The Mechanism of Aluminum-independent G-protein Activation by

0 downloads 0 Views 1MB Size Report
5 Feb 2006 - OF BIOLOGICAL. CHEMISTRY. 0 1993 by The American Society for Biochemistry and Molecular Biology, Inc ... AlF,(OH)-, rather than AlF;, is the species that reacts with ... influence of the bound magnesium on the chemical shifts of the a and p ...... accumulates in the intermediate state, and Equation. 6 be-.
Vol. 268, No. 4, Issue of February 5 , pp. 2393-2402,1993 Printed in U.S.A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1993 by The American Society for Biochemistry and Molecular Biology, Inc

The Mechanismof Aluminum-independent G-protein Activationby Fluoride and Magnesium 31PNMRSPECTROSCOPY

AND FLUORESCENCEKINETICSTUDIES* (Received for publication, July 28, 1992)

Bruno AntonnyS, Muppalla SukumarB,Joelle BigayS, Marc Chabretll, and TsutomuHigashijimas From the Slnstitut de Pharmacologie Moleculaire et Cellulaire, Centre National de la Recherche Scientifique, 660 Route des Lucioles, 06560 Valbonne, France and the §Department of Pharmacology, University of Texas Southwestern Medical Center, Dallas. Texas 75235

With magnesium present, fluoride andaluminum ionsactivate heterotrimeric G-proteins by forming AIF, complexes that mimic the y phosphate of a GTP. We report compelling evidence for a newly proposed process of G-protein activation by fluoride and magnesium, without A13+. Withmillimolar Mg2+ andF-, G, and Gt activate adenylylcyclase and cGMP-phosphodiesterase, respectively. In 31PNMR, addition of magnesium to GilaGDP or G,aGDP solutions containing fluoride, but no A13+,modifies the chemical shift of the GDP @ phosphorus, suggesting that magnesium interacts with the @ phosphate. Titration of this effect indicates that two Mg2+are bound per Ga. Biphasic activation kinetics,monitored by Ga tryptophan fluorescence,suggests the rapid binding of one Mg2+ to GuGDP and the slow association of another Mg2+,in correlation with fluoride binding and Ga activation. The deactivation rate upon fluoride dilution shows a second order dependence with respect to the residual F- concentration, suggesting the sequential release of at least three F-/Ga. Thus, in millimolar Mg2+and F-, and without AP+, two Mg2+and three F- bind sequentially to GaGDPand induce the switch to an active Ga(GDP-MgF3)Mgstate, which is structurally analogous to Ga(GDP-AlFJMg and to Ga(GTP)Mg.

The activation of G-protein by millimolar concentrations of fluoride, in the presence of magnesium, was observed to depend on the addition of trace amounts of aluminum or beryllium that form multifluorinatedcomplexes AlF, or BeF, (1).Based on the observation that fluoride activation also required the presence of a bound GDP in the G,a subunit of the inactive G-protein transducin, these fluorometallic complexes have been proposed to act as phosphate analogs that bind next to the GaGDP/3 phosphate and induce the switch to GaGDP.AIF, or GaGDP. BeF, forms that are analogous to active GaGTP (2, 3). As in the case of the activation of GaGDP by GDP/GTP exchange, a bound Mg‘+ in the nucleotide site is required for the activation of GaGDP by AlF, or BeF, (4). Recent “F and slP NMR studies of G-protein a subunits confirmed this model (5). In the31PNMR spectrum, AP+ and *This work was supported in part by National Institutes of Health Grant GM 40676 (to T. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. T To whom correspondence should be addressed. Tel.: 33-93 95 77 75; Fax: 33-93 95 77 10.

F- induced a dramatic change in the chemical shift of the /3 phosphorus of the bound GDP to a position similar to thatof the /3 phosphorus of a bound GTP. The analysis of the 19F NMR spectra indicated the bindingof three to five F-, one MgZ+, andone A13+/protein. All these experiments favor the existence of a GDP. AlF, complex (probably with x = 3), with one Mg2+ in the nucleotide site. From recent studies on the nature of the activespecies in solution, itseemsthat AlF,(OH)-, rather than AlF;, is the species that reacts with GotGDP: GaGDP + AlFS(0H)- + GaGDP.AIF3 + OH- (6). The conformationalswitch of the G-proteina subunit upon GDP/GTP exchange or AIF, binding is correlated with a large change in intrinsic tryptophanfluorescence that can serve as a very sensitive monitor with which to study the activation and deactivation kinetics (4). Using this technique, it was observed that although Al’+ (or Be2+)seemed to be an essential cofactor forG-protein activationby low F- concentrations (less than 1 mM) in the presence of Mg2+, higher concentrations of F- induce a similar fluorescence enhancement even in the absence ofAl”. The high fluorescence state induced by F- and Mg‘+ in the absence of Al”+differs in its stability from that induced in the presence of AI”; upon F- dilution, the latter deactivates in about1 h, whereas the former deactivates ina few seconds. This was thefirst evidence for another, A13+-independent, mechanism of G-protein activation by F- and Mgz+. We recentlyshowed that this activation promoted by F- without Al”+or Bez+ also requires the presence of a bound GDP in the nucleotide site of Gta (7). The Hill coefficient of the fluorescence enhancement as a function of [F-] and in the presence of 2 mM Mg” was about 3. Furthermore, the apparent rate of activation at highF- (15 mM) depended linearly on the concentrationof Mg2+when [Mg2+] exceeded 2 mM. These results lead us to propose a model in which one Mg2+ and three F- associate in the y phosphate site and form a GDP. MgF3 complex analogous to a GTP (7). In this model, Mg2+occupies a position corresponding to that of the y phosphorus in GTP. In all cases, Mg2+appears to be required for the activation of G-proteins. Although GDPIGTP-yS’ exchange in Ga can occur in the absence of Mg2+,as detected by a small fluorescence change,the active conformation characterizedby a high fluorescence level is reached only when Mg2+ is present (4). AlF, or BeF, do not activate GaGDP in the absence of Mg2+ (1-6). The rate of activation of G,aGDP with AlF:%(OH)-or Bel?,- depended in a hyperbolic manner on [Mg2+], withEC,,, of 500 P M (7). Thus, whatever G T P analog is bound, GTPyS, GDP.BeF3, or GDP.AlF,,, the switch to the active confor‘ T h e abbreviations used are: GTPrS, guanosine 5’-3-0(thio)triphosphate; ROS, rod outer segments; r, recombinant.

