RACK1 Binds to a Signal Transfer Region of Gβγ and Inhibits Phospholipase C β2 Activation*

Receptor for Activated C Kinase 1 (RACK1), a novel Gβγ-interacting protein, selectively inhibits the activation of a subclass of Gβγ effectors such as phospholipase C β2 (PLCβ2) and adenylyl cyclase II by direct binding to Gβγ (Chen, S., Dell, E. J., Lin, F., Sai, J., and Hamm, H. E. (2004) J. Biol. Chem. 279, 17861-17868). Here we have mapped the RACK1 binding sites on Gβγ. We found that RACK1 interacts with several different Gβγ isoforms, including Gβ1γ1, Gβ1γ2, and Gβ5γ2, with similar affinities, suggesting that the conserved residues between Gβ1 and Gβ5 may be involved in their binding to RACK1. We have confirmed this hypothesis and shown that several synthetic peptides corresponding to the conserved residues can inhibit the RACK1/Gβγ interaction as monitored by fluorescence spectroscopy. Interestingly, these peptides are located at one side of Gβ1 and have little overlap with the Gα subunit binding interface. Additional experiments indicate that the Gβγ contact residues for RACK1, in particular the positively charged amino acids within residues 44-54 of Gβ1, are also involved in the interaction with PLCβ2 and play a critical role in Gβγ-mediated PLCβ2 activation. These data thus demonstrate that RACK1 can regulate the activity of a Gβγ effector by competing for its binding to the signal transfer region of Gβγ.

Measurement of PLC␤2 Activity-The G␤␥ and peptide-mediated PLC␤2 activation was determined essentially as described before (12).
Construction, Expression, and Purification of G␤␥ Mutants-To construct G␤ 1 mutants for expression in Sf9 cells, a GATEWAY cloning system (Invitrogen) was used. Briefly, bovine G␤ 1 was first cloned into pENTR/SD/D-TOPO vector (Invitrogen) and site-directed mutagenesis was performed on pENTR/SD/D-TOPO G␤ 1 using a QuikChange site-directed mutagenesis kit (Stratagene). After the mutant G␤ 1 was transferred into a donor plasmid pDEST8 by homologous recombination reactions, recombinant baculoviruses encoding the G␤ 1 mutants were generated using a Bac-To-Bac baculovirus expression system (Invitrogen). Expression and purification of mutant G␤ 1 ␥ 2 from Sf9 cells were performed as described above.
Data Analysis-Unless indicated, data are representative of at least three independent experiments. Results are expressed as the mean Ϯ S.E. Student's t-tests were used to determine significant differences (two-tail p Ͻ0.05).

RESULTS
Interaction of RACK1 with Different G␤␥ Isoforms-To identify RACK1 binding sites on G␤␥, we first compared the binding of GST-RACK1 to different G␤␥ isoforms, including G␤ 1 ␥ 1 , G␤ 1 ␥ 2 , and G␤ 5 ␥ 2 . As shown in Fig. 1, A and B, GST-RACK1 interacts with purified G␤ 1 ␥ 1 , G␤ 1 ␥ 2 , and G␤ 5 ␥ 2 with similar affinities (ϳ0.3-0.5 M). G␥ 1 and G␥ 2 share only 37% amino acid identity and are post-translationally modified with different lipids, with G␥ 1 being farnesylated and G␥ 2 geranylgeranylated. The similar binding of RACK1 to G␤ 1 ␥ 1 and G␤ 1 ␥ 2 suggests that G␥ is not a major determinant for the RACK1/G␤␥ interaction. Consistent with this hypothesis, after forming a complex with its native partner, RGS9L, G␤ 5 can bind to GST-RACK1 as well as G␤ 5 ␥ 2 (data not shown).
RACK1 Binds to a Unique Region of G␤-G␤ 1 and G␤ 5 share only ϳ50% amino acid identity. However, the conserved residues between these two G␤ isoforms are clustered in the seven blades of G␤ that are known to be involved in interactions with diverse proteins (Fig. 2A). Given the similar binding of G␤ 1 ␥ 2 and G␤ 5 ␥ 2 to RACK1, we hypothesized that the RACK1 contact residues on G␤ are localized in these conserved regions. To test this hypothesis, we synthesized a series of peptides corresponding to surface-exposed residues of the conserved regions and determined their effects on the RACK1/G␤␥ interaction (Fig. 2B). For the purpose of synthesis, purification, and solubilization, we have kept the length of peptides at 9 -20 amino acids. Except for two peptides, p86 -105 and p265-275, that could not be dissolved in solution probably due to the high content of hydrophobic amino acids in the sequences, all other peptides were soluble.
