A CONSERVED PROTEIN INTERACTION INTERFACE ON THE TYPE 5 G PROTEIN β SUBUNIT CONTROLS PROTEOLYTIC STABILITY AND ACTIVITY OF R7 FAMILY REGULATOR OF G PROTEIN SIGNALING PROTEINS

Regulators of G protein signaling (RGS) proteins of the R7 subfamily limit signaling by neurotransmitters in the brain and by light in the retina. They form obligate complexes with the Gβ5 protein that are subject to proteolysis to control their abundance and alter signaling. The mechanisms that regulate this proteolysis, however, remain unclear. We used genetic screens to find mutations in Gβ5 that selectively destabilize one of the R7 RGS proteins in Caenorhabditis elegans. These mutations cluster at the binding interface between Gβ5 and the N terminus of R7 RGS proteins. Equivalent mutations within mammalian Gβ5 allowed the interface to still bind the N-terminal DEP domain of R7 RGS proteins, and mutant Gβ5-R7 RGS complexes initially formed in cells but were then rapidly degraded by proteolysis. Molecular dynamics simulations suggest the mutations weaken the Gβ5-DEP interface, thus promoting dynamic opening of the complex to expose determinants of proteolysis known to exist on the DEP domain. We propose that conformational rearrangements at the Gβ5-DEP interface are key to controlling the stability of R7 RGS protein complexes.

Both mammalian and C. elegans R7 RGS proteins share considerable sequence similarity and a common domain organization. In addition to the RGS homology domain that acts as a GTPase activator to terminate Gα signaling, these proteins possess an N-terminal DEP/DHEX (disheveled/EGL-10/pleckstrin similarity domain/DEP helical extension) module that recruits the SNARE-like membrane-binding protein R7BP/RSBP-1, followed by the GGL (G gamma-like) domain that binds the atypical G protein beta subunit Gβ5/GBP-2 (6,7,9). Association with Gβ5/GBP-2 and R7BP/RSBP-1 is essential for the stability of these complexes. While disruption of R7BP/RSBP-1 selectively destabilizes RGS9 and EAT-16 (9,10), knockout of Gβ5/GBP-2 essentially eliminates the expression of all R7 RGS proteins (11,12) leading to the hypothesis that interactions with Gβ5 play a central role in controlling the stability of R7 RGS proteins. The recent crystal structure of the RGS9 Gβ5 complex indicates that Gβ5 has three distinct interactions with R7 proteins: in addition to marginal contacts with the RGS domain, it forms extensive contacts with the GGL and the DEP domains (13).
Alterations to R7 RGS protein levels are thought to be an important mechanism underlying signaling plasticity and have been documented in response to changes in receptor activation status under pathological conditions such as Parkinson's disease (14) and addiction (15,16). Experiments in C. elegans (9,17) and mice (16,18,19) indicate that the abundance of R7 RGS proteins is critically important in determining the extent of their regulatory influence. The mechanisms that alter R7 RGS abundance remain unclear, but any mechanism that dynamically regulates protein abundance requires protein turnover. Given the sensitivity of R7 RGS proteins to degradation, their turnover may in fact be the regulated step.
In this study we have used the power of C. elegans genetics to identify an unusual mutation in Gβ5/GBP-2 that preferentially affects the stability of EAT-16 over EGL-10, resulting in a characteristic hyperactive phenotype caused by enhanced Gα q signaling. Interestingly, previous genetic screens reported several mutations in Gβ5/GBP-2 with similar functional properties but unexplained mechanisms (20,21). Our analysis shows that all of the identified "hyperactive" mutations are in residues conserved in mammals and disrupt the Gβ5-DEP interface, which serves as the hot-spot for the regulation of complex stability.

EXPERIMENTAL PROCEDURES
Nematode strains and culture C. elegans strains were maintained and double mutants generated using standard techniques (22). All strains used are listed in Supplemental Table  1, all gpb-2 mutations are listed in Supplemental Table 2. vs33 was isolated in a screen for hyperactive egg-laying mutants (23,24) and mapped using standard genetic techniques. Briefly, single nucleotide polymorphism (SNP) mapping (25) placed vs33 between -0.69 cM and 5.06 cM on the genetic map of chromosome I. A combination of three-factor mapping and SNP analysis further positioned vs33 between two visible markers at 0.00 (dpy-5) and 3.30 (unc-29). Subsequent sequencing of candidate G protein signaling genes within this region identified a 3 bp deletion in gpb-2 that eliminates the D263 codon. C. elegans behavior and morphology Egg-laying assays were performed as previously by guest on March 24, 2020 http://www.jbc.org/ Downloaded from described (26). To determine the number of unlaid eggs, adult animals were dissolved in bleach and the number of bleach-resistant eggs counted. To determine the percentage of eggs laid at each developmental stage, adult animals were allowed to lay eggs for 30 minutes and the eggs visually inspected. In the unlaid egg assay 30 animals per genotype were analyzed and the mean and 95% confidence intervals calculated. In the developmental-stage assay, ≥100 eggs per strain were analyzed, 95% confidence intervals and p-values were calculated using Wilson's estimates. Precisely-staged adults for both assays were obtained by isolating late L4 larvae and aging for 30 hours at 20˚C.
Mammalian cells from one well of a six-well plate were lysed in 500 μl of PBS buffer supplemented with 150 mM NaCl, 1 % Triton X-100, and Complete protease inhibitor (Roche). The homogenate was centrifuged at 15,000 x g for 15 min at 4°C.