2393

2394

G-protein Activation by Fluoride and Magnesium

mation requires the further binding of one Mg2+.According ATP, 6 mM MgCI,, 2.5 mM EDTA, 0.25 mg/ml bovine serum albumin, 7.5 mM to the work of Yamanaka et al. (8) with phosphorothioate 0.25 mM ascorbic acid, 0.75 mM 3-isobutyl-l-rnethylxanthine, 8, and 1 pg of pyruvate phosphoenolpyruvate, 125 mM HEPES pH analogs of GTP and insofaras Mg2+ is requiredfor the kinase), andthe mixture was incubated for 10 min for Sf9 cell hydrolysis of the bound G T P in Ga(9), this Mg2+ion interacts membranes and for 30 min for cyc- S49 cell membranes at 30 "C. with both GTP/3 and y phosphates. The active forms should The reaction was stopped by adding 800 p1 of stop solution (2.5% then be written as GaGTPyS-Mg, Ga(GDP.AlF,)Mg, and SDS, 50 mM ATP, 1.75 mM CAMP,and 100 rl of [3H]cAMP (-10,000 Ga(GDP.BeFa)Mg. These two latter forms will be abbrevi- cpm)), and the[32P]cAMPproduced was assayed as reported (17,18). Preparation of Protein Samples for NMR-Protein samples were ated here as GaGDP. AlF, and GaGDP BeF,, respectively. It by Centricon ultrafiltration using PM-30 membranes was thus reasonable to assume that the active form induced concentrated (5). Thedilution-concentration cycle was repeated several times with by Mg2+ and F- in the absenceof AP+ and Be2+ includes two HED buffer (50 mM HEPESpH 7.4, 1 mM EDTA, and 1 mM bound magnesium ions: one Mg", associated with three F- to dithiothreitol) and finally with the same buffer in D20, unless oththe terminal oxygen of GDP thatwould simulate the y phos- erwise noted. The concentrations of F-, free M$+, and AI3+ or Be3+ phate of a GTP analog, and another Mg2+that would bind to were adjusted by adding appropriate volumes of concentrated stock solutions. thisGTPanalogasintheother cases. ThisGa(GDP. NMR Meas~rernents-~~P NMR measurements were made on 10MgF3)Mg active form will be abbreviated here as GaGDP. 20-mg samples of rGila or an -6-mg sample of Gta in 0.7 ml of HED MgF,. buffer in 5-mm sample tubes. Spectra were recorded on a Varian In the present work, we bring further evidence for the VXR-500 spectrometer (31Presonance frequency, 202.33 MHz) with validity of this model for the activation by F- and Mg'+ of an acquisition timeof 1.6sand 10,000-50,000 transients (5). Chemical Gta, the subunitof transducin extracted from bovine retinal shifts were measured as parts/million downfield from 10% phosphoric rods, and of recombinantGsa,Goa,and Gia produced in acid used as an external standard.A peak at -10 ppm, for example, designates a peak 10 ppm upfield from the standard. Escherichia coli. We first checked whether in the GaGDP. Tryptophan Fluorescence-The intrinsic fluorescence of Goa, Glia, MgF, forms the G-proteins indeed activated theirrespective and G,a (-50 nM) was measured at 340 nm (bandwidth 30 nm) with effectors. Then, as in the case of aluminum- or beryllium- the excitation set at 292 nm (bandwidth 3 or 5 nm) on a Shimadzu dependent activation,we searched in31PNMR spectra for an RF5000 fluorometer. The sample in a 10 X 10 mm quartz cuvette was influence of the bound magnesium on the chemical shifts of thermostated at 25 "C and vigorously stirred with a magnetic bar. the a and p phosphorus of the GDP that would suggest the The buffer used was 20 mM Tris, pH 7.5, 120 mM KC1, and 0.5 mM (TK buffer). In the case of Goa, 2 p M GDP and 0.1% location of a magnesium at the y phosphorus site. By tryp- dithiothreitol Lubrol were added to the buffer in order to prevent the formation of tophan fluorescence, we further characterized the GaGDP. nucleotide-free protein. MgF, form from the F- and M P dependences of the activation and deactivation rates. RESULTS EXPERIMENTALPROCEDURES

M~terials-[~~S]GTPrS and [y3*P]ATPwere purchased from Du Pont-New EnglandNuclear. Other reagent sources are quoted in previous works (4,5).NaF, MgCI,, and KC1 were of the highest available purity (Merck, Suprapur grade). Protein Purification-G,a was extracted from bovine retinal rod outersegments (ROS) in its GDP form, purified ona Pharmacia polyanion column, and stored at -20 "C with 33% glycerol as described previously (10). rG.aGDP, rG,aGDP, andrGlIaGDP expressed in E. coli were purified according to reported procedures (11, 12) with minor modifications. In the case of rGila, the Gafractions obtained after Mono Q column chromatography were >90% pure by SDS-polyacrylamide gel electrophoresis and were used for NMR and fluorescence experimentswithout further purification (13). In the case of rG,a, the Mono Q fractions were further purified on a phenylSuperose column with a decreasing gradient of ammonium sulfate from 1.5 to 0 M in a buffer containing 20% glycerol (v/v) (14). The concentration of G-proteins was determined by the binding of ["SI GTPyS orby the amountsof bound GDP asreported previously (1114). Proteins were stored at -80 "C before use. Phosphodiesterase Assays-The activity of the cGMP phosphodiesterase of ROS was measured according to the pH metric method (15). The experiments were performed in the dark in500 pl of buffer (10 mM HEPES pH 7.5, 120 mM KCI, 2 mM MgC12, and 0.7 mM cGMP) with dark adapted ROS membranes (final concentration of rhodopsin, 5 p ~ )which , had been washed twice in an isotonic buffer. The reaction was started by adding appropriate concentrations of Fwith or without AI3+. Adenylate CyclaseAssays-Types I and I1 adenylylcyclase activities were measured in membranes from Sf9 cells, which express recombinant types I and I1 adenylylcyclase (16), or in membranes from G.a-deficient (cyc-) S49 cells. Cyclase activity was measured as described by Sternweis et al. (17), using plastic tubes rinsed with MilliQ water to avoid possible contamination of aluminum. In brief, 6.5 pmol of rG,a was preincubated with membranes (10 pg of protein for Sf9 cell membranes and 60 pg of protein for cyc- membranes) in 60 pl of buffer (0.67 mM ATP, 6 mM MgCI,, 0.1 mg/ml bovine serum albumin, 3 mM phosphoenolpyruvate, and 0.6 pg of pyruvate kinase, 50 mM HEPES pH8, different concentrations of NaF, and with and without AICI3 (33.3 ,AM))for 10 min a t 30 "C. Then the production of [32P]cAMPwas started by adding 40 p l of buffer (0.25 mM [a-3zP]

I n the Presence of Mgz+ and F- and without A13+,Gta and G,a Stimulate, Respectively, the cGMPPhosphodiesterase and Adenylylcyclase-cGMP phosphodiesterase activity of ROS membranes was measured in the dark in order to prevent the interaction of G, with photoexcited rhodopsin (Fig.1).All the experiments were performed in the presence of 2 mMMg". As shown in Fig. l a , no increase of phosphodiesterase activity was observed upon additionof 0.75 mM NaF. Further addition

FIG. 1. Effect of F- on Gt-cGMP phosphodiesteraseactivity. Phosphodiesteraseactivity was continuously monitored by a pH electrode as the decrease of pH accompanying cGMP hydrolysis. pH change was calibrated as the amountof hydrolyzed cGMP as shown. a, bovine ROS (5 G M rhodopsin) in 10 mM HEPES pH 7.5, 120 mM KC], and 2 mM MgCI? wasincubated in the dark with 0.7 mM cGMP. At 1 and 2 min, 0.75 mM NaF and 5 p~ AlC13 were added. b, same experiment as ina, but the buffer had been supplemented with 5 mM MgC12 and 5 mM EDTA. After 5 min, 20 pg/ml trypsin was added. c, same buffer as in a, but 10 mM NaF was added instead of 0.75 mM. d, same buffer as in b, but 10 mM NaF was added instead of 0.75 mM.

