Regulation of F-actin and Endoplasmic Reticulum Organization by the Trimeric G-protein Gi2 in Rat Hepatocytes

The roles of the heterotrimeric G-protein, Gi2, in regulating the actin cytoskeleton and the activation of store-operated Ca2+ channels in rat hepatocytes were investigated. Gαi2 was principally associated with the plasma membrane and microsomes. Both F-actin and Gαi2 were detected by Western blot analysis in a purified plasma membrane preparation, the supernatant and pellet obtained by treating the plasma membrane with Triton X-100, and after depolymerization and repolymerization of F-actin in the Triton X-100-insoluble pellet. Actin in the Triton X-100-soluble supernatant co-precipitated with Gαi2 using either anti-Gαi2 or anti-actin antibodies. The principally cortical location of F-actin in hepatocytes cultured for 0.5 h changed to a pericanalicular distribution over a further 3.5 h. Some Gαi2 co-localized with F-actin at the plasma membrane. Pretreatment with pertussis toxin ADP-ribosylated 70–80% of Gαi2 in the plasma membrane and microsomes, prevented the redistribution of F-actin, caused redistribution and fragmentation of the endoplasmic reticulum, and inhibited vasopressin-stimulated Ca2+ inflow. It is concluded that (i) a significant portion of hepatocyte Gαi2 associates with, and regulates the arrangement of, cortical F-actin and the endoplasmic reticulum and (ii) either or both of these regulatory roles are likely to be required for normal vasopressin activation of Ca2+ inflow.

In most nonexcitable and in some excitable cells, depletion of the inositol 1,4,5-trisphosphate (InsP 3 ) 1 -sensitive intracellular Ca 2ϩ stores in the endoplasmic reticulum (ER) activates a Ca 2ϩ influx pathway, a process known as store-operated Ca 2ϩ influx or capacitative Ca 2ϩ entry (1). Although it has been widely accepted that the key event initiating the opening of storeoperated Ca 2ϩ channels (SOCs) in the plasma membrane is the decrease in the concentration of Ca 2ϩ in the lumen of the ER, neither the mechanism that couples these two events nor the structures of SOCs are well understood (2). The results of recent experiments indicate that an essential prerequisite for the activation of SOCs is the close association between regions of the ER and the plasma membrane (3). It is proposed that this association is maintained by cytoskeletal elements such as the F-actin (4). There is evidence that, in some cell types, dismantling of the F-actin cytoskeleton (5), stabilization of the F-actin cytoskeleton (6), or inhibition of myosin light chain kinase (7) blocks Ca 2ϩ influx via SOCs while leaving Ca 2ϩ release from the intracellular stores unaffected (but see Ref. 8).
Hepatocytes are polarized epithelial cells in which the Factin cytoskeleton is distributed around the cortex, with a high concentration at the pericanalicular (apical) region (9). This cortical F-actin may play a role in maintaining subregions of the ER close to the plasma membrane (4). Evidence, including results obtained with a microinjected inhibitory anti-G␣ i2 antibody, indicates that the activation of SOCs in hepatocytes requires the trimeric G-protein G i2 (10) and a brefeldin A-sensitive protein, possibly a monomeric G-protein (11). It has been reported that some G␣ i2 co-localizes with F-actin in hepatocytes in primary culture (12). Moreover, studies with other cell types have provided evidence for an association between G␣ i2 and F-actin (13)(14)(15), and have suggested a potential role for G␣ i2 in organization of the F-actin cytoskeleton (16 -18). On the basis of these observations, we proposed that G i2 may regulate arrangement of the actin cytoskeleton and the arrangement of the ER by which both the intimate plasma membrane-ER association is achieved and the communication between different parts of the ER is maintained and allows the activation of SOCs.
The aims of the present experiments were to elucidate the role of G i2 in the activation of SOCs in hepatocytes by investigating the intracellular distribution of G␣ i2 and F-actin, the association of G␣ i2 with F-actin, and the requirement for G␣ i2 -F-actin interaction in regulation of the arrangement of F-actin and in the activation of SOCs. The results indicate that a significant proportion of the cellular G␣ i2 is associated with F-actin and regulates F-actin organization (especially the cortical actin layer near the canalicular membrane) and the arrangement of the ER. To our knowledge, this is the first demonstration of the role of G␣ i2 in regulating the arrangement of F-actin in an epithelial cell type. Taken together with previous evidence that the normal function of G␣ i2 is required for the activation of SOCs in rat hepatocytes (10), these observations suggest that G␣ i2 , either through regulation of cortical F-actin organization and/or arrangement of the ER, allows the normal activation of SOCs.
(John Curtin School of Medical Research, Australian National University, Canberra, Australia). Although this antibody detects both G␣ i1 and G␣ i2 , liver does not express detectable G␣ i1 (19,20), so that the Gprotein detected by this antibody in the present experiments is G␣ i2 . Peptides KENLKDCGLF and QLNLKEYNLV, synthesized as described in Ref. 10, were provided by Dr. Bruce Kemp (St. Vincent's Institute of Medical Research, Victoria, Australia). Purified phosphoprotein phosphatases 1 and 2A were kind gifts from Dr. Alistair Sim (University of Newcastle, Australia). Pertussis toxin, affinity-purified rabbit polyclonal anti-actin antibody, goat anti-rabbit IgG conjugated to alkaline phosphatase, actin standard for Western blotting, protein A-Sepharose, Triton X-100, nitro blue tetrazolium, and bromochloroindolyl phosphate were from Sigma, and Texas Red-X phalloidin, 3,3Јdihexyloxacarbocyanine iodide (DiOC 6 (3)), fura-2, and goat anti-rabbit IgG conjugated to Alexa TM 488 were from Molecular Probes, Inc. (Eugene, OR). Recombinant G␣ i2 protein was from Calbiochem (Alexandria, Australia). All other chemicals and materials were of the highest grade commercially available.
