin-de-pendent Ca 2ϩ inflow requires a pertussis toxin-sensitive trimeric G-protein in addition to the pertussis toxin-insensitive G-protein G q/11 , which is required for the activation of phospholipase C␤ and the subsequent generation of InsP 3 (11)(12)(13). The conclusion that vasopressin-and thapsigargin-stimulated Ca 2ϩ inflow requires the action of G i1␣ and/or G i2␣ is consistent with the observation reported here that the anti-G i1-2␣ antibody, but not the anti-G i1-2␣ antibody treated with blocking peptide, inhibited vasopressin-and thapsigargin-stimulated Ca 2ϩ inflow. Moreover, the observation that, when present at the same concentration as peptide G i␣2 , peptide G i1␣3 did not inhibit vasopressin-and thapsigargin-stimulated Ca 2ϩ inflow indicates that the observed inhibitory effects of peptide G i␣2 are most likely due to inhibition of the action of G i2␣ . Taken in conjunction with the observations that the only detectable pertussis toxin-sensitive trimeric G-proteins in hepatocytes are G i2 and G i3 (16,17), the results reported here indicate that the pertussis toxin-sensitive trimeric G-protein required for activation of the hepatocyte plasma membrane receptor-activated Ca 2ϩ channel is G i2 .
The conclusion that G i2 is required for store-activated Ca 2ϩ inflow in hepatocytes is consistent with the observation that G i2␣ is distributed in regions of the cytoplasmic space as well as at the plasma membrane, in contrast to G q/11␣ which was found at the plasma membrane of the hepatocyte. ADP-ribosylation of G i2␣ catalyzed by pertussis toxin is very slow (11,19), consistent with a location of some G i2␣ in the cytoplasmic space.
The partial inhibition of the thapsigargin-induced release of Ca 2ϩ from intracellular stores by the anti-G i2 antibody (which would not be expected to bind to other intracellular proteins) and the G i2␣ and G i3␣ peptides was unexpected especially in view of the absence of an inhibition of vasopressin-stimulated Ca 2ϩ release. This preferential inhibition of thapsigargin-induced Ca 2ϩ release may reflect some form of steric interaction between G i2␣ and the thapsigargin-sensitive (Ca 2ϩ ϩ Mg 2ϩ )-ATPase proteins.
The observations that (a) the anti-G q-11␣ antibody completely inhibited vasopressin-stimulated Ca 2ϩ inflow (present results) and (b) the only known action of G q/11 is to activate phosphoinositide-specific phospholipase C␤ (28) provide further evidence which indicates that an increase in InsP 3 is a necessary prerequisite for vasopressin activation of Ca 2ϩ inflow in hepatocytes (cf. the conclusion reached previously on the basis of studies with GDP␤S and heparin (11)). The observation that the anti-G i2␣ antibody completely inhibits thapsigargin-stimulated Ca 2ϩ inflow as well as vasopressin-stimulated Ca 2ϩ inflow also provides further evidence that the process of storeoperated Ca 2ϩ inflow is a necessary part of the mechanism of activation of the plasma membrane Ca 2ϩ channel by vasopressin in hepatocytes (cf. the conclusion reached previously on the basis of results with pertussis toxin which was also shown to block both vasopressin-and thapsigargin-stimulated Ca 2ϩ inflow (12)(13)).
