INTRODUCTION

Heterotrimeric G proteins communicate signals from activated receptors to effectors (1). Both  and subunits of G proteins regulate effectors (2, 3). The regions of G involved in interactions with effectors have been identified (4, 5). Identification of effector interaction domains on G has been more difficult, and it remains unclear if both the  and  subunits participate in regulating effectors. Most combinations of the five very similar  subunits and the six or more divergent  subunits activate effectors (6). Post-translational modification of the  subunit is required for effector regulation (7). Thus one could hypothesize that the  subunit is the key component in effector interactions. For AC2,¹ we had recently identified the region 956-982 as being important in G stimulation. A peptide (QEHA peptide) encoding this region blocks G regulation of several effectors (8), suggesting that different effectors interact with at least one common region in  or  subunits. To explore this possibility further, we have cross-linked the QEHA peptide to G. We find that the QEHA peptide selectively cross-links to  subunits of the free complex but not  subunits of an heterotrimer. Using the crystal structure of free subunits (9) and of the heterotrimer (10) and molecular modeling programs, we have developed a model of the QEHA peptide-G protein complex.

EXPERIMENTAL PROCEDURES

Peptides were synthesized on an Applied Biosystems peptide synthesizer (431A) and purified by high pressure liquid chromatography on an acetonitrile gradient. The purified peptides were lyophilized and stored at 20 °C. The peptides were dissolved in distilled water to a concentration of 1 mM. SIAB from Pierce was freshly dissolved in dimethyl sulfoxide (10 mM) prior to use. ECL was from Amersham Corp. All other chemicals were the highest chemical grade available. Antibodies to the C terminus of 2 (11) were the kind gift of Dr. N. Gautam. The anti- antibody (BC1) is against an epitope to the C-terminal 10-amino acid region common to all  subunits. Heterotrimeric G_o and subunits were prepared and stored as described (12).

G protein (0.3 µM) and peptides (0.5-50 µM) were incubated on ice (40 °C) for 60 min in a solution containing 10 mM of MgCl₂ and 10 mM Hepes, pH 7.4, in a volume of 50-100 µl. The incubations were then continued at room temperature (22-24 °C) for 30 min. The cross-linker was added to achieve a final concentration of 0.2 mM. When heterotrimeric G protein was used, G₀ (0.1 µM) was incubated with 1 mM GDP and 1 mM EDTA or 10 µM GTPS and 10 mM MgCl₂ in 10 mM NaHepes, pH 7.4, for 15 min at room temperature. Peptides were then added, and the incubations were continued on ice for 60 and 30 min at room temperature before the addition of the cross-linker. The cross-linking reaction was for 30 min at 22-24 °C. Reactions were quenched with 10 mM Tris buffer, pH 8.8, and 10 mM of 2-mercaptoethanol. The proteins were electrophoretically resolved on 10% SDS-polyacrylamide gels or on 14% SDS gels in the tricine buffer system (13) and transferred to Hydrobound-C nitrocellulose membranes (Amersham Corp.). An antibody against G subunit (BC1) or an anti- antibody against 2 were used as probes. The ECL was used to visualize the protein bands. For image storage and printing, the films were scanned with the image analyzer program Photolook. The images were exported to the program Canvas for labeling, storage, and printing. All experiments were repeated at least twice with similar results. Typical experiments are shown.

The co-ordinates for G and the heterotrimeric G protein were obtained from Drs. J. Sondek and P. Sigler. The structures were visualized using the program LOOK (Molecular Application Group, Palo Alto, CA). The residues on G that are in contact with G were identified within LOOK. A prediction of the secondary structure of the QEHA peptide was obtained with the program PHD (14) and was used to construct a three-dimensional model of the peptide using the program QUANTA (Molecular Simulations, Waltham, MA). The electrostatic surfaces of the QEHA peptide and the G subunit were visualized with GRASP (15). This information was then used in QUANTA to dock the peptide onto the G subunit by minimizing the distance between regions exhibiting electrostatic complementarity. The structure of the QEHA peptide docked to the G subunit was subjected to 1000 steps of energy minimization (conjugate gradient) followed by a short run of molecular dynamics simulations with the DISCOVER package within INSIGHT (Biosym Technologies, San Diego, CA). QEHA peptide and the side chains of residues 75-165 of G were relaxed in the frame of a fixed G protein backbone. The most favorable structure for the docked peptide interacting with G was selected for presentation.