Western Blot Analysis of G␣ i2 and Actin-SDS-PAGE was performed on 12% polyacrylamide resolving gels with the Laemmli discontinuous buffer system (21), and the resolved proteins were electrotransferred to nitrocellulose membranes by the method of Towbin et al. (22). Membranes were blocked with 1 M glycine containing 5% (w/v) nonfat milk powder, 5% (v/v) fetal calf serum, and 1% (w/v) ovalbumin for 1 h at room temperature and then washed three times (5 min each) at room temperature with 0.1% (v/v) Tween 20, 0.1% (w/v) nonfat milk powder, and 0.1% (w/v) ovalbumin dissolved in 137 mM NaCl, 2.7 mM KCl, 8 mM Na 2 HPO 4 , and 1.4 mM KH 2 PO 4 (pH 7.2). Membranes were incubated overnight at 4°C with either anti-G␣ i antibody (1:200 dilution in the above wash buffer) or anti-actin antibody (1:100 dilution) or, in some cases, both antibodies together followed by incubation with secondary antibody (goat anti-rabbit IgG conjugated to alkaline phosphatase, 1:1000 dilution) for 2 h at room temperature and finally developed for 5 min in 100 mM Tris-HCl (pH 9.5), 100 mM NaCl, and 5 mM MgCl 2 containing 0.33 mg/ml nitro blue tetrazolium and 0.16 mg/ml bromochloroindolyl phosphate. Quantitation of the bands was performed on a Bio-Rad model GS-700 imaging densitometer driven by the Molecular Analyst software package (Bio-Rad). SDS-PAGE in the presence of 6 M urea was conducted as described by Komatsu et al. (23).
Subcellular Fractionation and Marker Enzyme Assays-Rat livers were homogenized in a medium containing 250 mM sucrose, 5 mM HEPES/KOH (pH 7.4), and 1 mM EGTA (homogenization medium), supplemented with 1 mM dithiothreitol, 0.2 mM phenylmethanesulfonyl fluoride, 10 g/ml leupeptin, and 10 g/ml pepstatin A (protease inhibitor mixture) and subcellular fractions prepared by differential centrifugation (24), with the 100,000 ϫ g supernatant being designated the "cytosolic fraction." A purified plasma membrane fraction and a nucleicontaminated plasma membrane fraction were prepared by Percoll gradient centrifugation (25). Protein concentrations were determined by the Bradford method (26) with bovine serum albumin as a standard. The activities of the marker enzymes 5Ј-nucleotidase (plasma membrane) and glucose-6-phosphatase (ER) were determined as described by Aronson and Touster (27).
Treatment of a Liver Cytosolic Fraction with Phosphoprotein Phosphatases-The liver cytosolic fraction (100 l) was diluted with an equivalent volume of homogenization medium supplemented with 1% (w/v) Triton X-100, 1 mM dithiothreitol, and the protease inhibitor mixture. Either 5 l (5 units; 1 unit of the enzyme is defined as the amount that hydrolyzes 1 nmol of phosphate from the phosphorylated proteins per min at 30°C, pH 7.0) of phosphoprotein phosphatase 1 or 5 l (5 units) of phosphoprotein phosphatase 2A or 5 l of vehicle (control) was added to 25 l of the above diluted cytosolic extract. The mixture was incubated at 37°C for 1 h, mixed with 30 l of Laemmli sample buffer, boiled, and subjected to SDS-PAGE and Western blotting analysis.
Triton X-100 Extraction of the Plasma Membrane Fraction to Yield a Triton X-100-insoluble Pellet and a Triton X-100-soluble Supernatant and Preparation of a Repolymerized F-actin Fraction from the Plasma Membrane Triton X-100-Insoluble Pellet-Plasma membrane pellets were resuspended in lysis buffer, which consisted of 50 mM HEPES (pH 7.4), 1% (w/v) Triton X-100, 150 mM NaCl, 1 mM EGTA, 1 mM Na 3 VO 4 , 100 mM NaF, 10 mM Na 4 P 2 O 7 , 10% (w/v) glycerol, supplemented with the protease inhibitor mixture, and incubated on ice for 1 h. A Triton X-100-insoluble pellet and a Triton X-100-soluble supernatant were obtained by centrifugation at 14,000 ϫ g for 10 min. F-actin present in the plasma membrane Triton X-100-insoluble pellet was subjected to two cycles of depolymerization and repolymerization as described by Ueda et al. (15), and the final fraction was called the "repolymerized F-actin fraction." All fractions were quantitatively mixed with Laemmli sample buffer for Western blot analysis.
