Mechanism of Action of Monoclonal Antibodies That Block the Light Activation of the Guanyl Nucleotide-binding Protein, Transducin”

Seven monoclonal antibodies to the a subunit (G,) of the frog photoreceptor guanyl nucleotide-binding protein (transducin or G-protein) have been characterized as to their effect on G-protein function, and this has been correlated in the accompanying paper (Deretic, D., and Hamm, H. E. (1987) J. Biol. Chem. 262, 10839-10847) with the antibody-binding sites on G, tryptic fragments. Antibodies 4A, 7A, 7B, 7C, and 7D are members of a class of antibodies that block G- protein activation by light and therefore also block activation of the cGMP phosphodiesterase. All these blocking antibodies also block the interaction of G-protein with rhodopsin as measured by the light-scat- tering ”binding signal,” and as measured by the stabilization of meta-rhodopsin I1 by bound G-protein (ex- tra-meta-rhodopsin 11). The antibodies (or Fab fragments) also solubilize Gdy from the membrane in the dark under isosmotic conditions and thus interfere with G, interaction with the membrane. Antibody 4A also blocks the extra-meta-rhodopsin I1 generated by G-protein-rhodopsin

with guanyl nucleotide-binding proteins which interact with effectors such as adenylate cyclase, cGMP phosphodiesterase, or phosphoinositide phosphodiesterase. Recent evidence suggests that these receptors constitute a related class of proteins (Dixon et al., 1986;Martin et al., 1986); in the case of the guanyl nucleotide-binding proteins (G-proteins)' linking hormone receptors to their effectors, it is already clear that there is a family of G-proteins with similar structure and function (Bitensky et al., 1982;Manning and Gilman, 1983;Cerione et al., 1986;Lochrie et al., 1985;Medynski et al., 1985;Tanabe et al., 1985;Yatsunami and Khorana, 1985;Robishaw et al., 1986;Nukada et al., 1986aNukada et al., , 1986b. These G-proteins belong to a larger family of guanyl nucleotide-binding proteins which use GTP-GDP exchange as a triggering mechanism including tubulin, elongation factors (Halliday, 1984), and the ras oncogenic protein p21 .
In photoreceptors, the GTP-GDP exchange reaction of the GTP-binding protein (transducin or referred to here as Gprotein) is triggered by interaction with activated rhodopsin (Godchaux and Zimmerman, 1979;Fung and Stryer, 1980), and activated G-protein in turn activates a cGMP phosphodiesterase (Fung et al., 1981). G-protein has been purified and found to be a heterotrimer, cu(P7) (Kuhn, 1980;Fung et al., 1981;Baehr et al., 1982), and several functional sites have been identified. The GTP-binding regions of the molecule are found on the a subunit and have been identified by determining homologies with nucleotide-binding sites on other proteins McCormick et al., 1985), and cholera and pertussis toxin ADP-ribosylation sites have been determined (Abood et al., 1982;West et al., 1985). Functional sites that have not yet been determined include sites of interaction with rhodopsin and phosphodiesterase. The amino acid sequence of G, has recently been deduced from the cDNA sequence by several laboratories (Tanabe et al., 1985;Medynski et al., 1985;Yatsunami and Khorana, 1985).
1083 1 10832 Antibody Blockade of G-protein-Rhodopsin Interaction for elucidation of structure-function relationships of G-proteins. In this study, we examined the mechanism of action of mAb 4A blockade, as well as four independently derived monoclonal antibodies that block G-protein activation by light. All five antibodies appear to block a site on the Gprotein that interacts with rhodopsin. In the accompanying paper (Deretic and Hamm, 1987), the antigenic sites of the blocking antibodies are shown to be very near t h e COOH terminus of the G-protein ar subunit; the antigenic site of mAb 4H, which does not block activity, is on another region of t h e molecule.

MATERIALS AND METHODS
Preparation of Rod Outer Segments, Rod Membranes, and Proteins-Percoll-purified frog rod outer segments (ROS) were prepared, and fractionation of ROS proteins into soluble, peripheral, and membrane fractions was done as described by Hamm and Bownds (1986). Bovine ROS were prepared according to Papermaster and Dreyer (1974). Bovine ROS disc membranes were prepared according to Smith et al. (1975) with the modification of 2.5% Ficoll instead of 5% (Bauer and Mavromatti, 1980). Extraction of bovine soluble proteins was performed according to the method of Kihn (1980) as described by Emeis and Hofmann (1981). Bovine G-protein was prepared according to Fung (1983).
