Mitochondrial Dysfunction Induced by a Cytotoxic Adenine Dinucleotide Produced by ADP-ribosyl Cyclases from cADPR*

ADP-ribosyl cyclases were previously shown to produce three new adenine dinucleotides, P1,P2 diadenosine 5′-diphosphate (Ap2A) and two isomers thereof (P18 and P24), from cyclic ADP-ribose (cADPR) and adenine (Basile, G., Taglialatela-Scafati, O., Damonte, G., Armirotti, A., Bruzzone, S., Guida, L., Franco, L., Usai, C., Fattorusso, E., De Flora, A., and Zocchi, E. (2005) Proc. Natl. Acad. Sci. U. S. A. 102, 14509-14514). The Ap2A isomer P24, containing an unusual C1′-N3 N-glycosidic bond, is shown here to affect mitochondrial function through (i) opening of the permeability transition pore complex (and consequent proton gradient dissipation) and (ii) inhibition of Complex I of the respiratory chain. Whereas proton gradient dissipation is dependent upon the extracellular Ca2+ influx triggered by P24, the effect on oxygen consumption is Ca2+ independent. The proton gradient dissipation induces apoptosis in HeLa cells and thus appears to be responsible for the already described potent cytotoxic effect of P24 on several human cell types. The other products of ADP-ribosyl cyclase activity, Ap2A and cADPR, antagonize P24-induced proton gradient dissipation and cytotoxicity, suggesting that the relative concentration of P24, cADPR, and Ap2A in cyclase-positive cells may affect the balance between cell life and death.

ADP-ribosyl cyclases (ADPRCs) 2 are a family of multifunctional enzymes, present from protists to mammals and higher plants, that generate a number of products affecting the intracellular free calcium concentration ([Ca 2ϩ ] i ), arguably the most influential regulatory signal in cell physiology (1,2). Consequently, ADPRCs are involved in the regulation of several Ca 2ϩ -controlled cell functions of increasing complexity from protists (3) to lower Metazoa, mammals (4), and plants (5).
The Ca 2ϩ -active products of ADPRCs known so far include cyclic ADP-ribose (cADPR), ADP-ribose (ADPR), nicotinic acid adenine dinucleotide phosphate (NAADP ϩ ), and the ADP-ribose dimer (ADPR 2 ). cADPR is produced from NAD ϩ by means of the reaction typical of the ADPRC family that removes the nicotinamide moiety and creates an unusual C1Ј-N1 bond that cyclizes the ADPR molecule (4,6). cADPR binds to specific receptor channels on the endoplasmic reticulum, the ryanodine receptors, releasing Ca 2ϩ (4,7,8). All ADPRCs also display to a variable degree a cADPR-degrading activity that introduces water into the C1Ј-N1 bond, generating ADPR (4,6,7). ADPR has been recognized as an intracellular agonist gating some members of the transient receptor potential (TRP) family of non-selective, Ca 2ϩ -permeable plasma membrane channels, notably TRPM2, with consequent influx of extracellular Ca 2ϩ (9 -12). NAADP ϩ is produced by all ADPRCs from NADP ϩ and nicotinic acid (13). NAADP ϩ binds to specific Ca 2ϩ channels tentatively identified on lysosomal membranes in selected cell types (14). Finally, ADPR 2 , which is produced by ADPRC from NAD ϩ in the presence of ADPR, has been reported to synergize with cADPR in releasing Ca 2ϩ from sea urchin egg microsomes (15).
Recently, ADPRCs from Porifera, molluscs, and mammals have been shown to catalyze an additional reaction on cADPR, introducing adenine into the C1Ј-N1 bond of the cyclic nucleotide and generating three new adenine homodinucleotides, called P18, P24, and P31 from their high pressure liquid chromatography retention times (16). P31 proved to be P1,P2 diadenosine 5Ј-diphosphate (Ap2A). Though presence of Ap2A had been already reported in ADPRC ϩ platelets and cardiac myocytes (17,18), the enzyme responsible for its synthesis was unknown. P18 and P24 are two isomers of Ap2A, each containing an unusual N-glycosidic bond between the newly introduced adenine and one ribose, C1Ј-NI in P18 and C1Ј-N3 in P24 (16). They are the first dinucleotides featuring a non-canonical N-glycosidic bond demonstrated in (ADPRC ϩ ) animal cells (16,19).
