Partial G protein activation by fluorescent guanine nucleotide analogs. Evidence for a triphosphate-bound but inactive state.

N-methyl-3'-O-anthranoyl (MANT) guanine nucleotide analogs are useful environmentally sensitive fluorescent probes for studying G protein mechanisms. Previously, we showed that MANT fluorescence intensity when bound to G protein was related to the degree of G protein activation where MANT-guanosine-5'-O-(3-thiotriphosphate) (mGTP gammaS) had the highest fluorescence followed by mGTP and mGDP, respectively (Remmers, A. E., Posner, R., and Neubig, R. R. (1994) J. Biol. Chem. 269, 13771-13778). To directly examine G protein conformations with nucleotide triphosphates bound, we synthesized several nonhydrolyzable MANT-labeled guanine nucleotides. The relative maximal fluorescence levels observed upon binding to recombinant myristoylated Goalpha (myrGoalpha) and myrGialpha1 were: mGTPgammaS > MANT-5'-guanylyl-imidodiphosphate > MANT-guanylyl-(beta,gamma-methylene)-diphosphonate > MANT-guanosine 5'-O-2-(thio)diphosphate. Using protection against tryptic digestion as a measure of the activated conformation, the ability of the MANT guanine nucleotides to maximally activate myrGo alpha correlated with maximal fluorescence. Biphasic dissociation kinetics were observed for all of the MANT guanine nucleotides. The data were consistent with the following model, [formula: see text] where G protein activation (G*-GXP) is determined by a conformational equilibrium between two triphosphate bound states as well as by the balance between binding and hydrolysis of the nucleotide triphosphate. Compared with myrGialpha1, maximal mGTP fluorescence was only 2-fold higher for the myrGialpha1 Q204L mutant, a mutant with greatly reduced GTPase activity, and only 24% that of mGTPgammaS, indicating that partial activation by mGTP was not just due to hydrolysis of mGTP. These results extend our previous conclusion that GTP analogs do not fully activate G protein.

where G protein activation (G*-GXP) is determined by a conformational equilibrium between two triphosphate bound states as well as by the balance between binding and hydrolysis of the nucleotide triphosphate. Compared with myrG i␣1 , maximal mGTP fluorescence was only 2-fold higher for the myrG i␣1 Q204L mutant, a mutant with greatly reduced GTPase activity, and only 24% that of mGTP␥S, indicating that partial activation by mGTP was not just due to hydrolysis of mGTP. These results extend our previous conclusion that GTP analogs do not fully activate G protein.
Receptor-mediated activation of heterotrimeric GTP binding proteins is a common mechanism for signal transduction in biological systems. Receptors on the cell surface regulate G protein function by catalyzing the release of GDP from the ␣ subunit of the G protein, allowing GTP to bind and activate the G protein. G protein activation is thought to be concomitant with ␣ subunit dissociation from ␤␥ subunits. Activated G protein ␣ and ␤␥ subunits interact with effector proteins to modulate intracellular second messenger metabolism (for review, see Refs. [1][2][3]. G protein deactivation is mediated by ␣ subunit GTPase activity. The deactivation rate, which may be limited by the rate of phosphate release from ␣-GDP-P i (4) is catalyzed by membrane-bound GTPase accelerating proteins (5,6). Upon deactivation, ␣ and ␤␥ subunits reassociate. Perhaps the best understood G protein mechanism is the rhodopsin-mediated activation of transducin in the retinal rod outer segment. The rhodopsin conformations can be easily studied because of the retinal chromophore attached to the protein. In addition, changes in G protein intrinsic fluorescence (due largely to changes in tryptophan 207 (7)) can be used to monitor transducin activation and deactivation (8,9) and have been used to model the kinetic mechanism of the rhodopsin-stimulated activation/deactivation cycle of transducin (10).