G-protein Activationby Fluoride and Magnesium

2395

of 5 ,.LM A1C1, induced an increaseof phosphodiesterase activ4 ,I-. ity thatreached its maximum after80 s. This AP+-dependent effect of NaF was completely abolished if 5 mM EDTA-Mg" had been added to the buffer before the injections of NaF and AlCI, (truce b ) . EDTA hasa high affinity for A13+(conditional stability constant at pH 7.4,1015 M-') (19) and a lower one for magnesium (conditional stability constant at pH 7.4, IO6 M-') to which it wasstoichiometrically mixed inorder to maintain 2 mM free Mg2+in the reactionmedium. Under such conditions, the high concentration of EDTA did not affect the catalytic activity of phosphodiesterase as checked upon the later additionof trypsin, which activates phosphodiesterase by proteolyzing the inhibitory subunits of the phosphodiesterase (20). In truce c, a significant phosphodiesterase activity is seen upon the addition of 10 mM NaF, although no A13+was added. A fraction of this activity in the presence of 10 mM NaF could be due to aluminum contamination, since it is diminished by about 30% in thepresence of 5 mM EDTAMg (truce d ) . However, the persistenceof 70% of the activity in the presence of this strong chelatantof aluminum suggests that an aluminum-independent mechanism contributes to the activation of transducin in 10 mM fluoride. The effect of F- on G,-promoted activation of various types of adenylylcyclases was studied with and without AP'. For S49 (cyc-) membranes,inthe presence of added rG,a, a D significant adenylylcyclaseactivity was inducedby 3 mM NaF, NaF (mM) although no aluminum was present (Fig. 2 A ) . This activity FIG. 2. Effect of F- on adenylylcyclase activity in the presreaches its maximum with10 mM NaF and isfully dependent on the addition of rG,. In the presence of 20 PM A1C13, lower ence of 6 mM Mgz+.The measurements were performed as described under "Experimental Procedures." All data points were duplicated; fluoride concentrationsare efficientin promoting cyclase the error bars spanthe two measured values. A, adenylylcyclase activation.The dose-responsecurve with respect to [F-] activity of membranes from G,a-deficient (cyc-) S49 cells, suppleshows maximum activity for 3 mM NaF. Very similar results mented with 6.5 pmol of rG,a (short form), with different concentra20 PM A1CI3. For were obtained with membranesfrom Sf9 cells expressing type tions of F- (0-20 mM) and with (m)or without (0) I adenylylcyclase (not shown)or type I1 adenylylcyclase (Fig. control, two experiments with 10 mM NaF in the absence of added rG,a and with (0)or without (0)20 pM AlCl, are shown. B, same 2B). In this lattercase, the additionof bovine brain subunit experiment as in A with membranes from Sf9 cells expressing type I1 in equimolar amounts with respect to rG, increased 5-fold the adenylylcyclase, supplemented with 6.5 pmol of rG,a. At all fluoride specific adenylylcyclase activity for type I1 cyclase, as reported concentrations, the control activity in the absence of added rG,a was by Tang and Gilman (16), but the F- concentration depend- about half of that measured in the presence of 6.5 pmol of rG.a (not encies with or without AP' were hardly affected. It should be shown). This represents the contribution of endogenous G, present in these membranes. stressed here that the concentration dependencies of F- for the activation of adenylylcyclase with and without AP' are very similar to the activationof Ga asmonitored by intrinsic we studied the influence of high concentrations of F- and Mg2+ on the chemical shifts of the P and cy phosphorus, in fluorescence (see Fig. 8 of Ref. 4 for Goa). A difference is, 31P NMR spectrum of GDP however, that fluoridemonotonously enhances Ga fluores- rGilaGDP and G,aGDP. The cence whereas it seems, at high concentration, to have a small bound to rGila(Fig. 3A, b ) showed two peaks corresponding to the a (-8.6 ppm) and /3 (-2.0 ppm) phosphorus atoms of inhibitory effect on adenylylcyclase activity. Al"' had been proposed to be an essentialcofactor for G, to GDP as in the cases of rG,aGDP and bovine brain G,aGDP activate adenylylcyclase in the presence of F- and Mg'+ (1). (3), with chemical shifts distinctfrom those of free GDP (Fig. Our observations show that thisis true only whenthe concen- 3A, a ) . Addition of either 6 mM Mg2+ (Fig. 3A, c) or 20 mM tration of F- is less than 1 mM; when it is increased tomore F- (Fig. 3A, d ) did not cause anysignificant change in the "P spectrum. However, in the presence of both Mg2+and F- (Fig. than 3 mM, G, could activate adenylylcyclase without Al". chemical shifts (-7.5 and -4.3 The activation by NaF without added Al"' is not due to the 3A, e ) ,dramatic changes in the contamination ofAl". At 1 mM concentration, NaF showed ppm, for a and /3 phosphorus, respectively) are observed, the clear effects on G,a-promoted activation of adenylylcyclase change being larger for the/3 than for thea phosphorus atom. Addition ofAl"' (Fig. 3A, f-g) caused a further significant only in response to added A13+,suggesting that there is no change inthe chemical shift of the /3 phosphorus anda smaller contamination of A13+ from the reactionsolutions. 31PN M R Spectroscopy of the Mg2* + F - Activated Form of one for the a phosphorus, both of which reaching the values GaGDP-"P NMR spectroscopy had previously brought reported previously for the AlF, activation of rG,aGDP and strong support to the phosphate-analog model GaGDP-AlF, G,aGDP (5).These observations suggest that in the absence for G-protein activationby AlF, complexes, by demonstrating of aluminum, the activatingmagnesium and fluoride ions are large changes of chemical shifts for the peak corresponding inproximity to the GDP /3 phosphate and that this Gato the /3 phosphorus of the bound GDP upon the additionof activated form is chemically distinct from that obtained in F-, Mg2+, and A["+ to GaGDP (5); this confirmed that the the presence of aluminum. Given the high protein concentraaluminum ion interacted closely with the /3 phosphorus. To tion required for these NMR studies, there is no possibility check whether the activation by F- and Mg" in the absence that a contamination of Al"' plays any role for the effects of of A13+could be explained by a similar GaGDP.MgF, model, F- without Al". In fact,upon the additionof 0.5 M equivalent

G-protein Activation

2396

Magnesium by Fluoride and spectrum of G,aGDP. Despite the poor signal-to-noise ratio because of lower concentration of G,a compared with rGila, the peaks corresponding to athe and @ phosphorus atoms can be recognized. The chemical shifts of the bound GDP (Fig. 3, B, a ) and the shiftsobserved in the presence of F- and Mg2+ without and with A13+ (Fig. 3B, b and e, respectively) are remarkably similar to those observed for rGilaGDP. All these experiments show that the activationobserved for rG,aGDP and G,aGDP by F- in the absence of A13+ is mechanistically distinct from that in thepresence of A13+. In order to establish the stoichiometry of Mg2+ versus Gprotein for the activation by F- without AI3+,31PNMR spectra were measured as a function of Mg2+concentration. Spectra shown in Fig. 4A, a-e, correspond to 0.5, 1.0, 1.5, 2.0, and 5.0 M eq of M$+ with respect to the bound GDP concentration of rGila (0.7 mM). It can be seen that in the course of titration, the signals corresponding to the activespeciesincrease in intensity, while thosecorrespondingtothe inactiveform