RESULTS AND DISCUSSION

Purified G was incubated with and without QEHA peptide and the cross-linker SIAB. The mixtures were resolved by SDS-polyacrylamide gel electrophoresis and analyzed by immunoblotting with either anti-G or anti-G (₂) antibodies. For blotting with the anti-G antibody, the samples were resolved on 10% polyacrylamide gels. In the absence of cross-linker, G was at the expected position of 35 kDa (Fig. 1A, lanes 1 and 2). With cross-linker in the absence of QEHA peptide, an additional band was observed in the 48-50-kDa range, presumably a complex of the  and  subunits. Bands were also observed in the 80-kDa range that might be dimers of G subunits (Fig. 1A, lane 3). When G was incubated with the QEHA peptide and then with the cross-linker, a prominent band was observed at 36-38 kDa. We conclude that this additional band arises from cross-linking of the QEHA peptide to G subunits. Several homo- and hetero-bifunctional cross-linking agents were tested using a similar experimental paradigm. It was found that SIAB gave the most reproducible cross-linking. Because there are no cysteines in the QEHA peptide, the use of SIAB would allow the identification of cysteine residues on G subunits important for cross-linking. Further experiments were carried out with SIAB.

To study cross-linking of QEHA peptide to G subunits, samples were resolved on 14% acrylamide gels with tricine buffer. ₂ antibodies were chosen to visualize the G subunits, because this isoform is the most abundant one in bovine brain G preparations.² Without cross-linker, the ₂ band was seen (Fig. 1B, lanes 1 and 2), but no bands under 15 kDa were observed when the G subunits were incubated with cross-linker (Fig. 1B, lane 3).  immunoreactive bands in the 50-kDa range were observed when G subunits alone were incubated with cross-linker, in agreement with the previously known ability of  and  subunits to be cross-linked. Incubation with the QEHA peptide did not significantly alter this profile (Fig. 1B, lane 4). Although almost all of the  subunits disappear from the 8-10-kDa region upon incubation with cross-linker, not all of the G subunits are internally cross-linked, because a substantial portion of the  subunit runs at 35 kDa even after treatment with cross-linker (Fig. 1A, lanes 3 and 4). This suggests that after treatment with cross-linker, the  subunits may not recognized by the antibodies. These complexities indicate that chemical cross-linking with immunoblotting is not useful to study G interactions with effector peptides. So, we focused on the interaction with G subunits.

Cross-linking of the various peptides to G subunits of bovine brain G.

To test if the cross-linking of QEHA peptide was specific and dependent on peptide concentration, we examined the cross-linking of three peptides to G subunits. These were the QEHA peptide, the SKEE peptide (the 27-mer from the analogous region of AC3), and peptide C (an 11-amino acid peptide encoding region 568-578 of AC1). Only the QEHA peptide was extensively cross-linked (Fig. 2A). The extent of QEHA peptide cross-linking to G was proportional to the amount of peptide added and was most extensive at concentrations effective in blocking G regulation of effectors (Fig. 2B and Ref. 8).

Cross-linking of QEHA and SKEE peptides to G subunits of G_o in the unactivated and activated states.