Cyclic GMP Phosphodiesterase-This was measured as described by  using the proton evolution assay developed by Yee and Liebman (1978). Purified ROS (100 p~ rhodopsin) were suspended in Ringer's solution (115 mM NaCl, 2 mM KCl, 2 mM MgC12, 0.1 mM CaC12 adjusted to lo-' M with 0.39 mM EGTA, 10 mM HEPES, pH 7.8, with 50 kallikrein-inactivating units of Trasylol and 10 p~ leupeptin), disrupted by passage through a 26gauge needle in the presence of 50 p~ purified monoclonal antibody or control IgG, and incubated 30 min at room temperature. ROS were diluted (final rhodopsin concentration determined by difference spectroscopy, 5 p~) , and phosphodiesterase was assayed in the presence of 4 mM cGMP, 0.5 mM ATP, 0.5 mM GTP in Ringer's solution, pH 7.8, in the dark and after flashes of light of various intensities. Phosphodiesterase activity is expressed as moles of cGMP hydrolyzed per mole of rhodopsin per minute. The concentration of G-protein was estimated based on a stoichiometry of 1 G-protein/lO rhodopsins (Hamm and Bownds, 1986). Protein A-purified monoclonal antibodies (10 mg/ml) were added at a concentration calculated to be a 5:l molar ratio with G-protein. In other experiments (described below), calculations of rhodopsin G-protein, and antibody concentrations were always done in the same way. In some experiments, the antibody:(=-protein molar ratio was varied from 5:l to 0.2:l.
Light Scattering-This was measured as described by , who showed that the "binding signal" measured in ROS membranes in response to light is a physical consequence of stable G-protein interaction with rhodopsin in the absence of GTP. Frozen bovine ROS fragments were thawed and suspended in 120 mM KCl, 10 mM Tris/MOPS, 1 mM MgC12, 1 mM dithiothreitol, incubated with various concentrations of purified antibodies for 30 s to 30 min, and then diluted to a final rhodopsin concentration of 5 PM. The lightscattering binding signal was measured in a modified Durrum Dl17 spectrophotometer with a thermostated cuvette as the transmittance at 704 nm in response to a flash of light bleaching 2% rhodopsin.
Extra-meta-rhodopsin II Formation-This was measured according to the methods described by . This is a spectroscopic measure based on the fact that for T 5 12.0 'C and pH 2 7.5, photoexcited rhodopsin exists in an equilibrium between two spectroscopically different states, meta-rhodopsin I (Amm = 480 nm) and meta-rhodopsin I1 (Amax = 380 nm). In the absence of GTP, tight binding of G-protein to photoexcited rhodopsin stabilizes meta-rhodopsin I1 (Emeis et al., 1982). In the presence of GTP and its analogues, extra-meta-rhodopsin I1 forms transiently and subsequently decays ; this was used to monitor the kinetics of G-protein activation in the presence of monoclonal antibodies. Bovine disc membranes (5 p~ rhodopsin) recombined with the low ionic strength extract containing G-protein were incubated with antibodies in Ringer's solution, and then extra-meta-rhodopsin I1 was measured in a Shimadzu UV300 two-wavelength spectrophotometer (Fig. 3) or in the instrument described by Hofmann and Emeis (1981) (Fig. 5) as the absorbance change at 380 nm minus the absorbance change at 417 nm (meta-rhodopsin I1 isosbestic to meta-rhodopsin I) to correct for light scattering .
Disc membranes were solubilized in nonanoyl-N-methylglucamide, and rhodopsin was purified according to De Grip (1982). The nonanoyl detergent was then exchanged against the biphenyl detergent on a concanavalin A-Sepharose column.
Gel Electrophoresis-This was performed according to Laemmli (1970); and for densitometry, Coomassie Blue-stained gels were scanned with an EC densitometer and a Hewlett-Packard integrator.
Generation of Monoclonal Antibodies to G,-Antibodies 4A, 4C, and 4H were generated and characterized as described by Witt et al. (1984) and . Monoclonal antibodies of the 7 series were generated by in vitro stimulation of spleen cells from one BALB/c mouse with 100 ng of purified G, , using the method of Matthew and Patterson (1983). Fusion of immunized lymphocytes to the myeloma cell line P3-X63/Ag8.653, screening for antibody-producing hybrids, subcloning, characterization, and purification of antibodies were performed using the methods described by Witt et al. (1984). All antibodies (7A, 7B, 7C, and 7D) were of the IgG2b subtype.