Kinetic mechanisms of the predominant G proteins purified from brain (G o and G i ) have also been studied extensively in solution or when reconstituted into phospholipid vesicles (for review, see Ref. 11). G o is a G protein abundant in brain that contains two tryptophans in the ␣ subunit. Upon activation by GTP and magnesium, an increase in the intrinsic fluorescence of G protein tryptophan is observed (12). The effects of ␤␥ subunits, ions, and activating ligands on G o␣ tryptophan fluorescence have been well characterized in detergent solutions (13,14). Recent structural data of transducin bound to GDP and GTP␥S 1 indicate differences in structure in the ␣2 helix containing Trp 207 . In the GTP␥S-bound state the ␣2 helix rotates such that the Trp is in a more hydrophobic environment, which is presumably the cause of the increase in its quantum yield. However, the receptor-stimulated G protein activation/ deactivation cycle has not been studied, most likely due to the high background and relatively small signals available from intrinsic tryptophan fluorescence. One of our goals has been to study G protein activation and deactivation kinetics in real time. To that end, we have synthesized fluorescently labeled G protein ␣ and ␤␥ subunits (15) and studied the kinetics of their association (16). In addition, we have found that N-methyl-3Ј-O-anthranoyl (MANT) guanine nucleotide analogs are useful environmentally sensitive fluorescent probes to study heterotrimeric G proteins (17). MANT-GTP␥S (mGTP␥S) displays a marked increase in fluorescence upon binding to G o . The fluorescence increase is greater when the MANT moiety is excited via energy transfer from tryptophan (excitation, 280 nm). The fluorescence increase observed when mGTP is bound is lower. Previous modeling of the mGTP fluorescence kinetics indicated that the low peak fluorescence with mGTP was not just due to slow binding and rapid hydrolysis. We concluded that mGTP does not fully activate G protein even when bound as triphosphate. It is possible that the natural nucleotide, GTP, also does not fully activate the G protein and is a partial agonist regarding G protein activation. However, the intrinsic GTPase activity of G protein makes it difficult to discern whether a conformational change resulting in partial activation limits GTP effects in addition to the well known role of GTP hydrolysis. For example, GTP␥S and Gpp(NH)p inhibited cyc Ϫ S49 cell adenylyl cyclase activity more than GTP (18). However, GTP␥S and GTP activated cGMP phosphodiesterase to the same extent, suggesting that GTP can fully activate transducin in rod outer segment membranes (19).
To test the hypothesis that partial activation of G o by MANT-GTP was not just due to the balance between binding and hydrolysis and to explore the G protein activation/deactivation reactions, we synthesized several nonhydrolyzable MANT-labeled guanine nucleotides. The MANT-labeled nonhydrolyzable analogs displayed differing levels of maximal fluorescence, which correlated with their ability to activate G protein. In addition, the MANT guanine nucleotides with slower dissociation rates were able to more fully activate G protein, which is consistent with predictions of a two-step activation model. Partial activators showed biphasic dissociation kinetics, indicating that we are able to observe MANT-nucleotide fluorescence signals from both the inactive and active G protein states. We also compared mGTP fluorescence when bound to a hydrolysisdeficient G protein, G i␣1 Q204L, to that of wild type protein.
While mGTP fluorescence was greater when bound to the Q204L mutant than when bound to wild type protein, it was still only 24% of the fluorescence of mGTP␥S. Taken together, these results further support the use of MANT guanine nucleotide fluorescence as a measure of G protein activation and a probe of G protein conformational states. This also represents the first demonstration that nonhydrolyzable triphosphate nucleotides can produce partial activation of heterotrimeric G proteins. Regulation of this activating conformational change could be a mechanism of receptor or GTPase accelerating protein effects on G proteins. MANT Guanine Nucleotide Synthesis and Purification-N-Methylisatoic anhydride (6 mg/300 l of dimethylformamide) was mixed with the desired guanine nucleotide (4 mg/100 l of water) and 40 l of triethylamine and incubated for 1 h at 45°C (additional water was added, if necessary, to dissolve the guanine nucleotide after the addition of triethylamine). The product was purified using a 7.8-ml Source 15Q column (Pharmacia Biotech Inc.) followed by chromatography on a Mono Q HR5/5 column to concentrate the sample and remove residual fluorescent N-methylisatoic anhydride. The Source 15Q column was equilibrated with 20 mM Tris/bis-Tris propane, pH 7.0, at 6 ml/min. Following sample injection, the column was washed with 2 column volumes of equilibration buffer followed by a 0 -1.5 M NaCl gradient over 20 column volumes. The Mono Q column was equilibrated with 20 mM Tris/bis-Tris propane, pH 6.0, at 3 ml/min. The pooled peak from the Source 15Q column was diluted with equilibration buffer (1:1) and loaded onto the Mono Q column. Following a 2-column volume wash with equilibration buffer, the MANT guanine nucleotide was eluted with a 10-column volume NaCl gradient 0 -1.8 M NaCl. The MANTguanine nucleotides were quantitated by absorbance at 252 nm using 22,600 M Ϫ1 cm Ϫ1 as the molar absorption coefficient. Purity was determined by perfusion chromatography using a 4.6 ϫ 100-mm POROS HQ/H column (PerSeptive Biosystems, Inc.) equilibrated with 20 mM Tris/bis-Tris propane, pH 8.0, at 7 ml/min. Approximately 20 nmol of product was injected and eluted with a 15-column volume gradient from 0 to 2 M NaCl. Both absorbance (270-nm preparative and 252-nm analytical) and fluorescence (mercury lamp with excitation filter OD1 and pin hole filter and a GG 420 emission filter, Fluoro-tec filter fluorometer, St. John Associates, Beltsville, MD) were monitored. The product purity was determined based on the fraction of total fluorescence that corresponded to the desired MANT guanine nucleotide product (some products contained small amounts Ͻ 6% of contaminating mGDP).

