Ca2+ imbalance caused by ERdj5 deletion affects mitochondrial fragmentation

The endoplasmic reticulum (ER) is the organelle responsible for the folding of secretory/membrane proteins and acts as a dynamic calcium ion (Ca2+) store involved in various cellular signalling pathways. Previously, we reported that the ER-resident disulfide reductase ERdj5 is involved in the ER-associated degradation (ERAD) of misfolded proteins in the ER and the activation of SERCA2b, a Ca2+ pump on the ER membrane. These results highlighted the importance of the regulation of redox activity in both Ca2+ and protein homeostasis in the ER. Here, we show that the deletion of ERdj5 causes an imbalance in intracellular Ca2+ homeostasis, the activation of Drp1, a cytosolic GTPase involved in mitochondrial fission, and finally the aberrant fragmentation of mitochondria, which affects cell viability as well as phenotype with features of cellular senescence. Thus, ERdj5-mediated regulation of intracellular Ca2+ is essential for the maintenance of mitochondrial homeostasis involved in cellular senescence.

www.nature.com/scientificreports/ atherosclerosis, Cys649 is irreversibly oxidized, which prevents it from undergoing NO-induced S-glutathiolation and inactivates SERCA, which is involved in the progression of atherosclerosis 18,19 . Regarding redox-mediated regulation of cysteines on the luminal side of the ER, it was previously reported that the activity of SERCA2b, a ubiquitous isoform of SERCA, is negatively regulated by disulfide bond formation between two luminal cysteines of SERCA2b 20 . Thioredoxin Related Transmembrane Protein 1 (TMX1) is required for the formation of MAM and negatively regulates the Ca 2+ pumping activity of SERCA2b by forming disulfide bonds at SERCA2b luminal cysteines 21 . Moreover, we reported that ERdj5 activates the pump function of SERCA2b by cleaving its luminal disulfide bridge and that ERdj5 knockout in mouse embryonic fibroblasts (MEFs) caused a decrease in [Ca 2+ ] ER due to a reduction in SERCA2b activity 20 . These data suggested that the reductase activity of ERdj5 is involved in Ca 2+ homeostasis in the ER. These findings highlighted the importance of redox regulation in the ER by the function of ERdj5 in both protein and Ca 2+ homeostasis. In ERdj5-deficient mice, sensitivity to ER stress was reported to be dramatically increased in the salivary gland 22 . Recently, another group characterized the phenotypic traits of a Sjögren's syndrome animal model produced by the knockout of ERdj5 23 . Female mice with Sjögren's syndrome-like disorder showed inflammatory infiltrates in the salivary gland, serum autoantibodies, reduced saliva secretion, excessive cell death, and deregulated cytokine levels. On the other hand, the downregulation of ERdj5 in cancer cells increased cell death in response to treatment with fenretinide, a cancer chemopreventive and antiproliferative drug 24 .
A putative orthologue of mammalian ERdj5 was reported in C. elegans and named DNJ-27. Downregulation of DNJ-27 caused severe paralysis and abnormal mobility in worms that were phenotypically similar to human neurodegenerative diseases. Interestingly, the deletion of DNJ-27 caused aberrant mitochondrial fragmentation in the body wall muscle cells of C. elegans 25 , but the detailed mechanisms have never been revealed. In this study, we report that mitochondrial fragmentation was also induced by ERdj5 knockout in mammalian cells. In this process, we succeeded in showing the novel mechanism for the maintenance of transorganellar homeostasis mediated by Ca 2+ signalling originating from ERdj5.