Immunoprecipitation of G-protein-This was performed as follows. Purified G-protein, labeled with ['251]iodonaphthyl azide (Berkovici and Gitler, 1978) was incubated for 1 h at room temperature with different antibodies at a ratio of antibodies to G-protein of 2:l. Formalin-fixed Staphylococcus aureus cells (Bethesda Research Laboratories) were centrifuged at 3000 x g for 10 min, and the cell pellet was resuspended in an equal volume of phosphate-buffered saline, pH 7.2, containing 10% (w/v) P-mercaptoethanol and 3% (w/v) SDS and boiled for 30 min to reduce the protein background. After centrifugation at 3000 X g for 10 min, the cells were washed in 150 mM NaC1, 5 mM EDTA, 50 mM Tris, pH 7.4, 0.02% sodium azide, 0.5% Nonidet P-40 (NET buffer), centrifuged, and resuspended in the same buffer at 10% (w/v). The washed cell suspension (200 pl) was then added to protein-antibody complex containing 10 pg of antibody and incubated for 1 h at room temperature. Precipitates were centrifuged for 10 min at 3000 X g and washed three times with NET buffer. The immunoprecipitated proteins were eluted from S. aureus cells by boiling for 5 min in Laemmli (1970) sample buffer and run on 12.5% SDS-polyacrylamide gel electrophoresis. Autoradiography of the dried gel was performed using Kodak X-ARB film and an intensifying screen for 72 h at -70 "C.
Preparation of Fab Fragments-This was performed as follows. 10 mg of Protein A-purified mAb 4A in 1.8 ml of Ringer's solution was dialyzed against 0.1 M acetate, pH 5.5. This solution was brought to 2 mM EDTA, 1 mM cysteine, and 1 mg of papain/100 mg of antibody. The mixture was incubated for 6 h a t 37 "C, and digestion was stopped by addition of iodoacetic acid (final concentration, 10 mM). Fab fragments were then dialyzed against Ringer's solution and concentrated. Their purity was assessed by both reducing and nonreducing SDS-polyacrylamide gel electrophoresis. No residual complete antibody molecules were detected on Coomassie-Blue-stained gels.

RESULTS
A series of monoclonal antibodies was generated to the a subunit of the photoreceptor G-protein and was screened t o find antibodies that block functional sites on G-protein (Hamm and . Fig. 1 shows that one of these, antibody 4A, blocks light activation of cGMP phosphodiesterase, whereas another antibody of the series, 4H, has no effect. To investigate the mechanism of action of the blocking antibody, we followed the phosphodiesterase activation pathway backwards, examining the effect of the antibody on each step that could potentially be perturbed and could be measured. Antibody 4A blocks the light-activated GTP-GDP exchange reaction (Hamm and , whereas antibody 4H had no effect on this reaction (data not shown).
To examine whether the antibody blocked activation of G-protein

FIG. 1.
Antibody 4A blocks light-activated cGMP phosphodiesterase activity, whereas another antibody that binds to G,, antibody 4H, and control IgG from the parent myeloma, NS 1, have no effect. Percoll-purified frog ROS were suspended in Ringer's solution and disrupted by passage through a 26-g needle, and ROS membranes (5 PM rhodopsin) were incubated with a 5:molar excess of antibodies over G-protein for 30 min. Then phosphodiesterase (PDE) activity was measured at room temperature using the proton generation method of Yee and Liebman (1978) in the dark or after flashes of a calibrated light bleaching a known amount of rhodopsin (3 X 1 0 ' rhodopsin/outer segment). The initial rate of proton generation is plotted.
by blocking interaction with rhodopsin, two complementary methods were used light scattering (Kiihn et ul., 1981;Bennett et ul., 1982) and extra-rnetu-rhodopsin I1 formation Emeis et al., 1982). G-protein interaction with the membrane was examined by centrifugation experiments that quantitated the amount of G-protein in the supernatant or the pellet.
