Research ArticleCell Biology

High glucose–induced ROS activates TRPM2 to trigger lysosomal membrane permeabilization and Zn2+-mediated mitochondrial fission

See allHide authors and affiliations

Sci. Signal.  01 Aug 2017:
Vol. 10, Issue 490, eaal4161
DOI: 10.1126/scisignal.aal4161

Fragmented by diabetic stress

The high circulating glucose concentrations characteristic of diabetes induce the excessive production of reactive oxygen species (ROS), which triggers mitochondrial fragmentation. The cation channel TRPM2 is activated by ROS, leading Abuarab et al. to investigate the role of this channel in mitochondrial fragmentation in endothelial cells, which become dysfunctional in diabetics. In response to high glucose–induced oxidative stress, Ca2+ influx through TRPM2 channels caused lysosomal permeabilization and redistribution of lysosomal Zn2+ to mitochondria. The increase in mitochondrial Zn2+ led to the recruitment of the fission factor Drp-1, resulting in mitochondrial fragmentation. This pathway may play a role in the pathology of aging-associated diseases that are characterized by increased mitochondrial fragmentation.

Abstract

Diabetic stress increases the production of reactive oxygen species (ROS), leading to mitochondrial fragmentation and dysfunction. We hypothesized that ROS-sensitive TRPM2 channels mediated diabetic stress–induced mitochondrial fragmentation. We found that chemical inhibitors, RNAi silencing, and genetic knockout of TRPM2 channels abolished the ability of high glucose to cause mitochondrial fission in endothelial cells, a cell type that is particularly vulnerable to diabetic stress. Similar to high glucose, increasing ROS in endothelial cells by applying H2O2 induced mitochondrial fission. Ca2+ that entered through TRPM2 induced lysosomal membrane permeabilization, which led to the release of lysosomal Zn2+ and a subsequent increase in mitochondrial Zn2+. Zn2+ promoted the recruitment of the fission factor Drp-1 to mitochondria to trigger their fission. This signaling pathway may operate in aging-associated illnesses in which excessive mitochondrial fragmentation plays a central role.

INTRODUCTION

About 10% of the global population currently suffers from diabetes. Diabetes is a major risk factor for many late-onset diseases that include cardiovascular diseases, neuronal diseases, and cancer (13). In diabetic patients, tissues are exposed to abnormally high blood amounts of glucose, fats, and proinflammatory cytokines (collectively known as “diabetic milieu”) (1, 4, 5). Tissues exposed to diabetic milieu experience oxidative stress due to increased production of reactive oxygen species (ROS) (46). ROS target various mechanisms that contribute to diabetes-associated diseases, among which mitochondrial dynamics is emerging as an important disease mechanism (7, 8). By triggering abnormal mitochondrial fragmentation, ROS impair mitochondrial function, thereby contributing to disease states (812). That mitochondrial dynamics contributes to the disease state is highlighted by the rescue of oxidative stress–induced dysfunction of many cell types by inhibition of mitochondrial fragmentation (1315). Thus, mitochondrial dynamics represents an attractive therapeutic target for many late-onset diseases (8, 9, 16, 17). To fully realize this potential, however, requires a better understanding of the molecular and cellular mechanisms responsible for abnormal mitochondrial fragmentation.

Eukaryotic cells maintain a healthy mitochondrial network by regulating the balance between mitochondrial fusion and fission processes, collectively known as mitochondrial dynamics (710, 18). Oxidative stress, including that imposed by the diabetic milieu, tips this balance toward mitochondrial fission, leading to fragmented, dysfunctional mitochondria in the cell (79, 1113, 18, 19). Mitochondrial fusion is mediated by three guanosine triphosphatases (GTPases): mitofusin 1 (Mfn1), Mfn2, and optic atropy 1 (810, 16, 18). Fission is mediated by the dynamin-related protein–1 (Drp-1), another GTPase that forms oligomeric spirals to constrict mitochondria at sites where specific adaptors [mitochondrial fission factor (MFF), mitochondrial dynamics protein 49-51 (Mid49-51), and mitochondrial fission 1 protein (Fis1)] are located (9, 16). Under normal conditions, Drp-1 is mainly localized to the cytoplasm, but during oxidative stress, it is recruited to the network at sites marked and preconstricted by the endoplasmic reticulum (ER) tubules (20). Drp-1 recruitment to mitochondria is regulated and Ca2+-dependent (21, 22).

Here, we hypothesized that ROS-sensitive transient receptor potential channel, subtype melastatin 2 (TRPM2) ion channels mediated oxidative stress–induced mitochondrial fission because oxidative stress stimulates TRPM2 channels and that activation of TRPM2 increases intracellular cytosolic Ca2+ concentrations (2325) required for mitochondrial fission (21, 22). To address our hypothesis, we selected endothelial cells because these cells are harmed by the diabetic milieu. When exposed to high glucose, they display extensive mitochondrial fragmentation and do not respond to agonist-stimulated activation of nitric oxide synthase and cyclic guanosine 3′,5′-monophosphate (cGMP) production, an effect that is rescued by inhibition of fragmentation through silencing of Fis1 and Drp-1 (12). Furthermore, TRPM2 channels are present in endothelial cells (26). Our results demonstrate a role for TRPM2 channels in oxidative stress–induced mitochondrial fragmentation and reveal a signaling cascade that links oxidative stress to mitochondrial fission.

