The common features of type 2 diabetes mellitus (T2DM) include insulin resistance and pancreatic β-cell deterioration. The latter determines the onset of diabetes (Rorsman and Ashcroft, 2018). A number of studies show excessive free fatty acids (FFAs) contribute to β-cell dysfunction and cell death. Palmitate, stearate, and oleate are the major FFAs in human plasma (Hagenfeldt et al., 1972). When isolated human islets are exposed to high palmitate and high glucose, β-cells show endoplasmic reticulum (ER) dilation, swollen mitochondria, several enlarged autophagic vacuoles, and distorted nuclei (Martino et al., 2012). These morphologic changes result from lipotoxic pathways in which Ca²⁺ and reactive oxygen species (ROS) play key roles. Consistent with these mechanistic findings, the Paris Prospective Study reports increased plasma FFA concentration is one of independent risk factors of T2DM in subjects with a history of impaired glucose tolerance (Charles et al., 1997). Higher fasting FFA concentrations are associated with decreased insulin secretion after an oral glucose load in both children and adults (Salgin et al., 2012).

Palmitate is known to induce oxidative stress via effects on multiple metabolic pathways. These include increased mitochondrial FFA oxidation, ceramide-disrupted electron transport through complex I and III, misfolded protein-induced ROS generation via ER oxidoreductin 1α (ERO-1α) and protein disulfide isomerase (PDI), and cytosolic diacylglycerol-protein kinase C (PKC)-NADPH oxidase pathway (Ly et al., 2017; Oh et al., 2018). Oxidative stress is associated with ER Ca²⁺ release and store depletion (Eletto et al., 2014). Since protein folding machinery requires high luminal Ca²⁺, ER Ca²⁺ depletion accumulates misfolded proteins, which then trigger ER stress responses (Back and Kaufman, 2012). Several studies emphasize ER Ca²⁺ loss and ER stress as causative factors in β-cell death (Cnop et al., 2010; Cunha et al., 2008; Karaskov et al., 2006; Lee et al., 2010). However, fewer studies look into palmitate-elicited Ca²⁺ dyshomeostasis in cytosol and other organelles, e.g., mitochondria and lysosomes. Uncontrolled elevation of cytoplasmic Ca²⁺ concentration ([Ca²⁺]_i) may trigger many undesirable signaling pathways, including PKC, Ca²⁺/calmodulin-dependent protein kinase IV, c-Jun N-terminal kinase, and mitogen-activated protein kinase, which are followed by activation of transcription factors such as ATF2, CEBP/δ, CREB, Egr-1, and CHOP (da Cunha et al., 2015). Cytosolic Ca²⁺ overload also activates calpains, caspases, proteases, phosphatases, and endonucleases, which then trigger nitrosative and oxidative stress, lipid peroxidation, and mitochondrial dysfunction resulting in apoptosis (Carafoli, 2002; Szydlowska and Tymianski, 2010).

Increased [Ca²⁺]_i can be sequestered and dispersed by active mitochondria through mitochondrial Ca²⁺ uniporter (MCU) (De Stefani et al., 2015). MCU transports Ca²⁺ into the mitochondrial matrix driven by dissipation of the electrical gradient across the inner mitochondrial membrane and this accelerates mitochondrial metabolic flux and ATP synthesis. In insulin-secreting cells, MCU-mediated Ca²⁺ uptake plays a crucial role in coupling metabolism and insulin secretion (Quan et al., 2015; Tarasov et al., 2012). In mitochondria-associated ER membrane, MCU transports released Ca²⁺ from the ER to the mitochondrial matrix (Arruda and Hotamisligil, 2015). Oxidative stress augments this Ca²⁺ transport and leads to excessive matrix Ca²⁺ accumulation and cytotoxicity (Choi et al., 2017). Therefore, mitochondrial Ca²⁺ uptake has a double-edged consequences; the beneficial effect from sequestrating and relieving [Ca²⁺]_i overload and the detrimental action of matrix Ca²⁺ burden resulting in permeability transition pore opening and cytochrome c release. This is an interesting point worth investigating in the pathogenesis of β-cell lipotoxicity.