The light-scattering signal measured at 704 nm in the absence of GTP is based on the observation that flash illumination causes fast changes of the light scattering of disc membranes (Hofmann et al., 1976), representing an increase in the turbidity of the ROS suspension as a consequence of G-protein binding to rhodopsin. This signal is a stoichiometric measure of the G-protein bound to rhodopsin (Kuhn et ul., 1981;Bennett et ul., 1982). We examined the effect of several anti-G, antibodies on the light-scattering binding signal. Fig.  2A shows that antibody 4A blocks more than 80% of the binding signal in response to a flash of light bleaching 2% of the rhodopsin. In the presence of antibody 4H or nonspecific control IgG, the binding signal was undistinguishable from the control. The antibodies were present at an excess of 5:l over G-protein in this experiment; however, in experiments where the molar ratio of antibody to G-protein was varied, we showed that the blockade was complete at a 1:l stoichiometry between antibody 4A and G-protein (data not shown). In this experiment, the ROS were preincubated with antibody for 30 min, but varying the preincubation time showed that the antibody inhibition developed rapidly: the half-time for the inhibitory effect is less than 20 s, which was the shortest preincubation time we could reliably measure (data not shown). The lack of effect of antibody 4H is not due to a lower affinity for G-protein; Witt et al. (1984) showed that the antibody 4A and 4H affinities are very similar. Also, 2. A , the light-scattering binding signal, a consequence of Gprotein-rhodopsin interaction, was also blocked by antibody 4A, but antibody 4H had only a minor effect. Bovine ROS membranes were suspended at a concentration of 5 PM rhodopsin and incubated in the presence of Ringer's solution or 0.5 pM (1:l ratio with G-protein) control IgG or antibody 4H or 4A, and the binding signal was measured at room temperature at 704 nm after a flash of light bleaching 2% rhodopsin using the method of . The vertical scale is relative transmitted intensity change. B, antibodies 7A, 7B, 7C, and 7D also blocked the binding signal. All signals are from the first flash.
antibody 4H did not block the binding signal even after overnight incubation (data not shown). Although antibody 4H had no effect on the response to one flash ( Fig. 2 A ) , antibody 4H decreased the response to a second flash given to the same sample, as compared to control IgG. Fig. 2B shows the effect of four other G, antibodies on the binding signal. Antibodies 7A, 7B, 7C, and 7D all block the binding signal to a similar extent as antibody 4A at a molar ratio of 1:l with G-protein.
Because of its complex generation mechanism, suppression of the binding signal does not necessarily indicate blocking G-protein interaction with rhodopsin. Thus, it was possible that the antibody blocks this physical consequence without blocking G-protein interaction with rhodopsin. A different monitor of G-protein-rhodopsin interaction uses the fact that the binding conformation of rhodopsin coincides with the 380 nm intermediate rnetu-rhodopsin I1 (Emeis et ul., 1982;Bennett et ul., 1982). rnetu-Rhodopsin 11 is in equilibrium with its tautomeric form, metu-rhodopsin I (Matthews et ul., 1963;Parkes and Liebman, 1984). Stabilization of metu-rhodopsin I1 by bound G-protein expresses itself in a shift of this equilibrium toward metu-rhodopsin 11. The resulting enhanced formation of rnetu-rhodopsin I1 (so called extra-rneturhodopsin 11; Emeis and Hofmann, 1981) is easily measured because of the large spectral differences between metu-rhodopsin I (maximally absorbing at 480 nm) and rnetu-rhodopsin I1 (380 nm).
We examined the effect of the blocking antibody on extrarnetu-rhodopsin I1 formation, a more direct measure of Gprotein-rhodopsin interaction. The first two curves of Fig. 3A show the effect of light bleaching 3% rhodopsin on rneturhodopsin I1 measured spectrophotometrically in the absence or presence of 100 PM GTP. In the presence of high concentrations of GTP, no extra-rnetu-rhodopsin I1 can be observed because the level of transiently formed G-protein-rhodopsin complexes is too small. Under these experimental conditions, this measurement is therefore a good control for the equilibrium level of rnetu-rhodopsin I1 without interference of Gprotein. The difference between the first two curves represents the extra-metu-rhodopsin I1 generated by stable Gprotein-rhodopsin binding. The third curve of Fig. 3A shows that in the presence of low concentrations of the nonhydro-

FIG. 3. Formation of extra-meta-rhodopsin 11, a measure of rhodopsin-G-protein interaction in bovine disc membranes recombined with low ionic strength extract, is blocked by some antibodies.
A, extra-rneta-rhodopsin I1 is formed after a flash of light bleaching 3% rhodopsin in the absence of GTP (first curue), compared to the normal amount of rneta-rhodopsin I1 formed in the presence of 100 &M GTP (second curue). At 20 PM GppNHp, extrarneta-rhodopsin I1 forms but then decays at a rate proportional to the activation of G-protein (third curue). Experiments were performed at pH 7.5 and 3 "C; all signals are from the first flash. I?, two antibodies, 4H and 4C, have no effect on either extra-meta-rhodopsin I1 formation or its loss, whereas antibody 4A blocks the extra-rneta-rhodopsin I1 signal as well as its decay in a dose-dependent manner. Control IgG had no effect (not shown). C, comparison of the effect of different antibodies (1:l molar ratio) on stable extra-rneta-rhodopsin I1 formation. Change of the absorbance difference due t~ the amount of extra-mta-rhodopsin I1 formed by the first flash, compared to the 100 p~ GTP control, which gives the final absorbance level determined by the meta-rhodopsin I-rneta-rhodopsin I1 equilibrium without interference of G-protein. WM+EX, washed disc membranes reconstituted with protein extract, control without antibody; IgG, nonspecific control antibody. Conditions were: 2 "C, pH 8.0, flash bleaching 3% rhodopsin.