RESULTS

TRPM2 channels mediate oxidative stress–induced mitochondrial fragmentation

To test our hypothesis, we transfected human umbilical vein endothelial cells (HUVECs) with pMito-Cherry, a plasmid construct that allows labeling of mitochondria with the cherry fluorescent reporter protein. Consistent with previous reports, high glucose (33 mM) caused extensive breakdown of the mitochondrial network, resulting in small, rounded structures (Fig. 1A). By contrast, cells exposed to normal glucose concentration (5.6 mM) or normal glucose plus mannitol (to exclude potential osmotic effects by the excess glucose) displayed a healthy mitochondrial network comprising long, branched tubular networks. High glucose caused a significant reduction in both the aspect ratio (length to width ratio) and the form factor (a measure of degree of branching) of mitochondria (Fig. 1, B to D). Inhibition of TRPM2 channels with the nonspecific channel inhibitor 2-aminoethoxydiphenyl borate (2-APB) (27) or TRPM2 silencing RNA (fig. S1, A and B) prevented mitochondrial fragmentation (Fig. 1, A to E). By contrast, small interfering RNA (siRNA)–mediated silencing of stromal interaction molecule 1 and Orai-1, which play a major role in store-operated Ca2+ entry (SOCE) in endothelial cells, or their selective inhibition with Synta66 (28) failed to prevent high glucose–induced mitochondrial fragmentation (fig. S2, A and B). These data indicate that TRPM2 channels play a key role in high glucose–induced mitochondrial fragmentation.

Fig. 1 Inhibition of TRPM2 channels prevents high glucose–induced mitochondrial fragmentation.

(A) HUVECs expressing mitochondria-targeted Mito-Cherry protein were incubated with Endothelial Cell Growth Medium-2 (EGM-2) [control (CTRL); 5.6 mM] or EGM-2 supplemented with mannitol (27.4 mM) or high glucose (27.4 mM) for 24 hours. Where indicated, cells were cotransfected with siRNA or treated with 37.5 μM 2-APB. Representative confocal images are shown. Scale bars, 10 μm. Boxed regions are magnified in the bottom panels. Scale bar, 5 μm. (B) Plots of form factor against aspect ratio calculated from the images in (A). (C and D) Means ± SEM of aspect ratio (C) and form factor (D) calculated from the data in (A), analyzed as in (B). n = 3 independent experiments; N = 9 cells in total. (E) Means ± SEM of percent cells displaying mitochondrial fragmentation determined from data in (A). n = 3 independent experiments; N = 130 cells in total. (F) ROS production in HUVECs after the treatments as in (A). Cells were stained with H2DCFDA (DCF), and means ± SEM of fluorescence per cell are presented. n = 3 independent experiments; N = 180 cells in total. (G) Representative confocal images of HUVECs exposed to high glucose with and without the ROS scavenger N-acetyl-cysteine (NAC). (H) Means ± SEM of percent cells displaying mitochondrial fragmentation determined from the data in (G). n = 3 independent experiments; N = 70 cells in total. Statistical analysis was performed by one-way analysis of variance (ANOVA) with Tukey’s post hoc test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Scr, scrambled; NS, not significant.

High glucose is not a direct activator of TRPM2 channels, but as reported previously (12, 29), high glucose increased ROS production within cells (Fig. 1F). The high glucose–induced increase in ROS thus appeared to be sufficient to activate TRPM2 channels and thereby cause mitochondrial fragmentation. Consistent with this argument, quenching of ROS with N-acetyl-cysteine prevented high glucose–induced mitochondrial fragmentation (Fig. 1, G and H). Acute activation of TRPM2 channels with H2O2 caused mitochondrial fragmentation in a manner that was blocked by 2-APB or TRPM2-siRNA (fig. S1, C to E), supporting a role for these channels in mitochondrial dynamics.

To seek further evidence for the role of TRPM2 channels in mitochondrial dynamics, we also used human embryonic kidney (HEK) 293 cells, which lack TRPM2 channels (30, 31). H2O2 did not affect mitochondrial morphology in these cells, but heterologous expression of TRPM2 channels led to extensive fragmentation (fig. S1F). In an alternative approach, we compared the effect of high glucose on the mitochondrial network of primary endothelial cells from wild-type mice with those from TRPM2 knockout (TRPM2 KO) mice. We confirmed the identity of isolated endothelial cells by immunostaining for the endothelial marker CD31 (also known as platelet endothelial cell adhesion molecule–1). Similar to HUVECs, MitoTracker Red staining showed that high glucose caused extensive mitochondrial fragmentation in primary endothelial cells from wild-type mice (Fig. 2, A to C). Endothelial cells isolated from TRPM2 KO mice, however, were markedly resistant to high glucose–induced mitochondrial fragmentation (Fig. 2, D to G). Finally, mitochondria in the endothelial cells of intact aorta from TRPM2 KO mice did not undergo fragmentation in response to high glucose (fig. S3). Together, we provided several lines of evidence (pharmacological, siRNA, knockout, and HEK cell data) to support our hypothesis that TRPM2 channels play a key role in oxidative stress–induced mitochondrial fragmentation.

Fig. 2 Knockout of TRPM2 channels prevents high glucose–induced mitochondrial fragmentation in mouse pulmonary endothelial cells.

(A to F) Effect of high glucose on mitochondrial fragmentation of primary endothelial cells isolated from wild-type (WT) mice (A to C) or TRPM2 knockout (TRPM2 KO) mice (D to F). Cells were incubated with 5.6 mM glucose (CTRL), mannitol (Man; 27.4 mM), or high glucose (27.4 mM) for 72 hours and stained with MitoTracker Red (red, mitochondria), rabbit antibodies against CD31 (green, endothelial cells), and DAPI (4′,6-diamidino-2-phenylindole) (nuclei, blue). (A) and (D) show representative images. Scale bars, 10 μM. Boxed regions in the merged images are expanded in the bottom panels. Scale bar, 5 μm. (B and E) Plots of form factor against aspect ratio calculated from the images in (A) and (D), respectively. (C and F) Means ± SEM of aspect ratio and form factor calculated from data analyzed as in (B) and (E), respectively (n = 3 independent experiments; cells from two to three mice were pooled for each experiment). N = 9 cells in total. (G) Comparison of aspect ratio and form factor of mitochondria of endothelial cells isolated from WT and TRPM2 KO mice, following various treatments [data from (C) and (F)]. Statistical analysis was performed by one-way ANOVA with Tukey’s post hoc test. *P < 0.05, **P < 0.01, and ***P < 0.001.

