INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Heterotrimeric regulatory G proteins are associated with the inner face of the plasma membrane, where they are positioned to be activated by membrane-spanning, heptahelical receptors and to regulate a variety of intracellular effectors. A common view of the ligand-driven, protein-protein interactions that characterize G protein-coupled transmembrane-signaling systems includes random collisions between proteins that diffuse freely in the plane of the plasma membrane. However, there is mounting evidence for a higher level of organization and compartmentation of these signal-transducing molecules. These suggestions are based on demonstrations of restricted mobilities of certain receptors and G proteins, the possibility of interactions of signaling components with the cytoskeleton, and failure to replicate the high degree of specificity of signaling observed in vivo by reconstitution of purified proteins in vitro [for review, see Neubig (1994)]. Thus, G proteins may be restricted to particular compartments or specializations of the plasma membrane.

We (Chang et al., 1994; Smart et al., 1995b), and others (Sargiacomo et al., 1993; Schnitzer et al., 1995) have presented evidence that G proteins can be found in plasma membrane specializations called caveolae. Although these structures are most often identified morphologically in cross-section as flask-shaped invaginations of the plasma membrane, they also can exist in a flattened state. Caveolae can be opened or closed to the external mileau and play a role in transport processes such as transcytosis in endothelial cells and potocytosis in epithelial cells. A growing body of biochemical and morphological evidence also indicates that a variety of molecules that participate in signal transduction reactions are concentrated in caveolae [for review see Anderson (1993); Lisanti et al. (1994a); and Parton and Simons (1995)]. Furthermore, it has been reported by one group of investigators that G proteins interact directly with caveolin (S.W. Li et al., 1995; Scherer et al., 1996; Tang et al., 1996, 1997), a 21-kDa membrane protein that has been localized by immunocytochemistry to the membrane coat of caveolae. However, others (Stan et al., 1996) have recently questioned the specificity of subcellular fractionation procedures that have implied localization of G proteins (and several other molecules) in caveolae and suggest that this is not their predominant site of residence, at least in rat lung vasculature.

We have examined the organization of certain G protein subunits in the plasma membrane and, to a lesser extent, the localization of other components of a prototypical G protein-regulated signal transduction pathway, the hormone-sensitive adenylyl cyclase system. Receptors communicate with a pair of homologous G proteins, one of which (G_s) mediates stimulation of adenylyl cyclase, while the other (G_i) is responsible for inhibition. We provide additional evidence that these signaling molecules are localized to distinct domains, a portion of which colocalized with caveolin. Proper organization of these signaling proteins at the plasma membrane may optimize fidelity and efficiency of signal transduction in the intact cell.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cell Culture

Mammalian cells were cultured in DMEM (high glucose) supplemented with 10% fetal calf serum, 5 U per ml of penicillin, and 5 µg/ml of streptomycin (unless otherwise noted). Madin Darby canine kidney (MDCK) cells were transfected with Lipofectamine and either empty pCB6⁺ vector (Brewer, 1994) (clone 1) or the _o expression vector, _opCB6⁺ (clone 34). Cells were selected, cloned, and maintained in medium containing 500 µg/ml Geneticin (G418 sulfate). MA104 cells were derived from rhesus monkey kidney (Roth et al., 1987), and normal human fibroblasts were from skin biopsies. Murine lymphoma (S49) cells were grown suspended in medium supplemented with 10% heat-inactivated horse serum and no antibiotics. Fall army worm ovarian (Sf9) cells were propagated by suspension in IPL-41 medium supplemented with 10% heat-inactivated fetal calf serum as described (Tang et al., 1991). All cell culture reagents were purchased from Life Technologies, Inc. (Gaithersburg, MD) except for IPL-41 medium, which was from JRH Biosciences (Lenexa, KS).

Antibodies and Western Blotting

Caveolin antibodies (mouse monoclonal and affinity purified rabbit polyclonal) were purchased from Transduction Laboratories (Lexington, KY). The properties of G protein antibodies are summarized here and in Figure 1. B087 was made (in rabbit) against a synthetic peptide representing the last 10 amino acids of _i1 and _i2 (Linder et al., 1993); its reactivity by immunoblotting was _i1 = _i2i3, _o (Figure 1A). A569 was made (in rabbit) against a "common " peptide representing amino acids 40-54 of _i (Mumby et al., 1986); its reactivity by immunoblotting is _o>_i1 = _i2 = _i3 (Figure 1A). B600 (Linder et al., 1993) and T20 (Santa Cruz Biotechnology, Santa Cruz, CA) were made in rabbit against a peptide representing the highly homologous carboxyl terminus of  subunit isoforms. Antiserum 584 is specific for _s (Mumby and Gilman, 1991). 2A and R4 are mouse monoclonal antibodies that were generated against _o or _i1 proteins, respectively, and each was specific for the immunogen without cross-reaction with other  subunits when analyzed by western immunoblotting (X. Li et al., 1995).