ERdj5 deficiency causes mitochondrial fragmentation in mammalian cells. A previous study
demonstrated the protective role of DNJ-27/ERdj5 against the toxicity associated with the expression of human amyloid-β, α-synuclein and polyQ proteins in C. elegans 25 . At the same time, the knockdown of DNJ-27/ERdj5 caused mitochondrial fragmentation in body wall muscle cells in C. elegans. To confirm the mitochondrial morphology regulated by DNJ-27/ERdj5 in the intestinal cells of C. elegans, we obtained DNJ-27/ERdj5-deficient worms (DNJ-27/ERdj5 KO) and observed the mitochondrial structure in WT and DNJ-27/ERdj5 KO worms by MitoTracker Red staining (Fig. 1A). Mitochondrial fragmentation caused by the deletion of DNJ-27/ERdj5 in intestinal cells was observed, suggesting that this mitochondrial fragmentation is not specific to body wall muscle cells. To verify the fragmentation in mammalian cells, we observed mitochondrial morphology in ERdj5deficient MEFs (Fig. 1B) and HeLa cells in which ERdj5 had been knocked down by small interfering RNA (siRNA) (Fig. S1A). The data showed that the downregulation of ERdj5 caused mitochondrial fragmentation in mammalian cells. Next, we overexpressed ERdj5 in ERdj5-deficient MEFs or HeLa cells in which ERdj5 was knocked down. ERdj5 ± and -/-cells were treated with Carbonylcyanide m-chlorophenylhydrazone (CCCP). CCCP treatment causes mitochondrial fragmentation through the mitochondria membrane depolarization. As a result of CCCP treatment, mitochondrial fragmentation was observed in ERdj5 ± and −/− cells (Fig. 1C). Overexpression of ERdj5/WT-Myc and FLAG completely rescued the changes in mitochondrial structure in ERdj5deficient MEFs (Fig. 1C ) and knockdown HeLa cells ( Fig. S1B and C), whereas overexpression of an ERdj5/ AA-Myc mutant in which all CXXC motifs had been converted to Ala-X-X-Ala (AA), generating a reductase activity-null mutant, did not rescue mitochondrial fragmentation (Fig. 1C). These results suggest that ERdj5 is involved in the maintenance of mitochondrial morphology depending on its reductase activity through the CXXC motifs.

Maintenance of cytosolic calcium homeostasis through ERdj5. Previously, we demonstrated that
ERdj5 cleaved the disulfide bridge in the luminal loop of SERCA2b, a Ca 2+ pump on the ER membrane, and promoted calcium uptake into the ER from the cytosol 20 . Deletion of ERdj5 prevented calcium uptake by SERCA2b and decreased [Ca 2+ ] ER . Thus, we next examined the effect of [Ca 2+ ] i on the fragmentation of mitochondria. Treatment with Tg, an inhibitor of the SERCA family, was reported to prevent the uptake of [Ca 2+ ] ER via the SERCA pump and increase [Ca 2+ ] i due to leakage from the ER to the cytosol and release through IP3R or RyR. In the presence of 1.5 μM Tg, the number of fragmented mitochondria was increased in ERdj5 + /− MEF ( Fig. 2A). The steady-state level of [Ca 2+ ] i in ERdj5-deficient cells was approximately 1.4 times higher than that in WT cells, as revealed by analysis with Yellow Cameleon 3.6 (YC3.6), a fluorescence resonance energy transfer (FRET)based Ca 2+ sensor (Fig. 2B). Ca 2+ influx into mitochondria from the ER after IP3R stimulation by histamine was monitored by the mitochondria-localized Ca 2+ sensor CEPIA2 mt. There was no significant difference in Ca 2+ influx into the mitochondria between WT and ERdj5 KO cells (Fig. S2). Mitochondrial Ca 2+ was measured by the mitochondria-localized Ca 2+ probe Rhod-2. There was no significant difference between WT and ERdj5 KO cells (Fig. S3). Taken together, these results suggest that the deletion of ERdj5 causes the perturbation of Ca 2+ levels in the ER and cytosol of mammalian cells but not in mitochondria.