Materials
G Protein Expression and Purification-Qiagen vectors containing G o␣ and G i␣1 were cotransformed into the JM109 strain of Escherichia coli with the N-myristoyl transferase vector pBB131. Protein was expressed and purified according to Mumby and Linder (21) with the following exceptions. Following batch elution from the DEAE resin, the sample was loaded on a 100-ml heptylamine-Sepharose column using the same conditions as reported for the phenyl-Sepharose column (21). Fractions containing G protein were identified by a fluorescence assay utilizing mGTP␥S. An aliquot (10 -20 l) was incubated at room temperature with 400 nM mGTP␥S in HEDNML (50 mM Hepes, pH 8.0, 1 mM EDTA, 1 mM dithiothreitol, 100 mM NaCl, 10 mM MgSO 4 , 20 ppm deionized Lubrol) for either 10 min for recombinant myristoylated G o␣ (r-myrG o␣ ) or 30 min for recombinant myristoylated G i␣1 (r-myrG i␣1 ). Fluorescence was monitored at 440 nm (280-nm excitation) as described below. The background fluorescence of 400 nM mGTP␥S in buffer alone was subtracted to determine specific fluorescence.
Recombinant myristoylated G proteins were further purified by chromatography on Source 15Q (7.8 ml, Pharmacia), hydroxyapatite (Bio-Rad), and Mono Q HR5/5 (Pharmacia) columns using a BioCAD SPRINT System (PerSeptive Biosystems, Inc., Framingham, MA). The Source 15Q column was equilibrated with 50 mM Tris/bis-Tris propane, 1 mM dithiothreitol, pH 8.0, at 10 ml/min. Before loading onto the column the concentrated sample was filtered through a 0.45-m millipore filter. Following sample loading, the column was washed with 2 column volumes of equilibration buffer followed by a 0 -225 mM NaCl gradient over 20 column volumes. Hydroxyapatite column chromatography was performed as described previously (21). An additional final chromatography on the Mono Q HR5/5 column removed significant amounts of inactive G protein from the preparation.
Equilibrium Fluorescence of MANT Guanine Nucleotides-Recombinant myristoylated G protein (50 nM r-myrG i␣1 or 100 nM r-myrG o␣ ) was incubated with 0.01-7 M MANT guanine nucleotide at 30°C for r-myrG i␣1 and 20°C for r-myrG o␣ in HEDNML. Following a 20-min incubation, fluorescence measurements were made using a PTI Alphascan fluorometer (Photon Technology Intl., Monmouth Junction, NJ) with a water-cooled 150 W xenon arc lamp. The sample was placed in a 5 ϫ 5-mm quartz cuvette in a temperature-controlled (20°C) sample holder, and the fluorescence at 440 nm was determined (280-nm excitation). The fluorescence of MANT guanine nucleotides in the absence of G protein was subtracted from the total fluorescence to yield specific fluorescence, which represented 5-98% of total for mGDP␤S and mGTP␥S, respectively. K d and F max values were determined by fitting specific fluorescence values to a hyperbolic binding function (GraphPad Prism, GraphPad Software, Inc. San Diego, CA).
Equilibrium Nucleotide Binding-Binding constants for MANT guanine nucleotides were also determined from equilibrium [ 3 H]GDP competition experiments. The G protein (20 nM r-myrG o␣ in HEDNML) was preincubated with competing ligand for 20 min at 20°C before 60 -180 nM [ 3 H]GDP was added. Following incubation for 30 min at 20°C, the samples were filtered, and the data were fit to a sigmoid curve using GraphPad Prism. IC 50 values were converted to K d values by the method of Cheng and Prusoff (22) using a K d of 120 nM for [ 3 H]GDP binding to r-myrG o␣ . Although we propose a two-step model to describe G protein and guanine nucleotide interaction, the equilibrium behavior of that model is formally equivalent to binding with a single apparent affinity. Thus, K i values determined here are apparent binding affinities, which reflect a composite of the equilibrium constants for the two different steps in the model. The same is true for the equilibrium fluorescence K d values.
Trypsin Digestion-One microgram of recombinant myristoylated G o␣ was preincubated for 20 min at 20°C with 1, 10, or 100 M MANTguanine nucleotides in HEDNML buffer. Trypsin was added (0.02 g), and the incubation continued for 30 min. Samples were run on an 11% SDS-polyacrylamide gel prepared according to Laemmli (23) and stained with Coomassie Blue for visualization. Maximal G protein activation was determined in the presence of 100 M GTP␥S. Background was determined with no added guanine nucleotide. Four experiments were performed, the gels were scanned (Scan Jet IIC, Hewlett Packard), the 37-kDa band was quantitated (IMAGE QUANT, Molecular Dynamics), and the results were averaged.
