Regulation of cADP-ribose-induced Ca 2 (cid:49) Release by Mg 2 (cid:49) and Inorganic Phosphate*

cADP-ribose (cADPr) has recently been shown to re- lease Ca 2 (cid:49) from an intracellular store of permeabilized T lymphocyte cell lines (Guse, A. H., da Silva, C. P., Emmrich, F., Ashamu, G. A., Potter, B. V. L., and Mayr, G. W. (1995) J. Immunol. 155, 3353–3359). Using permeabi- lized Jurkat and HPB.ALL T lymphocytes, the effects of varying concentrations of inorganic phosphate and Mg 2 (cid:49) on cADPr-induced Ca 2 (cid:49) release were investigated. cADPr-induced Ca 2 (cid:49) release was dependent on the concentration of inorganic phosphate, showing very low Ca 2 (cid:49) release activity between 0.5 and 2 m M inorganic phosphate. At 4 to 5 m M inorganic phosphate, the cADPr- induced Ca 2 (cid:49) release was much more pronounced, reaching maximal values at 10 m M inorganic phosphate. The underlying mechanism for this stimulatory effect was an increased loading of the cADPr-sensitive Ca 2 (cid:49) store, which was demonstrated by enhanced reseques- tration of Ca 2 (cid:49) selectively into the cADPr-sensitive Ca 2 (cid:49) store. The free Mg 2 (cid:49) concentration also influenced cADPr-induced Ca 2

store of T lymphocyte cell lines (Guse et al., 1995). Moreover, Bourguignon et al. (1995) showed that a fraction of light density membrane vesicles of mouse T-lymphoma cells bound [ 3 H]ryanodine with high affinity and released Ca 2ϩ in response to cADPr. These authors also demonstrated the presence of RyRs in T-lymphoma cells by Western blot analysis. Evidence of the expression of brain type RyR in human Jurkat T cells was also presented by Hakamata et al. (1994). However, HPLC analysis of endogenous cADPr did not reveal any T cell receptor⅐CD3 complex (TCR⅐CD3)-induced elevation of this compound (Guse et al., 1995).
We report here that: (i) cADPr-induced Ca 2ϩ release in permeabilized Jurkat T lymphocytes is modulated by the concentration of P i and Mg 2ϩ ; and (ii) the stimulatory effect of high [P i ] on cADPr-induced Ca 2ϩ release was most likely due to enhanced loading of the cADPr-sensitive intracellular Ca 2ϩ store. In contrast, HPLC analysis revealed that the catabolism of cADPr at varying Mg 2ϩ and P i concentrations had only minor relevance for the modulatory effects observed. Importantly, the intracellular free Mg 2ϩ ([Mg 2ϩ ] i ) was not altered by stimulation of intact mag-Fura2-loaded Jurkat T cells, whereas stimulation of the TCR⅐CD3 resulted in a marked increase in the intracellular P i concentration.

EXPERIMENTAL PROCEDURES
Materials-cADPr was prepared enzymatically from ␤-NAD ϩ (see below). Saponin, ␤-NAD ϩ , adenosine, ADPr, malachite green, Tergitol NP-10, ATP, and phosphocreatine were obtained from Sigma. Creatine kinase was purchased from Boehringer Mannheim. Fura2/free acid and ionomycin were from Calbiochem. Mag-Fura2/AM and 4-bromo-A23187 were obtained from Molecular Probes Europe (Leiden, The Netherlands). Ammonium heptamolybdate and trisodium citrate were from Merck (Darmstadt, Germany). Buffers used in Ca 2ϩ release experiments or for HPLC were prepared using water that was first doubly distilled and then further purified using a MilliQ system (Waters-Millipore, Eschborn, Germany).
