GTP- and inositol 1,4,5-trisphosphate-activated intracellular calcium movements in neuronal and smooth muscle cell lines.

Recent evidence has revealed that a highly sensitive and specific guanine nucleotide regulatory process controls intracellular Ca2+ release within N1E-115 neuroblastoma cells (Gill, D. L., Ueda, T., Chueh, S. H., and Noel, M. W. (1986) Nature 320, 461-464). The present report documents GTP-induced Ca2+ release within quite distinct cell types, including the DDT1MF-2 smooth muscle cell line. GTP-induced Ca2+ release has similar GTP sensitivity and specificity among cells and rapidly mobilizes up to 70% of Ca2+ specifically accumulated within a nonmitochondrial Ca2+-pumping organelle within permeabilized DDT2MF-2 cells. Maximal GTP-induced release of Ca2+ is observed to be greater than inositol 1,4,5-trisphosphate (IP3)-induced Ca2+ release (the latter being approximately 30% of total releasable Ca2+). After maximal IP3-induced release, further IP3 addition is ineffective, whereas subsequent addition of GTP further releases Ca2+ to equal exactly the extent of Ca2+ release observed by addition of GTP in the absence of IP3. This suggests that IP3 releases Ca2+ from the same pool as GTP, whereas GTP also releases from an additional pool. The effects of GTP appear to be reversible since simple washing of GTP-treated cells restores their previous Ca2+ uptake properties. Electron microscopic analysis of GTP-treated membrane vesicles reveals their morphology to be unchanged, whereas treatment of vesicles with 3% polyethylene glycol, known to enhance GTP-mediated Ca2+ release, clearly induces close coalescence of membranes. In the presence of 4 mM oxalate, GTP induces a rapid and profound uptake, as opposed to release, of Ca2+. The findings suggest that GTP-activated Ca2+ movement is a widespread phenomenon among cells, which can function on the same Ca2+ pool mobilized by IP3, and although activating Ca2+ movement by a mechanism distinct from IP3, does so via a process that does not appear to involve fusion between membranes.


GTP-and Inositol 1,4,5-Trisphosphate-activated Intracellular Calcium Movements in Neuronal and Smooth Muscle Cell Lines*
(Received for publication, May 19, 1987) Sheau-Huei Chueh, Julienne M. Mullaney Recent evidence has revealed that a highly sensitive and specific guanine nucleotide regulatory process controls intracellular Ca2+ release within N1E-115 neuroblastoma cells (Gill, D. L., Ueda, T., Chueh, S . H., and Noel, M. W. (1986) Nature 320, [461][462][463][464]. The present report documents GTP-induced Ca2+ release within quite distinct cell types, including the DDTIMF-2 smooth muscle cell line. GTP-induced Ca2+ release has similar GTP sensitivity and specificity among cells and rapidly mobilizes up to 70% of Ca2+ specifically accumulated within a nonmitochondrial Ca2+-pumping organelle within permeabilized DDT2MF-2 cells. Maximal GTP-induced release of Ca2+ is observed to be greater than inositol 1,4,5-trisphosphate (IP3)-induced Ca2+ release (the latter being approximately 30% of total releasable Ca2+). After maximal IP3-induced release, further IP3 addition is ineffective, whereas subsequent addition of GTP further releases Ca2+ to equal exactly the extent of Ca2+ release observed by addition of GTP in the absence of IP3. This suggests that IP3 releases Ca2+ from the same pool as GTP, whereas GTP also releases from an additional pool. The effects of GTP appear to be reversible since simple washing of GTP-treated cells restores their previous Ca2+ uptake properties. Electron microscopic analysis of GTPtreated membrane vesicles reveals their morphology to be unchanged, whereas treatment of vesicles with 3% polyethylene glycol, known to enhance GTP-mediated Ca" release, clearly induces close coalescence of membranes. In the presence of 4 m M oxalate, GTP induces a rapid and profound uptake, as opposed to release, of Ca2+. The findings suggest that GTP-activated Ca2+ movement is a widespread phenomenon among cells, which can function on the same Ca2+ pool mobilized by IP,, and although activating Ca2+ movement by a mechanism distinct from IP3, does so via a process that does not appear to involve fusion between membranes.
The processes involved in calcium signaling within cells have been the focus of much recent attention, especially with * This work was supported by Grant NS19304 from the National Institutes of Health and Grant DCB-8510225 from the National Science Foundation (to D. L. G.). Electron microscopy (performed by A. L. 2.) was supported by Grant A111676 from the National Institutes of Health. 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.