-2

-4

a

a

-10

-2

-4 a Ommical")

a

-10

FIG. 3. 202.33-MHz 31P NMR spectra of rGllaGDP and GtaGDP in HED buffer (D,O) at 20 "C. A, rGilaGDP ( b ) , rGila in the presence of 6 mMMgC1, (c), 20 mM NaF (d),6 mMMgC12 and 20 mM NaF ( e ) ,6 mM MgCl,, 20 mM NaF, and 0.3 mMAIC13 ( f ) , 6 mM MgCl,, 20 mM NaF, and 3 mM AlC13 (g),and 6 mM MgCI,, 20 mM NaF, and 1 mM BeClz ( h )are shown. Spectra b, c, e, f, g, spectrum d , and spectrum h were recorded in separate experiments. Accordingly, the protein concentration is 0.6 mM for spectra b, c, e, /, g, 0.37 mM for spectrum d, and 0.23 mM for spectrum h. The spectrum of free GDP ( a ) recorded under identical conditions is shown for comparison. A line broadening of 20 Hz was applied. E, GtaGDP ( a ) , G,aGDP in the presence of 5 mM MgC12 and 20 mM NaF ( b ) ,and Gta with 5 mM MgCI2, 20 mM NaF, and 1 mMAIC13 (c). The protein concentration was 0.25 mM. A line broadening of 30 Hz was applied. 0.oy 0

.

I

0.5

.

0

1.o

.

1

1 . 5 ' 4

of AI3+(0.3 mM) with respect to GaGDP (Fig. 3A, f), one sees [Mg'?+l (mM) that thetwo types of activated states are stable and exchange slowly on NMR time scale; the spectrum exhibitsa superpoFIG. 4. Mg2+titration of 31PNMR spectra of rGilaGDP. A, 202.33-MHz 31P NMR spectra of rGilaGDP (0.7 mM) in HED buffer sition of the peaks observedfor the two activatedforms, GaGDP.AlF, (Fig. 3A, g) and GaGDP.MgF, (Fig. 3A, e ) . (DzO, 0.1 mM EDTA) at 20 "C was measured in the presence of 20 With F-, Mg+, and Be2+ (Fig. 3A, h ) , changes of chemical mM NaF with various concentrations of Mg2+.After subtracting the amount of MgZ' that is chelated by EDTA (0.1 mM), the corrected shifts are also observed for the a and p phosphorus, again concentration of M e is 0.35 mM ( a ) , 0.7 mM ( b ) , 1.05 mM (c), 1.4 suggesting a stable activated form of GaGDP.BeF,. These mM ( d ) , and 3.5 mM ( e ) .A line broadening of 40 Hz was applied. E, shifts appear qualitatively different from those of the AIF,- concentration of newly formed species was calculated from the area activated form but close to the MgF,-activated form. In all of peaks at -4.3 and -7.5 ppm of Fig. 4A and plotted uersus Mg" cases, the larger shift of the @ phosphorus compared with that concentration (0).The EC60 of M e was evaluated to be -0.1 mM of the (Y phosphorus is consistent with the suggestion that the by fluorescence measurements with different concentrations of M e in the presence of 20 mM F-. The number of M e binding sites ( n ) inorganic complex mimics the y phosphate group of a G T P and affinities ( K d ) used to calculate the theoretical curves are as in activating the G-protein. follows: t2 = 1, K d = 0 (. . . .); n = 1, K d = 0.1 mM (-----); n = 2, K d = Similar experiments have been carried out using G,aGDP 0.1 mM (---); n = 2, K d = 0; (.-.-.); n = 2, K d = 0.03 mM purified from bovine ROS. Fig. 3B, a shows the 31PNMR c-".

G-protein Activation

Magnesium by Fluoride and

2397

A1F3)Mgis supposed to form, no biphasic kinetics is observed whatever the 01 subunit subtype. The biphasic kinetics is hence specific to theF-- and Mg*+-dependentactivation mechanism. This biphasic kinetics could correspond to theactivation of two different populations of GaGDP or to a “two-step”mechanism affecting the whole GaGDP pool. The second hypothesis was supported by the experiments shown in Fig. 5B for Goa. The activation was started by the injection of 2 mM Mg2+ to a solution of G,GDP in 20 mM NaF. Theshape of the curve is similar to thatobtained before (Fig. 5A) when F- was added after Mg2+.When EDTA was added a few seconds after the jump (thin line), it induced a large instantaneous decrease, followed by the slow decay of the remaining small fraction of the original fluorescence enhancement. By contrast, when EDTA is added at the end of the slow rising phase (thick line), it induced almost only the slow fluorescence decrease. This suggests that theactivation consistsof at least two steps: a fast switch to a transitory conformation with a low affinity for Mg2+(high KOff) followed by a slow conformational change that results in a high affinity for Mg2‘ (low k d . Differences between the intermediate state and the final “active” state could also be observed upon fluoride dilution down to 1 mM; the deactivation of the intermediate state was too fast to be measured on the time scale of our fluorescence experiments, whereas the final state deactivated in several seconds (see Fig. 7). Therefore, it appears that the slow rising phase results in a stabilization of both F- and Mg*+ into the nucleotide site. The following experiments will suggest that thisstabilization correlates with the binding of a second Mg2+ ion. We then measured the amplitude of the jump and the apparent rate constant of the second phase as a function of the concentration of Mg2+.The Mg2+ dose response of these two parameters was obtained by adding 20 mM NaF to a 50 nM G,aGDP solution at various Mg2+concentrations (0-4 mM, Fig. 6 A ) . The amplitude of thejump depends in a Michaelis-type manner on [Mg”] (0). This suggests that one A Mg2+is involved in the first fast step of the activation mechanism. If the second activation (slow) phase corresponded to a monomolecular mechanism (for instance, a relaxation to a new conformation), its apparent rate should depend only on the fraction of G-protein in the intermediate state, which is proportional to the amplitude of the jump (see “Appendix”); the apparent rate constant should, thus, have the same dependence on magnesium concentration as that of the amplitude of the jump. This is not whatwas observed. Whereas the amplitude of the fluorescence jump saturated at -2 mM Mg2+ with EC5, ofabout 0.8 mM,the rateof the slow phase increased continuously with increasing Mg2+ and tended to depend linearly on Mg2+concentration when M e exceeded 1-2 mM. This linearity suggests a bimolecular reaction between one Mg2+and the transitory state toreach the final active state. The protein would switch from the intermediate state to the final active state through the slow binding of a second M$+. A more elaborate analysis of this bimolecular mechanism can be done, taking intoaccount that after the jump and for Mg2+ below 2 mM, a fractionof the protein is not inthe intermediate 0 o l o 200 300 state (see “Appendix”). time ( 5 ) We also studied the dependence of the biphasic activation FIG. 5. Biphasic kinetics of G-protein activation by Mg2+ kinetics upon F- concentration. The reaction was triggered and F-. A, the fluorescence of 50 nM G,aGDP, GilaGDP, orG,aGDP in TK buffer (1.4 ml) with 2 mM Mg2+was monitored, and, at the by the addition of 2 mM MgC12 to a solution containing 50 indicated time, 28 p1 of 1 M NaF (20 mM final) was added. Though nM G,aGDP and various concentrations of NaF. Here, both the proteinsolution was diluted by 2%, a fluorescence jump was the amplitude of the jump and the rate of the slow phase always observed, followed by a slow enhancement (see text). B, the (Fig. 6B). As effect of magnesium suppression on the two phases of the biphasic varied similarly with the F-concentration kinetics is shown. G,aGDP (50 nM) in TK buffer with 20 mM NaF discussed above, the dependence on F- concentration of the was activated by the addition of 2 mM MgC12;3 mM EDTA was added second phase is now fully explainable by its contribution to 10 s (thin trace)or 3 min (thick trace)after the magnesium addition. the first phase.This suggests that fluorides bind to the protein

