Ca2+/Calmodulin reverses phosphatidylinositol 3,4, 5-trisphosphate-dependent inhibition of regulators of G protein-signaling GTPase-activating protein activity.

Regulators of G protein signaling (RGS proteins) are GTPase-activating proteins (GAPs) for G(i) and/or G(q) class G protein alpha subunits. RGS GAP activity is inhibited by phosphatidylinositol 3,4,5-trisphosphate (PIP(3)) but not by other lipid phosphoinositides or diacylglycerol. Both the negatively charged head group and long chain fatty acids (C16) are required for binding and inhibition of GAP activity. Amino acid substitutions in helix 5 within the RGS domain of RGS4 reduce binding affinity and inhibition by PIP(3) but do not affect inhibition of GAP activity by palmitoylation. Conversely, the GAP activity of a palmitoylation-resistant mutant RGS4 is inhibited by PIP(3). Calmodulin binds all RGS proteins we tested in a Ca(2+)-dependent manner but does not directly affect GAP activity. Indeed, Ca(2+)/calmodulin binds a complex of RGS4 and a transition state analog of Galpha(i1)-GDP-AlF(4)(-). Ca(2+)/calmodulin reverses PIP(3)-mediated but not palmitoylation-mediated inhibition of GAP activity. Ca(2+)/calmodulin competition with PIP(3) may provide an intracellular mechanism for feedback regulation of Ca(2+) signaling evoked by G protein-coupled agonists.

Ca 2ϩ /calmodulin and PIP 3 may regulate RGS GAP activity to initiate [Ca 2ϩ ] i oscillations evoked by G protein-coupled agonists.
Production and Purification of Recombinant Proteins -G␣ i1 , RGS4, RGS10, GAIP, RGS4, and RGS4 mutants were expressed as His 6tagged proteins and purified using a Ni 2ϩ -nitrilotriacetic acid-agarose (9). RGS16 expression and purification were similar to RGS4 (9). Recombinant RGS1 and RGS2 were kindly provided by Drs. K. J. Blumer, S. Heximer (Washington University), and D. Forsdyke (Queens University), respectively (16). The fragment coding for mutant RGS4 K112E/ K113E was generated by polymerase chain reaction using the oligonucleotide containing the above mutation as a primer and cloned into the pQE60 expression vector (Qiagen) as described for RGS4 (9). RGS4 C95V protein was kindly provided by Dr. E. Ross (UT Southwestern).
Ca 2ϩ /Calmodulin Binding-Concentration of calmodulin was measured spectrophotometrically using the extinction coefficient 3060 M Ϫ1 cm Ϫ1 at 278 nm in presence of 1 mM EGTA (30). Concentration of RGS4 was measured using Bradford reagent from Bio-Rad. Band shift gel analysis of calmodulin binding to RGS4 and RGS2 in 4 M urea, in the presence of 0.1 mM Ca 2ϩ or 2 mM EDTA, was carried out as described (30). RGS binding to calmodulin-agarose was detected by SDS-PAGE of supernatants as follows: 15 l of wet calmodulin-agarose beads (26 g of CaM) washed with buffer (10 mM HEPES, pH 7.4, 0.1 mM CaCl 2 , 1 mM DTT) were pelleted; the supernatant was removed, and the beads were mixed with RGS4 (1.5 nmol) equal in amount to coupled calmodulin in a final volume 30 l of buffer. The suspension was incubated at room temperature with constant agitation. After 30 min, the beads were washed twice with 500 l of buffer. The suspension was brought to the original volume and incubated for 10 min. 500 mM EGTA and 5 M NaCl were added to final concentrations 2 and 50 mM, respectively, and the suspension was incubated for additional 10 min. Finally, 5 l of the SDS-PAGE loading buffer was added to the beads and incubated another 10 min. Equal volume aliquots of supernatant were withdrawn after each incubation step, followed by separation by SDS-PAGE and Coomassie Blue staining.
