Hydrogen Peroxide Induces Intracellular Calcium Overload by Activation of a Non-selective Cation Channel in an Insulin-secreting Cell Line*

Fura-2 fluorescence was used to investigate the effects of H2O2 on [Ca2+] i in the insulin-secreting cell line CRI-G1. H2O2 (1–10 mm) caused a biphasic increase in free [Ca2+] i , an initial rise observed within 3 min and a second, much larger rise following a 30-min exposure. Extracellular calcium removal blocked the late, but not the initial, rise in [Ca2+] i . Thapsigargin did not affect either response to H2O2, but activated capacitive calcium entry, an action abolished by 10 μm La3+. Simultaneous recordings of membrane potential and [Ca2+] i demonstrated the same biphasic [Ca2+] i response to H2O2 and showed that the late increase in [Ca2+] i coincided temporally with cell membrane potential collapse. Buffering Ca2+ i to low nanomolar levels prevented both phases of increased [Ca2+] i and the H2O2-induced depolarization. The H2O2-induced late rise in [Ca2+] i was prevented by extracellular application of 100 μm La3+. La3+(100 μm) inhibited the H2O2-induced cation current and NAD-activated cation (NSNAD) channel activity in these cells. H2O2 increased the NAD/NADH ratio in intact CRI-G1 cells, consistent with increased cellular [NAD]. These data suggest that H2O2 increases [NAD], which, coupled with increased [Ca2+] i , activates NSNAD channels, causing unregulated Ca2+ entry and consequent cell death.

Oxidative stress, through the production of oxygen metabolites, particularly H 2 O 2 and other reactive oxygen species (free radicals), results in destruction of many cell types through putative necrotic/apoptotic processes (1,2). Furthermore, excessive production of reactive oxygen species (e.g. via mitochondrial oxidation) has been causally related in the etiology of numerous degenerative disorders, including many age-related neurodegenerative diseases such as Parkinson's disease, Huntington's disease, and Alzheimer's disease (3)(4)(5). Although reactive oxygen species have been implicated in cell death, the exact mechanism(s) are, as yet, unclear. A favored hypothesis is that H 2 O 2 causes DNA strand breaks, leading to the activation of nuclear poly(ADP-ribose) polymerase, which critically depletes the cell of NAD, leading to eventual cell death (6). It has also been postulated that H 2 O 2 disrupts the cell membrane integrity in a nonspecific manner through lipid peroxidation (7). However, there is also a good correlation between oxidative stress (H 2 O 2 toxicity), induction of reactive oxygen species, and an increase in intracellular Ca 2ϩ levels immediately preceding the final destructive events (8).
Pancreatic beta cells have long been known to be particularly susceptible to oxidative stress-induced destruction (9), making these cells useful models for mechanistic studies. Indeed, alloxan, which is toxic to pancreatic beta cells through the production of H 2 O 2 and ultimately the highly reactive hydroxyl radical ( ⅐ OH), (10 -12), was observed to cause diabetes mellitus in experimental animals over 50 years ago (13). This susceptibility has been correlated with a reduced capacity to withstand free radical attack through a limited cellular defense mechanism, as pancreatic beta cells have been reported to be deficient in glutathione peroxidase, catalase, and superoxide dismutase (14,15) relative to other tissues. Therefore, we have combined Ca 2ϩ imaging and electrophysiological recordings of an insulinsecreting cell line (CRI-G1) to enable an investigation of the cellular consequences and mechanisms underlying mammalian cell responses to oxidative stress. Previously, it had been demonstrated, by intracellular recordings, that exposure to alloxan causes irreversible depolarization of mouse pancreatic beta cells (16). These initial observations have more recently been substantiated using whole-cell recordings, which show that alloxan and H 2 O 2 , through the production of reactive oxygen species, cause complete and irreversible depolarization of CRI-G1 insulin-secreting cells (17). This study also demonstrated that the H 2 O 2 -driven collapse of the membrane potential is mediated by the opening of a previously quiescent novel non-selective cation (NS NAD ) channel. Although activated by oxidative stress in intact cells, this non-selective cation channel requires the presence of both Ca 2ϩ and NAD on the cytoplasmic aspect of excised patches for channel activity to be observed (18,19). Permeation studies indicate that this channel has a significant conductance for divalent cations, most notably Ca 2ϩ (18); and therefore, oxidative stress-induced activation of this channel would be expected to allow a significant Ca 2ϩ influx associated with the collapse of the membrane potential.
We now report that exposure of CRI-G1 cells to concentrations of H 2 O 2 that activate NS NAD channels causes a biphasic rise in [Ca 2ϩ ] i . The second, late rise in [Ca 2ϩ ] i induced by H 2 O 2 reaches micromolar concentrations, indicating unregulated calcium influx. It is proposed that the second rise in [Ca 2ϩ ] i is caused by H 2 O 2 -induced activation of NS NAD channels, leading to concurrent depolarization and eventual cell death through calcium overload. Some of these data have been reported previously in preliminary form (20).

