Structural design principles that underlie the multi-specific interactions of Gαq with dissimilar partners

Gαq is a ubiquitous molecular switch that activates the effectors phospholipase-C-β3 (PLC-β3) and Rho guanine-nucleotide exchange factors. Gαq is inactivated by regulators of G protein signaling proteins, as well as by PLC-β3. Gαq further interacts with G protein-coupled receptor kinase 2 (GRK2), although the functional role of this interaction is debated. While X-ray structures of Gαq bound to representatives of these partners have revealed details of their interactions, the mechanistic basis for differential Gαq interactions with multiple partners (i.e., Gαq multi-specificity) has not been elucidated at the individual residue resolution. Here, we map the structural determinants of Gαq multi-specificity using structure-based energy calculations. We delineate regions that specifically interact with GTPase Activating Proteins (GAPs) and residues that exclusively contribute to effector interactions, showing that only the Gαq “Switch II” region interacts with all partners. Our analysis further suggests that Gαq-GRK2 interactions are consistent with GRK2 functioning as an effector, rather than a GAP. Our multi-specificity analysis pinpoints Gαq residues that uniquely contribute to interactions with particular partners, enabling precise manipulation of these cascades. As such, we dissect the molecular basis of Gαq function as a central signaling hub, which can be used to target Gαq-mediated signaling in therapeutic interventions.

Gα q can interact with its partners via different regions of the Gα subunit. However, the residue-level determinants of Gα q multi-specificity are still not sufficiently understood. Complexes of Gα q with representatives of these partners have been solved, namely Gα q -RGS2 37 , Gα q -RGS8 38 , Gα q -PLC-β3 39 , Gα q -GRK2 40 , and Gα q -p63RhoGEF 41 . Gα q contains two structural domains, the GTPase domain, which is also found in other G proteins, and the α-helical domain, which is unique to Gα subunits. The GTPase domain contains three flexible regions called "switch regions" (Sw I, II, and III), which undergo conformational changes, depending on whether the Gα subunit binds GTP or GDP 42 . These regions contain two residues that are critical for the GTPase reaction -a catalytic arginine in Sw I (Gα q -Arg183) and a catalytic glutamine in Sw II (Gα q -Gln209). In addition, Gα GTPase domains also include an "effector-binding site" that was previously defined as the C-terminal half of Sw II, the α3 helix, and the subsequent loop that connects the latter to the β5 strand 43 . This region was shown to participate in the binding of Gα subunits to effectors, such as the binding of Gα s to adenylyl cyclase 43 . RGS proteins were shown to substantially interact with the Gα GTPase domain, yet were also shown to interact with the Gα helical domain 37,38,[44][45][46] . More recently, it was suggested that the helical domain is a major determinant of the specific interactions between Gα q and RGS2 47 . PLC-β3 was shown by Waldo et al. to engage three Gα q regions in the GTPase domain, Sw I, Sw II and the effector-binding site, while interactions with the Gα q helical domain were not mentioned 39 . This study also suggested that a helix-turn-helix motif at the C-terminus of the PLC-β3 C2 domain determines its binding to Gα q as an effector. On the other hand, a loop that connects EF hands 3 and 4 in PLC-β3 was shown to mediate its GAP function 24,39 . Similar to PLC-β3, p63RhoGEF binds the Gα q effector-binding site via a conserved helix-turn-helix motif 19,39,41 , but also binds the C-terminal region of Gα q 41 . Diversely, while GRK2 also binds to the Gα q effector-binding site, this partner lacks a helix-turn-helix motif and, despite sharing an RGS homology (RH) domain with RGS proteins, GRK2 and RGS proteins were suggested to bind to non-overlapping surfaces of Gα q 40 . Indeed, a precise and quantitative definition of which Gα q residues contribute to the interface with each partner is lacking, as well as a clear-cut delineation of where these interfaces overlap.