To be able to identify peptides that possess low binding affinities to RACK1, we have developed a sensitive and quantitative fluorescencebased assay to monitor the RACK1/G␤␥ interaction in solution. In this assay, purified G␤␥ was labeled with an environmentally sensitive thiolreactive fluorescent probe, M8. Binding of purified MBP-RACK1 to M8-labeled G␤␥ was monitored by changes in the intensity and emission wavelength of fluorescence. This method allows us to monitor the interaction of RACK1 with G␤␥ in real time in solution and evaluate the ability of peptides to perturb the interaction. Because G␤ 1 ␥ 1 interacts with RACK1 with similar affinity as other G␤␥ isoforms, it was used in this study as it can be easily purified from bovine retina. As shown in Fig.  3, A and B, the addition of MBP-RACK1 to M8-labeled G␤ 1 ␥ 1 dose-dependently increased the intensity of fluorescence but did not cause a significant shift of the peak emission wavelength (data not shown). MBP alone had little effect. The increase in fluorescence is probably caused by the enhanced hydrophobicity around the fluorescent probe on G␤␥ upon RACK1 binding. The RACK1-mediated enhancement of the M8-G␤␥ fluorescence was immediate, suggesting the association of these proteins was rapid, i.e. within the mixing time. The binding affinity for the RACK1/G␤ 1 ␥ 1 interaction determined from this assay (EC 50 ϳ0.5 M) is similar to that from GST-RACK1 pulldown assays (EC 50 ϳ0.3 M), suggesting that the modification of G␤ 1 ␥ 1 with M8 does not perturb its binding to RACK1.
Results for some of the tested peptides are summarized in Fig. 4B. Peptides p86 -105 and p265-275 were excluded from this study as they could not be dissolved in solution unless using Me 2 SO, which itself interferes with the fluorescence of M8-G␤ 1 ␥ 1 . Except for peptides p44 -54, p309 -316 and p328 -337, p86 -98 and p287-295, which showed significant inhibitory effects on the RACK1/G␤␥ interaction, all other peptides had no effect. The inhibitory effects of the individual peptides on the RACK1/G␤␥ interaction ranged from 10 to 70%. When used in combination, peptides p44 -54 and 328 -337 caused almost complete inhibition of the RACK1/G␤␥ interaction, suggesting that they have additive inhibitory effects (data not shown). In addition, these peptides can also cause the dissociation of G␤␥ from RACK1 when added after the formation of the G␤␥-RACK1 complex, suggesting that the peptides may perturb the RACK1/G␤␥ interaction through competitive binding for RACK1 with G␤␥. Interestingly, when mapped to the crystal structure of G␤ 1 ␥ 1 , the corresponding residues of these inhibitory peptides are clustered at one side of G␤ 1 involving blades 1, 6, and 7 ( Fig. 4B, inset). Except for residues Lys-89, Ser-98, and Trp-332, which are contained in peptides p86 -98 and p328 -337, other G␣ subunit-interacting residues do not fall within the RACK1 binding sites in G␤␥ (Fig. 5A). This is in contrast to other G␤␥-interacting proteins such as GRK2 and phosducin whose binding sites on G␤␥ largely overlap with the G␣ binding interface (Fig. 5, B and C) (3,4).