This study clarifies the role of mitochondrial Ca²⁺ uptake via MCU in insulin-secreting cells experiencing lipotoxicity. Targeted reduction of MCU expression decreased mitochondrial ROS generation by palmitate but exacerbated both cytosolic Ca²⁺ overload and defective autophagic degradation. Surprisingly, we found palmitate upregulated MCU expression and thus enabled increased mitochondrial Ca²⁺ sequestration. This compensatory mechanism disappeared under high glucose condition. Attenuation of [Ca²⁺]_i overload by inhibiting Ca²⁺ influx protected palmitate-treated cells from autophagic blockage and cytotoxicity. These results suggest a novel compensatory mechanism in β-cells that protects against lipotoxicity and that is mediated by MCU maintenance of cytosolic Ca²⁺ homeostasis.

Reagents

All chemicals were purchased from Sigma-Aldrich (USA), unless otherwise stated. Krebs-Ringer bicarbonate buffer (KRBB) solution contains (mM): 5.5 glucose, 0.5 MgSO₄, 3.6 KCl, 0.5 NaH₂PO₄, 2 NaHCO₃, 140 NaCl, 1.5 CaCl₂, 10 HEPES, and pH 7.4 titrated with NaOH. Palmitate (#P9767; Sigma-Aldrich) was conjugated with bovine serum albumin (BSA) (#A6003; Sigma-Aldrich) in a molar ratio of 5.5:1 as described in (Xu et al., 2015).

Cell culture

Mouse insulinoma MIN6 cells (RRID:CVCL_0431) were grown in 5.5 mM glucose Dulbecco’s modified Eagle’s medium (DMEM) (#11885-084; Thermo Fisher Scientific, USA) supplemented with 10% fetal bovine serum (FBS) (#16000-044; Thermo Fisher Scientific), 100 IU/ml penicillin, 100 μg/ml streptomycin and 50 μM β-mercaptoethanol at 37°C with 5% CO₂. Before experiments, cells were seeded into corresponding plates and incubated overnight. Culture medium was then exchanged for either 5.5, 11, or 25 mM glucose DMEM with 1% FBS followed by the addition of palmitate or BSA. Experiments were performed with cells passaged 26–30 times.

Plasmid transfection

As a ER luminal Ca²⁺ fluorescent reporter, pCAG G-CEPIA1er was a gift from Franck Polleux (Addgene plasmid #105012; http://n2t.net/addgene:105012; RRID:Addgene_105012). Tandem fluorescent LC3 reporter, ptfLC3 was a gift from Tamotsu Yoshimori (Addgene plasmid #21074; http://n2t.net/addgene:21074; RRID:Addgene_21074). In the acidic environment of lysosomes, GFP is degraded but mRFP is not. Yellow and red LC3 puncta then represent autophagosomes and autophagolysosomes, respectively (Kimura et al., 2007). Approximately 6–8 h after cell seeding at 70–90% confluency, cells were transiently transfected with the plasmid using X-tremeGENE HP DNA Transfection Reagent (#6366236001; Roche, Germany) and Opti-MEM I Reduced Serum Medium (#31985-062; Thermo Fisher Scientific) as diluent according to the manufacturer’s protocol.

Confocal microscopy

Cells transfected with ptfLC3 were treated with palmitate or BSA followed by fixation with 4% paraformaldehyde for 15 min at room temperature in the dark. Images were captured using a confocal microscope (LSM 800; Zeiss, Germany) and its software (ZEN 2.3). Yellow and red puncta were quantified using ImageJ software (National Institutes of Health [NIH]; https://imagej.nih.gov/ij/) and red and green puncta were colocalized programmatically with a macro (https://imagejdocu.tudor.lu/plugin/analysis/colocalization_analysis_macro_for_red_and_green_puncta/start) (Satir and Pampliega, 2016).