Previous mutagenesis studies tested a limited number of Gα q residues, located in the switch regions and the effector-binding site, and showed them to be important for interactions with particular partners. Two residues in Gα q Sw II, three residues in Sw III, and three residues in the effector-binding site were identified as playing roles in PLC-β activation 39,48 . Shankaranarayanan et al. identified two residues in Gα q Sw III, one residue in the α3 helix, and one residue in the Gα q C-terminal region (Tyr356) as important only for activating p63RhoGEF 49 . Site-directed mutagenesis also assigned four Gα q residues in the GTPase domain as being important for binding GRK2 -one residue in Sw I, two residues in Sw III and one residue adjacent to Sw III 50 . This study also showed that mutations in four Gα q helical domain residues impaired GRK2 binding. Finally, Tesmer et al. showed that four Gα q residues in the Gα q effector-binding site were also required for GRK2 binding 40 . Nevertheless, which Gα q residues interact with all partners and which residues interact specifically with only one partner has yet to be defined.
Here, we used structural comparisons and energy-based calculations to produce a comprehensive map of the residue-level determinants of Gα q multi-specificity. We used structure-based Finite Difference Poisson-Boltzmann (FDPB) and burial-based energy calculations to accurately pinpoint which amino acids contribute to the interaction of Gα q with each of its different partners. We further identified unique Gα q regions that specifically interact with GAPs and disparate Gα q regions that interact with effectors. We also identified Gα q regions that contribute to interactions with multiple partners, and particular Gα q residues that specifically contribute to interactions with only one select partner.

Results
Delineation of structurally-similar domains and sub-structures in Gα q partners. Towards analyzing the multi-specificity determinants of Gα q with its partners at the individual residue level, we first characterized and precisely defined the structural building blocks used by Gα q partners to recognize Gα q using structural alignments -to compare the available experimentally-solved complexes of Gα q with GRK2, RGS2, RGS8, PLC-β3, and p63RhoGEF [37][38][39][40][41]51 . The structure-based ECOD classification database classifies GRK2 residues 29-185 and RGS2 residues 69-200 as homologous RGS homology domains (RH domains) belonging to the same structural family (i.e., the same ECOD F-group). On the other hand, Lodowski et al. defined the GRK2 RH domain as two discontinuous segments, the first being a nine-helix bundle (residues 30-185), and the second corresponding to an extended helix (residues 513-547) 52 . Structural alignment of GRK2 and RGS2 showed that the cores of the RH domains in RGS2 and GRK2 are similar. Specifically, GRK2 residues 52-176 and RGS2 residues 81-200 structurally aligned with a root mean square deviation (RMSD) of 2.8 Å (Fig. 1a). In contrast, a helical segment in the N-terminal region of the previously defined GRK2 RH domain (residues 36-52) and the extended helix at the C-terminus of this domain (residues 513-553) have no structural equivalents in RGS2 (Fig. 1a). Aligning the complexes using only the coordinates of Gα q showed that the RH domains of RGS2 and GRK2 indeed interact with distinct regions of Gα q (Fig. 1b). Furthermore, as noted previously 52 , we observed that the GRK2 and the RGS2/8 RH domains use different regions to interact with Gα q . In GRK2, the region encompassing helices α5 and α6 binds Gα q (Fig. 1c), while in RGS2 and RGS8 these helices are peripheral to the interface and helix α7 is an RGS-unique determent of the interaction (Fig. 1d). Taken together, these results show that only the core RH domain of GRK2 (residues 52-176) is homologous to the RGS domains of RGS2 and RGS8 and is relevant to a comparison of interactions with Gα q . Moreover, because of their disparate binding poses, dissimilar Gα q residues are expected to contribute to the binding of RGS proteins and GRK2.
We next compared the structures of Gα q with PLC-β3, RGS2, and RGS8 [37][38][39] , noting that their interfaces with Gα q are indeed structurally dissimilar (Fig. 2a,b). On the other hand, in all three of these interfaces, an asparagine residue (Asn260 in PLC-β3, Asn149 in RGS2 and Asn122 in RGS8) adopts the same orientation (Fig. 2c), interacting with the catalytic glutamine that is essential for GTP hydrolysis, as previously observed 39 . www.nature.com/scientificreports www.nature.com/scientificreports/ Finally, we compared the structures of Gα q with p63RhoGEF and PLC-β3 39,41 . Both of these proteins contain a structurally-similar pleckstrin homology (PH) domain. According to the ECOD database, p63RhoGEF residues 343-490 and PLC-β3 residues 12-146 adopt a PH domain-like fold. While this domain is structurally similar in both proteins, it interacts directly with Gα q in p63RhoGEF, whereas in PLC-β3, this domain is far from the interface with Gα q (Fig. 3a cf. b). On the other hand, a shorter helix-turn-helix motif in both p63RhoGEF and PLC-β3 binds Gα q similarly (Fig. 3c). These structurally-aligned helix-turn-helix motifs include residues 468-490 in p63RhoGEF and residues 852-874 in PLC-β3, binding Gα q at its previously-defined 43 effector-binding site (Fig. 3c).
