Activation of transient receptor potential (melastatin) 2 (TRPM2) channels during oxidative stress promotes pancreatic β-cell death, resulting in hyperglycaemia. Cell death is caused by TRPM2-mediated Ca2+-induced intracellular Zn2+ release, but not by Ca2+ alone.

Pancreatic β-cell death is a hallmark of Type 1 diabetes (T1D). It is caused by the autoimmune destruction of β-cells, leading to insulin deficiency and severe hyperglycaemia (T1D) [13]. Reactive oxygen species (ROS), including H2O2, play a central role in the immune-mediated β-cell death [1,2]. One mechanism by which ROS brings about cell death begins with nicks in ssDNA and activation of the poly(ADP-ribose) polymerase (PARP)–poly(ADP-ribose) glycohydrolase (PARG) pathway [4]. PARP catalyses the transfer of ADP-ribose moieties from NAD to acceptor proteins that are subsequently released by the action of PARG. Activation of PARP–PARG pathway thus results in NAD depletion and production of ADP-ribose. Since NAD is required for ATP generation, NAD depletion was thought to result in energy deficiency and cell death by necrosis [57]. A potential link between NAD depletion and T1D was demonstrated by the findings that knockout of PARP in mice prevents streptozotocin (STZ)-induced β-cell death and hyperglycaemia [5,7]. Subsequent studies however have shown that ADP-ribose, the by-product of NAD metabolism, is a potent activator of the transient receptor potential (melastatin) 2 (TRPM2) channel [8] and inhibition of TRPM2 channels prevents apoptosis [912]. These later results raised the possibility that TRPM2 (ADP-ribose)-mediated β-cell apoptosis could also contribute to T1D, but its role relative to energy depletion has not been investigated.

TRPM2 is a non-selective cation channel activated by H2O2. Although H2O2 can directly activate the channel, activation is largely mediated by ADP-ribose [8,1012]. The channels are also directly activated through the oxidation of a methionine located in the cytoplasmic N-terminal domain [13]. TRPM2 expression was first reported in the plasma membrane, but a previous study demonstrated its expression in the lysosomal membranes and a role in the lysosomal Ca2+ release [14]. Its activation thus can raise cytosolic Ca2+ through extracellular entry and lysosomal release. The TRPM2 channel is proposed to play a dual role in β-cell physiology with modest TRPM2-mediated Ca2+ elevation being implicated in glucose-stimulated insulin secretion [15,16], although the in vivo evidence in TRPM2-knockout mice is somewhat controversial [15,17]. However, strong activation of the channel causes excessive Ca2+ elevation resulting in β-cell death [9,11,12,14,18]. Interestingly, a previous study demonstrated that TRPM2 channels can be permeate by Zn2+ ions [19]. Although the relevance of this function to cell death has not yet been examined, Zn2+ is a bona fide cytotoxic ion [20,21] and could also contribute to TRPM2-mediated β-cell death.

The aim of the present study was to investigate the relative roles of ROS (H2O2) generated signals in β-cell death, with a focus on TRPM2-mediated Ca2+ and Zn2+ signals. Our results demonstrate that (1) activation of TRPM2 channels leads to intracellular Zn2+ release and that Zn2+ release is potentiated by extracellular Ca2+ entry; (2) rise in Ca2+ alone cannot cause cell death, but the concomitant Zn2+ release is essential for cell death; and (3) finally, knockout of TRPM2 channels protects mice from STZ-induced β-cell death and hyperglycaemia. Interestingly, the results exclude NAD (energy) depletion as the primary cause of β-cell death.

Animals

C57BL/6 (wild-type) and TRPM2 knockout mice (TRPM2−/−) were used in the present study. Generation of TRPM2−/− mice is as described previously [22]. These mice express a non-functional TRPM2 channel lacking the transmembrane domains 3 and 4 (encoded by exons 17 and 18). All procedures involving mice were performed under U.K. Home Office licence and ethical procedure.

Cell culture

Human embryonic kidney (HEK)-293 and insulin secreting (INS1) cells were cultured as described previously [23]. A HEK-293 cell line conditionally expressing TRPM2 under the control of a tetracycline-regulated promoter (a gift from Dr A.M. Scharenberg, University of Washington, Seattle, WA, U.S.A.) was cultured as for the HEK-293 cell line with the addition of blasticidin (5 μg·ml−1; Invitrogen) and zeocin (400 μg·ml−1; Invitrogen). Expression of TRPM2 was induced by overnight exposure to tetracycline (1 μg·ml−1). All cell lines were grown at 37°C in a humidified atmosphere containing 5% CO2.

Plasmid construction and transfection

A haemagglutinin (HA) epitope was introduced into the extracellular S1–S2 loop of TRPM2 using overlap extension PCR. The addition of the HA tag did not affect the function of the channel. The HA-tagged TRPM2 cDNA was subcloned into pCDNA3.1 and transfected into HEK-293 cells using Fugene® HD (Promega).

Antibodies and reagents

For immunostaining, rat anti-HA (clone 3F10) monoclonal antibodies (100 ng·ml−1, Roche Applied Science), mouse anti-CD63 primary antibodies (1:500, Abcam), guinea pig polyclonal anti-insulin (1:500, Dako) and rabbit polyclonal anti-glucagon (1:200, New England Biolabs) were used as primary antibodies. Alexa Fluor 488-conjugated anti-rat (1:500, Life Technologies), Cy3-conjugated anti-mouse (1:500, Jackson Immunoresearch), Alexa Fluor 488-conjugated anti-guinea pig (1:1000, Life Technologies) and Cy3-conjugated anti-rabbit (1:500, Jackson Immunoresearch) antibodies were used as corresponding secondary antibodies. FluoZin3™-AM, Fluo4-AM, Pluronic® F127, ER-Tracker™ Red, MitoTracker® Red CMXRos and LysoTracker® Red DND-99 were from Life Technologies. STZ and PJ34 were from Calbiochem. Collagenase (type IA), 2-aminoethoxydiphenyl borate (2-APB), N,N,N′,N′-tetrakis-(2-pyridylmethyl)ethylenediamine (TPEN), diethylenetriaminepenta-acetate (DTPA), clioquinol and propidium iodide were from Sigma–Aldrich. Z-VAD-FMK, IM-54 and necrostatin-1 were purchased from Enzo Life Sciences. Hoechst 33342 was from Cell Signaling Technology. All other chemicals were from Sigma–Aldrich.