Small interfering RNA (siRNA) transfection

Mouse MCU (siGENOME; #062849-01) and negative control (#SN-1002) siRNAs were purchased from Dharmacon (USA) and Bioneer (Korea), respectively. Cells were seeded on to a 6-well plate 24 h before siRNA transfection. Before the transfection, cells were washed twice with PBS and maintained in DMEM. Transfection was conducted using OptiMEM and Dharmafect 1 transfection reagent (#T-2001-03; Dharmacon) following the manufacturer’s guideline. Twenty-four hours after transfection, OptiMEM was replaced by complete medium. Cells were maintained 72 h before treatment with palmitate.

Western blot analysis

Cells were seeded on to a 6-well plate followed by treatment of palmitate with or without drugs as indicated. Cells were washed three times with cold PBS, lysed using cold RIPA buffer (#89900; Thermo Fisher Scientific) supplemented with protease and phosphatase inhibitors (#5892791001 and #4906837001, respectively; Roche), and centrifuged (17,000 rpm, 30 min). Supernatants were run on SDS-PAGE. Proteins were electroblotted on to polyvinylidene difluoride membranes (Merck Millipore, USA), blocked with either 5% BSA or 6% skim milk for 1 h at room temperature, and incubated with primary antibodies overnight at 4°C. The following primary antibodies were used: phospho-PERK (Thr980; #3179), total PERK (#3192), phospho-eIF2α (Ser51; #3597), total eIF2α (#5324), CHOP (#2895), cleaved caspase-3 (#9661), MCU (#14997), VDAC (#4866), LC3A/B (#4108), and p62 (#5114) from Cell Signaling Technology (USA); total OXPHOS antibody cocktail (#ab110413) and β-actin (#ab6276) from Abcam (UK); GAPDH (#sc-47724) from Santa Cruz Biotechnology (USA). All antibodies were applied diluted in buffer containing 0.1% Tris-buffered saline and Tween 20 (TBST) with 5% BSA. Membranes were washed with TBST four times (5 min/time), and then incubated with horseradish peroxidase-linked secondary antibody against rabbit or mouse IgG (#31460 or #31450, respectively; Thermo Fisher Scientific) supplemented with 6% skim milk for 1 h at room temperature. Membranes were washed again with TBST five times and then developed using ECL solution (Luminata Forte, #WBLUF0100; Millipore Corp., USA). Immunoreactive bands were visualized by Chemi Doc XRS+ imaging system and quantified using Image Lab 6.0 (Bio-Rad, USA).

MTT assay

MIN6 cells were plated at 5 × 10⁴ cells/well in a 96-well plate. Cells pre-treated for 30 min with drugs were then co-incubated with palmitate or control BSA for 24 h. MTT solution was prepared by dissolving 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide (#M2128; Sigma-Aldrich) in fresh DMEM medium at 0.5 mg/ml. MTT assay was initiated by replacing culture medium with 100 μl of MTT solution. The plate was wrapped in foil and placed in an incubator for 2 h. After careful removal of medium to avoid cell detachment, 100 μl of DMSO was added to each well to dissolve formazan crystals. The plate was then gently swirled for 10 min and the absorbance value of each well was measured at 570 nm and 650 nm (reference) by Epoch Microplate Spectrophotometer (Bio-Tek, USA).

Apoptotic assay

The number of cytoplasmic mono- and oligo-nucleosome-associated histone-DNA complexes in cell lysates was measured by sandwich ELISA assay (Cell Death Detection ELISA Plus kit, #11774425001; Roche). MIN6 cells (5 × 10⁴ cells/well) were seeded into 96-well plate. Cells were pre-treated with drug for 30 min then co-incubated with 500 μM palmitate or BSA for 24 h. The assay was performed by adding 200 μl of kit lysis buffer to each well and incubating for 30 min at room temperature. After centrifugation, 20 μl of supernatant was collected, and manufacturer instructions were followed. After a 5 min incubation with a peroxidase substrate, absorbance values were measured at 405 nm and 490 nm (reference) by Epoch Microplate Spectrophotometer.

