The GTPase-activating protein RGS4 stabilizes the transition state for nucleotide hydrolysis.

RGS proteins constitute a newly appreciated group of negative regulators of G protein signaling. Discovered by genetic screens in yeast, worms, and other organisms, two mammalian RGS proteins, RGS4 and GAIP, act as GTPase-activating proteins for members of the Gi family of G protein α subunits. We have purified recombinant RGS4 to homogeneity and demonstrate that it acts catalytically to stimulate GTP hydrolysis by Gi proteins. Furthermore, RGS4 stabilizes the transition state for GTP hydrolysis, as evidenced by its high affinity for the GDP-AlF4−-bound forms of Goα and Giα and its relatively low affinity for the GTPγS- and GDP-bound forms of these proteins. Consequently, RGS4 is most likely not a downstream effector for activated Gα subunits. All members of the Gi subfamily of proteins tested are substrates for RGS4 (including Gtα and Gzα); the protein has lower affinity for Gqα, and it does not stimulate the GTPase activity of Gsα or G12α.

We demonstrated that two of these proteins, RGS4 and GAIP, act as GAPs on several members of the G i subfamily of ␣ subunits (20). However, the mechanism of this effect and the possibility of other functional relationships between these GAPs and their target G proteins remained unclear. We demonstrate herein that purified recombinant RGS4 acts catalytically to accelerate the GTPase activity of G i protein family members, that RGS4 has high affinity for the transition state complex of the G protein ␣ subunit bound to GDP-AlF 4 Ϫ (but low affinity for GTP␥S-␣ and GDP-␣), and that the rank order of affinities of RGS4 for transition state complexes of ␣ subunits is G i family Ͼ G q Ͼ Ͼ G s . RGS4 has no detectable GAP activity toward G s␣ or G 12␣ .

EXPERIMENTAL PROCEDURES
Purification and Activation of Proteins-G s␣ , G i␣1 , and G o␣ were expressed and purified as described (21), as were G z␣ and G 12␣ (22); G i␣1 and G o␣ were myristoylated unless stated otherwise. Bovine retinal G t␣ was the generous gift of Dr. Heidi Hamm (University of Illinois College of Medicine), while recombinant G q␣ was supplied by Dr. John Hepler (this laboratory) (23). G o␣ was activated with 100 M GTP␥S and 10 mM MgSO 4 by incubation at 30°C for 60 min. The indicated GDP-bound ␣ subunits were activated with AlF 4 Ϫ by incubation with 20 M AlCl 3 , 10 mM NaF, and 10 mM MgSO 4 for 10 min on ice.
RGS4 (hexahistidine-tagged at the amino terminus) was synthesized in Escherichia coli as described previously (20), except that cells were incubated for 12 h at 30°C after induction. The bacterial lysate was purified on Ni-NTA resin (Qiagen) as described, except that 500 mM NaCl was included in the first wash. Ammonium sulfate (1.2 M final concentration) was added to the Ni-NTA column eluate, and this solution was applied to a 15-ml Phenyl-Sepharose FPLC column equilibrated with 50 mM Tris-HCl (pH 8), 2 mM dithiothreitol, and 1.2 M (NH 4 ) 2 SO 4 . The column was washed with 100 ml of equilibration buffer and eluted with a 100-ml continuous gradient of 50 mM Tris-HCl and 1.2 to 0.84 M (NH 4 ) 2 SO 4 . Fractions were analyzed electrophoretically, and protein was pooled and dialyzed into 50 mM NaHepes (pH 8), 1 mM EDTA, and 2 mM dithiothreitol prior to concentration and storage. The yield was 4 mg/liter of bacterial culture. Purified RGS4 was homogeneous based on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (see Fig. 5), had the anticipated amino-terminal amino acid sequence, and displayed a single mass spectroscopic peak (M r ϭ 24,183; predicted, 24,188).
GAP Assays-GAP assays were performed using G o␣ as the substrate as described previously (20). This assay relies on formation of a complex between G o␣ and [␥-32 P]GTP in the absence of Mg 2ϩ and subsequent initiation of nucleotide hydrolysis by incubation of this complex at 4°C with Mg 2ϩ and RGS4. A single round of hydrolysis of GTP to GDP is then monitored over 5 min by quantification of release of 32 P i .
Direct measurement of the k cat for GTP hydrolysis by G 12␣ and G z␣ required an alternative protocol because of their slow rates of nucleotide exchange. G 12␣ and G z␣ were incubated with 5 M [␥-32 P]GTP (20 -50 cpm/fmol) and 5 mM EDTA at 30°C for 30 min. Samples were then gel-filtered at 4°C through a Sephadex G-50 spin column (Pharmacia) equilibrated with 50 mM NaHepes (pH 8), 1 mM dithiothreitol, 5 mM EDTA, and 0.1% C 12 E 10 to remove free GTP and P i . GTPase activity was measured at 15°C in 50 l of the same solution with addition of 8 mM MgSO 4 , 1 mM GTP (unlabeled), and the indicated concentration of RGS4.