TRPM2 channels regulate mitochondrial fragmentation through Ca2+-induced changes in Zn2+ dynamics

Ca2+ is required for mitochondrial fragmentation (21, 22). However, TRPM2 channels not only conduct Ca2+ but also regulate intracellular Zn2+ dynamics (23, 24, 31, 32). To exclude a role for Zn2+, we used DTPA (diethylenetriamine pentaacetic acid) and TPEN [N,N,N′,N′-tetrakis(2-pyridinylmethyl)ethylenediamine] (31). Chelation of extracellular Zn2+ with DTPA failed to prevent H2O2-induced mitochondrial fission (fig. S4), suggesting that extracellular Zn2+ entry did not contribute to mitochondrial fission. By contrast, TPEN, which unlike DTPA also chelates intracellular Zn2+, abolished the ability of high glucose and H2O2 to induce mitochondrial fragmentation (Fig. 3, A to D, and fig. S5, A to D), suggesting that Zn2+ was likely released from an intracellular site. The effect of TPEN was not due to Ca2+ chelation because the concentration used here is too low (0.3 μM) to bind Ca2+ (30, 31, 33). Furthermore, the Zn2+ chelating agent clioquinol (31) also prevented H2O2-induced mitochondrial fission (fig. S5, A to D). In addition, raising the cytosolic concentrations of Zn2+ with the Zn2+-specific ionophore pyrithione (Zn-PTO) (30) caused extensive mitochondrial fragmentation, and this effect was rescued by TPEN (Fig. 3, E and F). Together, these data revealed a role for Zn2+ in oxidative stress–induced mitochondrial dynamics, prompting further investigation into the relative roles of Ca2+ and Zn2+.

Fig. 3 High glucose–induced mitochondrial fragmentation is mediated by Zn2+.

(A) Representative images of the effect of the Zn2+ chelator TPEN (0.3 μM) on HUVECs expressing Mito-Cherry and incubated with 5.6 mM glucose (CTRL) or high glucose (33 mM glucose). (B and C) Means ± SEM of form factor (B) and aspect ratio (C) calculated from the data in (A). n = 3 independent experiments; N = 9 cells in total. (D) Means ± SEM of percent cells showing mitochondrial fragmentation, calculated from the data in (A). n = 3 independent experiments; N = 130 cells in total. (E) Representative images of the effect of delivering Zn2+ through pyrithione (Zn-PTO, 0.7 μM Zn2+ and 0.5 μM pyrithione) on HUVECs expressing Mito-Cherry, in the presence or absence of TPEN (0.3 μM). (F) Means ± SEM of percent cells showing mitochondrial fragmentation, calculated from the data in (E). n = 3 independent experiments; N = 170 cells in total. (G) A23187 (1 μM) application, shown with a horizontal bar, causes an increase in cytosolic Ca2+ (Ca2+i). (H) Means ± SEM of change in Ca2+ fluorescence calculated from the data in (G). n = 3 independent experiments. (I) Representative confocal images of cells treated with vehicle (CTRL) or A23187 (1 μM) and stained for Ca2+. (J) Representative confocal images of HUVECs treated with vehicle (CTRL) or 1 μM A23187 for the indicated times and stained for mitochondria, in the presence or absence of TPEN (0.3 μM). (K) Means ± SEM of percent cells showing mitochondrial fragmentation, calculated from the data in (J). n = 3 independent experiments; N = 170 cells in total. Scale bars, 10 μm (in representative confocal images); 5 μm (in images in which boxed regions are expanded). Statistical analysis was performed by one-way ANOVA with Tukey’s post hoc test (B to D, F, and K) or Students t test (H). *P < 0.05, **P < 0.01, and ***P < 0.001.

We were unable to directly test the role of Ca2+ in mitochondrial fragmentation because the currently available Ca2+ chelators, including 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA)–tetraacetoxymethyl ester (AM), binds Zn2+ more avidly than Ca2+ (33) and thus are not selective for Ca2+ (30). As an alternative approach, we increased the cytosolic concentrations of Ca2+ using the Ca2+ ionophore A23187 (Fig. 3, G to I). Consistent with previous reports (34, 35), A23187 induced significant mitochondrial fragmentation (Fig. 3, J and K). However, unlike Zn-PTO, A23187 required longer incubation times to cause mitochondrial fragmentation. We have previously demonstrated that TRPM2-mediated Ca2+ entry stimulates intracellular Zn2+ release (31). We therefore asked whether A23187-induced mitochondrial fission could be mediated by Ca2+-induced changes in intracellular Zn2+ dynamics. Consistent with this possibility, TPEN significantly reduced A23187-induced mitochondrial fission (Fig. 3, J and K). Finally, RNA interference (RNAi) silencing of TRPM2 did not prevent mitochondrial fragmentation induced by A23187 (fig. S6, A and B) and Zn-PTO (fig. S6, A and C), indicating that TRPM2 channels did not contribute to ionophore-induced effects on mitochondria. Together, these data indicate that TRPM2-mediated Ca2+ entry affects intracellular Zn2+ dynamics and thereby mitochondrial fragmentation.

TRPM2 activation and Ca2+ entry cause lysosomal membrane permeabilization

To understand how oxidative stress affects the Zn2+ dynamics, we first examined the intracellular distribution of free Zn2+. Costaining for Zn2+ and organelle markers revealed that, as with other cell types (31, 36), free Zn2+ largely localized to lysosomes in HUVECs as was apparent from the overlap (yellow) of Zn2+ staining (FluoZin-3, green) with lysosomal staining (LysoTracker Red) (fig. S7A). Exposure to high glucose (but not to mannitol) caused a decrease in the number of vesicles containing free Zn2+, which was accompanied by a parallel decrease in the number of LysoTracker Red–positive vesicles (Fig. 4, A and B). Similar results were obtained when oxidative stress was directly imposed with H2O2 (fig. S7, B and C). The decrease in the number of LysoTracker Red–positive vesicles was not an artifact resulting from reduced uptake of the LysoTracker Red dye because H2O2 caused release of cathepsin B from lysosomes (fig. S8, A and B). These results suggest that high glucose and H2O2, like other oxidative insults (37), induce lysosomal membrane permeabilization and that lysosomal membrane permeabilization causes loss of lysosomal Zn2+.