lyzable GTP analogue GppNHp, extra-metu-rhodopsin I1 is generated and then decays at a rate that is proportional to Gprotein activation . Thus, at this temperature (3 "C), the rate of G-protein binding to rhodopsin and the rate of G-protein activation can be followed indirectly by measuring the amount of metu-rhodopsin 11.
Antibodies were added to reconstituted ROS membranes in this condition to look at the effect of antibodies on both these processes. The first two curves of Fig. 3B show that antibodies 4H and 4C do not affect either G-protein-rhodopsin binding or G-protein activation. Antibody 4A blocks both of these processes in a dose-dependent manner. Antibody concentration is expressed as a molar ratio to G-protein present in the ROS. At a molar ratio of 0.5 antibody to 1 G-protein, both Gprotein binding to rhodopsin and G-protein activation are partially blocked, and this blockade is almost complete at a 1:l ratio of antibody 4A to G-protein. At an antibody excess over G-protein of 51, extra-metu-rhodopsin I1 formation is completely blocked, and no G-protein activation occurs. Fig.  3C summarizes the effect of a series of antibodies on extrametu-rhodopsin I1 formation. Antibodies 7A, 7B, 7C, and 7D also block extra-metu-rhodopsin I1 formation at a 1:1 molar ratio. To test whether G-protein cross-linking by the bivalent antibody is important for its blocking action, Fab fragments of antibody 4A were tested in this assay. At an approximately 1:l molar ratio of Fab 4A to G-protein, extra-metu-rhodopsin I1 formation was considerably blocked (data not shown). This agrees with the G-protein solubilization data shown below.
The extra-metu-rhodopsin I1 signal measures the lightdependent tight binding of the G-protein to rhodopsin. To test whether antibody 4A recognizes this site or a site on Gprotein that is involved with the dark binding of G-protein to ROS membranes, we determined the effect of antibodies on the localization of G-protein to the membrane or cytoplasm in the dark or light. Rod outer segments were preincubated in Ringer's solution or with antibodies in the dark, and then a fractionation experiment was performed. ROS membranes were removed by centrifugation, and the amount of G-protein in the supernatant was assessed by SDS gel electrophoresis and densitometry (Fig. 4). Under control conditions at normal ionic strength in the dark, G-protein is bound to membranes, and very little is found in the supernatant (Fig. 4 4 , lane I ) . Preincubation with control IgG or antibody 4H in the dark did not significantly affect the amount of G-protein in the supernatant (lanes 2 and 7), whereas preincubation with antibody 4A caused the release of G-protein into the supernatant (lane 4 ) . The amount of G-protein in the supernatant was increased 6-fold compared with controls (Fig. 4B). The a and P-y subunits of G-protein were found in equal amounts in the supernatant (numbers below the burs in Fig. 4B). Thus, the antibody interferes with a site on G-protein which is involved with interaction of G-protein with ROS membranes in the dark. Fab fragments of 4A also elute G-protein from the membrane stoichiometrically (lune 9). Antibodies 7A, 7B, 7C, and 7D also eluted G-protein (Y and P subunits from the membrane (data not shown).
The G-protein released from the membrane by antibody 4A is not bound by light-activated rhodopsin. Fig. 4A (lane 5 ) shows that treatment of ROS membranes with light bleaching 10% rhodopsin, which would normally cause all G-proteins to be tightly bound to rhodopsin, does not cause a significant decrease in the soluble G-protein after treatment with antibody 4A. There is a parallel absence of G-protein on the bleached membranes (data not shown). There is no significant effect of light on G-protein solubility in the presence of control IgG or antibody 4H (lanes 3 and 8). To test whether the A. FIG. 4. Antibody 4A solubilizes G-protein from the disc membrane under isosmotic conditions in the dark, whereas antibody 4H or control IgG had no effect on G-protein localization. Frog ROS membranes were incubated in the dark for 30 min with Ringer's. A 2:l excess of control IgG, antibody 4A, Fab 4A or 4H, and then membranes were separated from soluble proteins by centrifugation. The supernatant was run on a 12.5% SDS-polyacrylamide gel and stained with Coomassie Blue, and the density of protein stain in the G, and GB bands was measured densitometrically; the relative densities of the Goand GBbands are plotted in B. The numbers below the bars represent the amount of G , as a percent of GeB Migration of molecular weight standards is indicated on the left side. HC and LC, are antibody heavy chain and light chain, respectively; HC', heavy chain dimer; 48K, ROS soluble 48-kDa protein, S-antigen. Lunes 1-8 are from one experiment, and lane 9 is from a different experiment.