PLC␤2 and RACK1 Share Overlapping Contact Residues on G␤␥-We have shown previously that RACK1 directly inhibits G␤␥-mediated PLC␤2 activation (12). To determine whether this is due to a competitive binding of RACK1 for the PLC␤2 contact residues on G␤␥, we further evaluated effects of the peptides on the PLC␤2/G␤␥ interaction. Similar to RACK1, the addition of PLC␤2 caused a dose-dependent increase in the fluorescence of M8-G␤ 1 ␥ 1 with an EC 50 ϳ1 M (data not shown). As with the interaction of RACK1 with G␤␥, peptides p44 -54, p86 -98, p287-296, p309 -316, and p328 -337 also inhibited the PLC␤2/G␤␥ interaction, with p44 -54 being most effective (Fig. 6). In addition to these peptides, peptide p177-189 also inhibited the PLC␤2/ G␤␥ interaction, although it had no effect on the binding of RACK1 to G␤␥. This suggests that the RACK1 and PLC␤2 binding sites on G␤␥ are overlapping but not identical.  (B). Pink and red colors in panel A indicate the similar and identical residues between G␤ 1 and G␤ 5 , respectively. The structure of the G␤ 1 ␥ 1 was generated using Swiss-Protein Data Bank viewer from the crystal coordinates of the G␤ 1 ␥ 1 (27)(28)(29). The G␤␥ Contact Region for RACK1 Is Critical for PLC␤2 Activation-Based on the ability of G␤-derived peptides to either perturb G␤␥mediated PLC␤2 activation or activate PLC␤2 in the absence of G␤␥, Buck and Iyengar (18) have proposed that contact regions of G␤␥ for an effector can be involved in either general binding or signal transfer functions. To determine the role of RACK1 binding sites in G␤␥-mediated PLC␤2 activation, we then evaluated effects of the G␤␥-derived peptides on the activity of PLC␤2. As shown in Fig. 7A, a majority of peptides, including those that can perturb the G␤␥/RACK1 and G␤␥/ PLC␤2 interactions, did not affect either the basal or the G␤␥-mediated PLC␤2 activation. However, peptide p177-189, which inhibited the interaction of G␤␥ with PLC␤2 but not RACK1, abolished G␤␥-stimulated PLC␤2 activity even better than GRK2-ct, suggesting that the corresponding residues of this peptide are involved in binding and signaling of G␤␥ to PLC␤2. Interestingly, peptides p44 -54 and p86 -105 stimulated PLC␤2 activity even in the absence of G␤␥ (Fig. 7B). These results are similar to previous reports using peptides from the same region (18,19), suggesting that the peptides are derived from regions of G␤␥ involved in signal transduction. Peptide p86 -105 seems to have a more potent ability to stimulate PLC␤2 than p44 -54 (Fig. 7B). However, the potency and efficacy of p86 -105 could not be accurately obtained in these studies because at higher concentration it precipitated out of solution.
The Charged Amino Acids within Residues 44 -54 Are Critical for Efficient G␤␥/RACK1 Interaction and PLC␤2 Activation-Because p44 -54 is the most effective peptide in perturbing the interaction of G␤␥ with either RACK1 or PLC␤2 and it can activate PLC␤2, we further determined the critical residues mediating its activity. We initially synthesized a series of p44 -54 mutant peptides with 2-3 alanine substitutions in each mutant and then determined their effects on the RACK1/ G␤␥ and PLC␤2/G␤␥ interactions (Fig. 8A). As shown in Fig. 8B, except for the mutant A53-54, which showed a reduced inhibitory effect (50 versus 70% for the original peptide) on the interaction of G␤␥ with RACK1, but not PLC␤2, all other mutant peptides have decreased abilities to inhibit both G␤␥/RACK1 and G␤␥/PLC␤2 interactions. Their inhibitory effects on the G␤␥/RACK1 and G␤␥/PLC␤2 interactions were reduced from 70 and 60% for the original peptide to 10 -20% and 20 -30%, respectively. The fact that substitution of residues 53-54 dif-FIGURE 5. Localization of structural determinants of G␤ 1 for interactions with RACK1, G␣ subunit, GRK2 and phosducin. Molecular surface of G␤ 1 ␥ 1 was generated using the crystal coordinates of the dimer (21) and the UCSF Chimera package (27,30). The surfaces that interact with RACK1 (A), GRK2 (B), and phosducin (C) are generated based on the data from Fig. 4B and previous reports (3,4) and are colored cyan, green, and blue, respectively. The surfaces that interact with G␣ t alone (21) are colored red, whereas surfaces that interact with both G␣ t and RACK1, G␣ t and GRK2, or G␣ t and phosducin are colored purple. ferentially affected the ability of p44 -54 to inhibit the G␤␥/RACK1 and G␤␥/PLC␤2 interactions further supports the notion that RACK1 and PLC␤2 share overlapping but not identical contact residues on G␤␥.