Pulse-Chase Degradation Experiments
Experiments to evaluate the stability of RGS proteins to proteolytic degradation were performed in HEK293FT cells as described previously (28). HEK293FT cells grown in T-25 flasks were transfected with RGS constructs with or without wild-type or mutant Gβ5 constructs. For assays involving RGS9, R7BP was additionally co-transfected. Total cDNAs were 5 μg per flask. Twenty-four hours after transfection, the cells were rinsed twice with PBS and placed into 5 ml of starvation medium (10% dialyzed fetal bovine serum (Invitrogen), DMEM without L-Methionine or L-Cysteine (21013-024, Invitrogen) and incubated at 37°C and 5% CO2 for 30 min. Subsequently, newly synthesizing proteins were labeled by adding 175 μCi of [ 35 S]methionine/cysteine (NEG772; Perkin-Elmer Life Sciences) per flask to the starvation medium and incubating at 37°C for 40 min. Cells were rinsed twice with 2 ml of DMEM and incubated with 5 ml of complete medium supplemented by additional 2 mM L-methionine and 2mM L-cysteine (Sigma-Aldrich) at 37°C for the indicated incubation times. At the end of the incubation, the cells were scraped into 5 ml of ice-chilled PBS, pelleted by centrifugation and resuspended in 900 μl of radioimmune precipitation assay buffer (50 mM Tris-HCl, pH 7.8, 300 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% sodium dodecylsulfate) supplemented with protease inhibitors (Complete; Roche Applied Science). After 20 min incubation at 4°C, the suspension was then centrifuged at 4°C and 14,000 x g for 30 min, and the resulting supernatant was incubated with 20 μl of protein G beads slurry and 3 μg of RGS9-2 CT antibody or HA antibody for 1 h at 4°C. After 3 washes immunoprecipitated proteins were eluted from the beads by 50 μl of SDS sample buffer. Samples were run on SDS-PAGE and transferred to a PVDF membrane (Bio-Rad). The membrane was air-dried and incubated on a phorphorimaging screen overnight. This screen was then scanned using a STORM phosphorimager (Molecular Dynamics), and the by guest on March 24, 2020 http://www.jbc.org/ Downloaded from bands were quantified using ImageQuant Software (Molecular Dynamics). Each experiment was repeated at least twice.

Recombinant proteins
RGS7, RGS9-2 and wild-type Gβ5 recombinant baculoviruses were described previously (29,30). Mutant Gβ5 constructs (Gβ5-D260A, Gβ5-C263Y) were used to generate recombinant bacluoviruses using the BacluoDirect C-Terminal Expression Kit (Invitrogen) following manufacturer's manual. Proteins were expressed in Sf-9 cells and protein complexes were purified by Ni-NTA chromatography taking advantage of the His-tag engineered at the N-terminus of RGS proteins as previously described (29). Recombinant GST-tagged R7BP protein was expressed in Escherichia coli and purified as described previously (30).

GST pull down assay
The assays were performed as previously described (27). Briefly, purified recombinant GST fusion R7BP or GST control protein (0.25 μM), purified R7 RGS Gβ5 proteins (0.25 μM) were co-incubated with 20 μl of 50 % glutathione agarose beads slurry (GE Healthcare) in binding buffer (20 mM Tris, pH 7.2, 300 mM NaCl, 0.25% n-dodecanoylsucrose, 50 μg/ml bovine serum albumin) for 1 hour at 4°C. The beads were washed with binding buffer 3 times. Proteins were eluted in SDS sample buffer, and proteins present in the eluates were detected by immunoblotting with anti-RGS7 (7RC1), anti-Gβ5 (ATDG) and anti-R7BP (NT) antibodies. In silico molecular dynamic simulations Molecular dynamics simulations were performed using GROMACS (31). An identical procedure was used for all the models studied in this work, and consisted of the following steps. A model with generated hydrogens was placed in the box of appropriate size and fully solvated. Protonation states of ionizable groups were assigned based on analysis of hydrogen bonding patterns. Sufficient numbers of sodium and chloride ions were placed in random positions to emulate physiological ionic strength (150 mM) and to neutralize the overall charge of the system. The fully solvated model was subjected to energy minimization followed by 20ps simulation with protein atoms restrained to their positions in the crystal structure in order to allow water relaxation. Finally, 100ps unrestrained simulation was performed at 300K. Snapshots were obtained every 500fs and it was found that initial equilibration period did not exceed 30ps (as judged by overall r.m.s.d. of aligned models). The second half of the unrestrained simulation was used for analysis.