Fluorescent detection of RGS4-calmodulin interaction was performed using Perkin-Elmer LS 50B and Hitachi F-2000 fluorescent spectrophotometers at 25 Ϯ 0.1°C. Both excitation and emission slit widths were 10 nm. For internal RGS4 tryptophan fluorescence measurements, the excitation and the emission wavelengths were 283 and 336 nm, respectively. Typically, aliquots of CaM solution were gradually added to the solution of RGS4 in the assay cuvette with constant agitation, and after each addition the mixture was equilibrated inside the instrument with the excitation beam shutter closed. The fluorescence measurements were made after 5-15 min. The excitation beam shutter was opened only long enough to get accurate readings. Measurements were repeated until reproducible readings were obtained. Data in Fig. 2 were corrected for dilution (which did not exceed 7%) and for fluorescence of calmodulin alone.
Calmodulin was dansylated as described (31) and extensively dialyzed against 10 mM HEPES, pH 7.4. Fluorescence measurements of dansylated calmodulin (dansyl-CaM) were at 335 (excitation) and 500 nm (emission). All measurements were corrected for background fluorescence observed in control experiments. The Scatchard plot analysis of the fluorescence affinity measurement data indicated a single binding site interaction (or 2 sites with similar affinities) and fit well to linear approximations within the experimental error.
RGS GAP Assays-GAP assays were carried out in a soluble single turnover system with G␣ i1 as described (2,9) with minor modifications. Briefly, 250 -500 nM G␣ i1 in the assay buffer (10 mM HEPES, pH 8.0, 5 mM EDTA, 2 mM DTT) was loaded with [␥-32 P]GTP at 30°C for 20 min. The solution was placed on ice for 5 min, and all subsequent reactions were carried out on ice in a cold room. A drop of RGS protein (5-10 l) was placed onto a wall of reaction tube next to a 5-l drop of solution, 500 mM MgCl 2 , 5 mM cold GTP, and the GAP reaction was started by addition of 175 l of G␣ i1 -GTP solution. The final concentration of free Mg 2ϩ ions in the reaction mix was 3-4 mM. To study the influence of Ca 2ϩ /calmodulin on RGS4 GAP activity with G␣ i1 -GTP, RGS4 was preincubated with calmodulin in 10 mM HEPES, pH 8.0, 1 mM CaCl 2 , 2 mM DTT on ice for 20 min. Then 175 l of G␣ i1 loaded with [␥-32 P]GTP was added with 5 l of 500 mM GTP, 475 mM MgCl 2 , 25 mM CaCl 2 .
Lipid Vesicles-To prepare small unilamellar lipid vesicles (SUVs), chloroform-soluble lipids were evaporated under a stream of nitrogen into a dry film, and resuspended by vortexing in a sonication buffer, 10 mM HEPES, pH 8.0, 0.1 mM EDTA, 2 mM DTT, essentially as described (32,33). SUVs were generated by sonication of a lipid suspension (2 mg/ml total lipid) in a water bath at room temperature for 10 -15 min. To prepare SUVs containing chloroform-insoluble diC16-PIP 3 , the suspension of diC16-PIP 3 in the sonication buffer (10 mg/ml) was added to the dry film of other lipids, vortexed vigorously for 1 min, and sonicated after addition of the appropriate amount of the sonication buffer. Aggregated material was removed from the preparations of SUVs by centrifugation at 10,000 ϫ g for 10 min. The micelles of diC16-PIP 3 were obtained by its brief sonication in a buffer solution (32).