Through different mechanisms, Ap2A and its isomers all affect the [Ca 2ϩ ] i when applied extracellularly to intact cells at micromolar (1 M) concentrations; P18 induces a decrease of the [Ca 2ϩ ] i , P24 conversely induces an increase of the [Ca 2ϩ ] i through extracellular calcium influx, and Ap2A synergizes with cADPR in releasing Ca 2ϩ from ryanodine receptor (16). Not unexpectedly, all three dinucleotides affect cell proliferation (a typically Ca 2ϩ -controlled cell function) in human cell lines and in hemopoietic progenitors. Specifically, P18 and P24 both show a potent cytotoxic effect on cell lines and an even more severe growth inhibitory effect on hemopoietic progenitor colony growth (with IC 50 values of 1.0 and 0.18 M, respectively). Ap2A conversely stimulates hemopoietic progenitor colony growth and synergizes with cADPR, recently demonstrated to behave as a hemopoietic growth factor (20), at suboptimal concentrations of both nucleotides (16).
These results prompted us to investigate the mechanisms underlying the cytotoxic effect of P18 and of P24, also in view of their potential use as novel antileukemic pharmaceuticals in clinical hematology. Results obtained demonstrate that mitochondria are a major target of P24 toxicity, with micromolar concentrations of the dinucleotide inducing (i) dissipation of the mitochondrial proton gradient (⌬⌿ m ) in intact cells through opening of the permeability transition pore (PTP) and (ii) inhibition of the respiratory chain in isolated mitochondria, acting on Complex I. Conversely, P18 and Ap2A do not show significant effects on any of these functions, indicating the absolute requirement for a C1Ј-N3 bond in the adenine dinucleotide to exert its mitochondrial effects. Whereas dissipation of ⌬⌿ m is dependent upon the [Ca 2ϩ ] i increase, inhibition of Complex I is Ca 2ϩ independent.
These results identify P24 as a novel endogenous regulator of mitochondrial function in mammalian cells, thus extending the physiological functions of the family of ADPRC products to include control of cell respiration. Interestingly, cADPR and Ap2A partially antagonize P24-induced cytotoxicity and proton gradient dissipation, suggesting that the relative intracellular concentrations of these ADPRC products may affect cell life and death.
Microscopical Evaluation of Necrosis and Apoptosis-HeLa cells (obtained from ATCC, Manassas, VA) were grown in Dulbecco's modified medium, supplemented with 2 mM glutamine and 10% fetal calf serum plus 50 units/ml penicillin and 50 g/ml streptomycin in a humidified, 5% CO 2 atmosphere at 37°C. Cells (2 ϫ 10 4 /well) were seeded into 24-well plates and exposed to either P18 or P24 at the concentrations and for the times indicated in the legend to Fig. 1. Cells were then incubated with 10 M Hoechst 33258 and 1 M PI for 5 min to estimate the percentage of apoptotic and necrotic cells, respectively, as described in Ref. 22. After washing with Hank's balanced salt solution (HBSS), stained cells were identified under a fluorescence microscope using excitation/emission wavelengths of 340/440 Ϯ 25 and 568/585 Ϯ 25 nm for Hoechst 33258 and PI, respectively. Three randomly selected fields were acquired from each treatment. The corresponding bright field images were also acquired, and the three channels were overlaid using the appropriate function of the Metamorph software. At least 100 cells were counted for each treatment, and the percentage of apoptotic or necrotic cells was determined.
Swelling and Oxygen Consumption of Isolated Mitochondria-Rat liver mitochondria were isolated from albino Wistar rats by standard centrifugation techniques as described (23). Mitochondria (0.5 mg of protein/ml) were suspended in standard medium containing 250 mM sucrose, 1 mM KH 2 PO 4 , 20 M EGTA, 10 mM MOPS, pH 7.4, and 5/2.5 mM glutamate/malate or 5 mM succinate as substrates. Mitochondrial swelling was followed as the change of 90°light scattering of the mitochondrial suspension at 540 nm in a PerkinElmer 650 -40 fluorescence spectrophotometer (24). Oxygen consumption was determined polarographically using a Clark oxygen electrode (24). All assays were performed at 25°C on instruments equipped with thermostatic control and magnetic stirring.