A novel allele of GPB-2/Gβ5 causes hyperactive behavior in C. elegans
To further our understanding of the mechanisms that regulate G protein signaling, we performed a genetic screen for animals hyperactive for egg laying, a behavior controlled by R7 RGS Gβ5 complexes (23). We identified a novel allele, vs33, and mapped it to a small interval on chromosome I containing the gene gpb-2/Gβ5. This mutation deletes the codon for aspartate 263 from this gene.
Wild-type animals hold their fertilized eggs in utero for a few hours while the embryos develop past the eight-cell stage. As a result wild-type animals accumulate 9.1±0.6 eggs in utero and lay only 12.3%±6.4% of their eggs at an early developmental stage of less than eight cells ( Figure 1B). Wild-type animals also move across a bacterial lawn in a sinusoidal wave, make few reversals and forage over much of the available bacterial lawn ( Figure 1C). These behaviors are the result of a balance between the opposing effects of Gα o and Gα q signaling ( Figure 1D).
GPB-2/Gβ5 is an essential subunit of the two C. elegans R7 RGS proteins, EGL-10 and EAT-16, which inhibit Gα o and Gα q signaling, respectively (12,20,32). In animals lacking GPB-2/Gβ5, there is no inhibition of Gα o or Gα q and signaling by both G proteins is increased ( Figure 1G). However, as this does not alter the net balance of signaling, gpb-2(null) mutants appear grossly wild-type in appearance and behavior. Consistent with previously published data, we observed that gpb-2(null) mutants accumulate 6.1±0.8 eggs in utero, lay only 16.0%±7.3% of their eggs at an early developmental stage ( Figure 1E) and exhibit locomotion and foraging behavior similar to that of wild-type animals ( Figure 1F).
In surprising contrast to gpb-2(null) mutants, we found that gpb-2(vs33) mutants are strongly hyperactive for egg-laying and locomotion behaviors. gpb-2(vs33) animals lay their eggs almost as soon as they are fertilized and thus accumulate only 2.3±0.3 unlaid eggs and lay 88.5%±6.3% of their eggs at an early developmental stage ( Figure 1H). In addition, gpb-2(vs33) mutants move with much deeper body bends than wild-type animals, make more reversals and restrict their foraging to a smaller area of the bacterial lawn ( Figure 1I). gpb-2(vs33) mutants are also thin and pale compared to the wild-type or gpb-2(null) animals ( Figure 1B,E,H). Finally, whereas wild-type animals are paralyzed by exogenously applied serotonin and dopamine (26,33), gpb-2(vs33) mutants but not gpb-2(null) animals are resistant to these effects (data not shown).
The defects exhibited by gpb-2(vs33) animals are characteristic of animals in which either signaling by the inhibitory G protein, Gα o , is decreased or signaling by the stimulatory G protein, Gα q , is increased ( Figure 1J). Interestingly, two other point mutations in gpb-2, the alleles sa833 and sa604, have been reported to cause the very same set of defects as observed in the gpb-2(vs33) mutants (20), Supplemental Table 2). We have named these gpb-2(hyp) mutations.