Intracellular and ER Ca²⁺ measurement

MIN6 (3 × 10⁴) cells suspended in 500 μl of complete medium were seeded on to 12 mm L-poly-lysine coated coverslips embedded in a 4-well plate then treated with palmitate (500 μM, 24 h) or BSA. To measure cytosolic Ca²⁺, cells were incubated with 5 μM Fura-2 AM dissolved in KRBB for 1 h at room temperature, washed three times with KRBB, and perfused with free-Ca²⁺ KRB buffer in the presence of 200 μM sulfinpyrazone unless noted otherwise. The chamber was placed into the IX-73 inverted microscope platform (Olympus, Japan) with a camera attachment (Prime-BSI CMOS camera; Teledyne Photometrics, USA) and an illuminator (pe-340Fura; CoolLED, UK). The fluorophore was alternately excited at 340 and 380 nm while emission was recorded at 510 nm using MetaFluor 6.1 (Molecular Devices, USA). Background-subtracted 340/380 Fura-2 AM ratio reflecting [Ca²⁺]_i changes was calculated. At the end of each experiment, 10 μM ionomycin dissolved in 1.5 mM Ca²⁺ KRBB was added to induce a maximal ratio for comparison between control and palmitate-treated cells. For ER Ca²⁺ measurement, we transfected cells with G-CEPIA1er 48 h before the experiment. Fluorescence emission at 535 nm was recorded and analyzed following excitation with 488 nm.

ROS measurement

MIN6 (3 × 10⁴) cells were plated on to 12 mm L-poly-lysine coated coverslips embedded in a 4-well plate before treatment with palmitate (500 μM, 24 h) or BSA. Cells were incubated either with 2.5 μM CM-H₂DCFDA (2′-7′ dichlorofluorescin diacetate) (#C6827; Thermo Fisher Scientific) or 5 μM mitoSOX (#M36008; Thermo Fisher Scientific) for 10 min at 37°C to detect intracellular ROS or mitochondrial superoxide production, respectively. Excitation/emission wavelengths for DCF and mitoSOX were 490/535 nm and 514/560 nm, respectively. Fluorescence intensity was quantified using the IX81 inverted microscope (Olympus) with MetaMorph 6.1 (Molecular Devices).

Statistical analysis

Data were analyzed in GraphPad Prism 6.01 (GraphPad Software, USA) and presented as mean ± SD or SEM if available. Representative Ca²⁺ traces were the average of two or more Ca²⁺ recordings. Cell responses from at least three independent experiments were displayed in jitter plots in combination with bar graphs. Student’s t-test or one- or two-way ANOVA with Bonferroni post-hoc test was used. P values less than 0.05 were considered statistically significant.

Palmitate elicited Ca²⁺ and ROS-related pathogenic changes and apoptosis in MIN6 cells

First, we established a cytotoxic dose profile for palmitate in a mouse clonal pancreatic β-cell line, MIN6. Palmitate decreased cell viability, measured by MTT assay, in a dose-dependent manner (Fig. 1A). Lipotoxic cell death was confirmed by increased apoptotic DNA fragments (Fig. 1B). Palmitate markedly elevated intracellular and mitochondrial ROS accumulation (Figs. 1C and 1D). In addition, palmitate incubation abrogated cyclopiazonic acid (CPA)-induced [Ca²⁺]_i response, which reflects ER Ca²⁺ content (Figs. 1E and 1F). To demonstrate the loss of Ca²⁺ stores in the ER, we measured intraluminal Ca²⁺ using a fluorescent reporter, G-CEPIA1er. As expected, the majority of control BSA-treated cells showed the ATP-induced ER Ca²⁺ release mediated by IP₃. Subsequent application of CPA completely depleted ER Ca²⁺ stores. Interestingly, heterogeneous responses to ATP and CPA were observed among different cells with palmitate pre-incubation. One-third of palmitate-treated cells possessed an empty ER Ca²⁺ pool and agonist treatment did not release any further ER Ca²⁺. The remaining two-thirds of cells had lower ER Ca²⁺ and showed a smaller Ca²⁺ response upon ATP stimulation compared to control BSA-treated cells (Figs. 1G–1I). Depletion of ER Ca²⁺ stores strongly activated store-operated Ca²⁺ entry, especially in cells with completely emptied ER Ca²⁺ pools (Fig. 1G).