View larger version (24K): [in this window] [in a new window]   Figure 1.   Specificity of antibody preparations. (A) Preparations of purified recombinant G protein  subunits (25 ng) were resolved by SDS-PAGE and analyzed by silver staining or Western immunoblotting. Only the region of the gel where G protein  subunits migrate is shown (7 cm long, 11% polyacrylamide gel). If the film is exposed to the blots for an extended time (not shown), a reaction with s can be observed with the A569 antibody preparation (but not with B087 or R4). (B) Crude membrane fractions (25 µg protein) were resolved on a 16-cm long, 9% polyacrylamide gel, and immunoblots are shown. Numbers at left indicate the position of prestained molecular mass standards in kilodaltons. The crude membrane preparations are from MDCK (vector control cells, lanes 1), MA104 (lanes 2), or fibroblasts (lanes 3). The  subunit reactivities of the antibodies are indicated in parentheses near the top of the panel. A569 detects only i in cell membranes because it reacts better with i than s and there is approximately 10-fold more i than s in most cells (and there is little or no o expressed in these cells). Mab R4 is specific for i1 (X. Li et al., 1995). Monoclonal antibody R4 does not react well for immunoblotting and is not sensitive enough, by this method, to detect i1 in MDCK cells or fibroblasts; only MA104 membranes are shown for this antibody. The polyclonal caveolin antibody preparation reacts primarily with one protein band but also with other minor isoforms of caveolin in the 21-25 kDa region of the blot. Please note that the 40-kDa band labeled i (to the right of lane 3 of the caveolin blot, panel B) is residual signal remaining from a previous incubation of the blot with the i reactive A569 antibody preparation and is not from reactivity with the caveolin antibodies. Antibodies for Western immunoblotting were as follows: affinity-purified A569 and B087 at 100 ng/ml, the purified polyclonal caveolin antibodies at 500 ng/ml, culture medium from monoclonal antibody R4-producing cells diluted 1:25 (vol/vol).

For many experiments G protein-specific antibodies were affinity purified (Mumby et al., 1988). All antibody preparations employed for immunolocalization were tested for specificity of reactivity by immunoblotting of purified recombinant  subunits (Lee et al., 1994) and crude membrane fractions prepared from cultured cells (Mumby et al., 1990). Samples were treated with N-ethylmaleimide (50 mM) (Sternweis and Robishaw, 1984) and were resolved by SDS-PAGE (Laemmli, 1970). Immunoblotting of proteins transferred to nitrocellulose (Towbin et al., 1979) was performed with enhanced chemiluminescence reagents from Amersham (Arlington Heights, IL).

Immunofluorescence

To obtain plasma membranes, MA104 cells were grown on poly-L-lysine-coated coverslips. Coverslips were soaked in sterile poly-L-lysine (0.5 mg/ml) in 0.1 M sodium borate, pH 8, for 30 min or longer and then rinsed twice with sterile water before cells were plated. Fibroblasts and MDCK cells required no coating of the coverslips. Cells on coverslips were sonicated with a Vibra Cell sonicator (Sonics and Materials, Danbury, CT) set to 40% output/10 J for MA104 cells and fibroblasts and 80% output/15 J for MDCK cells (Muntz et al., 1992). Fixation was performed with 4 or 10% paraformaldehyde with similar results. Coverslips were processed for immunofluorescence using secondary antibodies (20 µg/ml) labeled with Oregon Green (Molecular Probes, Eugene, OR) or Texas Red (Zymed, South San Francisco, CA) (Muntz et al., 1992). Labeled membranes were viewed and photographed using a Zeiss epifluorescence photo microscope III RS with a 100 W DC mercury lamp (Carl Zeiss, Thornwood, NY).

Fractionation

Triton X-100 extraction and sucrose gradient centrifugation of cells were performed as described by Lisanti and colleagues (S.W. Li et al., 1995). Detergent-free caveolae were prepared on OptiPrep gradients (fibroblast, MDCK, and MA104 cells) by the method of Smart et al. (1995b). S49 cells (harvested at 2-4 × 10⁶ cells/ml) were disrupted by nitrogen cavitation (3 × 10⁷ cells/ml), and plasma membranes were collected from the interfacial areas of 20/30% and 30/40% sucrose step gradients as described (Ross et al., 1977). Two milligrams of S49 cell membranes were suspended in 23% OptiPrep and applied below a linear OptiPrep gradient as described (Smart et al., 1995b).

Adenylyl Cyclase Assay

Enzyme activity was quantified as described (Salomon et al., 1974; Smigel, 1986). Samples were assayed in a volume of 100 µl (S49) or 200 µl (fibroblasts) for 30 min at 30°C in the presence of 10 mM MgCl₂. [-³²P]ATP was purchased from DuPont NEN (Boston, MA). Five micrograms of S49 cell plasma membranes isolated from the sucrose gradients were assayed as a control for several experiments. These plasma membranes had basal adenylyl cyclase-specific activities of 0.03 ± 0.01 nmol/min/mg protein. Activity was raised four- to eightfold by 2 µM isoproterenol and 15 µM guanosine triphosphate (GTP) (0.12 ± 0.03 nmol/min/mg) or 50 µM forskolin (0.23 ± 0.09 nmol/min/mg).

Electron Microscopy

Plasma membranes were isolated from the upper surface of cells by the method of Sanan and Anderson (1991). Immunogold labeling was performed on plasma membranes that had been fixed in 3% paraformaldehyde. Secondary antibodies (goat anti-mouse or anti-rabbit from Zymed Laboratories, Inc., South San Francisco, CA) were diluted to 50 µg/ml. Rabbit anti-goat 10 nm gold conjugate (Sigma Chemical, St. Louis, MO) was used at 2 × 10¹¹ particles per ml. Grids were viewed and photographed using a JEOL JEM-100CX electron microscope.