Zinc and calcium imaging

Cells grown to ∼ 60% confluency on 35 mm glass-bottomed dishes (FluoroDish™, World Precision Instruments) were loaded with 500 nM Fluozin3™-AM or Fluo4-AM in standard bath solution (SBS: NaCl 130 mM, KCl 5 mM, HEPES 10 mM, D-glucose 8 mM, MgCl2 1.2 mM, CaCl2 1.5 mM, pH 7.4) supplemented with 0.01% Pluronic® F127 at 37°C for 1 h. The loading medium was replaced with SBS and incubated for 30 min at 37°C, repeating this wash step a second time. Where required, H2O2 and drugs were included during the loading and subsequent wash steps so that the total H2O2 exposure time with and without drugs was 2 h. At the end of 2 h, the dishes were rinsed with SBS and fluorescence recorded using a DeltaVision microscope (Applied Precision) or an LSM 510 inverted confocal microscope. Cells were excited at a wavelength of 488 nm and emitted light was collected at 510 nm. Mean cellular fluorescence intensity was measured offline with ImageJ software (NIH). Data were quantified from at least three independent experiments, each performed in two dishes with three fields of views imaged.

Organelle staining

Organelles containing Zn2+ were identified by co-labelling for Zn2+ and organelles. Cells were loaded with FluoZin-3 as above and treated with markers of intracellular organelles for 30 min at 37°C: ER-Tracker™ Red (1 μM), MitoTracker® Red CMXRos (50 nM) or LysoTracker® Red DND-99 (50 nM). FluoZin-3 fluorescence was recorded as above; organelle markers were excited at 587 (ER-Tracker) and 579 (MitoTracker and LysoTracker) nm and emission collected at 615 (ER-Tacker) and 599 (MitoTracker and LysoTracker) nm respectively.

Immunofluorescence staining

HEK-293 cells, grown on coverslips, were transfected with the TRPM2–HA construct. Forty-eight hours after transfection, cells were fixed with 2% paraformaldehyde for 10 min, permeabilized with 0.25% Triton X-100/PBS for 5 min, blocked with 1% ovalbumin in PBS, stained with rat anti-HA and mouse anti-CD63 primary antibodies and labelled with Alexa Fluor 488-conjugated anti-rat and Cy3-conjugated anti-mouse secondary antibodies [23]. Images were acquired with a Zeiss LSM 700 confocal microscope using 63×/1.4 NA (numerical aperture) oil objective and excitation (Alexa Fluor 488, 494 nm; Cy3, 548 nm) and emission (Alexa Fluor 488 519 nm; Cy3, 562 nm) wavelengths.

Cell death assays

Cells were incubated with 200 μM H2O2 for 2 h with or without drugs (see figure legends) in dye-free RPMI 1640 medium. The medium was gently removed and the cells were incubated for 30 min in RPMI 1640 containing propidium iodide (Sigma, 5 μg·ml−1) and Hoechst 33342 (Cell Signaling Technology, 4 μM) and 10 mM N-acetylcysteine (Sigma) prior to imaging using an epifluorescent microscope (EVOS) fitted with a 20× lens. The number of blue (Hoechst; total cells) and red (propidium iodide; dead cells) cells were counted using Image J. Percentage cell death was calculated from the ratio of propidium iodide/Hoechst-stained cells.

Induction of diabetes with streptozotocin and glucose assays

Diabetes was induced by multiple low-dose STZ (MLDS) [24]. Eight-week-old male wild-type C57BL/6 or TRPM2−/− mice were injected intraperitoneally (i.p) with STZ prepared in 0.1 M sodium citrate buffer, pH 5.0, for five consecutive days at a dose of 40 mg·kg−1 body weight. Control mice were injected with vehicle alone. Mice were allowed free access to water and food. Blood was collected from tail veins at weekly intervals, after overnight fasting and glucose levels were determined with a blood glucose monitor (Accu-Chek® Compact Plus).

Immunohistochemistry

Five weeks after the last injection, mice were killed by cervical dislocation. Pancreata were isolated, fixed in 4% buffered paraformaldehyde, embedded in paraffin and sectioned (10 μm). Sections were mounted on slides and incubated overnight with anti-insulin and anti-glucagon antibodies in blocking buffer (3% BSA in PBS) and stained with Alexa Fluor 488- and Cy3-conjugated secondary antibodies. Images were taken with an LSM 510 inverted confocal microscope using a 40× oil-immersion objective.

Mouse islets and streptozotocin-induced cell death

Islets were isolated from wild-type and TRPM2−/− mice as described previously [25]. Briefly, pancreata were dissected from mice, minced into pieces (∼ 1 mm3) and digested with collagenase [type IA; 1 mg·ml−1 in Hanks balanced salt solution (HBSS)] for 10 min with gentle agitation at 37°C. After washing with RPMI 1640, islets were hand-picked under a dissecting microscope and incubated overnight in the RPMI 1640 medium prior to drug treatments. For toxicity assays, islets were incubated with freshly prepared STZ (2 mM) with or without drugs as appropriate. Islets were photographed after 48 h.

Data analysis

Statistical significance was determined using one-way ANOVA followed by Tukey post-hoc means comparison test or unpaired Student's t test. Statistical significance was set at P<0.05 and all error bars represent S.E.M.

β-cell apoptosis results from TRPM2-mediated rise in cytosolic Zn2+

We have investigated the mechanism by which H2O2 activation of TRPM2 channels causes β-cell death. For this, we have used the rat-derived model pancreatic β-cell line, INS1 [26]. H2O2 application (200 μM, 2 h) caused a marked increase in INS1 cell death, as measured by the uptake of propidium iodide (Figures 1A and 1B). Cell death was blocked by an inhibitor of apoptosis (Ac–DEVD), but not by inhibitors of necrosis (IM-54) and necroptosis (necrostatin-1), indicating that cell death occurred by apoptosis (Figure 1C; Supplementary Figure S1). Since current evidence suggests that β-cell death in diabetes occurs by apoptosis [1,2], results obtained from this model system are deemed relevant for the following mechanistic studies. PJ34 (10 μM) and 2-APB (50 and 150 μM), agents used to inhibit PARP and the downstream TRPM2 channel function [18,27], almost completely inhibited cell death (Figures 1A and 1B; Supplementary Figures S2, S3A and S3B), supporting a role for TRPM2 channels in β-cell apoptosis.