The abbreviations used are: IPS, inositol 1,4,5-trisphosphate; ER, endoplasmic reticulum; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; EGTA, [ethylene bis(oxyethylenenitrilo)]tetraacetic acid; GTP-yS, guanosine 5'-0-(3-thio)triphosphate; GppNHp, guanosine 5'-(&-y-imido)triphosphate; GDPPS, guanyl-5"yl thiophosphate. regard to the production and action of the Ca"-mobilizing messenger, inositol 1,4,5-trisphosphate. Many studies now show that IP31 mediates Ca2+ release from an internal organelle believed to be endoplasmic reticulum (1)(2)(3). Recent evidence has revealed that in addition to the release of Ca'+ activated by IP3, a highly sensitive and specific guanine nucleotide regulatory process functions to promote Ca2+ release from a similar intracellular Ca'+ pool (4,5). Using either permeabilized cells of the neuronal cell line N1E-115 (4) or microsomal vesicles derived therefrom (5), our experiments have revealed that the latter process promotes a rapid and substantial release of Ca2+ accumulated within permeabilized cells or vesicles via internal (ATP + Me)-dependent Ca2+ pumping activity characterized earlier (6). Activation of Ca2+ release is induced by submicromolar GTP concentrations and shows high specificity for guanine nucleotides (4,5). The release process appears to be dependent on terminal phosphate hydrolysis of GTP. This is concluded from the following observations: first, nonhydrolyzable GTP analogues are ineffective in inducing Ca2+ release; second, such analogues block the action of GTP; third, GTP-mediated Ca2+ release is competitively inhibited by GDP indicating that the two nucleotides can bind to a common site (4). At present it is unclear whether the activation process involves terminal phosphate transfer to water (as in the case of a GTPase reaction), or whether a kinase-mediated process transfers phosphate to another substrate molecule. Evidence for the former was recently presented by Nicchitta et al. (7), whereas evidence for a GTP-induced protein phosphorylation possibly associated with Ca2+ release was reported by Dawson et al. (8).
In earlier studies, Dawson et al. (8,9) observed that GTP could promote the effectiveness of IP3 in inducing Ca'+ release from liver microsomes, whereas GTP or IP3 alone induced little or no release of Ca2+. In contrast, our results with either permeabilized N1E-115 cells (4) or isolated microsomal vesicles (5), reveal that GTP promotes substantial ca'+ release without any exogenously added IPS. Moreover, we recently reported that a number of important distinctions exist between the actions of IP3 and GTP in releasing intracellular Ca'+ (10). First, IP3-activated Ca2+ release is unaffected by either GDP or GTP-yS, both of which block GTP-activated Ca'+ release. Second, polyethylene glycol promotes GTPactivated release but does not alter the action of IP3. Third, IP3-activated Ca'+ release is clearly modified by the free Ca2+ concentration being completely inhibited at 10 PM Ca'+; in contrast, GTP induces Caz+ release independent of the free Ca2+ concentration. Fourth, the actions of IP3 and GTP show very distinct temperature dependence: GTP-induced Ca2+ release is completely blocked by decreasing the temperature from 37 to 4 "C, whereas the rate of IP3-induced Ca2+ release is decreased by only 20% at 4 "C, a result consistent with the observations of Smith et al. (11). Several of these distinctions between the actions of GTP and IP3 have also been reported by Henne and Soling (12) using liver and parotid microsomes, and by Jean and Klee (13) using microsomes from NG108-15 neuroblastoma X glioma hybrid cells. In summary, the relative temperature insensitivity and rapidity of Ips-induced Ca2+ release are consistent with its probable direct activation of a Ca2+ channel in ER. In contrast, GTP appears to effect release by a temperature-sensitive process probably involving enzymic hydrolysis of the terminal phosphate from GTP.
Although the modes of activation of Ca2+ release by IP3 and GTP are clearly distinct, their effects are similar; indeed, differences in their activation mechanisms do not preclude possible coupling between their actions, as indicated by Dawson (8,9). Thus, one goal of the studies described here is to ascertain more information on the relationship between the actions of GTP and IP3, particularly, whether the two effectors operate upon the same pool or distinct pools of Ca2+. A second aim of the studies reported here is to gain further information on the actual movements of Ca2+. The activation of GTP-dependent Ca2+ release probably involves an enzymic GTP hydrolytic mechanism. But what is the actual Ca2+ ion translocation process induced by GTP? Does the mechanism involve channel activation, or, alternatively, could some form of membrane fusion or communication between membranes explain the GTP-activated Ca2+ release? The enhancing effect of polyethylene glycol suggests that a possible interaction between membranes may be involved (4, 5, 10). Indeed, electron microscopic evidence described here directly reveals the effectiveness of polyethylene glycol in promoting close appositions between membranes, presumably by dehydrating membrane surfaces. However, the results reported here indicate that fusion of membranes is unlikely to explain the actions of GTP. Importantly, as a direct result of approaching the question how Ca2+ ions are released, we observed that GTP can induce a profound and apparently entireIy distinct activation of uptake as opposed to release of Ca2+. Subsequent characterization of this uptake phenomenon leads us to postulate that GTP may activate a conveyance of Ca2+ across membranes and between organelles. The latter studies on GTP-activated Ca2+ uptake are described separately in the following report (14).

EXPERIMENTAL PROCEDURES
Culture of NlE-115 Cells-Procedures for the culture of the N1E-115 neuroblastoma cell line have been described previously (6). Cells were originally obtained from Dr. Marshall Nirenberg (National Institutes of Health, Bethesda) and were utilized in studies between passage 16 and 30. Cells were passaged every 14 days, and experiments were undertaken on cells grown for 8-12 days after subculture, at which time Ca2+ pumping activity is maximally expressed (6).