Construction of Transfer and Expression Vectors

The vector for stable expression of _o in MDCK cells was synthesized by inserting cDNA encoding rat _o (from clone 31 in pGEM2 provided by Dr. Randall Reed, Johns Hopkins, Baltimore, MD) (Jones and Reed, 1987) into the BglII and EcoRI sites of pCB6⁺ (Brewer, 1994). A full-length cDNA encoding rat VIP21/-caveolin-1 in pBluescript (Kurzchalia et al., 1992) was kindly provided by Drs. D. Zacchetti and K. Simons (European Molecular Biological Laboratory, Heidelberg, Germany). -Caveolin-1, tagged with six histidine residues at the amino terminus (NH₆-caveolin), was generated by synthesizing an oligonucleotide cassette for substitution insertion. This cassette encodes an initiator ATG followed by six histidines and the first seven amino acid residues of caveolin; NotI and AccI sites were placed at its 5 and 3 ends, respectively. After annealing, the cassette was subcloned into the unique AccI site near the extreme amino terminus of the caveolin-coding region and the 5 NotI site of pBluescript (Strategene, La Jolla, CA). The resulting plasmid was linearized with XhoI, end-filled with Klenow fragment, and then digested with NotI to yield a fragment containing the full-length NH₆-caveolin-coding sequence. This fragment was subcloned into the NotI and SmaI sites of the Sf9 baculovirus expression vector pVL1392 (Summers and Smith, 1987) to yield a transfer vector encoding full-length -caveolin-1 (Sf9 NH₆:FL-caveolin). FL-caveolin or the first 101 amino-terminal residues of caveolin (NT-caveolin) were also fused to the carboxyl terminus of glutathione S-transferase (GST) for expression in Escherichia coli. Three oligonucleotides were designed for use in polymerase chain reaction amplification of caveolin: Oligo 1, a sense oligonucleotide encoding an XbaI site in frame with the first thirty 5 coding bases of -caveolin; oligo 2, an antisense oligonucleotide encoding bases 273-303, a glycine and six histidines, followed by a stop codon and a HindIII site; oligo 3, an antisense oligonucleotide encoding bases 507-534 and a hexahistidine tag at the carboxyl terminus followed by a stop codon and a HindIII site. Oligo 1 and oligo 2 were used to generate, by polymerase chain reaction amplification, a DNA fragment encoding NT-caveolin with a hexahistidine tag appended to the carboxyl terminus. Oligos 1 and 3 were used to generate a DNA fragment encoding full-length caveolin with a hexahistidine tag appended to the carboxyl terminus. These fragments were each digested with XbaI and HindIII and subcloned into the pGEX-2T (Pharmacia Biotechnology, Piscataway, NJ) vector in frame with GST to yield an E. coli expression vector that encoded full-length -caveolin-1 fused to the carboxyl terminus of GST (GST-FL-caveolin:CH₆) or the amino-terminal 101 residues of -caveolin-1 (GST-NT-caveolin:CH₆), each with a hexahistidine tag at the carboxyl terminus (CH₆). The nucleotide sequence of all caveolin constructs was confirmed by dye terminator sequencing using an Applied Biosystems 373A automated sequencer (Perkin Elmer-Cetus, Foster City, CA).

Purification of Caveolin and GST-Caveolin Fusion Proteins

Sf9 NH₆:FL-caveolin was extracted from recombinant virus-infected Sf9 membranes (3.5 mg/ml protein) by addition of TX-100 to 1% (vol/vol) and isolated by Ni-NTA affinity chromatography (Qiagen, Santa Clarita, CA). The highly enriched NH₆:FL-caveolin recovered from the first column was purified to near homogeneity by Q-Sepharose anion exchange chromatography as described (Hepler et al., 1996). GST-FL-caveolin:CH₆ and GST-NT-caveolin:CH₆ expressed in E. coli were purified by sequential Ni-NTA and glutathione-agarose affinity chromatography and were judged to be greater than 95% pure by SDS-PAGE and Coomassie blue staining. These proteins were used for experiments shown in Figures 8 and 9. GST-caveolin constructs without the CH₆ tag were also synthesized, but they yielded proteins from glutathione-agarose that appeared (by Western immunoblotting) to have been proteolyzed to varying extents and were thus considered poor candidates for further analysis.

View larger version (22K): [in this window] [in a new window]   Figure 8.   Assay for interaction between G protein subunits and the amino-terminal domain of caveolin. (A) Purified GST (upper panel) or GST-NT-caveolin containing caveolin:CH6 residues 1-101 (lower panel) immobilized on glutathione agarose beads (3500 pmol) was incubated with purified myristoylated o (410 pmol), bovine brain (580 pmol), or o heterotrimer (410 pmol) as indicated at the top of the figure. After allowing the proteins to interact overnight, the resin was washed six times, and bound GST and any associated proteins were eluted with 25 mM reduced glutathione. Equal portions of the load samples (L), protein that did not bind the affinity resin and was present in the flow through (FT), the first wash (W1), the last wash (W6), and glutathione eluates (E) were analyzed for the presence of G protein subunits by immunoblotting with specific G protein  and/or  antisera. Film was exposed to the blot for 2 min. (B) Samples from glutathione eluates from each binding condition described above were analyzed side-by-side for comparison and analyzed for the presence of G protein subunits by immunoblotting. Film was exposed to immunoblots for either 2 min or, to maximize detection of the chemiluminescence signals, for 15 h. There is a hint of signal for o perceptible in the GST NT-caveolin eluate only after prolonged exposure of the blot to film. This signal represents an extremely small portion of that which was loaded on the resin. An ink dot was placed to the right of each of three bands detected after the 15-h exposure: two bands corresponding to  (one each for GST and GST-NT-caveolin:CH6 elutions) and one for o in the elution from GST NT-caveolin:CH6.

