AMPK Regulates ER Morphology and Function in Stressed Pancreatic β-Cells via Phosphorylation of DRP1

Abstract

Experimental lipotoxicity constitutes a model for β-cell demise induced by metabolic stress in obesity and type 2 diabetes. Fatty acid excess induces endoplasmic reticulum (ER) stress, which is accompanied by ER morphological changes whose mechanisms and relevance are unknown. We found that the GTPase dynamin-related protein 1 (DRP1), a key regulator of mitochondrial fission, is an ER resident regulating ER morphology in stressed β-cells. Inhibition of DRP1 activity using a GTP hydrolysis-defective mutant (Ad-K38A) attenuated fatty acid–induced ER expansion and mitochondrial fission. Strikingly, stimulating the key energy-sensor AMP-activated protein kinase (AMPK) increased the phosphorylation at the anti-fission site Serine 637 and largely prevented the alterations in ER and mitochondrial morphology. Expression of a DRP1 mutant resistant to phosphorylation at this position partially prevented the recovery of ER and mitochondrial morphology by AMPK. Fatty acid–induced ER enlargement was associated with proinsulin retention in the ER, together with increased proinsulin/insulin ratio. Stimulation of AMPK prevented these alterations, as well as mitochondrial fragmentation and apoptosis. In summary, DRP1 regulation by AMPK delineates a novel pathway controlling ER and mitochondrial morphology, thereby modulating the response of β-cells to metabolic stress. DRP1 may thus function as a node integrating signals from stress regulators, such as AMPK, to coordinate organelle shape and function.

The endoplasmic reticulum (ER) is the site of proinsulin biosynthesis and folding and thus plays an important role in insulin production and secretion (1). Considering the vital functions of the ER in β-cells, it may be expected that ER dysfunction plays an imperative role in the pathophysiology of type 2 diabetes (T2D); indeed, recent studies assign an important role to the ER in the development of β-cell failure (2).

Accumulation of misfolded proteins in the ER results in ER stress and consequently activation of the unfolded protein response (UPR). The latter is an adaptive signaling cascade aimed to decrease ER protein load and enhance folding capacity, while continuous UPR activation may lead to apoptosis (3). Morphological alterations of the ER alongside increased expression of UPR markers in β-cells have been documented in animal models of diabetes and humans with T2D (4, 5). ER stress is often accompanied by changes in ER morphology; however, the mechanisms involved and the impact of ER morphology on cellular function are poorly understood (6, 7).

The ER consists of a membrane network surrounding the nucleus and forming tubular and flat or sack-like structures (called sheets or cisternae) in the cytoplasm (8). Several proteins involved in ER membrane curvature stabilization and in regulation of interconnections of ER tubules have been identified (9). Little is known about the formation of cisternae, although some of the tubule-shaping proteins may also be involved in stabilizing the edges of these structures (10). The ER and mitochondria exhibit tightly coupled dynamics and have extensive contacts. A recent report suggested that ER tubules define the position of mitochondrial fission sites, preceding the recruitment of dynamin-related protein 1 (DRP1) (11). DRP1 is a ubiquitous protein involved in the regulation of mitochondrial fission in physiology and stress (12). The mechanisms that modify ER morphology in stressed mammalian cells are less identified. The saturated free fatty acid (FFA) palmitate has been shown to induce ER expansion and stress in several cell types including pancreatic β-cells (6, 7, 13). Palmitate-induced β-cell stress leads to dysfunction and apoptosis in vitro and in vivo (14); this may have implications for obesity and T2D that are often associated with elevated FFA concentrations (15).

Herein we studied the mechanisms involved in the regulation of ER morphology and function in β-cells, as well as the interactions with mitochondria under stressful conditions. We found that the cellular energy sensor AMP-activated protein kinase (AMPK) regulates FFA-induced ER morphology changes, as well as mitochondrial fragmentation. We identified DRP1 as an ER resident protein and show it is a downstream target of AMPK. Stimulation of AMPK inhibited DRP1, prevented ER and mitochondrial morphology changes, and improved β-cell function and survival under lipotoxic conditions.

Materials and Methods

Islet isolation and cell culture

Islets were isolated from 12-week-old male C57BL6 mice by collagenase injection to the bile duct. Animal use was approved by the Institutional Animal Care and Use Committee of the Hebrew University and the Hadassah Medical Organization. Human islets (from four donors aged 45 ± 9 y) were obtained from the Clinical Islet Laboratory (University of Alberta/Alberta Health Services). The islets were cultured 27 to 98 hours before shipment to Israel in CMRL1066 media. β-Cell viability in the different isolations was 78% to 85%. Islets were cultured intact overnight and then dispersed by 0.05 mg/mL trypsin (Biological Industries) and plated on eight-chamber borosilicate plates (Nunc). RPMI 1640 culture media were supplemented with 10 mmol/L glucose, 10% fetal bovine serum (FBS), 100 IU/mL penicillin, 100 μg/mL streptomycin, and 2 mmol/L l-glutamine (Biological Industries) and used for islets. Experiments were performed after culture for 2 to 3 days at 37°C in a humidified atmosphere containing 5% CO₂.