Guanine-Nucleotide Dissociation Kinetics-Recombinant myristoylated G o␣ (300 nM) was incubated for 20 min at 20°C with 1 M MANT guanine nucleotides in HEDNML buffer. Fluorescence measurements were made as described above using the PTI ALPHASCAN fluorometer, and the sample was continuously stirred. Dissociation was initiated by the addition of 10 M GTP␥S. Data were acquired every second, and the data fit with a one-phase exponential decay with floating end point. Due to the fast MANT-GDP␤S dissociation kinetics and a fast component of the dissociation kinetics of other guanine nucleotides, dissociation was also measured using a stopped-flow spectrofluorometer (SX-17MV, Applied Photophysics Ltd., Leatherhead, UK). Excitation was at 280 nm (Photomultiplier tube voltage, 750 V) with 4.65-nm slits, and emission light was measured with a photomultiplier tube behind a KV 399 filter. Recombinant myrG o␣ (200 nM) was incubated for 20 min at 20°C with 2 M MANT guanine nucleotide in HEDNML buffer and then loaded into one syringe. The other syringe contained 10 or 50 M GTP␥S in HEDNML. Fifty l of each reactant was used per shot. All measurements were performed at 20°C, and samples were equilibrated at that temperature for at least 3 min before beginning measurements. For mGDP␤S, data were collected every 12.5 ms for 5 s with a 0.2-ms filter. Dissociation data for mGpp(CH 2 )p and mGTP␥S, mGpp(NH)p were acquired for 100, 500, and 500 s, respectively, on a log time base. Nine or 13 traces were collected and averaged. Data were analyzed using one and two component exponential dissociation equations, and an F test was used to determine the minimal model that best fit the data (Graph-Pad, Prism 1.03). Dissociation experiments were performed two times with similar results.

RESULTS
Maximal MANT Fluorescence Is Dependent upon the Guanine Nucleotide Analog-We previously observed that at saturating concentrations of guanine nucleotide, peak fluorescence due to mGTP bound to G o was 19% that of mGTP␥S. Modeling the mGTP kinetics suggested that mGTP does not fully activate G protein, even when bound (17). To test the hypothesis that partial activation was not just due to the balance between binding and hydrolysis, the fluorescence and activation properties of hydrolysis-resistant MANT-labeled guanine nucleotides were examined. Specific fluorescence saturated with increasing MANT-nucleotide concentrations (Fig. 1A). The magnitude of MANT equilibrium fluorescence when bound to r-myrG o␣ or r-myrG i␣1 (Fig. 1, A and B) was dependent upon the type of guanine nucleotide analog (mGTP␥S Ͼ mGpp(NH)p Ͼ mGpp(CH 2 )p Ͼ mGDP␤S). A summary of maximal fluorescence and fluorescence K d values for r-myrG i␣1 and r-myrG o␣ is listed in Table I. With the exception of mGDP␤S, the apparent affinities of the MANT guanine nucleotides were greater for r-myrG o␣ than for r-myrG i␣1 .
Affinities of Fluorescent Analogs for G Protein-As another method to determine affinity for G protein, the ability of the MANT guanine nucleotide analogs to compete with [ 3 H]GDP for binding to recombinant myristoylated G o␣ was determined (Fig. 2). The K i values (mean Ϯ S.E., n ϭ 4) for mGTP␥S, mGpp(NH)p, mGpp(CH 2 )p, and mGDP␤S were 3 Ϯ 1 nM, 160 Ϯ 90 nM, 1.6 Ϯ 0.5 M, and 250 Ϯ 200 M, respectively. There is a discrepancy between the K d values for some nucleotides determined by fluorescence saturation (Table I) and by competition versus [ 3 H]GDP. Because mGTP␥S binds G protein with very high affinity in both experiments, the mGTP␥S is titrating the G protein. For example, when 100 nM r-myrG o␣ was present in the equilibrium fluorescence assay, the K d was 20 nM, and when 20 nM G protein was used in the [ 3 H]GDP competition experiment, the calculated K i was 3 nM. Thus, both values are probably overestimates of the true K d . In contrast, a shortcoming of the equilibrium fluorescence measurements for ligands with low affinity (e.g. mGDP␤S) is the relatively small amount of total fluorescence that is due to specific binding. Therefore, another method to determine affinity for G protein was employed. The same order of affinity was observed for recombinant myristoylated G i␣1 ; however, the affinities were lower (data not shown). We used r-myrG o␣ in subsequent experiments.
