A GTP-dependent Step in the Activation Mechanism of Capacitative Calcium Influx*

Calcium influx in electrically non-excitable cells is regulated by the filling state of intracellular calcium stores. Depletion of stores activates plasma membrane channels that are voltage-independent and highly selective for Ca2+ ions. We report here that the activation of plasma membrane Ca2+ currents induced by depletion of Ca2+ stores requires a diffusible cytosolic factor that washes out with time when dialyzing cells in the whole-cell configuration of the patch-clamp technique. The activation of calcium release-activated calcium current (IcRAc) by ionomycin-or inositol 1,4,5-trisphosphate-in-duced store depletion is blocked by guanosine 5’-3-0-(thi0)triphosphate (GTPyS) and guanyl-5’-yl imidodi-phosphate, non-hydrolyzable analogs of GTP, suggesting the involvement of a GTP-binding protein. The inhibition by GTP+ occurs at a step prior to the activation of ZcmC and is prevented by the addition of GTP. We conclude that the activation mechanism of depletion-in-duced Ca2+ influx encompasses a GTP-dependent step, possibly involving an as yet unidentified small GTP-binding protein. Capacitative

lar depletion-activated Ca2+ currents have been identified in J u r k a t T cells (7, 8) and other cell types (2).In this study, we investigated IcMc in rat basophilic leukemia cells (RBL-2H31, a mast cell line that is advantageous for several reasons: 1) it is an easily accessible cell line; 2) cell size and Ca2+ currents are larger compared with peritoneal mast cells; 3) secretory activ- ity is smaller and much slower than in peritoneal mast cells.Store depletion in RBL cells by InsP3, ionomycin, or Ca2+ chelators activates a current that is essentially indistinguishable from ZCmc in peritoneal mast cells (6) in terms of kinetics, current-voltage relationship, Ca2+ selectivity, Ca2+-induced inactivation, and block by divalent ions (data not shown).In the experiments of this study, ZCmc was activated by photoreleasing intracellular caged InsP3 or applying ionomycin extracellularly at appropriate times after establishment of the wholecell configuration of the patch-clamp technique.Pipette solutions were designed to probe the involvement of diffusible messengers, in particular GTP-binding proteins.[Ca2+li was buffered to about resting levels with a mixture of 8 m~ BAPTA and 2 mM Ca-BAPTA (free [Ca2+li -60 nM) to prevent spontaneous activation ofZCMC, which occurs when [Ca2+li is buffered to very low values (5).Pipette solutions contained Cs+ instead of K+ to avoid G protein-mediated activation of outward potassium currents (9).

MATERIALS AND METHODS
Rat basophilic leukemia cells (RBL-2H3) were cultured on glass cov- erslips with Dulbecco's modified Eagle's medium supplemented with fetal calf serum (lo%), NaHC0, (45 mM), glucose (5 mM), streptomycin (0.12 mglml), and penicillin (0.60 mglml).Patch-clamp experiments were performed in the tight-seal whole-cell configuration at 23-27 "C in a Ringer's solution containing (in mM): NaCl, 140; KC1, 2.8; CaC12, 10; MgCl,, 2; glucose, 11; HepewNaOH, 10, pH 7.2.Sylgard-coated patch pipettes had resistances between 2 and 4 megaohms after filling with the standard intracellular solution, which contained (in mM): cesium glutamate, 145; NaC1,8; MgCl,, 1; Fura-2, 0.1; Mg-ATP, 0.5; BAPTA (8 mM); Ca-BAPTA (2 mM); Hepes.CsOH, 10, pH 7.2.This solution was supplemented with GTPyS, GDPPS, GDP, GTP, GMP-PNP, ATPyS, ATP, NaF, and AlCl,, or caged InsP, (Calbiochem) when needed.For nucleotide concentrations above 0.5 mM, equimolar amounts of MgClz were also included.Extracellular solution changes were made by short application (5-10 s) from a wide-tipped micropipette that contained standard Ringer's supplemented with ionomycin (14 PM, Calbiochem).The concentration of [Caz+li was monitored with a photomultiplier-based system as described (10) in order to control the efficacy of Ca2+ buffers in the cell.All cells included in the analysis had no significant changes in [Ca2+li.For experiments with caged InsP,, the filter wheel (360/390 nm) was stopped in the dark position before starting an experiment, and a short illumination at 360 nm was delivered when needed.A 10-s flash was enough to fully activate I c u c when 400 PM caged InsP, was included in the patch pipette.Low resolution currents were sampled at 2 Hz and filtered at 500 Hz.High resolution current recordings were acquired by a computer-based patch-clamp amplifier system (EPC-9; HEKA, Lambrecht, Federal Republic of Germany).Capacitive currents and series resistance were determined and corrected before each voltage ramp using the automatic capacitance compensation of the EPC-9.Holding potential was usually 0 mV.Most of the current amplitudes were determined at a potential of -40 mV, taken from high resolution currents in response to voltage ramps.Currents were filtered at 2.3 kHz and digitized at 100-PS intervals.For analysis and presentation, currents were filtered digitally to l kHz.

