Modular Design of Gβ as the Basis for Reversible Specificity in Effector Stimulation*

The G protein Gβγ subunit complex stimulates effectors by direct interactions utilizing extensive Gβ regions over the surface of its propeller structure that faces the Gα subunit. Our previous experiments have shown the resolved functions of signal transfer and general binding for Gβ regions involved in stimulation of the effector phospholipase C-β2, PLC-β2, within the region Gβ-(86–135), which comprises three β strands arranged in a structurally contiguous fashion (Buck, E., Li, J., Chen, Y., Weng, G., Sacarlata, S., and Iyengar, R. (1999) Science283, 1332–1335). This raises an important question as to why mutagenesis studies indicate that an extensive set of sites all over the Gβ propeller structure and outside the 86–135 region are involved in Gβ regulation of PLC-β2. Using peptides to define functions of these Gβ regions, we find that Gβ signaling to PLC-β2 relies on a collection of modular signal transfer and general binding units, each with lower apparent affinity relative to Gβγ-PLC interactions. Gβ-(42–54) functions as a signal transfer region, Gβ-(228–249) and Gβ-(321–340) function in general binding, and Gβ-(64–84) and Gβ-(300–313) seem to play a structural role rather than a direct contact with the effector. A substitution within the Gβ-(42–54) signal transfer region that increases the K act of this peptide for PLC-β2 is accompanied by an increase in the observed maximal extent of signal transfer. We conclude that the lower K act for individual signal transfer regions may result in a decrease in the maximal effect of signal transfer. The spatial resolution of the signal transfer and general binding regions over a wide surface of Gβ allow geometrical constraints to achieve specificity even with relatively low affinity interactions.

Many receptors transmit their signals through heterotrimeric guanine nucleotide-binding proteins, G proteins. Both the ␣ and ␤␥ subunits of the G protein are capable of regulating effectors. Effectors for G␤␥ include classical second messengerproducing enzymes such as adenylyl cyclase and phospholipase C-␤2, PLC-␤2. 1 Signal transfer from the G protein subunits to these effectors occurs through direct protein-protein interac-tions (1). A noteworthy feature of G␤␥ in regulation of effectors is that it occurs with a much lower affinity (2) compared with the G␣ subunit (3). Nevertheless, the interaction between G␤␥ and effectors is quite specific. For instance, it stimulates PLC-␤2, but not PLC-␤4 (4). These observations lead to a central question in G protein signaling of how specificity is achieved even when interactions are of low affinity. This study addresses this question for G␤␥ regulation of PLC-␤2.
Molecular biological techniques, such as yeast two-hybrid screening and site-directed mutagenesis have been extensively used to map the regions of G␤ involved in interactions with effectors including PLC-␤2 (5)(6)(7)(8)(9). The sites mapped by these approaches cover a large surface of G␤ on the side that interacts with G␣. These interaction sites are spread through all seven blades in the structure of the G␤ subunit.
