Figure 4. Membrane targeting strategies employed by multi-domain RGS proteins.
(A) The R7 RGS proteins form obligate heterodimers with Gbeta5 via a Ggamma-like sequence (the "GGL" domain) N-terminal to the RGS-box [37]. This GGL/Gbeta5 interaction could allow R7 RGS proteins to act as conventional Gbeta/gamma subunits in coupling Galpha subunits to 7TM receptors, thereby localizing RGS-box-mediated GAP activity to particular receptors [44]. The DEP domain of RGS9-1 interacts with a membrane-anchoring protein (R9AP) [47]; analogous interactors may exist for the DEP domains of other R7 subfamily members [89].  (B) The PDZ domain of RGS12 is able to bind the C-terminus of the IL-8 receptor CXCR2 (at least in vitro) [57]. The RGS12 PTB domain binds the synprint (“synaptic protein interaction”) region of the N-type calcium channel (Cav2.2); this interaction is dependent on neurotransmitter-mediated phosphorylation of the channel by Src [29].

2.a. R7 RGS proteins as novel Ggamma subunits

     In 1998, we identified a polypeptide sequence, N-terminal to the RGS-box within RGS6, RGS7, and RGS11, with similarity to conventional Ggamma subunits [27]. This Ggamma-like or "GGL" domain was subsequently shown by us [27, 37, 38] and others [39, 40, 41] to bind the neuro-specific outlier Gbeta subunit: Gbeta5. This constitutive GGL/Gbeta5 interaction was also found to hold true for the C. elegans counterparts:  the R7 subfamily RGS proteins EGL-10 and EAT-16 each form obligate dimers with the Gbeta5-homolog, GPB-2 [42, 43]. This GGL/Gbeta5 pairing presents the possibility that R7 RGS proteins not only serve as GAPs for activated Galpha subunits, but also serve to couple inactive Galpha subunits to 7TM receptors (Fig. 4A) akin to the function of conventional Gbeta/gamma subunits (Fig. 1) (reviewed in [44, 45]).

    R7 RGS proteins also have an N-terminal DEP (Dishevelled/EGL-10/Pleckstrin homology) domain [46]. At least for the retinal-specific R7 RGS protein RGS9-1, a membrane-associated binding partner has been identified for the DEP domain: "RGS9 anchor protein" or R9AP [47, 48, 49, 50]. Proper functioning of RGS9-1 as a GAP for the Galpha coupled to the retinal photoreceptor (rhodopsin), as well as proper membrane targetting of this GAP activity by R9AP, appear critical to normal vision; Dryja and colleagues have recently reported that loss-of-function mutations to RGS9-1 or R9AP are found in people with abnormalities in photoresponse recovery ("bradyopsia") that include an inability to see moving objects accurately, especially in low-contrast lighting, and difficulty adjusting to light intensity changes [51]. Although this is the first identified human mutation in an RGS protein causing pathological changes to the timing of 7TM signaling, we surmise that it will not be the last. Indeed, one could speculate that a component of essential hypertension might be due to loss-of-function mutation to RGS2; we and others have recently found that Rgs2-deficient mice [52] exhibit constitutive hypertension [53, 54], consistent with an earlier finding that RGS2 establishes an important negative feedback circuit on vasoconstrictive hormone signaling in vascular smooth muscle as mediated by 7TM receptors coupled to Gq heterotrimers [55].