Because ER Ca²⁺ disturbance has a detrimental effect on protein folding (Back and Kaufman, 2012), we investigated alterations in the PERK-mediated ER stress response during the course of palmitate incubation. Both PERK and eIF2α showed early activation, before 12 h of palmitate treatment. This was followed by an increase in CHOP expression (Figs. 1J and 1K). Upregulation of CHOP was concomitant with a decrease in phospho-eIF2α, which is a dephosphorylation target of CHOP (Novoa et al., 2001). To demonstrate the cross-talk between ER stress and oxidative stress, we tested ROS scavenger treatment to see if their reduction of oxidative stress also reduces the severity of palmitate induced ER stress. Indeed, mitoTEMPO, a mitochondria-targeted antioxidant, reduced CHOP protein levels and reduced the extent of caspase 3 activation (Fig. 1L). Furthermore, mitoTEMPO reduced DNA fragments indicative of lipotoxic cell death (Fig. 1M).

MCU knockdown did not improve palmitate-induced ER stress nor cell death

We have previously reported cellular oxidative stress triggers MCU-mediated Ca²⁺ uptake and ER Ca²⁺ release in myocytes of Drosophila melanogaster (Choi et al., 2017). To understand the role of mitochondrial Ca²⁺ uptake in glucolipotoxicity, we first checked MCU protein levels under high glucose or palmitate conditions. MCU expression was downregulated when cells in normal glucose (5.5 mM) medium were exposed to high glucose (25 mM) for 24 h (Figs. 2A and 2B). We then incubated MIN6 cells with palmitate (500 μM) for 24 h in either 5.5 or 25 mM glucose medium. Interestingly, palmitate significantly upregulated MCU protein expression under normal glucose medium but not under high glucose, while VDAC expression did not vary in either condition (Figs. 2C and 2D). To elucidate the pathophysiological role of upregulated MCU in palmitate-induced lipotoxicity, we reduced MCU expression by RNAi targeting, and reduction in protein level was confirmed by immunoblotting (Figs. 2E and 2F). MCU knockdown led to the downregulation of oxidative phosphorylation proteins, a phenomenon that was also previously demonstrated in a rat β-cell line, INS-1E (Quan et al., 2015). Silencing of MCU abolished mitochondrial superoxide generation in response to palmitate; this suggests a critical role for Ca²⁺ uptake in mitochondrial oxidative stress (Figs. 2G and 2H). However, MCU knockdown did not rescue cells from palmitate-induced cytotoxicity, as estimated by MTT and DNA fragments assays (Figs. 2I and 2J). Furthermore, MCU silencing did not attenuate palmitate-induced CHOP upregulation and caspase 3 cleavage (Figs. 2K–2M). These results suggest that mitochondrial Ca²⁺ uptake does not participate in palmitate-induced cytotoxicity despite its contribution to ROS generation.

MCU silencing aggravated palmitate-induced cytosolic Ca²⁺ overload and defective autophagic degradation

In a pancreatic β-cell, glucose or other secretagogues increase [Ca²⁺]_i via voltage-gated Ca²⁺ channel (VGCC)-mediated influx or via Ca²⁺ release from the ER, which triggers insulin granule exocytosis (Gilon et al., 2014). The consequent removal of secretagogue relieves cytosolic Ca²⁺ load and returns the cytosol to the resting [Ca²⁺]_i level. We recorded changes in [Ca²⁺]_i using Fura-2 dye in response to high K⁺-induced depolarization. Palmitate slightly elevated basal [Ca²⁺]_i levels but strongly attenuated the [Ca²⁺]_i response upon depolarization (Figs. 3A–3C). Suppression of mitochondrial Ca²⁺ uptake by MCU knockdown further elevated basal [Ca²⁺]_i in palmitate-treated cells, which could be caused by reduced mitochondrial Ca²⁺ sequestration (Figs. 3A and 3B).