As identified by site-directed mutagenesis, regions of G␤ that mediate regulation of PLC-␤2 could serve three distinct functions. Some regions may make direct physical contact with the effector protein and transfer signals to it. Other regions, which form direct contact sites, might contribute to overall binding affinity but not be involved directly in signal transfer. Still other regions may be important from a structural standpoint for G␤ while not making any direct physical contacts with this effector protein. Such a region could be responsible for the global folding of the protein or could serve to stabilize local secondary structure or side chain alignment important for mediating an interaction with PLC-␤2, even though the region by itself makes no direct contact with this effector. Mutagenesis studies are not useful for ascribing such functions to various regions. However, peptides that encode these various regions may be used for this purpose. In previous work we demonstrated that the functions of two sites in a spatially contiguous region on G␤ (G␤-(86 -135)), important for regulation of PLC-␤2, could be resolved and separately identified as signal transfer and general binding regions (10). This finding leads to the question of why the mechanism of G␤␥ stimulation of PLC-␤2 relies extensively on sites outside the G␤-(86 -105) region to coordinate stimulation of PLC-␤2. Here we use the peptide approach to define the functions of the various regions of G␤ outside G␤-(86 -135) that mutagenesis studies indicate are involved in PLC-␤2 function. We find that G␤ utilizes a modular design of multiple signal transfer and general binding domains to allow for stimulation of PLC-␤2. These findings suggest a mechanism that ensures that PLC-␤2 regulation is both specific and reversible. D246S were synthesized on an Applied Biosystems peptide synthesizer (Model 431A). Peptides were lyophilized and stored at Ϫ20°C. Peptides G␤-(42-54), G␤-(42-54)-R48A, G␤-(42-54)-R49A, and G␤-(46 -54) were synthesized at the Tufts University Core Facility. These peptides were all purified by HPLC. When needed, peptides were dissolved in HED buffer (10 mM Hepes (pH 7.0), 1 mM EDTA (pH 8.0), 1 mM dithiothreitol) at a stock concentration of 3 mM. All peptides were prepared fresh at the time of the experiment. The pH of each peptide in solution was tested to ensure that the pH of the peptide stocks was pH 7.0. The identity and purity of peptides were verified by mass spectrometry and HPLC. We had found with the G␤-(42-54) region peptides that extended storage even at Ϫ20°C resulted in dimerization of the peptide. Therefore, to ensure accurate identification of the peptides, their effect was tested within 1 week of mass spectrometry analysis. For controls we used substituted peptides. Substitutions were made at the same residues that had been shown by mutagenesis to affect G␤␥ regulation of PLC-␤2. This allowed for direct comparison of our experiments with the mutagenesis studies.
G␤␥ Expression-Recombinant G␤ 1 ␥ 2 -His 6 was expressed in High 5 insect cells by co-infection of recombinant baculovirus. Three to 4 days postinfection, the cells were lysed by decompression in a Parr bomb after equilibration at 600 p.s.i. The lysate was centrifuged, and the cytosolic fraction was collected. The G␤ 1 ␥ 2 was then purified on a Ni 2ϩ column as previously described (5). The K act for this G␤␥ on PLC-␤2 is 300 nM. When indicated a final concentration of 400 nM of this purified recombinant G␤␥ was used in the assay reaction.
PLC-␤2 Expression-Recombinant human PLC-␤2 was expressed in High 5 insect cells by infection with recombinant baculovirus. Cells were grown in suspension to a density of ϳ1.5 ϫ 10 6 cells/ml before infection with the virus. Three days after infection, the cells were harvested. The cells were spun at 1,000 rpm for 10 min, the media was then decanted, and the cell pellet was washed once in cold phosphatebuffered saline. The washed cell pellet was resuspended in lysis buffer (20 mM Hepes, pH 7.0, 5 mM EDTA, pH 8.0, 5 mM EGTA, pH 8.0, 2.5 mM KCl, 1 mM dithiothreitol, 10 g/ml aprotinin, 10 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride). The volume of lysis buffer was 20 ml/ 500 ml suspension cell growth. The cells were then lysed by decompression in a Parr bomb after equilibration at 600 p.s.i. The lysate was ultracentrifuged for 1 h at 35,000 rpm at 4°C. The supernatant was decanted and stored in aliquots at Ϫ70°C.