Mitochondrial Membrane Potential in Intact Cells-For measurements of mitochondrial membrane potential, cells were washed and incubated in HBSS and 10 mM HEPES in the presence of 10 nM TMRM and 1.6 M cyclosporin H at 37°C for the times indicated in Fig. 2. Treatment of control cells with cyclosporin H is necessary because the extent of cell, and hence of mitochondrial, loading with potentiometric probes is affected by the activity of the plasma membrane multidrug resistance P-glycoprotein, which is inhibited by cyclosporin H (25). Cell fluorescence images were acquired on a Leica TCS SL confocal microscope, equipped with a HCX PL APO CS 63.0 ϫ 1.40 oil objective, at 1-min intervals.
Cell Calcium Measurements-For cytosolic Ca 2ϩ measurements, adherent HeLa cells (2 ϫ 10 4 /well in 96-well plates) were incubated in complete medium with Fluo 3-AM (10 M) for 45 min at 37°C. Cells were then washed twice in HBSS, and cell fluorescence was monitored at 1-min intervals on a Fluostar Optima microplate reader (BMG Labtechnologies, Offenburg, Germany) at ex 485 nm and em 520 nm after addition of the various dinucleotides as indicated in the legend to Fig. 5A. For mitochondrial Ca 2ϩ measurements, adherent HeLa cells in complete medium (10 5 /well, in LabTek chambers; Nalge Nunc Int. Corp., Naperville, IL) were co-incubated with the fluorescent Ca 2ϩ probe Rho 2-AM (4.5 M) and the mitochondrial tracer Mitofluor green (20 nM) for 30 min at 22°C. Cells were washed twice in HBSS, and images of the regions of probe colocalization (4 -5/cell, Ն10 cells/field) were acquired at 1-min intervals on a Leica TCS SL confocal microscope.
Cytotoxicity Assays-HeLa cells were seeded in triplicate (10 4 /well) in 96-well plates and exposed to the various treatments described in the legends to Figs. 4 and 6. After 24 h of culture, cells were washed twice in HBSS and incubated for 20 min in the presence of calcein green (2.5 mg/ml). A standard curve of cells, freshly seeded at increasing density (0.5-2.0 ϫ 10 4 /well), was prepared and stained in parallel. Cells were washed three times in HBSS, and the fluorescence was measured with a FluoStar Optima microplate reader at ex 485 and em 520 nm. The number of viable cells was estimated by comparison of sample fluorescence with the linear standard curve.

P24 Reduces Cell Viability by Inducing Apoptosis-Previous
work had shown that both P18 and P24 are cytotoxic on a number of cell lines, including HeLa cells (16). Indeed, HeLa cell viabilitywasseverelyreducedbyP24inatime-andconcentrationdependent manner; the IC 50 of P24 was 10 M after 24 h of incubation, similar to that already reported (16). At the same concentrations and for the same duration of exposure, P18 was ineffective, in agreement with the reported IC 50 value of P18 in HeLa cells, which is 1 log higher than that of P24 (16). The occurrence of necrosis and/or apoptosis was investigated by assessing the staining of nuclei with PI and Hoechst 33258, respectively. Fig. 1A shows representative images of HeLa cells incubated in the absence (control) or in the presence of 10 M P24 for 24 h. The P24-treated cultures showed a marked decrease of the cell number and an increase of the proportion of blue, i.e. apoptotic, cells. The percentage of necrotic and apoptotic cells in P24-and P18-treated cultures was determined, and results are shown in Fig. 1, B and C, respectively. An incubation time shorter than 24 h was chosen to avoid loss of adherent cells due to cell death. The decrease of cell viability in P24-incubated cultures occurred mostly because of apoptosis, although at higher P24 concentrations a significant proportion of cell death was due to necrosis. Conversely, at the same concentrations, P18 did not significantly modify the percentage of necrotic and apoptotic cells as compared with untreated controls.
P24 Dissipates the Mitochondrial Proton Gradient in Intact Cells-Because mitochondrial alterations are known to play a major role in determining cell death, we investigated the effect of P24 on mitochondrial functions to elucidate the mechanisms underlying its cytotoxicity. Thus, intact cells were incubated in the presence of P24 and ⌬⌿ m was measured with the fluorescent probe TMRM. As shown in Fig. 2, A (blue trace) and C, P24 induced a progressive decrease in TMRM fluorescence, reflecting a fall of ⌬⌿ m . The slight decrease in fluorescence obtained upon addition of the uncoupler FCCP indicated that ⌬⌿ m was already almost completely abolished after incubation of the cells for 30 min with 20 M P24. The effect of P24 on ⌬⌿ m was concentration dependent, with as low as 50 nM P24 dissipating ϳ30% of the TMRM fluorescence after 2 h of incubation of the cells (not shown).