EXPERIMENTAL PROCEDURES
Cell Culture-Cells from the insulin-secreting cell line CRI-G1 were grown in Dulbecco's modified Eagle's medium containing sodium pyruvate (0.01%) and glucose (0.1%) and supplemented with 10% fetal calf serum and 1% (v/v) penicillin/streptomycin at 37°C in a humidified * 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.
atmosphere of 95% air and 5% CO 2 . Cells were passaged at 2-5-day intervals as described previously (21), plated either onto glass coverslips or directly onto 3.5-cm Petri dishes at dilutions that provided a subconfluent cell density (Falcon 3001), and used 1-4 days after plating.
Electrophysiological Recording and Analysis-Experiments were performed using whole-cell current and voltage clamp and single channel recording modes. Recording electrodes were pulled from borosilicate glass capillaries and had resistances of 8 -12 megaohms for outside-out recordings and 2-6 megaohms for whole-cell experiments when filled with electrolyte solution. Recordings were made using an Axopatch-1D or List EPC-7 patch clamp amplifier. Data were recorded onto digital audio tape and replayed for illustration onto a Gould TA 240 chart recorder. During current clamp experiments, hyperpolarizing current pulses (50 pA and 0.2-s duration) were applied every 5 s to monitor changes in input resistance. All voltage clamp experimental protocols were generated, and the resultant data were stored using PCLAMP6 (Axon Instruments, Inc.) and a Viglen PS/200 computer. In whole-cell voltage clamp recording mode, the membrane potential was held at Ϫ70 mV, and current-voltage relations were obtained by either applying 10-mV voltage steps of 200-ms duration with 100 ms between steps over the range Ϫ130 to Ϫ50 mV for H 2 O 2 -activated currents or 10-mV steps from Ϫ40 to ϩ40 mV every 5 s when investigating Ba 2ϩ currents. The mean current amplitude of the final 50 ms of the voltage clamp current response for each voltage jump was plotted against the applied voltage to generate current-voltage relations. Typical values for the series resistance during whole-cell recordings were 10 -18 megaohms. All values in the text are expressed as mean Ϯ S.E. Statistical significance between data sets was determined using unpaired Student's t test.
For outside-out patch experiments, the pipette solution contained 140 mM KCl, 5.02 mM CaCl 2 , 1 mM MgCl 2 , 5 mM EGTA, and 10 mM HEPES (pH 7.4), resulting in a free Ca 2ϩ concentration of 50 M; and the bathing medium consisted of 140 mM NaCl, 1 mM MgCl 2 , 1 mM CaCl 2 , and 10 mM HEPES (pH 7.2). The pipette solution for whole-cell recordings contained 140 mM KCl, 0.6 mM MgCl 2 , and 10 mM HEPES (pH 7.2), resulting in a free Ca 2ϩ concentration in the low nanomolar range, before stimulation with H 2 O 2 , and allowing an increase in free Ca 2ϩ concentration during stimulation. Some whole-cell recordings were performed with a pipette solution containing 140 mM KCl, 0.6 mM MgCl 2 , 10 mM EGTA, and 10 mM HEPES in order to clamp intracellular Ca 2ϩ in the low nanomolar range. The bath solution consisted of normal saline (135 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , and 10 mM HEPES (pH 7.4)). Additionally, when investigating voltage-dependent Ca 2ϩ channel currents, the pipette contained 140 mM CsCl, 10 mM EGTA, 0.6 mM MgCl 2 , 2.73 mM CaCl 2 , and 10 mM HEPES (pH 7.2), resulting in a free Ca 2ϩ concentration of 100 nM; and the bath contained normal saline supplemented with Ba 2ϩ (135 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 10 mM BaCl 2 , and 10 mM HEPES (pH 7.4)).
Imaging Intracellular Calcium of Single Cells-After washing with normal saline supplemented with 3 mM glucose, the plated cells were incubated with 2 M fura-2/AM for 45-120 min at room temperature (22-25°C) prior to experiments in which [Ca 2ϩ ] i was measured in intact cells. Single coverslips were mounted in a chamber on top of an inverted fluorescence microscope and perfused with normal saline plus 3 mM glucose. Measurements of [Ca 2ϩ ] i in individual cells were made from the fluorescence ratio (excitation at 340/380 nm and emission at Ͼ510 nm for fura-2) using a specially designed filter wheel assembly, incorporating a CCD camera, a photomultiplier, and a suite of software (MAGICAL, Applied Imaging, Sunderland, United Kingdom) that samples emission following excitation at 340-and 380-nm wavelengths at 15-s intervals. [Ca 2ϩ ] i was calculated from a calibration curve using the equation [Ca 2ϩ ] i ϭ K d ⅐␤⅐((R Ϫ R min )/(R max Ϫ R)), where R max , R min , and R are the maximum ratio, minimum ratio, and measured ratio, respectively; ␤ is minimum 380/maximum 380; and K d represents the Ca 2ϩ binding affinity of fura-2. R max , R min , and ␤ were determined from free standing solutions of 2 and 0 mM Ca 2ϩ o (ϩ5 mM EGTA) in HEPES buffer solution as described previously (22,23); these values did not vary significantly from day to day. Zero extracellular Ca 2ϩ was obtained by the removal of CaCl 2 from the normal saline as well as by the addition of 5 mM EGTA. All intact cell imaging was performed on at least three separate cultures, and numbers quoted (n) represent the number of individual cells analyzed. The intrinsic autofluorescence of cells, induced by the presence of reduced pyridine nucleotides, was examined without loading with fura-2/AM. Significant changes in autofluorescence were observed using excitation at 360 nm and emission at Ͼ510 nm.