diminish. With the same amount of Mg2+uersus G-protein, half of the protein pool is in the MgF, bound form (Fig. a, b). With two M$+/protein, nearly 100% of the protein is in the MgF, bound form (Fig. 4 A , d ) . Fig. 4B shows signal intensity of active species uersus [Mg2+]. Unless the affinity of Mg2+is very low, this indicates that two Mg2+/protein are required for the conformational change. We have checked that the apparent affinity of Mg2+for GaGDP, in the presence of 20 mM F-, is sufficiently high to rule out the possibility that only one MgZ+ binds toGa.This control has been performed using the fluorescence enhancement that occurs when a G-protein a subunit switches to an activated form. With 20 mM NaF, the variation of the intrinsic fluorescence of G,aGDP as a function of [Mg2+]gives an apparentaffinity of 0.1 mM. Given the concentration of G&DP (0.7 mM) used in the NMR experiments, this affinity is not compatible with the binding of one Mg+/protein (Fig. 4B).On the contrary, a simulationwith two M$+ with Kd 0.1 mM for both fits fairly well. Kinetic Analysis of Fluoride-induced Tryptophan Fluorescence Changes; Biphasic Actiuation of Ga by F- and Mgz+As in the case of the activation by aluminum fluoride complexes, G-proteinactivation by F- and Mg2+ without A13+ correlates with a large tryptophan fluorescence change. Fig. 5A shows the intrinsic tryptophan fluorescence change that occurs upon the addition of 20 mM NaF to 50 nM G,aGDP, GilaGDP, andG,aGDP in the presence of 2 mM Mg2+.Clear biphasic fluorescence increases are observed; a rapid “jump” is followed bya slow rising phase. The amplitude of the jump is different for each Ga. It corresponds to 10, 20, and 40% of the total fluorescence increase for G,a, Gila, andGoa, respectively. As shown previously (4, 6), thetotal fluorescence increases were similar to that induced by AP+, Mg2+,and Fmixtures. However, in the presence of AP+, when Ga(GDP.

G-protein Activation

2398 A

30

I

0.05

NaF 20 mM

Magnesium by Fluoride and

r

-0.01

g m

X

0

-2 "0 --

2

1

MgCG (mM)

B 32

-

22

-

m

&

MgC122mM

0

; 12a

0

0

3

0.02

C 0

01

-1

L

- 0.01 5 C

R R

a

'0

X

2 0

--il

0.03

0

a

0

3

hand, the dilution of F- down to 1 mM promoted a much faster deactivation (witha time constant of 10 s) as shown in Fig. 7A. After the fluorescence decay, the readdition of NaF up to 15 mM caused full reactivation of G,aGDP with a time constant of 25 s. The fact that the activation rate is 2.5 times slower than the deactivation rateis apparently contradictory with the fact that15 mM NaF and2 mM MgC12 fully enhance the fluorescence of G,aGDP. This suggests that the deactivation of the a subunit upon F- dilution proceeds through different steps than those involved in its activation. In other words, some intermediate states required for the activation

1 03 0 20 NaF (mM)

' 0 40

0

FIG. 6. Effect of increasing concentrations ofNaFand MgClz on the two phases of the activation kinetics. G,aGDP (50 nM) in TK buffer with2 p~ GDP and 0.1 % Lubrol was incubated with various concentrations of MgCl, ( A ) or NaF ( B ) .The activation was triggered by the addition of 20 mM NaF ( A ) or 2 mM MgCL ( B ) . The amplitude of the fast phase is normalized to the fluorescence level before fluoride ( A ) or magnesium ( B ) addition. In A , the fast phase amplitudes (0)are fitted with AF/F = (AF,,,.,/F). [Mp"']/(Kd + [ M e ] )with AF,,,,,/F = 32%and Kd = 0.87 mM, and theslow phase + rate constants (0) are fitted with kapparent= k,. [Mg"]'/(& [Mg"]) with k, = 12.5 M". s-' and Kd = 0.87 mM.

0

0

50

100

150

200

30

40

time (s)

60

10

20

time (s) only during the first step and that no further binding of C fluoride occurs when the protein switches from the intermediate state to the final activated form. 15 All these experimentssuggest that in the intermediate state, one Mg2+ and all the necessary F- ions (probably three) are s 10 bound to the protein. The binding of a second Mg2+ to this P metastable state induces the switch to theactive form. 5 Both millimolar F- and Mg2+are required for the activation mechanism and especially for the formation of the intermediate state. Does this mean that these two ions cannot bind 0 independently to GaGDP? In fact,a very small fluorescence NaF (ItW enhancement observed when Mg2+ is added to G,aGDP sugFIG. 7. Fluorescence monitoring of the dependence of Ggests that Mg2+ can bind to GaGDP in the absence of F- protein deactivation rate on residual NaF concentration. A, (data not shown). This signal is instantaneous, small (2% the fluorescence cuvette initially contained1.8 ml of TK buffer with fluorescence enhancementatsaturation),anddependson 2 mM MgCl, and 0.9 mM NaF, G,aGDP (8 PM) was fully activated by incubation (5 min, 25 "c)with 2 mM MgCl, and 15 mM NaF. At time M e concentration with an EC,,of 0.5 mM, suggestinga M$+ affinity for GaGDP of about 0.5 mM. By contrast, no t = 10 s, 9 pl of the activated protein solution was injected in the fluorescence change was observed when NaF was added to cuvette. The NaF concentration in the cuvette is then 0.975 mM. At 60 s, 27 p1 of NaF 1 M was injected giving a final NaF concentration G,aGDP in the absenceof Mg". of 15 mM. One notices that the fluorescence decay (corresponding to Analysis of Deactivation Kinetics; Evidence for the Sequen- deactivation) is faster than the fluorescence enhancement (corretialRelease of at Least Three F"To further analyze the sponding to activation).B, fluorescence decay of G,aGDP.MgF, was interactions of F- and Mg2+ with GaGDP, we investigated, monitored, as in A , by diluting activatedprotein solutions (in 15 mM down to residual NaF concentrationsof 975,875,775,675,475, by monitoring the decay of tryptophan fluorescence, the ki- NaF) 275, and 75 p ~ from , the slowest to the fastest decay.Similar netics of deactivation upon removal of either of these ions experiments were performed for G,GDP and GilGDPexcept that from the solution. Deactivation was induced by (i) chelating preactivation of these Ga were with 20 mM NaF and 2 mM MgCl'. C, allfree Mg2+ by excess EDTA while keeping F- at high dependence of the deactivation rate on residual NaF concentration concentration or by (ii) diluting F- while keeping Mg2+ con- for G,, Go, and Gil. The time constants of fluorescence decays, as measured in B, were plotted uersus the residual concentration of F-. centration constant. The addition of 3 mM EDTA toa solution The fits were made with second order polynomials (T = a + b[F] + of G,aGDP that had been activated to saturation by 15 mM c[F]'; see "Appendix"). The values for a-c were 0 s, 1.1 X 10' s. M-', NaF and 2 mM MgC12 induced a slow decay of fluorescence 2.2 x IO6 S. M-' (GJ, 0.5 s, 8.9 x 10' S. M-', 6.1 x lo6 s . ~ "(Gd, and with a time constant of 400 s (data not shown). On the other o s, 7.9 x 103 S.M-', I x 107 S. M-2 (GJ. A