antibody 4A-induced blockade of G-protein-rhodopsin interaction could be reversed by activating more rhodopsin, a brighter light stimulus was given. Light bleaching 40% of the rhodopsin was effective in causing a slight decrease in soluble G-protein, but most G-protein (85%) remained in the supernatant even with large bleaches. The strong antibody blockade of G-protein-rhodopsin interaction is borne out by measurements of extra-meta-rhodopsin I1 formation showing that even after bleaching 20% rhodopsin, the antibody completely blocks extra-meta-rhodopsin I1 resulting from G-protein-rhodopsin interaction (data not shown). It appears that antibody 4A blocks interaction of G-protein with rhodopsin by interfering with its dark binding site to ROS membranes. To test whether this binding site involves attachment to rhodopsin or to phospholipids, we tested the effect of antibody 4A on extra-meta-rhodopsin I1 formation in detergent-solubilized rhodopsin. Extra-meta-rhodopsin I1 resultant from G-protein-rhodopsin interaction is formed after a flash of light bleaching 16% rhodopsin, even when rhodopsin is solubilized in the biphenyl detergent ( Fig. 5A and B).' Antibody 4A blocks extra-meta-rhodopsin I1 formation under these conditions as well (Fig. 5C). Thus, the

FIG. 5. Extra-meta-rhodopsin I1 is blocked by antibody 4 A in detergent-solubilized rhodopsin.
Bovine ROS membranes were dissolved in 600 p~ biphenyl detergent (critical micellar concentration,* 190 p M ) to a rhodopsin concentration of 1.35 pM. ROS low ionic strength extract containing approximately 0.75 pM G-protein was added back to solubilized rhodopsin in the absence (A and B ) or presence ( C ) of 5:1 molar excess of antibody 4A. Then the extrameta-rhodopsin I1 signal was measured (at pH 7.5 and 8 "C) after a flash of light bleaching approximately 16% rhodopsin. A, first flash without mAb 4A; B, fourth flash of the same sample normalized to the same absolute amount of photolyzed rhodopsin as in A (1/(1 -0.16)3 = 1.69). The increased noise level is due to this normalization. C, first flash with mAb 4A. The intermediate level of meta-rhodopsin I1 formed in the presence of antibody 4A in response to a flash of light is roughly equal to the amount of meta-rhodopsin I1 generated without the influence of G-protein ( B , Emeis et al., 1982). antibody blockade of G-protein-rhodopsin interaction does not appear to require membrane phospholipids.
An alternative hypothesis is that this antigenic site is a region of interaction between the a and subunits. Fung (1983) showed that the subunits of the G-protein must be together in an a& complex for effective interaction with the ROS membrane. Interruption of interaction between the a and subunits would result in all the functional effects shown in Figs. 1-5, resulting from an inability of the individual subunits to bind to the membrane. If antibody 4A interrupted cu-or interaction, one would expect it would immunoprecipitate only the a subunit, with the subunit left behind in the supernatant. Immunoprecipitation experiments attempting to test this hypothesis are shown in Fig. 6. Purified G-protein radiolabeled with ['2sII]iodonaphthyl azide was incubated in the presence of antibodies 4A, 4H, or I@, and then formalin-fixed S. aureus cells were used to precipitate G-protein-antibody complexes. The first lane shows the starting amount of protein used for immunoprecipitation. In the absence of antibody (second lane) or in the presence of control IgG (third lane), no G-protein was immunoprecipitated. Both antibodies 4A and 4H immunoprecipitated the GnR7 holoprotein as judged by the presence of similar amounts of the a and p subunits in the gel. This suggests that mAb 4A does not disrupt the interaction between the a and subunits. Similar amounts of a and / 3 subunits were immunoprecipitated by antibody 4A, which blocks G-protein binding to rhodopsin, and antibody 4H, which does not. Antibodies 7A, Antibody Blockade of G-protein-Rhodopsin Interaction  FIG. 6. Anti-G, antibodies 4A and 4H immunoprecipitate both the a and B subunits of the G-protein. Purified bovine Gprotein previously labeled with ['2sI]iodonaphthyl azide was incubated with nonspecific IgG, mAb 4A, mAb 4H, or buffer (-). Immunocomplexes were precipitated with fixed S. aureus cells as described under "Materials and Methods." Precipitated proteins were separated on 12.5% SDS-polyacrylamide gel electrophoresis, and autoradiographed. The first lone represents the total labeled protein used in the immunoprecipitation. Immunoprecipitation of G-protein was not quantitative due to the relatively low affinity of these antibodies for G-protein (Witt et al., 1984). The smaller amount of G-protein immunoprecipitated by mAb 4H is due to a smaller amount of mAb 4H relative to mAb 4A present during the immunoprecipitation as assessed by Coomassie Blue staining of the gel. 7B, 7C, and 7D also immunoprecipitated the holoprotein (data not shown).