Combined Ca 2ϩ Imaging and Electrophysiology-Methods for detecting [Ca 2ϩ ] i in cells while recording in the whole-cell configuration were essentially the same as those used for intact cells, except that the pipette solution contained 140 mM KCl, 0.6 mM MgCl 2 , 10 mM HEPES, and 50 M fura-2 (pH 7.2), and in some experiments, 10 mM EGTA was added to buffer intracellular Ca 2ϩ to low nanomolar levels. The bath solution consisted of normal saline (135 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 10 mM HEPES, and 3 mM glucose (pH 7.4)). Under these conditions in the cell-attached configuration, the fluorescence ratio indicated Ͻ10 nM free Ca 2ϩ i contamination of the pipette solution from stock reagents. In these experiments, the cells were viewed on an upright Zeiss Axioskop microscope with the appropriate dichroic mirror (405 nm) and trinocular mounting to send the emitted light to the photomultiplier tube (PHOCAL, Life Sciences Resources, Basingstoke, United Kingdom).
All solution changes were achieved by superfusing the bath with a gravity-feed system at a rate of ϳ10 ml/min, which allowed complete bath exchange within 2 min. All experiments were performed at room temperature (22-25°C).

RESULTS
Intact CRI-G1 cells were loaded with fura-2/AM, and the effect of H 2 O 2 on the fura-2 ratio ([Ca 2ϩ ] i ) was monitored. The addition of either 1 or 10 mM H 2 O 2 to the extracellular medium resulted in a biphasic increase in [Ca 2ϩ ] i (Fig. 1A). Although Only the initial small sustained rise in intracellular calcium levels occurred in the absence of extracellular calcium. The re-addition of extracellular calcium resulted in an immediate massive calcium increase (near-dye saturation (Ͼ1 M)). Note that exposure to H 2 O 2 in the absence of extracellular Ca 2ϩ was longer in duration than the time taken to reach a maximal response in A.
the [Ca 2ϩ ] i response to 1 and 10 mM H 2 O 2 differed temporally (data not shown), the magnitude of both the early and late phases of increased [Ca 2ϩ ] i was indistinguishable (p Ͼ 0.05). The initial rise occurred within 3 min, and the fura-2 ratio increased from an initial resting level of 0.44 Ϯ 0.01 (ϳ57 nM) to 0.93 Ϯ 0.02 (190 nM; n ϭ 72). The second, late response occurred ϳ30 min later (ϳ40 min for 1 mM H 2 O 2 ) and induced a rise to near-dye saturation (fura ratio Ͼ 2; Ͼ1 M Ca 2ϩ i ). Removal of extracellular calcium by the addition of 5 mM EGTA decreased the initial rise in [Ca 2ϩ ] i (ratio rising from 0.40 Ϯ 0.01 (48 nM) to 0.73 Ϯ 0.02 (130 nM); n ϭ 52; p Ͻ 0.01) and completely abolished the second rise induced by 10 mM H 2 O 2 (Fig. 1B). The re-addition of extracellular calcium resulted in an immediate increase in [Ca 2ϩ ] i , again rising to near-dye saturation, indicating that a calcium permeation pathway had been activated by the H 2 O 2 exposure. These results indicate that the H 2 O 2 -induced early increase in intracellular calcium levels results predominantly from mobilization of Ca 2ϩ i from an intracellular source and that the second, late increase in intracellular calcium levels is a result of extracellular calcium influx.