Preparation of cADPr-cADPr was synthesized from NAD ϩ in a broadly similar fashion to that reported . Thus, 1.5 mM ␤-NAD ϩ in 2.5 ml of HEPES buffer, pH 6.8, was incubated with 10 l of crude ADPr cyclase from Aplysia californica  at room temperature for approximately 10 min. This conversion was monitored by ion-exchange HPLC, and when cyclization was judged to be complete, the cADPr was purified by ion-exchange chromatography on Sepharose Q Fast Flow using a gradient of triethylammonium bicarbonate. cADPr was used as its triethylammonium salt. cADPr exhibited satisfactory 1 H and 31 P NMR spectra and HPLC data and was quantified by UV spectroscopy.
Cell Culture-Jurkat and HPB.ALL T lymphocytes were cultured as described in an earlier report (Guse and Emmrich, 1991).
Ca 2ϩ Release Experiments in Permeabilized Cells-Permeabilized cells were prepared, and the Ca 2ϩ concentration was measured as described (Guse et al., 1992). In brief, cells were permeabilized in the presence of saponin (30 g/ml) for 17.5 min (Jurkat) or 10 min (HP-B.ALL) in an intracellular buffer (20 mM HEPES, 110 mM KCl, 2 mM MgCl 2 , 5 mM KH 2 PO 4 , 10 mM NaCl, pH 7.2) at 37°C. The total concentrations of KH 2 PO 4 and MgCl 2 were different where indicated. To maintain the same osmolarity, the concentration of KCl was changed inversely; e.g. when KH 2 PO 4 was added at 10 mM (instead of 5 mM), KCl was reduced to 105 mM (instead of 110 mM). The proportion of free [Mg 2ϩ ] depends on the concentration of total (added) Mg 2ϩ and Mg 2ϩcomplexing ligands (ATP and creatine phosphate; for concentrations, see below) as well as on the pH of the solution; therefore, the free [Mg 2ϩ ] was calculated using TOT2FREE software (generously supplied by R. Thieleczek, Institut f. Physiologische Chemie Bochum, Germany), which is based on the mathematics described by Fabiato (1991), e.g. at [Mg 2ϩ ] total ϭ 2 mM, [ATP] ϭ 1 mM, and [creatine phosphate] ϭ 20 mM, pH 7.2, the free Mg 2ϩ concentration of the solution was calculated to be 1.07 mM.
An aliquot containing 3 ϫ 10 7 cells was transferred to a cuvette, and fluorescence was measured in a Hitachi F-2000 fluorometer (Colora, Lorch, Germany) with wavelength settings alternating between 340 Ϯ 2.5 and 380 Ϯ 2.5 nm (excitation ratio mode) and 510 Ϯ 5 nm (emission) at 37°C in the presence of Fura2/free acid (1.5 M) with automated gentle stirring. Reuptake of Ca 2ϩ into stores was achieved by addition of ATP (1 mM), creatine phosphate (20 mM), and creatine kinase (20 units/ml). At the end of each experiment, the free Ca 2ϩ concentration was calibrated by addition of Ca 2ϩ /ionomycin (1 mM/2 M) and subsequently by addition of EGTA/Tris (4/40 mM). The free Ca 2ϩ concentration was calculated according to the formulas given by Thomas and Delaville, 1991. HPLC Analysis of the Metabolism of Exogenous cADPr-Permeabilized T cells (3 ϫ 10 7 cells in 750 l of intracellular buffer) were incubated with cADPr (final concentration, 7 M) for 0, 1, and 3 min. cADPr and its metabolites were extracted from permeabilized cells in a fashion similar to that described recently (Guse et al., 1995). Briefly, to an aliquot of the cell suspension (250 l) 0.75 ml of ice-cold HClO 4 (3 M) was added, and the resulting suspension was vortex mixed and freeze thawed twice in liquid N 2 . After addition of 5 ml H 2 O and ultrasonication for 10 s, precipitated protein was removed at 15,000 ϫ g (10 min, 4°C). The supernatant was adjusted to pH 8.0 by addition of KOH. The KClO 4 precipitate was pelleted at 15,000 ϫ g (10 min, 4°C). The supernatant was lyophilized and stored at Ϫ70°C. The samples were reconstituted in HPLC buffer A (see below) and filtered through disposable 0.45-m filters with low dead volume directly before placing them into the HPLC autosampler. cADPr was analyzed on an automated Kontron Instruments (Neufahrn, Germany) HPLC system equipped with a strong anion-exchange HPLC column (MonoQ HR 5/5, Pharmacia Biotech Inc.) using a gradient from buffer A (1 mM Tris-HCl, pH 8.0) to 150 mM trifluoroacetic acid at a flow rate of 0.75 ml/min. The gradient was (in percent buffer B): 0 min, 0%; 5 min, 0%; 7.5 min, 15%; 8.5 min, 15%; 13 min, 50%; 15 min, 50%; 17.5 min, 0%; and 22.5 min, 0%. The wavelength of the detector was adjusted to 270 nm. Retention times for standard compounds were: adenosine, 1.82 min; NAD ϩ , 2.56 Ϯ 0.05 min (n ϭ 4); cADPr, 9.62 Ϯ 0.05 min (n ϭ 6); and ADP-ribose, 10.13 Ϯ 0.01 min (n ϭ 6).
