Brain is highly demanding for metabolism and energy and thus is prone to damage by ischemia, if it lasts long or becomes severe, and also by reperfusion after transient ischemia (McCord, 1985; Kalyanaraman, 2013). Hippocampus is crucial for brain functions such as learning and memory, and studies of both animal models of ischemia-reperfusion and ischemic stroke patients have documented that hippocampal neuronal death during reperfusion plays an important role in progressive decline in the cognitive function after ischemia stroke (Kitagawa et al., 1990; Doyle et al., 2008). Reperfusion-induced brain damage is strongly and causatively related to generation of excessive reactive oxygen species (ROS) upon re-introduction of oxygen molecule during reperfusion (McCord, 1985; Chan, 2001; Uttara et al., 2009; Chen et al., 2011; Sanderson et al., 2013). To gain a better understanding of ROS-induced neuronal death is therefore critical in developing therapeutics targeting reperfusion-related neuronal death, which is currently non-existent, to ameliorate post-ischemia cognitive dysfunction.

It is known that ROS cause DNA damage and the poly(ADP-ribose) polymerase-1 (PARP-1) activity is critical in the process of repairing DNA damage (Dantzer et al., 2006). However, it is also well recognized that prolonged activation or hyper-activation of PARP-1 induces cell death, including neurotoxicity (Cole and Perez-Polo, 2002; Lo et al., 2003; Alano et al., 2010; Virág et al., 2013). ADP-ribose (ADPR), long known as a biochemical by-product of the PARP-1-dependent DNA repair process, has been relatively recently established as an intracellular signaling molecule due to its capacity of selectively gating the transient receptor potential melastatin-related 2 (TRPM2) Ca²⁺-permeable cationic channel (Perraud et al., 2001; Sano et al., 2001). An early study showed that activation of the TRPM2 channel mediates ROS-induced cell death (Hara et al., 2002) and recent studies using transgenic TRPM2-knockout (TRPM2-KO) mice have revealed a critical role for ROS-induced TRPM2-mediated cell death in diverse cell types that contributes to pathologies such as diabetes (Manna et al., 2015), ischemic kidney damage (Gao et al., 2014) and paracetamol-induced liver injury (Kheradpezhouh et al., 2014). In the brain, it has been shown that the TRPM2 channel is expressed in hippocampal (Olah et al., 2009; Verma et al., 2012; Ye et al., 2014), cortical (Kaneko et al., 2006), striatal (Fonfria et al., 2005; Kaneko et al., 2006) and dopaminergic neurons (Sun et al., 2016) as well as microglial cells (Kraft et al., 2004; Fonfria et al., 2006; Mortadza et al., 2017), neurovascular endothelial cells (Park et al., 2014) and pericytes (Jiang et al., 2017). Previous studies support a role for the TRPM2 channel, particularly TRPM2-mediated increase in the cytosolic Ca²⁺ concentration ([Ca²⁺]_c), in ROS-induced neuronal death (Fonfria et al., 2005; Kaneko et al., 2006; Verma et al., 2012). Furthermore, TRPM2-dependent neuronal death has been related to post-ischemia brain damage (Jia et al., 2011; Nakayama et al., 2013; Shimizu et al., 2013, 2016; Ye et al., 2014) and Alzheimer’s disease (Ostapchenko et al., 2015). The molecular mechanisms underlying ROS-induced TRPM2-dependent neuronal death however remain less well-defined.

Zn²⁺ is a trace metal ion that is biologically important, serving an essential enzyme cofactor and transcription factor, but Zn²⁺ is also well-known for being neurotoxic. In agreement with the finding from a recent study that the TRPM2 channel selectively mediates brain damage induced by reperfusion, not by ischemia (Alim et al., 2013), our recent study has revealed an exclusive role for the TRPM2 channel in sustaining the cytosolic Zn²⁺ concentration ([Zn²⁺]_c) during reperfusion that is causatively associated with post-ischemia neuronal death (Ye et al., 2014), implying TRPM2-dependent rise in the [Zn²⁺]_c is critical in driving ROS-induced neuronal death. As has been well recognized, Zn²⁺-induced neuronal death results in significant part from the potency of cytosolic Zn²⁺ in inducing lysosomal and mitochondrial dysfunction (Jiang et al., 2001; Dineley et al., 2003, 2005; Hwang et al., 2008; Medvedeva et al., 2009; Sensi et al., 2011; Shuttleworth and Weiss, 2011). Therefore, in the present study, we performed experiments using cultured hippocampal neurons to investigate the contribution of TRPM2 channel, particularly TRPM2-dependent alterations in the intracellular Zn²⁺ homeostasis, lysosomal function and mitochondrial function, in ROS-induced neuronal death.