To establish whether QEHA peptide interaction with G is biologically relevant, we tested the ability to cross-link QEHA peptide to G when the complex was free and when the complex was part of the heterotrimer. Only free G regulates effectors, and this regulation can be inhibited by excess G (16, 17), suggesting that the effector interaction domains of G are occluded by interactions with G. Hence, if the interaction of QEHA with G is biologically relevant, then cross-linking should be observed only when G is dissociated from G. Cross-linking of QEHA peptide to G was studied when heterotrimeric G_o was treated with GDP and EDTA to maintain the protein in the heterotrimeric state or with GTPS and Mg²⁺ to dissociate  from subunits prior to exposure to the QEHA peptide and cross-linker. The SKEE peptide was used as control. After the cross-linking reaction the samples were electrophoretically resolved and then blotted with the anti- antibody. G_o alone or G_o pretreated with GTPS and Mg²⁺ prior to exposure to cross-linker did not yield any additional bands in the 35-40-kDa region (Fig. 3, lanes 1 and 2). When QEHA peptide was added to uninactivated G_o, only a trace band was observed above the 35-kDa  band (Fig. 3, lane 3). However, extensive cross-linking was observed when the QEHA peptide was added to G_o pretreated with GTPS and Mg²⁺ (Fig. 3, lane 4). No cross-linking of the SKEE peptide with unactivated G_o was observed (Fig. 3, lane 5), and very little cross-linking with activated G_o was seen (Fig. 3, lane 6). These results indicate that the QEHA peptide can be cross-linked to the  subunit only when it is part of the free G complex but not in the heterotrimer.

Because the QEHA peptide comprises a sequence of AC2, which interacts directly with G subunits (18), we had proposed that the QEHA peptide would interact with G and prevent its interactions with effectors. The experiments in Figs. 1, 2, 3 provide evidence for this mechanism. The experiment in Fig. 3 also indicates that the cross-linking site for QEHA (i.e. the reactive cysteine) and possibly the region of interaction on G are inaccessible when the G subunit is interacting with G subunit. The crystal structures of free G subunits and the heterotrimer (9, 10) were used to identify cysteines in the  subunit that are hidden in the heterotrimer but are exposed upon dissociation of the  subunit. A space-filling molecular model of the  subunit is shown in Fig. 4A. The atoms of the  subunit identified by Lambright et al. (10) to be in contact with the  subunit in the heterotrimer structure, are shown in purple. Visual inspection indicated that the SH group of Cys-204 is fully exposed in free . The ready accessibility of the SH group of Cys-204 in contrast to buried SH side chains of all of the other Cys residues makes it a likely site of attachment. Support for the notion that such a reactive Cys in free G came from two experiments: first, the observation that no background cross-linking with the control peptide (SKEE) was observed when the intact heterotrimer was used (Fig. 3, lane 5); and second, the observation that following their reaction with SIAB, QEHA, and other unrelated control peptides exhibited equivalent cross-linking (Fig. 4B, lanes A2, QEHA2, and C2). The lack of specificity following pretreatment with SIAB is in marked contrast to the specific cross-linking of the QEHA peptide to G subunits, when the peptides are first mixed with the G complex and then exposed to the cross-linker SIAB (Fig. 4B, lanes A1, QEHA1, and C1). When the peptides are first reacted with SIAB they are essentially converted to Cys-reactive agents with the active iodo-acetamide group. In such a system, all peptides have an equal probability of reacting with the exposed SH group and hence the observed lack of specificity. In contrast, without SIAB pretreatment, the cross-linker has to react with the peptide and G simultaneously for the cross-link to be successful. This occurs only with QEHA peptide, presumably because the QEHA peptide is bound to G, so that the reactive  amino group is at the correct distance from the reactive Cys in G. If the control peptides do not bind to G, their  amino groups may not be correctly positioned for cross-linking.