Immunoprecipitation-The plasma membrane Triton X-100-soluble supernatant prepared as described above was incubated on ice for 2 h with either an anti-G␣ i or an anti-actin antibody or with normal rabbit serum (as control). Samples were mixed with swollen protein A-Sepharose (5 mg, dry weight), and the incubation continued for a further 1 h. Immune complexes bound to protein A-Sepharose were collected by centrifugation (12,000 ϫ g, 1 min). The pellets were washed three times in 0.2 M NaCl, 50 mM Tris-HCl (pH 7.4), resuspended in Laemmli sample buffer, boiled for 5 min, and centrifuged (12,000 ϫ g, 1 min), and the supernatant was retained for SDS-PAGE and Western blotting analysis.
Treatment of Rats with Pertussis Toxin and Isolation and Culture of Hepatocytes-Pertussis toxin (25 g in 50 mM Tris, pH 7.5, 10 mM glycine, 0.5 M NaCl, 50% (v/v) glycerol/100 g of body weight) or vehicle was administered to Hooded Wistar rats by intraperitoneal injection (28). After 24 h, hepatocytes were isolated by collagenase perfusion (29) and grown in primary culture on type I collagen-coated coverslips (30).
The Localization of the F-actin Cytoskeleton, G␣ i2 , and the Endoplasmic Reticulum-The locations of the F-actin and ER were determined using Texas Red-X phalloidin and DiOC 6 (3), respectively, and confocal microscopy as described previously (31). Negative controls for ER and F-actin staining were carried out systematically by omitting DiOC 6 (3) and Texas Red-X phalloidin, respectively. Determination of the location of G␣ i2 by immunofluorescence was performed as described previously (10). Controls were performed by omitting either the primary antibody or the secondary antibody or both and by incubating the primary antibody with excess blocking peptide before use.
For double labeling of F-actin and G␣ i2 in the same cell, F-actin staining was first performed as described above. The cells were then washed with phosphate-buffered saline containing 0.05% (v/v) Tween 20 and 1% (w/v) bovine serum albumin (Tween solution) and incubated overnight at 4°C with anti-G␣ i antibody (5 g/ml in Tween solution). Thereafter, cells were washed six times with the Tween solution, incubated with secondary antibody (Alexa TM 488-conjugated goat anti-rabbit IgG, 1:100 dilution in Tween solution), and washed twice with Tween solution and four times with phosphate-buffered saline before the coverslips were mounted on slides in 50% glycerol in phosphatebuffered saline.
Confocal microscopy was performed using a Bio-Rad MRC-1000 laser-scanning confocal microscope system in combination with a Nikon Diaphot 300 inverted microscope and a ϫ 40 NA 1.15 water immersion objective lens. The excitation and emission wavelengths were set at 568/10 and 605/35 nm, respectively, for Texas Red-X, and at 488/10 and 522/32 nm, respectively, for DiOC 6 (3) and Alexa TM 488. To standardize the fluorescence intensity measurements among experiments, the time of image capturing, the image intensity gain, the image enhancement, and the image black level were optimally adjusted at the outset and kept constant for each of Texas Red-X, DiOC 6 (3), and Alexa TM 488. In most cases, only images of the optical sections near the middle of the z axis were collected.
Quantitative examination of the captured images was performed using CoMOS (Bio-Rad) image analysis software. To quantitate F-actin distribution, for each experimental condition, 60 hepatocyte doublets were randomly selected from the images obtained from three separate cell preparations (20 doublets from each preparation), and the fluorescence (pixels) in the total doublet and in the pericanalicular area was measured. The fluorescence in the pericanalicular area was expressed as a percentage of the total doublet fluorescence. This percentage indicates the relative amount of F-actin around the bile canaliculus and hence the degree of reorganization of F-actin during primary culture (cf. Ref. 32). To avoid the subjectivity of this measurement, it was verified that the elliptical area designated as "pericanalicular area" occupied 9.95 Ϯ 0.06% (mean Ϯ S.E., n ϭ 60) of the total area of control doublets and 9.92 Ϯ 0.06% (mean Ϯ S.E., n ϭ 60) of the total area of pertussis toxin-treated doublets, respectively.
Electron Microscopy-Pellets (3000 ϫ g for 2 min) of the plasma membrane fraction (ϳ1 mg) were fixed in 1 ml of 1% (w/v) glutaraldehyde in 25 mM HEPES buffer (pH 7.4) for 30 min on ice. After washing three times with 25 mM HEPES buffer, the samples were postfixed with 1% (w/v) OsO 4 in the same HEPES buffer for 1 h on ice. Freshly isolated intact hepatocytes (pelleted by centrifugation at 80 ϫ g for 30 s) were fixed for 2 h at room temperature in a solution containing 1% (w/v) OsO 4 and 0.1 M Na 2 HPO 4 /NaH 2 PO 4 (pH 7.4). Fixed samples were dehydrated by stepwise exposure to increasing concentrations of ethanol (50, 75, 85, 95, and 100% (v/v)) and embedded in Durcupan with propylene oxide as an intermediate transition medium. The ultrathin sections were cut on an ultramicrotome, stained with aqueous uranyl acetate and Reynold's lead citrate, and examined with a JEOL 1200 EX transmission electron microscope.
Measurement of Ca 2ϩ Inflow-Cytoplasmic free Ca 2ϩ concentrations ([Ca 2ϩ ] cyt ) and initial rates of Ca 2ϩ inflow (measured using a Ca 2ϩ add-back protocol) in rat hepatocytes loaded with fura-2 by microinjection were determined using fluorescence microscopy (31).