Fig. 4 High glucose induces lysosomal membrane permeabilization through TRPM2-mediated Ca2+ rise.

(A) Representative confocal images of HUVECs incubated with 5.6 mM glucose (CTRL), mannitol (27.4 mM), or high glucose (27.4 mM) for 24 hours and stained for Zn2+ (FluoZin-3, green) and lysosomes (LysoTracker Red, red). (B) Means ± SEM of the percentage of LysoTracker Red–positive lysosomes per cell, calculated from the data from (A). n = 3 independent experiments; N = 100 cells in total. (C) HEK-293–TRPM2tet were not induced (−Tet) or induced (+Tet) with tetracycline before exposure to H2O2 (200 μM; 90 min) and staining for lysosomes. (D) Means ± SEM of the percentage of LysoTracker Red–positive lysosomes per cell, calculated from the data from (C). n = 3 independent experiments; N = 100 cells in total. (E) Representative confocal images of cells exposed to EGM-2 alone or EGM-2 supplemented with 1 μM A23187 for 2 or 4 hours before staining with LysoTracker Red. (F) Means ± SEM of the percentage of LysoTracker Red–positive lysosomes per cell, calculated from the data in (E). n = 3 independent experiments; N = 100 cells in total. (G) HUVECs transfected with scrambled siRNA or TRPM2-siRNA were treated with A23187 for 4 hours and stained as in (E). Means ± SEM percentage of LysoTracker Red–positive lysosomes per cell from three independent experiments is shown. N = 100 cells in total. Scale bars, 10 μm (in representative confocal images); 5 μm (in images in which boxed regions are expanded). Statistical analysis was performed by one-way ANOVA with Tukey’s post hoc test. *P < 0.05, **P < 0.01, and ***P < 0.001.

Our data suggested that TRPM2-mediated Ca2+ entry affects intracellular Zn2+ dynamics (Fig. 4A and fig. S7B). Because lysosomal membrane permeabilization mobilized lysosomal Zn2+, we examined the role of TRPM2 channels and Ca2+ entry on lysosomal membrane permeabilization. TRPM2-siRNA, but not scrambled siRNA, inhibited lysosomal membrane permeabilization induced by both high glucose (Fig. 4, A and B) and H2O2 (fig. S7, B and C) and the associated loss of lysosomal Zn2+. To confirm the role of TRPM2 in lysosomal membrane permeabilization, we used HEK-293–TRPM2tet cells, which express TRPM2 channels only when induced with tetracycline. In the absence of tetracycline, H2O2 did not affect the number of lysosomes, but when TRPM2 expression was induced with tetracycline, cells displayed robust lysosomal membrane permeabilization (Fig. 4, C and D). Together, these results provide evidence that lysosomal membrane permeabilization is not a nonspecific process but is regulated by a Ca2+ channel. To demonstrate that Ca2+ entry drives lysosomal membrane permeabilization, we increased the cytosolic concentrations of Ca2+ with A23187, which triggered lysosomal membrane permeabilization (Fig. 4, E and F). siRNA-mediated silencing of TRPM2 channels failed to prevent A23187-induced lysosomal permeabilization, indicating that TRPM2 acts upstream in the signaling cascade (Fig. 4G). Together, our data demonstrate that oxidative stress causes lysosomal membrane permeabilization by stimulating TRPM2-mediated extracellular Ca2+ entry. Although the precise mechanisms by which Ca2+ entry induces lysosomal membrane permeabilization remain to be investigated, these results demonstrate that Ca2+-induced lysosomal membrane permeabilization leads to the mobilization of lysosomal Zn2+.

TRPM2-induced lysosomal membrane permeabilization is accompanied by an increase in mitochondrial Zn2+

Although lysosomal membrane permeabilization caused release of lysosomal Zn2+, there was no detectable increase in cytosolic Zn2+ (Fig. 4A and fig. S7B). Because mitochondria can sequester free Zn2+ (38) and because mitochondrial fission was prevented by Zn2+ chelation (Fig. 3, A to F and J and K), we asked whether Zn2+ released during lysosomal membrane permeabilization was removed by mitochondria. To test this notion, we costained high glucose–treated cells with FluoZin-3 and MitoTracker Red to label Zn2+ (green) and mitochondria (red), respectively. The numerous yellow puncta in merged images of high glucose–treated cells indicated the presence of Zn2+ in fragmented mitochondria. By contrast, in control and mannitol-treated cells, the mitochondrial network was intact, and there was no detectable free Zn2+ in mitochondria (Fig. 5, A to C). We also examined high glucose– and H2O2-induced increases in mitochondrial Zn2+ using cells expressing modest amounts of Mito-Cherry. We selected cells showing partial fragmentation to visualize mitochondria on their way to full fragmentation using instant structured illumination microscopy (iSIM) (39). The resulting high-resolution images showed the presence of Zn2+ in partially broken mitochondria (Fig. 5D). These data suggest that an increase in mitochondrial Zn2+ likely triggered mitochondrial fragmentation. To confirm that lysosomal membrane permeabilization contributed to the increase in mitochondrial Zn2+, we blocked the lysosomal membrane permeabilization–mediated lysosomal Zn2+ release with TRPM2-siRNA, which attenuated high glucose–induced mitochondrial Zn2+ accumulation and, as expected, was accompanied by inhibition of mitochondrial fragmentation (Fig. 5, A to C). Similar results were obtained in H2O2-treated cells (fig. S9, A and B). Together, these data indicate that Zn2+ released during lysosomal membrane permeabilization is redistributed to mitochondria.

Fig. 5 High glucose increases mitochondrial Zn2+ and fragmentation through TRPM2 channel activation.