Our analyses suggest that because of the structural dissimilarities between Gα q partners and the dissimilarities in their binding poses in relation to Gα q , an alternative approach to precisely delineate the multi-specificity determinants of Gα q should be used. This approach involves assessing which residues are common and which are unique to such interactions by focusing on the Gα q side of the interface and by analyzing which Gα q residues contribute to each interaction using a quantitative energy-based approach.  PLC-β3 and RGS domains are structurally dissimilar, except for one asparagine residue in both partners that interacts similarly with Gα q . (a) The complex of Gα q with PLC-β3 (PDB ID: 3OHM). Gα q is shown as in Fig. 1b, PLC-β3 is shown as a magenta ribbon diagram. (b) Superimposition of the complexes of Gα q with RGS2 and RGS8 (PDB IDs: 4EKD and 5DO9, respectively). Gα q is shown as in panel a, RGS2 and RGS8 are shown as blue and light blue ribbon diagrams, respectively. (c) Superimposition of the complexes of Gα q with PLC-β3, RGS2 and RGS8, using only the coordinates of Gα q for the superimposition. One asparagine residue ("Asn", namely Asn260/149/122 in PLC-β3/RGS2/RGS8) adopts the same orientation towards Gα q in all of these structures and interacts with Gα q similarly in all structures -in particular with the Gα q catalytic residue Gln209 (shown in sticks, with hydrogen bonds shown as dashed black lines).

Residue-level mapping of Gα q interactions with individual partners.
To map the individual residues that contribute to the interactions of Gα q with each of its partners, we characterized the five complexes detailed above using an energy-based computational methodology developed previously by our lab [45][46][47][53][54][55] . The FDPB method was used to calculate the net electrostatic and polar contributions (ΔΔG elec ) of each residue within 15 Å of the Gα q -partner interface in each complex. For each residue, we separately calculated the electrostatic contributions from the side chain and/or those originating from the main chain of each residue. Residues that substantially contribute to the interaction were defined as those contributing ΔΔG elec ≥ 1 kcal/mol to the interactions (i.e. twice the numerical error of the electrostatic calculations) 56 . Note that this approach calculates the net difference between the interaction of a residue with its protein partner in relation to its interaction with the water and ions in the solvent, and thereby identifies only residues that are calculated to substantially contribute to binding. Non-polar energy contributions (ΔΔG np ) were calculated as a surface-area proportional term by multiplying the per-residue surface area buried upon complex formation by a surface tension constant of 0.05 kcal/mol/Å 2 . Residues with substantial non-polar contributions were defined as those contributing ΔΔG np ≥ 0.5 kcal/mol to the interactions (namely, more than 10 Å 2 of each protein surface is buried upon complex formation). To reduce false positives and negatives, we applied a consensus approach across comparable biological replicates in multiple PDB structures or across multiple dimers in an asymmetric unit (see Methods and Supplementary Figs S1 and S2), which substantially improved the accuracy of our predictions. Residues thus calculated to contribute substantially to intermolecular interactions were mapped to the structure of each individual protein (Fig. 4).