Surface Plasmon Resonance-Surface plasmon resonance measurements were performed using BIAcore 1000 instrument (BIAcore, Inc.) at 25°C. RGS protein was immobilized on the surface of the carboxymethylated dextran chip (CA-5) using standard carbodiimide chemistry in accordance with manufacturer's instructions. Lipid dissolved in the running buffer, 10 mM HEPES, pH 8.0, 150 mM NaCl, 3 mM EDTA, 0.005% surfactant NP20, was injected over the chip surface with a flow rate 5 l/min. At least three different concentrations were used. Control injections were made with a blank chip without coupled RGS. The amount of coupled protein was 2500 (RGS4), 1580 (RGS4 K112E/ K113E), and 2900 response units (RU) (RGS16). The regeneration of the chip after each binding experiment was achieved by injecting 0.01% SDS in running buffer. The data were analyzed using the BIAevaluation 2.1 software (BIAcore, Inc.). The binding curves showed a moderate heterogeneity of both association and dissociation phases. The minor components identified in the analysis of association and dissociation (less than 20% of total signal) were ignored. A relatively fast transition process evident at the beginning of the dissociation phase was not studied. The dissociation constant, K d , was calculated for each sensorgram using kinetic constants k a and k d . The average values are shown in Table IV.

RESULTS AND DISCUSSION
Ca 2ϩ /Calmodulin Binds RGS Proteins-To investigate the role of Ca 2ϩ in feedback regulation of G protein signaling by RGS proteins, we characterized two potential calmodulin binding regions in RGS4 as follows: one in the N-terminal 33 amino acids and another between residues 99 -113, in helixes 4 and 5 of the RGS domain (4Box). These regions contain amphiphatic sequences with bulky hydrophobic residues at certain positions and clusters of positively charged amino acids similar to calmodulin-binding sites in other proteins (Table I;   The amino acid residues used are: Ϫ, negative; ϩ, positive; h, hydrophobic; X, any; -, gap in alignment. found that RGS4 bound calmodulin in a Ca 2ϩ -dependent manner (the complex was dissociated by 2 mM EGTA) in a standard band shift assay for calmodulin binding ( Fig. 1; Ref. 30). RGS4 binding to Ca 2ϩ /calmodulin was corroborated using calmodulin coupled to agarose beads. RGS4 binding to calmodulin-agarose beads required Ca 2ϩ and was stable to buffer washes, but bound RGS4 could be eluted from beads either with EGTA or SDS (Fig. 1C). We found that a previously characterized amphipathic calmodulin-binding peptide from CaM kinase II (34) competed with RGS4 binding to calmodulin-agarose beads (Fig.  1E, 4th and 5th lanes). An RGS4 N-terminal 33 amino acid peptide (P 1-33 ), which conveys high affinity and receptor-selective regulation of G q signaling (12), bound Ca 2ϩ /calmodulin ( Fig. 1D) and competed with RGS4 for binding to calmodulinagarose beads (Fig. 1E, 6th and 7th lanes). By contrast, a scrambled sequence composed of the same amino acids (P 1-33 ) did not bind to calmodulin beads (data not shown) and did not compete with RGS4 binding (Fig. 1E, 8th and 9th lanes). Interestingly, RGS4 apparently formed a heterotrimeric complex with G␣ i1 -GDP-AlF 4 Ϫ and Ca 2ϩ /calmodulin ( Fig. 1F), consistent with their predicted distinct binding sites on RGS4.
We used fluorescence spectroscopy to quantitate binding interactions between RGS4 and Ca 2ϩ /calmodulin. The fluorescence of two tryptophan residues within the RGS domain of RGS4 (4Box) was quenched by titration with Ca 2ϩ /calmodulin (which lacks tryptophan), indicative of RGS4-Ca 2ϩ /calmodulin binding in solution (Fig. 2). A sharp inflection in the titration curve indicates formation of a stable, equimolar complex between Ca 2ϩ /calmodulin and 4Box (20 mM NaCl, 10 mM HEPES, pH 7.4). Upon further addition of Ca 2ϩ /calmodulin to 4Box, the slope of the fluorescence titration curve paralleled that of unbound RGS4 and Ca 2ϩ /calmodulin (calculated as the sum of their fluorescence signals measured separately). This behavior is consistent with the prediction of a single Ca 2ϩ /calmodulinbinding site in the RGS domain. By contrast, in the absence of salt, the fluorescence intensity of RGS4 (Fig. 2) or 4Box (data not shown) did not change at higher molar ratios of Ca 2ϩ / calmodulin, suggesting additional, low affinity calmodulinbinding sites on RGS4. These low affinity interactions are probably electrostatic because they were not detected at higher ionic strength (Fig. 2, 4Box, and data not shown).