Correlation of Maximal Fluorescence with Limited Proteolysis of r-myrG o␣ Bound to MANT Guanine Nucleotides-G protein activation is characterized by a protein conformational change causing increased intrinsic tryptophan fluorescence (12), a decreased affinity for ␤␥ subunit (9), and decreased trypsin cleavage at arginine 209 in G o␣ (24). To test the hypothesis that fluorescence of bound MANT-nucleotide was related to G protein activation, the degree of activation of G o␣ by MANT guanine nucleotide analogs was assessed by their ability to protect against trypsin cleavage. Following preincubation with guanine nucleotides, r-myrG o␣ was subjected to trypsin digestion. In the absence of added guanine nucleotide (Fig. 3A,  0 lanes), the G protein is digested to small fragments not shown The fluorescence of MANT guanine nucleotide in buffer was subtracted to determine specific fluorescence (counts per second ϫ 10 Ϫ5 (cps ϫ 10 Ϫ5 )). B, recombinant myristoylated G i␣1 (50 nM) or G o␣ (100 nM) was incubated with MANT guanine nucleotides for 20 min at 30°C for G ␣1 and 20°C for G o␣ in HEDNML buffer, and fluorescence levels were measured as described above. Maximal fluorescence and apparent K d values were determined by fitting the data to a one site binding hyperbola as described under "Experimental Procedures." Maximal fluorescence values and S.E. (relative to mGTP␥S) are shown here for recombinant myristoylated G i␣1 (left) and G o␣ (right). Maximal fluorescence and K d values are listed in Table I. in the figure. Preincubation with GTP␥S (Fig. 3A, * lanes), induces a conformation such that trypsin only released a short peptide of 21 amino acids from the amino terminus of G o␣ . The remaining 37-kDa fragment is insensitive to further trypsin digestion (24). Even at saturating concentrations of MANTguanine nucleotides, 2 there was a difference in the ability of the MANT-nucleotides to activate r-myrG o␣ as compared with 100 M GTP␥S (Fig. 3B). mGTP␥S gave the greatest protection followed by mGpp(NH)p and mGpp(CH 2 )p, which were both much more effective than mGDP␤S. 3 These data support the hypothesis that fluorescence is related to the amount of activated G protein (Fig. 4).
MANT Guanine Nucleotide Dissociation Kinetics-Nucleotide triphosphate binding and G protein activation most likely involve two steps (12,17,25,26). Since the MANT nucleotides only partially activate the G protein, we used them as a tool to examine the mechanism of partial activation and to probe the kinetics of the activating conformational change. Within the framework of a two-step activation model, there would be an equilibrium between a triphosphate-bound but inactive state (G-mGXP) and a bound active state (G*-mGXP, Reaction 1) The different MANT nucleotides would exhibit different equilibrium constants (k act /k deact ) for the activation step. The initial binding step is governed by the forward rate constant k 1 and dissociation by k i1 , while the activation and deactivation of the triphosphate-bound G protein is determined by k act and k deact .
We examined the rates of fluorescence decrease of prebound fluorescent guanine nucleotide, following the addition of competing ligand, GTP␥S. The rates of fluorescence decrease varied greatly among the MANT guanine nucleotide analogs (Fig.  5A). MANT-GTP␥S displayed very slow kinetics, followed by mGpp(NH)p and mGpp(CH 2 )p with intermediate rates and mGDP␤S with the fastest kinetics. In addition, all four nucleotides showed biphasic kinetics with results of nonlinear least 2 Based on the apparent K d and K i values measured here (i.e. 1.6 M for mGpp(CH 2 )p) and the nucleotide concentrations used (100 M in trypsin digest), greater than 98% of G protein is bound to ligand. 3 In contrast to the MANT-nucleotides, the unmodified nucleotides, GTP␥S, Gpp(NH)p, and Gpp(CH 2 )p at 100 M, were all equally effective at activating r-myrG o␣ as monitored by intrinsic tryptophan fluorescence (A. Remmers, data not shown). However, 100 M GTP only increased tryptophan fluorescence to 16% of that observed with the other nonhydrolyzable guanine nucleotides. Because the MANT moiety quenches tryptophan fluorescence, similar experiments could not be performed with the MANT guanine nucleotides. If the trypsin protection and intrinsic fluorescence both detect the same activation state, then it appears that the MANT moiety may hinder the ability of mGpp(NH)p and mGpp(CH 2 )p to fully induce the active conformation of the G protein.

TABLE I Binding affinities and maximal fluorescence of MANT guanine nucleotide analogs for r-myrG o␣ and r-myrG i␣1
Fluorescence data similar to that in Fig. 1A were fit with a one-site binding hyperbola. Shown are the K d and maximal fluorescence values (mean Ϯ range, n ϭ 2). Relative maximal fluorescence was determined by dividing by the maximal fluorescence observed upon mGTP␥S binding.

FIG. 3. Activation of r-myrG o␣ by MANT guanine nucleotides.