INS-1E β-cells (passage 65–95) were grown in RPMI 1640 medium supplemented with 10% FBS, 1 mmol/L sodium pyruvate, 2 mmol/L l-glutamine, 10 mmol/L HEPES, 0.05 mmol/L 2-betamercaptoethanol, 100 IU/mL penicillin, and 100 μg/mL streptomycin (Biological Industries). Mouse embryonic fibroblasts (MEFs) were grown in DMEM supplemented with 10% FBS, 100 IU/mL penicillin, and 100 μg/mL streptomycin.

Reagents

The palmitate-BSA, methylpalmitate-BSA, or oleate-BSA solutions were prepared by dissolving the sodium salt of FFA (Sigma) at a concentration of 10 mmol/L in 11% BSA (Sigma) in a stirred water bath at 37°C for 16 hours. The FFA-BSA solution pH was adjusted to 7.4, filtered through a sterile 0.22-μm filter, and stored at −20°C; it was diluted 1:20 in culture media before use. The molar ratio of FFA:BSA was 6:1. The following compounds were obtained from Sigma: BSA, thapsigargin, tunicamycin, palmitic acid, methylpalmitate, oleic acid, 5-amino-1-β-d-ribofuranosyl-imidazole-4-carboxamide (AICAR), metformin, and carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP). AICAR and metformin were applied to cells overnight prior to exposure to FFA.

Fluorescent constructs and dyes

The following constructs were used to express fluorescent proteins: GFP-wild type and S656A Drp1 construct (S637 in human DRP1), in which the serine residue was substituted by alanine (Stefan Strack, University of Iowa) (16), eYfp-ER (Clontech), Rfp-ER (Felipe Pimentel Muinos, University of Salamanca) (17), Gfp-proinsulin (Louis Philipson, University of Chicago) (18), mCherry-calnexin (Volodymyr Korkhov) (19), turquoise-ER (Dorus Gadella, Swammerdam Institute for Life Sciences) (20). Mitochondria were labeled with the fluorescent dyes tetramethylrhodamine-ethyl-ester (TMRE) (10 nmol/L, 1 hour preincubation, present during imaging) or Mitotracker red (200 nmol/L, 30 min incubation, washed out before imaging) (Invitrogen).

Viruses

Adenovirus for expression of human AMPKα1 dominant negative (DN) and cytomegalovirus null as control was purchased from Applied Biological Materials and used 3 days after infection at multiplicity of infection (MOI) 100. Inhibition of AMPK activity was verified by Western blotting for phosphorylated acetyl-CoA carboxylase (ACC). Adenovirus for expression of DRP1 dominant-negative (Ad-K38A) and photo-activated GFP (Ad-PAGFP) as control were previously described (21). An adenovirus expressing the DRP1 mutant, S656A DRP1 (S636 in human DRP1), in which the serine residue was substituted by alanine, was obtained from Dr Stefan Strack, University of Iowa (15). β-Cells were treated according to the experimental protocol 3 days after infection at multiplicity of infection (MOI) 250. Expression was verified by Western blotting for total DRP1. Baculovirus expressing ER-Gfp was purchased from Invitrogen.

Confocal microscopy

Primary mouse and human islet cells and INS-1E β-cells were imaged with a ×63 oil objective on a Zeiss LSM 710 confocal microscope equipped with an incubator. ER and mitochondrial morphology were analyzed using Metamorph software (Molecular Devices). Image analysis was performed on multiple consecutive 0.2-μM confocal sections, allowing definitive distinction between cells with continuous thin ER structures (defined as tubular ER) or with large cisternae-like structures (expanded ER). To quantify ER expansion, we analyzed ER images of cells expressing ER-Gfp. Images were thresholded above background and the average diameter of tubular ER in control cells was calculated and used as a reference. ER was defined as expanded when >50% of the ER showed a diameter at least three times greater than the reference. These criteria were validated in INS-1E cells and MEFs treated with the ER stressor thapsigargin that was used as a positive control. The images were analyzed by two independent observers: one of them was blinded to treatment conditions. ER imaging criteria were uniformly used in all experiments, allowing objective assessment of ER morphology under the different treatment conditions. Mitochondria morphology was analyzed and designated tubular or fragmented, as previously described (21, 22). Scatter-plot graphs of GFP-DRP1 vs turquoise-ER were obtained with ImageJ software (NIH) and overlap coefficients and correlation R were obtained by Zen software (Zeiss).