Next, we examined the ability of the mutant peptides to activate PLC␤2. As compared with the original peptide p44 -54, the mutant peptides A44 -46, A47-49 and A50 -52 showed decreases in both the potency and efficacy in PLC␤2 activation (Fig. 8C). The EC 50 (potency) is 585, 371, or 185 M for A44 -46, A47-49, or A50 -52 versus 21 M for the original peptide p44 -54, and the efficacy (E max ) is 84, 23, or 64% of the maximal response of the original peptide for A44 -46, A47-49, or A50 -52. Interestingly, mutant peptide A53-54 displayed a decrease in potency (172 versus 21 M for the original peptide) but enhanced efficacy (120% of the maximal response of the original peptide) in activating PLC␤2. This suggests that the last 2 residues, Gly-53 and His-54, play differential roles in mediating efficient interaction (potency) and maximal signal transduction (efficacy) between G␤␥ and PLC␤2.
Notably, peptide p44 -54 contains a high content of positively charged residues (5 of 11 amino acids) (Fig. 8A). Because all the mutant peptides have 1 or 2 charged residues substituted with alanine, we questioned whether the charged residues play a role in the interactions of G␤␥ with RACK1 and PLC␤2 and in PLC␤2 activation. We synthesized a peptide (5A) with all the 5 charged residues in p44 -54 replaced with alanines. Unfortunately, this peptide could not be dissolved in solution, probably because of the increased hydrophobicity following the alanine substitution. Although it can be dissolved in solvent such as Me 2 SO, this peptide could not be used in the spectrofluorometric assays because the solvent itself interfered with the fluorescence of M8-G␤ 1 ␥ 1 . However, we can determine the effect of this peptide on PLC␤2 activity because a small amount of Me 2 SO did not affect PLC␤2 activity. As shown in Fig.  8B, peptide 5A lost the ability to activate PLC␤2, suggesting that the charged residues are critical for the signaling function of p44 -54.
To further characterize the overall contribution of the charged amino acids within residues 44 -54 to the RACK1/G␤␥ interaction and PLC␤2 activation in the context of full-length G␤, we constructed G␤ 1 mutants with residues Arg-48 and Arg-49, His-54, or all five charged residues Arg-46, Arg-48, Arg-49, Arg-52, and His-54 substituted with alanines. After these mutants were expressed together with His 6 -G␥ 2 in Sf9 cells and purified to near homogeneity, their abilities to interact with RACK1 and to activate PLC␤2 were evaluated. As shown in Fig. 9, A and B, mutations of Arg-48 and Arg-49 or His-54 (data not shown) did not affect the binding of G␤␥ to RACK1. The mutant G␤ 1 (5A)␥ 2 showed a slight decrease in affinity but not binding capacity (B max ) for RACK1, suggesting that the 5 charged residues Arg-46, Arg-48, Arg-49, Arg-52, and His-54 contribute to the tight association of G␤␥ with RACK1. To ensure that the reduced binding affinity of the mutant G␤ 1 (5A)␥ 2 to RACK1 is not secondary to global structural changes, we have also evaluated the binding of this mutant to the C terminus of GRK2 and phosducin. Based on the crystal structures of the G␤ 1 ␥ 2 -GRK2 and G␤ 1 ␥ 1phosducin complexes, none of the 5 residues is involved in interaction with GRK2 but Arg-42 and Arg-46 are engaged in direct contacts with the C-terminal domain of phosducin (3,4). As expected, the mutant G␤ 1 (5A)␥ 2 interacts with GRK2-ct as well as the wild-type G␤ 1 ␥ 2 (Fig.  9C). By contrast, G␤ 1 (5A)␥ 2 showed a reduced binding to phosducin (Fig. 9D). These findings suggest that mutations of the 5 charged residues did not result in misfolding of G␤ 1 ␥ 2 .

DISCUSSION
In this study, we have identified the key contact region of G␤␥ for RACK1. Moreover, we have shown that the G␤␥ binding sites for RACK1 are shared by PLC␤2 and contribute to G␤␥-mediated PLC␤2 activation. These findings thus provide direct evidence for the ability of RACK1 to compete for a common effecter binding site on G␤␥ and inhibit G␤␥-mediated effector activation.