Our results show that in all of the complexes analyzed, the majority of Gα q residues contribute to interactions with the cognate partners via non-polar interactions (Fig. 4a,c,e,g, Supplementary Fig. S3, Supplementary Tables 1-4). A similar majority of non-polar contributing residues was also observed in PLC-β3 (Fig. 4d), p63RhoGEF (Fig. 4f), and GRK2 (Fig. 4h). In contrast, the majority of RGS residues that contribute to interactions with Gα q do so via electrostatic contributions (Fig. 4b). The electrostatic dominance in interactions of RGS domains with Gα subunits was also observed in interactions of Gα o and Gα i with various RGS domains 46 . The number of Gα q residues that contribute to interactions with RGS2 and RGS8 are 27 and 25, respectively. Complexes with PLC-β3 and p63RhoGEF involve a larger number of Gα q residues, namely 36 and 31 residues, respectively. In contrast, in the complex with GRK2, only 16 Gα q residues contribute to the interaction. While about a quarter of the Gα q residues contributing to interactions with RGS domains are located in the Gα q helical domain, only three Gα q helical domain residues contribute to interactions with PLC-β3, and no contributions with p63RhoGEF and GRK2 originate from the Gα q helical domain.
On the opposing face of these interfaces, the structurally similar helix-turn-helix motifs in PLC-β3 and p63RhoGEF (Fig. 3) contain 12 residues that contribute to interactions with Gα q ; four of these residues are identical and contribute similarly to interactions with Gα q in p63RhoGEF and in PLC-β3 ( Supplementary Fig. S4). As mentioned above (Fig. 2c), Asn260 in PLC-β3 and the corresponding Asn149/122 in RGS2/8 adopt essentially the same orientation and interact similarly with the catalytic Gα q Gln209 residue. Our calculations predict that this residue contributes to interaction with Gα q via side-chain electrostatic and non-polar interactions in all three structures.
Comparison of the multi-specific interactions of Gα q with its different partners. To precisely define the shared and unique determinants responsible for interactions of Gα q with its partners, we compared which Gα q residues contribute to the interaction with each partner (Fig. 5). We thus identified a single residue in the Gα q P-loop that contributes to interactions with RGS proteins, and one or two residues in the Gα q β1  2RGN). p63RhoGEF is shown as a ribbon diagram, with its Dbl homology (DH) domain colored turquoise, its Pleckstrin homology (PH) domain colored dark blue, and its helix-turn-helix motif (residues 468-490) colored light blue. (b) The structure of Gα q -PLC-β3 (PDB ID: 3OHM). PLC-β3 is shown as a magenta ribbon diagram, except for the PH domain that is colored purple. The PLC-β3 helix-turn-helix motif (residues 852-874) is colored blue. (c) Superimposition of the complexes of Gα q with PLC-β3 and p63RhoGEF, using only the coordinates of Gα q for the superimposition. p63RhoGEF is colored turquoise and PLC-β3 is colored magenta, with the helix-turn-helix motifs colored light blue and blue, as in panels a and b, respectively. Gα q in all panels is shown as in Fig. 1b. www.nature.com/scientificreports www.nature.com/scientificreports/ Residues that contribute substantially to interactions in Gα q complexes with RGS2/8, PLC-β3, p63RhoGEF, and GRK2. (a) Gα q residues that substantially contribute to interactions with RGS2 and RGS8. The Gα q -RGS2/8 crystal structures (PDB IDs: 4EKD and 5DO9) were superimposed using the Gα q coordinates. (b) RGS2 and RGS8 residues that substantially contribute to interactions with Gα q . The crystal structures of Gα q -RGS2 (PDB ID: 4EKD) and Gα q -RGS8 (PDB ID: 5DO9) were superimposed using RGS coordinates, shown as gray ribbons. (c) Gα q residues that contribute substantially to interactions with PLC-β3 (PDB ID: 3OHM). (d) PLC-β3 (maroon ribbon) residues that contribute substantially to interactions with Gα q . (e) Gα q residues that contribute substantially to interactions with p63RhoGEF (PDB ID: 2RGN). (f) p63RhoGEF (cyan ribbon) residues that contribute substantially to interactions with Gα q . (g) Gα q residues that contribute substantially to interactions with GRK2 (PDB ID: 2BCJ). (h) GRK2 (pink ribbon) residues that contribute substantially to interactions with Gα q . In all panels, residues that contribute substantially to interactions with the cognate partner are shown as spheres and colored according to the type of energy