Previous studies attributed oxidative-stress-induced cell death to a rise in cytosolic Ca2+ [1,11,12,14,28]. A previous study however reported that TRPM2 is not only capable of conducting Ca2+, but also Zn2+ [19], raising the possibility that Zn2+ could also contribute to β-cell death. To test this possibility, we first measured the effect of H2O2 on cytosolic free Zn2+ by loading INS1 cells with FluoZin-3, a Zn2+-selective fluorescent dye (FluoZin-3 binds free Zn2+ only) [20,21]. In untreated cells, Zn2+ fluorescence was restricted to punctate structures. In H2O2 treated cells, Zn2+ staining was found throughout the cytoplasm. Using a calibration curve generated from cells loaded with different concentrations of Zn2+ using zinc pyrithione, we found that H2O2 causes an increase in Zn2+ levels from 7.66±2.71 to 134.37±40.61 nM (Supplementary Figure S4). This effect was suppressed by PJ34 and 2-APB (Figures 1D and 1E; Supplementary Figures S3C and S3D), suggesting that TRPM2 activation contributes to the rise in cytosolic Zn2+. To test whether this rise in Zn2+ contributes to cell death, we used TPEN (1 μM) [20,21] and clioquinol (10 μM) [29] at concentrations at which they chelate Zn2+, but can hardly bind Ca2+. Both reagents effectively prevented cell death (Figures 1F and 1G). We tested the role of Zn2+ in native β-cell death using pancreatic islets isolated from mice. We imposed oxidative stress by exposing the islets to STZ. The results show that TPEN, as well as PJ34, significantly prevented STZ-induced islet destruction (Figures 1H and 1I). Collectively, these results suggest that TRPM2-mediated elevation of cytosolic Zn2+ plays a significant role in pancreatic β-cell apoptosis.

The source of TRPM2-targeted Zn2+ stores is intracellular, likely lysosomes

We next asked where the free Zn2+ could be coming from. In the above experiments, Zn2+ was absent from the extracellular buffer, suggesting that Zn2+ release occurred from an intracellular source. To exclude the contribution of any contaminating Zn2+, we have included the cell-impermeable Zn2+ chelator, DTPA [30], in the extracellular buffer. DTPA failed to prevent the H2O2-induced Zn2+ response; by contrast, the cell-permeable Zn2+ chelator, TPEN, completely prevented the Zn2+ response (Figures 2A and 2B). These results confirm Zn2+ release from an intracellular source. Previous studies suggested that oxidants trigger Zn2+ release from metallothioneins, proteins that have a strong buffering capacity for cytosolic Zn2+ [20,21,31]. However, the finding that PJ34 and 2-APB are able to prevent H2O2-induced Zn2+ response (Figures 1D and 1E) suggests that Zn2+ release occurs by a specific mechanism, requiring TRPM2 channel activation. PJ34 and 2-APB however are not sufficiently selective to fully exclude Zn2+ release from metallothioneins. To address the contention, we sought to use a recombinant approach. HEK-293 cells, which do not express TRPM2 channels [8,9,32], showed no detectable Zn2+ response (Figure 2C). Induction of TRPM2 expression using a tetracycline-inducible expression system (HEK-293–TRPM2tet) [8] (Supplementary Figure S5) however resulted in a robust, 2-APB-sensitive Zn2+ rise (Figures 2D and 2E). These results exclude non-specific Zn2+ release from metallothioneins and more importantly, for the first time, demonstrate the existence of a specific mechanism for Zn2+ release involving TRPM2 channels.

In untreated cells, FluoZin-3 staining was seen in punctate structures (Figures 1D and 2C), indicating the existence of free Zn2+ in an intracellular organelle. Co-staining of cells with markers of intracellular organelles revealed extensive co-localization of Zn2+ with a marker of lysosomes (LysoTracker), but not with markers of the ER (ER-Tracker) and mitochondria (MitoTracker) (Figure 2F). Similar lysosomal co-localization was also apparent in HEK-293 cells (Figure 2G) on which Zn2+ response could be conferred by heterologous TRPM2 expression (see above). Collectively, these data suggest that significant amounts of free Zn2+ exist in lysosomes and more importantly that Zn2+ release is TRPM2-dependent.

Interplay between TRPM2-mediated Ca2+ influx and Zn2+ release underlies apoptotic cell death

H2O2 activation of TRPM2 channels raises the cytosolic levels of both Ca2+ and Zn2+. The Ca2+ rise results from its entry from the extracellular medium as well as release from lysosomes [14]. The Zn2+ rise, on the other hand, is due to intracellular release. Since both the ions are cytotoxic, we next asked: what are the relative roles of Ca2+ and Zn2+ in TRPM2-mediated apoptosis? To address this, we have used HEK-293–TRPM2tet cells, as this system allowed controlled expression of TRPM2 channels and easy detection of apoptosis. We loaded the cells with FluoZin-3 or Fluo-4 to detect changes respectively in the levels of Zn2+ and Ca2+ and stained with Hoechst to detect nuclear condensation (an indicator of apoptosis). In the absence of TRPM2 expression (−tetracycline), H2O2 (200 μM, 2 h) had no detectable effect on cytosolic Zn2+ levels and apoptosis (Figure 3A). In TRPM2-expressing cells (+tetracycline) however H2O2 caused Zn2+ rise as well as marked nuclear condensation (41.3±10.5%; n=3). Most remarkably, all cells that showed Zn2+ response displayed apoptosis; apoptosis was virtually absent from cells showing no Zn2+ response (Figures 3A and 3C).