Nature and Distribution of G␣ i2 in Rat Liver Subcellular
Fractions-When rat liver homogenates were subjected to Western blot analysis, two forms of G␣ i2 , with apparent molecular masses of 41 and 43 kDa, were detected (Fig. 1A). The plasma membrane fraction contained predominantly the 41-kDa band, which co-migrated with recombinant G␣ i2 (Fig. 1B,  lanes 1 and 2), while the cytosolic fraction contained predominantly the 43-kDa band (Fig. 1B, lanes 3 and 4). Treatment of cytosolic fraction with phosphoprotein phosphatase 1 converted the 43-kDa form of G␣ i2 to a form that co-migrates with recombinant G␣ i2 (Fig. 1C, lanes 1, 2, and 5). By contrast, treatment with phosphoprotein phosphatase 2A did not alter the mobility of the 43-kDa band (Fig. 1C, lanes 1, 3, and 6). These results indicate that (i) the 41-kDa form of G␣ i2 (subsequently referred to as G␣ i2 ) corresponds to the form of G␣ i2 (nonphosphorylated) normally detected in most cell types and (ii) the species of G␣ i2 with an apparent molecular mass of 43 kDa (subsequently referred to as phosphorylated G␣ i2 ) is a phosphorylated form of G␣ i2 (cf. Ref. 33).
G␣ i2 was found in the plasma membrane, the nuclearplasma membrane, and the heavy and light microsomal fractions of the liver ( Fig. 2A) but was barely detectable in the cytosolic fraction. The amount of G␣ i2 associated with the microsomes was estimated to be 40% of total cellular G␣ i2 . G␣ i2 (41-kDa) was the predominant form of G␣ i2 found in the plasma membrane and the nuclear plasma membrane fractions. Phosphorylated G␣ i2 was principally found in the cytosolic fraction, but some was also associated with the heavy and light microsomes ( Fig. 2A). In order to determine how tightly G␣ i2 is associated with the microsomal membranes, the microsomes were treated with KCl, which has been shown to cause the dissociation of loosely bound proteins from liver microsomal membranes (34). Phosphorylated G␣ i2 , but not the non-phosphorylated form, could be removed from microsomes by treat- ment with KCl (Fig. 2B). These results indicate that G␣ i2 is tightly associated with microsomal vesicles, whereas phosphorylated G␣ i2 is only loosely associated.
The distribution of the phosphorylated and nonphosphorylated forms of G␣ i2 within hepatocytes was further analyzed by determining the degrees of enrichment of the liver subcellular fractions in the two forms of G␣ i2 , 5Ј-nucleotidase (a plasma membrane marker enzyme) and glucose 6-phosphatase (an ER marker enzyme) (Fig. 3). The degree of enrichment of the purified plasma membrane fraction with G␣ i2 is similar to that for 5Ј-nucleotidase, indicating that, as shown previously (12), considerable G␣ i2 is located at the plasma membrane of hepatocytes. A small amount of glucose-6-phosphatase activity was found to be associated with the purified plasma membrane fraction. This may reflect either contamination of the plasma membrane fraction with microsomes derived from the ER or the attachment of small regions of the ER to the plasma membrane (cf. Ref. 24).
The degree of enrichment of the heavy and light microsomal fractions with G␣ i2 is similar to that for glucose-6-phosphatase (Fig. 3). Consideration of the degrees of enrichment of these two fractions with 5Ј-nucleotidase, together with the observation that the purified plasma membrane fraction is equally enriched in 5Ј-nucleotidase and G␣ i2 , indicates that the presence of G␣ i2 in the microsomal fractions is unlikely to be due to the contamination of these fractions by plasma membrane vesicles. The total amounts of phosphorylated G␣ i2 and G␣ i2 in the cytosolic fraction were estimated to be 84 Ϯ 5 and 13 Ϯ 3% (means Ϯ S.E., n ϭ 3 rat livers), respectively, of the total amount present in the homogenate.
Evidence for the Association of G␣ i2 and Actin in a Purified Rat Liver Plasma Membrane Fraction-It has previously been shown that a purified liver plasma membrane fraction (prepared in a manner similar to that described above) contains F-actin, which is attached to the plasma membrane (35). Experiments were undertaken to determine whether G␣ i2 is associated with this plasma membrane-associated actin. First, the quality of the plasma membrane fraction was further assessed by electron microscopy (Fig. 4). This showed numerous extended sheets of membrane (large arrow), the presence of small vesicles adherent to some sheets (small arrows), and numerous other vesicles of varying size. The preparation was largely free of mitochondria and nuclei.
The plasma membrane fraction was treated with 1% (w/v) Triton X-100 to solubilize membrane lipids and integral proteins and thereby to obtain, by centrifugation, a plasma membrane Triton X-100-insoluble pellet enriched in F-actin and other cytoskeletal components (15). G␣ i2 and actin were detected by Western blotting in both the Triton X-100-insoluble pellet (predominantly F-actin) and the Triton X-100-soluble supernatant (predominantly G-actin) (Fig. 5). It was estimated by densitometric analysis that approximately 27 Ϯ 3% (mean Ϯ S.E., n ϭ 4) of the total plasma membrane G␣ i2 and approximately 45 Ϯ 1% (mean Ϯ S.E., n ϭ 3) of the total plasma membrane actin were recovered in the plasma membrane Triton X-100-insoluble pellet.