G-protein Activation

Magnesium by Fluoride and

2399

process would be bypassed in the deactivation mechanism. propose a model in which one magnesium associated with the Another puzzling feature of deactivation kinetics upon flu- three F- mimics the y phosphate group of a GTP, and the oride dilution was its dependence on the residual concentra- second magnesium is bound to this GTP analog and confers to the protein an active Ga(GDP.MgFJMgconformation tion of F-. Since G,aGDP cannot be activated by submillimolar F- concentration in the presence of 2 mM Mg2+, we identical to that of GaGTP .Mg. This mechanism seems to expected, and indeedobserved, a complete fluorescencedecay be valid for all types of G-protein a subunits, although, asfor upon dilution of the initial 15 mM NaF to a final F- concen- the other activation and deactivation processes (nucleotide tration below 1 mM. Deactivation could always be fitted to a exchange, GTPase), the values of the kinetics parameters single exponential, but the time constant strongly depended depend on the G-protein subtype. We shall discuss the assembly of this multiionic complex on the residual F- concentration, as shown in Fig. 7B. With a residual F- concentration of 0.075 mM, the relaxation time inside the nucleotide pocket of inactive GaGDP, its stability was of T = 1 s, that is 10 times faster than that in0.975 mM when the protein has switched to the active form, and its F-. The time constant could be well fitted by a second order multiple dissociation pathways. In 31PNMR spectroscopy, the large effects of F- and Mg" polynomial of the final F- concentrations. Total exponential on the chemical shifts of the p phosphorus of the proteinfluorescencedecays were alsoobserved for GilaGDPand weaker effect on the a phosG,aGDP upon NaF dilution below 1 mM (Fig. 7C). Thevalues bound GDP, compared with the of decay time constantsdiffered from that obtained under thephorus chemical shift, were consistent with the suggestion same conditions for G,a but showed analogous parabolic de- that a combination of F- and Mg' ionsmimicked the y phosphate of a bound GTP, as in the case of the AlF, complex pendences on finalF- concentrations. by "F NMR spectroscopy of The fact that under conditions in which fluoride dilution (1-7). On theotherhand, GaGDP in thepresence of F- and Mg2+,we could not detect induces total deactivation its rate remains sensitive to the possibility that any new peaks from protein-bound "F (datanotshown), final free F- concentrationrulesoutthe deactivation follows, irreversibly, the dissociation of a single while a discrete protein-bound"F peak had been observed for F- ion. The sensitivity to residual F- concentration implies AlF,-activated G a (5). Thissuggested either that the chemical that F- from the solution can bind back and slow down the shifts of Ga-bound F- in the absence ofA13+ are similar to deactivation process before the irreversible step occurs. The that of free F- or, more likely, that the chemical shifts are slightly different, but theexchange of Ga-bound F- with free second order dependence of the deactivation time constant ( T ) upon residual fluoride concentration further suggests that F- in solution is fast. Indeed, the fluorescehce kinetics measdeactivation can be slowed down by the rebinding of two F- urements supported the hypothesisof a very fast exchange of ions from the solution. More generally, it can be shown that F- in theabsence of Al'+ (see Fig. 7B and Refs. 4 and 6). Both the numberof successive reversible F- dissociations preceding 31PNMR spectroscopy, from the titration of the shifted Pp the irreversible step is equal to the order of the polynomial peak, and fluorescence measurements, from the biphasic acdependence of the deactivation time constant on theF- con- tivation kinetics,suggested the binding of two MgZ+.One centration (see "Appendix"). In the present case, two revers- Mg2+, associatedto theF-, mimics the y phosphate of a GTP, ible F- dissociation steps can occur and, thus, at least three and the other interacts with this GTP analog as the Mg2+ high fluorescence species with a different number of bound bound toactive GaGTP. Mg. The biphasic activation kinetics suggests that thefluorides F- are in fast equilibria before athird dissociation step triggers and one Mg'+ binds rapidly to Ga to form an intermediate the irreversible deactivation. state, which then reactsslowly with a secondMg". The switch k-3 k-z kto the active conformation correlates with this slow binding. GcY*F~ .k+, Ga*FZ + F e Ga*F + 2F 2Ga + 3F k+2 We propose that the intermediate state that forms rapidly corresponds toweak hydrogen bonding of the fluorides to the SCHEME 1 protein y phosphate site and is contingent to the bindingof In thisscheme, Mg2+is ignored in order tosimplify, and G* the first magnesium to GDP. The second step would corredesignates the activated high fluorescence states. Three F- spond to theslow binding of the Mg2+,which stimulates the are bound in the fully activated state, andtwo of them canbe y phosphorus (Fig. 8). The negatively charged F- are then released and rapidly exchanged without inducing the deacti- stabilized by the centralMg2+,which also associates with the vation. This gives the parabolic dependence of the time con- terminal oxygen of the phosphate. The rate constant for the stant of deactivation withrespect to theresidual free fluoride binding of Mg2+is very low ( k , = 12.5 M-' s") (see Fig. 6). Concentration. The association and dissociation rateconThus, this step is probably not limited by the diffusion of stants of the three fluorides are related to the polynomial Mg"+ to the nucleotide site but by low saturation of the three coefficients of the parabolas, as detailed in the "Appendix," F- sites or by a slow protein conformational change. and differ for the various G a subtypes. The zero order coefThe structureof the y phosphate analog, Pp-O-MgFr3, where ficient is the sum of the dissociation time constants of the Mg2+is at the center of a tetrahedron and theF- at the apex, three fluorides, T~ = l / k 1 l / k Z I/&. The parabolic may account for the large difference in the deactivation rates extrapolation of the apparent time constant down to [F-] = depending on whether it is induced by the dilution of the 0 sets an upper limit at 0.5 s for T,, in Gta and even lower fluoride or the chelationof the magnesium in the medium. In values for Goa and Gila. Thus, the individual fluorides disso- the tetrahedon,Mg2+forms ionic bonds with four anions.The ciate and exchange rapidly on a subsecond time scale. probability that the three Mg"-F- bonds and the Mg2+-0bond dissociate simultaneously is extremely low; thus, the DISCUSSION rate of Mg2+dissociation must be slow. By contrast, each All the experiments presented here confirm that fluoride fluoride is involved in only one ionic bond with the central and magnesium, in the absenceof any othercomplex-forming M$+ and is likely to exchange fast. When fluoride is supmetal ions, can activate GaGDP subunits of heterotrimeric pressed from the solution, the fast sequential exit of the three G-proteins andsuggest that the concurrent binding of 2 Me2+ fluorides will lower the bindingof the central Mg2+ and allow and at least three F-ions isrequired for activation. We its dissociation. To account for the parabolic dependence of

+

+

2400

G-protein Activationby Fluoride and Magnesium Mg2: (2 mM) I

-

F-

I

””

” ” -

(go mM) 11

I c -” “ ” ”