By sharp contrast, there was no relationship between the rise in cytosolic Ca2+ levels and apoptosis (Figure 3B). In non-induced cells (−tetracycline), H2O2 caused a small rise in Ca2+, but had no effect on nuclear morphology. This small rise in Ca2+ could be attributed to activation of other Ca2+ channels or leaky expression of TRPM2 channels. In the induced cells (+tetracycline), H2O2 caused a robust increase in cytosolic Ca2+ in almost all cells. However, nuclear condensation was limited to ∼ 40% of these cells (Figure 3C). These results suggest that this TRPM2-mediated rise in Ca2+ alone is not sufficient to induce apoptosis, but a rise in Zn2+ is indispensable. Consistent with this argument, PJ34 and 2-APB prevented both Zn2+ rise and nuclear condensation, whereas chelation of Zn2+ with TPEN (1 μM) inhibited nuclear condensation (Figures 3D and 3E; Supplementary Figures 3E and 3F). To confirm that at this concentration, TPEN chelates Zn2+ only, we have tested the ability of the chelator to quench fluorescence resulting from metal ion binding to Fluo-4 and FluoroZin-3 (Figures 3F and 3G). TPEN caused a small, but significant, decrease in Fluo-4 fluorescence, a decrease that could be attributed to the exceptional ability of Fluo-4 to bind Zn2+, co-released [20,21]. The TPEN-insensitive Fluo-4 signal can be attributed to Ca2+ binding to the probe. By contrast, TPEN, as expected, almost completely prevented the FluoZin-3 signal, which is due to binding of Zn2+ only. Taken together, these results indicate that the rise in Ca2+ alone cannot induce apoptosis in the absence of Zn2+.

If Zn2+ release is the principal cause of apoptosis, what then is the role of Ca2+? There is overwhelming evidence in the literature that Ca2+ plays a significant role in apoptosis [11,28]. We suspected that Ca2+ could play an indirect role because Ca2+ is an activator of TRPM2 channels [32,33]. Consistent with this expectation, removal of extracellular Ca2+ resulted in significant attenuation of H2O2-induced Zn2+ release (Figures 3H and 3I). This manoeuvre, however, does not prevent intracellular Ca2+ release, so we do not know the extent to which intracellular Ca2+ release has augmented Zn2+ release. Unfortunately, there are no known reagents capable of chelating Ca2+ without chelating Zn2+ to address this problem. Nevertheless, our data uncover an important mechanism whereby TRPM2 channels orchestrate interplay between Ca2+ and Zn2+ in promoting cell death.

TRPM2-deficient mice are protected from MLDS-induced pancreatic β-cell loss and hyperglycaemia

We next investigated the in vivo role of TRPM2 in pancreatic β-cell death. For this, we have used the MLDS mouse model [34,35]. This model is thought to closely resemble human T1D because MLDS causes β-cell death by promoting DNA breakdown, PARP activation, T-cell-dependent immune reaction and cytokine-induced oxidative stress [34]. We injected wild-type and TRPM2-deficient mice (TRPM2−/−) [17,22] with MLDS or vehicle and followed changes in blood glucose levels as a marker of β-cell death. The results (Figure 4A) show that, unlike the wild-type mice, TRPM2-deficient mice are markedly resistant to MLDS-induced hyperglycaemia. Eight of 16 TRPM2−/− mice showed little change in glycaemia, i.e. glucose levels were similar to pre-STZ treatment levels; the rest of the animals however showed varying degrees of increase in glucose levels. By contrast, all wild-type mice (16 out of 16) showed highly elevated levels of glucose. Immunostaining of pancreata showed marked loss of insulin-positive cells in STZ-injected wild-type mice, but not the corresponding TRPM2-deficient mice (Figure 4B). Consistent with the in vivo data, pancreatic islets isolated from wild-type mice are destroyed by ex vivo STZ treatment (2 mM, 48 h), whereas those from TRPM2−/− mice are completely resistant to STZ (Figures 4C and 4D). These data indicate that TRPM2 plays a crucial role in β-cell death and development of T1D phenotype in the mouse model.

Taken together with the published reports, our data support the mechanism outlined in Figure 4(E), where oxidant-induced DNA breakdown leads to β-cell death via a cascade of events that include PARP–PARG activation, ADP-ribose production, activation of TRPM2 channels and intracellular Zn2+ release. The physiological consequence of this chain of events is hyperglycaemia. Results of the present study provide new insights into the mechanisms of β-cell death and address some of the uncertainties surrounding β-cell death.

Activation of PARP–PARG pathway utilizes NAD and generates ADP-ribose. NAD depletion and ADP-ribose both cause β-cell death, but by fundamentally different mechanisms. By creating energy (ATP) deficiency, NAD depletion is thought to cause necrosis [57]. ADP-ribose, on the other hand, leads to apoptosis by activating the TRPM2 channels [11,12]. Although oxidative stress can cause cell death via both necrosis and apoptosis, immune-mediated β-cell death appears to occur largely by apoptosis [1,2]. For this reason, we chose experimental paradigms that promote β-cell death predominantly by apoptosis. In in vitro experiments, we used modest concentrations of H2O2 to ensure that cell death occurred largely by apoptosis. Under these conditions, pharmacological inhibition of TRPM2 channels prevented INS1 β-cell apoptosis (Figures 1A and 1C). These effects could be replicated in HEK-293 cells through heterologous expression of TRPM2 channels (Figure 3A). For in vivo experiments, we have used the MLDS mouse model of T1D, which is also thought to cause cell death by apoptosis [34]. Using this model, we demonstrated that knockout of TRPM2 channels prevents β-cell death and hyperglycaemia (Figures 4A–4D). Thus both in vitro and in vivo evidence indicates that activation of TRPM2 channels is essential for apoptotic cell death. Interestingly, these findings exclude NAD depletion as the primary cause of β-cell death because TRPM2 activation occurs downstream of NAD depletion (Figure 4E). We cannot, however, exclude the possibility that NAD depletion and necrosis occur at high (non-physiological) concentrations of oxidants.

Apoptosis occurs by two main pathways: the extrinsic or cell death receptor pathway and the intrinsic or mitochondrial pathway [36]. The former involves the tumour necrosis factor superfamily of receptors and activation of the initiator caspase, caspase 8. The latter involves release of pro-apoptotic factors from mitochondria and activation of the initiator caspase, caspase 9. There is some degree of cross-talk between the two pathways [36,37], with caspase 8 affecting the mitochondrial pathway. Both pathways ultimately lead to the activation of executioner caspases, including caspase 3, which are responsible for the final steps of apoptosis [37]. Knockout of caspases 3 and 8 has been shown to protect mice from MLDS-induced islet destruction and hyperglycaemia [38,39]. Up-regulation of ROS is a common feature of both pathways [2,36,37]. Indeed, a number of antioxidants have been shown to inhibit β-cell death mediated by both pathways [40,41]. It therefore seems likely that ROS and, hence, the TRPM2 channel act upstream of caspases. Such interpretation is consistent with our finding that TRPM2 knockout mice are resistant to MLDS-induced β-cell death. Precisely where within the apoptotic pathways TRPM2 channels play a role remains to be determined.