Measurement of [Mg 2ϩ ] i -Jurkat T cells were centrifuged (Hereaus Varifuge 3.0R, 6 min, 1600 rpm, 18°C), and the cell pellet was resuspended in fresh, warm RPMI 1640 medium containing 10% newborn calf serum at a density of 10 7 /ml. The cells were then incubated for 5 min at 37°C. Subsequently, mag-Fura2/AM was added (final concentration, 4 M), and the tube was protected from light and left for 20 min at 37°C. The cell suspension was then diluted 5-fold and incubated for another 20 min at 37°C. Finally, the cells were washed and centrifuged twice (Hereaus Varifuge 3.0R, 2 min, 3900 rpm, 18°C), and the cell pellet was resuspended in a buffer containing 140 mM NaCl, 5 mM KCl, 1 mM MgSO 4 ,1 mM CaCl 2 , 20 mM HEPES, 1 mM NaH 2 PO 4 , and 5.5 mM glucose, pH 7.4 (termed "extracellular buffer") at a density of 2 ϫ 10 6 cells/ml. 750 l of the cell suspension was transferred into a cuvette containing another 750 l of the extracellular buffer. The cuvette was placed into a Hitachi F-2000 spectrofluorometer, and [Mg 2ϩ ] i was measured at room temperature at alternating excitation wavelengths of 330 Ϯ 5 and 370 Ϯ 5 nm and at the corresponding alternating emission wavelengths of 491 Ϯ 5 and 511 Ϯ 5 nm. Ratios of the fluorescence intensities of the wavelength pairs were automatically calculated by the Hitachi software. Calibration was carried out at the end of each single measurement by addition of MgCl 2 (35 mM final concentration) and the Mg 2ϩ ionophore 4-bromo-A23187 (2.7 M final concentration) to achieve the maximal ratio (R max ). The minimal ratio (R min ) was then determined by addition of Tris-EDTA (50 and 50 mM final concentrations).