Zummo et al. (2017) showed palmitate decreases autophagic flux in insulin-secreting cells by inhibiting the fusion of lysosomes and autophagosomes. Bafilomycin A1 is an inhibitor of lysosomal V-ATPase, which blocks luminal acidification and degradation of auophagolysosomes. We also observed that the LC3-II/β-actin ratio was elevated in palmitate-treated cells but not increased any further during bafilomycin A1 treatment, implying the formation of autophagolysosomes was suppressed by palmitate (Fig. 3D). Consistent with increases in LC3, p62 protein was increased during the treatment with palmitate (Fig. 3E). We further examined palmitate-elicited autophagic inhibition by using tandem fluorescent-tagged LC3 (ptfLC3) plasmid. Confocal microscopy revealed autophagosome and autophagolysosome puncta, and these are rendered in yellow and in red, respectively. Palmitate drastically increased the average cell area covered by yellow puncta while it modestly affected that of red puncta (Figs. 3F and 3G). As a result, the ratio of yellow to red puncta was significantly higher in palmitate-treated cells compared to BSA-treated cells (Fig. 3G, inset graph). We further estimated the effects of palmitate on autophagic flux in MCU-silenced and control cells. Immunoblots revealed that MCU ablation did not rescue p62 accumulation during lipotoxicity (Fig. 3H). Consistently, the extent of yellow puncta and the ratio of yellow to red puncta in palmitate-treated cell were not reduced but instead were augmented by MCU knockdown (Figs. 3I and 3J). Collectively, these data suggest that upregulation of MCU by palmitate mitigates cellular Ca²⁺ overload and alleviates autophagic block during β-cell lipotoxicity.

Attenuation of extracellular Ca²⁺ influx via VGCC alleviated Ca²⁺ overload and cytotoxicity by palmitate

One hypothesis suggests excessive Ca²⁺ influx from extracellular space through VGCCs could contribute to the palmitate-induced toxicity in insulin-secreting cells (Choi et al., 2007). Our data also indicate that inhibition of mitochondrial Ca²⁺ sequestration results in detrimental consequences of Ca²⁺ cytotoxicity by palmitate. Hence we next tested whether attenuating [Ca²⁺]_i overload could relieve the pathologic consequences of β-cell lipotoxicity. To reduce VGCC-mediated Ca²⁺ influx, we exposed cells to a Ca²⁺ chelator, EGTA, or to VGCC inhibitors verapamil or nifedipine. As expected, co-treatment with verapamil for 24 h abated basal [Ca²⁺]_i rise by palmitate (Figs. 4A and 4B). Preventive action of verapamil on [Ca²⁺]_i overload was more prominent in MCU silenced cells (Fig. 4B). In these experiments, we washed out verapamil thoroughly before measuring maximum voltage-gated Ca²⁺ flux, which we induced using high K⁺. Pre-incubation with verapamil ameliorated the [Ca²⁺]_i response to depolarization both in control and MCU knockdown cells (Fig. 4C).

We observed that extracellular Ca²⁺ chelation with EGTA or VGCC inhibition with verapamil suppressed palmitate-induced cytosolic (Figs. 4D and 4E) and mitochondrial (Figs. 4F and 4G) ROS production. Moreover, verapamil reduced p62 accumulation during the entire period of palmitate treatment (Fig. 4H). Confocal imaging analysis with tandem fluorescent-tagged LC3 transfection revealed that verapamil pre-incubation decreased the number of yellow puncta and the ratio of yellow to red puncta in palmitate-treated cells (Figs. 4I and 4J). Finally, extracellular EGTA and VGCC inhibitors each rescued cell viability (Fig. 4K) and prevented apoptotic cell death (Fig. 4L). These results suggest that cytosolic Ca²⁺ burden due to extracellular Ca²⁺ influx is a critical pathogenic mechanism for autophagy defects and β-cell lipotoxicity. This Ca²⁺ overload by palmitate could be compensated by MCU-mediated Ca²⁺ sequestration.