DISCUSSION
We show that stoichiometric interaction between G, and monoclonal antibody 4A blocks the light-dependent activation of the G-protein and subsequent activation of the cGMP phosphodiesterase. Antibody 4A and four other antibodies also interfere with light-induced G-protein interaction with rhodopsin as measured by the light-scattering binding signal and meta-rhodopsin I1 stabilization and G-protein binding to the ROS membrane in the dark. The light-induced interaction between G-protein and rhodopsin takes place even in detergent-solubilized rhodopsin depleted of membrane phospholipids, and mAb 4A also blocks this interaction. Two monoclonal antibodies, 4H and 4C, do not block these events.
The blocking antibodies described in this report could have their inhibitory effect by several possible mechanisms. They could bind to the site of G-protein interaction with rhodopsin or close enough to this site that the G-protein-rhodopsin binding is sterically hindered. Alternatively, they could bind to a site on G-protein normally involved in a conformational change and block the change. Finally, they could bind to a site normally involved in binding to Gg. Fung (1983) showed that Goy is important for G, interaction with rhodopsin. Thus, interruption of this site of interaction could be expected to remove GnBr from the membrane and therefore effectively block G-protein interaction with rhodopsin. However, immunoprecipitation experiments show that antibody 4A can effectively immunoprecipitate the a8-y complex (Fig. 6). If antibody 4A interrupted a-py interaction, one would expect it would immunoprecipitate only (Y subunit with P-y subunit left behind in the supernatant. Antibodies that bind to sites of conformational changes in proteins have been reported to block protein activation by blocking conformational changes (Djavadi-Ohaniance et al., 1984). This mechanism of action seems unlikely in this case because mAb 4A as well as other blocking antibodies remove G-protein from the membrane in the dark before any light activation and resultant conformational change occur. Antibody-induced removal of the Gprotein from the membrane does not induce the activated conformation since mAb 4A treatment alone does not activate phosphodiesterase in the dark . Therefore, we favor the first mechanism, that the epitopes of mAbs 4A, 7A, 7B, 7C, and 7D are the site of G-proteinrhodopsin interaction or close enough to that site to sterically inhibit that interaction. Since Fab fragments of mAb 4A also solubilize G-protein and block its interaction with rhodopsin, the antigenic site should be relatively close to the G-proteinrhodopsin interaction site (a Fab interaction site is approximately 20 A in diameter; Amit et al., 1986). The fact that Fab fragments of antibody 4A are as effective as the bivalent antibody also indicates that cross-linking of G-protein is not necessary for solubilizing G-protein from the membrane.
The finding that G-protein cannot bind to pure phospholipid vesicles but binds well to phospholipid vesicles containing pure rhodopsin, both in the dark and the light (Fung, 1983), suggests that the major site of interaction of G-protein with the membrane is rhodopsin and not an interaction with phospholipids, although phospholipids may play some role. mAb A4 removes G-protein from the ROS membrane in the dark (Fig. 4) and thus disrupts the major site of G-protein binding to the membrane. mAb 4A can also block the lightactivated G-rhodopsin binding measured by extra-meta-rhodopsin I1 formation in reconstituted membranes (Fig. 3) as well as in purified rhodopsin in the absence of phospholipids (Fig. 5), suggesting that mAb 4A directly blocks the interaction between G-protein and rhodopsin. These findings are consistent with the notion that in the dark, the membranebinding site for G-protein is rhodopsin and that the mAb 4A antigenic site is at or near the site of G-protein-rhodopsin interaction.