To further test that G␣ i2 associates specifically with F-actin among the various cytoskeletal components of the plasma membrane, a repolymerized F-actin fraction was prepared from the plasma membrane Triton X-100-insoluble pellet by a twostep depolymerization-polymerization procedure (15). Analysis by SDS-PAGE and Western blotting with anti-G␣ i and antiactin antibodies demonstrated the presence of G␣ i2 in the repolymerized F-actin fraction (Fig. 5, lane 4). Approximately 44 Ϯ 0% of the G␣ i2 and 47 Ϯ 2% of the actin in the plasma membrane Triton X-100-insoluble pellet were recovered in the final repolymerized F-actin fraction. This corresponds to 12 Ϯ 0 and 21 Ϯ 1% (means Ϯ S.E., n ϭ 3) of the total plasma membrane G␣ i2 and actin, respectively.
The idea that G␣ i2 and actin associate near the plasma membrane was also investigated using a co-immunoprecipitation approach. When an anti-G␣ i antibody was used to precipitate G␣ i2 from the Triton X-100-soluble supernatant of the purified plasma membrane fraction, the precipitate was found to contain actin, identified using an anti-actin antibody and Western blot analysis (Fig. 6A). When an anti-actin antibody was used to precipitate actin from the Triton X-100-soluble supernatant of the purified plasma membrane fraction, the precipitate was found to contain G␣ i2 , identified using an anti-G␣ i antibody and Western blot analysis (Fig. 6B). When a similar co-immunoprecipitation experiment was performed with a liver cytosolic fraction (which is enriched in phosphorylated G␣ i2 ), no co-immunoprecipitation of phosphorylated G␣ i2 and actin was observed (data not shown).
Distribution of F-actin and G␣ i2 in Hepatocytes in Primary   FIG. 3. The relative distribution of 41-kDa G␣ i2 , 43-kDa G␣ i2 , and the plasma membrane (5-nucleotidase) and endoplasmic reticulum (glucose-6-phosphatase) markers in subcellular fractions of rat liver. The homogenization of rat liver; preparation of subcellular fractions; and determination of protein concentration, relative amounts of 41-kDa G␣ i2 and 43-kDa G␣ i2 (by Western blot analysis and densitometry), and marker enzyme activity were conducted as described under "Experimental Procedures." The degree of enrichment of a given fraction by G␣ i2 or marker enzyme was determined by dividing the amount of G␣ i2 (densitometry units) or marker enzyme (enzyme units) per mg of protein in the given subcellular fraction by the amount of G␣ i2 or marker enzyme per mg of protein in the total homogenate. The results are the means Ϯ S.E. of three separate experiments involving separate rat liver homogenates. FIG. 4. Electron micrograph of a purified liver plasma membrane fraction. The preparation of a plasma membrane fraction from rat liver, processing of the fraction for electron microscopy, and transmission electron microscopy were performed as described under "Experimental Procedures." Scale bar, 500 nm. The image shown is representative of 10 electron micrographs from two different membrane preparations.
Culture-The intracellular distribution of G␣ i2 and F-actin and the interaction between these proteins was further investigated using hepatocytes attached to collagen-coated coverslips, and Texas Red-X phalloidin and immunofluorescence to detect F-actin and G␣ i2 , respectively. In freshly isolated rat hepatocytes allowed to attach to coverslips for 0.5 h, F-actin was observed around the cortex, in both single hepatocytes and in hepatocyte doublets (Fig. 7A). When cultured for a further 3.5 h, the amount of F-actin in single cells and in doublets decreased in most regions of the cortex. In single cells, areas of high F-actin remained in some small regions of the cortex. In doublets, a pronounced concentration of F-actin at the canalicular membranes was observed (Fig. 7C). This most likely corresponds to the re-establishment of F-actin polarity and cell polarity, as described previously (32,36). Hepatocytes cultured for 4 h appeared to be more flattened and to have a larger diameter compared with cells cultured for 0.5 h (Fig. 7, compare C with A).
Substantial amounts of G␣ i2 (presumably both phosphorylated and nonphosphorylated forms) were found in the cytoplasmic space as well as at the plasma membrane of most hepatocytes examined, as shown previously (10, 12) (Fig. 7, E and G). In order to investigate the possible co-localization of G␣ i2 and F-actin, hepatocytes were double stained with Texas Red-X phalloidin and anti-G␣ i antibody (Fig. 8, A-C). The results indicate that there are regions of the cortex where the fluorescence signals representing G␣ i2 and F-actin overlap (indicated by the orange-yellow regions in Fig. 8C).