Culture of DDTIMF-2 Cells-Cells of the smooth muscle cell line DDT,MF-2 were obtained from Dr. James Norris (University of Arkansas Medical Center). These cells were originally cloned from a steroid-induced tumor in hamster vas deferens by Norris et al. (15). The cells had been maintained in continuous culture prior to our receiving them and were of indefinite passage number after the original cloning. DDTIMF-2 cells were grown under similar conditions to those we described previously for N1E-115 cells (6), with the following exceptions: the Dulbecco's modified Eagle's medium (high glucose) was supplemented with 5% as opposed to 10% fetal bovine serum; cells were passaged every 7 days, and medium changes were performed on days 3, 5, and 6 after subculture. Cells used in experiments had been grown for between 3 and 7 days after subculture.
Permeabilization of Cells-The procedures for cell permeabilization were based on those of Burgess et al. (16) and were exactly as described in previous reports (4)(5)(6)10). Briefly, N1E-115 or DDT,MF-2 cells removed from culture dishes using Ca2+/Mg2+-free medium, were treated with 0.005% saponin at 37 "C (in uptake medium; see below), followed by three thorough washes with saponin-free medium. The exact time of saponin treatment was determined by assessing cell permeability by trypan blue exclusion, and treatment was terminated when 95% of cells were rendered permeable to the dye. Cells were saponin treated at a concentration of approximately 2 X lo6 cells/ml, and the time for treatment varied according to cell type, generally between 5-10 min for either N1E-115 cells or DDTIMF-2 cells. After permeabilization cells were kept slowly stirring at 4 "C and were used in experiments within 2 h.
Calcium Flux Experiments-Measurements of Ca2+ fluxes using permeabilized cells were undertaken essentially as described in previous reports (4)(5)(6)10). During uptake and release experiments, cells were maintained in suspension by gentle stirring in glass vials which contained medium approximating intracellular ionic conditions and comprising: 140 mM KCl, 10 mM NaCl, 2.5 mM MgCL, 10 mM Hepes-KOH, pH 7.0 (referred to as uptake medium). All experiments were conducted at 37 "C, with cells present at 2-5 X lo5 cells/ml in a total reaction volume of between 2 and 4 ml. CaZ+ flux incubations were started by adding permeabilized cells suspended in medium and prewarmed to 37 "C for 1 min, to an equal volume of uptake medium containing sufficient uptake components to give the following final conditions (unless otherwise stated): 1 mM ATP, 50 pM CaClz (containing 80 Ci/mol 45Ca2+), 3% polyethylene glycol, and EGTA to buffer free Ca2+ to exactly 0.1 p~ (see below). At the desired times, 200-pl aliquots were removed from the stirred uptake medium, diluted immediately into 3 ml of uptake medium containing 1 mM LaCI3, and vacuum filtered on glass fiber filters which were then rapidly washed three times with 3 ml of the same La3+-supplemented uptake medium. Where stated, mitochondrial inhibitors (5 p~ ruthenium red and/or 10 p~ oligomycin) were included in the uptake medium. Unless otherwise stated, these inhibitors were present in all experiments undertaken in the presence of free Ca2+ levels in excess of 0.1 p~. In most experiments conducted at 0.1 p~ free CaZ+, mitochondrial inhibitors were not included since previous studies (6) conclusively demonstrated a negligible mitochondrial component of Ca2+ uptake under this condition. ATP-regenerating conditions (creatine phosphate and creatine phosphokinase) used in some previous studies (6) were avoided since they interfered with the conversion of nucleoside diphosphates and since they were found to be unnecessary in view of the minimal decline of ATP concentrations during experiments.
Ekctron Microscopy-Thin section electron microscopy was performed on samples of crude microsomal vesicles (prepared from N1E-115 cells exactly as described in Ref. 5) which had been treated under identical conditions to cells in the Ca2+ uptake procedure described above. Samples of the treated vesicles were fixed by adding one volume of suspended vesicles (at approximately 0.5 mg of membrane protein/ml) in uptake medium to an equal volume of 2% glutaraldehyde in 20 mM phosphate buffer, pH 7.2. After 50 min at 4 "C, vesicles were pelleted at 12,000 X g for 5 min, and treated with 1% buffered osmium tetroxide for 18 h at 4 "C. The samples were then dehydrated using a graded acetone series and embedded in Epon 812 epoxy plastic. Thin sections were cut and stained with uranyl acetate and lead citrate prior to visualization in a Siemens 1A electron microscope.
Free Ca2+ concentrations were controlled using EGTA and computing all complexes between EGTA, ATP, Ca2+, M e , monovalent cations, and protons, as described previously (6) using the stability constants and computer program described by Fabiato and Fabiato (17). Trypan blue dye exclusion, observed under phase contrast microscopy was measured using 0.5% trypan blue in uptake medium during every cell permeabilization procedure. Both cells and media were mycoplasma-free as tested by isolation from cell culture and Hoechst stain (Microbiological Associates). ATP and EGTA solutions were adjusted to pH 7.0 with NaOH and KOH, respectively.