contribution: side-chain polar/electrostatic and non-polar contributions, magenta; side-chain polar/electrostatic contribution only, red; main-chain polar/electrostatic contribution only, yellow; main-chain polar/electrostatic and non-polar contributions, blue; non-polar contributions only, green. In panels a,c,e, and g -Gα q is shown as a gold ribbon and the cognate partner as a transparent gray molecular surface. In panels b,d,f, and h -Gα q is shown as a transparent gray molecular surface. Figure 5. Comparison of Gα q residues contributing to interactions with different partners. (a) Structural regions in Gα q that can interact with its partners. Gα q is shown as a ribbon diagram colored light orange (GTPase domain) and gold (helical domain). The α3 helix and the subsequent loop are colored maroon. The P-loop is colored magenta and the three switch regions are marked as follows: Sw I, teal; Sw II, blue; and Sw III, purple. The nucleotide is shown as balls and sticks, colored green. (b) Gα q residues that substantially contribute to the interaction with RGS2 and RGS8. (c) Gα q residues that substantially contribute to the interaction with PLC-β3. (d) Gα q residues that substantially contribute to the interaction with p63RhoGEF. (e) Gα q residues that substantially contribute to the interaction with GRK2. Gα q structures (as in the complexes analyzed in Fig. 4) are depicted as gold ribbon diagrams, with partner structures omitted for clarity. Gα q residues that substantially contribute to interactions with each partner are shown as spheres and colored according to the type of energy contribution, as in Fig. 4.  (2019) 9:6898 | https://doi.org/10.1038/s41598-019-43395-0 www.nature.com/scientificreports www.nature.com/scientificreports/ strand, which immediately precedes the P-loop, that contribute to interactions with PLC-β3 and p63RhoGEF. As mentioned above, the Gα q helical domain makes no contributions to interactions with p63RhoGEF or GRK2. Rather, this domain mostly interacts with RGS domains, and in a limited fashion with PLC-β3 (Fig. 5b,c cf. d,e). Residues in the Gα q Sw II contribute to interactions with all partners, while Sw I and III residues only contribute to interactions with RGS domains and PLC-β3 (Fig. 5). Between nine and 12 Gα q residues in the region immediately following Sw III (residues 248-265, termed here the α3 motif, Fig. 5a) only contribute to interactions with PLC-β3, p63RhoGEF and GRK2 (Fig. 5c-e). Several Gα q residues closer to the C-terminus (residues 319-321 and 353-357) only contribute to the interaction with p63RhoGEF (Fig. 5d).
The majority of Gα q residues contribute to interactions with only one partner (Fig. 6a,b). Most residues in Sw II and some residues in Sw III contribute to interactions with both effectors and GAPs (Fig. 6c), while the helical domain and Sw I contribute to interactions only with GAP proteins (Fig. 6c,d). Moreover, the Gα q helical domain contains five residues that contribute to interactions with RGS proteins alone (Fig. 6b,d). Residues in the Gα q effector-binding site do not contribute to interactions with RGS proteins, and true to the name of this site, contribute only to interactions with effectors (Fig. 6d). Overall, there are only two Gα q residues, located in Sw II, that contribute to interactions with all four partners (Fig. 6a,d).
To gain a wider perspective on the Gα family in terms of multi-specific interactions with different partners, we compared the interactions of Gα q analyzed above with the interactions of Gα i with RGS proteins and GoLoco motifs (Fig. 7). Between 25 and 28 Gα i residues contribute to interactions with different RGS proteins (Fig. 7b), while 35-41 Gα i residues contribute to interactions with the GoLoco motifs in RGS14 and LGN (Fig. 7c). Similar to Gα q , the majority of Gα i residues contributing to interactions with RGS proteins rely on electrostatic interactions (Fig. 7b), with Gα i regions interacting with RGS proteins being similar to Gα q regions that engage RGS proteins (Figs 5b cf. 7b). In contrast, the majority of Gα i interactions with the GoLoco motifs involve non-polar interactions (Fig. 7c). Unlike the interactions of RGS domains with either Gα i or Gα q , we found six to eight residues in the Gα i P-loop that contribute to interactions with GoLoco motifs. A third of Gα i residues contributing to interactions with the RGS14 GoLoco motif are located in the helical domain, while a