The fragmentation of mitochondria in ERdj5-deficient cells depends on Drp1 activation. Mitochondrial morphology is dynamic and changed by coordinated fission and fusion. To address the mechanism of [Ca 2+ ] i -dependent mitochondrial fragmentation, we focused on the role of dynamin-related protein 1 (Drp1), The cells were immunostained with anti-Tom20 antibody (green). The graph shows the circularity, aspect ratio and relative extent of the mitochondria from Tom20, as analysed by ImageJ. Scale bars = 10 µm (C) Mitochondrial morphology after the rescue of ERdj5 in ERdj5 −/− cells. Twenty-four hours after transfection of the indicated constructs into MEFs, the cells were incubated in the presence or absence of 20 μM CCCP for 1 h. Then, the cells were immunostained with anti-Tom20 (green) and anti-Myc (red) antibodies. The graphs show the relative extent, number of mitochondria, aspect ratio, and circularity of the mitochondria from Tom20, as analysed by ImageJ. Scale bars = 10 µm (A-C) Insets show high-magnification views of the boxed areas. *P < 0.05 by t test. The results are reported as the means of 30 worms ± SDs (A), 50 cells ± SDs (B) or the means of 25 cells ± SDs (C). www.nature.com/scientificreports/ a cytosolic GTPase involved in mitochondrial fission. cAMP-dependent protein kinase A (PKA) phosphorylates Ser637 in the GTPase effector domain (GED) of human Drp1. This phosphorylation inhibits mitochondrial fission by inhibiting the intramolecular interaction between the GTPase domain and GED domain of Drp1, GTPase activity, and eventually the mitochondrial recruitment of Drp1 26,27 . Conversely, calcineurin, an [Ca 2+ ] i -dependent phosphatase, dephosphorylates phosphorylated Ser637 (Ser637 (P)) in Drp1 and stimulates Drp1 translocation to the mitochondrial membrane 28 .

Scientific
In addition, Ser616 of Drp1 is phosphorylated by Calmodulin kinase II (CamKII), a calcium-dependent phosphatase. Phosphorylation of Ser616 in Drp1 promotes translocation of Drp1 to mitochondria and mitochondrial cleavage 29,30 .
To examine the activation of Drp1 in response to [Ca 2+ ] i , we examined the ratio of phosphorylated vs total Drp1 using a specific antibody for Ser637(P) or Ser616(P). The results showed that phosphorylation of Ser637 was decreased and phosphorylation of 616 was increased in ERdj5−/− MEFs. (Fig. 2C,D). Next, we examined mitochondrial morphology in ERdj5-deficient cells after Drp1 knockdown (Fig. S4). Drp1 knockdown in ERdj5−/− cells restored the mitochondrial morphology from a fragmented morphology to a tube-like morphology (Fig. 2E,F). On the other hand, mammalian mitochondrial morphology was also reported to be regulated through the processing of OPA1 in mitochondrial inner membrane potential (ΔΨ-dependent manner 31 . However, we did not observe a significant difference in the processing of OPA-1 between ERdj5 ± and ERdj5-/-MEFs (Fig. S5). Additionally, we confirmed that the ΔΨ was no difference between ERdj5 ± and ERdj5−/− MEFs, which was examined with the JC-1 MitoMP detection kit (Fig. S6). This finding suggests that mitochondrial fission by the deletion of ERdj5 is independent of the dissipation of the ΔΨ.
The sensitivity of ERdj5-deficient cells to apoptosis. Previous reports have revealed that aberrant mitochondrial fragmentation enhances sensitivity to apoptosis 27 . Treatment with staurosporine (STA), a protein kinase C inhibitor, is known to induce aberrant mitochondrial fragmentation and apoptosis 32 . We examined whether the deletion of ERdj5 would enhance the sensitivity of the cells to apoptosis in the presence of STA. To evaluate the sensitivity to apoptosis, we observed the activation of caspase-3 using the NucView 488 assay. The caspase-activated NucView-positive signal was increased in ERdj5−/− MEFs treated with STA ( Fig. 3A). This suggests that ERdj5 deficiency caused cells to be more sensitive to apoptosis through mitochondrial fission.