Here we provide evidence to show that a significant but not exclusive role of the TRPM2 channel in ROS-induced neuronal death and reveal TRPM2-dependent dynamic alterations in the intracellular Zn²⁺ homeostasis, lysosomal and mitochondrial functions that are important in ROS-induced neuronal death.

Previous studies using cortical, striatal and hippocampal neurons (Fonfria et al., 2005; Kaneko et al., 2006; Verma et al., 2012) support a role for the TRPM2 channel in H₂O₂-induced neuronal death, but the exact contribution and the underlying molecular mechanisms remained not well-defined. Here we presented genetic and pharmacological results to confirm the previous findings (Figure 1A and Supplementary Figures S1, S2). More specifically, our results indicate that the TRPM2 channel has a significant but not exclusive role in mediating ROS-induced neuronal death. Furthermore, we provide several lines of evidence to support that TRPM2-dependent alteration in the intracellular Zn²⁺ homeostasis is critical in ROS-induced hippocampal neuronal death. First of all, H₂O₂-induced neuronal death was strongly inhibited by TPEN at a concentration that specifically acts as a Zn²⁺ chelator (Figure 1F). Second, single live cell imaging showed that exposure to H₂O₂ for 30 min evoked a prominent increase in the [Zn²⁺]_c (Figure 2) or alteration in the intracellular Zn²⁺ homeostasis in neurons (Figures 3, 4). Immunostaining showed that such exposure to H₂O₂ resulted in no discernible effect on the overall expression of TRPM2 protein, but appeared to slightly change its subcellular location (Supplementary Figure S4), and further study may help to clarify the implication of such a change. It is well recognized that an increase in the [Zn²⁺]_c, as a result from extracellular Zn²⁺ influx or internal Zn²⁺ release through Zn²⁺-specific transporters as well as diverse Ca²⁺-transporting mechanisms, can induce neuronal death (Colvin et al., 2003; Medvedeva et al., 2009; Sensi et al., 2011; Shuttleworth and Weiss, 2011; Li et al., 2015). In the present study, we demonstrated a critical role for the TRPM2 channel in ROS-induced increase in the [Zn²⁺]_c (Figure 2 and Supplementary Figure S3) or alteration in the intracellular Zn²⁺ homeostasis (Figures 3, 4). Finally, inhibition of such Zn²⁺ signaling significantly attenuated ROS-induced neuronal death (Figure 1F). Considered that reperfusion stimulates generation of excessive ROS (McCord, 1985; Kalyanaraman, 2013), the findings from the present in vitro study, together with those from our recent in vivo study (Ye et al., 2014), consistently supports the notion that an alteration in the intracellular Zn²⁺ homeostasis via activation of the TRPM2 channel contributes to ROS-induced and reperfusion-related neuronal death.

We showed in the present study that H₂O₂-induced hippocampal neuronal death was also attenuated by removing extracellular Ca²⁺ (Figure 1E), as reported in a previous study for H₂O₂-induced cortical neuronal death (Kaneko et al., 2006). These findings suggests TRPM2-mediated Ca²⁺ influx is critical in determining H₂O₂-induced neuronal death. It is well recognized that intracellular Ca²⁺ is important for full activation of the TRPM2 channel by ADPR (Du et al., 2009; Tóth and Csanády, 2010). Our recent study shows functional expression of the TRPM2 channel on the cell surface, particularly requirement of extracellular Ca²⁺ influx for H₂O₂-induced increase in the [Zn²⁺]_c in hippocampal neurons (Ye et al., 2014). Thus, one explanation, which can readily reconcile the role of the [Ca²⁺]_c and [Zn²⁺]_c in ROS-induced neuronal death, is that TRPM2-mediated Ca²⁺ influx promotes further activation of the TRPM2 channel that leads to TRPM2-dependent alteration in the intracellular Zn²⁺ homeostasis, as recently proposed by studies examining the role of TRPM2 channel in H₂O₂-induced pancreatic β-cell death (Manna et al., 2015).