RESULTS AND DISCUSSION
To test for washout of diffusible cytosolic constituents required for signaling the filling state of Ca2+ stores to plasma membrane Ca2+ channels, we dialyzed RBL cells for variable periods of time prior to emptying the stores by the Ca2+ ionophore ionomycin.As shown in Fig. IA, at the instance of breakin, there was an initial outward current, likely due to a small K+ conductance at 0 mV, which rapidly decayed to zero as the cesium-containing pipette solution equilibrated in the cytosol.Thereafter, depletion of Ca2+ stores by ionomycin activated a small inward current at 0 mV.Voltage dependence as assessed by voltage ramps (Fig. lA, right panel), and kinetic behavior of the inward current identified it as the previously characterized Z C ~C (5,6).Depletion-activated Ca2+ currents were larger when depleting stores early in the experiment as compared with currents activated at later times.To compare the amplitude among different cells we corrected the traces for cell capacitance (which is a measure for cell size) and for series conductance (which determines the rates of diffusional exchange between cell and pipette) (11).The details are given in the figure legend.conductance of 174 nS of all these experiments, the time constant for diffusible factor is a fairly large molecule rather than a nucleotide, inositol phosphate, or other small second messenger; these would washout in less than 1 min (11).Nevertheless, we tested a number of second messengers and metabolites (CAMP, 50 p ~; inositol 1,3,4,5-tetrakisphosphate, 100 VM) and found none to significantly activate or inhibit Zcmc when applied intracellularly.However, although there was no apparent requirement for GTP in the pipette solution, we did find an inhibitory effect of non-hydrolyzable analogs of GTP on the activation of Zcmc.Fig. 2A illustrates that intracellular application of GTPyS (100 PM) prevents the ionomycin-induced activation of ZcmC normally seen in control cells.The relationship between the time of store depletion (after establishment of the whole-cell configuration) and the resulting amplitude of Zcmc in the presence of GTPyS is shown in Fig. 2B ( not shown).GDPPS (up to 500 PM) had no inhibitory effect on I C ~C (Fig. 2A).Indeed, in the presence of GDPPS current amplitudes were slightly larger than control currents (Fig. 2B, right panel, open triangles).One possible reason for the increased amplitudes in the presence of GDPPS may be that I C ~C is tonically inhibited by basal trimeric G protein activity, possibly through kinase phosphorylation (see AlF; effects below).The inhibition of trimeric G proteins by GDPPS may relieve this tonic modulation, resulting in an enhanced response following store depletion.
To exclude that GTPyS interferes with ionomycin-induced store depletion we used photohydrolyzable caged InsP3 to activate I C ~C at later times.As depicted in Fig. 3A  calculated with ionomycin, both in the absence and presence of GTPyS.
We asked whether the inhibition of Icmc by GTPyS occurred prior to the activation of Icmc or whether it was due to a modulatory effect, possibly resulting from activation of G protein-mediated negative feedback mechanisms, which might be anticipated to regulate a process as paramount as Ca2+ influx.Interestingly, however, GTPyS does not significantly inhibit ICmc once it is activated as evidenced by Fig. 3B.In this experiment GTPyS was applied through the patch pipette, and ionomycin was applied shortly after break-in, resulting in a fast and sustained activation ofZcmc.Likewise, GTPyS was unable to reduce ICmc when current activation was induced even earlier by InsP3 co-applied through the patch pipette (Fig. 3, B and   C).Together, these results suggest that a GTP-dependent step is involved in activating I C ~C and that non-hydrolyzable analogs interfere with the activation at some crucial early step in the signal transduction pathway.A channel block by GTPyS itself, a G protein, or some other second messenger is highly unlikely, as one would have to postulate that channels can only be blocked in their closed conformation and, once activated, they prevail in a state of high open probability, rendering them no longer susceptible to GTPyS inhibition.In fact, however, open probability of activated channels seems to be low, as suggested by noise analysis of Icmc in Jurkat cells (7).
Primary targets of non-hydrolyzable GTP analogs are heterotrimeric (40-45 kDa) and monomeric GTP-binding proteins of small molecular mass (19-21 kDa) (12, 13).The former are activated irreversibly by non-hydrolyzable GTP analogs whereas the latter require GTP hydrolysis for functioning and are inhibited by GTPyS.Our results show that the main effect of GTPyS is to increase the rate of washout of the Icmc response to ionomycin, which we interpret to be the result of interference with the activation cycle of small G proteins.An alternative interpretation would be that GTPyS effects are mediated by a trimeric G protein whose washout is accelerated by dissociation of a and Py-subunits.Three arguments make this possibility seem unlikely.1) Washout occurs under non-stimulatory conditions and GDPPS, which should prevent dissociation of trimeric G proteins, does not alter the rate of run-down.