ALL T cells depend on ionic conditions
Jurkat or HPB.ALL T cells were permeabilized and free [Ca 2ϩ ] was determined as outlined under "Experimental Procedures." The composition of the intracellular buffer was changed as indicated. To keep the osmolarity constant, the molarity of KC1 (normally set at 110 mM) was changed inversely. Pools were completely depleted by stepwise addition of cADPr (7 M), Ins(1,4,5)P 3 (4 M), and ionomycin (4 M). Data are presented as mean Ϯ S.D. (n ϭ 4-8).  [P i ] in the intracellular buffer was changed from the standard composition ( [P i ] ϭ 5 mM) to the values indicated. To maintain a similar osmolarity, the concentration of KCl was changed inversely. After charging of the Ca 2ϩ pools by addition of ATP (1 mM) and an ATP-regenerating system, cADPr (7 M), Ins(1,4,5)P 3 (4 M), and ionomycin (IM, 4 M) were added where indicated. A-C, single representative Ca 2ϩ tracings at the P i concentrations indicated. D, data are presented as mean Ϯ S.D. (n ϭ 4 -8). Shaded bars in D, physiological range of [P i ] as determined in intact unstimulated T cells and T cells stimulated by OKT3 (10 g/ml) for 30 min. 50-l volume, followed by 100 l of phosphate reagent. The phosphate reagent was produced as follows: 735 volume parts of a malachite green solution (0.045%, w/v) were mixed with 245 volume parts of an ammonium heptamolybdate solution (4.2%, w/v, in 4 M HCl), stirred overnight, and then mixed with 20 volume parts of Tergitol NP-10 (2%, v/v), and the resulting mixture was finally filtered. One minute after addition of the phosphate reagent to the well, 20 l of trisodium citrate solution (34%, w/v) was added. The microtiter plates were incubated for 30 min at room temperature, and the absorption at 585 nm was then measured in a V max microtiter reader (Molecular Devices Corp, Menlo Park, CA). Linear calibration curves for P i were obtained in the range of 0.25-4 nmol (r Ͼ 0.99). The intracellular concentration of P i was calculated assuming an intracellular volume of 198 nl/10 5 cells (Guse et al., 1993).

Effect of Varying [P i ] and [Mg 2ϩ ] on the Total Intracellular
Ca 2ϩ Pool Content in Permeabilized T Cells-During the first experiments using an altered composition of the intracellular buffer, it became clear that changing the parameters [P i ] and [Mg 2ϩ ] free would also influence the Ca 2ϩ pool size of the permeabilized cell preparation (Table I). The size of the total intracellular Ca 2ϩ pool (i.e. Ca 2ϩ that was releasable by 4 M ionomycin) under standard conditions ([P i ], 5 mM; [Mg 2ϩ ] free , 1.07 mM) was about 400 -450 nM/3 ϫ 10 7 cells. When lowering [P i ] from 5 to 0.5 mM, the absolute pool size decreased by a factor of about 2 (Table I). Increasing [P i ] to 10 mM did not change the pool size significantly (Table I). Changes in [Mg 2ϩ ] free , either as decreases from 1.07 down to 0 mM as well as increases up to 8.58 mM, reduced the absolute pool size by a factor of about 4 (Table I). For [P i ] these effects may be explained as follows; the lower concentration of P i , as one product of ATP hydrolysis, may have facilitated ATPase-dependent hydrolysis of ATP, thereby leaving less ATP for loading of Ca 2ϩ pools. On the other hand, Mg 2ϩ is needed for formation of a Mg⅐ATP complex by most enzymes that use ATP as a (co)substrate. Therefore, low [Mg 2ϩ ] may have left too much ATP useless for the Ca 2ϩ ATPases of the endoplasmic reticulum, since the ATP could not form sufficient Mg⅐ATP complexes. Similar results for the effects of [P i ] and [Mg 2ϩ ] free were obtained for another T lymphocyte cell line, namely HPB.ALL cells (Table I). However, under standard conditions (5 mM [P i ] and 1.07 mM [Mg 2ϩ ] free ) the absolute Ca 2ϩ pool size calculated for 3 ϫ 10 7 cells was about 2-fold larger for HPB.ALL cells compared with Jurkat cells (Table I).
To demonstrate the effects of [P i ] and [Mg 2ϩ ] free on cADPrand Ins(1,4,5)P 3 -induced Ca 2ϩ release, we present the data normalized as a percentage of the total Ca 2ϩ pool content. [P i ] and [Mg 2ϩ ] free on cADPr-and Ins(1,4,5)P 3 -induced Ca 2ϩ Release-The magnitude of cADPrinduced Ca 2ϩ release was dependent on [P i ], showing only small release activity below 2 mM (about 10% of the total Ca 2ϩ pool content; Fig. 1A), whereas at [P i ] of 4-10 mM considerably higher Ca 2ϩ release was observed (about 25-38% of the total Ca 2ϩ pool content; Fig. 1). A slightly negative relationship was observed for Ins(1,4,5)P 3 -induced Ca 2ϩ release, showing highest release activity at low [P i ] and reduced release activity above 4 mM [P i ] (Fig. 1). In HPB.ALL cells a comparable dependence of cADPr-and Ins(1,4,5)P 3 -induced Ca 2ϩ release was observed (Table II). However, high [P i ] (10 mM) did not result in a further enhancement of cADPr-induced Ca 2ϩ release in these cells (Table II).