111

” ” ” _

-”c

IV

“””

low

fluorescence

. )

high

FIG. 8. Model for the mechanism of G-protein activation by F- and Mg2+. Thick and thin arrows indicate, respectively, fast and slow steps. State I is M f l and F- free GaGDP and shows the lowest intrinsic tryptophan fluorescence level. The formationof state I1 can be detected through a small fluorescence change (2%),which occurs instantaneously upon the addition of MgZf (E& of Mg2+ -0.5 mM for G,aGDP). Since there is millimolar Mg2+ in physiologicalconditions, state I1 corresponds tothe basal state of a G-protein.The intermediate state I11 was detected by the instantaneous fluorescence jump occurring upon addition of both M$+ and F- (see Fig. 5 A ) . In this state, three F- are supposed to bindto charged residues of the nucleotide y phosphate binding site, but with low affinities and very fast exchange rates. Since fluorescence changes areinstantaneous between states I, 11, and 111, these states are in fast equilibria.State IV is activeand corresponds to the highest fluorescence form. It is in slow equilibrium with state 111, as demonstrated by the slow fluorescence enhancement describing its formation from state I11 and by the slow fluorescence decay following MgZf dilution (Fig. 5 ) . The switch from state 111 to IV correlates with the binding of a second Mfl, as suggested in Fig. 6A. This M e , as A13+ in the case of AlF,, is the central cation that associates with the terminal oxygen of the (3 phosphate. It stabilizes the three negative charges of F-, forms the y phosphate analog,and, hence, induces the switch to the active “GTP” form. Then the interactions between Ga, GDP, the three F-, and the two Mg2+ ions dramatically increase. The two Mg2+ ionsdissociate slowly (>IO0 s), and the apparent affinities of the F- reach the submillimolar range. Upon Mg2+ removal,state IV deactivates slowly back to I via states 111 and I1 (Fig. 5B). By contrast, upon F- dilution, the bound F-s are rapidly released, destabilizing M e binding, and state IV decays rapidly to state I via the fast transient state V, which still has the two magnesiums bound. State V is assumed to keepthe same fluorescence levelas the active form, Ga(GDP. MgFs)Mg.The fact that the deactivation rates following F- dilution are well fitted by a second order polynomialof the residual fluoride concentration indicates that at least three fluorides are released inthis process. There could be more. the deactivation kinetics on the residual NaF concentration, nism, because they are already associated in a tetrahedral complex insolution. However, deactivationkinetics were we must assume that (i) in the active state, the bound fluorides rapidly exchange with the free F- in the solution and (ii) the weakly sensitive to theresidual free fluoride concentration in active conformation remains metastable when up to two flu- the medium, suggesting that the fluorides of a bound alumiorides are temporarily absent. Only the dissociation of a third num fluoride or beryllium fluoride complexare stillexchangeF- triggers an irreversible switch to the inactive state, prob- able with free fluoride a t a somewhat faster rate than the ably by allowing the fast release of the central Mg”. It must whole complex. Thus, the mechanismsby which the activated forms GaGDP-MgF, and GaGDP-AlF, deactivate have some be noticed that the comparatively slow deactivation ofG, upon diluting the F- down to only 1 mM (Fig. 7A) was not qualitative analogies, although their kineticsdiffer widely. The processof activation by fluoride and magnesium might accelerated if Mg2+had been simultaneously diluted down to 50 K M . This indicates that the dissociationof the three fluo- concern all nucleotide-binding proteins, kinase, or phosphatase that are sensitive tofluoride complexes of aluminum or rides must precede that of the central Mg2+. beryllium (for review, see Ref. 21) including mitochondrialF1 This deactivation process upon F- dilution in the presence of Mg2+ is too rapid to be explained by areversal of the ATPase (22, 23), actin, and tubulin (24-26). This might be activation process upon addition of F- in the presence of relevant for the recently reported inhibitionof Ca2+-ATPase from sarcoplasmic reticulum bymillimolar concentrations of Mg2+. This agrees with the assumption made above that in F- and Mg2+in the absence of AP+ (27). the activation process, the binding of the three fluorides did not follow but ratherpreceded that of the central magnesium. Acknowledgments-We thank Drs. Hiroshi Itoh, Maurine Linder, The binding of three F- to GaGDP is a prerequisite for the and Alfred G. Gilman for providingthe strain of recombinant E. coli, binding of the “ y phosphorus” Mg2+ and the switch to the W-J. Tang formembranes of Sf9 cells that expresstype I or I1 active conformation. Only the slow deactivation promoted by adenylylcyclase,andIsabelleLenoirforpurifyingtheverylarge Mg2+suppression in the continued presence of F- proceeds quantity of transducin used here. through the reversalof the activation pathway. APPENDIX In a recent study we showed that aluminum fluoride and beryllium fluoride complexes bind to GaGDP proteins by a Model for the Biphasic Kinetics of G-protein Activation by mechanism that is close to a bimolecular reaction (6). This MgFx-When a protein (G) associates instantaneously with a suggested that the metal(A13+or Be2+) and the three fluorides ligand (A) and then switches slowly to a final conformation do not bind separately as in the caseof the Mg’+/F- mecha- (G*)

G-protein Activation

+A

G

Kd

GA

5 G*A.

Magnesium by Fluoride and (Scheme A l )

The rate of the second phase is given as follows. d[G*A]/dt = k.[GA]

(Eq. 1)

Thefirst reaction is instantaneousascompared second one and, hence, is a t equilibrium,

with the

[GI/[GAl = &/A, (A, >> Co)

(Eq. 2)

where C0 and A. are the total concentrationsof G and A. [GI

+ [GA] + [G*A] = Co

(Eq. 3)

Then one gets from Equations 2 and 3, (Eq. 4)

The differential Equation 1 is then, (Eq. 5 )

The apparent rate constantof the slow phase is then, (Eq. 6)

where Ao/(An + K d ) is the fraction of protein in the intermediate state after the fast phase. If the slow phase corresponds to the binding of asecond molecule of A, G

Kd

+ A e GA + A

k, "-f

G*A,.

(Scheme A2)

Using the same equations as previously, the apparent rate constant of the slow phase is,

2401

saturation by 20 mM [F-] in the presence of 2 mMMg" induced a total exponential deactivation whose rate strongly depended on the residual fluoride concentration. This sensitivity to residual fluoride concentration implies that fluoride from the solution can bind back and temporarily replace a dissociatedfluoride and, thus, slow down the deactivation process before an irreversible step occurs. We propose that the irreversible switch to the inactive state occurs when the magnesium, which associated to the fluorides mimics the phosphorus, is released. The release rate of this magnesium is very low when all the fluoride sites are saturated, as indicated by the very slow deactivation observed upon diluting magnesium in the presence of a high fluoride concentration. It is reasonable to assume that the magnesium dissociation rate increases when some of the fluoride sites are empty and becomes very fast when the last fluoride is gone. Our model is based on the simple assumption that the dissociation rate of magnesium is negligible with respect to that of fluoride until the last fluoride is released and becomes predominant afterward,thus triggering inactivation only afterthelast fluoride is released. This means that with submillimolar Fin the milieu, the rebinding of fluoride to a partially dissociated MgF, complex is faster than the dissociation of magnesium.However, as soon asthelast fluoride exits,the dissociation of the lone magnesium becomes faster than the reassociation of fluorides, which would stabilize the active state. For a fully activated state with three bound fluorides, the scheme is, k-:>