Effects of the Ablation of G␣ i2 Function by Pretreatment with Pertussis Toxin on the Intracellular Distribution of F-actin, G␣ i2 , and the Endoplasmic Reticulum and the Activation of Ca 2ϩ
Inflow-In order to further elucidate the role of G␣ i2 in regulation of the arrangement of the actin cytoskeleton and to study the roles of G␣ i2 and F-actin in the activation of SOCs, the treatment of rats with pertussis toxin was used to ablate G␣ i2 function. The effectiveness of pertussis toxin treatment was assessed by determining the degree of ADP-ribosylation of G␣ i2 , using SDS-PAGE in the presence of 6 M urea to identify ADP-ribosylated G␣ i2 (23). Pertussis toxin treatment caused ADP-ribosylation of G␣ i2 , as shown by the appearance of a new band in the urea/SDS-PAGE gel with a slower mobility than that of G␣ i2 (Fig. 9). Treatment with pertussis toxin did not result in any change in the mobility of the phosphorylated (43-kDa) G␣ i2 band (results not shown). The slower band (ADPribosylated G␣ i2 ) was observed in the plasma membrane fraction (Fig. 9A, lower panel, lane 2), the plasma membrane Triton X-100-insoluble pellet (lane 4), the plasma membrane Triton X-100-soluble supernatant (lane 6), and the heavy and light microsomal fractions (Fig. 9B). Quantitation of the bands using densitometry showed that pertussis toxin treatment resulted in ADP-ribosylation of 60, 80, and 50% of G␣ i2 in the total plasma membrane fraction, the plasma membrane Triton X-100-insoluble pellet, and the plasma membrane Triton X-100-soluble supernatant, respectively, and approximately 70% of G␣ i2 associated with the heavy plus the light microsomes.
Pertussis toxin pretreatment caused no detectable changes in the total amount of actin in the plasma membrane fraction (Fig. 9A, upper panel, compare lane 2 with lane 1). Further, since the Triton X-100-insoluble pellet contains predominantly F-actin and the Triton X-100-soluble supernatant contains mainly G-actin (6,15), the results also indicated that pertussis toxin treatment did not change the relative distribution of the two forms of actin in the plasma membrane fraction ( Cells from rats treated with pertussis toxin (pertussis toxintreated cells) that had been cultured for 0.5 h exhibited no substantial differences in the intracellular distribution of Factin compared with cells from vehicle-treated rats (control cells) cultured for this time (Fig. 7, compare B and A). However, the treatment with pertussis toxin prevented the redistribution of F-actin from the cortex to the bile canaliculus and other parts of the cell observed in control cells cultured for 4 h (Fig. 7, compare D and C). To quantitatively compare the differences in the distribution of F-actin in 4-h cultured doublets from control and pertussis toxin-treated rats, the pericanalicular fluorescence due to the F-actin-Texas Red-X phalloidin complex was expressed as a percentage of the total doublet fluorescence. This value was 18.87 Ϯ 0.70% (mean Ϯ S.E., n ϭ 60) in control doublets compared with 11.27 Ϯ 0.26% (mean Ϯ S.E., n ϭ 60) in pertussis toxin-treated doublets (p Ͻ 0.001, heteroscedastic t test). Pertussis toxin treatment also inhibited the spreading of cells observed at 4 h (Fig. 7, compare D and C). Thus, the total doublet area was 1153 Ϯ 49 m 2 (mean Ϯ S.E., n ϭ 60) in control doublets compared with 936 Ϯ 25 m 2 (mean Ϯ S.E., n ϭ 60) in pertussis toxin-treated doublets (p Ͻ 0.001, heteroscedastic t test).
Pertussis toxin-treated hepatocytes cultured for both 0.5 and 4 h exhibited noticeable differences in the distribution of G␣ i2 (Fig. 7, compare F and E; compare H and G). In contrast to control cells, where considerable G␣ i2 was present in the cytoplasmic space as well as at the plasma membrane, in pertussis toxin-treated cells, G␣ i2 was principally located at the plasma membrane and in the cortical region (Fig. 7, compare F and H  with E and G).
Pertussis toxin-treated hepatocytes exhibited more intense staining of the ER, monitored using DiOC 6 (3), than that observed in control cells (Figs. 10, compare B and D with A and  C). Moreover, the DiOC 6 (3) signal was more evenly distributed in pertussis toxin-treated cells. These differences were observed in cells cultured for both 0.5 and 4 h. Examination of the cells by electron microscopy revealed that pertussis toxintreated hepatocytes had largely lost the regular parallel arrangement of sheets of rough ER that were observed in control hepatocytes (Fig. 11, compare B and A). These differences can be seen more clearly at higher magnification (Fig. 11, compare D and C). Moreover, in pertussis toxin-treated cells the smooth ER appeared less dense than that in control hepatocytes (Fig. 11, compare B and A).
As shown previously, treatment with pertussis toxin inhibited vasopressin-stimulated Ca 2ϩ inflow (Fig. 12). There was no detectable effect of pertussis toxin treatment on vasopressin-induced release of Ca 2ϩ from intracellular stores (results not shown).

Role of G i2 in Regulating the Organization of F-actin and the
Endoplasmic Reticulum-In keeping with the observations of others (37), a 43-kDa phosphorylated form of G␣ i2 as well as the nonphosphorylated 41-kDa form were detected in hepatocytes. The present study has focused on G␣ i2 (the 41-kDa form), which is bound to the plasma membrane and ER (microsomes), rather than on the phosphorylated 43-kDa G␣ i2 , for the following reasons: (i) the phosphorylated G␣ i2 is hardly detectable in the plasma membrane fraction and is only loosely associated with the microsomes, (ii) there was no evidence from co-immunoprecipitation studies of an association between actin and phosphorylated G␣ i2 , and (iii) there was no evidence that the phosphorylated G␣ i2 was ADP-ribosylated by pertussis toxin treatment.