Under physiological conditions, ⌬⌿ m is maintained by the proton pumping activity of the respiratory chain complexes. However, when the electron flow is hampered, ⌬⌿ m maintenance is allowed by the proton pumping activity of the F 0 F 1 -ATPase, powered by ATP hydrolysis. The contribution of respiratory chain or ATP hydrolysis to ⌬⌿ m maintenance can be assessed by using selective inhibitors. Addition of oligomycin, an inhibitor of F 0 F 1 -ATPase, did not affect ⌬⌿ m (Fig. 2, A (black trace) and C), indicating that in the absence of any impairment of the respiratory chain ATP hydrolysis does not contribute to the generation of mitochondrial membrane potential, as expected. Conversely, the addition of oligomycin to P24-treated cells caused an abrupt fall of ⌬⌿ m (Fig. 2, A (red trace) and C). On the one hand, this result suggests an inhibi-  FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7 tory effect of P24 on the respiratory chain. On the other hand, the effect elicited by oligomycin rules out the possibility that adenylate translocase or F 0 F 1 -ATPase are also severely affected by P24, as oligomycin would have been without further effect if ATP uptake or hydrolysis were already inhibited. The addition of rotenone alone, an inhibitor of Complex I, resulted in only a slight decrease of the ⌬⌿ m (Fig. 2, A (purple trace) and C) due to the progressive consumption of ATP for proton pumping by the F 0 F 1 -ATPase. The addition of rotenone to P24-treated cells, however, induced a faster and greater decrease of ⌬⌿ m (Fig. 2, A (green trace) and C) than either that of P24 (blue trace) or of rotenone (purple trace) alone, suggesting that a mechanism other than respiratory chain inhibition is primarily responsible for ⌬⌿ m dissipation triggered by P24.

Mitochondrial Effects of P24
P24 has been shown to induce an increase of the [Ca 2ϩ ] i through extracellular Ca 2ϩ influx (16). Preincubation of the cells for 1 h with the intracellular Ca 2ϩ chelator EGTA-AM (0.5 mM) prior to exposure to P24 prevented ⌬⌿ m dissipation (Fig. 2, B (red trace) and C), as did presence of EDTA (not shown), demonstrating a causal role of the [Ca 2ϩ ] i increase in mediating the effect of P24 on ⌬⌿ m . P18, cADPR, and Ap2A, at concentrations ranging between 1.0 and 10 M, did not significantly affect ⌬⌿ m (not shown).
P24 Affects Oxygen Consumption of Isolated Rat Liver Mitochondria-To verify an inhibitory effect of P24 on the respiratory chain and to identify the site of inhibition, experiments were carried out on isolated rat liver mitochondria. Oxygen consumption was measured in the presence of CsA to avoid possible interference of the mitochondrial PTP. In fact, PTP opening can affect oxygen consumption due to its uncouplinglike effect and to the release of pyridine nucleotides (26,27).
In the presence of the NAD ϩ -dependent substrates glutamate/ malate, the increase over basal val- ues of the oxygen consumption stimulated by the uncoupler FCCP was reduced by 86% after preincubation of mitochondria with 10 M P24 for 10 min (Table 1). Conversely, P24 did not affect the utilization of succinate in the presence of rotenone (Table 1). Therefore, P24 inhibits electron transfer at the level of Complex I, but not of Complex II. P24 also inhibited ADPstimulated respiration (State 3) in the presence of the Complex I substrates glutamate/malate (Table 1) nearly as efficiently as it inhibited the uncoupler-induced respiration (88 versus 86%). The fact that P24 failed to inhibit the ADP-stimulated oxygen consumption in the presence of the Complex II substrate succinate unequivocally demonstrates that P24 does not affect F 0 F 1 -ATPase or adenine nucleotide translocase as their inhibition would have affected also the activity of Complex II.
Basal oxygen consumption (State 4) was significantly higher in P24-treated mitochondria, compared with controls, in the presence of substrates of either Complex I (20% increase) or Complex II (37% increase) (Table 1). Thus, P24 behaves as a mild uncoupler, in line with the decrease of the ⌬⌿ m described in the previous paragraph. The effect of P24 on uncoupled respiration displayed a dose dependence, because oxygen consumption was inhibited by 20 and 40% in the presence of 0.1 and 1.0 M P24, respectively. Similar results as those described above were obtained in the presence of EGTA, indicating that the effects of P24 on respiration were not Ca 2ϩ -dependent. P18, cADPR, and Ap2A, at concentrations ranging between 1.0 and 10 M, did not significantly affect oxygen consumption (not shown).