MLDS has been shown to cause selective destruction of β-cells, which turn induces immune response against islets, resulting in insulitis [34]. Although not described in β-cells, a role for TRPM2 has been described in the inflammation through up-regulation of chemokines [42]. Inhibition of TRPM2 channels indeed prevents inflammation in a mouse model of colitis and inflammatory and neuropathic pain [42,43]. Thus TRPM2 channels could also contribute to MLDS-induced insulitis.

TRPM2 channel activation leads to a rise in both Ca2+ and Zn2+ in the cytoplasm. Since both ions are implicated in apoptosis, we asked: which ion plays a dominant role? Our evidence suggests that Zn2+, rather than Ca2+, plays the main role. First, sequestration of Zn2+ with chelating agents prevented oxidant-induced β-cell death (Figures 1F and 1G) and islet destruction (Figures 1H and 1I). Secondly, in HEK-293 cells engineered to express TRPM2 channels, virtually all cells that displayed a rise in Zn2+ showed apoptotic phenotype (Figures 3A and 3C). By sharp contrast, not all cells that showed a rise in Ca2+ exhibited apoptotic phenotype (Figures 3B and 3C). Finally, when we prevented the rise in Zn2+ with TPEN, leaving the Ca2+ rise unaffected, no apoptosis was apparent (Figures 3D–3G). These data provide compelling evidence that rise in Ca2+ alone is not enough, but a rise in Zn2+ is indispensable for cell death. Although Zn2+ appears to play the primary role, the rise in Ca2+ is also important because in the absence of extracellular Ca2+ entry, the Zn2+ rise was significantly attenuated (Figures 3H and 3I). Thus our data indicate an interesting interplay between Ca2+ and Zn2+ in cell death.

We next examined the source of oxidant-sensitive Zn2+. Free Zn2+ is highly toxic to cells. For this reason, cytoplasmic Zn2+ is either buffered by metallothioneins or sequestered into organelles [20,21]. It has been widely reported that oxidative stress triggers Zn2+ release from metallothioneins through non-specific protein oxidation [20,21,31]. However, in the absence of TRPM2 channel function, with 200 μM H2O2 (2 h), we failed to detect Zn2+ release (Figures 1D, 1E, 2D and 2E). Our data thus exclude Zn2+ release from metallothioneins at modest concentrations of oxidants. More importantly, they demonstrate the existence of a specific mechanism for release of cytotoxic Zn2+ in the form of TRPM2 channels. A search for other sources revealed copious amounts of free Zn2+ in lysosomes (Figure 2F). Although we have not demonstrated Zn2+ release from lysosomes directly, the fact that TRPM2 channels are expressed in lysosomes [14,44] (Supplementary Figure S6) and mediate lysosomal Ca2+ release [14] strongly suggest that lysosomal Zn2+ probably exits via TRPM2 channels. Although lysosomes express ZIP members of Zn2+ transporters (ZIP3 and ZIP8) [45], the finding that H2O2 failed to trigger Zn2+ release in HEK-293 cells in the absence of TRPM2 channel expression (Figures 2D and 2E) indicates that expression of TRPM2 channels alone is enough to cause Zn2+ release. However, we cannot fully exclude functional coupling between TRPM2 channels and ZIPs. Since extracellular Ca2+ entry potentiates ROS-induced Zn2+ release and Ca2+ is an activator of TRPM2 channels [32,33], we propose that extracellular Ca2+ enters through the plasma membrane TRPM2 channels and co-stimulates (together with H2O2) lysosomal TRPM2 channels to promote Zn2+ release (Figure 4E). Although TRPM2-channel-mediated Zn2+ influx could be demonstrated in HEK-293 cells using FluoZin-3 [19], our attempts to demonstrate Zn2+ permeation by electrophysiology have been unsuccessful [46], largely because concentration of Zn2+ required to demonstrate Zn2+ currents were found to inhibit the channel profoundly.

Whole animal studies support a role for Zn2+ in the patho-physiology of T1D. First, excess dietary Zn2+ supplementation potentiates the incidence of diabetes and mortality in non-obese diabetic mice (genetic models of T1D), whereas dietary Zn2+ restriction has the opposite effect [47]. Secondly, Zn2+ chelators (calcium EDTA and clioquinol) reduce STZ-induced β-cell death and hyperglycaemia [48,49]. Finally, knockout of zinc transporter 5 (Zn5T) reduced β-cell Zn2+ levels and improved β-cell mass and glucose homoeostasis in non-obese diabetic mice [47]. Thus there is significant in vivo evidence in support of a role for Zn2+ in the patho-physiology of T1D. Taken together with the present study, we conclude that β-cell death in animal models of T1D is caused by TRPM2-channel-mediated rise in intracellular Zn2+ and that TRPM2 is a promising target for prevention of T1D.

Besides the effect on β-cell viability, TRPM2 channels have also been implicated in insulin secretion and peripheral insulin resistance. Uchida et al. [15] reported that TRPM2 knockout mice exhibit slightly elevated basal insulin, impaired glucose tolerance, but normal fasting blood glucose levels. Zhang et al. [17], by contrast, reported that TRPM2 knockout improves glucose tolerance (due to improved insulin sensitivity), but has no effect on basal glucose or insulin levels. The reason for the discrepancy between the two reports is unclear, but the finding that TRPM2 knockout improves insulin sensitivity suggests that TRPM2 channels could also be a potential target for T2D.

In summary, our in vitro and in vivo studies provide novel insights into how oxidative stress causes pancreatic β-cell death and hyperglycaemia. The mechanism involves activation of TRPM2 channels, entry of extracellular Ca2+, Ca2+-potentiated Zn2+ release and, ultimately, Zn2+-induced β-cell apoptosis. Whether a similar mechanism underlies cell death in other degenerative diseases remains to be investigated.