Previous studies examining the consequences of mammalian cell exposure to reactive oxygen species have generally been inconclusive with respect to the identity of the permeation pathway for the oxidative stress-induced rise in [Ca 2ϩ ] i consequent to Ca 2ϩ entry from the extracellular medium. There are reports that implicate voltage-gated calcium channels (through the use of organic Ca 2ϩ channel blockers) as the source of calcium influx (24,25), whereas others indicate that such a pathway is unlikely (26,27). Even if voltage-gated calcium channels were not the primary path, reactive oxygen speciesderived depolarization of cells could activate these channels, which could then secondarily contribute to the increased intracellular calcium levels. To examine whether voltage-gated calcium channels contribute in any way to the H 2 O 2 -induced increase in [Ca 2ϩ ] i in CRI-G1 cells, 100 M Cd 2ϩ was applied to the extracellular environment of the cells. This concentration of Cd 2ϩ completely inhibits voltage-gated calcium currents (n ϭ 5) in this cell line, an action partially reversible on washout of the Cd 2ϩ ( Fig. 2A). As further verification that Cd 2ϩ inhibits the function of voltage-gated calcium channels in these cells, the increased [Ca 2ϩ ] i response associated with a depolarization-induced (40 mM K ϩ ) calcium influx was shown to be completely prevented by 100 M Cd 2ϩ (n ϭ 12) (Fig. 2B). However, in the presence of extracellular calcium, co-application of 100 M Cd 2ϩ and 10 mM H 2 O 2 failed to inhibit the H 2 O 2 -induced rise in [Ca 2ϩ ] i , with the fura-2 ratio increasing to dye saturation (n ϭ 13) (Fig. 3A) A, mean current-voltage relation, from a cell at a holding potential of Ϫ70 mV, obtained over the voltage range Ϫ40 to ϩ40 mV with 140 mM Cs ϩ in the pipette (q) and the 1 mM extracellular Ca 2ϩ replaced with 10 mM Ba 2ϩ . This current had a threshold for activation of near Ϫ40 mV and was maximal at 0 mV. The addition of 100 M Cd 2ϩ (E) to the bathing medium completely abolished this current. Note the presence of a small residual current at both hyperpolarized and depolarized potentials, probably due to contamination by non-selective cation or anion currents. Representative barium currents were observed at Ϫ10 mV; the initial inward spike is due to activation of voltage-gated sodium currents. The application of 100 M Cd 2ϩ resulted in complete and reversible inhibition of the barium current, but had no effect on the sodium current. B, fluorescence measurements of a single fura-2/AM-loaded CRI-G1 cell illustrating the rise in [Ca 2ϩ ] i caused by depolarization elicited by 40 mM extracellular K ϩ . The additional presence of 100 M Cd 2ϩ completely abolished the 40 mM K ϩ , depolarization-induced calcium increase. Inhibition produced by Ca 2ϩ was reversed upon removal (data not shown).
form reactive oxygen species (9). The effects of Cd 2ϩ on [Ca 2ϩ ] i were also examined in the absence of extracellular calcium, and Cd 2ϩ was demonstrated to have no effect on [Ca 2ϩ ] i per se and did not alter the initial H 2 O 2 -induced rise in [Ca 2ϩ ] i (ratio rising from 0.37 Ϯ 0.01 (42 nM) to 0.83 Ϯ 0.06 (155 nM); p Ͼ 0.05 compared with data in the absence of Cd 2ϩ ). However, readdition of 1 mM Ca 2ϩ -containing extracellular solution in the continued presence of Cd 2ϩ induced a rapid, large increase in [Ca 2ϩ ] i (n ϭ 13) (Fig. 2B). Consequently, it appears highly unlikely that activation of voltage-dependent calcium channels is responsible for, or contributes significantly to, the H 2 O 2induced second, late phase of increased [Ca 2ϩ ] i in CRI-G1 cells.
Another possible contributor to this second, large rise in [Ca 2ϩ ] i is by means of an intracellular Ca 2ϩ -triggered capacitive entry system for Ca 2ϩ (I CRAC ) that has somehow become unregulated by the presence of H 2 O 2 . The data described above indicate that the initial phase of increased [Ca 2ϩ ] i in response to H 2 O 2 challenge is due to mobilization of calcium from intracellular stores. Consequently, the possibility that the calcium originates from microsomal stores was investigated using thapsigargin, an endoplasmic reticulum Ca 2ϩ -ATPase inhibitor that depletes intracellular Ca 2ϩ stores by blocking the uptake of calcium (28). The addition of 1 M thapsigargin to intact cells resulted in a rapid rise in [Ca 2ϩ ] i , which peaked and then declined, but did not return to initial pre-thapsigargin levels (Fig. 4A), with the fura-2 ratio increasing from 0.41 Ϯ 0.01 (50 nM) to 0.83 Ϯ 0.03 (155 nM) and then plateauing to a new level of 0.59 Ϯ 0.02 (90 nM) (n ϭ 32). Following thapsigargin treatment, removal of extracellular calcium decreased [Ca 2ϩ ] i to a level below that seen in control experiments, 0.32 Ϯ 0.01 (32 nM; p Ͻ 