Measurement of PLC-␤2 Activity-10 g of protein from the cytosolic fraction of H5 cells expressing PLC-␤2 was used/100 l of reaction volume. Phospholipid substrate is a mixture of [ 3 H]PIP 2 and unlabeled phospholipids. Unlabeled phospholipids (Sigma-P-6023) were crude lipids from bovine brain. The total diphosphoinositide and triphosphoinositide content was 20 -40%. The remainder was a mixture of phosphatidylinositol and phosphatidylserine. Phospholipids were stored in chloroform stocks. When needed an aliquot of phospholipids was dried under nitrogen and then resuspended and sonicated in 10 mM Hepes (pH 7.0) to form lipid micelles. A total of 0.01 Ci of [ 3 H]PIP 2 , corresponding to 3000 -7000 cpm, and 5 g of unlabeled mixed phospholipids was used per reaction. The PLC assay was done as previously described (11). Briefly, substrate, PLC-␤2 (10 g of protein cytosolic fraction), peptide, and G␤␥ subunits (400 nM recombinant G␤ 1 ␥ 2 ) were mixed on ice in 100 l of buffer containing 10 mM NaCl, 2 mM EGTA, 1 mM EDTA, and 1 mM MgCl 2 . Reactions were started by the addition of 25 l of 2.5 mM CaCl 2 in 10 mM Hepes (pH 7.0) and incubated at 32°C for 15 min. Reactions were stopped by the addition of 1 ml CMH (chloroform: methanol:HCl mixed 100:100:1 by volume) and 250 l of 10 mM EDTA. After extraction, the aqueous phase (400 l) was counted on a Beckman scintillation counter to indicate PLC activity. All dose response curves have been fitted with the curve for a sigmoidal dose response curve using the equation y ϭ minimum ϩ (maximum Ϫ minimum)/(1 ϩ 10 ٙ (LogEC 50 Ϫ X)), where X is the logarithm of concentration of peptide and Y is the activity of PLC. The PLC activity is expressed as pmol inositol 1,4,5-trisphosphate/min/g total protein. Over the course of this study two different batches of PLC-␤2 were used. Basal activities between the batches varied by about 2-fold. All experiments were repeated a minimum of three times, and typical results for each set of experiments are shown.

RESULTS
Blade 7 in the structure of the G␤ propeller has been implicated by a number of studies to be important for the stabilization of the G␤ structure and for effector regulation (9,12,13).
The crystal structure of G␤ shows that blade 7 is the only blade of G␤ that is not composed of a contiguous amino acid sequence (13). The outermost ␤ strand (the d-strand) of this blade is formed from a sequence near the amino terminus of G␤, G␤-(47-54), while the remaining ␤ strands (a, b, and c) are formed by the carboxyl terminus of G␤ (Fig. 1A). Here, amino-terminal residues in the d-strand make contact with carboxyl-terminal resides in the c-strand of this blade to form a so-called snap to close the propeller structure. In addition to this structural role some data suggest that residues within the d-strand of blade 7 may play a direct role in G␤ effector interactions (7)(8)(9). Yeast two-hybrid data show the first 100 amino acids of G␤ to bind to a number of effectors including PLC-␤2 (8). Site-directed mutagenesis data also suggest this region to be involved in G␤␥ regulation of PLC-␤2 as select mutations within this region decrease the observed maximal effect for G␤␥ stimulation of PLC-␤2 activity by up to 85% (7). To ascertain if direct contact with PLC-␤2 occurs with this region of G␤ and what function this contact might serve in regulation of PLC-␤2, we tested a peptide that includes this region of G␤, G␤-(42-54), on PLC-␤2 activity. This peptide, highlighted in magenta in Fig. 1A, has the ability to modestly stimulate PLC-␤2 activity on its own, with a K act value of ϳ7 M. This indicates that direct contact with PLC-␤2 may occur within this region of G␤ and that this contact is capable of stimulating PLC activity (Fig. 1B). We also tested a truncated peptide from this region, G␤-(46 -54) (Fig. 1B). The G␤-(46 -54) region composes just the outermost d ␤ strand of blade 7. This peptide has a 3-fold higher K act value than G␤-(42-54); however, the observed maximal effect of this truncated peptide for stimulating PLC-␤2 is the same as that for G␤-(42-54). These data indicate that the amino acids necessary for signal transfer within the G␤-(42-54) region reside in the blade 7 d ␤ strand.