In addition to the RGS domain, Ca 2ϩ /calmodulin apparently binds to the N-terminal 33 amino acids of RGS4 because the peptide P 1-33 binds in a Ca 2ϩ -dependent manner both to cal-modulin beads (Fig. 1) and in solution, as detected by twodimensional heteronuclear single quantum coherence NMR using [ 15 N]calmodulin. 2 Although P 1-33 has no residues with fluorescent properties convenient for affinity measurements, we used a dansylated derivative of calmodulin (dansyl-CaM, see Ref. 31) which allowed us to compare the binding properties of P 1-33 and RGS4 in similar conditions. The apparent K d values (Table II) were calculated from Scatchard linear transformations of titration curves (Fig. 3). In the presence of 100 mM NaCl, RGS4 and P 1-33 bind to Ca 2ϩ /dansyl-CaM with similar affinities (K d Ϸ5 M). The affinity of P 1-33 binding within the accuracy of measurements did not change with ionic strength, whereas RGS4 bound Ca 2ϩ /dansyl-CaM almost 5 times stronger than P 1-33 in 20 mM NaCl, consistent with our observations that a decrease in salt concentration strengthened the interaction of RGS4 with Ca 2ϩ /calmodulin beads.
Calmodulin binds to many RGS proteins in a Ca 2ϩ -dependent manner but with different salt dependences. Binding assays indicated that RGS4, RGS16, and GAIP had similar affin-  (C and D). Peptides CaM kinase II and P 1-33 but not P 1-33 compete with RGS4 for Ca 2ϩ /CaM binding. Pairs of lanes for each peptide show the relative yield of RGS4 from supernatants after incubation with beads (s) and after SDS elution from beads (b) (E). Binding of the RGS4-G␣ i1 -GDP-AlF 4 Ϫ complex to Ca 2ϩ /CaM-agarose beads (F). Proteins were visualized on Coomassie-stained PAGE gels.
ities toward calmodulin-agarose beads in 20 mM NaCl, whereas RGS10 interaction with Ca 2ϩ /calmodulin was relatively weak (even without salt). Only RGS1 and RGS2 bound to calmodulinagarose beads in high salt (150 mM KCl). Stable interaction was corroborated by band shift analysis that revealed an RGS2-Ca 2ϩ /calmodulin complex in high ionic strength running buffer with 4 M urea, 275 mM Tris-HCl, pH 8.3 (data not shown). Calmodulin Does Not Influence RGS4 GAP Activity-To test the biological relevance of Ca 2ϩ /calmodulin binding to RGS proteins, we studied its effect on RGS4 and RGS1 GAP activity. We found that neither 1 mM Ca 2ϩ alone nor 1.6 M Ca 2ϩ / calmodulin preincubated with either RGS protein altered their GAP activity toward G␣ i1 in a single turnover assay ( Fig. 4 and data not shown). Because the N terminus of RGS4 is not required for GAP activity (9) but binds calmodulin (Fig. 1, D and  E), we also tested GAP activity of an N-terminal deletion mutant that lacked the first 57 residues of RGS4 but retained the calmodulin-binding site within the RGS domain (R4⌬N; Fig.  4B). The GAP activities of full-length RSG4 and R4⌬N were unaffected by Ca 2ϩ /calmodulin in the single turnover assay (Fig. 4). These results indicated that Ca 2ϩ /calmodulin binding neither sterically blocked G␣ i1 binding nor irreversibly changed the conformation of the RGS4-G␣ i1 interface. Considering the stability of RGS4 binding to Ca 2ϩ /calmodulin beads (Fig. 1), it seems improbable that GAP activity resulted from the rapid and transient displacement Ca 2ϩ /calmodulin from RGS4 by G␣ i -GTP. Indeed, we found that G␣ i1 -GDP-AlF 4 Ϫ , which mimics a transition state of RGS4-catalyzed G␣-GTP hydrolysis (K d ϭ 0.6 nM at 25°C, Ref. 9), binds as a complex with RGS4 and Ca 2ϩ /calmodulin-agarose beads (Fig. 1E). This complex was stable to extensive washing with buffer, but RGS4-G␣ i1 -GDP-AlF 4 Ϫ was eluted from Ca 2ϩ /calmodulin beads by 1% SDS. G␣ i1 -GDP-AlF 4 Ϫ did not bind to Ca 2ϩ /calmodulin in the absence of RGS4. These results indicate that a heterodimeric complex of Ca 2ϩ /calmodulin-RGS4 retains GAP activity. We therefore favor a model in which calmodulin binds surface residues of the RGS domain without substantially altering the RGS4 conformation.