The deletion of ERdj5 induces phenotype with features of cellular senescence. A previous report showed that mitochondrial fission through the hyperactivation of Drp1 increases the production of reactive oxygen species (ROS) in the mitochondria and finally accelerates cellular senescence 33 . To examine whether ERdj5 is involved in cellular senescence, we investigated the activity of β-galactosidase, a marker of cellular senescence, using the SPiDER-βGal kit in ERdj5 ± and ERdj5-/-MEFs. As shown by this assay, the deletion of ERdj5 increased the fluorescence of SPiDER-βGal to the same extent as 1 mM H 2 O 2 treatment for 6 h (Fig. 3B). Moreover, we assessed the involvement of ERdj5 in individual ageing with C. elegans. To monitor the lifespan of the worms, WT and DNJ-27/ERdj5 KO worms were incubated at 20 °C, and the number of surviving worms was scored every other day. The deletion of DNJ-27/ERdj5 slightly decreased the lifespan of the worms (Fig. 3C), consistent with the results regarding phenotype with the feature of cellular senescence (Fig. 3B). These results suggest the possibility that ERdj5 has important roles in mitochondrial homeostasis, which involves the maintenance of cellular senescence.

Discussion
We have shown that the downregulation of ERdj5 causes aberrant mitochondrial fission, resulting in the production of fragmented mitochondria, and revealed its mechanism; the knockdown of ERdj5 caused an increase in [Ca 2+ ] i in mammalian cells, followed by the activation of Drp1, which assembles around the mitochondria and constricts both outer and inner mitochondrial membranes.
Another group showed that an ERdj5 orthologue in worms (DNJ-27) elicited a protective effect against proteotoxicity caused by aggregation in the cytosol 25 . Moreover, the authors showed that the deletion of DNJ-27 caused severe paralysis and slow mobility in neurodegenerative disease model animals. Here, we observed  www.nature.com/scientificreports/ mitochondrial fission in the intestine in C. elegans. Cytosolic Ca 2+ signalling is essential for the contraction of muscle and peristaltic motion of the intestines. Therefore, since calcium homeostasis in muscle and intestinal cells should be maintained more strictly than in other tissues, it is reasonable that intracellular calcium homeostasis in intestinal cells was significantly disrupted due to ERdj5 deficiency and that remarkable mitochondrial disruption was observed in intestinal cells. Furthermore, in this report, we showed that the downregulation of ERdj5 also caused aberrant mitochondrial fragmentation in mammalian cells. A previous report demonstrated  www.nature.com/scientificreports/ that ER stress was strongly induced, especially in the salivary gland, by just the disruption of ERdj5. Interestingly, a recent study showed that the deletion of ERdj5 resulted in a Sjögren's syndrome-like phenotype in mice. Sjögren's syndrome induces salivary gland dysfunction that leads to xerostomia. Although the detailed mechanism is still unknown, salivary fluid secretion is prevented by abnormal calcium signals in salivary gland acinar cells 34 . Our conclusion that the downregulation of ERdj5 results in the perturbation of [Ca 2+ ] i is consistent with this pathogenetic mechanism of Sjögren's syndrome.