We substituted two of the residues within the G␤-(42-54) region, Arg-48 and Arg-49, which were indicated by the mutagenesis studies to be important (7) to see how these specific amino acids are involved in stimulating PLC-␤2 within this signal transfer region. Fig. 2A shows the positions of Arg-48 and Arg-49 within blade 7 of G␤. Fig. 2B, upper panel, shows the effect of the G␤-(42-54)-R48A-substituted peptide as compared with wild type peptide sequence. This substituted peptide displayed very interesting behavior. The R48A substitution leads to a nearly 8-fold increase in K act ; however, this increase in K act is accompanied by a 2-fold increase in the maximal observed stimulation of PLC-␤2. This observation suggests Arg-48 to be important for binding to PLC-␤2 as well as for contributing to signal transfer capabilities. On the other hand, the G␤-(42-54)-R49A-substituted peptide, while also displaying an increase in the K act value, does not show an increase in the maximal observed effect seen with the G␤-(42-54)-R48Asubstituted peptide (Fig. 2B, middle panel). Even though the G␤-(42-54)-R48A-substituted peptide has a higher K act , it is able to evoke an enhanced enzymatic response from PLC-␤2. This observation may indicate for residue Arg-48 of G␤ the subtle balance between the potential ability to tightly bind to PLC-␤2 and the ability to initiate the presumed allosteric change in PLC-␤2 required for stimulation of enzyme activity. Substituting Ala for Arg at amino acid position 48 causes two major changes in the side chain properties of the amino acid at this position. First, the size of the side chain is a much smaller one. Second, there is a loss of positive charge and replacement with a hydrophobic side chain. We wondered which property of Arg-48 was important for its binding and signal transfer capabilities, and so we tested another substituted peptide from the G␤-(42-54) region in which the Arg at position 48 was substituted with citrulline (Cit), an amino acid which bears structural similarity with Arg but is uncharged. If the effect of the R48Cit peptide on stimulation of PLC-␤2 is similar to the wild type peptide then we can conclude that it is probably the size of the side chain at this position that is more important. If, however, the effect of the R48Cit peptide is similar to the R48A-substituted peptide, then we can conclude that it is probably the charge of the amino acid side chain at this position that is more important. As seen in Fig. 2B, lower panel, the G␤-(42-54)-R48Cit-substituted peptide displays the same increase in K act and increase in the observed maximal effect seen with the G␤-(42-54)-R48Asubstituted peptide. This observation suggests that the positive charge of the side chain of the amino acid at position G␤48 is likely the critical determinant in controlling the binding and signal transfer capabilities of this specific amino acid of G␤.
␤ strands b and c of blade 7, composed of carboxyl-terminal G␤ sequence, are also likely candidates to function in G␤␥mediated stimulation of PLC-␤2. Mutation of the conserved Asp residue in this blade, Asp-333, shows the most dramatic loss in the rate of G␤␥ complex formation as compared with analogous Asp mutations in other blades (12). Trypsin digestion shows that G␤␥ complexes, which do form, seem at least on a global scale to fold properly; however, local structural changes at the level of this individual blade might disrupt G␤␥ signaling even if the gross structure of the G␤ propeller remains largely intact. When Trp-332, a G␣ contact point, is mutated to Ala, the mutant G␤ is much less effective at modulating a number of effectors including PLC-␤2 (6). A chimeric G␤ constructed by substituting the last 20 amino acids of G␤ 1 with those of Dictyostelium G␤, an isoform of G␤ that poorly regulates the effector PLC-␤2, resulted in a G␤ that had lost nearly all ability to regulate PLC-␤2 activity but was still fairly effective at G␤ modulation of some other pathways, such as the MAPK pathway (9). Truncation of six amino acids from the carboxyl terminus of STE4, a yeast homolog of G␤, rendered G␤ incapable of interaction with downstream effectors, although it could still associate with STE18, the yeast homologue of G␥ (14). These data suggest that while the carboxyl-terminal 20 amino acids of G␤, ␤ strands b and c of blade 7, are important from a gross structural standpoint, they may also play roles in direct protein-protein interactions. To test whether these last 20 residues of G␤ form direct interactions with PLC we tested a peptide derived