0.05). The subsequent addition of 10 mM H 2 O 2 resulted in a response indistinguishable from that of cells not exposed to thapsigargin (Fig. 1A), i.e. an initial sustained increase in [Ca 2ϩ ] i , with the fura-2 ratio rising from 0.32 Ϯ 0.01 (32 nM) to 0.76 Ϯ 0.02 (137 nM) (p Ͼ 0.05), followed by a second, larger rise following re-addition of extracellular Ca 2ϩ , reaching near-dye saturation. It is therefore clear that H 2 O 2 mobilizes intracellular calcium from a thapsigargin-insensitive source. Previous studies have demonstrated that emptying of intracellular Ca 2ϩ stores in pancreatic beta cells triggers capacitive calcium entry  through I CRAC (29,30) and that low (10 -100 M) concentrations of La 3ϩ inhibit this process (31). In the CRI-G1 cells, emptying of intracellular Ca 2ϩ stores with thapsigargin resulted in the activation of I CRAC . This can be seen (Fig. 4, A and  B) by removal and re-addition of extracellular calcium, which resulted in [Ca 2ϩ ] i transients not observed without exposure to thapsigargin (n ϭ 32 and 24 with and without thapsigargin, respectively). The induction of [Ca 2ϩ ] i transients indicative of I CRAC activation in these cells was maximally and reversibly inhibited by the presence of 10 M La 3ϩ (Fig. 4B). In contrast, the presence of 10 M La 3ϩ did not abolish the influx of extracellular Ca 2ϩ (n ϭ 13) induced by 10 mM H 2 O 2 , although there was an ϳ30% inhibition (Fig. 4B). However, application of 100 M La 3ϩ did eradicate the second [Ca 2ϩ ] i transient due to H 2 O 2 -induced Ca 2ϩ influx (n ϭ 24) (Fig. 4C), an action that was reversible, as washout of the La 3ϩ in the continued presence of extracellular Ca 2ϩ resulted in eventual massive influx of Ca 2ϩ . Consequently, these data indicate that the H 2 O 2 -induced Ca 2ϩ influx is pharmacologically distinct from the I CRAC pathway in these cells and also exclude a nonspecific membrane breakdown as mediator of Ca 2ϩ entry.
The identity of the Ca 2ϩ entry pathway responsible for the second, late increase in [Ca 2ϩ ] i induced by H 2 O 2 in CRI-G1 cells was investigated using whole-cell current clamp recordings with 50 M fura-2 in the intracellular solution in order to monitor simultaneously membrane potential and the levels of free intracellular calcium. After initiation of the whole-cell recording mode, the cell interior was dialyzed with an ATP-free solution, which resulted in cell hyperpolarization (Fig. 5A, upper trace). The membrane potential changed from an initial resting value of approximately Ϫ40 mV to a new stable potential of Ϫ74 Ϯ 1.8 mV (n ϭ 6). Concomitant with the hyperpolarization was a decrease in cell input resistance. This hyperpolarization and decreased input resistance are due to the opening of ATP-sensitive K ϩ (K ATP ) channels as internal ATP is washed out of the cell (17,32). Following stabilization of the membrane potential and input resistance, application of 10 mM H 2 O 2 caused a sustained increase in [Ca 2ϩ ] i (Fig. 5A, lower trace), which preceded any effect on membrane potential. This increase occurred within 3 min (2.6 Ϯ 0.2 min; n ϭ 6), and the fura-2 ratio rose from 0.45 Ϯ 0.02 (46 nM) to 0.87 Ϯ 0.08 (150 nM). There was then a further delay of ϳ10 -12 min before a second phase of increased [Ca 2ϩ ] i was observed (fura-2 ratio rising from 0.87 Ϯ 0.08 to 1.94 Ϯ 0.16 (Ͼ600 nM)) ( Fig. 5A, lower trace). These [Ca 2ϩ ] i responses to 10 mM H 2 O 2 almost exactly recapitulated those obtained in intact cells. The decreased delay for the second phase of increased [Ca 2ϩ ] i during whole-cell recordings in comparison with intact cells is likely due to dialysis of the cells' natural antioxidant protective mechanisms (e.g. glutathione). The second, late phase of increased [Ca 2ϩ ] i coincided with the appearance of a slow, irreversible depolarization to ϳ0 mV. This H 2 O 2 -induced depolarization is consistent with our previous report on the action of H 2 O 2 in this cell line (17).
To determine whether the later large increase in [Ca 2ϩ ] i induced by H 2 O 2 is dependent upon the initial increase in [Ca 2ϩ ] i , simultaneous recordings of membrane potential and [Ca 2ϩ ] i were made with 10 mM EGTA included in the intracellular pipette solution to clamp the [Ca 2ϩ ] i to low nanomolar levels. As illustrated in Fig. 5B, the inclusion of 10 mM EGTA in the pipette clamped the [Ca 2ϩ ] i below the resting level (0.26 Ϯ 0.01; Ͻ1 nM), and this did not change after application of 10 mM H 2 O 2 . Furthermore, although the reduction in intracellular calcium had no effect on the hyperpolarization induced by the washout of the internal ATP (the mean cell hyperpolarization was to Ϫ75 Ϯ 2.4 mV; n ϭ 4), it completely prevented the H 2 O 2 -induced depolarization (Fig. 5B, lower trace).