2)
GTPyS does not reduce the current once it has been activated, which would have been expected assuming a trimeric G protein mechanism.3) Similarly, GTP (like GTPyS) is expected to activate and dissociate trimeric G proteins, which in turn would enhance the washout and reduce Icmc.However, we find that GTP antagonizes GTPyS effects, a key feature of small G protein mechanism.
Pharmacological tools to discern between trimeric and small GTP-binding proteins are sparse.AlF; ions are thought to be specific activators of trimeric G proteins and ineffective at stimulating small GTP-binding proteins (141, prompting us to test AlF; for its effects on IcmC.Fig. 3 illustrates that AlF; indeed exhibited a different response pattern than GTPyS: perfusing cells with AlF; consistently activated Icmc.However, the MF;-induced current underwent a fast inactivation that could not be recovered by ionomycin addition (n = 8, Fig. 3B).In contrast to GTPyS, the AlF;-induced fast inactivation was still evident with early activation of Icmc by InsP3-induced store depletion (Fig. B and C).Furthermore, the effects of AlF; were not antagonized by 5 m~ GTP (data not shown).
The activation of Icmc by AlF; could be the result of store depletion by a massive stimulation of InsP3 production via activation of phospholipase C through a trimeric G protein.The inability of GTPyS to induce store depletion through trimeric G proteins could be accounted for by insufficient build-up of InsP3 to induce store depletion.We noted that in RBL cells, the minimal InsP3 concentration required to activate ZCmc (10 g ~) is about 1 order of magnitude higher than in peritoneal mast cells under identical conditions (K+-based pipette solutions, EGTA as Ca2+ chelator).In addition, the presence of Cs+ and BAPTA in the pipette further reduce the efficacy of InsP3, such that at least 25 p InsPs are required to activate Zcmc.Possible explanations for a reduced efficacy of GTPyS as compared to AIF; could be as follows.1) the GDP/GTP exchange rate of the G protein that couples to phospholipase C, which constitutes the rate-limiting step in GTPyS action, may be too slow to produce levels of InsP, that overwhelm removal processes; 2) GTPyS effects might be inhibited by BAPTA; in peritoneal mast cells Br2-BAPTA causes inhibition of Ca2+-independent GTPySinduced secretion (15).
The AlF;-induced inactivation of ZcCRAc is clearly different from the GTPyS-induced block of activation, as AIF; was still able to inhibit ZCMc after it had been activated.For the reasons outlined above, GTPyS does not seem to effectively stimulate trimeric G proteins and therefore would not induce this inhibitory modulation.The inhibitory effects of AIF; may be due to a modulation of the current itself, possibly through activation of negative feedback mechanisms secondary to trimeric G protein stimulation (e.g.phosphorylation through protein kinases).In addition, NaF is known to inhibit phosphatases (16) that might act synergistically to promote phosphorylation processes.Hence, while it is apparent that GTPyS and AlF; exhibit obvious differences in their effects on cellular responses, they seem to do so with respect to both trimeric and monomeric G proteins.
In summary, our results are consistent with the notion that the signal for capacitative Ca2+ entry following depletion of intracellular Ca2+ stores involves a GTP-dependent step.It is tempting to speculate that this step encompasses a small G protein that is blocked by non-hydrolyzable GTP analogs.If this interpretation is correct one could imagine at least three, admittedly speculative, models of how capacitative Ca2+ influx might be activated by small G proteins. 1) For direct G protein activation of Zcmc there must be a Ca2+ sensor of the luminal Ca2+ concentration that translates store depletion into activation of the G protein.This is possibly aided by an accessory protein, such as GDS (GDP dissociation stimulator), which catalyzes GDP-GTP exchange (17).Then the G protein acts as a messenger to gate the Ca2+ channels in the plasma membrane.Washout of the small G protein itself, or of accessory proteins, may be responsible for the observed current rundown.Termination of the activated state of the G protein in the intact cell may involve other accessory proteins such as GAPS (GTPase-activating proteins), which will catalyze hydrolysis of GTP (13, 17) and render the G protein inactive.This accessory protein may also be subject to washout in whole-cell experiments, which would result in an irreversible activation of Zcmc.2) For indirect G protein activation of Zcmc the G protein could act as an intermediate signal to produce another as yet unidentified diffusible messenger.3) G protein-induced fusion activates Zcmc; this could be proposed on the basis of the well documented role of small G proteins in vesicular traffic (13,18-20).It is conceivable that activation of a small G protein induces fusion of specialized vesicles that carry Ca2+ channels.These channels would then insert in the plasma membrane, and Ca2+ influx can occur as long as they remain in the membrane.Termination of the current through these new channels could be accomplished by endocytic activity.