RESULTS AND DISCUSSION
Cell Specificity of GTP-mediated Calcium Release-Using permeabilized cells of the N1E-115 neuronal cell line we have conducted a detailed examination of the Ca2+ accumulation and release properties of an intracellular nonmitochondrial compartment believed to be the ER (4-6). These studies reveal that Ca2+ can be accumulated via a high affinity (ATP + M&+)-dependent Ca2+ pump (6), shown by a variety of criteria to be distinct from that we established earlier as functioning in the neural plasma membrane (18)(19)(20). More recent studies clearly indicate that a fraction of this accumulated Ca2+ is released by IP3 (4,5). Half-maximal Ca2+ release is effected with 0.5 p M IPS (4), a sensitivity very similar to values reported in a number of other systems (1-3). However, not seen before was the extraordinarily sensitive and profound release of Ca2+ induced by GTP we observed using either permeabilized N1E-115 cells (4,5) or microsomal membrane vesicles derived from these cells (5). In contrast, the studies of Dawson et al. (8,9) using liver microsomes have shown that GTP, while significantly enhancing the effectiveness of IP3 in inducing Ca2+ release, by itself induced only a very modest effect.
In view of these differences, it was important to establish whether the effectiveness of GTP in directly induced Ca2+ release was an anomoly, perhaps restricted to the N1E-115 neuronal cell line. Experiments undertaken on a quite unrelated cell type, the DDTlMF-2 smooth muscle cell line derived from hamster vas deferens (15), suggest this is not the case. Thus, as shown in Fig. 1, a sensitive, specific, and substantial GTP-dependent release of Ca2+ is observed using permeabilized DDT1MF-2 cells loaded with Ca2+ via intracellular Ca2+ pumping activity. Such release is very similar to that observed with N1E-115 neuroblastoma cells, with pronounced effectiveness of submicromolar GTP concentrations, even in the presence of 1 mM ATP. Maximally effective GTP levels (5 p~ or above) effect a rapid release of approximately 70% of the ionophore-releasable Ca2+. In addition to the DDTI-MF-2 cell line, we have measured almost identical effects of GTP on Ca2+ release from permeabilized rat BC3H-1 smooth muscle cells and human WI-38 normal embryonic lung fibroblasts. Using microsomal membrane vesicle fractions prepared from DDTIMF-2 cells by methods similar to those described for NlE-115 cell-derived microsomes (5), we have observed GTP effects on Ca2+ release almost identical to those seen with permeabilized cells. Furthermore, using microsomes derived from guinea pig parotid gland, Henne and Soling (12) have observed very similar effects on release of accumulated Ca2+ induced by GTP; in this study, GTP-activated Ca2+ movements were followed using Ca2+ electrodes. The observations of Jean and Klee (13) on GTP-and IP3-mediated Ca2+ release from microsomes derived from NG108-15 neuroblastoma X glioma hybrid cells are also very consistent with each of the above studies.
Therefore, it appears that the effectiveness of GTP in directly activating Ca2+ release is a widespread phenomenon, not restricted to particular cell types or to the use of specific cellular or subcellular preparations. However, it does appear from the observations of Dawson (8,9) using liver microsomes, and from preliminary studies of Thomas and Rubin (21) using permeabilized hepatocytes, that the liver responds somewhat differently, GTP giving a slower and less substantial release of Ca2+ relative to those systems mentioned above, and more significantly, a GTP-mediated potentiation of the effectiveness of IP,. Recent preliminary observations of our own' using liver microsomal membrane fractions confirm certain of these observations on the actions of GTP. Possible explanations for such differences are commented upon in the accompanying paper (14)   Subcellular Specificity of GTP-activated Calcium Release-It is widely held that IP3 releases Ca2+ from a nonmitochondrial intracellular organelle presumed to be the ER (1-3). This view is based largely on circumstantial evidence that ER is known to be a major Ca'+-sequestering organelle able to accumulate Caz+ via a high affinity pump (2,6). Our previous studies on the GTP-activated Ca2+ release process (4,5) suggest that Ca2+ is also released from a nonmitochondrial organelle since release was observed in the presence of mitochondrial inhibitors and under free Ca2+ conditions (0.1 p~) which do not permit significant mitochondrial uptake. However, it was important to determine whether the effect of GTP on release was specific to a nonmitochondrial Ca2+ pool, or whether Ca2+ accumulated within mitochondria could also be released or affected in any way by GTP. As shown in Fig. 2, it is clear that GTP does not alter mitochondrial Ca2+. In this experiment, Ca2+ uptake and release were undertaken at a high free Ca2+ concentration (10 p~) , using permeabilized N1E-115 cells that were loaded under otherwise standard conditions in the presence of ATP. In the presence of the release from within permeabilized N1E-115 cells. N1E-115 cells were permeabilized and loaded with 'Ta2+ as described under "Experimental Procedures," except that the total Ca2+ concentration during Ca2+ flux incubations was 10 p~, and EGTA was not present. Both uptake and release were undertaken either in the presence of 10 pM oligomycin (A, A) or without oligomycin (0, 0), the latter condition permitting uptake into both mitochondrial and nonmitochondrial pools. At the end of a 10-min uptake period, either 10 p~ GTP (0, A) or control buffer (0, A) was added to vials. Samples of the mixture were withdrawn at the intervals shown after these additions followed by rapid filtration and washing as described under "Experimental Procedures." mitochondrial ATPase inhibitor oligomycin at 10 pM, mitochondrial Caz+ uptake is completely abolished (6), and under such conditions 10 p~ GTP effects more than 50% release of accumulated Caz+ (Fig. 2, lower curues). Without oligomycin, considerably higher (approximately 4-fold) Ca2+ uptake is observed since mitochondria can accumulate Ca2+ at the higher free Ca2+ level included in the experiment. However, upon addition of GTP, only approximately 12% of the accumulated Ca2+ is released (Fig. 2, upper curues). In fact, this represents an almost identical amount of GTP-mediated Ca2+ release (approximately 3 nmol/106 cells) to that observed in the presence of oligomycin. Therefore, the effects of GTP are indeed specific to a nonmitochondrial Ca2+-sequestering pool. It should be noted that a similar experiment cannot be undertaken to examine the effects of IP3 because, as we showed in a recent report (lo), at this concentration of free Ca2+ (10 FM) the effect of IPS is completely inhibited. Such inhibition reflects a potentially important Ca2+-dependent feedback on IP3-mediated release which may have much significance in the physiological role of IPS within cells.