Gα q signaling is increased in gpb-2(hyp) mutants
To determine whether the behavioral defects seen in gpb-2(hyp) mutants are the result of decreased Gα o signaling or increased Gα q signaling, we first took a genetic approach. We hypothesized that gpb-2(hyp) mutations may either increase EGL-10 activity, causing decreased inhibitory Gα o signaling, or may decrease EAT-16 activity, resulting in increased stimulatory Gα q signaling. To distinguish between these alternatives, we combined each gpb-2(hyp) mutation with a null allele of egl-10. If the gpb-2(hyp) mutations increase EGL-10 activity, then in the absence of EGL-10 they should have no effect and the double mutants should phenocopy egl-10(null) single mutants. However, if gpb-2(hyp) mutants decrease EAT-16 activity, resulting in increased Gα q signaling, then in the absence of EGL-10, Gα o signaling will also be increased to restore the balance of signaling, as in gpb-2(null) mutants, so that the double mutants should resemble both wild-type and gpb-2(null) animals. We analyzed the stage of the eggs laid by each strain and this data is presented in Figure 2.
We first observed the egg-laying behavior of a number of control strains and classified the eggs laid by these animals as follows: early stage, less than eight cells; normal stage, nine cells to pre-comma; late-stage, comma and later (Figure 2A). We found that the majority of eggs (84%) laid by wild-type animals were in the normal range of stages ( Figure 2B, bar 1), whereas 100% of eggs laid by egl-10(null) mutants, in which inhibitory Gα o signaling is increased, were of a late stage ( Figure 2B, bar 2). We next examined two hyperactive egg-laying mutants, an EGL-10 overexpressing strain, in which inhibitory Gα o signaling is strongly inhibited, and an eat-16(null) mutant in which stimulatory Gα q signaling is increased. We observed that both strains lay the majority of their eggs (66% and 79%, respectively) at early developmental stages ( Figure 2B, bars 3 and 4). Finally, we combined the hyperactive eat-16(null) mutation (too much Gα q signaling) with the egl-10(null) mutation (too much Gα o signaling) and as predicted found these animals had a statistically significant decrease in the number of early-stage eggs laid and now laid a majority (93%) of normal stage eggs ( Figure 2B, bar 5), similar to gpb-2(null) mutants and to wild-type animals.
Next we put all three gpb-2(hyp) mutations through our genetic analysis. We found that all three single mutants laid the majority of their eggs at an early developmental stage: vs33, 88%; sa833, 83%; sa604, 72% ( Figure 2B, bars 6, 8 and 10). However when combined with the egl-10(null) mutation, all three double mutants exhibited dramatic and statistically significant decreases in the number of early-stage eggs laid, with gpb-2(vs33);egl-10(null) and gbp-2(sa604);egl-10(null) animals now laying the majority of their eggs, 73% and 89% respectively, within the normal range ( Figure 2B, bars 7, 9 and 11). This is consistent with gpb-2(hyp) mutations causing an increase in Gα q signaling and not a decrease in Gα o signaling.

EGL-10 and EAT-16 levels are differentially altered in gbp-2(hyp) mutants
The increase in Gα q signaling could be explained by the gpb-2(hyp) mutations specifically decreasing activity of the inhibitor of Gα q signaling, EAT-16, rather than increasing the activity of the inhibitor of Gα o signaling, EGL-10. As both R7 RGS proteins require GPB-2/Gβ5 for stability, gpb-2(hyp) mutations might have this effect by specifically decreasing cellular EAT-16 levels. We investigated this possibility by analyzing the levels of EGL-10, EAT-16 and GPB-2/Gβ5 in whole worm lysates of wild-type, gpb-2(null) and gpb-2(hyp) animals. As expected, we found in our control experiments that each protein was present in wild-type animals and absent in the corresponding null mutants, confirming antibody specificity ( Figure 3A-C, lanes 1-4). In addition, consistent with our previous data (9,12), we observed that both EGL-10 and EAT-16 were virtually absent in gpb-2(null) mutants ( Figure 3A and B, lane 4).
Our results suggest gpb-2(hyp) mutations lead to greater decreases in EAT-16 levels than of EGL-10 levels, ultimately shifting the balance in favor of Gα q signaling relative to Gα o signaling.

gpb-2(hyp) mutations affect the interface of GPB-2/Gβ5 with the DEP domain of R7 RGS proteins
We analyzed the position of each gpb-2(hyp) mutation within the R7 RGS GPB-2/Gβ5 complex. We first aligned the sequence of C. elegans GPB-2 with that of the mouse Gβ5 protein, used in the published crystal structure of the RGS9 Gβ5 complex (Cheever et al., 2008). As shown in Figure 4A and Supplemental Table 2, GPB-2/Gβ5 forms a seven WD40 repeat propeller and all three gpb-2(hyp) mutations, vs33 (D263Δ), sa833 (C266Y) and sa604 (D307N) affect conserved residues. When we mapped the gpb-2(hyp) mutations onto this structure, we found that the residues mutated in vs33 and sa833 are located within the fifth β-propeller blade, while the residue mutated in sa604 is located nearby on blade six ( Figure 4B). Furthermore, we found that both the residues mutated in vs33 and sa604 are positioned at the interface with the R7 RGS DEP domain ( Figure  4C). These observations strongly suggest that the amino acids mutated in gpb-2(vs33) and gpb-2(sa604) play important roles in specifying the interaction of GPB-2/Gβ5 with either EAT-16 or EGL-10.