In this report, we found that the steady-state [Ca 2+ ] i was approximately 1.4 times higher in ERdj5-deficient cells than in WT cells (Fig. 2B). Previously, we reported the redox-assisted regulation of SERCA2b mediated by ERdj5 by showing that the deletion of ERdj5 negatively regulated the pump function of SERCA2b and resulted in a low [Ca 2+ ] ER 20 . Therefore, one of the causes by which ERdj5 deficiency increases the steady-state [Ca 2+ ] i is a decrease in SERCA2b pump function. This result is consistent with the inhibition of the SERCA2 pump with thapsigargin, and indeed, the prevented SERCA2b pump function causes mitochondrial fragmentation. However, it cannot be excluded the other possibility that the increase in [Ca 2+ ] i due to ERdj5 deficiency may change the release and leakage of Ca 2+ from the ER to the cytosol, including the change in MAM formation. It would be interesting to examine if ERdj5 is involved in cellular Ca 2+ dynamics in addition to regulating the SERCA2b pump. Previous report showed that STA treatment enhanced the release of Ca 2+ through IP3R. This activation by STA increased the transport of Ca 2+ into the mitochondria and caused depolarization of mitochondria, resulting in apoptosis 35 . As shown in Fig. 3A, the vulnerability to STA in ERdj5-deficient cells may be due to mitochondrial fragmentation and changes in Ca 2+ dynamics by ERdj5 deficiency. Additionally, in Fig. 3B The knockdown of DNJ-27/ERdj5 was reported to cause severe paralysis in C. elegans 25 . The authors of this report confirmed that DNJ-27/ERdj5 overexpression prevented paralysis in neurodegenerative model worms. Previous studies suggested a close correlation between some neurodegenerative diseases and the perturbation of intracellular calcium homeostasis [37][38][39] ; the impairment of ER membrane permeability by the accumulation of amyloid proteins was reported to increase intracellular free Ca 2+ levels 37 , and ROS caused by the accumulation of amyloids preferentially damaged some ion pumps, including SERCA family proteins, resulting in an increase in [Ca 2+ ] i 40 . Our data suggest that the decrease in cell viability under higher [Ca 2+ ] i conditions caused by the deletion of ERdj5 could enhance sensitivity to proteotoxicity (Fig. 4). Hence, the ERdj5-mediated maintenance of Ca 2+ homeostasis in both the ER and cytosol might be critical for a potential therapeutic strategy for neurodegenerative disease. Other groups have also noted the possibility that ERdj5 could serve as a novel chemotherapeutic target for cancer; the retinoid analogue fenretinide was shown to be a cancer-preventive and chemotherapeutic drug. Unlike most retinoids, fenretinide induces apoptosis in vitro 24 . The knockdown of ERdj5 by RNAi in neuroectodermal tumour cells and melanoma cells increased the apoptotic response to fenretinide. While the collapse of ER protein quality control due to the downregulation of ERdj5 was described to enhance the sensitivity www.nature.com/scientificreports/ to apoptosis in this report, we would also like to note the possibility that the perturbation of calcium by ERdj5 caused proteostasis instability, leading to apoptosis in cancer cells.
The mammalian ER contains nearly 20 members of the protein disulfide isomerase (PDI) family, including ERdj5, which together constitute a redox environment that maintains ER homeostasis. Although a luminal cysteine pair in SERCA2b is cleaved by ERdj5 via its reductase activity, the functional redundancy of ERdj5 is well unknown. A previous report showed that some ERdj5 downregulation-mediated phenotypes, including general paralysis and slow motility, were more clearly observed in C. elegans than in ERdj5-deficient mice. This suggests that some PDI family proteins or other mechanisms may complement the functions of ERdj5 in mammals but not in C. elegans. Therefore, further studies are needed to understand the maintenance of ER homeostasis through the redox network centred on ERdj5.
Cell culture and transfection. HeLa cells were kindly provided by Dr. Tamotsu Yoshimori (Osaka University, Japan). MEFs, HeLa cells (HeLa Kyoto strain), and HEK293 cells were maintained in DMEM (Thermo Scientific) containing 10% fetal bovine serum (Sigma-Aldrich), 100 U/mL penicillin, and 100 μg/mL streptomycin. Plasmids were transfected into MEFs using Lipofectamine LTX (Invitrogen). In HeLa cells, plasmids were transfected by using PEI MAX (Polysciences). The amount of DNA was adjusted to obtain equivalent expression levels of the introduced proteins in each experiment. siRNAs were transfected into MEFs using Lipofectamine RNAiMax (Invitrogen).