Effect of Varying
High [P i ] did not enhance the sensitivity of the Ca 2ϩ release system to cADPr. Dose-response curves for cADPr recorded at different [P i ] did not result in a leftward shift of the curves at high [P i ]. Instead, remarkable differences in the amplitude of the Ca 2ϩ release were observed ( Fig. 2A).
When increasing [P i ], the absolute Ca 2ϩ release by cADPr paralleled the increase of the total Ca 2ϩ pool content, whereas for Ins(1,4,5)P 3 a saturation was observed at 5 mM P i (Fig. 2B). In other words, high [P i ] selectively increased the pool size of the cADPr-sensitive Ca 2ϩ pool. The underlying mechanism is demonstrated in Fig. 2C. The rate of Ca 2ϩ resequestration after addition of cADPr was higher compared with Ins(1,4,5)P 3 and, most importantly, increased at higher [P i ]. In contrast, the rate of Ca 2ϩ resequestration after addition of Ins(1,4,5)P 3 was generally lower compared with cADPr and did not increase significantly at higher [P i ] (Figs. 1, A-C, and 2C). It should be noted that the absolute Ca 2ϩ concentrations observed after addition of cADPr at high versus low [P i ] (e.g. 348 nM [Ca 2ϩ ] at 10 mM P i versus 180 nM [Ca 2ϩ ] at 0.5 mM P i ; see Fig. 1, A and C) only had a minor influence on Ca 2ϩ reuptake by the cADPr-sensitive stores (data not shown).
To determine whether the observed effects were specific for P i , either oxalate or AMP were used in substitution for P i . Fig.  2B shows that using 0.5 mM P i plus 4.5 or 9.5 mM oxalate or AMP resulted in reduced uptake of Ca 2ϩ into the Ca 2ϩ stores. When oxalate was used in substitution for P i , the magnitudes of cADPr-and Ins(1,4,5)P 3 -induced Ca 2ϩ release were lower than with 5 mM P i and, in the case of Ins(1,4,5)P 3 , even lower than with 0.5 mM P i (Fig. 2B). When AMP was used, neither cADPr nor Ins(1,4,5)P 3 could release any Ca 2ϩ from the permeabilized cell suspension (Fig. 2B). Under such conditions, Ca 2ϩ could only be released by the ionophore ionomycin, indicating that either the second messenger-sensitive Ca 2ϩ pools were not charged or, less likely, high concentrations of both oxalate and AMP somehow inhibited second messenger-mediated Ca 2ϩ release. Likewise, Ca 2ϩ reuptake was lower or not detectable when oxalate or AMP was used in substitution for P i (Fig. 2C).
Low and very high [Mg 2ϩ ] free also resulted in reduced cADPr-induced Ca 2ϩ release (Fig. 3), whereas Ins(1,4,5)P 3induced Ca 2ϩ release was not significantly influenced by different [Mg 2ϩ ] free (Fig. 3). Similar results for [Mg 2ϩ ] free were obtained using HPB.ALL cells (Table II), although as for [P i ], the effects of low and high [Mg 2ϩ ] free were not as pronounced as in Jurkat cells (Table II).