(A) HUVECs were incubated with 5.6 mM glucose (CTRL), mannitol (27.4 mM), or high glucose (27.4 mM) for 24 hours and stained for Zn2+ (FluoZin-3, green) and mitochondria (MitoTracker Red, red). Representative confocal images are shown. Scale bars, 10 μm. Boxed regions in the merged images are magnified in the far right panels. Scale bar, 5 μm. (B) Means ± SEM of percent localization of Zn2+ with mitochondria calculated from the data in (A). N = 3 independent experiments. CTRL, 155 cells; mannitol, 91 cells; high glucose, 119 cells; scrambled siRNA, 108 cells; siRNA-TRPM2, 148 cells. (C) Percentage of cells showing mitochondrial fragmentation, calculated from the data in (A). n = 3 independent experiments; N = 160 cells in total. Statistical analysis for (B) and (C) was performed by one-way ANOVA with Tukey’s post hoc test. *P < 0.05, **P < 0.01, and ***P < 0.001. (D) iSIM images of HUVECs showing localization of Zn2+ in mitochondria following high glucose– and H2O2-induced stress. HUVECs expressing Mito-Cherry were treated with high glucose (33 mM; 24 hours) or H2O2 (200 μM, 2 hours) before staining for Zn2+ with FluoZin-3. Representative images (n = 3 independent experiments) are shown. Scale bars, 10 μM.

TRPM2-dependent increase in mitochondrial Zn2+ promotes Drp-1 recruitment

Mitochondrial fragmentation is initiated by the recruitment of Drp-1, a GTPase that catalyzes mitochondrial fragmentation (8, 9, 16, 18, 22, 40). We postulated that the TRPM2-dependent increase in mitochondrial Zn2+ induced Drp-1 recruitment. To test this notion, we transfected HUVECs with Drp-1–green fluorescent protein (GFP) (a construct that suppresses the activity of endogenous Drp-1) (22) and monitored its recruitment from the cytoplasm to mitochondria. High glucose, but not mannitol, promoted Drp-1 mitochondrial recruitment, as assessed by the colocalization of fluorescence of MitoTracker Red with that of Drp-1–GFP (Fig. 6, A and B). TRPM2-siRNA suppressed recruitment of Drp-1–GFP, but not of dominant negative Drp-1–GFP, in response to high glucose (Fig. 6, A and B). Furthermore, RNAi silencing of Drp-1, as well as its partners, Fis1 and MFF, prevented high glucose–induced mitochondrial fragmentation (Fig. 6, C and D). Together, these results suggested that TRPM2 activation leads to Drp-1 recruitment and mitochondrial fragmentation.

Fig. 6 High glucose–induced TRPM2 activation and rise in Zn2+ promotes mitochondrial Drp-1 recruitment.

(A) Representative confocal images of HUVECs transfected with Drp-1–GFP or dominant negative (DN)–Drp-1–GFP and incubated with 5.6 mM glucose (CTRL), mannitol (27.4 mM), or high glucose (27.4 mM) for 24 hours. TPEN (0.3 μM) was included as indicated. (B) Means ± SEM of percent colocalization of GFP with MitoTracker Red calculated from the data in (A). n = 3 independent experiments; N = 50 cells in total. (C) Representative iSIM fluorescence images of mitochondrial fragmentation in HUVECs cotransfected with pMito-Cherry and siRNA targeting Drp-1, Fis1, and/or MFF and exposed to high glucose for 48 hours. (D) Means ± SEM data for percent cells with mitochondrial fragmentation from experiments performed as in (C). n = 3 independent experiments; N = 60 cells in total. (E) Representative confocal images of HUVECs transfected with Drp-1–GFP or DN–Drp-1–GFP, treated with Zn-PTO for 1 hour, and stained for mitochondria. (F) Means ± SEM of percent colocalization of GFP with MitoTracker Red, calculated from the data in (E). n = 3 independent experiments; N = 50 cells in total. Scale bars, 10 μm (in representative confocal images); 5 μm (in images in which boxed regions are expanded). Statistical analysis was performed by one-way ANOVA with Tukey’s post hoc test. *P < 0.05, **P < 0.01, and ***P < 0.001.

We next asked whether Zn2+ contributes to Drp-1 recruitment. Chelation of Zn2+ with TPEN markedly inhibited high glucose–induced Drp-1–GFP recruitment (Fig. 6, A and B). Furthermore, delivery of Zn2+ through Zn-PTO stimulated mitochondrial recruitment of Drp-1–GFP but not its dominant negative version (Fig. 6, E and F). Together, we conclude that TRPM2-mediated increase in mitochondrial Zn2+ promotes Drp-1 recruitment and subsequent mitochondrial fragmentation.

DISCUSSION

Hyperglycemia is a major risk factor for various human diseases. Multiple studies have reported that mitochondrial dynamics plays a key role in the pathophysiology of diabetes and diabetes-associated complications (7, 8). Cell-based studies have shown that hyperglycemic conditions cause abnormal mitochondrial fragmentation by increasing the production of ROS (11, 12, 29). However, the underlying signaling mechanisms are not fully understood. Using endothelial cells as a model system (12), we report a signaling pathway that linked high glucose–induced ROS production to mitochondrial fragmentation (fig. S10). The pathway entailed extracellular Ca2+ entry through ROS-activated TRPM2 channels, Ca2+-induced lysosomal membrane permeabilization, redistribution of lysosomal Zn2+ to mitochondria, Zn2+-induced mitochondrial recruitment of Drp-1, and mitochondrial fragmentation.

Using pharmacological, RNAi and gene knockout approaches, we demonstrated that ROS-sensitive TRPM2 channels mediate oxidative stress (high glucose and H2O2)–induced mitochondrial fragmentation (Figs. 1, A to E, and 2, A to G, and figs. S1, C to F, and S3). To gain insight into the underlying mechanism, we examined the roles of Ca2+ and Zn2+ because the intracellular concentrations of both of these ions are increased by TRPM2 activation (23, 24, 3032). Although Ca2+ is implicated in mitochondrial fragmentation (21, 22, 41, 42), our results demonstrated that chelation of Zn2+ alone was sufficient to prevent mitochondrial fragmentation (Fig. 3, A to D, and fig. S5, A to D). This role for Zn2+ was further supported by the finding that delivery of Zn2+ through the zinc ionophore, Zn-PTO, caused mitochondrial fragmentation (Fig. 3, E and F). The lack of Ca2+-specific chelators (30, 33) prevented us from directly testing the role of TRPM2-mediated Ca2+ entry in mitochondrial fission. We have therefore used the Ca2+ ionophore A23187, which raises cytosolic concentrations of Ca2+ but not Zn2+ (30). Consistent with the previous reports (34, 35), A23187 caused mitochondrial fission but required longer incubation times (>2 hours) to elicit this effect (Fig. 3, J and K), which was inhibited by the Zn2+ chelator TPEN (Fig. 3, J and K). These results imply that Zn2+ plays a crucial role in Ca2+-induced mitochondrial fission.