PIP 3 Inhibits RGS GAP Activity-Previous studies indicated that brief dialysis of recombinant RGS4 into patch clamped pancreatic acinar cells potently inhibited Ca 2ϩ signaling evoked by G i -and G q -coupled receptor agonists (12,16). This suggested the possibility that endogenous RGS proteins might be relatively inactive prior to agonist stimulation of Ca 2ϩ signaling and that recombinant RGS proteins escaped this inhibition. To identify inhibitors of RGS4 GAP activity on G␣ i1 in the single turnover assay, we tested various compounds related to and including either the substrate or products of PLC␤. We found that RGS4 GAP activity was inhibited by preincubation with an analog of PIP 3 , diC16-PIP 3 (30 M PIP 3 , at 0°C for several minutes, Fig. 5A), or phosphocholine vesicles containing 20% PIP 3 (Fig. 5B). Inhibition of RGS4 GAP activity was dependent on the concentration of PIP 3 (Fig. 5C). The kinetic curve of GTP hydrolysis in the presence of PIP 3 closely approximated the basal activity of G␣ without RGS4. This behavior indicated negligible dissociation of PIP 3 from RGS4 during the assay (5 min) because GTP hydrolysis was initiated by a rapid 20-fold dilution of the RGS-PIP 3 incubation mix into a solution containing G␣ i1 -GTP. In control experiments, the intrinsic GTPase activity of G␣ i1 was unaffected by PIP 3 (Fig. 5D). The GAP activity of each RGS protein that was tested in the single turnover assay, except RGS16, was inhibited by PIP 3 (Fig. 5D and data not shown). By contrast to PIP 3 inhibition of RGS GAP activity, no effect was observed following coincubation of RGS4 with 400 M PIP 2 from bovine brain (Fig. 5B), and only 2-3-fold inhibition was observed following coincubation with PIP 2 micelles (9 mM). We barely detected the inhibitory activity of 4-mono phosphorylated phosphatidylinositol phosphate, PIP (9 mM). As summarized in Table III, several synthetic PIP 2 lipids were also without effect (assayed at 400 M in phosphocholine vesicles), including dioctanoylphosphatidylinositol 3,4-bisphosphate (diC8 -3,4-PIP 2 ), dioctanoylphosphatidylinositol 3,5-bisphosphate (diC8 -3,5-PIP 2 ), and dioctanoylphosphatidylinositol 4,5bisphosphate (diC8 -4,5-PIP 2 ). RGS4 GAP activity was also not affected by a lipid head group derivative, 1-stearoyl-2-arachidonoyl-sn-glycerol (diacylglycerol, DAG), which is one of the reaction products of PLC␤. No inhibitory activity was detected using PI or highly charged head group derivatives of PIP 3 lacking the fatty acyl moieties, including inositol 1,3,4,5-tetrakisphosphate (IP 4 ) and glycerophosphoinositol 3,4,5trisphosphate (IP 3 , the other reaction product of PLC␤; data not shown). PIP 3 binding to Rac1 and RhoA was shown to have similar requirements for both electrostatic and hydrophobic interactions (35). Surprisingly, we found that dioctanoylphosphatidylinositol 3,4,5-trisphosphate (diC8-PIP 3 ), which only differed from PIP 3 (diC16-PIP 3 ) in the length of the fatty acid chains, did not inhibit RGS4 GAP activity; nor did diC16-3,5-  PIP 2 . Thus, diC16-PIP 3 is the only phospholipid which inhibited RGS4 GAP activity, and its activity appears to require both the long chain fatty acid moiety and the highly charged head group.