We previously observed ( 5 ) that IP3 and GTP were without any effects on the movement of Ca2+ across isolated inverted plasma membrane vesicles derived from neural tissue (18)(19)(20). However, it is important to note that there may be a flaw in this more conventional approach to determining subcellular specificity by isolating discrete fractions enriched in certain organelles. Thus, as described in the accompanying report (14), it is possible that interactions between more than one type of membrane may be involved in the Caz+ releaseinducing effects of GTP; hence, isolation and separation of organelles may reduce interactions between specific membranes and therefore prevent the effects of GTP. Although we have no direct proof that ER is a source of GTP-releasable CaZ+, interpretation of the effects of oxalate, a known permeator of the ER membrane, may indicate that the ER is indeed a site of action of GTP, as described later.
Relationship between GTP-and IP3-releasable Calcium Pools-Perhaps as important as establishing the actual identity of the source(s) of Ca2+ released in response to IP3 and GTP, is the determination of whether the two agents activate release from the same or different CaQC pools. In addition, this determination had fundamental significance to the question of whether coupling between the actions of GTP and IP3 could exist in cells. Thus, although our recent studies clearly establish that GTP and IP3 activate Ca2+ release via distinct mechanisms (lo), Dawson's results using liver microsomes indicate a potentiating action of GTP on IPa-induced Ca2+ release (8,9). It became clear to us that while the activation processes for Ca2+ release in response to GTP or IP3 are distinct, coupling between their actions at some level was still a possibility; thus, of major significance to the question of whether or how such coupling could occur was to establish if IP3 and GTP have a common site of action.
The question of whether IP3 and GTP activate the same or different Ca2+ pools is directly addressed in the study depicted in Fig. 3; however, the answer is not as straightforward as might be predicted. In this experiment, the effects of sequential addition of maximally effective concentrations of IP3 (10 p~) and GTP (10 p~) were examined in permeabilized N1E-115 cells. When IPS is added to fully Ca*+-loaded cells (Fig.   3A), approximately 30% of the accumulated Ca2+ is rapidly lost, a fractional release which is very constant for these cells. After maximal release is attained, if a further addition of IP3 is made, little further Ca2+ is released. This result suggests that it is unlikely that IP3 degradation is a limiting factor in Ca2+ release. Indeed, if this were the case, Ca" reaccumulation would be expected after maximal release, as has been observed in other systems where significant IP, metabolism presumably does occur (22)(23)(24). Under our conditions, using washed permeabilized cells in the presence of more than 1 mM free Mg+, significant IP3 breakdown due to particulate 1P3 5-phosphatase activity would be expected (3). However, two other factors probably contribute to diminished IP, degradation. First, IP3 3-kinase activity, predominantly a cytosolic enzyme (3), would be largely lost in washed permeabilized cells, and second, the dilute cell suspension used (approximately 2 X lo5 cells/ml) would lessen the effectiveness of IP3-modifying enzymes. As shown in Fig. 3A, if GTP is added after maximal IP,-induced Ca2+ release has occurred, than additional release of Ca2+ is observed until finally about 55% of the originally accumulated Ca2+ is released. Significantly, this degree of Ca2+ release is almost exactly the same as that observed when GTP is the initial addition, without any IPS (Fig. 3B). In this case, a further addition of GTP gives little additional Ca2+ release; moreover, addition of IP, at this time gives almost no greater release than that seen with GTP. It is clear that no change in the releasability of accumulated Ca2+ has occurred since A23187 added at any stage of the experiment causes similar release of Ca2+ down to the passive equilibration level. These results suggest that the releasable Ca2+ accumulated within permeabilized N1E-115 cells exists in three distinguishable subcompartments. First, there appears to be a pool of Ca2+ which is releasable with either GTP or IPS, second, a Ca2+ pool which is releasable only with GTP, and third, a pool which cannot be released with either effector. The fact that comparable results have been obtained using the DDT,MF-2 cell line lends support to this conclusion. Thus, it is apparent that, although in both cell types the GTPreleasable pool differs from the IP3-releasable pool in being larger, at least a significant proportion of accumulated Ca2+ lies within a pool which can be released by both agents. In other words, it appears that all of the Ca2+ within the IP3sensitive Ca2+ pool is also releasable by the GTP-activated process, even if additional GTP-releasable Ca2+ also exists. This implies a probable proximal relationship between the IP3-and GTP-activated Ca2+ release processes and permits us to consider the existence of a possible coupling event linking their modes of action.