Homologous interface mutants of mammalian Gβ5 fail to protect RGS9 and RGS7 from proteolytic degradation
Given the conservation of the Gβ5 R7 RGS interaction across species, we next asked whether the mutations at this interface have similar effects in the mammalian system. Based on the alignment between C. elegans GBP-2 and its mammalian homolog Gβ5, mutations at the corresponding positions in the mammalian Gβ5 short isoform (Supplemental Table 2: D260A, C263Y, D304N) were introduced by site-direct mutagenesis. We then checked the ability of these mutants to stabilize two representative mammalian R7 RGS proteins: RGS7 and RGS9. Consistent with published observations (34), co-transfection with Gβ5 in HEK293FT cells led to substantial stabilization of RGS7 as evidenced by the up-regulation of its levels ( Figure 5A). Unlike RGS7, association of Gβ5 with RGS9 is not sufficient to protect it from proteolytic degradation and additionally requires interaction with another binding partner, R7BP (10,27). Consequently, we found that when co-transfected with R7BP, wild type Gβ5 markedly enhanced the expression of RGS9 ( Figure 5B). In the presence of R7BP, the protection of RGS7 by wild-type Gβ5 was further increased. In contrast, neither D260A nor C263Y mutants of Gβ5 could enhance the expression of RGS7 and RGS9 in the absence or presence of R7BP. The third mutant D304N exerted effects similar to those of the wild-type Gβ5 suggesting that this mutation does not affect the stability of the complex. Importantly, the mutations did not seem to affect the expression levels of Gβ5 itself, analogous to our results with mutant GPB-2 proteins in C. elegans lysates (Figure 3C), so the lack of R7 RGS stabilization by the D260A and C263Y mutants of Gβ5 can not be explained by lower expression levels.
We next performed pulse-chase degradation experiments to determine whether the inability of the mutant Gβ5 constructs to enhance R7 RGS expression results from higher proteolytic susceptibility of the complexes ( Figure 5C, D). Consistent with earlier reports (27,34), we found that in the absence of Gβ5, both RGS7 (Figure 5 C) and RGS9 ( Figure 5D) have very short half-lives in cells (< 1 hr) due to high rates of proteolytic degradation (10,27). Co-expression with wild-type Gβ5 increased the lifetime of both R7 RGS proteins by as much as four to six fold. However, when D260A or C263Y mutants were used instead of wild-type Gβ5, no increase in the lifetime of the RGS7 or RGS9 was evident ( Figure 5C, D). Correlating with these results, we found that co-expression of wild-type Gβ5 largely protects RGS9 from ubiquitination, while the D260A and C263Y Gβ5 mutants failed to do so (Supplemental Figure 1). These data indicate that both D260A and C263Y mutants fail to provide proteolytic protection to RGS7 and RGS9 accounting for loss of their ability to support R7 RGS expression.

Intact formation of RGS Gβ5 R7BP complexes with Gβ5 mutants.
Since Gβ5 directly binds to R7 RGS proteins ( Figure 6A) we next asked whether the loss of physical interaction could account for the inability of the Gβ5 mutants to provide proteolytic protection to the complexes. We first investigated the interaction between Gβ5 and full-length RGS9, as a representative R7 RGS protein, by co-immunoprecipitation. An AU5 epitope tag was appended to all Gβ5 constructs to discriminate them from endogenous Gβ5 present in HEK293 cells. As is evident from the data presented in Figure 6B, wild-type Gβ5 as well as Gβ5-D260A and Gβ5-C263Y were co-immunoprecitated by RGS9 antibodies. Similarly, reciprocal immnoprecipitation using anti-AU5 antibody pulled down readily detectable levels of RGS9 confirming the interaction ( Figure 6C). Lower binding levels observed with Gβ5 mutants match lower expression levels of these complexes and suggest that the mutations do not eliminate the interaction of Gβ5 with R7 RGS proteins.
The recently solved crystal structure of the RGS9 Gβ5 complex (13) reveals that Gβ5 binds to three sites in RGS9 (Fig. 6A). In addition to a coiled-coil interaction with the GGL domain, it also makes contacts with the DEP domain of RGS9, an interaction sufficiently robust to be detected by biochemical methods (35). The third rather limited point of contact is with the RGS domain. We therefore reasoned that if Gβ5 mutations affect interaction only at the DEP-binding interface, their effects could be masked when the binding is examined with RGS9 molecules in which the other binding interface remains intact. To investigate this possibility we tested the binding of Gβ5 to either a DEP domain-containing N-terminal fragment (RGS9NT) or a GGL-containing C-terminal fragment (RGS9CT) separately. We found that both D260A and C263Y mutants could associate with RGS9NT or RGS9CT as well as did wild-type Gβ5, indicating the preservation of binding interactions at both sites.
Finally, we investigated the possibility that mutations in Gβ5 affect the ability of RGS9 to recruit R7BP, association with which is also necessary for providing proteolytic protection to the complex (27). We obtained purified recombinant preparations of the RGS9 Gβ5WT, RGS9 Gβ5D260A and RGS9 Gβ5C263Y complexes and studied their interactions with GST-tagged R7BP immobilized on the beads by pull-down assays. As shown in Figure 6F, the ability of R7BP to effectively retain RGS9 was not detectably affected by the mutations in Gβ5.
Together, these results demonstrate that mutations in Gβ5 destabilize RGS complexes by sensitizing them to proteolytic degradation without detectably affecting the initial formation of the trimeric R7 RGS Gβ5 R7BP complexes.