View larger version (21K): [in this window] [in a new window]   Figure 9.   Failure of purified full-length caveolin or caveolin derivatives to alter steady-state GTPase activity of o or Go heterotrimer. (A) The steady state GTPase activity of purified myristoylated o (3 nM), myristoylated Go heterotrimer (o + , 3 nM each), or (3 nM) was measured (hatched bars). The relative capacities of purified GST fusion protein containing residues 1-101 of caveolin:CH6 (GST-NT-cave; 1 µM), purified full-length caveolin:CH6 fused to GST (GST-FL-cave; 1 µM), full-length NH6:caveolin purified from Sf9 cells (Sf9-FL-cave; 1 µM), or a peptide corresponding to residues 82-101 of caveolin (cave peptide; 10 µM) to alter the steady-state GTPase activity of myristoylated o (black bars) or Go heterotrimer (gray bars, o + ) were assayed but were found to be negligible. (B) The effects of increasing concentrations of caveolin peptide (residues 82-101) on the steady-state GTP hydrolysis of myristoylated o (0.6 nM) were assayed but were also found to be negligible.

Preparation of G Proteins and G Protein Binding to GST-NT-Caveolin Affinity Resin

Myristoylated _o was synthesized in E. coli and purified as described (Linder and Mumby, 1994). To examine G protein binding to GST-caveolin:CH₆, protein affinity resins were prepared following a published procedure (S.W. Li et al., 1995). Analysis of starting material (load samples), bound, and recovered GST and GST-NT-caveolin:CH₆ proteins indicated that essentially all of the applied GST fusion proteins bound to the glutathione-agarose (not shown). Purified G proteins were incubated with GST-agarose, GST-FL-caveolin:CH₆, or GST-NT-caveolin:CH₆-agarose overnight at 4°C to facilitate G protein interaction with caveolin. Unbound material, washes, and eluted proteins were analyzed by immunoblot analysis using G protein antibodies or goat anti-GST (Pharmacia, Piscataway, NJ). No G protein subunits, GST-NT-caveolin:CH₆, or GST remained bound to the agarose after glutathione elutions as assessed by immunoblot analysis of resin boiled in SDS-PAGE sample buffer (not shown).

Preparation and Analysis of Caveolin Peptide and Steady State GTP Hydrolysis Assays

A 20-amino acid peptide corresponding to -caveolin-1 residues 82-101 (DGIWKASFTTFTVTKYWFYR) was synthesized in the Howard Hughes Medical Institute Biopolymer Facility at the University of Texas Southwestern Medical Center. The identity and purity of the peptide was confirmed by high pressure liquid chromatography and mass spectral analysis. This peptide was identical in sequence to the peptide designated caveolin-2 by Lisanti and co-workers (S.W. Li et al., 1995). Peptide was solubilized in water and further resolved from low molecular weight impurities by size exclusion chromatography; recovered peptide was detected and quantified by measurement of the optical density at 280 nm. This peptide and the purified caveolin proteins were assayed for their effects on the steady-state rate of hydrolysis of [-³²P-GTP] (Hepler et al., 1996). Guanosine triphosphatase (GTPase) reactions were carried out in a final assay volume of 50 µl. Myristoylated _o was incubated together with either G alone or together with Sf9 FL-NH₆:caveolin, GST-FL-caveolin:CH₆, GST-NT-caveolin:CH₆, or caveolin peptide for 15 min at 4°C in reaction buffer (20 mM HEPES, pH 8, 1 mM EDTA, 1 mM dithiothreitol, 0.5 mM MgSO₄, 0.05% C₁₂E₁₀). Duplicate reactions were started by addition of 30 µl of reaction buffer containing 1.67 µM [³²P]GTP (50 pmol GTP/assay final; 10,000-15,000 cpm/pmol) and incubating samples for 20 min at 30°C. Reactions were terminated by addition of 750 µl of a Norit A-activated charcoal suspension (5% in 50 mM NaH₂PO₄). Samples were centrifuged and 500 µl of supernatant fraction were assayed for the presence of inorganic [³²P]phosphate by liquid scintillation counting. These experiments were conducted twice with similar results obtained.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Antibody Specificity

The specificities of the antibody preparations employed for Western immunoblotting, immunofluorescence, and immunogold labeling of _i and caveolin are summarized in Figure 1 and MATERIALS AND METHODS. The reactivity of peptide antibodies A569 and B087 with purified G protein  subunits is shown in Figure 1A; the specificity of monoclonal antibody R4 for _i1 has been demonstrated previously (X. Li et al., 1995). The antibodies to G protein  subunits interact selectively with _i in crude membrane preparations (Figure 1B), and they are thus useful reagents for immunolocalization experiments.