A, one microgram of recombinant myristoylated G o␣ was preincubated with the indicated concentrations of mGTP␥S, mGpp(NH)p, mGpp(CH 2 )p, and mGDP␤S followed by incubation with trypsin and 11% SDS-polyacrylamide electrophoresis as described under "Experimental Procedures." In the lanes marked with 0 and *, recombinant myristoylated G o␣ was preincubated with no additional guanine nucleotide or 100 M GTP␥S, respectively. Shown is a representative experiment that was performed 4 times. B, following trypsin digestion in the presence of GTP␥S or the MANT guanine nucleotide analogs, the amount of 37-kDa r-myrG o␣ was quantitated for mGTP␥S (f), mGpp(NH)p (å), mGpp(CH 2 )p (ç), and mGDP␤S (ࡗ) as described under "Experimental Procedures," and the results from four experiments were averaged (ϮS.E.). Background and maximal protection from trypsin digestion was defined using no added guanine nucleotide and 100 M GTP␥S, respectively. square fits summarized in Table II. We propose that the two kinetic components reflect the disappearance of the G-mGXP and G*-mGXP states. The fraction of fluorescence that decreased rapidly was dependent on the guanine nucleotide where mGpp(CH 2 )p (10.3 Ϯ 0.2%) had a larger fast component than mGpp(NH)p (4.1 Ϯ 0.2%) (Fig. 5B and Table II). The half-times for the fast and slow processes were 3.6 and 33 s for mGpp(CH 2 )p and 16 and 330 s for mGpp(NH)p. The fluorescence decrease for mGTP␥S never reached a plateau, so the rapidly decreasing fraction was estimated to be very small (ϳ0.4%; see Fig. 5 legend for details), and GDP␤S had the greatest fast component (63%). If there was an equilibrium between nucleotide-bound and activated G protein, guanine nucleotide analogs that more completely activate the G protein would have a smaller amount of fast component, which reflects direct dissociation of mGXP from the nonactivated G-mGXP state.
One puzzling aspect of the dissociation kinetics was why the fast phase of the mGpp(CH 2 )p fluorescence (which we attribute to dissociation of the nonactivated G-mGXP) was only 10% when it appeared from the steady state fluorescence and trypsin protection data that mGpp(CH 2 )p left ϳ50% of the G protein inactive. Similarly, the 4% fast component appeared to underestimate the 30% inactive for mGpp(NH)p. This paradox is easily understood when we take into account the relative quantum yields of the two states. If the inactive G-mGXP state has lower fluorescence than the activated G*-mGXP state, then we would expect the fast phase of fluorescence decrease to be proportional to the amount of G-mGXP present at equilibrium but to be quantitatively less. Based on a comparison of the fraction of MANT guanine nucleotide fluorescence that decreases rapidly and the fraction inactive measured by trypsin digest, we estimate that the relative quantum yield of G-GXP is approximately one-fourth that of G*-GXP. The relative quantum yields for the G-mGXP state are comparable with the relative maximal fluorescence observed for mGDP␤S bound G o␣ , indicating that the fluorescence of G-mGXP is similar to that of an inactive G protein MANT-nucleotide complex. The biphasic kinetics and the relative amount of fluorescence that decreased rapidly for each MANT guanine nucleotide suggest that these data can be described by a two-step G protein activation model (Reaction 1).
Maximal mGTP Fluorescence When Bound to r-myrG i␣1 and r-myrG i␣1 Q204L-We employed an additional approach to determine if mGTP also partially activates G protein by using a hydrolysis-deficient mutant of r-myrG i␣1 , Q204L. The steady state binding of [␥ 32 P]GTP and [ 35 S]GTP␥S was identical and indicative of extremely reduced GTPase catalytic activity (Ͼ100-fold lower) in this mutant (27). We found a 10-fold decrease in steady state GTPase using buffer conditions employed for our fluorescent guanine nucleotide experiments (data not shown). The rate of mGTP␥S binding was identical for wild type and Q204L G i␣1 (Fig. 6) and is presumably limited by the rate of GDP release. For r-myrG i␣1 , the mGTP fluorescence signal is 10% that of mGTP␥S. The maximal mGTP fluorescence is doubled when bound to the hydrolysis-deficient mutant (Fig. 6, inset). Thus, hydrolysis does contribute to the low amount of activated G protein at steady state; however, it does not appear to be the sole limiting factor (see "Discussion").