Such specific interaction of the QEHA peptide with G were seen when the pellets of the biotinylated G avidin beads were examined for interactions between G and G. The effect of the QEHA and SKEE peptides on interactions between G and G are shown in Fig. 5A. This experiment is similar to that shown in Fig. 3 of Chen et al. (8). In Fig. 5A the area of interest is the Coomassie Blue-stained region below the dye front. In the absence of any added peptide, no staining was observed (Fig. 5A, lane 1). Very little staining was seen when 100 µM SKEE peptide was used (Fig. 5A, lane 2). In contrast, considerably greater staining was observed when QEHA peptide was used, indicating that the QEHA peptide was retained by the biotinylated G avidin bead complex irrespective of the presence of G subunit (Fig. 5A, lanes 3 and 4). Although these observations are qualitative in nature and do not allow us to measure directly the binding of QEHA peptide to G complex, the results indicate that the QEHA peptide can selectively interact with G subunits. Taken together the experiments in Figs. 1, 2, 4B, and 5A suggest that it is the bound QEHA peptide that is specifically cross-linked. From the results in Fig. 3 we conclude that the interaction of the QEHA peptide with G involves regions that are at least in part occluded when G is in contact with the G and that the N-terminal of the QEHA peptide, which reacts the hydroxy-succinimidyl group of the cross-linker, would be positioned within the reactive distance of 12-14 Å from Cys-204. With these constraints we attempted to dock the QEHA peptide on to the  subunit.

We first obtained a predicted structure for the QEHA peptide and used it to calculate the energy minimized structure that served in docking to G (see ``Experimental Procedures''). Assuming the electrostatic long range interactions to be the likely guiding forces in the initial docking of the peptide to its binding site, the electrostatic potential on the molecular surfaces of QEHA peptide and G subunit were calculated to identify mutually complementary patches of positive and negative potentials. The electrostatic complementarity served to guide the docking of the QEHA peptide onto the G subunit using molecular graphics routines (see ``Experimental Procedures''). Once docked onto the surface of the G subunit, the structure of the QEHA peptide was relaxed with molecular dynamics simulations and a second round of minimization. The result of the docking procedure is shown in Fig. 5B, where the N-terminal Gln of the peptide is seen to be 12.4 Å from Cys-204. From this model it appears that the QEHA peptide is in contact with a surface of the G subunit encoded by residues 75-165.

We show that the QEHA peptide can be cross-linked to  subunits of the free G complex but not the heterotrimer, suggesting that QEHA peptide interacts with a region(s) of G normally occluded by G. This conclusion agrees with previous observations that only free G and not the heterotrimer regulate effectors (16, 17). However, the QEHA peptide does not block G interactions with G, although it blocks G regulation of several effectors (8). Thus, the experiment in Fig. 5A is seemingly at odds with the experiment in Fig. 3 and with the observations that the C terminus domain of ARK blocks association of G with G (19). These differences are explained by the relative affinities of G and effectors for G and by partial overlaps between regions of G involved in G and effector interactions. If GDP-G subunits have a much higher affinity for G than the effector, the QEHA peptide in the useful concentration range would not be able to compete with G for G binding. This may account for the observations in Fig. 5A. Indeed, such a difference in the affinity of G and effectors for G may be necessary if there is to be GTPase-dependent termination of signaling when the signal is communicated by G.

The biochemical cross-linking and the molecular modeling provide insight into the mode of interaction between QEHA peptide and G. The molecular model allows us to propose the region 75-165 of G as likely to be involved in effector interactions. Several residues within this region, such as Lys-78, Trp-99, and Asn-119, which are in contact with G (10) and hence would be unavailable for effector interactions in the heterotrimeric state, are also predicted to interact with the QEHA peptide. The proposed QEHA binding site (i.e. the effector interaction region) of G is not identical to the G binding site, but it is likely that the sites overlap partially. Hence, the binding of the relatively small QEHA peptide may not eliminate interactions with G, but that binding of G or a large fragment of ARK would preclude G from participating in other interactions due to steric hindrance.

Using the yeast two-hybrid system, Yan and Guatam have identified the first 100 amino acids of G as being involved in interactions with a probe encoding the region 956-982 of AC2 (20). Thus, two independent approaches now identify some common regions on G as being involved in effector interactions. It should be emphasized that the QEHA interaction regions on G identified here from the modeling are not a definitive definition of the exact effector interaction domains on G subunits. Rather, these specific predictions of G regions that are likely to be involved in effector interactions should be subjected to detailed experimental verification.