Super-resolution imaging

For super-resolution imaging we used a Nikon N-SIM confocal microscope that allows a resolution of 85 nm. INS-1E β- cells were seeded on Ibidi 35-mm plates, with a 1.523 refractive index glass coverslip bottom (170 ± 10 μm thickness). Before imaging, the point-spread function was visualized with 100-nm fluorescence beads to adjust the correction ring of the objective to the coverslip thickness. A ×100 oil objective with a numerical aperture value of 1.49 was used. The raw data were examined carefully, and the optical grid pattern was clearly visible in all sets of images that were used. Images were reconstructed with NIS-Elements software (Nikon).

Electron microscopy

Cultured mouse islet cells or INS-1E β-cells were fixed using a solution of cold 2.5% glutaraldehyde, 2% formaldehyde, and 0.1 mol/L acodylic acid, cut into ultrathin sections, and imaged with a transmission electron microscope (Tecnai 12 TEM, Phillips).

Isolation of ER and mitochondria

Cells were broken by osmotic swelling and Dounce homogenization, and remaining intact cells and nuclei were spun down at 1000g for 10 minutes. The supernatant was centrifuged at 12 000g for 15 minutes to obtain a mitochondrial pellet. The postmitochondrial supernatant was centrifuged at 100 000g to obtain an ER pellet. ER enrichment was verified with a kit measuring the ER resident reduced NAD phosphate–cytochrome C reductase activity (Sigma). Mitochondrial enrichment was verified by Western blotting for the mitochondrial proteins complex II and cytochrome C. The cytosol marker chosen was glyceraldehyde-3-phosphate dehydrogenase (GAPDH).

AMPK-DRP1 phosphorylation assay

Phosphorylation of human DRP1 by AMPK was tested using human recombinant Drp1 (Novus) and human AMPKα1β1γ1 (GenScript) using an HTRF Transcreener ADP kit (Cisbio) and the known AMPK substrate SAMS as positive control. Data were recorded by Pherastar Plus (BMG LABTECH). The experiment was performed at the Genscript facility.

Western blotting

Protein expression in INS-1E β-cells was assessed by standard Western blotting using antibodies against IRE1α, phosphorylated PERK, phosphorylated eIF2α, phosphorylated JNK (Thr183/Tyr185), total and phosphorylated (S79) ACC, total and phosphorylated (Thr172) AMPK, total and phosphorylated (S616/637) Drp1, cytochrome C, HSP90, and caspase-3 (Cell Signaling); GAPDH, C/EBP homologous protein (CHOP) (Santa Cruz Biotechnology); and Complex II (Mitosciences). X-ray film densitometry was used for quantification (ImageMaster VDS-CL; Amersham Pharmacia Biotech).

Quantitative real-time RT-PCR

RNA was extracted from INS-1E cells using Bio Tri RNA (Biolab). Samples of 1 μg total RNA were reverse-transcribed using Moloney murine leukemia virus reverse transcriptase (Promega). Quantitative real-time RT-PCR for spliced Xbp1 was performed with StepOnePlus Real-Time PCR System, using the Power SYBR Green Gene Expression Assay (Applied Biosystems). All samples were analyzed in triplicate; GAPDH (Applied Biosystems) was used as internal control.

Apoptosis ELISA assay

INS-1E cells plated in 96-well plates were treated and then lysed, and oligonucleosomes in the cytosol, indicative of apoptosis-induced DNA degradation, were quantified using the Cell Death Detection ELISA^PLUS kit (Roche Diagnostics) according to the manufacturer's instructions.

Oxygen consumption

Oxygen consumption was measured in INS-1E cells in V7 culture plates in an XF24 respirometer (Seahorse Bioscience). One hour before measurements, culture medium was replaced with assay medium containing 3 mmol/L glucose, 0.8 mmol/L Mg²⁺, 1.8 mmol/L Ca²⁺, 143 mmol/L NaCl, 5.4 mmol/L KCl, 0.91 mmol/L NaH₂PO₄, and 15 mg/mL phenol red (Seahorse Bioscience). Measurements were recorded before and consecutively after stimulation with 20 mmol/L glucose and 1 μmol/L FCCP.

Insulin and proinsulin content and secretion

Islets were recovered overnight and then groups of islets were treated with and without 0.5 mmol/L palmitate and/or 1 mmol/L AICAR for 16 hours. Insulin response to glucose was evaluated by static incubation of islets in four-chamber culture plates (Nunclon δ Multidishes; Nunc A/S) with 20 islets per chamber in quadruplicate. Batches of islets were preincubated for 60 minutes in RPMI 1640 containing 3.3 mmol/L glucose and then were consecutively incubated at 3.3 mmol/L and 16.7 mmol/L glucose for 1 hour at 37°C in 1 mL of modified Krebs-Ringer bicarbonate buffer containing 20 mmol/L HEPES and 0.25% BSA. Medium was collected at the end of the basal (3.3 mmol/L glucose) and stimulated (16.7 mmol/L glucose) incubations, centrifuged, and frozen at −20°C pending insulin assay. Islet insulin and proinsulin content were determined by RIA and ELISA, respectively, in extracts of islets subjected to repeated freeze-thaw cycles in 1.5-mL microfuge tubes containing 450 μL 0.1% BSA in 0.1 N HCl. Mouse insulin immunoreactivity in islet extracts and in the medium was determined using a rat RIA kit (Millipore) and islet proinsulin content using an ELISA kit (Mercodia). The antiserum used for proinsulin measurement does not cross-react with insulin or C-peptide.