from these 20 residues on basal and G␤␥stimulated PLC activity. This peptide is very effective at blocking nearly all G␤␥ stimulation, whereas basal activity is largely unaffected (Fig. 3A). This effect, at varying concentrations of the G␤-(321-340) peptide, is shown in Fig. 3B, upper panel. The results show that G␤-(321-340) has roles beyond participating in the folding of the G␤ propeller structure or in maintaining a local conformation important for forming interactions with a G␤ effector protein. Specifically, our data indicate that this region is part of a general binding domain and not a signal transfer region because the 20-mer peptide was incapable of regulating PLC-␤2 on its own. The last six amino acids within this carboxyl-terminal region of G␤, G␤ 335-340, appear to play dual roles. By forming the c ␤ strand within blade 7 they seem to form a snap through interactions with the d ␤ strand within this blade to structurally close the G␤ propeller. Also, this region seems to make direct binding contact with PLC-␤2. We tested a peptide from the G␤-(321-340) region that was substituted at position 335, F335A. Phe-335 is a residue implicated by the homologous scanning mutagenesis experiments to be important in the ability of G␤␥ to stimulate PLC-␤2 (9) on G␤␥ stimulation of PLC-␤2. This substitution greatly reduced the ability of the peptide to inhibit G␤␥ stimulation as compared with the wild type peptide. This observation highlights Phe-335 of G␤ as an important binding contact within this general binding domain (Fig. 3B, lower panel).
The region G␤-(228 -249), ␤ strands a and b of blade 5, has been shown to be important for the folding of G␤ into its propeller structure and the formation of the G␤␥ dimer. When the conserved Asp residue in this blade is mutated, dimer formation is reduced (12). Mutation of residues within the region G␤-(228 -249) leads to a decrease in G␤␥ stimulation of PLC-␤2 (7). To determine whether this reduction in effector regulation is the result of a structural change, either affecting the global folding of the propeller or altering the conformation of another PLC-␤2 contact site, or if G␤-(228 -249) is a direct contact region for PLC-␤2, we tested the effect of the G␤-(228 -249) peptide on basal and G␤␥ stimulation of PLC-␤2. G␤-(228 -249) had no measurable effect on PLC-␤2 basal activity, but it could inhibit close to 100% the stimulation by G␤␥ with an apparent K i of about 200 M (Fig. 4, upper panel). A G␤-(228 -249) peptide that was substituted at two positions, D228R and D246S, was ineffective at modulating G␤␥ stimulation of PLC-␤2 (Fig. 4, lower panel). This is consistent with the mutagenesis studies in which these same substitutions led to a nearly complete loss of the effect of G␤␥ on PLC-␤2 (5). This region, thus, seems to be part of a general binding domain that makes direct physical contact with PLC-␤2 but plays no role in signal transfer. Asp residues at positions 228 and 246 are important in this binding interaction.
Mutagenesis studies have predicted two additional regions of G␤ to be potentially important in forming protein-protein interactions with PLC-␤2. These are the regions G␤-(300 -313) and G␤-(64 -84) (6 -8). When chimeras are made with analogous Dictyostelium sequence both showed a reduction in G␤␥ stimulation of PLC-␤2 (9). For one of these regions, G␤-(64 -84), mutation of amino acids at positions Ser-72, Asp-76, and Trp-82 renders G␤␥ with a loss of function for modulating PLC activity (8). However, the G␤-(64 -84) region appears to be important from a structural standpoint. Amino acids Ser-72, Asp-76, and Trp-82 form part of a hydrogen bonding network that is conserved in all WD repeats, and so mutations at these positions might disrupt G␤␥ regulation of PLC-␤2 by causing a structural change that prevents proper folding of this blade (13). Also, residues within the G␤-(64 -84) region make intramolecular contacts with the G␤ 86 -105 region, a region of G␤ that we have previously found to be a signal transfer region for PLC stimulation (10). Mutations in the G␤-(64 -84) region could alter G␤␥ regulation of PLC indirectly by changing the structure of a signal transfer region on G␤. We tested a peptide encoding the G␤-(64 -84) region and found it to have no measurable effect on either basal or G␤␥ stimulation of PLC (Fig. 5A). This experiment suggests that this region might not be important for making a direct contact with PLC and supports its structural role by either maintaining local secondary structure or the structure of another signal transfer region.