sixth of the residues contributing to the interaction of Gα i with the GoLoco motif of LGN are located in the helical domain ( Supplementary  Fig. S5). Moreover, the majority of the contributing residues in the Gα i helical domain are involved only in interactions with GoLoco motifs (Fig. 7d). On the other hand, the Gα i Sw I and Sw II regions contain numerous residues that contribute to interactions with either the GoLoco motifs or with RGS proteins, while Gα i Sw III makes only limited contributions to interactions with either the GoLoco motifs or the RGS proteins (Fig. 7b,c). Lastly, the Gα i α3 motif contains eight residues that contribute only to interactions with the GoLoco motifs but not with RGS proteins. Figure 6. Multi-specificity analysis of Gα q . (a) Gα q residues that substantially contribute to interactions with its partners, classified according to the number of binding partners interacting with each residue, colored as in the key. Gα q is shown as a gold ribbon. The contributions of RGS2 and RGS8 residues were combined into a consensus map representing both RGS proteins. (b) Gα q residues that uniquely contribute to interactions with only one partner (i.e. those marked with yellow spheres in panel a), shown as spheres and colored according to the identity of the partner with which they interact: p63RhoGEF, orange; PLC-β3, yellow; RGS2/8, cyan; GRK2, light blue. (c) Gα q residues that interact uniquely with GAPs (PLC-β3/RGS proteins) as opposed to non-GAPs (p63RhoGEF/GRK2). Gα q residues that contribute to interactions with these partners are shown as spheres and colored as follows: residues that contribute to interactions with p63RhoGEF and/or GRK2 (contributions to "effectors" only) are colored purple, residues that contribute to interactions with PLC-β3 and/ or RGS2/8 (contributions to "GAPs" only) are colored green, and residues that contribute to interactions with both effectors and GAPs are colored teal. (d) Gα q residues that interact with particular effector combinations. Contributing Gα q residues are shown as spheres and colored as follows: residues that contribute to interactions with all three effectors (p63RhoGEF, GRK2, and PLC-β3) are colored lilac, residues that contribute to interactions with PLC-β3 and with RGS2/8 are colored green, residues that contribute to interactions with PLC-β3 only are colored yellow, residues that contribute to interactions with only p63RhoGEF are colored orange, residues that contribute to interactions with only RGS2/8 are colored cyan and residues that contribute to interactions with all four partners are colored maroon.

Discussion
Our energy-based computational methodology provides a quantitative framework to compare the multi-specific interactions of Gα q with RGS2/8, PLC-β3, p63RhoGEF, and GRK2 at the individual residue level. Our results revealed that only residues from one Gα q region, Sw II, interact with all of its partners, GAPs and effectors alike -suggesting that Sw II is a necessary and central motif for the recognition of the activated state of Gα q by its partners, rather than a major specificity determinant towards a specific partner. The Gα q effector-binding site is also rather promiscuous, interacting with all partners except for RGS proteins, while the Gα q helical domain, Sw I and parts of Sw III only interact with RGS proteins and PLC-β3. Our multi-specificity analysis identified numerous residues across the surface of Gα q that interact with only one partner, such as the Gα q C-terminal region that exclusively interacts with p63RhoGEF and several helical domain residues that only interact with RGS domains. This analysis enables precise manipulation of individual interactions. Our results also show that most of the Gα q effector-binding site interacts only with effectors, as opposed to the Gα q regions that uniquely interact with GAPs. In particular, the Gα q effector-binding site contains eight residues that contributed to interactions with all three effectors -PLC-β3, p63RhoGEF and GRK2; four Gα q positions contribute to interactions with two effectors. The effector-binding site also contains two residues that contributed specifically to interactions only with PLC-β3, three residues that contributed to interactions only with p63RhoGEF, and one residue that contributed to interactions only with GRK2. Therefore, the Gα q effector-binding site also includes specificity determinants for particular Gα q effectors.