G*MF, G*MF, k+:i

k-2

+ F === G*MF k+,

k-, +2F* GSM+3Fk_M.G+M+3F k+L

(Scheme A3)

where the G* designates the active (high fluorescence) states, G designates the inactive state (low fluorescence), M stands for magnesium, and F stands for fluoride. The rate of deactivation is given by,

d[G]/dt = k-M[G*M] = - dZC*/dt (Eq. 1') of the second phase where k,Ao is the first order rate constant and Ao/(Ao + k d ) is the mole fraction of theintermediate with state, after the fast phase. BG* = [G*M] + [G*MF] + [G*MF,] + [G*MF,]. (Eq. 2') G-proteinactivation by F- and Mg2+ proceeds through Scheme A2 as compared with Mg". Scheme A l , although it Since k-M >> k+l[F], assuming thatG*M, G*MF, and G*MF, does not take into account the multiplicity of fluoride binding, are at steady state, onegets the three following equations. adequately describes the fact that thedependences on [F-] of k-M[G*M] = k-,[G*MF] (Eq. 3') the amplitudeof the fast phase and the apparent rate constant of the slow phase are similar. Indeed, this similaritydoes not (k,,[F] + k-l)[G*MF] = k-z[G*MF?] + k+l[F][G*M] (Eq. 4') depend on the number of fluorides that bind but on the fact (Eq. 5') that all fluoride binding occurs during the first fast phase and (k+,[F] + k-z)[G*MF,] = k-,[G*MF:3] + k+,[F][G*MF] thatduringthe following slow phase,there is nofurther From Equations 3 , 4 , and 5 , the concentration of each species fluoride binding. The general scheme would be, can be calculated as a function of [G*M]. [G*MF] =

k-M [G'M] k-

(Eq. 6')

I

In thisscheme, the apparent rate constant of the slow phase is again proportionaltothefraction of theproteinthat 6 beaccumulates in the intermediate state, and Equation comes, kpp

=

k..[GAJ/([GAnI + [GI) = k..Ao"/(Kd + An")

(Eq. 8 )

MultipleFluoride Dissociation Models for GaGDP. MgFx Deactivation upon Fluoride Dilution-Diluting fluoride below 1 mM in a G-protein solution that had been activated at

Introducing these expressions in Equation 2, one obtains,

G-protein Activation

2402

Magnesium by Fluoride and depends linearly on the concentration of F-. And withasingleboundfluoride, thetimeconstant is obviously independent of the finalfluoride concentration and reduces to, 7

1 k-M

s=-+-.

The differential Equation 1 can then be written as, (Eq. 10')

with 1 k"

1 1 k-1 k-3k-2

+

(Eq. 2")

REFERENCES 1. Sternweis, P. C., and Gilman, A. G. (1982) Proc. Natl. Acad. Sci. (1. S. A. 79,4888-4891 2. Bigay, J., Deterre, P., Pfister, C., and Chahre, M. (1985) FEES Lett. 191, 181-185 3. Bigay, J., Deterre, P., Pfister, C., and Chahre, M. (1987) EMBOJ. 6,2907991 R

I"_-

1

7 = - + - + - + -

1 k-1

(Eq. 11')

(&

+ &)[F] k-1k-z k-lk-2k-a k-zk-3

+

~

k+h+3

W .

4. Higashijima, T., Fergusson, K. M., Sternweis, P. C., Ross, E. M., Smigel, M. D., and Gilman, A. G. (1987) J. Biol. Chem. 2 6 2 , 752-756 5. Hi ashijima, T., Graziano, M. P., Suga, H., Kainosho, M., and Gilman, A. (1991) J. Biol. Chem. 266,3396-3401 6. Antonny, B., and Chabre, M. (1992) J. Biol. Chem. 267,6710-6718 7. Antonny, B., Bigay, J., and Chahre, M. (1990) FEES LRtt. 268,277-280 8. Yamanaka, G., Eckstein, F., and Stryer, L. (1985) Biochemistry 24,8094-

6.

Rlnl

Thus, if three bound fluorides must dissociate before the switch to the inactive, low fluorescence state, the time constant of the fluorescence change observed upon fluoride dilution depends on a second order polynomial of the final concentration of F-. More generally, if n fluorides have to dissociate, the time constant of the fluorescence change will depend on a ( n - 1) order polynomial of [F-1. For example, if only two fluorides have to dissociate, the scheme reduces to, G*MF2

k-2

G*MF

k-1

+FEG*M + 2F 2 G + M + 2F

(Scheme A4)

and the expression for the time constant of the fluorescence change is, 1 1 k-M k-2k-1

r=-+-+-+-

1

k+z IF]. k-1k-2

(Eq. 1")

9. Higashjjima, T., Fergusson, K. M., Smigel, M. D., and Gilman, A. G. (1987) J. Bud. Chem. 262,757-761 10. De,t,errs,P,., Bigay, J., Pfister, C., and Chabre, M. (1984) FEES Lett. 178, LL0"LJL

11. Linder, M., Pang, I. H., Duronio, R., Gordon, J., Sternweis, P. C., and Gilman, A. G. (1991) J. Bid. Chem. 266,4654-4659 12. Itoh, H., and Gilman, A. G. (1991) J . Bid. Chem. 2 6 6 , 16226-16231 13. Sukumar, M., and Higashijima,T. (1992) J. Biol. Chem. 267,21421-21424 14. Ferguson, K. M., Higashijima, T,. Smigel, M. D., and Gilman, A. G. (1986) J. Biol. Chem. 261,7393-7399 15. Yee, R., and Liebman, P. A. (1978) J . Biol. Chem. 252,8902-8909 16. Tang, W-J., and Gilman, A. G. (1991) Science 2 5 4 , 1500-1503 17. Sternweis, P. C., Northup, J. K., Smigel, M. D., and Gilman, A. G. (1981) J. Biol. Chem. 2 5 6 , 11517-11526 18. Salomon, Y., Londos, C., and Rodbell, M. (1974) Anal. Biochem. 5 8 , 541548 19. Martin, R. B. (1988) in Metal Ions in Biological Systems (Sigel, H., ed) Vol. 24, pp. 1-57, Marcel Dekker, Inc., New York 20. Hurley, J. B., and Stryer, L. (1982) J. Biol. Chem. 2 5 7 , 11094-11099 21. Chahre, M. (1990) Trends Biochem. 15,6-10 22. Lunardi,. J., Du uis, A,, Garin, J., Issartel, J. P:, Michel, L., Chabre, M., and Vlgnals, V. (1988) Proc. Natl. Acad. SCLU . S. A. 85,8958-8962 23. Dupuis, A,, Issartel, J. P., and Vi nais, P. (1989) FEES Lett. 2 5 5 , 47-52 24. Carlier, M. F., Didry, D., Melki, k , Chahre, M., and Pantalonl, D. (1988) Biochemistry 2 7 , 3555-3559 25. Combeau C. and Carlier, M. F. (1988) J. Biol. Chem. 2 6 3 , 17429-17436 26. Cornbeau: C.' and Carlier, M. F. (1989) J. Bid. Chem. 2 6 4 , 19017-19021 27. Murphy, A. d., and Coll, R. J. (1992) J. Biol. Chem. 2 6 7 , 5229-5235

F.