Clearly, activation of the late phase of increased [Ca 2ϩ ] i by

FIG. 5. H 2 O 2 depolarizes CRI-G1 cells as well as causes a biphasic calcium response.
Shown are the results from the simultaneous measurement of fura-2 fluorescence and whole-cell current clamp recording of the membrane potential of CRI-G1 cells. A, application of 10 mM H 2 O 2 induces a biphasic rise in intracellular calcium levels (lower trace), with the second, larger rise coinciding exactly with cellular depolarization (upper trace). B, simultaneous measurement of fura-2 fluorescence and cell membrane potential with nominally calcium-free electrode solution (10 mM EGTA). Application of 10 mM H 2 O 2 had no effect on either intracellular calcium or cell membrane potential. Note the initial hyperpolarization of the cell membrane potential on initiation of the whole-cell recording configuration (due to intracellular dialysis of ATP). The downward deflections of the membrane potential traces in A and B were caused by hyperpolarizing current pulses (50 pA and 0.2-s duration) applied every 5 s to monitor input resistance changes. H 2 O 2 is dependent upon the initial rise in calcium levels, is through a specific permeation pathway, and is consequent to the depolarization of the cells. Previous studies have demonstrated that the H 2 O 2 -induced depolarization of CRI-G1 cells is due to the opening of a novel non-selective cation conductance (NS NAD channel), which also allows permeation of Ca 2ϩ (17,18). Activation of this channel by H 2 O 2 is shown in Fig. 6, where cell-attached single channel recordings from CRI-G1 cells were performed. Following application of H 2 O 2 (4.4 mM), there was a delay of ϳ9 -10 min before the appearance of single channel currents (Fig. 6A). This current was characterized by a linear current-voltage relation with a conductance of 70 pS 1 and a reversal potential of 0 mV (Fig. 6, B and C). Excision of the patch into the inside-out configuration resulted in loss of channel activity, which could only be recovered by application of NAD to the cytoplasmic aspect of the membrane patch (Fig.  6C), identifying the channel as NS NAD (17). We also examined the likelihood that this channel is responsible for the late phase of Ca 2ϩ entry induced by H 2 O 2 by determining the sensitivity of this conducting pathway to inhibition by La 3ϩ . Following attainment of the whole-cell recording configuration (Fig. 7A), the conductance of the cell was small (1.46 Ϯ 0.54 nS; n ϭ 3), with a reversal potential of Ϫ42.3 Ϯ 4.0 mV (n ϭ 3), which is close to the cell resting membrane potential of approximately Ϫ40 mV. After dialysis with the ATP-free pipette solution, the conductance increased to 18.2 Ϯ 2.80 nS, with a reversal potential of Ϫ70 Ϯ 3.6 mV, which is attributed to activation of K ATP channels. In contrast, subsequent to the H 2 O 2 -induced depolarization of CRI-G1 cells, voltage clamp recordings dem-onstrated the activation of a non-selective cation current characterized by a linear slope conductance (Fig. 7A) of 4.6 Ϯ 1.1 nS with a concurrent loss of K ATP current, indicated by a shift in the reversal potential to 5.7 Ϯ 8.5 mV (n ϭ 3). Application of 100 M La 3ϩ caused a 50 -80% inhibition of the H 2 O 2 -activated current, decreasing the slope conductance to 2.1 Ϯ 0.1 nS with only a slight shift in the reversal potential (to Ϫ1 mV). The inhibition of this current by La 3ϩ was partially reversed after a Ͼ5-min wash, with the slope conductance increasing to a value of 3.7 Ϯ 0.2 nS (n ϭ 3) (Fig. 7A). The inhibitory effect of La 3ϩ was also determined on single channel currents activated by 1 mM NAD ϩ in the presence of 50 M Ca 2ϩ in isolated outside-out membrane patches. In all outside-out patches examined, the addition of 100 M La 3ϩ to the extracellular membrane aspect caused complete cessation of channel activity (Fig. 7B), an action not reversible even with prolonged washing (n ϭ 6).
The data presented above are consistent with the notion that H 2 O 2 induces the release of Ca 2ϩ from an as yet undefined intracellular Ca 2ϩ store, which yields the first phase of increased [Ca 2ϩ ] i , and this is obligatory for the subsequent activation of a substantial Ca 2ϩ influx through NS NAD channels, producing the second phase of increased [Ca 2ϩ ] i . The activity of the NS NAD channel in isolated patches is Ca 2ϩ -dependent, but the presence of intracellular Ca 2ϩ per se is not sufficient to sustain channel activity; there is also an absolute requirement for intracellular NAD. Therefore, the effect of H 2 O 2 exposure on pyridine nucleotide levels was examined in intact cells utilizing the autofluorescence properties of these nucleotides (30,33). Application of 10 mM H 2 O 2 caused a rapid (within 2-3 min) but transient decrease in NAD(P)H autofluorescence both in the absence and presence of extracellular calcium (n ϭ 24; p Ͼ 1 The abbreviation used is: S, siemens. 0.05) (Fig. 8). Cellular autofluorescence decreased by 17.4 Ϯ 1.8% within the first 5 min and then gradually returned to control levels during the remainder of the experiment (1 h). This indicates that H 2 O 2 causes an initial oxidation of pyridine nucleotides, resulting in an increase in the NAD/NADH ratio, indicating a transient increase in NAD levels. Such a change in NAD combined with the initial rise in [Ca 2ϩ ] i could conspire to create an environment conducive to NS NAD channel activation.