Immunofluorescence

We examined the distribution of _i on the inner face of isolated plasma membranes (prepared by sonicating cells adherent to coverslips) by immunofluorescence. Antibodies to G protein _i or  subunits revealed punctate patterns of immunofluorescent staining suggesting a concentration of these proteins at distinct sites on the plasma membrane (Figure 2). We compared the distribution of _i with caveolin by conducting double immunofluorescence on plasma membranes prepared from three different cultured cell types: MDCK and MA104 cells [renal epithelial cells that have been used previously for isolation and functional characterization of caveolae (Anderson et al., 1992; Sargiacomo et al., 1993; Smart et al., 1994, 1995a)] and primary human fibroblasts (which have excellent morphological characteristics and many invaginated caveolae). Although there are clearly areas where the patterns of fluorescence are coincidental, consistent with colocalization of the two proteins, the coincidence is not complete (Figure 3). The extent of colocalization appeared to be considerably less for fibroblasts than for the two epithelial cell lines. The punctate patterns for both proteins are composed of dots that are generally smaller and more variable in size than those observed with antibodies to clathrin, which is localized in coated pits (data not shown). G proteins did not appear colocalized appreciably with clatherin in double immunofluorescence experiments with antibodies to both proteins (not shown). The fluorescent dots that represent _i often appear finer than those for caveolin, and they can be so closely juxtaposed that it is difficult to reproduce them faithfully for publication. Similar patterns were observed with different antibodies to _i or caveolin; the pattern for _i was indistinguishable with antibodies P960 (Casey et al., 1990), A569, B087, and R4. Fluorescence was not observed if antibodies to the G proteins were first incubated with immunogen peptide (or relevant G protein subunit), whereas the staining pattern was preserved if the antibodies were incubated with a nonreactive G protein subunit or peptide (not shown).

View larger version (135K): [in this window] [in a new window]   Figure 2.   Immunofluorescence of plasma membranes performed with antibodies specific for G protein subunits. En face views of the inner side of plasma membrane fragments were obtained by sonicating MA104 cells adherent to coverslips. Oregon green-conjugated secondary antibodies were used to visualize primary antibodies to detect: (A) i subunits with B087 antibodies (10 µg/ml) or (B)  subunits with T20 antibodies (1 µg/ml). The larger areas of the photographs that are devoid of fluorescent signal represent spaces where plasma membrane fragments are absent. Scale bar, 2 µm.

View larger version (111K): [in this window] [in a new window]   Figure 3.   Double immunofluorescence of plasma membranes performed with antibodies specific for i and caveolin. Plasma membranes similar to those in Figure 2 were prepared from MDCK cells (A and B), MA104 cells (C and D), and fibroblasts (E and F). Oregon green-conjugated secondary antibodies were used to visualize i in the lefthand panels and Texas red-conjugated antibodies were used to detect caveolin in the righthand panels. The spillover of signal between the two fluorophores was insignificant (determined by single immunofluorescence, not shown). Affinity-purified B087 (10 µg/ml) was used for panels A and C, R4 (1:100 dilution of culture medium from antibody-producing cells) for panel E, caveolin monoclonal antibody (5 µg/ml) for B and D, and affinity-purified polyclonal antibody (10 µg/ml) for panel F. Scale bar, 4 µm.

Subcellular Fractionation of G Proteins, Caveolin, and Adenylyl Cyclase Activity

Methods designed to isolate low-density membrane fractions from tissues or cultured cells have permitted enrichment of both G proteins and caveolin. MDCK cells, Triton X-100 extraction, and sucrose gradients have been employed frequently by others (Sargiacomo et al., 1993; Lisanti et al., 1994b; S.W. Li et al., 1995; Song et al., 1996) with results similar to those shown in Figure 4. Most of the cellular proteins (assayed by Ponceau S staining of the blot, not shown) remained in the region where the Triton X-100 extract was loaded at the bottom of the tube (fractions 8-11), while Triton X-100-insoluble membranes of low density floated up into the sucrose gradient (lower numbered fractions). Immunoblotting revealed that endogenous _i and  subunits, exogenously expressed _o protein, and the bulk of the caveolin fractionated closely together. G proteins and caveolin (from MDCK cells) also fractionated together when this procedure was modified by replacement of detergent with sodium carbonate, pH 11 (Song et al., 1996) (not shown).

View larger version (50K): [in this window] [in a new window]   Figure 4.   Fractionation of endogenous and ectopically expressed G proteins and caveolin from TX-100 extracts of MDCK epithelial cells. Stably transfected MDCK cells that ectopically express o (panel A) were compared with G418-resistant control cells that had been stably transfected with the empty vector (panel B). Four milliliters of detergent extract, adjusted to 40% sucrose, was loaded at the bottom of a tube followed by a 7-ml linear gradient of 30-5% sucrose. After centrifugation, 0.8-ml fractions were collected from the top of the gradients, half of which was acetone-precipitated and analyzed by Western immunoblotting. Most of the cellular protein (assayed by Ponceau S staining of the blots, not shown) remains below the gradient in the five bottom fractions (numbered 10-14). Ectopically expressed o (A) consistently comigrates with endogenous i, , and caveolin (cav) in detergent-resistant, low-density fractions centered around fraction 6 (A and B). i was detected by B087 antiserum (1:10,000 dilution),  by B600 antiserum (1:10,000), o by culture medium from Mab 2A-producing cells (1:500), and caveolin (cav) by the purified polyclonal antibodies (30 ng/ml). Not shown are results for s, which comigrates with the other G protein subunits from MDCK and MA104 cells.