RESULTS AND DISCUSSION
RGS4 Acts Catalytically-We reported previously that RGS4 and GAIP accelerate the rate of the GTPase reaction catalyzed by selected members of the G i subfamily of ␣ subunits by at least 40-fold; the concentrations of RGS proteins utilized in that study were in excess of those of the G ␣ protein substrates. Other GAPs have been shown to act catalytically (4, 24 -26), and the same is true of RGS4. Thus, concentrations of RGS4 as low as 2.9 nM were sufficient to enhance the GTPase activity of 70 nM GTP-G o␣ (Fig. 1A), and the initial slopes of the curves in   catalyzed the turnover of 6 mol of GTP-G o␣ /min at a substrate (GTP-G o␣ ) concentration that is probably more than 20-fold below the K m of GTP-G o␣ for RGS4. It would be necessary to saturate the substrate with RGS4 to estimate the maximal enhancement of the rate of GTP hydrolysis; the rate of hydrolysis of GTP is then too fast to measure with the methods used herein to monitor release of 32 P i . The maximal enhancement of the rate of GTP hydrolysis observed in this experiment was greater than 40-fold. Variation of the substrate (GTP-G o␣ ) concentration permitted estimation of the K m for the interaction of GTP-G o␣ with RGS4 and the maximal rate of turnover of GTP-G o␣ by RGS4 (Fig. 1B). These values are 2.5 M and 14/s, respectively, but should be considered estimates, since the substrate concentration was varied over a relatively limited range. This turnover rate could be limited by the maximal intrinsic rate of GTP hydrolysis by G o␣ .
RGS4 Stabilizes the Transition State for GTP Hydrolysis-The crystal structures of the GDP-AlF 4 Ϫ complexes of G i␣1 (27) and G t␣ (28) revealed that AlF 4 Ϫ occupies the position normally taken by the ␥-phosphate of GTP but, remarkably, that the fluorine atoms assume a square planar configuration about the central aluminum atom, in contrast to the tetrahedral geometry of a phosphate group. The AlF 4 Ϫ complex is octahedrally coordinated to a ␤-phosphate oxygen and to the putative hydrolytic water molecule. Furthermore, an Arg residue and a Gln residue that are known to be critical for catalysis are dramatically reoriented in the GDP-AlF 4 Ϫ structure (compared to their positions in the GTP␥S-bound protein), contacting the fluorine atoms and the hydrolytic water. These facts (and oth- Inset, the initial rates of P i release are plotted against the amount of RGS4. B, varying concentrations of GTP-G o␣ were incubated with 2.1 nM RGS4 and 15 mM MgSO 4 at 4°C, and the linear release of 32 P i was monitored over 1 min (7 time points). These initial rates are plotted against the concentration of GTP-G o␣ (determined by the maximal amount of P i released). Inset, Lineweaver-Burk analysis of these data. These data represent one of two similar experiments.