P24 Induces Mitochondrial Swelling by Opening the Permeability Transition Pore-The fact that P24 and rotenone together induced a larger dissipation of ⌬⌿ m than that caused by either compound alone (Fig. 2, A and C) suggested presence of an additional mechanism of ⌬⌿ m dissipation triggered by P24, besides inhibition of the respiratory chain. Because neither F 0 F 1 -ATPase nor adenylate translocase was inhibited by P24, we investigated whether P24 induced the opening of the mitochondrial PTP, which could account for the observed decrease of ⌬⌿ m (27). Swelling of the mitochondrial matrix was monitored as decrease in absorbance at 540 nm, reflecting a decrease in light scattering of the mitochondrial suspension. Intact mitochondria were suspended in the presence of the highest Ca 2ϩ concentration insufficient per se to trigger mitochondrial swelling, yet necessary to ensure PTP opening in response to specific stimuli (28). Indeed, P24 at 5 M caused mitochondrial swelling (Fig. 3), which could be attributed to PTP opening based on the significant (75%) inhibition exerted by CsA (29). The fact that CsA also significantly reduced (by ϳ70%) the drop of TMRM fluorescence by P24 (Fig. 2, B (green trace) and C) indicates a causal role of PTP opening in ⌬⌿ m dissipation. Thus, the larger drop in ⌬⌿ m caused by the concomitant addition of rotenone and P24 to HeLa cells, as compared with the effect of either rotenone or P24 alone (Fig. 2, A and C), was likely the consequence of both respiratory chain inhibition and PTP opening. Indeed, the simultaneous occurrence of these two alterations hampers the ability of F 0 F 1 -ATPase to maintain ⌬⌿ m (27). P18, cADPR, and Ap2A, at concentrations ranging between 1.0 and 10 M, did not induce mitochondrial swelling (not shown).
P24 Cytotoxicity Is Causally Related to the Dissipation of ⌬⌿ m -The proton gradient dissipation and the inhibition of Complex I exerted by P24 at concentrations (5-20 M) in the same range as those inducing cytotoxic effects on HeLa cells (Fig. 1B) suggested that the mitochondrial effects of the dinucleotide could be responsible for cell death. Indeed, CsA at 1 M remarkably reduced P24 cytotoxicity (Fig. 4). Moreover, extracellular Ca 2ϩ chelation during exposure of the cells to P24 completely prevented cytotoxicity (Fig. 4), demonstrating the fundamental role of the [Ca 2ϩ ] i increase in mediating the cytotoxic effect of P24. This experiment also highlights the fact that a brief (2 h) exposure of HeLa cells to P24 is sufficient for the dinucleotide to exert its cytotoxic effect, similar to what has been observed on colony-forming cells (16). As P24-induced ⌬⌿ m dissipation is Ca 2ϩ dependent (Fig. 2, B (red trace) and C), whereas Complex I inhibition is not (see above), the protective effects of EDTA and of CsA against P24 cytotoxicity indicate that PTP opening and ⌬⌿ m dissipation play a major role in inducing cell death. In line with this conclusion, the Complex I  At a 5-fold higher concentration than P24 (i.e. 100 M), P18, but not Ap2A, almost completely prevented the P24-induced increase of the [Ca 2ϩ ] cyt , which is due to extracellular Ca 2ϩ influx, being abrogated by EDTA (Ref. 16 and Fig. 5A). cADPR also prevented the P24-induced [Ca 2ϩ ] cyt rise (Fig. 5A), as the slight increase of the [Ca 2ϩ ] cyt observed in the presence of cADPR and P24 together was similar to that produced by 100 M alone (Fig. 5A, inset). A sustained cytosolic [Ca 2ϩ ] increase is known to result also in a mitochondrial [Ca 2ϩ ] rise (30). In fact, P24 also induced a sustained [Ca 2ϩ ] mit increase (Fig. 5B) as determined by confocal microscopy on HeLa cells loaded with the mitochondria-targeted fluorescent Ca 2ϩ probe Rho-2 AM (Fig. 5C). In line with their above-described effects on the [Ca 2ϩ ] cyt , cADPR, but not Ap2A, also reduced the P24-induced increase of the [Ca 2ϩ ] mit (Fig. 5B).