HPLC Analysis of cADPr Metabolism in Permeabilized T Cells-The effects of varying [P i ] and [Mg 2ϩ ] free on cADPrinduced Ca 2ϩ release could be explained by: (i) a direct effect of the different ionic conditions on the putative cADPr-responsive Ca 2ϩ release system, including the Ca 2ϩ pool size; (ii) modulation of the proportion of free cADPr, e.g. cADPr that is not bound unspecifically to cellular proteins; or (iii) the differences  cADPr and Ins(1,4,5

)P 3 in permeabilized Jurkat and HPB.ALL T cells depends on the ionic conditions
Jurkat or HPB.ALL T cells were permeabilized and free [Ca 2ϩ ] was determined as outlined under "Experimental Procedures." The composition of the intracellular buffer was changed as indicated. To keep the osmolarity constant, the molarity of KC1 (normally set at 110 mM) was changed inversely. cADPr (7 M) and Ins(1,4,5)P 3 (4 M) were added to monitor Ca 2ϩ release in response to these compounds. The remaining Ca 2ϩ pools were then completely depleted by a 2-fold addition of 4 M ionomycin. Data are presented as percentage of the whole pool content (i.e. the sum of cADPr-, Ins(1,4,5)P 3 , and ionomycin-releasable Ca 2ϩ ) and are given as mean Ϯ S.D. (n ϭ 4-8). in catabolism of cADPr by endogenous cADPr hydrolases under different ionic conditions. To investigate the latter, we incubated permeabilized cells under such different ionic conditions with cADPr and analyzed the metabolism of this exogenously added cADPr by HPLC. The incubation conditions were identical to the experiments carried out to study Ca 2ϩ release fluorometrically, except that no Fura2/free acid, ATP, creatine phosphate, and creatine kinase were added. Indeed, increasing [P i ] resulted in a reduced catabolism of cADPr compared with low [P i ]. However, since the effects of cADPr on Ca 2ϩ release are very rapid (see Fig. 2), and since even at low [P i ] after 1 min about 85% of the initial cADPr (about 5.95 M) was still present, it is not very likely that the effects of [P i ] on cADPr-induced Ca 2ϩ release were due to the enhanced catabolism at low [P i ].
As demonstrated for low [P i [P i ] i in Intact T Cells-At least two mechanisms for the regulation of cADPr-induced Ca 2ϩ release are possible: (i) the concentration of cADPr is regulated by stimulation of receptors in the plasma membrane, as is known for Ins(1,4,5)P 3 , e.g. by the TCR⅐CD3 complex; or (ii) receptor-dependent changes of other compounds may regulate cADPr-induced Ca 2ϩ release. Since we observed such dependencies on both [P i ] and [Mg 2ϩ ], we measured the intracellular concentrations of these ions under resting and activated conditions. The [Mg 2ϩ ] i was measured using a fluorescent indicator for FIG. 2. Characteristics of the regulation of cADPr-induced Ca 2؉ release by [P i ]. Jurkat T lymphocytes were permeabilized and [Ca 2ϩ ] was measured as described in Fig. 1. A, dose-response curves for cADPr were recorded at the concentrations of P i as indicated. Data are presented as mean Ϯ S.D. (n ϭ 3-6). B, the composition of the intracellular buffer was altered as indicated below the graph. Bars, Ca 2ϩ as released by cADPr (7 M) plus Ins(1,4,5)P 3 (4 M) plus ionomycin (4 M), cADPr alone (7 M), or Ins(1,4,5)P 3 alone (4 M). Data are presented as mean Ϯ S.D. (n ϭ 3-8). C, the composition of the intracellular buffer was altered as indicated below the graph. Bars, Ca 2ϩ reuptake within the first min after the Ca 2ϩ peak in response to either cADPr (7 M) or Ins(1,4,5)P 3 (4 M). Data are presented as mean Ϯ S.D. (n ϭ 3-8). Inset, Ca 2ϩ reuptake phase as 340:380 ratio tracings under the following conditions (variable intracellular buffer components): 1, 0.5 mM P i , 9.5 mM AMP; 2, 0.5 mM P i ; 3, 5 mM P i ; 4, 10 mM P i . Mg 2ϩ , mag-Fura2. In resting, mag-Fura2-loaded Jurkat T cells, [Mg 2ϩ ] i amounted to 0.941 Ϯ 0.348 mM (mean Ϯ S.D., n ϭ 12) as indicated by a shaded bar in Fig. 3D. Stimulation of the TCR⅐CD3 complex using the anti-CD3 monoclonal antibody OKT3 did not result in a significant change of [Mg 2ϩ ] i (data not shown). Interestingly, other unphysiological activation procedures, such as addition of thapsigargin or cytosolic alkalinization using 4-aminopyridine (Guse et al., 1994), resulted in a more pronounced elevation of [Mg 2ϩ ] i (data not shown).