Our identification of the RACK1 binding sites on G␤␥ was facilitated by the use of a combination of peptide and spectrofluorometric approaches. The use of peptides to compete for the binding of G␤␥ to RACK1 allows us to screen a large surface area of G␤␥ in a short period. Moreover, it avoids potential structural changes of G␤␥ when mutagenesis experiments are performed on G␤␥ to identify the RACK1 contact residues. The caveat of this approach, however, is that the binding affin- . Interactions of wild-type and mutant G␤ 1 ␥ 2 with RACK1, GRK2-ct, and phosducin. A, binding of GST-RACK1 (ϳ20 nM) to purified wild-type G␤ 1 ␥ 2 and its mutants G␤ 1 (R48/49A)␥ 2 and G␤ 1 (5A)␥ 2 at the indicated concentrations. The binding was determined by GST binding assays in vitro. Pellets containing GST-RACK1 and G␤␥ were resolved by SDS-PAGE and probed with antibodies against G␤ (upper panel) or GST (middle panel). One percent of G␤␥ inputs were also examined by immunoblotting to show the equal loading of wild-type and mutant G␤ 1 ␥ 2 (lower panel). B, affinities of the interaction of RACK1 with wild-type and mutant G␤ 1 ␥ 2 . The amount of G␤␥ bound to GST-RACK1 was determined by densitometric scanning and calculated using the known amount of purified G␤ 1 ␥ 2 as a standard. The EC 50 and B max are 1.23 Ϯ 0.7 M and 1.71 Ϯ 0.5 pmol for wild-type G␤ 1 ␥ 2 , 0.85 Ϯ 0.3 M and 1.36 Ϯ 0.5 pmol for G␤ 1 (R48/48A)␥ 2 , and 6.4 Ϯ 3.2 M and 1.38 Ϯ 0.6 pmol for G␤ 1 (5A)␥ 2 . C and D, binding of GST-GRK2-ct (C) or GST-phosducin to wild-type G␤ 1 ␥ 2 and its mutant G␤ 1 (5A)␥ 2 . ϳ20 -50 nM of GST-GRK2-ct and GST-phosducin and the indicated concentrations of wild-type and mutant G␤ 1 ␥ 2 were used in the GST binding assays. Lanes 1 in panels C and D, GST control. ity of the peptides to RACK1 may be significantly lower than when they reside in the native proteins. To circumvent this problem, we have developed a sensitive fluorescence competition approach. This allowed identification of peptides such as p86 -98 and p309 -316, which have small inhibitory effects, as little as 10%, on the RACK1/G␤␥ interaction that may be neglected by using non-equilibrium approaches such as GST pulldown assays. It is noteworthy that of 13 peptides we studied, only 5 (p44 -54, p86 -98, p287-298, p309 -316, and p328 -337) significantly inhibit the G␤␥/RACK1 interaction, suggesting that the corresponding residues of these peptides are specifically involved in the binding of G␤␥ to RACK1. However, it should be pointed out that we cannot completely exclude the potential contribution of other regions of G␤␥ in binding RACK1, because in this study we only evaluated the conserved domain between G␤ 1 and G␤ 5 and it is possible that other regions also play a role in the G␤␥/RACK1 interaction. In addition, peptides with small effects on the G␤␥/RACK1 interaction that are beyond the sensitivity of our assays will not be detected. However, based on the following considerations, we believe that the region identified on G␤␥ in this study is the key contact between G␤␥ and RACK1. First, although the corresponding residues of the peptides p44 -54, p86 -98, p287-298, p309 -316, and p328 -337 are not contiguous in amino acid sequences, they are close in space and, particularly, clustered on a same surface of G␤␥ and confined to three blades, blades 1, 6, and 7. This is consistent with the fact that, like G␤, RACK1 is a member of WD40 repeat proteins that may form a rigid circular ␤ propeller structure and thus its interaction with G␤␥ may be mediated through limited contacts (13). Second, peptide p44 -54 alone can inhibit up to 70% of the G␤␥/ RACK1 interaction, and a combination of this peptide with p328 -337 from the same blade 7 where p44 -54 resides almost completely abolished the interaction, suggesting that residues 44 -55 and 328 -337 are the major molecular determinants of G␤␥ for binding RACK1. In further support of this notion, mutations of 5 charged residues within amino acids 44 -54 of G␤ 1 resulted in a weaker interaction between G␤ 1 ␥ 2 and RACK1.