Reversibility of GTP-activated Calcium Release-An important question to answer concerns the nature of the Ca2+ translocation process activated by GTP. Either of two distinct possibilities appeared likely. First, GTP could activate a chan-nel process to permit the flow of Ca2+ out of the organelle(s) into which Ca2+ is sequestered; alternatively, GTP could activate a fusion or communication between organelle membranes and result in the release or transfer of Ca2+. Whereas the rapidity and temperature insensitivity of IP3-activated release are suggestive of channel activation (10)(11)(12), the relative temperature sensitivity of the effect of GTP could reflect a quite different process. Perhaps more significantly, the promotion of the effects of GTP by polyethylene glycol (4,5,10) strongly suggests that membrane interactions are involved in this process, as addressed below. However, certain observations on the reversibility of the GTP-activated process militate against the view that simple membrane fusion is induced by GTP resulting in Ca2+ release from a closed organelle. Thus, we recently reported that GDP, which competitively blocks the action of GTP, could at least partially reverse the effect of GTP and allow a small reuptake of Ca2+ into the previously GTP-activated permeabilized cells (4). If membrane fusion were the mechanism for Ca2+ liberation, then even partial reversibility of the effect would not be expected. However, the GDP reversal of GTPs effect was never complete and was frequently small perhaps due to rapid nucleotide diphosphokinase-mediated GDP to GTP conversion known to occur within the permeabilized cells (4,5).

FIG. 4. Reversibility of GTP-activated Ca" release from permeabilized DDTIMF-2 cells effected by washing of GTPtreated cells in GTP-free medium.
After permeabilization, all cells were subjected to pretreatment in the presence of standard uptake medium (containing 3% polyethylene glycol, but without labeled Ca2+) for 10 min at 37 "C, during which time cells were kept in suspension by gentle stirring at a concentration of approximately 1 X lo6 cells/ml. In A and C, pretreatment did not contain GTP, whereas in B and D, pretreatment was in the presence of 10 p~ GTP.
At the end of pretreatment, cells were added either directly to uptake vials ( A and B ) , or were subjected to three washes in the same uptake medium (still containing 3% polyethylene glycol) but without any GTP (C and D). In the latter case cells were washed each time by sedimentation at 800 X g for 2 min, followed by gentle resuspension. Labeled Ca2+ uptake assays were performed under standard conditions, exactly as described under "Experimental Procedures." In each case, Ca2+ accumulated within cells was measured at the times indicated either under control conditions (O), or in the presence of 10 p~ GTP (0) or 5 p~ A23187 (A) added at the beginning of the uptake incubation. but without labeled Ca2+. If such pretreated cells, without washing, are now included in an uptake assay (that is, in the presence of label), Ca2+ uptake proceeds as shown in Fig. 4 A ; in this case the cells had not previously been treated with GTP. Thus, inclusion of GTP in the uptake assay prevents uptake into the GTP-releasable pool, just as A23187 prevents uptake into all pools (compare with Fig. 3). Unwashed GTPpretreated cells, as expected, show no uptake above that seen when GTP is included in the uptake assay, since GTP from the pretreatment has fully activated the release process (Fig.   4B). However, when GTP-pretreated cells are washed three times with GTP-free uptake medium (still containing polyethylene glycol) uptake closer to that seen with cells not pretreated with GTP is observed (Fig. 40). When these same pretreated and washed cells are included in an uptake assay containing GTP, there is a similar inhibition of uptake, indicating reactivation of the release process. As a control, washed cells not pretreated with GTP behave the same as unwashed cells when included in the uptake assay (Fig. 4C). Similar reversibility by washing has been observed even when cells were pretreated with up to 100 PM GTP. These results suggest that simple washing with GTP-free medium effectively reverses a previous GTP-mediated activation of Ca2+ release. Even though uptake after washing is not restored completely to 100% of that observed without pretreatment, the fact that GTP gives a large effect on the washed GTP-pretreated cells indicates that most of the GTPstimulated release activity has been restored. The lack of complete restoration of uptake after washing probably results from a small amount of residual GTP which dissociates perhaps more slowly from within the permeabilized cells. It would be difficult to reconcile this observed reversibility with a simple membrane fusion process activated by GTP; in other words, the effects of a direct membrane fusion event would not be washed away and result in the restoration of almost normal Ca2+ retention, as observed.