Although TRPM2 channels can promote Zn2+ entry (31, 32), chelation of extracellular Zn2+ with the membrane-impermeable DTPA reagent failed to prevent mitochondrial fragmentation (fig. S4), suggesting that the free Zn2+ required for mitochondrial fragmentation must come from an intracellular site. Most of the intracellular Zn2+ is protein-bound (33) except in lysosomes where the acidic pH allows Zn2+ to exist in its free state (31, 36). We asked whether lysosomal Zn2+ was mobilized to affect mitochondrial dynamics. We found that both high glucose and H2O2 trigger lysosomal membrane permeabilization, resulting in the loss of lysosomal Zn2+ (Fig. 4, A and B, and fig. S7, B and C). These results were not surprising because oxidative stress causes lysosomal membrane permeabilization (37, 43, 44) and because both high glucose and H2O2 are stress-inducing substances. Unexpectedly, however, we found that pharmacological inhibition or knockdown of TRPM2 channels prevented high glucose– and H2O2-induced lysosomal membrane permeabilization and the consequent loss of lysosomal Zn2+ (Fig. 4, A and B, and fig. S7, B and C). These findings demonstrate that lysosomal membrane permeabilization is not a nonspecific process (37) but is regulated by an ion channel. Because the TRPM2 channel is primarily a Ca2+ channel, we suspected that extracellular Ca2+ entry could stimulate lysosomal membrane permeabilization by activating lipases such as phospholipase A2, sphingomyelinase, and phospholipase C (43, 45). Increased cytosolic Ca2+ with A23187 caused a marked increase in lysosomal membrane permeabilization (Fig. 4, E and F). Together, our results indicate that TRPM2-mediated Ca2+ entry stimulates lysosomal membrane permeabilization. These findings are important from a pathophysiological perspective because lysosomal membrane permeabilization is linked to various human diseases, including vascular diseases (44).

Although lysosomal membrane permeabilization led to the loss of lysosomal Zn2+, we found an increase in mitochondrial Zn2+, rather than in cytoplasmic Zn2+. Inhibition of lysosomal membrane permeabilization with TRPM2-siRNA attenuated both high glucose– and H2O2-induced increases in mitochondrial Zn2+ (Fig. 5, A and B, and fig. S9, A and B), indicating mobilization of lysosomal Zn2+ to mitochondria. Although how this transfer occurs remains to be investigated, mitochondria have several transport mechanisms, such as the mitochondrial Ca2+ uniporter (38), to facilitate Zn2+ uptake. Regardless of how Zn2+ enters mitochondria, we did not see mitochondrial fragmentation in the absence of an increase in mitochondrial Zn2+. Thus, our results indicate that by raising the mitochondrial Zn2+, TRPM2-mediated lysosomal membrane permeabilization causes mitochondrial fission.

A critical step in mitochondrial fission is the recruitment of Drp-1 from the cytoplasm to mitochondria (8, 9, 16). Accordingly, high glucose increased the recruitment of heterologously expressed Drp-1–GFP to mitochondria (Fig. 6, A and B). Moreover, inhibition of TRPM2 channels and hence of lysosomal membrane permeabilization or chelation of Zn2+ prevented Drp-1–GFP recruitment (Fig. 6, A and B), suggesting that mobilization of lysosomal Zn2+ to mitochondria is essential for mitochondrial fission. The importance of Zn2+ was confirmed by the robust recruitment of Drp-1–GFP to mitochondria by direct delivery of Zn2+ through Zn-PTO (Fig. 6, E and F). Together, our data indicate that during oxidative stress, TRPM2-mediated mobilization of lysosomal Zn2+ to mitochondria promotes Drp-1 recruitment and subsequent mitochondrial fragmentation.

In summary, we described a signaling pathway by which oxidative stress causes mitochondrial fragmentation (fig. S10). We showed that plasma membrane TRPM2 channels responded to oxidative stress to generate Ca2+ signals, which induced lysosomal membrane permeabilization leading to the mobilization of lysosomal Zn2+ to mitochondria, where Zn2+ promotes Drp-1 recruitment and mitochondrial fission. For Drp-1 to catalyze fission, ER tubules need to wrap around and preconstrict the mitochondria. Here, we found that in addition to the ER, plasma membrane and lysosomes play crucial roles in transmitting oxidative stress signals to mitochondria. Thus, our study illustrates how the interplay between various organelles, in conjunction with Ca2+ and Zn2+ signals, regulates mitochondrial dynamics. Furthermore, our study raises several questions. First, how is lysosomal Zn2+ mobilized to mitochondria? Does it require close proximity of lysosomes to mitochondria? Second, what is the molecular route through which Zn2+ enters mitochondria? Third, how does Zn2+ promote mitochondrial Drp-1 recruitment? Does Zn2+ influence posttranslational modifications of Drp-1 (8, 9, 18) required for its recruitment to mitochondria? Or does it affect other aspects of mitochondrial fission, such as ER-assisted constriction (20) or the recruitment of dynamin-2 (46)? Notwithstanding these questions, given the growing recognition that abnormal mitochondrial fragmentation is a recurring theme in the pathophysiology of various late-onset human diseases, the findings presented in this study may have translational potential for age-related illnesses in which mitochondrial dynamics plays a crucial role.