ers) argue that GDP-AlF 4
Ϫ is not a GTP analog but rather mimics the trigonal-bipyramidal species that is presumed to appear at or near the transition state of the S N 2 reaction.
These considerations have prompted examination of the interactions of RGS4 with the GDP-, GTP␥S-, and GDP-AlF 4 Ϫbound forms of G o␣ by testing the capacity of these proteins to compete with GTP-G o␣ for its GAP. The basic assay is seen in Fig. 2A. Addition of 2 M GDP-G o␣ or GTP␥S-G o␣ to 200 nM [␥-32 P]GTP-G o␣ had little or no observable effect on nucleotide hydrolysis stimulated by 29 nM RGS4. However, 2 M GDP-AlF 4 Ϫ -G o␣ completely blocked the effect of RGS4, presumably by formation of a high affinity complex with the GAP. Free guanine nucleotide or the combination of AlCl 3 , NaF, and MgSO 4 had no effect on nucleotide hydrolysis in these assays.
Competition assays (at 150 nM GTP-G o␣ ) with different concentrations of GTP␥S-G o␣ or GDP-AlF 4 Ϫ -G o␣ highlight the preferential affinity of RGS4 for the transition state conformation of the G protein (Fig. 2B). Stimulated GTP hydrolysis is nearly completely inhibited at concentrations of GDP-AlF 4 Ϫ -G o␣ that approximate those of RGS4. The apparent K d of RGS4 for GDP-AlF 4 Ϫ -G o␣ is thus below 100 nM. Again, GTP␥S-G o␣ is a poor competitor (Fig. 2B), consistent with the K m for GTP-G o␣ estimated in Fig. 1B, as were the GTP␥S-bound forms of G z␣ , G i␣1 , and G i␣3 (not shown).
RGS4 Interacts with G i␣ Subfamily Members and with G q␣ but Not with G s␣ or G 12␣ -The competition assay just described was utilized to examine the interactions of RGS4 with other G protein ␣ subunit family members. We demonstrated previously that RGS4 stimulates GTP hydrolysis by G i␣1 , G i␣2 , G i␣3 , and G o␣ . G i␣1 and G o␣ were similarly effective in the competition assay as the GDP-AlF 4 Ϫ -bound species (Fig. 3). Two additional G i subfamily members were also tested for interactions with RGS4. GDP-AlF 4 Ϫ -G t␣ was an effective competitor (Fig. 3), indistinguishable from G i␣1 and G o␣ . The k cat for GTP hydrolysis by G z␣ was measured directly and was increased from its normal low value of 0.02/min to 0.12/min by 12 nM RGS4 (Fig.  4A). It thus seems likely that the GTPase activity of all G i subfamily members is stimulated by RGS4. We have attempted direct estimation of affinities between GDP-AlF 4 Ϫ -bound ␣ sub-units and RGS4 by observation of surface plasmon resonance (Pharmacia Biosensor). Appropriate protein-protein interactions were detected, but the rate of dissociation of hexahistidine-tagged RGS4 from the Ni-NTA-derivatized chips of the Biosensor instrument was faster than the rate of dissociation of GDP-AlF 4 Ϫ -G o␣ from RGS4. Thus, we can only estimate an upper limit for the K d from these experiments, roughly 100 nM.
We also noted previously the inability of RGS4 (or GAIP) to stimulate GTP hydrolysis by G s␣ ; accordingly, the GDP-AlF 4 Ϫ bound form of G s␣ did not compete with G o␣ (Fig. 3). Of interest, GDP-AlF 4 Ϫ -G q␣ did interact with RGS4, although its apparent affinity for the protein is 10-to 100-fold or more lower than are those of the G i subfamily members. However, the assumption that GDP-AlF 4 Ϫ -bound complexes of all G protein ␣ subunits are transition-state mimics may be unwarranted. G q must be reconstituted with an appropriate receptor to examine the effect of RGS4 on nucleotide hydrolysis; the rate of nucleotide exchange is too low to permit preparation of GTP-G q␣ substrate by the present methods. A high concentration of RGS4 did not stimulate GTP hydrolysis by G 12␣ , which proceeded with a k cat of 0.06/min at 15°C (Fig. 4B).