The following observations indicate that G␣ i2 (the 41-kDa form) associates with actin at the periphery of the hepatocyte: (i) the detection of both G␣ i2 and F-actin in a Triton X-100insoluble pellet prepared from a highly purified liver plasma membrane fraction; (ii) the detection of G␣ i2 in repolymerized actin obtained after F-actin in the plasma membrane Triton X-100-insoluble fraction was de-polymerized and re-polymerized; (iii) co-precipitation of G␣ i2 and actin from the plasma membrane Triton X-100-soluble fraction using either an anti-G␣ i antibody or an anti-actin antibody; and (iv) the observed co-localization of some G␣ i2 and F-actin at the cell periphery.
The results of experiments that employed pertussis toxin to ablate the action of G␣ i2 indicate that this trimeric G-protein is FIG. 6. Western blot analysis of anti-G␣ i and anti-actin immunoprecipitates from a Triton X-100-soluble supernatant prepared from a purified liver plasma membrane fraction. A plasma membrane fraction was treated with Triton X-100 and centrifuged to obtain a Triton X-100-soluble supernatant. A, co-immunoprecipitation of actin by an anti-G␣ i antibody. The Triton X-100-soluble supernatant was treated with anti-G␣ i antibody (lane 1) or normal rabbit serum as a control (lane 2), as described under "Experimental Procedures." Immunoprecipitates were resolved by SDS-PAGE, Western blotted, and probed first with an anti-G␣ i antibody and subsequently an anti-actin antibody. B, coimmunoprecipitation of 41-kDa G␣ i2 by an anti-actin antibody. The Triton X-100-soluble supernatant was treated with anti-actin antibody (lane 1) or normal rabbit serum as a control (lane 2) as described under "Experimental Procedures." Immunoprecipitates were resolved by SDS-PAGE, Western blotted, and probed with first anti-G␣ i antibody and subsequently with an anti-actin antibody. The upper band labeled IgG HC is immunoglobulin heavy chain. The results shown are those from one of two experiments, each of which gave similar results.
FIG. 7. The distribution of F-actin and G␣ i2 monitored using fluorescence microscopy, in hepatocytes derived from control and pertussis toxin-treated rats. Hepatocytes derived from vehicletreated rats (Control) and rats treated with pertussis toxin (PTX) were cultured for 0.5 or 4 h, and the locations of F-actin (using Texas Red-X phalloidin) or G␣ i2 (using immunofluorescence) were determined as described under "Experimental Procedures." Panels I and J are images obtained when the anti-G␣ i antibody was omitted from the procedure used to detect G␣ i2 . Images were obtained by confocal microscopy. The scale bars correspond to 20 m. The images shown are representative of more than 300 cells examined from three separate control and pertussis toxin-treated cell preparations.

FIG. 8. The localization of F-actin and G␣ i2 in hepatocytes.
Freshly isolated hepatocytes from untreated rats were cultured for 0.5 h, fixed, stained first for F-actin with Texas Red-X phalloidin (A), incubated with primary and secondary antibodies for the detection of G␣ i2 (B), and then examined by confocal microscopy, as described under "Experimental Procedures." C, images in A and B are superimposed, revealing regions of double labeling, indicated by orange-yellow color. Scale bars, 20 m. The images shown are representative of more than 100 cells examined from two separate cell preparations.
involved in regulating the organization of cortical F-actin in hepatocytes. Pertussis toxin specifically ADP-ribosylates and inactivates the ␣ subunit of G i1 , G i2 , G i3 , G o , and transducin (38). Since neither transducin, G o , nor G␣ i1 is expressed at detectable levels in hepatocytes (19,20), G␣ i2 and G␣ i3 are the only two known targets for pertussis toxin in these cells. Moreover, there is evidence that the time course for ADP-ribosylation of G␣ i3 by pertussis toxin treatment in vivo (72 h) is longer than that for G␣ i2 (24 -48 h) (23). Therefore, the in vivo pertussis toxin treatment employed in this study (24 h) is likely to result chiefly in inactivation of G␣ i2 . Moreover, urea/SDS-PAGE and Western blotting confirmed that the majority of the G␣ i2 on the plasma membrane, in particular the G␣ i2 associated with F-actin, was ADP-ribosylated and hence inactivated. It is clear from our results that this pertussis toxin treatment inhibited the redistribution of F-actin from the cortex to the bile canaliculus in hepatocyte doublets and the redistribution of F-actin to specific regions of the plasma membrane in single hepatocytes. Normally, cell polarity, which is lost during isolation of hepatocytes, can be restored within 3-4 h in monolayer culture (36). This re-establishment of cell polarity has been found to be closely associated with the redistribution of F-actin from the entire cortex to the canalicular pole (i.e. the polarization of F-actin) (32). The present results indicate that G i2 may be part of the machinery that governs the maintenance of a polarized distribution of F-actin in hepatocytes. The observation that pertussis toxin pretreatment prevented the spreading of hepatocytes in primary culture provides further evidence that G i2 regulates F-actin organization, since it has been shown that hepatocyte spreading in culture requires F-actin organization (39).
Studies with several other types of cells have also shown that G i2 interacts with F-actin (13)(14)(15) and is likely to play a role in regulating the organization of F-actin (16 -18). For example, the degree of actin polymerization in differentiating U937 cells was found to correlate well with an increase in the amount of G␣ i2 at the plasma membrane (16). In human airway smooth muscle cells, it has been shown that G␣ i2 is required for carbachol-induced stress fiber formation (18). In experiments employing pertussis toxin, evidence has also been obtained that the dysfunction of G␣ i causes a 40 -50% decrease in the cortical F-actin content in chromaffin cells (40) and diminishes fMet-Leu-Phe-induced actin polymerization in neutrophils (41). Furthermore, evidence for a link between the activity of G␣ i , the basal concentration of intracellular cyclic AMP, and the assembly of stress fibers in primary human granulosa-lutein cells has recently been reported (42). These observations, together with our present results with hepatocytes, suggest that trimeric G-proteins such as G i2 are involved in regulating the organization of the actin cytoskeleton in a variety of cell types.