Electron Microscopic Analysis of GTPand Polyethylene
Glycol-treated Membrane Vesicles-More direct evidence to FIG. 5. Electron microscopic analysis of thin sections of microsomal membrane vesicles derived from permeabilized N1E-115 cells after treatment with polyethylene glycol and/or GTP. Microsomal membrane vesicles were isolated from permeabilized N1E-115 cells as described previously (5). Vesicles had been stored frozen at -70 "C before use, a treatment following which they still retain ATP-dependent Ca2+ pumping activity and GTP-mediated Ca2+ release properties, as described (5). Vesicles from a single aliquot were thawed, and samples of the vesicles subjected to the exact Ca2' uptake conditions described under "Experimental Procedures," with the exception that labeled Ca2+ was not included in any samples, and polyethylene glycol was not included in samples A and C (3% polyethylene glycol was present in samples B and D). In samples C and D, 10 p~ GTP was also included. After 10-min exposure to uptake conditions, vesicle samples were prepared for and examined by electron microscopy, as described under "Experimental Procedures." Bars in each micrograph represent 1 pm.
suggest that membrane fusion is not an obvious explanation for the effects of GTP is derived from electron microscopic visualization of membrane vesicles. Moreover, this approach gives an important indication about the type of process occurring when cells are treated with polyethylene glycol; thus, the studies reveal that polyethylene glycol promotes a very obvious coalescence of membranes. Electron microscopic studies were performed using isolated microsomal membrane vesicles derived from N1E-115 cells and are shown in Fig. 5. These vesicles have been used extensively in previous studies (5) and have been shown to accumulate Ca2+ via ATP-dependent Ca2+ pumping activity and to release Ca2+ by a polyethylene glycol-promoted GTP-dependent process identical to those activities observed with permeabilized cells (4,5 ) . Microsomal vesicles were treated under conditions that exactly correspond to those present during Ca2+ uptake and release experiments, either with or without GTP in the presence or absence of 3% polyethylene glycol (at this concentration polyethylene glycol maximally promotes the effects of GTP; see Ref. 10). As can be seen from Fig. 5 , A and B, polyethylene glycol induces a very obvious coalescence between the membrane vesicles. Thus, whereas in Fig. 5A it is clear that vesicles are dispersed, the polyethylene glycoltreated vesicles in Fig. 5B exist almost totally as closely associated aggregates; in fact, it was difficult to locate any unattached membrane vesicles in this sample. While vesicles appear closely associated with each other, there is no obvious evidence that they are no longer intact or that their membranes have fused. Thus, if fusion were a major event under this condition it might be expected that larger membrane sacs would appear, whereas the vesicle profiles appear either the same or perhaps slightly decreased in size. It is important to remember that, as we have previously clearly demonstrated, ATP-dependent Ca2+ uptake into polyethylene glycol-treated membrane vesicles or polyethylene glycol-treated permeabilized cells is almost completely unaltered relative to nonpolyethylene glycol-treated preparations, indicating that polyethylene glycol does not significantly alter either the intravesicular volume or the functional integrity of the membranes (4, 5, 9) despite any differences in the appearance of vesicles induced by polyethylene glycol. When vesicles are treated with GTP in the absence of polyethylene glycol (Fig. 5C), there is no significant alteration in their appearance. Furthermore, when the vesicles are treated with both polyethylene glycol and GTP under conditions that exactly correspond to those known to promote Ca2+ release, the vesicles appear identical to those treated only with polyethylene glycol (compare Fig. 5, B and D). It may thus be concluded that GTP itself does not induce any observable alteration in vesicle structure or association. However, the striking effectiveness of polyethylene glycol is good evidence to suggest that the effect of GTP in inducing Ca2+ movements is promoted by a condition that increases the close association between membranes. This important point is extended below and in the following report (12).