Gβ5 mutants are normally processed by the CCTε/PhLP1 chaperone assembling complex
The assembly of βγ complexes requires the assistance of the cytosolic chaperonin complex TriC/CCT, which acts in cooperation with the co-chaperone phosducin-like protein 1 (PhLP1) to facilitate folding of β subunits (36). The crystal structure of the RGS9 Gβ5 complex revealed that the interaction of Gβ5 with the GGL domain of RGS9 shares the same organization as that seen in conventional βγ complexes (13). Furthermore, the protein interaction interface engaging in the interaction with phosducin-like proteins is also conserved in Gβ5 (13). Indeed, a recent study (37) has demonstrated that folding of Gβ5 and its assembly with R7 RGS proteins is mediated by the coordinated action of TriC/CCT and PhLP1 ( Figure 7A). These observations raise the possibility that mutations in Gβ5 compromise the stability of the complex by affecting how it is processed by the CCT-PhLP1 machinery. For example, the rate of complex assembly with RGS proteins would be slowed if the interaction of Gβ5 with CCT and/or PhLP1 is disrupted.
We examined this possibility by comparing the association of wild-type and mutant Gβ5 proteins with PhLP1 and CCT complexes using immunoprecipitation assays. As shown in Figure  7B, we detected no differences in the ability of wild-type Gβ5 and its D260A and C263Y mutants to interact with PhLP1 and CCT. We therefore conclude that Gβ5 mutants are normally processed by the CCT-PhLP1 machinery to form intact complexes with R7 RGS proteins.

Molecular dynamic simulations show substantial re-arrangements of the DEP-Gβ5 interface
Our biochemical observations indicate that Gβ5 mutants retain its ability to associate with the R7 RGS proteins, suggesting subtle rearrangements in the complex that lead to its proteolytic destabilization. To gain insight into potential conformational changes in the RGS9 Gβ5 complex induced by the D260A and C263Y amino acid changes, we performed molecular dynamics (MD) simulation studies on the Gβ5 mutants structure alone and in complex with RGS9 (Figure 8).
Analysis of the intermolecular interactions in the RGS9 Gβ5 complex using the EBI PISA server (38) indicates that Gβ5 forms much weaker complex with the DEP domain (~8.4 kcal/mol solvation energy) as compared to the GGL domain (~32 kcal/mol). Both mutation sites reside in the β-turn loop that connects the second and third β-strands of the fifth blade of Gβ5 ( Figure 4B) and our analysis shows that the mutations leave the direct contacts of this mutated loop with the DEP domain intact. However, the mutated loop is adjacent to and contacts a "DEP domain interaction loop" that has much more significant interactions with the DEP domain ( Figure 8A), contributing about half of the contact surface of Gβ5 with the DEP domain and traversing the center of the binding epitope.
MD simulation of the mutant Gβ5 indicates that both mutations significantly influence the conformation of the DEP domain interaction loop. For example, D260A mutation converts the D260 loop from a β-turn to a less tight α-turn (39). This conformational change may modify the backbone hydrogen bonding between the mutated loop and the DEP interaction loop, affecting the conformation of the I283 and F284, two key residues at the interface with the DEP domain ( Figure 8B). In the C263Y mutant, the mechanism affecting the DEP domain interaction loop is different, yet it leads to a similar disruption of the I283/F284 residues ( Figure 8C).
It thus appears that D260A and C263A mutations cause re-arrangement of the interface, perhaps by making the interactions more transient, without having large influence on the affinity of the DEP-Gβ5 binding. It needs to be noted that this MD simulation is not a direct experimental observation and some deviations from these predictions are expected in the actual structure of the complex between Gβ5 mutants and RGS9. The major conclusion is instead that the D260A and C263Y mutations, while not located at the sites directly involved in forming the Gβ5/DEP interface, have the potential of significantly disrupting the conformation of the DEP-domain interaction loop and thus modulating the overall interaction.