DISCUSSION
Previous studies have demonstrated that H 2 O 2 , most probably through the generation of reactive oxygen species, causes an increase in intracellular calcium levels that precedes, if not causes, cell death in a wide variety of cell types, including cardiac (27) and smooth muscle cells (26), pancreatic acinar cells (34), alveolar macrophages (35), and central nervous system neurons (36). Here, we describe the effects of H 2 O 2 on [Ca 2ϩ ] i in the insulin-secreting cell line CRI-G1 and clearly demonstrate that H 2 O 2 also disrupts calcium homeostasis in these cells. There is no clear consensus in the literature indicating the likely mechanism by which H 2 O 2 causes an increase in [Ca 2ϩ ] i ; suggestions that have been promulgated include influx through voltage-dependent calcium channels (25,35), nonspecific changes in membrane calcium permeability (37,38), alteration in Na ϩ -Ca 2ϩ exchange (39), or changes in calcium release from intracellular stores (40,41 A, current-voltage relations obtained from a voltage-clamped cell, over the voltage range Ϫ130 to Ϫ50 mV, from a holding potential of Ϫ70 mV prior to and following H 2 O 2 -induced (10 mM) cell depolarization. Initially, the current was linear with a reversal potential near Ϫ40 mV (q), and following dialysis of the cell with an ATP-free solution (E), the current dramatically increased, and the reversal potential shifted more negative, indicating activation of the K ATP current. In the presence of H 2 O 2 , the K ATP current was abolished, and there was an inward current characterized by a linear current-voltage relation (f), which, on application of 100 M external La 3ϩ , was significantly reduced (Ⅺ), an action only partially reversed on washout of La 3ϩ (ࡗ). Representative families of currents were generated in the same experiment. B, representative continuous outside-out recording illustrating the irreversible inhibition of NAD-activated (NS NAD ) single channel activity by 100 M external La 3ϩ . This experiment was performed on an outside-out patch with 1 mM internal NAD and 50 M free Ca 2ϩ in the pipette at a membrane potential of Ϫ40 mV.
ization. The depolarization elicited by H 2 O 2 is not simply due to a reduction in K ATP channel current in these cells, as in separate cell-attached recordings, it was demonstrated that the collapse of the membrane potential is associated with the induction of a separate conducting pathway (see below). This result indicates that the initial rise in [Ca 2ϩ ] i activates a calciumdependent process that leads to both cell depolarization and, as a consequence, a second rise in [Ca 2ϩ ] i . Further substantiation of this hypothesis was obtained by chelating intracellular calcium to low nanomolar levels, an action that prevented both phases of the Ca 2ϩ response and the H 2 O 2 -induced depolarization. The source of this later, larger increase in [Ca 2ϩ ] i is clearly from the extracellular milieu, and consequently, a number of conducting pathways were considered as possible contributors.
Activation of voltage-gated calcium channels has been implicated in the autoimmune-associated destruction of pancreatic beta cells (44) and has been suggested to contribute to cellular response to oxidative stress (25,35,45). Therefore, a possible contribution to the observed effects of H 2 O 2 from this permeation pathway was investigated in CRI-G1 cells using the inorganic calcium channel blocker Cd 2ϩ . A Cd 2ϩ concentration (100 M) that maximally inhibited the voltage-gated calcium current in these cells was shown to completely inhibit the Ca 2ϩ influx induced by depolarization. However, this concentration of Cd 2ϩ had no effect on the magnitude (but did accelerate the response) of the H 2 O 2 -induced rise in [Ca 2ϩ ] i , indicating that voltage-dependent calcium channels are not involved in the H 2 O 2 -induced late phase of calcium influx observed in these cells. Another possible calcium permeation pathway, in this case linked to intracellular calcium stores, is the capacitive calcium entry (I CRAC ), a non-selective cation current present in pancreatic beta cells and activated by emptying of intracellular stores (30,46,47). Depletion of intracellular Ca 2ϩ stores in CRI-G1 cells with thapsigargin resulted in the activation of a calcium influx pathway, indicative of the presence of I CRAC in these cells. I CRAC is known to be inhibited by low concentrations (Ͻ100 M) of extracellular metal ions, particularly La 3ϩ (31,48). In CRI-G1 cells, the capacitive calcium entry pathway was completely blocked by the application of low (10 M) concentrations of La 3ϩ . However, this concentration of La 3ϩ had only a small inhibitory effect on the H 2 O 2 -induced late increase in [Ca 2ϩ ] i , indicating that the calcium influx activated by H 2 O 2 is distinct from that activated by emptying of intracellular Ca 2ϩ stores. Although low concentrations of La 3ϩ failed to block the H 2 O 2 -induced calcium influx, a higher concentration of La 3ϩ (100 M) completely abolished the calcium influx. This also indicates that although the calcium influx is independent of both I CRAC and voltage-dependent calcium channels, it cannot be explained by nonspecific membrane degradation.