Interestingly, P18, which antagonized the increase of the [Ca 2ϩ ] cyt triggered by P24, synergized with P24 in inducing an even higher rise of the [Ca 2ϩ ] mit than that observed with P24 alone (Fig. 5B). In fact, P18 alone also induced an increase of the [Ca 2ϩ ] mit , similar to that observed with P24 alone (not shown).
In line with its inhibition of the P24-induced [Ca 2ϩ ] i rise and with the causal role of the [Ca 2ϩ ] i increase in determining ⌬⌿ m dissipation, cADPR also prevented the drop of TMRM fluorescence induced by P24 (Fig. 2, B (black trace) and C). Interestingly, Ap2A showed a protective effect similar to cADPR on P24-induced ⌬⌿ m dissipation (Fig. 2B), although it did not affect either the cytosolic or the mitochondrial [Ca 2ϩ ] increase induced by P24. This result suggests that Ap2A antagonizes a Ca 2ϩ -independent effect of P24 on the PTP that is necessary together with the [Ca 2ϩ ] cyt increase, for PTP opening. That both Ca 2ϩ and P24 are required to ensure PTP opening is also confirmed by the following observations, (i) in the absence of P24, the addition of 10 M Ca 2ϩ was not sufficient for PTP opening (Fig. 3) and (ii) cADPR and Ap2A both induce a [Ca 2ϩ ] cyt increase similar to P24 (16), but they do not have any effect on the ⌬⌿ m (not shown).
At a 5-fold higher concentration than P24, both cADPR and Ap2A significantly reduced P24 cytotoxicity on HeLa cells, in line with their protective effects on the ⌬⌿ m dissipation; the IC 50 value of P24 increased from 10 to 16 and 23 M in the presence of cADPR and Ap2A, respectively (Fig. 6). Conversely, presence of a 5-fold excess of P18 together with P24 increased cell death (Fig. 6), as expected from the intrinsic cytotoxic properties of P18 (16).

DISCUSSION
Here we identify the mitochondrial effects of a recently described, novel product of ADPRC, an enzyme activity known to synthesize several (di)nucleotides active on the [Ca 2ϩ ] i . At micromolar concentrations (1-20 M), P24, an isomer of Ap2A containing an unusual C1Ј-N3 glycosidic bond, induces ⌬⌿ m dissipation and inhibition of Complex I of the respiratory chain (Fig. 7). Dissipation of the ⌬⌿ m by P24 on intact HeLa cells appears to be mainly the consequence of PTP opening, as it is almost completely prevented (70%) by CsA (Fig. 2, B and C). The conclusion that P24 induces PTP opening is also confirmed by results obtained on isolated rat liver mitochondria showing that P24 causes a CsA-inhibitable matrix swelling (Fig. 3), the most direct evidence of PTP opening. However, presence of a CsA-insensitive effect of P24 on the PTP, and hence on the ⌬⌿ m , is indicated by the P24-induced stimulation of basal oxygen consumption (State 4) in isolated mitochondria, an uncoupling effect on respiration that occurs despite presence of CsA ( Table 1). Presence of CsA-insensitive effects of P24 on the ⌬⌿ m does not necessarily imply other sites of action of the dinucleotide besides the PTP (31)(32)(33).
Opening of the PTP by P24 is caused by the [Ca 2ϩ ] i increase, resulting from Ca 2ϩ influx triggered by the dinucleotide, as it is prevented by chelation of extra-or intracellular Ca 2ϩ (Figs. 2, B  and C, and 7). On the other hand, Ca 2ϩ per se is not sufficient to induce PTP opening, not even at micromolar concentrations (Fig. 3); thus, P24 appears to sensitize the PTP to Ca 2ϩ . The fact that Ap2A antagonizes P24-induced ⌬⌿ m dissipation (Fig. 2B) without affecting the [Ca 2ϩ ] i rise suggests a competition between Ap2A and P24 for direct binding to the PTP. In this respect, a regulatory, adenylic nucleotide-binding site is present on the PTP protein complex (34) that might be targeted by P24/Ap2A given the structural similarity between these dinucleotides and ADP. Extracellularly added P24 and Ap2A have been shown to cross the plasma membrane of intact HeLa cells, allowing access of these dinucleotides to the mitochondria (16).