DISCUSSION
In this study we provide insights into the molecular mechanisms involved in posttranslational control of R7 RGS protein expression. Volatility of R7 RGS expression in the nervous system is well documented. In neurons, the levels of RGS7 protein are up-regulated by stimulation of α2 adrenergic (40), 5-HT2A (41) and TNFα1 (42) receptors. Similarly, levels of the best studied R7 RGS protein, RGS9 have been reported to change upon administration of addictive drugs such as cocaine (16), acute morphine (15) and amphetamine (43). Furthermore, alteration in striatal dopamine signaling in Parkinson's patients also lead to the elevation of RGS9 expression (14). We have recently reported that RGS9 expression is sensitive to changes in membrane potential and oxygenation (44). In several cases, it has now been documented that changes in R7 RGS protein levels occur as the result of regulated proteolytic degradation. For example, TNFα1 signaling upregulates RGS7 levels by preventing proteasome-dependent degradation of RGS7, (Benzing et al., 1999), and downregulation of RGS9 protein levels occur as the result of rapid proteolytic degradation triggered by calcium entry (44). Increased susceptibility of R7 RGS proteins to proteolytic degradation has been shown to involve posttranslational modifications such as ubiquitination (27,42) and phosphorylation (42,45) and/or recruitment of destabilizing factors such as the Hsc70 chaperone (46) that could serve to target RGS9 to lysosomal cysteine proteases for degradation (10). Thus it appears that regulation of R7 RGS protein stability serves as a general mechanism by which neurons dynamically adjust the levels of these signaling molecules.
It is now well recognized that the stability of R7 RGS proteins requires their association with their binding partners R7BP (or its homologue R9AP) and Gβ5. Binding to R9AP or the homologous protein R7BP (RSBP-1 in C. elegans) is required for the stable expression of RGS9, RGS11 and EAT-16 (9,28,47). This regulation is rather selective because R9AP/R7BP/RSBP-1 do not influence the expression of RGS6, RGS7 and EGL-10. In contrast, complex formation with Gβ5 was found to be essential for the stability of all R7 RGS proteins, both in C. elegans (12) and mammals (11). Although this apparent hierarchical relationship suggests that Gβ5 is critical for the stability of R7 RGS proteins, it has remained unclear whether regulated association with Gβ5 has any role in the dynamic regulation of R7 RGS protein stability. The prevailing view has been that the association between R7 RGS and Gβ5 is constitutive in nature and that in the absence of Gβ5, GGL-containing RGS proteins simply fail to fold properly (48,49) and are thus subject to destruction by the proteolytic systems that serve as the quality control checkpoint in the assembly of obligate oligomeric complexes. In this view, the Gβ5 and R7 RGS proteins are obligatorily and constitutively associated in exactly the same manner as conventional G protein β and γ subunits, with these associations depending on analogous coiled-coil interactions between Gβ5 and the GGL (G gamma like) domain of R7 RGS, or the conventional Gβ and G γ subunits, respectively. In this report we challenge this view by demonstrating in both C. elegans and mammals that association of the GGL domain with Gβ5 is not sufficient for providing proteolytic protection to the R7 RGS complexes. We identified and characterized specific point mutations in Gβ5 that do not affect the coiled-coil interaction with the GGL domain, do not affect Gβ5 assembly into complexes with R7 RGS and R7BP proteins, but do nevertheless abolish the proteolytic stabilization that normally results from the interaction by perturbing the Gβ5-DEP interface.
From unbiased genetic screens for C. elegans mutants with the hyperactive phenotype characteristic of increased Gα q signaling, we identified and characterized three independent mutations in C. elegans Gβ5 that result in degradation of the EAT-16 R7 RGS protein that inhibits Gα q . These screens selected for a very special class of Gβ5 mutations, since seeing the hyperactive phenotype requires that the mutant Gβ5 retains the ability to form a functional complex with the second R7 RGS protein in C. elegans, EGL-10. Our western analysis shows that two of the three mutations selected also partially destabilized EGL-10, just not to the same extent that they destabilized EAT-16.