2-APB

2-aminoethoxydiphenyl borate

DTPA

diethylenetriaminepenta-acetate

HA

haemagglutinin

HEK

human embryonic kidney

i.p.

intraperitoneally

MLDS

multiple low-dose streptozotocin

PARG

poly(ADP-ribose) glycohydrolase

PARP

poly(ADP-ribose) polymerase

ROS

reactive oxygen species

SBS

standard bath solution

STZ

streptozotocin

T1D

Type 1 diabetes

TPEN

N,N,N′,N′-tetrakis(2-pyridylmethyl)ethylenediamine

TRPM2

transient receptor potential (melastatin) 2

Asipu Sivaprasadarao and Paul Manna conceived the study. Paul Manna, Tim Munsey, Nada Abuarab, Fangfang Li, Aruna Asipu, Gareth Howell, Asipu Sivaprasadarao, Wei Yang and Alicia Sedo performed the experiments and analysed data. Jacqui Naylor contributed to the design of the experiments. Lin-Hua Jiang and David Beech provided the intellectual input into the design of several experiments and contributed to the writing of the manuscript. Asipu Sivaprasadarao wrote the manuscript.

This work was supported by the Medical Research Council [grant number G0802050]; the British Heart Foundation [grant number PG/10/68/28528]; Alzheimer Research U.K. [grant number ART/PPG2009A/2]; and the King Saud University for Health Sciences, Ministry of Higher Education for Saudi Arabia [Studentship] to N.A.

1
Cnop
 
M.
Welsh
 
N.
Jonas
 
J.C.
Jorns
 
A.
Lenzen
 
S.
Eizirik
 
D.L.
 
Mechanisms of pancreatic beta-cell death in type 1 and type 2 diabetes: many differences, few similarities
Diabetes
2005
, vol. 
54
 
Suppl 2
(pg. 
S97
-
S107
)
[PubMed]
2
Mathis
 
D.
Vence
 
L.
Benoist
 
C.
 
Beta-cell death during progression to diabetes
Nature
2001
, vol. 
414
 (pg. 
792
-
798
)
[PubMed]
3
Atkinson
 
M.A.
Eisenbarth
 
G.S.
Michels
 
A.W.
 
Type 1 diabetes
Lancet
2014
, vol. 
383
 (pg. 
69
-
82
)
[PubMed]
4
Krishnakumar
 
R.
Kraus
 
W.L.
 
The PARP side of the nucleus: molecular actions, physiological outcomes, and clinical targets
Mol. Cell
2010
, vol. 
39
 (pg. 
8
-
24
)
[PubMed]
5
Pieper
 
A.A.
Brat
 
D.J.
Krug
 
D.K.
Watkins
 
C.C.
Gupta
 
A.
Blackshaw
 
S.
Verma
 
A.
Wang
 
Z.Q.
Snyder
 
S.H.
 
Poly(ADP-ribose) polymerase-deficient mice are protected from streptozotocin-induced diabetes
Proc. Natl. Acad. Sci. U.S.A.
1999
, vol. 
96
 (pg. 
3059
-
3064
)
[PubMed]
6
Charron
 
M.J.
Bonner-Weir
 
S.
 
Implicating PARP and NAD+ depletion in type I diabetes
Nat. Med.
1999
, vol. 
5
 (pg. 
269
-
270
)
[PubMed]
7
Burkart
 
V.
Wang
 
Z.Q.
Radons
 
J.
Heller
 
B.
Herceg
 
Z.
Stingl
 
L.
Wagner
 
E.F.
Kolb
 
H.
 
Mice lacking the poly(ADP-ribose) polymerase gene are resistant to pancreatic beta-cell destruction and diabetes development induced by streptozocin
Nat. Med.
1999
, vol. 
5
 (pg. 
314
-
319
)
[PubMed]
8
Perraud
 
A.L.
Fleig
 
A.
Dunn
 
C.A.
Bagley
 
L.A.
Launay
 
P.
Schmitz
 
C.
Stokes
 
A.J.
Zhu
 
Q.
Bessman
 
M.J.
Penner
 
R.
, et al 
ADP-ribose gating of the calcium-permeable LTRPC2 channel revealed by Nudix motif homology
Nature
2001
, vol. 
411
 (pg. 
595
-
599
)
[PubMed]
9
Hara
 
Y.
Wakamori
 
M.
Ishii
 
M.
Maeno
 
E.
Nishida
 
M.
Yoshida
 
T.
Yamada
 
H.
Shimizu
 
S.
Mori
 
E.
Kudoh
 
J.
, et al 
LTRPC2 Ca2+-permeable channel activated by changes in redox status confers susceptibility to cell death
Mol. Cell
2002
, vol. 
9
 (pg. 
163
-
173
)
[PubMed]
10
Miller
 
B.A.
 
Inhibition of TRPM2 function by PARP inhibitors protects cells from oxidative stress-induced death
Br. J. Pharmacol.
2004
, vol. 
143
 (pg. 
515
-
516
)
[PubMed]
11
Takahashi
 
N.
Kozai
 
D.
Kobayashi
 
R.
Ebert
 
M.
Mori
 
Y.
 
Roles of TRPM2 in oxidative stress
Cell Calcium
2011
, vol. 
50
 (pg. 
279
-
287
)
[PubMed]
12
Sumoza-Toledo
 
A.
Penner
 
R.
 
TRPM2: a multifunctional ion channel for calcium signalling
J. Physiol.
2011
, vol. 
589
 (pg. 
1515
-
1525
)
[PubMed]
13
Kashio
 
M.
Sokabe
 
T.
Shintaku
 
K.
Uematsu
 
T.
Fukuta
 
N.
Kobayashi
 
N.
Mori
 
Y.
Tominaga
 
M.
 
Redox signal-mediated sensitization of transient receptor potential melastatin 2 (TRPM2) to temperature affects macrophage functions
Proc. Natl. Acad. Sci. U.S.A.
2012
, vol. 
109
 (pg. 
6745
-
6750
)
[PubMed]
14
Lange
 
I.
Yamamoto
 
S.
Partida-Sanchez
 
S.
Mori
 
Y.
Fleig
 
A.
Penner
 
R.
 
TRPM2 functions as a lysosomal Ca2+−release channel in beta cells
Sci. Signal.
2009
, vol. 
2
 pg. 
ra23
 
[PubMed]
15
Uchida
 
K.
Dezaki
 
K.
Damdindorj
 
B.
Inada
 
H.
Shiuchi
 
T.
Mori
 
Y.
Yada
 
T.
Minokoshi
 
Y.
Tominaga
 
M.
 
Lack of TRPM2 impaired insulin secretion and glucose metabolisms in mice
Diabetes
2011
, vol. 
60
 (pg. 
119
-
126
)
[PubMed]
16
Togashi
 
K.
Hara
 
Y.
Tominaga
 
T.
Higashi
 
T.
Konishi
 
Y.
Mori
 
Y.
Tominaga
 
M.
 