RGS4 Is a PIP 3 -binding Protein-Protein-lipid binding affinities were estimated by surface plasmon resonance measurements on the BIAcore (Biacore, Inc). RGS4 coupled to the carboxymethylated dextran surface of the BIAcore chip bound diC16-PIP 3 (Fig. 6A) but not diC8-PIP 3 , PIP 2 , or other phosphoinositides, consistent with the observation that RGS GAP activity was most sensitive to inhibition by diC16-PIP 3 . The association phase of diC16-PIP 3 binding was typically complete within several minutes, whereas dissociation was slow (Fig.  6A). No diC16-PIP 3 binding was detected on a blank chip. The K d value of diC16-PIP 3 binding to RGS4 (44 Ϯ 19 nM; Table IV) was calculated from the on and off rates extracted from the binding curves (Fig. 6A, and data not shown).
Ca 2ϩ /CaM Antagonizes the PIP 3 Inhibition of RGS4 GAP Activity-The positively charged patch on the surface of helixes 4 and 5 in the RGS domain of RGS4 (residues 99 -113; Ref. 26) appeared to be a good candidate for binding not only to calmodulin but also to PIP 3 . We found that PIP 3 inhibition of RGS4 GAP activity was reversed by coincubation of RGS4 and PIP 3 (micelles or phosphocholine vesicles) with Ca 2ϩ /calmodulin (Fig. 7). Ca 2ϩ /calmodulin and PIP 3 apparently compete for binding to helixes 4 and 5 in RGS4. To test this model further, we introduced amino acid substitutions of glutamate for lysine residues at positions 112 and 113 in helix 5 of RGS4. This mutant protein (RGS4 K112E/K113E) retained GAP activity and calmodulin binding, but it bound PIP 3 with almost 10-fold lower affinity than did wild type RGS4 (Table IV). Similarly, low binding affinity was observed with RGS16. The comparatively weak binding of these proteins to PIP 3 correlated with their relative insensitivity to PIP 3 inhibition of GAP activity (Figs. 5D and 6B). Helixes 4 and 5 are conserved in many RGS domains (Table I) and may provide an important regulatory feature of RGS proteins because Ca 2ϩ /calmodulin has been shown to serve as a molecular switch that regulates lipidprotein interactions (19 -21). PIP 3 Inhibits GAP Activity of Palmitoylation-resistant RGS Proteins-We propose that PIP 3 -mediated inhibition of GAP activity acts by a concerted mechanism in which the highly charged head group interacts with the RGS domain residues in helixes 4 and/or 5 to position the palmitoyl moiety of PIP 3 near its binding site. Palmitoylation of a nearby cysteine residue inhibited RGS4 and RGS10 GAP activity (36). This cysteine residue in helix 4 (Cys 95 in RGS4) is conserved in the RGS domain of all mammalian RGS proteins except RGS6 and RGS7 (6). Substitution of cysteine for valine (C95V) prevents covalent modification by palmitate at this position in RGS4. GAP activity of the C95V mutant protein is equivalent to wild type protein, but it is not inhibited by palmitoylation (36). By contrast, the GAP activity of RGS4 C95V is as sensitive to  Ϫ ϩ /Ϫ a 4-PIP (brain) NT ϩ/Ϫ b PI (brain) NT ϩ diC8-3,4,5-PIP 3 Ϫ ϩ diC8-3,4-PIP 2 Ϫ ϩ diC8-3,5-PIP 2 Ϫ ϩ diC8-4,5-PIP 2 Ϫ ϩ Gro-3,4,5-IP 3 NT ϩ 1,3,4,5-IP 4 NT ϩ DAG NT ϩ a Inhibition detectable above 1 mM 4,5-PIP 2 . b Marginal effect at 7 mM PI, ϩ, binding detected or RGS4 GAP activity is normal; Ϫ, binding or GAP not detected; NT, not tested. PIP 3 -mediated inhibition as is wild type protein (Fig. 8A). The GAP activities of RGS16 and the mutant