The special Gβ5 mutations selected in the C. elegans genetic screens affect the conformation of the so called "protein interaction" interface in Gβ5 (50,51). This surface is formed by the loops connecting individual WD-40 repeat units. Interestingly, the analogous surface in conventional Gβ subunits serves as a critical contact point that mediates βγ interactions with most of its partners including α subunits (52,53), effectors (51) and regulatory proteins, e.g. phosducin (54). Moreover, analysis of the RGS9 Gβ5 crystal structure indicates that many of the key residues involved in these interactions are also conserved in Gβ5 but are instead involved in binding to the DEP domain of RGS9 (13). The interactions of conventional Gβ subunits via the protein interaction interface are all reversible binding interactions in which Gβ is required to repeatedly bind and then dissociate from its partners as signaling is initiated or terminated. Our modeling of the Gβ5-DEP interaction suggests a relatively weak binding interface that would similarly allow binding at this interface to be reversible, raising the possibility that R7 RGS complexes undergo conformational rearrangements in which this interface opens up. Indeed, (35)  In addition to engaging in the interaction with Gβ5, the DEP domain of R7 RGS proteins also mediates binding to the membrane-targeting subunits R9AP/R7BP/RSBP-1, an interaction that can also affect stability of the complex (9,30,55,56). Given that R9AP/R7BP/RSBP-1 protect only some R7 RGS members from proteolysis (10,27) and that all R7 RGS proteins as well as R7BP are severely destabilized when Gβ5 is eliminated (10,11,57), we propose a model in which the primary stabilizing effect is provided by the interaction of Gβ5 with the DEP domain. The unique organization of the DEP interface of the "unstable" R7 RGS proteins (RGS9, RGS11 and EAT-16) might not allow Gβ5 to fully engage in the interaction without the assistance of R9AP/R7BP/RSBP-1, whose function might be to merely modulate the Gβ5 DEP interaction. Indeed, in biochemical experiments using stringent wash conditions, Gβ5 binds to the DEP domain of RGS7 but not to that of RGS9 (35). Consistent with this model, mutations destabilizing the Gβ5 DEP interface in C. elegans appear to have more pronounced effects on the stability of the R7 RGS protein EAT-16 that requires assistance from R7BP/RSBP-1. This phenomenon led to the imbalance between EAT-16 and EGL-10 protein levels that allowed us to identify Gβ5 DEP interaction mutations in the first place.
The results of our study allows us to conclude that the Gβ5 DEP interface of the R7 RGS complexes is a conserved "hot-spot" that is critically involved in setting the stability of R7 RGS complexes to proteolytic degradation. It prompts speculation that a range of signaling events known to modulate R7 RGS function and protein levels do so at least in part by affecting the Gβ5 DEP interface through conformational changes, post-translational modifications, or the recruitment of various host factors. The percentage of eggs laid at each developmental stage by the indicated genotypes. egl-10 XS indicates a strain carrying the nIs51 transgene that overexpresses EGL-10 from a transgene carrying multiple copies of the egl-10 gene [17]. Asterisks indicate significant differences (p≤0.05) between the percent of early stage eggs laid by bracketed strains. When eat-16(null) or gpb-2(hyp) mutations are combined with an egl-10(null) mutation, wild-type egg-laying behavior is restored. Figure 3. Western analysis of EGL-10, EAT-16 and GPB-2 protein levels in whole worm lysates of the indicated genotypes. EGL-10 runs just above 66kDa, the EAT-16 antibody detects a single specific band just below 66 kDa and a higher cross-reactive band (X), GPB-2 runs as a doublet at ~40kDa, UNC-64 runs at ~33 kDa and β-Tubulin runs at ~55k Da. EAT-16 is virtually absent in all gpb-2 mutants, whereas EGL-10 levels are higher in gpb-2(hyp) mutants than in the gpb-2(null).   the DEP domain and the GGL domain. Binding to R7BP, which is important for the proteolytic stability of the complex, is mediated by the DEP domain of RGS9. (B and C) Full-length RGS9 can associate with both wild-type and mutant Gβ5 subunits as shown by co-immunoprecipitation assays. Indicated constructs were co-transfected into HEK293 cells and proteins were immunoprecipitated with antibodies against either RGS9 (forward IP, panel B) or the AU5 tag engineered in Gβ5 constructs (reverse IP, panel C). (D and E) Mutations in Gβ5 do not detectably affect its binding to N-or C-terminal domains of RGS9. Indicated constructs containing either N-terminal (D) or C-terminal (E) fragments were separately co-transfected with wild-type or mutant Gβ5 in HEK293 cells and interactions were followed by immunoprecipitation assays. The control lanes shown indicate that the low levels of endogenous Gβ5 present in the HEK293 cells did not interfere with these assays. (F) pull-down assays using recombinant GST-tagged R7BP show the interaction of full-length recombinant RGS9 bound to Gβ5 mutants is intact. Schematic diagram of the R7 RGS/Gβ5 assembly process. Gβ5 folding and assembly with RGS7 requires assistance from CCT and PhLP1. Gβ5 is folded on chaperonin CCT, and the co-chaperon protein, PhLP1, binds to Gβ5 to stabilize the complex until Gβ5 is fully folded. (B) Wild-type and mutant Gβ5 co-immunoprecipitate with CCTε equally well. (C) Wild-type and mutant Gβ5 bind to PhLP1 with similar efficiency.