The affinity of the putative dark G-protein-rhodopsin site must be relatively low because Kiihn (1984) showed that in the dark, G-protein does not bind to columns of purified rhodopsin bound to concanavalin A-Sepharose, but that in the light, it does bind to the column. The dissociation constant has been estimated as KD = lo-' to M (depending on the preparation) from membrane mixing experiments (Liebman and Sitaramayya, 1984) and M from light scattering measurements." The dark membrane-binding site saturates a t 25% rhodopsin, i.e. 2.5-fold more G-protein can bind than is bound in native membranes (Liebman and Sitaramayya, 1984). The only protein present a t such abundance in washed membranes is rhodopsin (Hamm and Bownds, 1986). There is known to be a considerably increased affinity of G-protein for bleached rhodopsin (KO = 2 X lo-' to M, depending on GDP binding (Bennett and Dupont, 1985)). Whether the dark binding site is similar or different from the tight binding site between photoactivated rhodopsin and G-protein is not known; however, it could be the same site whose affinity increases upon the light-induced conformational change in rhodopsin or a neighboring region. A reactive cysteine is located in the vicinity of the light binding site of G, for rhodopsin (Hofmann and Reichert, 1985). Modification of one S H group a t G, by N-ethylmaleimide does not disturb dark binding but blocks light binding completely; however, the smaller cyanide group does not block light binding.
If this is so, monoclonal antibodies against rhodopsin a t the site of interaction with G-protein should disrupt G-protein binding in the dark. Preliminary experiments4 have been done using a monoclonal antibody against the COOH-terminal region of rhodopsin, antibody 1D4. G-protein is indeed eluted from the membrane by preincubation with this antibody, although the release is not q~antitative.~ Proteolytic digest studies have shown that the first cytoplasmic loop near the COOH-terminus of rhodopsin is an important region of G-Schleicher, A., and Hofmann, K. P. (1987)  protein binding on rhodopsin (Kuhn and Hargrave, 1981).5 Monoclonal antibodies are currently being generated to this region, and their effects on G-protein binding will be tested. The question arises whether such a G-protein-rhodopsin interaction in the dark would result in a relatively long-lived preformed complex between G-protein and rhodopsin so that 10% of the rhodopsin might be activated more rapidly than freely diffusing components. There is evidence in other systems of a complex between receptors and GTP-binding proteins. In a purification of the a,-adrenergic receptor, the inhibitory guanyl nucleotide regulatory protein, Gi, copurified through several steps with the receptor (Cerione et ul., 1986). In hormone-sensitive cells, which contain only a few thousand hormone receptors, a preformed complex of receptor and Gprotein could have a major effect on the kinetics of hormone activation of intracellular events. In the photoreceptor membrane, the implications of such a preformed complex on the kinetics of G-protein activation are not clear. In the typical operating range of this receptor (-1 photoactivated rhodopsin/disc membrane), the probability of hitting a precomplexed rhodopsin would only be 10%. On the other hand, stimulation of one of the preformed complexes would activate only one G-protein with the reaction mode of the complex. A sufficiently fast exchange of photoactivated rhodopsin and Gprotein therefore appears to be required for efficient excitation of the photoreceptor.
The transient formation and then decay of extra-meturhodopsin I1 in the presence of 20 pM GppNHp (Fig. 3A, third curve) provide a kinetic measure of the guanyl nucleotide exchange rate, leading to the release of G-protein from rhodopsin, with the consequent decay of metu-rhodopsin I1 . GppNHp exchanges into the guanyl nucleotidebinding site more slowly than does GTP; therefore, under these conditions, the rate is slow enough to be easily measurable. If the G-protein diffused to an activated rhodopsin, the bound antibody could be expected to decrease its diffusion rate, with a consequent slowing of activation. It is striking that antibodies 4H and 4C bind to G-protein but do not change the kinetics of its activation. Under the conditions of these experiments, the G-protein can carry a large load (IgG, M , 150,000) without a visible effect on its activation kinetics. However, the actual G-protein-rhodopsin binding reaction is kinetically buried in the rate-limiting rnetu-rhodopsin I1 formation under the conditions of the measurements (Emeis et ul., 1982). It is of interest to know whether this antigenic site is conserved in other GTP-binding proteins. Dot-blotting experiments show that all the antibodies described here do crossreact to some degree with purified G,, Gi, and Go, although cross-reactivity is substantially lower than with photoreceptor G-protein. Functional experiments also show that antibodies that block photoreceptor G-protein activation also block stimulation and inhibition of adenylate cyclase by hormones and guanyl nucleotides in pineal, brain, and S49 cyc-mernbranes.'j Thus, the site of interaction between receptors and GTPbinding proteins may be conserved.