TRPM2 activation by cyclic ADP-ribose at body temperature is involved in insulin secretion
EMBO J.
2006
, vol. 
25
 (pg. 
1804
-
1815
)
[PubMed]
17
Zhang
 
Z.
Zhang
 
W.
Jung
 
D.Y.
Ko
 
H.J.
Lee
 
Y.
Friedline
 
R.H.
Lee
 
E.
Jun
 
J.
Ma
 
Z.
Kim
 
F.
, et al 
TRPM2 Ca2+ channel regulates energy balance and glucose metabolism
Am. J. Physiol. Endocrin. Metab.
2012
, vol. 
302
 (pg. 
E807
-
E816
)
18
Fonfria
 
E.
Marshall
 
I.C.
Benham
 
C.D.
Boyfield
 
I.
Brown
 
J.D.
Hill
 
K.
Hughes
 
J.P.
Skaper
 
S.D.
McNulty
 
S.
 
TRPM2 channel opening in response to oxidative stress is dependent on activation of poly(ADP-ribose) polymerase
Br. J. Pharmacol.
2004
, vol. 
143
 (pg. 
186
-
192
)
[PubMed]
19
Yu
 
P.
Wang
 
Q.
Zhang
 
L.H.
Lee
 
H.C.
Zhang
 
L.
Yue
 
J.
 
A cell permeable NPE caged ADP-ribose for studying TRPM2
PLoS One
2012
, vol. 
7
 pg. 
e51028
 
[PubMed]
20
Sensi
 
S.L.
Paoletti
 
P.
Bush
 
A.I.
Sekler
 
I.
 
Zinc in the physiology and pathology of the CNS
Nat. Rev. Neurosci.
2009
, vol. 
10
 (pg. 
780
-
791
)
[PubMed]
21
Shuttleworth
 
C.W.
Weiss
 
J.H.
 
Zinc: new clues to diverse roles in brain ischemia
Trends Pharmacol. Sci.
2011
, vol. 
32
 (pg. 
480
-
486
)
[PubMed]
22
Zou
 
J.
Ainscough
 
J.F.
Yang
 
W.
Sedo
 
A.
Yu
 
S.P.
Mei
 
Z.Z.
Sivaprasadarao
 
A.
Beech
 
D.J.
Jiang
 
L.H.
 
A differential role of macrophage TRPM2 channels in Ca2+ signaling and cell death in early responses to H2O2
Am. J. Physiol. Cell Physiol.
2013
, vol. 
305
 (pg. 
C61
-
C69
)
[PubMed]
23
Manna
 
P.T.
Smith
 
A.J.
Taneja
 
T.K.
Howell
 
G.J.
Lippiat
 
J.D.
Sivaprasadarao
 
A.
 
Constitutive endocytic recycling and protein kinase C-mediated lysosomal degradation control KATP channel surface density
J. Biol. Chem.
2010
, vol. 
285
 (pg. 
5963
-
5973
)
[PubMed]
24
Wu
 
K.K.
Huan
 
Y.
 
Enna
 
S.J.
Williams
 
M.
Porsolt
 
R.E.
Sullivan
 
J.P.
Ferkany
 
J.W.
Kenakin
 
T.
 
Streptozotocin-induced diabetic models in mice and rats
Current Protocols In Pharmacology
2008
Chichester
John Wiley & Sons
25
Li
 
D.S.
Yuan
 
Y.H.
Tu
 
H.J.
Liang
 
Q.L.
Dai
 
L.J.
 
A protocol for islet isolation from mouse pancreas
Nat. Protoc.
2009
, vol. 
4
 (pg. 
1649
-
1652
)
[PubMed]
26
Asfari
 
M.
Janjic
 
D.
Meda
 
P.
Li
 
G.
Halban
 
P.A.
Wollheim
 
C.B.
 
Establishment of 2-mercaptoethanol-dependent differentiated insulin-secreting cell lines
Endocrinology
1992
, vol. 
130
 (pg. 
167
-
178
)
[PubMed]
27
Togashi
 
K.
Inada
 
H.
Tominaga
 
M.
 
Inhibition of the transient receptor potential cation channel TRPM2 by 2-aminoethoxydiphenyl borate (2-APB)
Br. J. Pharmacol.
2008
, vol. 
153
 (pg. 
1324
-
1330
)
[PubMed]
28
Ermak
 
G.
Davies
 
K.J.
 
Calcium and oxidative stress: from cell signaling to cell death
Mol. Immunol.
2002
, vol. 
38
 (pg. 
713
-
721
)
[PubMed]
29
Kaur
 
D.
Yantiri
 
F.
Rajagopalan
 
S.
Kumar
 
J.
Mo
 
J.Q.
Boonplueang
 
R.
Viswanath
 
V.
Jacobs
 
R.
Yang
 
L.
Beal
 
M.F.
, et al 
Genetic or pharmacological iron chelation prevents MPTP-induced neurotoxicity in vivo: a novel therapy for Parkinson's disease
Neuron
2003
, vol. 
37
 (pg. 
899
-
909
)
[PubMed]
30
Yamasaki
 
S.
Sakata-Sogawa
 
K.
Hasegawa
 
A.
Suzuki
 
T.
Kabu
 
K.
Sato
 
E.
Kurosaki
 
T.
Yamashita
 
S.
Tokunaga
 
M.
Nishida
 
K.
Hirano
 
T.
 
Zinc is a novel intracellular second messenger
J. Cell Biol.
2007
, vol. 
177
 (pg. 
637
-
645
)
[PubMed]
31
Bossy-Wetzel
 
E.
Schwarzenbacher
 
R.
Lipton
 
S.A.
 
Molecular pathways to neurodegeneration
Nat. Med.
2004
, vol. 
10
 
Suppl
(pg. 
S2
-
S9
)
[PubMed]
32
McHugh
 
D.
Flemming
 
R.
Xu
 
S.Z.
Perraud
 
A.L.
Beech
 
D.J.
 
Critical intracellular Ca2+ dependence of transient receptor potential melastatin 2 (TRPM2) cation channel activation
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
11002
-
11006
)
[PubMed]
33
Du
 
J.
Xie
 
J.
Yue
 
L.
 
Intracellular calcium activates TRPM2 and its alternative spliced isoforms
Proc. Natl. Acad. Sci. U.S.A.
2009
, vol. 
106
 (pg. 
7239
-
7244
)
[PubMed]
34
Wu
 
K.K.
Huan
 
Y.
 