Mutagenesis experiments highlight specific amino acids in the G␤-(300 -313) region as critical in G␤␥ regulation of PLC-␤2 (7). For example the L300A mutation can affect regulation of PLC-␤2. However, Leu-300 is also a direct G␥ subunit contact. The L300A mutation could affect the structure of the G␤␥ dimer. We tested a peptide from the G␤-(300 -313) region and found it to have no observed effect on either basal or G␤␥ stimulation of PLC (Fig. 5B). Thus, it seems that this region also does not make contact with PLC-␤2. The G␤-(300 -313) region may be important in G␤␥ stimulation of PLC-␤2 because it functions to stabilize a certain structure whether it be through interactions with G␥ or through local interactions within the blade, which is necessary for effective signal transfer to PLC-␤2. DISCUSSION The experiments in this study show that G␤␥ interactions with PLC have evolved in modular domains so that multiple sites of interaction can regulate binding affinity and signal transfer. The roles of the various sites are summarized in Fig.  6. The multiplicity of both signal transfer and general binding domains raises questions about the advantage of such a configuration.
It is rather surprising that we have identified two signal transfer regions of G␤. Each appears sufficient to transfer signals because each can independently activate PLC-␤2. Why have two such regions? One explanation is that multiplicity of signal transfer regions increases the probability of signal transfer even when interaction affinities are low. This same reasoning can explain why signal transfer regions are not contiguous with general binding domains. It is likely that these functions have been distributed to different effector contact sites because the affinity required to enable these two proteins to interact fruitfully with specificity might not allow the flexibility necessary to induce local allosteric changes in the effector protein required for change in activity. Protein interaction regions must compromise between tight specific binding and efficacious signal transfer. This idea is supported by the analysis of Arg-48 substitutions within the G␤-(42-54) peptide. Here it seems that while the R48A substitution renders the G␤-(42-54) peptide with a higher K act for stimulating PLC-␤2, it increases the observed maximal extent of this stimulation. Specifically, eliminating the positive charge from the amino acid side chain at this position might prevent this region of G␤ from interacting tightly with PLC. This weaker binding configuration might allow more effective signal transfer from this region of G␤ to the effector and hence increase the extent of stimulation. Here nature, it seems, has sacrificed this enhanced extent of stimulation to gain a tighter, more specific interaction with this signal transfer region.
Multiple general binding domains may exist to perform a variety of functions. Some domains might function in adhesion and act as the "velcro" to hold the two protein partners together. Here, multiple dispersed and lower affinity interactions may be important to permit reversibility while utilizing spatial geometry to achieve specificity. Other general binding domains may function in orientation, perhaps to align a signal transfer region into the proper configuration. The possibility that some general binding domains function as anti-adhesion regions to destabilize strong binding interactions between signal transfer regions and the effector cannot be ruled out. Such a counterintuitive interaction would be useful in obtaining reversible interactions with the effector while still retaining fine-tuned specificity.
The modular design of G␤ for regulation of PLC may have evolved to allow for reversibility of interaction and effective signal transfer while maintaining specificity of interaction with PLC-␤2. The geometrical constraint imposed by spatially separated sites can confer a unique level of specificity for low affinity interactions between G␤ and its effector. Such mecha-nism will easily promote the transient and reversible regulation of specific effector enzymes, achieving an essential requirement for signal flow within the cell.
Our studies highlight the flexibility with which nature has designed protein-protein interactions to achieve specific and transient signal flow within signaling pathways. Such information is important not only to define the fundamental principles of the molecular mechanisms by which these two signaling proteins communicate but also to serve as a foundation for the design of small molecules that may mimic or inhibit such signal flow.