In this study, we demonstrated ER Ca²⁺ release and store depletion by palmitate related to oxidative stress. ER Ca²⁺ depletion by palmitate leads to ER stress responses and cell death in mouse clonal pancreatic β-cells. These results are consistent with reports in other insulin-secreting cells (Gwiazda et al., 2009; Hara et al., 2014) and our previous observation in mouse podocytes (Xu et al., 2015). Several mechanisms have been suggested for palmitate-induced ER Ca²⁺ depletion. Binding to the fatty acid receptor coupled to Gq-protein (GPR40) can trigger IP₃-mediated ER Ca²⁺ release (Mancini and Poitout, 2013). Oxidative stress induced by palmitate may stimulate ER Ca²⁺ release channels directly (Li et al., 2009) or promote IP₃ generation via phospholipase Cγ (Weissmann et al., 2012). Released Ca²⁺ is taken up by sarco-/endoplasmic ER Ca²⁺ ATPase (SERCA) with ATP-consuming active processes. However, palmitate is known to decrease SERCA expression and activity, consequently reducing ER Ca²⁺ uptake and luminal Ca²⁺ content (Gustavo Vazquez-Jimenez et al., 2016; Hara et al., 2014; Zhang et al., 2014). Lowered Ca²⁺ in the ER deteriorates chaperone function. The accumulation of misfolded protein further aggravates oxidative stress via ERO-1α and PDI (Tu and Weissman, 2004). This positive feedback loop between oxidative stress and ER stress could activate pro-apoptotic downstream effectors in β-cell lipotoxicity.

We intended to scrutinize the alterations in ER Ca²⁺ content by using Ca²⁺ sensing fluorescent protein expressed exclusively in the ER lumen. This sensor successfully detected purinoceptor-mediated ER Ca²⁺ release, emptying of Ca²⁺ stores by ER Ca²⁺ ATPase inhibition, and Ca²⁺ refilling through store-operated Ca²⁺ entry (SOCE). We observed two patterns of ER Ca²⁺ disturbance by palmitate in insulin-secreting cells: (i) marked reduction but not depletion and (ii) complete loss of Ca²⁺ in the ER reservoir (Fig. 1G). Particularly, cells with the latter pattern have a steep [Ca²⁺]_i rise upon extracellular Ca²⁺ exposure; this implies full activation of SOCE. Previously, we observed a marked STIM1 oligomerization leading to persistent SOCE activation by palmitate in mouse podocytes (Xu et al., 2015). Furthermore, VGCC-mediated Ca²⁺ influx was also triggered by palmitate in insulin-secreting cells (Choi et al., 2007). The Ca²⁺ influx through SOCE and VGCC together with ER Ca²⁺ release might result in sustained elevated [Ca²⁺]_i in palmitate-treated cells (Gwiazda et al., 2009). Therefore, it is reasonable to infer that cytosolic Ca²⁺ overburden plays a pathogenic role in β-cell lipotoxicity.

In this study, we observed an increase in MCU expression induced by palmitate that has not been previously reported as far as we know. This upregulation could be a compensatory mechanism for cells to handle cytosolic Ca²⁺ overload in lipotoxicity. We directly measured the palmitate-induced elevation of basal [Ca²⁺]_i and noted that MCU silencing remarkably exacerbated this [Ca²⁺]_i burden (Figs. 3A and 3B). Mitochondrial Ca²⁺ uptake via MCU could sequester Ca²⁺ and relieve Ca²⁺ accumulation in peri-ER microdomains. Mitochondria are known to decrease movement and gather at high Ca²⁺ microdomains. Therefore, mitochondria can sequester Ca²⁺ from localized microdomains to avoid cytotoxicity (Yi et al., 2004). It has been well known that high glucose exacerbates lipotoxicity in a synergistic manner in the progress of T2DM. Interestingly, our study showed that a high glucose environment decreased MCU expression and abated palmitate-mediated MCU upregulation. We speculate that high glucose eliminates the compensatory mechanism involving MCU and causes further serious consequences related to Ca²⁺ overload in β-cell lipotoxicity. This is the first proposal of a hypothesis for glucolipotoxicity.