Dissipation of ⌬⌿ m and inhibition of Complex I by P24 appear to be two independent effects. PTP opening and ⌬⌿ m dissipation by P24 are not secondary to inhibition of Complex I, as rotenone does not induce a significant loss of ⌬⌿ m (Fig. 2, A  and C). Inhibition of Complex I by P24 is not caused by dissipation of ⌬⌿ m , as it occurs also in the presence of EDTA and CsA (Table 1 and see "Results"), i.e. under conditions where PTP opening and ⌬⌿ m dissipation are largely prevented.
The cytotoxic effect of P24 appears to be causally related to the dissipation of ⌬⌿ m through PTP opening rather than to inhibition of Complex I. In fact, the Complex I inhibitor rotenone did not show any cytotoxic effect on HeLa cells. Conversely, PTP opening is recognized as a pivotal process in cell death, both in apoptosis and in necrosis (35). Indeed, P24 cytotoxicity was significantly reduced by CsA and completely prevented by EDTA (Fig. 4).
The effects of P24 on the PTP and on Complex I appear to be strictly related to the peculiar chemical structure of this dinucleotide, as neither of its isomers, P18 and Ap2A, shows any effect on ⌬⌿ m or on mitochondrial respiration. Interestingly, however, P18, Ap2A, and also cADPR affect P24 cytotoxicity; P18 decreases the IC 50 of P24 from 10 to 6 M, and Ap2A and cADPR instead increase the IC 50 value to 16 and 23 M, respectively (Fig. 6). The additive effect of P18 and P24 on cytotoxicity confirms the conclusion, drawn from the above-described experiments on mitochondrial function, that P18 and P24 have different targets of action. Ap2A and cADPR both prevent the P24-induced ⌬⌿ m dissipation (Fig. 2B) that is the principal cause for cell death, but through different mechanisms; cADPR  , which is antagonized by EDTA, cADPR, and P18 but not by Ap2A, leads to an increase of cytosolic Ca 2ϩ (1[Ca 2ϩ ] cyt ). This in turn induces a rise of the mitochondrial Ca 2ϩ (1[Ca 2ϩ ] mit ), synergized by P18 and antagonized by EDTA, EGTA-AM, and by cADPR, but not by Ap2A. The increase of the [Ca 2ϩ ] mit and P24 synergize in opening the permeability transition pore complex (PTP), thus triggering proton gradient dissipation (2⌬⌿ m ). P24 induces PTP opening also through a Ca 2ϩ -independent mechanism (possibly direct binding to the PTP complex), which is antagonized by Ap2A. Proton gradient dissipation is the principal cause of cell death, which occurs mainly through apoptosis and can be prevented or significantly reduced by co-incubation of cells with EDTA, EGTA-AM, the PTP desensitizer cyclosporin A (CsA), cADPR, or Ap2A together with P24. P24 also inhibits cell respiration at the level of Complex I (I), an effect that appears to be unrelated to ⌬⌿ m dissipation, as it is not affected by either EDTA or CsA, and not responsible for induction of apoptosis. Indeed, rotenone is not cytotoxic on HeLa cells. P18 synergizes with P24 in inducing a higher rise of the [Ca 2ϩ ] mit and an increase of the cytotoxic effect.
prevents the [Ca 2ϩ ] i rise triggered by P24 (Figs. 5 and 7) that is necessary for PTP opening whereas Ap2A does not affect the P24-induced [Ca 2ϩ ] i increase (Fig. 5), thus apparently interfering with a Ca 2ϩ -independent effect of P24 on the PTP (Fig. 7).
Among the important issues that remain to be elucidated at a molecular level, the channel responsible for Ca 2ϩ entry triggered by P24 (and antagonized by cADPR and P18) and the site of action of P24 on the PTP deserve priority. The extracellular Ca 2ϩ influx triggered by P24 is not dependent on Complex I inhibition and consequent possible ATP shortage to plasma membrane Ca 2ϩ -extruding pumps, because rotenone does not induce any [Ca 2ϩ ] i increase in HeLa cells (not shown).
In conclusion, results presented here identify the ADPRC product P24 as a novel endogenous regulator of mitochondrial function. The fact that Ap2A and cADPR antagonize P24-induced ⌬⌿ m dissipation and cytotoxicity suggests that the relative concentration of these ADPRC products may affect the delicate balance between survival and death in ADPRC ϩ cells.