FIG. 1 .
figure legend.Fig. LB (left panel shows the normalized traces of Fig. IA (left panel).The relationship between the time of store depletion after break-in (normalized for series conductance) and the resulting amplitude of Zcmc (normalized for cell capacitance) is depicted in Fig. LB (right panel) for all tested cells.The relationship follows an exponential time course with a time constant of about 250 s (considering the average series conductance of 174 * 74 nS, mean * S.D., n = 91).The gradual decrease in the maximum amplitude of Zcmc with prolonged whole-cell dialysis suggests that activation of Zcmc requires a cytosolic component that diffuses out of the cell during the time of dialysis.The rather slow time course of washout indicates that the

FIG. 2 . 1 )
FIG. 2. GTP-@ inhibits ionomycin-induced ZCRllc.A, ionomycininduced ICac under different experimental conditions (holding potential, 0 mV).As indicated in the graphs, the standard pipette solution was supplemented with GTPyS (100 p ~) , GTPyS (100 PM) plus GTP (5 mM), GDPpS (500 PM), GMP-PNP (500 PM), or ATPyS (100 PM).The arrows indicate the start of the ionomycin application (5-10 8).B, current amplitudes of ICmc at -40 mV as a function of time delay of panel the data for GTPyS ( n = 42) were pooled (0) and fitted with a ionomycin application (normalized as described in Fig. 1).In the left single exponential function A exp (-t/T), where A = -2.9pNpF and T = 16.3 ~. 1 0 -~ S, yielding a time constant of about 80 s for the current decrease (considering the mean series conductance of 201 nS, 42 experiments).The fitted curve under control conditions (taken from Fig. 1) The right panel depicts the two fitted curves (control and GTPyS) and and the points for GTPyS plus GTP (0) are also shown in the left panel.the points for GDPpS (A, = 100 PM, V = 500 PM), ATPyS (01, and GMP-PNP (A).

FIG. 3 .
FIG. 3. GTP+ and AlFi act in different ways.A, GTPyS inhibitsInsP,-induced activation of IcRAc.Data points represent current amplitudes of Icmc at -40 mV in the absence (A) and presence of 100 p.~ GTPyS (0) as a function of time delay of photoreleased InsP, (400 p ~) .Currents were normalized as described in Fig.1.For comparison, the two curves for ionomycin-induced current amplitudes under control conditions and in the presence of GTPyS are also shown (taken from Fig.2B).The insets illustrate two examples of current responses (A = control, 0 = 100 PM GTPyS) recorded at 0 mV upon release of caged InsP, (time of photolysis is indicated by arrows).B, GTPyS does not significantly inhibit Icuc when it is activated by applying ionomycin quickly after break-in or by co-perfusion with InsP, (upper truces).In contrast, AIF; (5 mM NaF, 50 PM AlC1,) induces a fast inactivation of the current activated by AlF; alone, or by co-perfusion of InsP, (lower truces).C, the relative amount of current inactivation at a holding potential of 0 mV after 390 s of perfusion of the cell is shown for different activation procedures (as labeled in the graph).Current inactivation represents the ratios of current amplitude after 390 s and the maximal current.Data are means * S.D. of experiments in which ICmc was activated by ionomycin (n = 9).InsP, (n = lo), ionomycin + GTPyS (n = 7), InsP, + GTPyS (n = 12), AlF; (n = 9), AlFa + InsP, (n = 10).Only experiments in which activation of the current was induced within 100 s following break-in were taken into account.