Oxalate Effects on GTP-mediated Calcium Movements-The evidence provided in Figs. 4 and 5 suggests that simple membrane fusion is not likely to account for the observed release of Ca2+ induced by GTP. However, in pursuit of a more definitive approach to this question, a dramatic and unexpected result was derived which, as described in the following report (14), has subsequently afforded us a much better understanding of the mechanism of action of GTP. Previous studies have shown that Ca2+ pumping into the ER within permeabilized cells is greatly enhanced when oxalate is present (6). Such a process is well recognized in many cell types and derives from the permeability of the ER membrane to anions, including oxalate and phosphate, which can form insoluble complexes with Ca2+. In the presence of such anions, precipitation of accumulated Ca2+ effectively reduces Ca2+ efflux resulting in a sustained linear rate of uptake, as was observed previously using permeabilized N1E-115 cells (6). We sought to test the releasability of Ca2+ sequestered within permeabilized cells in the presence of oxalate. If oxalateprecipitated CaZ+ could be released by GTP, it would favor the idea of some nonselective "emptying" event, as expected from a process such as membrane fusion. On the other hand, if GTP could not effect release of the Ca2+-oxalate precipitate, this would support a more selective efflux event being mediated by GTP. As shown in Fig. 6, in fact neither of these two predictions occurred. Instead, in the presence of oxalate, a rapid and profound increase in the accumulation of Ca2+ was observed upon addition of GTP in the presence of oxalate. In this experiment, permeabilized DDT,MF-2 cells were used, and the free Ca2+ concentration was increased to 30 PM to sustain the larger accumulation of Ca2+ occurring in the presence of oxalate (see Ref. 14). Under such conditions, in the absence of oxalate, 10 PM GTP induces greater than 50% release of accumulated Ca2+ (Fig. 6A), a result typical of those described in the above experiments. If 10 mM oxalate were included in the same assay, a linear rate of Ca2+ uptake would be attained, approximately equalling the initial rate of uptake observed without oxalate (not shown, but see Ref. 6). With oxalate at a lower concentration (4 mM), only a slight en- measured under the the conditions described under "Experimental Procedures" with the exception that total Ca2+ was present at 30 p~, and EGTA was not added. No oxalate was present in experiment A , whereas 4 mM K-oxalate was present throughout experiment B. Uptake was measured from zero time (that is, the time of cell addition), and aliquots were removed at the times shown. After 5 min of uptake, either 10 p~ GTP (0) or control buffer (0) was added. Aliquots of cells were rapidly filtered and washed as described under "Experimental Procedures." hancement of the final rate of Ca2+ uptake is observed (Fig.  6B). However, in the presence of 4 mM oxalate, upon the addition of 10 MM GTP, a rapid and dramatic increase in the rate of ca2+ accumulation is observed, as shown in Fig. 63. This result certainly does indicate again that simple GTPinduced fusion of membranes is an unlikely mechanism to account for the effects of GTP. Investigation of this apparently paradoxical effect of GTP in promoting sustained Ca2+ uptake and its relationship to GTP-mediated Ca2+ release forms the substance of the following report (14).
Concluding Remarks-Two major areas of investigation have been addressed in this report. The initial area concerns the cellular and subcellular specificity of the GTP-activated Ca2+ pool and its relationship to the pool of Ca2+ released by IPS. Three important conclusions are drawn from these studies. First, we have observed that the process of GTP-mediated Ca2+ release is a general phenomenon, not restricted to one particular cell type. Nor is its existence restricted to the use of particular cellular or subcellular systems, it being observable using either permeabilized cells or microsomal vesicles derived from different cell types. Second, it is clear that the effect of GTP is specific to a subcellular compartment which is not the mitochondrion. Therefore the process does not involve a nonspecific induction of Ca2+ release from any organelles that happen to have accumulated Ca". Previous studies ( 5 ) revealed that purified inverted plasma membrane vesicles which pump and accumulate Ca2+ were also unresponsive to GTP. However, as noted above, should more than one type of membrane be involved in the action of GTP, studies using purified membrane fractions may yield misleading results (this point is emphasized in the following report). Even though localization of the site of action of GTP to a particular organelle has not yet been achieved, the effects of oxalate do provide at least indirect evidence for the involvement of ER. Thus, it is known that the ER is permeable to oxalate and that when oxalate is introduced into a variety of different cell types, extensive and very visible Ca2+ oxalate precipitates can be observed within ER (25-28). The third conclusion concerns the relationship between pools of Ca2+ releasable in response to IPS and GTP. Thus, evidence supports the view that at least a function of accumulated Ca2+ resides in a pool that is releasable by either IP, or GTP. However, it seems clear that the GTP-activatable pool is larger, and hence may include an additional discrete component. This is an important inference with considerable relevance to the data and conclusions presented in the following report (14).
The second area addressed in the present report concerns the nature of the actual translocation process activated by GTP which results in the observed movements of Ca2+. We have concluded from three quite distinct approaches that release of Ca2+ is unlikely to occur by simple fusion of membrane resulting in the release of Ca2+ from a closed organelle. However, it should be pointed out that we do not as yet have definitive proof against the involvement of a fusion process. Indeed, since polyethylene glycol, which promotes the action of GTP, can itself induce membrane fusion, this type of process appeared an attractive hypothesis. However, such fusion which may involve a combination of membrane surface dehydration and bilayer disruption occurs only in the presence of polyethylene glycol concentrations of 25% or higher (29). It should also be pointed out that studies from Paiement and co-workers (30-32) have shown that GTP itself can induce what appears to be fusion between ER and nuclear membranes within cells. However, in these studies the effective GTP levels were approximately three orders of magnitude higher than those shown to be effective in inducing Ca2+ movements in the present studies. Thus, at present it seems more plausible to consider that polyethylene glycol enhances the effectiveness of GTP by promoting a close association between membranes, as has been observed in the present report. As discussed in detail in the following report (14), it is envisaged that GTP activates a conveyance of Ca2+ across and/or between different membranes. Such a model adequately explains most of the observations so far reported by ourselves and others on the movements of Ca2+ induced by GTP.