The intracellular [P i ] was determined by a colorimetric assay for P i in neutralized perchloric acid-quenched cell samples. As indicated in Fig. 4, there was a rapid increase in [P i ] in cells stimulated by OKT3, which remained elevated for at least 30 min. The [P i ] in quiescent cells was about 2.8 mM, whereas on addition of OKT3, values up to about 4.5 mM were observed.

DISCUSSION
In this report we demonstrate that: (i) the Ca 2ϩ pool content of permeabilized T lymphocytes was significantly altered by changing the standard composition of the intracellular buffer used; (ii) [P i ] and [Mg 2ϩ ] both can modulate cADPr-induced Ca 2ϩ release in permeabilized T lymphocyte cell lines; (iii) the stimulatory effect of increasing [P i ] is likely due to an increased loading of the cADPr-sensitive intracellular Ca 2ϩ pool; and (iv) [P i ] i increased on stimulation of TCR⅐CD3, whereas [Mg 2ϩ ] i did not.
The dependence of cADPr-induced Ca 2ϩ release on [P i ] has not been investigated so far. Not only did [P i ] influence cADPrinduced Ca 2ϩ release, it also had a pronounced effect on Ca 2ϩ pool loading in permeabilized T cell lines. Therefore, four dif- ferent mechanisms for the effect of [P i ] are possible: (i) the catabolism of cADPr was reduced at high [P i ]; (ii) the loading of cADPr-sensitive Ca 2ϩ stores was selectively increased compared with Ins(1,4,5)P 3 -sensitive pools at high [P i ]; (iii) elevated levels of [P i ] may have prevented binding of cADPr to proteins other than the cADPr-dependent Ca 2ϩ release system and thus may have increased the proportion of free versus unspecifically bound cADPr; and (iv) low [P i ] inhibited cADPrinduced Ca 2ϩ release selectively by disturbing the interaction with the Ca 2ϩ release system, whereas the charging of intracellular Ca 2ϩ pools in general was reduced. As shown by our HPLC data, a major effect of enhanced catabolism of cADPr at low [P i ] can be excluded. Second, we demonstrated that charging of both the Ins(1,4,5)P 3 -and cADPr-sensitive Ca 2ϩ pools was reduced at low [P i ] and increased at higher [P i ]. However, increased loading of the cADPr-sensitive pool was much more pronounced; e.g. a 10-fold increase in the cADPr-sensitive Ca 2ϩ pool but only a 2-fold increase in the Ins(1,4,5)P 3 -sensitive Ca 2ϩ pool at 0.5 versus 5 mM P i and another 2-fold increase in the cADPr-releasable Ca 2ϩ between 5 and 10 mM were found. In contrast, the Ins(1,4,5)P 3 -sensitive Ca 2ϩ pool was not further loaded when [P i ] was increased to 10 mM. Most importantly, the Ca 2ϩ resequestration after addition of cADPr increased at high [P i ], whereas after addition of Ins(1,4,5)P 3 such an effect was not observed, supporting the view that enhanced loading exclusively of the cADPr-sensitive intracellular Ca 2ϩ store was the major mechanism underlying the stimulatory effect of high [P i ]. When oxalate or AMP was used in substitution for P i , the loading of the intracellular Ca 2ϩ pools was not enhanced but reduced. Under such conditions the loading was so low that Ca 2ϩ release effects of cADPr or Ins(1,4,5)P 3 were very small or not detectable at all. This finding was accompanied by decreased Ca 2ϩ reuptake into the intracellular Ca 2ϩ stores. According to several years of experience with permeabilized T cells, these data confirm our view that most of the alterations in the composition of the intracellular buffer lead to reduced charging and thereby to reduced Ca 2ϩ release by Ca 2ϩmobilizing second messengers. Thus, there is some specificity of high [P i ] to support charging of the Ca 2ϩ pools, especially the cADPr-sensitive Ca 2ϩ pool.