RGS4 K112E/K113E, which were relatively insensitive to inhibition by PIP 3 , responded like wild type protein to inhibition by palmitoylation (Fig. 8, B and C). In contrast to RGS4 interaction with PIP 3 , inhibition of GAP activity by palmitoylation was not prevented by addition of Ca 2ϩ /calmodulin (Fig. 8D). This is presumably because Ca 2ϩ /calmodulin competes with PIP 3 binding to RGS4 but cannot displace the covalent modification of RGS4 by palmitate. The feedback mechanisms that regulate the palmitoylation of RGS proteins in vivo are unknown, but we propose that PIP 3 -mediated inhibition of RGS GAP activity may be reversed in a Ca 2ϩ -dependent manner through binding of Ca 2ϩ /calmodulin.   Feedback Regulation of RGS GAP Activity by Ca 2ϩ /Calmodulin and PIP 3 -Negative regulation of RGS GAP activity by PIP 3 may be part of a reset mechanism that allows a new wave of G protein signaling in response to agonist. G protein-coupled receptors, such as formyl-methionyl-leucyl-phenylalanyl receptors in neutrophils, can elicit both Ca 2ϩ -release and a rapid and large accumulation of PIP 3 by activating the effector proteins PLC␤ and PI-3 kinase, respectively (reviewed in Ref. 37). Ca 2ϩ release from internal stores is one of the initial responses to PLC␤ activation either by G␣ q or G␤␥ from G i class G proteins. As the local concentration of Ca 2ϩ elevates in response to PLC␤ activity, we postulate that Ca 2ϩ /calmodulin binding to RGS4 and other RGS proteins displaces PIP 3 to restore GAP activity (modeled in Fig. 9). Ca 2ϩ /calmodulin binding may also enhance GAP activity by relocating RGS proteins within the receptor signaling complex to be in proximity to their G␣-GTP substrates. Feedback inhibition of G protein-mediated PLC␤ activation by RGS proteins would allow [Ca 2ϩ ] i to decrease and promote the dissociation of calmodulin from RGS proteins. We hypothesize that PIP 3 released after receptor stimulation could then bind RGS proteins to inhibit their GAP activity. If agonist stimulation persists, this may reactivate G protein signaling and allow another burst of Ca 2ϩ release from internal stores. If agonist is no longer present, PIP 3 -mediated inhibition of RGS GAP activity may reset the signaling pathway to allow a robust cellular response to subsequent agonist stimulation. Feedback regulation of RGS GAP activity may provide an intracellular mechanism to initiate oscillations over a wide range of frequencies.
FIG. 9. A model of RGS regulation by Ca 2؉ calmodulin and PIP 3 . Agonist-bound receptor promotes GTP binding to the G␣ subunit and subsequent dissociation of G␣ and G␤␥. The effector protein PLC␤ may be activated by either G␣ q or G␤␥ from G i class proteins. PLC␤ catalyzes the hydrolysis of PIP 2 to produce DAG and IP 3 , which binds IP 3 R to release Ca 2ϩ from intracellular stores. We propose that endogenous RGS proteins may be inactive prior to agonist-evoked Ca 2ϩ signaling, but as the local concentration of intracellular Ca 2ϩ elevates, it binds calmodulin, which can displace PIP 3 from helixes 4 and 5 of RGS proteins, and thereby restores RGS GAP activity. Calmodulin binding may also reposition RGS within the receptor complex to enhance activity. RGS-mediated inhibition of G protein signaling would decrease [Ca 2ϩ ] i allowing dissociation of calmodulin and rebinding of PIP 3 to inhibit RGS GAP activity.