Immunoblotting. Samples were separated by SDS-PAGE using a 7.5% acrylamide gel and transferred to nitrocellulose membranes (GE Healthcare) for 1 h at 100 V and 4 °C. The membranes were treated with Blocking One (Nacalai Tesque) for 30 min at room temperature and then incubated sequentially with primary and secondary antibodies diluted in Blocking One in PBST. Detection was performed with SIGMAFAST BCIP/NBT (Sigma-Aldrich).
Uncropped blots are shown in (Fig. S7). . Then, the maximum fluorescence ratio R max was obtained by the fluorescence ratio of YC3.6 after the medium was replaced with HBSS containing 20 mM CaCl 2 and 1 mM ionomycin. The minimum fluorescence ratio R min was obtained after depleting the intracellular calcium ions by washing the cells with HBSS containing 1 mM EGTA, 5 mM MgCl 2 , and 1 mM ionomycin. Finally, [Ca 2+ ] i was calculated using the following equation in which the K' d value of YC3.6 and the Hill coefficient n were to be 0.25 and 1.7, respectively.

Fluorescence microscopy.
The CEPIA2 mt fluorescence in HEK 293 cells was monitored with a Varioskan LUX multi-plate reader (Thermo Scientific) with 487-nm excitation and 508-nm emission wavelengths.
Analysis of mitochondrial Ca 2+ levels was performed as previously described.
Briefly, cells were loaded with 0.8 mM Rhod-2 AM (Molecular Probes), 100 nM MitoTracker Green FM (Thermo Scientific) and 0.05% Pluronic F-127 (Molecular Probes) in DMEM containing 5 mM glucose for 30 min. Cells were washed once with recording medium [0.05% Pluronic F-127, 1.25 mM Probnecid, 1 mM CaCl 2 in HBSS], followed by incubation at room temperature for 60 min in recording medium. Culture medium was changed to recording medium, and the cells were exposed to either 20 mM CCCP or 1 mM EGTA. Fluorescence signals were observed with an LSM700 (Zeiss) confocal laser scanning microscope. Fluorescence images labeled with Rhod-2 were collected using an excitation wavelength of 555 nm. Rhod-2 fluorescence was detected using single excitation and emission filters. Quantification was performed in the area where Rhod-2 and Mito-tracker green co-localized.
Measurement of mitochondrial membrane potential. The cells were stained with JC-1 following the protocol. JC-1 was excited at 405 nm. The wavelength of JC-1 fluorescence emission changes from 555 to 488 nm in response to mitochondrial depolarization. The ratio of emission at 488 nm to that at 555 nm was evaluated as an index of mitochondrial depolarization. Fluorescence signals were observed with an LSM700 (Zeiss) confocal laser scanning microscope.
Senescence-associated β-galactosidase staining. SPiDER-βGal was purchased from Dojindo Laboratories. After MEFs had been harvested on glass-bottom dishes (Matsunami), the cells were stained with SPiDER-βGal and Hoechst 33,342 (Thermo Scientific) at 37 °C for 30 min following the protocol of the SPiDER-βGal kit. Imaging was performed with an LSM 700 confocal microscope (Zeiss) with an excitation wavelength of 488 nm (SPiDER-βGal) or 361 nm (Hoechst 33,342) and an emission wavelength of 550 nm (SPiDER-βGal) or 497 nm (Hoechst 33,342).
Lifespan assay. The embryos in hermaphroditic worms were synchronized by treatment with alkaline hypochlorite and allowed to develop for 3 days. Adult worms were incubated at 20 °C, transferred to new nematode growth medium (NGM) plates and scored every other day. A worm was judged as dead when its pharyngeal pumping ceased, and the worm did not respond to prodding with a platinum wire. Lifespan curves were plotted using Kaplan-Meier survival curves and analysed using log-rank tests. Mean lifespans were plotted and analysed using OASIS 2 (Online Application for Survival Analysis 2; https:// sbi. poste ch. ac. kr/ oasis2) 42  www.nature.com/scientificreports/