Recently, it has been demonstrated that application of H 2 O 2 to CRI-G1 cells evokes the activation of a calcium-dependent non-selective cation (NS NAD ) channel, which results in the complete and irreversible collapse of the cell membrane potential (17). The NS NAD channel is permeable to calcium (18), and therefore, the possibility that the second, larger rise in [Ca 2ϩ ] i resulted from influx of calcium through the NS NAD channel was examined. Dual calcium imaging and whole-cell current clamp recordings support the contention that the activation of the NS NAD channel by H 2 O 2 is causally related to the late increase in [Ca 2ϩ ] i . There is an excellent temporal coincidence between the cellular depolarization and the late phase of Ca 2ϩ influx caused by H 2 O 2 , and removal of extracellular Ca 2ϩ prevents the depolarization. Although H 2 O 2 also inhibits K ATP currents (most likely due to Ca 2ϩ entry initiating K ATP current rundown/inactivation) concomitant with the appearance of the non-selective cation current (see Fig. 7A), the overriding influence for cell depolarization is the activation of the NS NAD channel as evidenced by the cell-attached recordings. Furthermore, strong evidence for the NS NAD channel providing the depolarizing driving force is provided by the inhibition of the H 2 O 2 -induced Ca 2ϩ influx by 100 M La 3ϩ , a concentration of La 3ϩ that was also observed to inhibit the H 2 O 2 -activated macroscopic current as well as the NAD-activated single channel currents. In the absence of a specific inhibitor of the NS NAD channel, we cannot exclude other explanations for the above results, but activation of this permeation pathway appears to be the most parsimonious interpretation. It should be noted that the NS NAD channel appears to be a separate conductance pathway from that activated by incretin hormones and maitotoxin, which has the characteristics of a small conductance (30 pS), calcium-activated, non-selective cation channel (49).
The diabetogenic compound alloxan and H 2 O 2 have been postulated to destroy pancreatic beta cells through DNA damage, which consequently activates nuclear poly(ADP-ribose) synthase, leading to critical depletion of cellular NAD pools (6). This appears to contradict the mechanism proposed above, as cellular NAD is presumably required to activate the NS NAD channel. However, the autofluorescence results indicate that H 2 O 2 causes an immediate oxidation of NAD(P)H, resulting in an increased NAD/NADH ratio, which then declines again within 1 h. Therefore, although the final outcome of H 2 O 2 exposure may well be NAD depletion, the acute effect is an increase in NAD levels, which, coupled with a rise in [Ca 2ϩ ] i , may well be sufficient to activate the NS NAD channel, causing the cell to depolarize, to flood with calcium, and ultimately to die.
Consequently, we propose that H 2 O 2 causes a loss of intracellular calcium homeostasis in CRI-G1 cells by inducing a biphasic rise in [Ca 2ϩ ] i , with the first phase caused by mobilization of intracellular calcium and the second, larger phase due to influx from the extracellular medium. The source of intracellular calcium is not the inositol 1,4,5-trisphosphate-or cyclic ADP-ribose-sensitive endoplasmic reticulum stores, and its identity is as yet elusive. However, the second phase of increased [Ca 2ϩ ] i is consistent with the influx of extracellular calcium permeating through the NS NAD channel. This represents a novel calcium influx pathway activated by oxidative stress. The wider implications of this are unclear as yet, but there have been many studies in which a large influx in calcium has been observed following oxidative stress, but in which the influx pathway has not been identified. For example, Bielefeldt et al. (26) have recently shown that intestinal smooth muscle cells exposed to H 2 O 2 produce a biphasic rise in [Ca 2ϩ ] i , reminiscent of the data presented within. Additionally, their response was insensitive to both depletion of intracellular Ca 2ϩ stores and inhibition of voltage-dependent calcium channels. Similarly, the human astrocytoma cell line UC11MG has been shown to respond with a biphasic rise in [Ca 2ϩ ] i to exposure to H 2 O 2 , which was also insensitive to calcium channel blockers (50). Therefore, it is possible that these and other unattributed Ca 2ϩ influxes resulting from cell death-inducing stimuli are caused by the activation of the NS NAD channel or by a similar non-selective cation channel.