Pertussis toxin treatment also caused fragmentation and redistribution of the ER, detected using DiOC 6 (3) and fluorescence microscopy and by electron microscopy. Furthermore, 40% of the total cellular G␣ i2 was found to be associated with microsomes, and approximately 70% of microsome-associated G␣ i2 was ADP-ribosylated by pertussis toxin treatment. These results indicate that G i2 is likely to be directly or indirectly involved in regulating the structure and intracellular distribution of the ER in hepatocytes. Moreover, considering the evidence of Hajnóczky et al. (43) that the luminal communication between intracellular Ca 2ϩ stores is cooperatively modulated by GTP and the cytoskeleton, an intriguing possibility is that G i2 is involved in maintaining the luminal continuity of the ER in hepatocytes, either via the actin cytoskeleton or by interaction with other proteins.
Pertussis toxin treatment caused a noticeable redistribution of G␣ i2 immunofluorescence from the cytoplasmic space to the cell periphery. This observation may reflect the redistribution of some G␣ i2 from the cytoplasmic space to the cell periphery. However, others have shown, using Western blotting, that compared with native G␣ i2 , ADP-ribosylated G␣ i2 has a higher affinity for the anti-G␣ i antibody employed in the present studies (44). Therefore, some of the substantial increase in G␣ i2 immunofluorescence at the cortex of pertussis toxin-treated hepatocytes may be due to an enhanced affinity of the anti-G␣ i antibody for ADP-ribosylated G␣ i2 (compared with native G␣ i2 ).
Role of Actin and G i2 in Activation of Ca 2ϩ Inflow-Pertussis toxin treatment caused a substantial inhibition of vasopressininduced Ca 2ϩ inflow with little effect on vasopressin-induced release of Ca 2ϩ from intracellular stores (present and previous (30) results). Previous studies have shown that pertussis toxin treatment completely inhibits thapsigargin-induced Ca 2ϩ inflow without a substantial effect on thapsigargin-induced release of Ca 2ϩ from the ER (45) and have shown that the effects of pertussis toxin can be mimicked by the microinjection of an anti-G␣ i2 antibody or peptide corresponding to the carboxyl region of G␣ i2 , which inhibits G i2 function (10). These results provided substantial evidence to indicate that G i2 (rather than G i3 , which is also present in rat hepatocytes and can be ADPribosylated by pertussis toxin (20,38)) is necessary for the activation of SOCs in rat hepatocytes (10). Moreover, the previous experiments also indicate that the ablation of G␣ i2 action by pertussis toxin does not substantially affect the formation of InsP 3 catalyzed by phospholipase C␤, the interaction of InsP 3 with InsP 3 receptors, the ability of InsP 3 receptors to release Ca 2ϩ from most regions of the ER, or the interaction of thapsigargin with the ER (Ca 2ϩ ϩ Mg 2ϩ )-ATPase and the inhibition of this Ca 2ϩ pump. (The possibility that ablation of G␣ i2 affects the release of Ca 2ϩ from a small region of the ER near the plasma membrane that is central to the activation of SOCs but was not detected as a reduction in vasopressin-induced release of Ca 2ϩ from intracellular stores cannot be excluded.) The present results show that two of the functions of G i2 in hepatocytes are to regulate F-actin assembly at the cortex and arrangement of the ER. It is possible that one or both of these functions is essential for the activation of SOCs. Thus, as suggested by others, the activation of SOCs may require maintenance of a region of the ER near the plasma membrane (e.g. "docking" of regions of the ER with the plasma membrane and/or the fusion of vesicles containing SOC proteins with the plasma membrane (6,46)). In this respect, it is interesting to note that the effects of G␣ i2 ablation (pertussis toxin treatment) in stabilizing F-actin at the hepatocyte cortex and inhibiting SOC activation are similar to results recently reported by Patterson et al. (6). These authors showed that, in a smooth muscle cell line, the stabilization of F-actin by different procedures (treatment with jasplakinolide or calyculin A, which induced the formation of a dense ring of F-actin around the cell periphery) also inhibited SOC activation (6).
A requirement for G i2 in the activation of SOCs has not been reported in studies of most other mammalian cells (47). This suggests that the requirement for G i2 in SOC activation in hepatocytes (10) reflects one or more aspects of the specific structure and function of these cells, such as maintenance (via G i2 regulation of the actin cytoskeleton or interaction of G i2 with another ER-associated protein) of cell polarity and/or a specific distribution of the ER throughout the cell, which is critical for the activation of SOCs. This may be due to a requirement for G i2 in the regulation of F-actin organization that is more accentuated in hepatocytes than in other cell types. Another possibility is that, in the hepatocyte, the InsP 3 receptors principally involved in inducing a decrease in Ca 2ϩ in the lumen of the ER are located some distance from the SOCs so that normal intraluminal communication through the ER is required for SOC activation (cf. Ref. 48).