An alternative method of fractionation involves floatation of plasma membrane-containing fractions (from a Percoll gradient) through a continuous gradient of OptiPrep (Smart et al., 1995b). Endogenous G protein _i subunits and  subunits, as well as exogenously expressed _o protein, did not cofractionate completely with caveolin through the OptiPrep gradients. We routinely observed that a substantial portion of G protein migrated into lower density fractions than did caveolin. These observations were made for MDCK cells (Figure 5A-C), MA104 cells (not shown), and fibroblasts (not shown) and are consistent with the results of double immunofluorescence experiments shown in Figure 3.

View larger version (44K): [in this window] [in a new window]   Figure 5.   OptiPrep gradient fractionation of detergent-free plasma membranes from MDCK epithelial or S49 lymphoma cells. Sonicated plasma membranes were brought to 23% OptiPrep in 4 ml and were placed at the bottom of the tube. A linear gradient of 20-10% OptiPrep was poured on top of the plasma membranes. After centrifugation, either 1 ml (panels A, B, C, and F) or 0.7-ml fractions (panels D and E) were taken from the top of the tube. These fractions were analyzed for total protein content by Bradford assay (panels A and D), G protein or caveolin content by Western immunoblotting (panels B, C, and E), or adenylyl cyclase activity (panel F). The specific activity of isoproterenol- (Iso-) and forskolin- (Fsk-) stimulated adenylyl cyclase in the individual fractions 1-6 and combined fractions 7-11 are shown in panel F. Antibody preparations utilized for immunoblotting were the same concentration as Figure 4 but, in addition, 584 antiserum was employed for detection of long and short isoforms of s at a dilution of 1:5,000.

We extended our fractionation studies to S49 lymphoma cells because they lack detectable caveolin (by immunoblotting) and have been utilized extensively to study G protein-mediated signal transduction. The lack of caveolin in these cells is consistent with reports from others who have studied lymphocytes (Fra et al., 1994). Plasma membranes from these cells were fractionated by the OptiPrep method (Figure 5, D and E), with results similar to those obtained with MDCK cells (Figure 5, A-C). G protein subunits (_i, both long and short isoforms of _s, and ) were found in fractions that migrated up into the gradient and were resolved from the bulk of the protein (Figure 5E). Although antibodies of sufficient quality to detect -adrenergic receptors and adenylyl cyclase in these fractions are not available, the highest specific activities of -adrenergic receptor- and forskolin-stimulated cAMP synthesis were found in fractions 3-6 (Figure 5F). Thus, all components of a hormone-sensitive adenylyl cyclase system are present in membrane fractions with low densities and can be separated from the bulk of the plasma membrane protein prepared on sucrose gradients from cells that do not contain caveolin. Similar results were obtained for fibroblasts. The highest total activity of forskolin-stimulated adenylyl cyclase in fibroblast membranes was present in fractions 3-7 (Figure 6B), and the highest specific activity was found in fractions 3-6 (Figure 6C). Fibroblast G proteins were found (by Western immunoblotting, not shown) evenly spread, as in MDCK cells (Figure 5, B and C), over more fractions (1-8) than was the bulk of caveolin (not shown) and adenylyl cyclase activity (Figure 6B). Isoproterenol-stimulated adenylyl cyclase activity was not detected in membranes from fibroblasts. Depending on the cell type and variations between independent experiments, 50-80% of total adenylyl cyclase activity was found in fractions 1-6 whereas only 15-20% of total plasma membrane protein was present in these fractions.

View larger version (17K): [in this window] [in a new window]   Figure 6.   Adenylyl cyclase activity in OptiPrep gradient fractions prepared from fibroblasts. Plasma membranes were fractionated similarly to those shown in Figure 5. (A) Protein content of each 1-ml fraction. (B) cAMP produced in an assay tube containing 120 µl of gradient fraction incubated with 25 µM forskolin. (C) Specific activity of forskolin-stimulated adenylyl cyclase.

Immunogold Electron Microscopy

We examined plasma membranes of fibroblasts and other cells to determine sites of localization of _i and to evaluate the extent to which this protein is associated with morphologically recognizable caveolae. Similar immunogold labeling patterns were obtained with peptide antibodies directed to two different regions of _i (Figure 7B and C). This pattern differed from that obtained with an anti-caveolin antibody (Figure 7A). Both anti-_i antibodies yielded clusters of gold particles that were most frequently associated with structures of irregular shape and size that we were unable to identify. Coated pits were not labeled by _i antibodies (Figure 7, B and C) consistent with our double immunofluorescence results with clathrin antibodies (not shown). Immunogold labeling of _i was reduced dramatically (74-88% in two experiments) when the primary antibodies were combined with the immunogen peptide but not with an irrelevant peptide. Only about 10-20% of morphologically identifiable caveolae in fibroblasts were labeled by _i antibodies (see solid arrows, Figure 7; Table 1). By contrast, essentially all of the morphologically identifiable (invaginated, doughnut-shaped) caveolae were labeled by the preparation of caveolin antibodies. Similar patterns of immunogold staining of _i were also seen with plasma membranes of MDCK and MA104 cells, although fewer invaginated caveolae could be identified (not shown).