Diabetic atherosclerosis mouse models
Atherosclerosis.
2007
, vol. 
191
 (pg. 
241
-
249
)
[PubMed]
35
Like
 
A.A.
Rossini
 
A.A.
 
Streptozotocin-induced pancreatic insulitis: new model of diabetes mellitus
Science
1976
, vol. 
193
 (pg. 
415
-
417
)
[PubMed]
36
Choi
 
D.
Woo
 
M.
 
Executioners of apoptosis in pancreatic β-cells: not just for cell death
Am. J. Physiol. Endocrinol. Metab.
2010
, vol. 
298
 (pg. 
E735
-
E741
)
[PubMed]
37
Circu
 
M.L.
Aw
 
T.Y.
 
Reactive oxygen species, cellular redox systems, and apoptosis
Free Radic. Biol. Med.
2010
, vol. 
48
 (pg. 
749
-
762
)
[PubMed]
38
Liadis
 
N.
Salmena
 
L.
Kwan
 
E.
Tajmir
 
P.
Schroer
 
S.A.
Radziszewska
 
A.
Li
 
X.
Sheu
 
L.
Eweida
 
M.
Xu
 
S.
Gaisano
 
H.Y.
Hakem
 
R.
Woo
 
M.
 
Distinct in vivo roles of caspase-8 in beta-cells in physiological and diabetes models
Diabetes
2007
, vol. 
56
 (pg. 
2302
-
2311
)
[PubMed]
39
Liadis
 
N.
Murakami
 
K.
Eweida
 
M.
Elford
 
A.R.
Sheu
 
L.
Gaisano
 
H.Y.
Hakem
 
R.
Ohashi
 
P.S.
Woo
 
M.
 
Caspase-3-dependent beta-cell apoptosis in the initiation of autoimmune diabetes mellitus
Mol. Cell. Biol.
2005
, vol. 
25
 (pg. 
3620
-
3629
)
[PubMed]
40
Gurgul
 
E.
Lortz
 
S.
Tiedge
 
M.
Jorns
 
A.
Lenzen
 
S.
 
Mitochondrial catalase overexpression protects insulin-producing cells against toxicity of reactive oxygen species and proinflammatory cytokines
Diabetes
2004
, vol. 
53
 (pg. 
2271
-
2280
)
[PubMed]
41
Lim
 
S.
Rashid
 
M.A.
Jang
 
M.
Kim
 
Y.
Won
 
H.
Lee
 
J.
Woo
 
J.T.
Kim
 
Y.S.
Murphy
 
M.P.
Ali
 
L.
, et al 
Mitochondria-targeted antioxidants protect pancreatic β-cells against oxidative stress and improve insulin secretion in glucotoxicity and glucolipotoxicity
Cell. Physiol. Biochem.
2011
, vol. 
28
 (pg. 
873
-
886
)
[PubMed]
42
Yamamoto
 
S.
Shimizu
 
S.
Kiyonaka
 
S.
Takahashi
 
N.
Wajima
 
T.
Hara
 
Y.
Negoro
 
T.
Hiroi
 
T.
Kiuchi
 
Y.
Okada
 
T.
, et al 
TRPM2-mediated Ca2+ influx induces chemokine production in monocytes that aggravates inflammatory neutrophil infiltration
Nat. Med.
2008
, vol. 
14
 (pg. 
738
-
747
)
[PubMed]
43
Haraguchi
 
K.
Kawamoto
 
A.
Isami
 
K.
Maeda
 
S.
Kusano
 
A.
Asakura
 
K.
Shirakawa
 
H.
Mori
 
Y.
Nakagawa
 
T.
Kaneko
 
S.
 
TRPM2 contributes to inflammatory and neuropathic pain through the aggravation of pronociceptive inflammatory responses in mice
J. Neurosci.
2012
, vol. 
32
 (pg. 
3931
-
3941
)
[PubMed]
44
Sumoza-Toledo
 
A.
Lange
 
I.
Cortado
 
H.
Bhagat
 
H.
Mori
 
Y.
Fleig
 
A.
Penner
 
R.
Partida-Sanchez
 
S.
 
Dendritic cell maturation and chemotaxis is regulated by TRPM2-mediated lysosomal Ca2+ release
FASEB J.
2011
, vol. 
25
 (pg. 
3529
-
3542
)
[PubMed]
45
Jeong
 
J.
Eide
 
D.J.
 
The SLC39 family of zinc transporters
Mol. Aspects Med.
2013
, vol. 
34
 (pg. 
612
-
619
)
[PubMed]
46
Yang
 
W.
Manna
 
P.T.
Zou
 
J.
Luo
 
J.
Beech
 
D.J.
Sivaprasadarao
 
A.
Jiang
 
L.H.
 
Zinc inactivates melastatin transient receptor potential 2 channels via the outer pore
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
23789
-
23798
)
[PubMed]
47
Sheline
 
C.T.
Shi
 
C.
Takata
 
T.
Zhu
 
J.
Zhang
 
W.
Sheline
 
P.J.
Cai
 
A.L.
Li
 
L.
 
Dietary zinc reduction, pyruvate supplementation, or zinc transporter 5 knockout attenuates β-cell death in nonobese diabetic mice, islets, and insulinoma cells
J. Nutr.
2012
, vol. 
142
 (pg. 
2119
-
2127
)
[PubMed]
48
Kim
 
B.J.
Kim
 
Y.H.
Kim
 
S.
Kim
 
J.W.
Koh
 
J.Y.
Oh
 
S.H.
Lee
 
M.K.
Kim
 
K.W.
Lee
 
M.S.
 
Zinc as a paracrine effector in pancreatic islet cell death
Diabetes
2000
, vol. 
49
 (pg. 
367
-
372
)
[PubMed]
49
Priel
 
T.
Aricha-Tamir
 
B.
Sekler
 
I.
 
Clioquinol attenuates zinc-dependent beta-cell death and the onset of insulitis and hyperglycemia associated with experimental type I diabetes in mice
Eur. J. Pharmacol.
2007
, vol. 
565
 (pg. 
232
-
239
)
[PubMed]
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC-BY) (http://creativecommons.org/licenses/by/3.0/) which permits unrestricted use, distribution and reproduction in any medium, provided the original work is properly cited.

Supplementary data