MATERIALS AND METHODS

Reagents and plasmid constructs

LysoTracker Red DND-99, MitoTracker Red CMXRos, Opti-MEM, Pluronic F-127, Fura-2 AM, Fluo-4 AM, FluoZin-3 AM, Hoechst 33342, H2DCFDA, and Lipofectamine 2000 were purchased from Life Technologies. DAPI Fluoromount-G was purchased from SouthernBiotech. Human TRPM2-siRNA [ON-TARGETplus human TRPM2 (7226)] was from Thermo Fisher Scientific. TRPM2-siRNA–2 (5′-GAAAGAAUGCGUGUAUUUUGUAA-3′) was custom-made by Dharmacon. Scrambled siRNA was from Ambion (4390846). siRNA for Drp-1 (catalog no. S1I02661365; 5′-CAGGAGCCAGCTAGATATTAA-3′), Fis1 (catalog no. SI04356751; 5′-AAGGCCATGAAGAAAGATGGA-3′), and MFF (catalog no. SI04320386; 5′-AACGCTGACCTGGAACAAGGA-3′) were from Qiagen. All other chemicals were either from Sigma-Aldrich or Calbiochem. Stock solution of zinc pyrithione was prepared by mixing aqueous solution of ZnCl2 with pyrithione made up in ethanol. pMito-Cherry was constructed from pECFP-Mito (Clontech Laboratories). Drp-1–GFP clone containing short hairpin RNA to knock down the endogenous Drp-1 and its dominant negative (K38A) version (22) was a gift from S. Strack (University of Iowa).

Isolation of lung endothelial cells

Mice were sacrificed by cervical dislocation. Lung microvascular endothelial cells (lung ECs) were isolated from 8- to 10-week-old wild-type (C57BL/6) and TRPM2 KO male mice by immunoselection with anti-CD146–coated magnetic beads (Miltenyi Biotec) according to the protocol described previously (47). Generation of TRPM2 KO mice has been described (31). Mice were bred and maintained under UK Home Office license and ethical procedure.

Cell culture and transfections

Freshly isolated lung ECs and HUVECs (Lonza) were grown in EGM-2 [Endothelial Cell Basal Medium-2 (EBM-2) supplemented with endothelial growth supplements (Lonza)]. The medium was changed every 24 and 48 hours for lung ECs and HUVECs, respectively. For the experiments, lung ECs were seeded onto 1:500 fibronectin (F1141, Sigma-Aldrich)–coated coverslips and HUVECs onto 0.1% gelatin–coated glass-bottomed dishes (35 mm; FluoroDish) and grown in EGM-2. HUVECs were used within passages 3 to 6. HEK-293 cells expressing tetracycline-inducible human TRPM2 (HEK-293–TRPM2tet cells) (31) were cultured in Dulbecco’s modified Eagle medium (DMEM with GlutaMAX; Invitrogen) supplemented with 10% fetal bovine serum (Sigma), penicillin (50 units/ml), streptomycin (50 μg/ml), Zeocin (200 μg/ml), and blasticidin (0.4 μg/ml; InvivoGen). To induce TRPM2 expression, cells were incubated for 48 hours with tetracycline (1 μg/ml). All cells were grown at 37°C under 5% CO2 and humidified atmosphere.

HUVECs were grown on FluoroDish dishes to 50 to 70% confluency. Cells were transfected with Drp-1–GFP or pMito-Cherry using Lipofectamine 2000. Where appropriate, cotransfections were performed with 25 nM human TRPM2-siRNA or scrambled siRNA. Medium was replaced 7 hours after transfection, and incubation continued for 48 to 72 hours during which cells were treated as required before imaging.

Immunostaining

For immunostaining experiments, cells were grown on coverslips. Following the desired treatments (see figure legends), cells were washed with phosphate-buffered saline (PBS), fixed with 2% paraformaldehyde (10 min), and permeabilized with 0.25% Triton X-100, 10 mM tris, and 150 mM NaCl (pH 7.4; 5 min). Nonspecific binding sites were blocked with 1% ovalbumin/PBS for 1 hour before incubation for 2 hours with rabbit antibody against CD31 (1:300; Abcam) diluted in 1% ovalbumin/PBS. After washing thrice with PBS, cells were incubated in the dark with Alexa Fluor 488–conjugated donkey antibody against rabbit immunoglobulin G (1:500; Life Technologies) diluted in 1% ovalbumin/PBS. After washing, coverslips were mounted onto microscope slides in DAPI Fluoromount-G and imaged.

Intracellular Zn2+ redistribution

Intracellular distribution of Zn2+ was assessed by live imaging after staining the cells with FluoZin-3 for Zn2+ and organelles with vital stains (31). HUVECs grown on FluoroDish dishes were washed with EBM-2 and incubated for 4 hours at 37°C in the dark in EBM-2 containing 1 μM FluoZin-3 AM and 0.02% (w/v) Pluronic F-127. Cells were then washed twice (15 min each) with EBM-2 and incubated for 30 min at 37°C with organelle marker dyes, MitoTracker Red CMXRos (200 nM) or LysoTracker Red DND-99 (200 nM) diluted in EBM-2. After washing with EBM-2, cells were imaged. HEK-293 cells were similarly loaded but using the relevant medium.

To examine the effect of glucose, HUVECs were incubated in EGM-2 (5.6 mM glucose) or EGM-2 containing 33 mM glucose or 5.6 mM glucose plus 27.4 mM mannitol for 42 hours at 37°C. Cells were then loaded with FluoZin-3 AM and organelle markers in the respective media as described above. The total incubation time including all steps was 48 hours. For ionophore-mediated Zn2+ loading, cells were incubated in EGM-2 or EGM-2 containing 0.7 μM ZnCl2 plus 0.5 μM pyrithione for 2 hours at 37°C, before washing and loading FluoZin-3 AM (2 hours) and organelle marker dyes. Where appropriate, cells were cotreated with TRPM2 channel inhibitors or metal chelators or pretransfected with siRNA (see relevant figure legends for details). Images were captured at 37°C using Zeiss LSM 700 confocal microscope fitted with a 63× oil objective. Colocalization of Zn2+ fluorescence and organelle fluorescence was determined using the Imaris software.