In the present study, we have provided new insights into TRPM2-dependent alteration in the intracellular Zn²⁺ homeostasis leading to ROS-induced neuronal death. As previously reported in cortical (Colvin et al., 2003) and hippocampal neurons (Ye et al., 2014), the present study revealed that hippocampal neurons under control conditions contained a very low level of Zn²⁺ that was largely present in puncta (Figure 2), and further demonstrated that a majority of such Zn²⁺ puncta were strongly co-localized with LysoTracker (Figures 3A,B), suggesting lysosomal localization. Similar findings were described in insulin-secreting cells (Manna et al., 2015). Moreover, H₂O₂-induced increase in the [Zn²⁺]_c in hippocampal neurons was accompanied with lysosome dysfunction as evidenced by the massive reduction in LysoTracker fluorescence intensity (Figure 3A). Genetic deletion of the TRPM2 channel not only prevented H₂O₂-induced increase in the [Zn²⁺]_c (Figures 2A,B) but also H₂O₂-induced lysosomal dysfunction (Figure 3A), leading us to propose that lysosomal dysfunction is at least in part responsible for lysosomal Zn²⁺ release to increase the [Zn²⁺]_c. However, other Zn²⁺ release mechanisms and sources remain possible. For example, the TRPM2 channel was shown to function as a lysosomal Ca²⁺ release channel in insulin-secreting cells (Togashi et al., 2006; Lange et al., 2009) and has been recently suggested to mediate lysosomal Zn²⁺ release that is responsible for ROS-induced increase in the [Zn²⁺]_c and subsequent pancreatic-β cell death (Manna et al., 2015). Further studies are required to examine whether a similar mechanism operates in hippocampal neurons. It is noteworthy to mention that increasing indirect evidence supports a role for the TRPM2 channel in modulating intracellular Zn²⁺ homeostasis (Ye et al., 2014; Manna et al., 2015; Abuarab et al., 2017; Li et al., 2017). However, unlike the Ca²⁺-permeability (Xia et al., 2008), it remains difficult to demonstrate whether the TRPM2 channel is permeable to Zn²⁺ because of the potent inhibition of TRPM2 channel by micromolar concentrations of Zn²⁺ (Yang et al., 2011). There is evidence that ROS induces Zn²⁺ release from cytosolic Zn²⁺-binding proteins such as metallothioneins in hippocampal neurons (Lee et al., 2003), which cannot be ruled out as a possible source for H₂O₂-induced increase in the [Zn²⁺]_c observed in the present study.

In this study, we also provide evidence to suggest that mitochondrial Zn²⁺ accumulation and ensuring mitochondrial ROS generation are important in ROS-induced hippocampal neuronal death. The Zn²⁺ puncta in untreated neurons were poorly co-localized with MitoTracker, but there was a considerable increase in the col-localization between Zn²⁺ and MitoTracker in H₂O₂-treated neurons, suggesting mitochondrial Zn²⁺ accumulation (Figure 4). H₂O₂-induced Zn²⁺ translocation into mitochondria was however not observed in TRPM2-KO neurons (Figure 4). Furthermore, H₂O₂ induced mitochondrial dysfunction (Figures 5A,B), mitochondrial fragmentation (Figures 5A,C) and release of Cyt-c (Figures 5D,E). Similar Zn²⁺-induced mitochondrial effects were previously reported in cortical neurons (Jiang et al., 2001; Dineley et al., 2003; Sensi et al., 2003, 2011; Medvedeva et al., 2009; Shuttleworth and Weiss, 2011). Here we provide the first evidence to show that these H₂O₂-induced mitochondrial effects in hippocampal neurons were strongly dependent of the TRPM2 channel (Figure 5). Of notice, a recent study reports that genetic depletion of the TRPM2 expression in human neuroblastoma SH-SY5Y cells led to an increased mitochondrial ROS level as well as reduced cell proliferation (Bao et al., 2016), although it was unclear regarding the implication to ROS-induced cell death. Nonetheless, another recent study shows that activation of the TRPM2 channel mediates SH-SY5Y cell death induced by H₂O₂ as well as 1-methyl-4-phenylpyridinium ion (Sun et al., 2016). We found negligible production of mitochondrial ROS in both WT and TRPM2-KO neurons under control conditions (Figures 6A,B). However, H₂O₂ induced substantial production of mitochondrial ROS in hippocampal neurons that was prevented by genetic deletion or pharmacological inhibition of the TRPM2 channel (Figures 6A–C) as well as TPEN (Figures 6C,D). These results strongly support mitochondrial ROS generation as a sequela of mitochondrial Zn²⁺ accumulation.

In summary, this study shows that TRPM2-dependent dynamic alteration in the intracellular Zn²⁺ homeostasis and lysosomal and mitochondrial dysfunctions are important in contributing to ROS-induced hippocampal neuronal death. These novel insights should be useful in facilitating better understanding ROS-induced neuronal death implicated in ischemia-reperfusion brain damage.

L-HJ conceived the research. L-HJ, XL and WY designed the experiments. XL performed experiments and analyzed data, and WY carried out initial experiments and contributed to discussion. L-HJ and XL wrote the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.