We observed a significant reduction in mitochondrial superoxide generation by MCU knockdown, consistent with results from many other studies (Panahi et al., 2018; Ren et al., 2017; Tomar et al., 2019). This reduction could be attributed to the decrease in mitochondrial metabolism related to suppressed TCA cycle dehydrogenases and NADH shuttle system, which is sensitive to matrix Ca²⁺ (Denton, 2009). In addition, MCU knockdown correlates with downregulation of electron transport chain proteins (Fig. 2E), and this in turn could also decrease superoxide production. Unexpectedly, however, palmitate-induced ER stress and apoptosis were not prevented nor aggravated by MCU silencing. The ratio of yellow to red puncta from tandem fluorescent-tagged LC3 also became substantially higher in palmitate-treated MCU knockdown cells, indicating a possibly worsened defect in autophagolysosome formation (Park et al., 2014). Increased p62 accumulation upon MCU silencing supports the notion of deterioration in blocking autophagy. It is conceivable the benefits of attenuating mitochondrial Ca²⁺ uptake were offset by the negative consequences of impaired cytosolic Ca²⁺ clearance. Notably, all these pathologic alterations in palmitate-treated cells were rescued by agents that reduce Ca²⁺ influx from extracellular compartments, e.g., EGTA or VGCC inhibitors such as verapamil and nifedipine (Fig. 4). Our findings are consistent with a previous report that cytosolic Ca²⁺ overload impedes the fusion between autophagosomes and lysosomes in HepG2 cells (Park and Lee, 2014).

In pancreatic β-cells, ATP-sensitive K⁺ channel (K_ATP)-dependent [Ca²⁺]_i increase via VGCC is an important signal for glucose-stimulated insulin secretion (Wollheim and Maechler, 2002). In this study, depolarization-induced [Ca²⁺]_i increase was blunted by palmitate in MIN6 cells (Fig. 3A), and cells in this state displayed impaired insulin secretion upon nutrient stimulation. It has been known that [Ca²⁺]_i increase via VGCCs can suppress itself due to Ca²⁺-dependent inactivation (Haack and Rosenberg, 1994). We infer that elevated [Ca²⁺]_i levels under lipotoxic conditions could inactivate VGCC directly and suppress K_ATP-dependent Ca²⁺ signal generation for insulin secretion. This may be the pathogenic mechanism responsible for insufficient insulin secretion and glucose intolerance in T2DM patients. Surprisingly, pretreatment of verapamil effectively recovered not only palmitate-induced [Ca²⁺]_i overload but also restored depolarization-triggered [Ca²⁺]_i response to normal. Preventive treatment with verapamil on β-cell lipotoxicity might result from beneficial actions of [Ca²⁺]_i regulation, making verapamil an effective preventive or therapeutic target for T2DM.

We demonstrated in this study that mitochondrial Ca²⁺ uptake via MCU upregulation could be an essential compensatory mechanism whereby cells alleviate cytosolic Ca²⁺ overload. Therefore, decreased mitochondrial Ca²⁺ sequestration due to mitochondrial dysfunction may aggravate palmitate-induced autophagy defects. Based on our observations, we suggest cytosolic Ca²⁺ overload in lipotoxicity could be relieved by i) suppressing Ca²⁺ influx, ii) accelerating mitochondrial Ca²⁺ clearance, or iii) enhancing Ca²⁺ recycling into the ER. Further investigation related to maintaining Ca²⁺ homeostasis may provide the best strategy against pancreatic β-cell failure and metabolic diseases.

We thank the following authors: Daniel J. Shiwarski (University of Pittsburgh), Ruben K. Dagda (University of Nevada School of Medicine), and Charleen T. Chu (University of Pittsburgh) for sharing their ImageJ macro.

This work was supported by the Medical Research Center Program (2017R1A5A2015369) and National Research Foundation of Korea (NRF) Grant (2016R1A2B4014565) from Ministry of Science, ICT.

The authors have no potential conflicts of interest to disclose.