The failure of the anti-G q-11␣ antibody to completely inhibit vasopressin-induced release of Ca 2ϩ from intracellular stores may be due to a failure of the injected antibody to bind to all G q/11␣ molecules, possibly because the affinity of the anti-G q-11␣ antibody for G q/11␣ is low or G q/11␣ is in an environment that restricts the accesses of the antibody. Others have also reported that, relative to the effects of anti-G i␣ antibodies, longer incubation times are required in order to detect the inhibition by anti-G q-11␣ antibodies of phospholipase C ␤ in intact cells (19,20). Another possible explanation for incomplete inhibition by the anti-G q-11␣ antibody of vasopressin-induced release of Ca 2ϩ is that, in hepatocytes, there is a species of phosphoinositidespecific phospholipase C which can be activated by seven trans-  (Table I). b and c, effects of G i2␣ and G i3␣ peptides on thapsigarginstimulated Ca 2ϩ inflow. The solid (G i2␣ peptide present) and broken (G i3␣ peptide present) traces in b are representative of the results obtained for 1 of 11 cells (20 cells tested) and for 1 of 8 cells (12 cells tested), respectively (Table I). The trace shown in c (control, no peptide present) is representative of the results obtained for 1 of 11 cells (15 cells tested) (Table I). membrane-spanning receptors through a mechanism which does not involve G q/11 (28). However, no evidence for such a pathway in hepatocytes has so far been reported. Furthermore, G i2 is unlikely to be involved in the activation of phosphoinositide-specific phospholipase C␤ in hepatocytes, since first, this function of G i2 in hepatocytes has not been reported, and second, the anti-G i1-2␣ antibody and the G i2␣ peptide caused no inhibition of vasopressin-stimulated release of Ca 2ϩ from intracellular stores. It is noteworthy that in several cells the anti-G q-11␣ antibody inhibited vasopressin-stimulated Ca 2ϩ inflow with little effect on the release of Ca 2ϩ from intracellular stores. One possible explanation for this observation is that only a small region of the intracellular Ca 2ϩ stores (most likely the endoplasmic reticulum near the plasma membrane) is involved in activation of the plasma membrane Ca 2ϩ channels.
Based on the results obtained, the sequence of events emerging for the activation by vasopressin of RACCs in hepatocytes is likely to include the following steps: formation of the vasopressin-receptor complex, activation of G q/11 , activation of phospholipase C ␤ , an increase in InsP 3 at the periphery of the cell, release of Ca 2ϩ from the endoplasmic reticulum in this region, activation of G i2 , and activation of one or more RACCs. Since the anti-G i1-2␣ antibody and the G i2 peptide inhibit thapsigargin-stimulated Ca 2ϩ inflow (in the absence of added vasopressin and hence in the absence of the formation of the vasopressin-receptor complex) activation of G i2 would not involve its interaction with a seven-transmembranespanning receptor protein. One possible role of G i2 may be to regulate the movement of Ca 2ϩ between components of the endoplasmic reticulum (8,29,30) or interaction of the endoplasmic reticulum with the plasma membrane.
The proposed role of G i2 in store-activated Ca 2ϩ inflow in hepatocytes does not exclude a role for a low molecular weight G-protein, as proposed for mast and mouse lacrimal acinar cells, in part, on the basis of the observation that GTP␥S inhibits store-activated Ca 2ϩ inflow in these cell types (8,9). Indeed, other studies conducted in this laboratory have also shown that a relatively high concentration of GTP␥S inhibits thapsigargin-stimulated Ca 2ϩ inflow in hepatocytes. 2 Furthermore, the action of G i2 may be complimentary to that of a Ca 2ϩ influx factor (5-7).  3. Intracellular localization of G i2␣ and G q/11␣ in hepatocytes using immunofluorescence. Fluorescence images of cells treated with anti-G i1-2␣ antibody (a and b) and anti-G q-11␣ antibody (c and d). Freshly isolated hepatocytes were cultured for 24 h, fixed, permeabilized, treated with the indicated primary rabbit anti-G ␣ polyclonal antibody and a goat anti-rabbit IgG secondary antibody conjugated to Cy3, and the fluorescence viewed by scanning confocal fluorescence microscopy, using a ϫ 60 (a, c) and ϫ 100 (b, d) objective lens, as described under "Experimental Procedures." The images are from a scan 1 m in depth in a plane located approximately half way between the bottom and top of the cell. The scale bars represent 10 m. The results shown are those obtained for a representative sample of the cells observed in one of five experiments which gave similar results.