Therefore, the major mechanism by which low [P i ] downmodulated and high [P i ] up-regulated the activity of cADPr is the charging of the cADPr-sensitive Ca 2ϩ store. However, the other possibilities, increased concentration of free cADPr at high [P i ] and inhibition of the interaction of cADPr with its specific binding protein(s) at low [P i ], may have contributed to the effects observed.
Most importantly, when analyzing the [P i ] i in intact Jurkat T cells, a TCR⅐CD3-mediated increase was observed. In unstimulated T cells [P i ] i was about 2.8 mM, in other words, in a concentration range in which the Ca 2ϩ release by cADPr was still small in permeabilized cells (Fig. 1D). Therefore, the TCR⅐CD3-mediated increase of [P i ] i to about 4.5 mM, at which cADPr released considerably more Ca 2ϩ in permeabilized cells (Fig. 1D), may be a switch to turn on cADPr-induced Ca 2ϩ release inside the intact T cell. A prerequisite for such a model would be a constantly high intracellular concentration of cADPr. Although an exact quantification of endogenous cADPr is very difficult due to its chemical lability in extraction media (e.g. 3 M HClO 4 ) and the fact that various endogenous compounds show very similar behavior in anion-exchange as well as ion pair reversed phase HPLC, 2 such a high endogenous concentration has been measured in Jurkat cells using two different HPLC systems, amounting to about 16 M (Guse et al., 1995), assuming the intracellular volume to be 198 l/10 8 cells (Guse et al., 1993). However, it is not clear: (i) which proportion of this total amount of cADPr is freely available in the cytosol, in other words, how much cADPr is bound unspecifically to proteins or structures other than the cADPr-dependent Ca 2ϩ release system; and (ii) whether the observed increase of [P i ] i could have significantly increased the amount of free cADPr.
A modulatory effect of [Mg 2ϩ ] on cADPr-induced Ca 2ϩ release has already been described for sea urchin egg homogenates (Graeff et al., 1995). Similar to our results, Graeff et al. (1995) described an inhibitory effect of high [Mg 2ϩ ] (Ͼ6 mM) on cADPr-but not on Ins(1,4,5)P 3 -induced Ca 2ϩ release. In sea urchin egg homogenates 1 mM Mg 2ϩ was the optimal concentration for cADPr-induced Ca 2ϩ release (Graeff et al., 1995); this result and the fact that higher [Mg 2ϩ ] acted in an inhibitory way was confirmed in our permeabilized Jurkat T cells. Since [Mg 2ϩ ] at Ͼ5 mM is a known blocker of the RyR (Meissner, 1994) our data from T lymphocytes add another piece of evidence that RyR is one of the central players in cADPrinduced Ca 2ϩ release. However, in contrast to [P i ] i , which increased in response to stimulation of TCR⅐CD3, no changes in [Mg 2ϩ ] i were observed, indicating that Mg 2ϩ -dependent modulation of cADPr-induced Ca 2ϩ release does not play a major role, if any, in vivo.
In conclusion, our results indicate that cADPr-induced Ca 2ϩ release may be regulated by [P i ] i in Jurkat T lymphocytes in vivo, since: (i) the filling state of the cADPr-sensitive Ca 2ϩ store and, thereby, cADPr-induced Ca 2ϩ release in permeabilized T cells was modulated by [P i ]; and (ii) [P i ] i increased in intact Jurkat T cells stimulated via the TCR⅐CD3 complex. Such a model would also explain our recent finding of high endogenous concentrations of cADPr, which remained unchanged during stimulation of the TCR⅐CD3 complex (Guse et al., 1995).