2.b. Membrane targeting of other RGS proteins

    With the sheer number of RGS proteins identified (e.g., at least 37 RGS proteins encoded by the human genome; Fig. 3), a central question has arisen as to how (and if) receptor selectivity is engendered for the GAP activity of specific RGS proteins. Similar to the relationship between RGS9‑1 and R9AP, other mechanisms of localizing RGS proteins to membranes and/or specific 7TM receptors are being uncovered. For example, we identified an N-terminal PDZ (PSD-95/Discs-large/ZO-1 homology) domain within the longest isoform of the R12 subfamily member RGS12 (Fig. 2); this PDZ domain is capable of binding peptides derived from the C‑termini of 7TM receptors, including from one of the interleukin-8 receptors (CXCR2) [56, 57]. With our colleague Dr. María Diversé-Pierluissi, we have also shown that, in dorsal root ganglion neurons, the phosphotyrosine-binding (PTB) domain of RGS12 mediates its recruitment to the alpha-1B pore-forming subunit of the N‑type calcium channel (Cav2.2) in a neurotransmitter- and phosphorylation-dependent manner (Fig. 4B) [29]. We have since mapped the RGS12 PTB docking site to the SNARE-binding or “synprint” region of the Cav2.2 channel (Siderovski & Diversé-Pierluissi; manuscript submitted); the channel/PTB domain interaction, while phosphotyrosine-dependent, does not occur within a canonical Asn-Pro-X-(p)Tyr binding motif common to many PTB docking sites [58]. With the ability to interact with a multitude of proteins by virtue of its PDZ, PTB, and Ras-binding domains, along with G-alpha interactions via its RGS-box and GoLoco motif (described below), RGS12 in particular appears to be a signaling nexus or ‘hub’ capable of coordinating signal transduction from receptor and/or non-receptor tyrosine-kinases and both monomeric (Ras-superfamily) and heterotrimeric G-protein subunits [13].

    An important finding in the RGS field was made by Wilkie and colleagues when they observed receptor selectivity of RGS proteins [59]. As an example, RGS1 was observed to be a 1000-fold more potent inhibitor of carbachol- than cholecystokinin-stimulated Ca2+ mobilization in pancreatic acinar cells, despite both agonists having similar coupling profiles (i.e., via Gq/11 family G-alpha subunits signaling to phospholipase C-beta) [60]. Strikingly, the closely related RGS2, also a potent GAP for Galpha-q/11 family members [61], was equipotent at inhibiting carbachol and cholecystokinin signaling. The molecular determinants of receptor-selective inhibition of G-protein signaling by RGS4 have been delimited to the N-terminal 58 amino acids of this protein [59]. Full-length RGS4 and RGS4(deltaN58) are equipotent GAPs in vitro, however RGS4 is a 10,000-fold more potent inhibitor of muscarininc signaling in the context of pancreatic acinar cells [59]. 

    These early findings by Wilkie et al. helped initiate the concept that domains outside the RGS-box may have significant functional effects on signaling specificity and potency. Studies using ribozyme and RNA-interference (RNAi)-mediated knockdown of endogenous RGS proteins have reinforced this notion of complementary selectivity between GPCRs and RGS proteins. For example, in studies employing A-10 rat aortic smooth muscle cells, Neubig and colleagues found that ribozyme-mediated depletion of RGS3 selectively enhances carbachol signaling via the M3 muscarinic receptor, whereas analogous depletion of RGS5 only potentiates angiotensin II signaling via the AT1a receptor; RGS2-directed ribozyme treatment had no effect on either 7TM receptor signaling pathway [62]. An obvious molecular mechanism for engendering such receptor selectivity would be direct interaction between a 7TM receptor and an RGS protein. The multidomain architecture of RGS proteins (Fig. 2) provides several potential means by which this could occur. However, to date, there is only limited evidence that 7TM receptors directly interact with RGS proteins. As previously mentioned, we found the RGS12 PDZ domain can bind peptides corresponding to the C-terminal tail of the interleukin-8 (CXCR2) 7TM receptor in vitro [56, 57], but this interaction has not yet been tested with full-length proteins in vivo. A more recent report cites a finding of in vitro binding between an intracellular loop of the M1 muscarinic receptor and RGS2 as evidence for direct coupling between the M1 receptor and RGS2 GAP activity [63]. It would be more informative if loss-of-function mutants for the M1-RGS2 interaction (i.e., mutations that do not disrupt G-protein activation or GAP activity) could be created and analyzed in a cell biological context. It must be emphasized that no report has yet demonstrated interaction between a full-length 7TM receptor and a full-length RGS protein in cells.