View larger version (69K): [in this window] [in a new window]   Figure 7.   Immunogold labeling of plasma membranes with caveolin or i antibodies. Electron micrographs show en face views of the inner side of plasma membrane fragments that have been torn from the upper surface of cultured fibroblasts. The location of some of the morphologically identifiable caveolae (full or partial doughnut shapes indicated by closed arrows) and coated pits (flat or curved honeycomb patterns indicated by open arrows) are shown. (A) Caveolae are decorated well by gold particles when the polyclonal (panel A, 1 µg/ml) or monoclonal (not shown) caveolin antibodies are utilized. (B) Morphologically identifiable caveolae are infrequently labeled by i reactive B087 antibodies (10 µg/ml) or (C) A569 antibodies (10 µg/ml). The letters at the upper left corner of each panel are placed over a small area that is devoid of plasma membrane or gold particles. Scale bar, 0.5 µm.

View this table: [in this window] [in a new window]   Table 1.  Percentage of caveolin and Gi immunogold localized to invaginated caveolae of fibroblasts

Invaginated caveolae are not evenly distributed on the fibroblast plasma membrane, and we sought to determine whether _i is enriched in areas that have high densities of these membrane specializations. Square micron areas of the plasma membrane that were obviously caveolae-rich or caveolae-poor were chosen arbitrarily for counting the total number of caveolae and the total number of gold particles observed after labeling with either _i or caveolin antibodies. As expected, the caveolin antibody yielded total numbers of gold particles in the caveolae-rich areas that were 10- to 20-fold greater than those in the caveolae-poor areas (Table 2). By contrast, immunogold labeling with either of two _i antibodies was greater in areas of plasma membrane that were relatively poor in invaginated caveolae.

View this table: [in this window] [in a new window]   Table 2.  Quantification of caveolin and Gi immunogold in caveolae-enriched and caveolae-poor areas of fibroblast plasma membranes

Interaction of Caveolin with G Proteins

Caveolin has been proposed to act as a scaffold that localizes G proteins to caveolae and, further, as a regulator of the activity of these signal transducers (S.W. Li et al., 1995; Scherer et al., 1996; Tang et al., 1996, 1997). Li, S.W. et al. (1995) demonstrated qualitatively that recombinant G protein  subunits (_o and _i2) from extracts of transfected MDCK or baculovirus-infected Sf21 insect cells bind to GST-FL-caveolin or GST-NT-caveolin. Because these experiments were not performed with purified proteins, the interaction of  subunits with the caveolin constructs could have been indirect or require additional factors. If G proteins bind to caveolin directly, pure  subunits should bind to GST-caveolin attached to glutathione agarose beads. We investigated such binding with pure myristoylated _o in the presence or absence of . The _o and appeared to be quantitatively recovered in the flow through and first wash of the GST-caveolin resins in a manner indistinguishable from GST without appended caveolin (Figure 8A for GST-NT-caveolin:CH₆; GST-FL-caveolin:CH₆ not shown). Only after prolonged exposure of the blot to film could we detect a hint of _o in the glutathione elution from the GST-NT-caveolin:CH₆ agarose (Figure 8B), and we estimate that the eluted material represented much less than 1% of that added to the caveolin constructs. In experiments conducted similarly to those published (S.W. Li et al., 1995), we were unable to detect binding of recombinantly expressed unpurified _o or _i2 (from extracts of baculovirus-infected Sf9 cells) to FL-caveolin:CH₆ (data not shown). Thus, we were unable to demonstrate a substantial interaction between G protein subunits and caveolin in the presence or absence of other cellular components.

It has been reported that synthetic peptides corresponding to specific regions of caveolin (e.g., residues 82-101 of caveolin-1) modulate the basal GTPase activity of _i and _o and alter binding of GTPS to these proteins (S.W. Li et al., 1995; Scherer et al., 1996; Tang et al., 1996, 1997). If these represent physiologically relevant reactions, intact caveolin should cause similar effects. We found that neither the caveolin-1 peptide (reported to inhibit GTPase activity), Sf9 NH₆:FL-caveolin, GST-FL-, nor GST-NT-caveolin:CH₆ influenced the basal GTPase activity of _o significantly in the presence or absence of (Figure 9); the inhibitory effect of on the GTPase activity of _o is characteristic (Figure 9A, hatched bars). We also failed to observe effects of the peptide or proteins on binding of GTPS to _o (not shown). Finally, we tested two other peptides that were said to modulate GTPase activity (Scherer et al., 1996; Tang et al., 1996). These peptides represent regions of caveolin-2 and caveolin-3 isoforms that correspond to the caveolin-1 sequence specified above. These two peptides were difficult to solubilize and gave irreproducible results. In summary, we cannot reproduce the published effects of caveolin peptides on the GTPase activity of _o or on GTPS binding to the protein, and we also observe no effects of FL- or NT-caveolin on these parameters.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Individual cells express an imposing array of proteins that participate in G protein-mediated signal transduction reactions: a large number of receptors; a substantial variety of G protein , , and  subunits; many distinct effectors that exist in multiple isoforms; and regulators such as members of the large families of receptor kinases, arrestins, regulators of G protein signaling (RGS) proteins, and probably others. Although the critical phenomena that result from the interactions between these components can and at times must be studied using purified proteins reconstituted in vitro, other aspects of G protein-mediated signaling are not yet observable in such well-defined systems. For example, the magnitude of stimulation of adenylyl cyclase activity by appropriate hormones is much greater in intact cells than in membranes; the specificity of receptor-G protein interactions in reconstituted systems appears to be less stringent than that observed in membranes or cells; requirements for specific combinations of isoforms of G protein subunits to observe appropriate responses to individual receptor agonists cannot yet be duplicated in vitro (Neubig, 1994). These and other phenomena suggest the possibility of or need for nonrandom, functional organization of these components in the plasma membrane to achieve the requisite speed, magnitude of response, functional specificity, and regulation that characterize these systems.