Mitochondrial fragmentation

Morphology of mitochondria was assessed from images of cells stained with MitoTracker Red or transfected with pMito-Cherry. The aspect ratio (ratio between the major and minor axis) and form factor (degree of branching) were determined using NIH ImageJ 1.44p from individual mitochondria after reducing nonspecific noise of the fluorescence signal as reported previously (48). Form factor is defined as [(perimeter2)/(4π*area)]. Mitochondrial morphology measurements were made from more than 70 individual mitochondria of cells sampled from three independent experiments. Mitochondria were counted as fragmented if the form factor was below 2.5. A cell is considered as having its mitochondria fragmented when ≥50% of its total number of mitochondria is fragmented (49).

Drp-1 recruitment to mitochondria

To examine high glucose–induced Drp-1 recruitment to mitochondria, cells were transfected with Drp-1–GFP or dominant negative Drp-1 (K38A)–GFP with and without human TRPM2-siRNA or scrambled siRNA. Forty-eight hours after transfection, cells were incubated in EGM-2 or EGM-2 containing desired additives for a further 24-hour period. Cells were then stained with MitoTracker Red, washed, and imaged in EBM-2 for Drp-1–GFP and MitoTracker Red.

ROS measurement

Total ROS was determined by staining cells with the H2DCFDA reagent. Cells grown in a 96-well plate were treated or not treated (control) with the test substances (see relevant figure legends). They were then incubated with 10 μM H2DCFDA diluted in EBM-2 for 30 min at 37°C followed by three washes with EBM-2. Cells were imaged using the EVOS FL Cell Imaging System (Life technologies) fitted with a 40× lens. Images were analyzed with ImageJ software. The results were expressed as the mean fluorescence intensity per cell.

Image acquisition and analysis

Images were collected with a Zeiss LSM 700 inverted laser scanning confocal microscope equipped with an oil-immersion 63×/numerical aperture (NA) 1.3 objective lens. DAPI (345-nm excitation, 458-nm emission) was excited with a diode laser at 405 nm, fitted with a 420- to 440-nm emission filter. MitoTracker Red, LysoTracker Red, and Mito-Cherry (548-nm excitation, 562-nm emission) were excited using a He-Ne laser fitted with 543-nm filters. Alexa Fluor 488 and FluoZin-3 (494-nm excitation, 519-nm emission) were excited with an Argon laser at 488 nm, fitted with a 500- to 530-nm emission filter. Images were acquired with ZEN lite 2011 and analyzed using ImageJ/Imaris software. Some images were collected using iSIM fitted with an Olympus Water Immersion Objective 60×/NA 1.2 Uplsapo 60xw and 488- and 561-nm lasers (39).

Intracellular Ca2+ measurements

Intracellular changes in [Ca2+] were monitored using Fura-2 AM (30). HUVECs grown in 96-well plates were incubated with Fura-2 AM (2 μM) in SBS containing 0.02% Pluronic F-127 for 1 hour at 37°C. After washing twice with SBS [10 mM Hepes, 130 mM NaCl, 1.2 mM KCl, 8 mM glucose, 1.5 mM CaCl2, and 1.2 mM MgCl2 (pH 7.4)] for 30 min, 200 μl of SBS was added to each well. Fluorescence was recorded using the FlexStation II multimode microplate reader (Molecular Devices). Fluorescence was measured at 5- to 10-s intervals using excitation wavelengths of 340 and 380 nm and an emission wavelength of 510 nm for Ca2+. After taking several control measurements, 50 μl of the desired reagents made up at fivefold the required final concentration was added to wells. The ratio of fluorescence intensities at 340 and 380 nm (F340/F380) was calculated. For imaging, cells were loaded with Fluo-4 AM (1 μM), and images were collected with EVOS FL Cell Imaging System (Life Technologies).

Data analysis

Colocalization analysis was performed using Imaris software (Bitplane). ImageJ was used for quantification of mitochondrial morphology, Zn2+ uptake by mitochondria, and ROS levels. All experiments were performed at least three times (n), and the values were presented as means ± SEM; n and N in the figure legends indicate the number of independent experiments (n) over the number of cells (N) analyzed. Statistical significance was assessed using Student’s t test or one-way ANOVA, followed by Tukey’s post hoc test. P values are indicated with *, **, ***, ****, which correspond to values of 0.05, 0.01, 0.001, and 0.0001, respectively.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/10/490/eaal4161/DC1

Fig. S1. TRPM2 channels mediate mitochondrial fragmentation.

Fig. S2. SOCE channels do not contribute to high glucose–induced mitochondrial fragmentation.

Fig. S3. Genetic deficiency of TRPM2 prevents high glucose–induced mitochondrial fragmentation in endothelial cells of intact aorta.

Fig. S4. Extracellular Zn2+ does not contribute to the H2O2-induced increase in mitochondrial Zn2+ and mitochondrial fragmentation.

Fig. S5. Zn2+ chelation prevents H2O2-induced mitochondrial fragmentation.

Fig. S6. Ca2+ and Zn2+ ionophores induce mitochondrial fragmentation independently of TRPM2.

Fig. S7. Activation of TRPM2 channels reduces the number of lysosomes.

Fig. S8. H2O2-induced lysosomal permeabilization and release of cathepsin B.

Fig. S9. H2O2-induced increase in mitochondrial Zn2+ is TRPM2-dependent.

Fig. S10. Signaling cascade associated with oxidative stress–mitochondrial fragmentation.

REFERENCES AND NOTES

Acknowledgments: We thank M. Ludlow and G. Howell (University of Leeds) for advice with data analysis using Imaris/ImageJ software. We thank A. Curd and R. Hughes (University of Leeds) for advice with iSIM and S. Strack (University of Iowa) for providing the Drp-1–GFP clone. Funding: This work was supported by the British Heart Foundation (PG/10/68/28528) and King Saud bin Abdulaziz University for Health Sciences, Ministry of Higher Education for Saudi Arabia for a studentship (N.A.). Author contributions: A.S. and N.A. conceived the study. A.S., N.A., J.L., and L.-H.J. designed the experiments. N.A. performed the experiments. N.A. and T.M. analyzed the data. N.A. and A.S. wrote the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article