RGS4 Forms a High-affinity Complex with GDP-AlF 4
Ϫ -G i␣1 -The data presented above strongly imply formation of a high affinity complex between RGS4 and the GDP-AlF 4 Ϫ -bound forms of various G i␣ proteins. To demonstrate this directly and to prepare material for crystallographic analysis, we have performed gel filtration chromatography on mixtures of RGS4 and nonmyristoylated G i␣1 ; the latter protein has been crystallized previously in various conformations. Gel-filtered separately, G i␣1 and RGS4 elute from a Superdex 200 column at positions consistent with their monomeric molecular weights (Fig. 5A). After incubation of GDP-AlF 4 Ϫ -G i␣1 with a modest molar excess of RGS4 on ice for 10 min, the gel filtration profile reveals formation of a complex of the two proteins, migrating with an apparent molecular weight of 70,000; free, excess RGS4 is also seen in its monomeric position (Fig. 5B). The GTPase activity of the Gln 204 3 Leu mutant of G i␣1 is impaired severely. This protein does not interact with AlF 4 Ϫ , and its GTPase activity is not affected by RGS4 (20,27). Consistent with these observations, this protein does not form a high affinity complex with RGS4 when incubated with AlCl 3 , NaF, and MgSO 4 (Fig. 5C); the elution profiles of the two proteins correspond to those observed in Fig. 5A.
We conclude that RGS4 has a relatively low affinity (greater than 1 M) for its substrates, GTP-G i␣ family members, but interacts directly and with high affinity with the transitionstate conformations of these ␣ subunits. Stabilization of the transition state lowers the activation energy barrier for hydrolysis of GTP, accounting for the large rate enhancements that are seen. We surmise that RGS4 is not an effector for G i␣ proteins, based on the relatively poor affinity of RGS4 for their GTP␥S-bound forms compared to those of other known effectors of G protein ␣ subunits. We note, however, that some other RGS proteins are considerably larger than are RGS4 or GAIP, and generalization of this point may be unwarranted. Finally, we call attention to the recent observation by Mittal et al. (29) that wild-type p21 ras protein interacts with AlF 4 Ϫ only in the presence of its GAPs. Thus, the ground state of p21 ras must be too distant, conformationally, from the transition state to recognize AlF 4 Ϫ (unlike heterotrimeric G proteins), but GAPs for low molecular weight GTPases and RGS proteins both appear to act predominantly by stabilizing the transition states for nucleotide hydrolysis and not by elevating the energy level of the enzyme-substrate complex.

FIG. 5. RGS4 and GDP-AlF 4
؊ -G i␣1 form a high-affinity complex. A, RGS4 (2.2 mg) or G i␣1 (2.2 mg; activated with 20 M AlCl 3 , 10 mM NaF, and 5 mM MgSO 4 for 10 min at 4°C) was applied to a 16/60 Superdex gel filtration column equilibrated and eluted with 50 mM NaHepes (pH 8), 1 mM EDTA, 10 mM NaF, 20 M AlCl 3 , 5 mM MgSO 4 , 2 mM dithiothreitol, and 50 mM NaCl. Fractions (1.5 ml) were collected, and A 280 was monitored. Arrows indicate the position of molecular weight standards. B, as in A, except that RGS4 and GDP-AlF 4 Ϫ -G i␣1 were incubated together prior to gel filtration. Fractions were analyzed electrophoretically, and gels were stained with Coomassie Brilliant Blue. The higher molecular weight band is G i␣1 ; the lower is RGS4. C, as in B, except the Gln 204 3 Leu mutant of G i␣1 was substituted for the wild type protein.
RGS4 Stabilizes the Transition State of G o␣ 27212