A substantial fraction of the total activity and the highest specific activity of isoproterenol-stimulated adenylyl cyclase activity in S49 cells was found in low-density subfractions of the plasma membrane with buoyant properties similar to those of caveolae (Figure 5F). Thus, G_s-coupled receptors, G_s itself, and adenylyl cyclase appear to be localized in similar domains  perhaps together to increase the efficiency and fidelity of signal transduction at distinct sites on the plasma membrane. In this regard, investigators have come to different conclusions about the distribution of endogenous -adrenergic receptors in the plasma membrane. Some have indicated that the distribution of unstimulated receptors was random and homogeneous (Kaveri et al., 1987; Muntz et al., 1988; Raposo et al., 1989), while others have observed nonrandom patterns (Zemick and Strader, 1988; Wang et al., 1989). Cytochemical experiments have also suggested that adenylyl cyclase activity is associated with structures resembling caveolae (Wagner et al., 1972; Slezak and Geller, 1984; Rechardt and Hervonen, 1985).

We and others (Wang et al., 1989; Lewis et al., 1991) have focused on G proteins and found them to be clustered at the inner surface of the plasma membrane, indicating their concentration at particular sites. Our membrane fractionation data indicate localization of G proteins in low-density subfractions of the plasma membrane and are consistent with colocalization of a substantial fraction of G protein subunits with caveolin. These experiments are at least semiquantitative, in that all of a particular G protein subunit in the cell can presumably be detected by immunoblotting after denaturation; however, the resolution of this technique is relatively low. Immunofluorescence experiments indicate that there is significant colocalization of G protein subunits and caveolin, but that it is incomplete and variable among cell types. This is consistent with recent reports demonstrating coimmunoprecipitation of G proteins or related signaling components with caveolin using cell extracts (Chun et al., 1994), some of which included detergent-insoluble particles (de Weerd and Leeb-Lundberg, 1997; Feron et al., 1997). Incomplete and/or cell type-dependent colocalization of G proteins and caveolin may explain the results of Stan et al. (1996), who concluded that G proteins are not enriched in caveolae. Although this may be true in the cells studied, those of rat lung vasculature, the techniques utilized for isolation of caveolae (perfusion of lung with cationized silica and the immunoisolation of caveolin-containing membranes) may have altered the distribution of the proteins in question. Our electron microscopy studies suggest that most of the G protein _i subunits reside in irregular structures that could not be identified. Although the resolution of the microscopic techniques is of course intrinsically high, quantitation can be lost. It is possible that populations of G protein subunits are not visualized in these experiments because of fixation artifacts or masking of epitopes. These caveats aside, we would conclude that morphologically distinct, invaginated caveolae are a minor location of _i in the cells we studied by electron microscopy.

Caveolin per se is not required for localization of G proteins in low-density subfractions of the plasma membrane (Figure 5E). In this regard we must question a premise for colocalization of caveolin and G proteins  direct protein-protein interactions. We were unable to observe interactions between G protein  or subunits and caveolin, and we were unable to detect functional effects of caveolin or caveolin-based peptides on guanine nucleotide binding or hydrolysis of G protein subunits. Our use of recombinant myristoylated _o purified from bacteria (protein that could lack important covalent adducts such as palmitate), may have prevented our detection of a substantial direct interaction between G proteins and caveolin. We are unable to explain why, in experiments conducted similarly to those published (S.W. Li et al., 1995), we did not detect interaction between caveolin and unpurified _o or _i2 from extracts of baculovirus-infected Sf9 cells (in which G proteins are properly palmitoylated) or effects of caveolin-based peptides on G protein function.

Our working hypothesis on localization is that G proteins in particular (and probably entire G protein-coupled signaling systems) are not randomly distributed in the plasma membrane but are physically and functionally segregated into distinct membrane domains. If caveolae are defined narrowly as caveolin-containing invaginations of the plasma membrane (Simons and Ikonen, 1997) then we conclude that these structures are not the predominant location of _i [Table 1 and Stan et al., (1996)]. If, instead, caveolae are defined more broadly as a membrane system that includes low-density subdomains of the plasma membrane (Anderson, 1998), then our data show that such structures are major sites for localization of G proteins and adenylyl cyclase. Using this definition, caveolae may be similar or identical to the protein-containing clusters of glycosphingolipids and cholesterol that Simons and Ikonen (1997) have described as mobile rafts within the fluid membrane bilayer. It would be of great interest to examine the signaling properties of cells in which these structures were specifically disrupted.