DISCUSSION
In this report we demonstrate that two fluorescent nucleotide triphosphate analogs, mGpp(NH)p and mGpp(CH 2 )p, only partially activate the G protein as reflected in their maximal fluorescence when bound to G protein and their ability to induce a trypsin-resistant state. Rapid measurements of dissociation kinetics show nucleotide fluorescence decreases from two pools of G protein, which appear to represent triphosphatebound but inactive as well as activated G protein, G-mGXP and G*-mGXP, respectively. Thus, the MANT guanine nucleotides reveal a novel aspect of G protein mechanisms and provide a means to dissect the detailed kinetics of the G protein activation step and study their regulation.  (Table I). On the ordinate, the ability of 100 M MANT guanine nucleotides to protect against trypsin digestion is plotted where protection by 100 M GTP␥S is 1.00 (Fig. 3B).
FIG. 5. Kinetics of MANT guanine nucleotide fluorescence decrease. A, MANT guanine nucleotides were prebound to r-myrG o␣ , and dissociation was initiated by the addition of GTP␥S as described under "Experimental Procedures." The data were best fit with a two-phase exponential decay with floating end point. The fast and slow rate half-times and the fraction of fluorescence that decreased rapidly are listed in Table II G Protein Activation Model-Three lines of evidence suggest that we are observing not only MANT guanine nucleotide binding but also G protein activation: 1) the fluorescence increase is magnesium-dependent (17), 2) the magnitude of the fluorescence signal correlates with the degree of G protein activation as measured by protection from trypsin digestion (Fig. 3), and 3) the magnitude of the fluorescence signal inversely correlates with the nucleotide dissociation rate (see below). It is generally assumed that G protein activation is a separate step from nucleotide triphosphate binding and differential magnesium dependence of these two steps (12,25,26) is one of the strongest pieces of evidence in support of this model. However, there has not been a good method to study the kinetics of the activation step because GDP release is slow with purified G proteins. Consequently, most analyses of G protein activation have simply considered the balance between the GDP release step and the GTP hydrolysis steps to determine the amount of G protein activation. With the MANT-nucleotides, we clearly demonstrate a role for the conformational activation step in regulating G protein activation. While the MANT-nucleotides may not activate as well as the natural nucleotides (see below), they do show that the different triphosphate analogs have distinct activation properties, which must be considered separately from their affinities. Subsequent studies will be needed to better understand the factors (e.g. magnesium concentration, ␤␥ subunits, and receptor activation) that control these steps in the G protein.
The strongest support for the G protein activation model lies in the biphasic dissociation kinetics observed for mGpp(NH)p and mGpp(CH 2 )p, which we predicted based on their incomplete activation of the ␣ subunits. Although more complicated models can describe G protein activation, we cannot distinguish among them and thus discuss the data in the context of a simple two-step equilibrium model. Based on a two-step reversible reaction where the first (binding) step is fast and the second (activation) step is slow (Reaction 1), it is possible to assign molecular rate constants to the observed fast and slow dissociation. The slow rate of fluorescence decrease is equal to k deact * (k i1 /(k i1 ϩ k act )), and the fast rate is equal to k act ϩ k i1 (corresponding to the disappearance of G-mGXP). The MANT derivatives with higher fluorescence dissociated more slowly, which would be expected because a small k deact would both increase the steady state activation and decrease the rate of dissociation. The observation that mGpp(CH 2 )p had a larger fraction of fluorescence decrease rapidly compared with mGpp(NH)p is consistent with mGpp(CH 2 )p being less effective in forming G*-mGXP than mGpp(NH)p. Although only 10% of the mGpp(CH 2 )p fluorescence decreases rapidly, it appears, based on the steady state mGpp(CH 2 )p fluorescence and the trypsin digest experiment, that approximately 50% of the G protein is activated. Thus, the small fast signal appears to account for approximately half of the G protein because G-mGXP most likely has a lower quantum yield than G*-mGXP. This lower relative quantum yield (20 -30% of that of G*-mGXP) is mostly dependent on the G protein conformational state. Likewise, although ϳ60% of mGDP␤S fluorescence decreases rapidly, most of the protein is in the bound but not activated state. The ability to detect conformational transitions with a spectral probe that is more selective than intrinsic tryptophan fluorescence will enable us to examine the mechanisms of receptor, ␤␥, and effector proteins in the cycle of G protein activation and deactivation. To evaluate potential models of G protein activation, "on kinetics" with GDP-free G protein will be measured and the molecular rate constants determined.
MANT moiety renders mGpp(CH 2 )p and mGpp(NH)p partial agonists-Yamanaka et al. (19) described different affinities of the guanine nucleotides in interacting with transducin in which GTP␥S was most potent and Gpp(CH 2 )p had a rather low affinity. In addition GTP␥S is more potent than Gpp(NH)p to compete for the GTP binding site on G i (28) and to inhibit G s GTPase (29). Differing affinities of hydrolysis-resistant GTP analogs were also observed in vivo in cardiac atrial myocytes. The GTP analogs activated G k (one of the G i subtypes) and opened an inwardly rectifying potassium channel with the following relative affinities GTP␥S Ͼ GTP Ͼ Gpp(NH)p Ͼ Gpp(CH 2 )p (30). However, all produced the same maximal response. In contrast, Gpp(CH 2 )p was only able to partially activate phosphodiesterase in rod outer segment membranes (19). We observed strikingly different abilities of MANT-labeled Since only a small fraction of mGTP␥S dissociated at all in the rapid mix experiments, the fraction with rapid dissociation kinetics was estimated. The fluorescence amplitude with fast dissociation kinetics (0.0068 Ϯ 0.0002 V) was divided by an estimate of total mGTP␥S fluorescence.
c Shown are mean and S.E. from three experiments.