Based on the crystal structure of the heterotrimer G␣t␤ 1 ␥ 1 and G␣ i ␤ 1 ␥ 2 , there are 16 residues from G␤ 1 engaged in interaction with the G␣ subunit (21)(22)(23). The peptides tested in this study contain 13 of these residues. However, only 3 G␣-interacting residues, Lys-89, Ser-98, and Trp-332, are found to be within the RACK1 contact region of G␤␥ (Fig. 5A). Because we could not clearly define the boundary of the RACK1 binding sites on G␤␥ by using the peptide approach, it remains to be determined whether these residues are involved in direct interac-tions with RACK1. Nevertheless, these residues do not seem to be essential for the binding of G␤␥ to RACK1, as peptide p86 -98 that contains residues Lys-89 and Ser-98 inhibited only 10% of the RACK1/ G␤␥ interaction. Moreover, a G␤ 1 ␥ 2 W332A mutant binds to GST-RACK1 as well as the wild-type G␤ 1 ␥ 2 (data not shown). Furthermore, the G␤ 1 ␥ 1 subunit can still interact with RACK1 after it forms a heterotrimeric complex with the G␣ t subunit (11). These findings suggest that the G␣-interacting residues do not play an essential role in the binding of RACK1 to G␤␥. This is in contrast to GRK2 and phosducin, whose binding sites on G␤␥ significantly overlap with the G␣ binding interface (3,4). Because the G␣ contact residues on G␤␥ are known to be involved in activation of all known G␤␥ effectors, the lack of significant overlapping of the G␤␥ binding sites for RACK1 with the G␣ binding interface may have important implications for the physiological role of RACK1. Thus, unlike GRK2 and phosducin, whose binding to G␤␥ resulted in inhibition of all known G␤␥ effectors, RACK1 may regulate only a subset of G␤␥ effectors whose binding sites overlap with those of RACK1. Indeed, we have found previously that overexpression of RACK1 affects G␤␥-mediated PLC␤2 and ACII activation, but not the activation of mitogen-activated protein kinases nor chemotaxis of human embryonic kidney 293 cells via G␤␥, although in the last two cases the direct effectors of G␤␥ are unknown (12).
Residues within the RACK1 contact region of G␤␥ have been implicated in the activation of a variety of effectors, including PLC␤2, ACII, and the G protein-coupled inwardly rectifying potassium (GIRK) channels. For example, mutations of 2 (Arg-49 and Thr-50) or 4 residues (Arg-46, Arg-48, Arg-49, and Thr-50) in the outer strand of blade 7 (residues 42-52) abolished G␤␥-mediated PLC␤2 activation (8). Substitution of Lys-89 with alanine inhibits the activation of PLC␤2, ACII, and the GIRK channels, and an S98A mutant of G␤ 1 ␥ 2 displayed a reduction in the activation of ACII (5). A C-terminal chimera mutant of G␤ 1 with 4 residues, Val-327, Ala-328, Phe-335, and Asn-340 replaced with the corresponding residues of Dictyostelium G␤ was unable to stimulate PLC␤2 (24). Finally, mutagenesis studies of the C-terminal domains of the G␤ that either forms a hydrophobic binding pocket for the G␥ subunit prenyl group or undergoes conformational change upon the binding of G␤␥ to phosducin indicate that residues in this region are very important in the activation of both PLC␤2 and ACII (25). Taken together, these findings suggest that amino acids in the RACK1 binding sites of G␤␥ are critically involved in effector activation. In support of this notion, we have provided direct evidence that RACK1 and PLC␤2 share overlapping contact residues on G␤␥, because peptides that inhibited the RACK1/G␤␥ interaction have the same effects on the PLC␤2/G␤␥ interaction. Although the binding of PLC␤2 to G␤␥ involves multiple domains of G␤␥ as noted here and by other reports (5,6,8,26), the RACK1 contact region is critical for PLC␤2 activity. This is demonstrated by the fact that peptides p44 -54 and p86 -105 can directly activate PLC␤2. Moreover, mutations of the charged residues in the domain between amino acids 44 and 54 affect the potency and efficacy of PLC␤2 activation. These findings thus unambiguously demonstrate that by competitive binding to the region of G␤␥ critical for effector interaction and activation, RACK1 can regulate the function of G␤␥.