Our calculations also delineate which interactions with particular Gα q regions can underlie GAP activity. The Gα q Sw I region contains residues that contribute to interactions only with PLC-β3 and RGS proteins, suggesting a functional role for Sw I interactions with proteins possessing GAP activity towards Gα q . This hypothesis is sustained by a previous study 57 that used FTIR spectroscopy to show that RGS4 interactions with the intrinsic arginine finger of Gα i , which is located in Sw I, are important for GAP activity. We also found that the Gα q helical domain contains seven residues that contribute to interactions with RGS proteins. We note that, in addition to RGS proteins, the Gα q helical domain also contributes to interactions with PLC-β3 -interactions that were not discussed or investigated in previous studies. This suggests that interactions with the Gα q helical domain play a common role in mediating the function of these different GAPs. On the other hand, we observed no interactions between residues in the Gα q effector-binding site and RGS proteins. Relevantly, the extended loop in PLC-β3, which connects EF hands 3 and 4 and was shown to be a critical component for GAP activity 24,39 , interacts mostly with the Gα q Sw I and the N-terminus of Sw II. Taken together, this suggests that the Gα q effector-binding site interacts solely with effectors or with regions responsible for effector activation and does not play a role in interactions with GAPs. On the other hand, no residues in Gα q Sw I or in the Gα q helical domain contribute to interactions with GRK2. Although GRK2 contains an RH domain that is similar to RGS domains in RGS proteins, all GRK2 interactions with Gα q are with Sw II and the effector-binding site. Therefore, while Carman et al. suggested that GRK2 has weak GAP activity towards Gα q 31 , our results did not identify any interactions that might underlie www.nature.com/scientificreports www.nature.com/scientificreports/ a GAP activity of GRK2. Overall, our results suggest that the Gαq effector-binding site interacts solely with downstream effectors, while the Gα q helical domain and Sw I region interact uniquely with GAPs.
Our analysis of Gα q multi-specificity also pinpointed specific determinants responsible for particular partner interactions. Several residues in the Gα q helical domain consist of a unique specificity determinant towards RGS proteins, while only one residue in this domain is specific to PLC-β3. We found eight residues in the Gα q C-terminal region (319-321 and 353-357) that contribute solely to interactions with p63RhoGEF. The role of residues in this Gα q region in mediating interactions with p63RhoGEF is supported by a previous study that mutated Tyr356 in the Gα q C-terminal region and showed that p63RhoGEF binding was impaired 49 . However, our results suggest the Gα q C-terminal region that contributes to specific interactions with p63RhoGEF is more extensive than previously suggested. Furthermore, the Gα q N-terminal region preceding the P-loop also contains two residues that contribute to interactions only with p63RhoGEF, while one residue in this region contributes only to the interaction with PLC-β3. Most of these contributing residues were not investigated previously and represent new Gα q specificity determinants towards these partners.
Finally, we compared the interactions of Gα i and Gα q with multiple partners and showed that interactions with RGS proteins involving the switch regions were nearly identical, while RGS interactions with the helical domain differed substantially. Our analysis showed that RGS proteins contributed essentially the same to interactions with the Sw I and Sw III regions in both Gα q and Gα i , suggesting these regions do not play a role in determining RGS domain specificity towards Gα q and Gα i . This contrasts with a previous study that suggested, based on visual inspection of the crystal structure, that the Gα Sw I and Sw III regions contain key residues responsible for the selectivity of RGS domains for Gα q and Gα i 38 . Relevantly, both GoLoco motifs and RGS domains interacted with Sw I in the Gα subunits. Taken together, these commonalities suggest that proteins whose function involves the guanine nucleotide, possessing either GAP or GDI activity, bind Sw I as part of their function. On the other hand, the extensive interactions of GoLoco motifs with the Gα i helical domain, combined with their interactions with the effector-binding site, are unique to the Gα i sub-family and stand out from the interactions of Gα q with its partners. More general conclusions regarding the multi-specificity determinants of the entire Gα family will require applying the approach used here to additional members.
In summary, the energy-based computational analysis described here presents a precise comparison of Gα q interactions with multiple partners using a common quantitative framework. This framework allows extension of such analyses to other Gα subunits involved in interactions with different partners and to additional multi-specific proteins. Our analysis suggests that multiple Gα q residues contribute to the discrimination between different protein partners, and provides a structural basis for precisely mutating Gα q residues in order to manipulate and unravel its interactions in vivo and in cells. From a wider perspective, our results provide residue-level insight into protein-protein interactions that drive cellular signaling processes and lay the basis for specifically targeting Gα q -mediated signaling in therapeutic interventions.

Data Availability
The datasets generated during or analyzed during the current study are available from the corresponding author on reasonable request.