FIG. 6. Maximal mGTP fluorescence when bound to r-myrG i␣1 and r-myrG i␣1 Q204L. Seven M mGTP␥S or mGTP was added to 100 nM r-myrG i␣1 (black line) or r-myrG i␣1 Q204L (gray line), stirring in HEDNML at 30°C and monitoring fluorescence as described under "Experimental Procedures." Data were fit to a one-component exponential rise, and the maximal fluorescence was plotted in the inset. Shown is a representative experiment that was repeated once with similar results. The relative maximal mGTP fluorescence compared with that of mGTP␥S when bound to r-myrG i␣1 and r-myrG i␣1 Q204L was 10.2 Ϯ 1.2% and 23.6 Ϯ 0.6%, respectively (mean Ϯ range, n ϭ 2). guanine nucleotides to maximally activate r-myr G o␣ , although affinities are similar to those described previously for nonfluorescent nucleotides. In the presence of saturating concentrations of mGpp(NH)p and mGpp(CH 2 )p, G protein was only partially activated as indicated by protection against trypsin digestion. Thus, we conclude that these fluorescent analogs only partially activate G protein. It is possible that faster dissociation rate constants for mGpp(NH)p and mGpp(CH 2 )p could contribute to the greater trypsin sensitivity compared with mGTP␥S. However, if the nucleotides fully activated G o , one would expect little nonactivated G o to be present for tryptic digestion. More direct methods to measure G protein activation such as use of fluorescent subunit probes (14) or adenylyl cyclase inhibition by G i (31) will be useful to definitively test this possibility.
Is partial activation of G proteins only a property of MANTnucleotides? Additional data suggest that GTP may also function as a partial activator. Slepak et al. (32) generated r-myrG o␣ Q205L, a GTP hydrolysis-deficient G protein mutant. This corresponding mutation reduced the rate of GTP hydrolysis (k cat ) approximately 20-fold in Ras (33) and greater than 100-fold in G i␣1 (27). They showed that GTP was only able to partially protect the G o␣ Q205L mutant from tryptic digestion as compared to GTP␥S. The time course of the experiment indicated that GTP hydrolysis was not a factor in the partial protection. However, G i␣1 Q204L was protected from trypsin digest in the presence of GTP to a greater extent than was G o␣ Q205L (27).
Implications for the G Protein GTPase Cycle-Based on the classical model of steady state G protein activation by a hydrolyzable nucleotide (34) in which binding and activation are considered to occur simultaneously, the fractional activation, G*-mGTP/G total, would equal k off /(k off ϩ k cat ). In this equation k off is the GDP dissociation rate, and k cat is the rate of hydrolysis of bound GTP. If we assume the fluorescence signal represents G*-mGTP, then when decreasing the G protein hydrolysis rate by a factor of 10 as seen with G i␣1 Q204L, one should see a 5-fold increase of activated G protein (G*-mGTP) from 0.1 to 0.5. Since we observed an increase only from 0.1 to 0.2 of maximum mGTP␥S fluorescence, we conclude that mGTP, like mGpp(CH 2 )p and mGpp(NH)p, is unable to fully activate G protein even with the reduced GTPase activity of the Q204L mutant.
We pointed out previously (17) that estimates of k cat from "single turnover" experiments (26,27) may overestimate the true value of the rate constant of GTP hydrolysis. The actual value determined in such experiments is k cat ϩ k i1 . The fact that the amount of P i released is less than the amount of GTP␥S bound indicates that the dissociation term k i1 is not negligible as previously assumed. If k cat is not as large as previously estimated, then some other factor (i.e. the activation/deactivation equilibrium) is required to account for the low degree of steady state activation of G proteins by GTP.
The proposal that GTP may be a partial agonist regarding G protein activation suggests a novel role of agonist-liganded receptor. In addition to stimulating GDP release from G pro-tein, the receptor may also shift the activation/deactivation equilibrium toward activated G protein when GTP is bound. These fluorescent GTP analogs are novel tools to discern guanine nucleotide binding from activation. In future experiments they will permit us to provide estimates of the kinetics of these processes and to assess the role of receptor, ␤␥, and effector in regulation of the G protein activation/deactivation conformational equilibrium.