Statistical analysis

Data are given as means ± SEM. The differences between multiple group means were analyzed by one-way ANOVA with post-hoc Bonferroni corrected two-tailed t test. Two-tailed unpaired Student t test was used to compare differences between two groups. A P value of less than .05 was considered significant.

Results

Induction of ER expansion by ER stressors in mouse islets and clonal β-cells

We employed several imaging techniques to study ER morphology in ER stress. Primary mouse β-cells were infected with a baculovirus expressing ER-Gfp and then treated with palmitate or the Sarco/endoplasmic reticulum Ca²⁺-ATPase inhibitor thapsigargin for different periods of time followed by confocal microscopy and EM analysis. In nonstressed β-cells, ER morphology was mainly tubular; treatment with palmitate (0.5 mmol/L) or thapsigargin (300 nmol/L) for 6 hours induced marked ER expansion (Figure 1). EM, super-resolution confocal microscopy, and detailed z-stack imaging of sections spanning 1 to 2 μm showed that under stressful conditions the fine tubular structure of ER was replaced by sack-like cisternae (Figure 1, A–C). At this time point, 60% and 45% of cells treated with palmitate or thapsigargin, respectively, acquired ER cisternae shape (Figure 1D). Similar findings were observed in INS-1E cells labeled with eYFP-ER and by EM (Supplemental Figure 1, published on The Endocrine Society's Journals Online web site at http://mend.endojournals.org). ER enlargement appeared in ∼70% of INS-1E cells already after 3 hours of palmitate or thapsigargin treatment; thereafter, the ER appeared fragmented in a minority of the cells, most probably secondary to apoptosis (Supplemental Figure 1). The polyunsaturated FFA oleate and the nonmetabolizable FFA methylpalmitate did not elicit morphological changes (not shown). Of note, palmitate-induced ER expansion appeared prior to the development of apoptosis and was reversible following removal of palmitate from the culture medium (Supplemental Figure 2A). By contrast, the morphological changes induced by thapsigargin were not reversed, probably because thapsigargin binding to Sarco/endoplasmic reticulum Ca²⁺-ATPase is irreversible. ER changes were similar in mouse β-cells and in INS-1E cells; therefore, the latter were used for further mechanistic studies.

Figure 1.

Effects of ER stressors on ER morphology in primary β-cells. Primary mouse β-cells expressing GFP-ER were treated at 11.1 mmol/L glucose with 0.5 mmol/L palmitate or 300 nmol/L thapsigargin for 6 and 24 hours. ER morphology was analyzed by confocal microscopy (A–C). Distance between sections in the z-stack confocal imaging is indicated (μm). Note that cisternal structures are continuous throughout several sections under palmitate and thapsigargin treatment. (D) Quantification of the number of primary mouse β-cells with expanded (cisternal) ER morphology (C, control; P, palmitate; T, thapsigargin). Results are means ± SEM of four (6 h) and six (24 h) separate experiments analyzing 20 to 40 cells imaged per treatment. *, P < .05; **, P < .01. (E) EM images of primary mouse β-cells. Yellow square indicates zoom area. Arrows point to ribosomes on ER membrane. Size bar, 500 nm.

ER stress and regulation of ER morphology

ER expansion peaked at 3 to 6 hours of palmitate treatment. We therefore analyzed UPR activity at these time points. As expected, treatment with palmitate stimulated the UPR, evident by increased eIF2α phosphorylation, CHOP expression, spliced Xbp1, as well as JNK and cJUN phosphorylation (Figure 2A). We have previously shown that high glucose amplifies palmitate-induced ER stress (glucolipotoxicity) (23); however, treatment with high glucose alone or adding high glucose to palmitate did not modify ER morphology (Supplemental Figure 2B). To study further whether ER stress per se causes ER expansion, we treated β-cells with a cytokine cocktail (IL-1β+TNFα+interferon γ) or tunicamycin, which induces ER stress by inhibiting N-linked protein glycosylation. Although as expected cytokines and tunicamycin stimulated different parameters of cellular stress (Supplemental Figure 3 and Figure 2B), they did not induce any ER expansion. At 24 hours, cytokines and tunicamycin induced ER fragmentation, most probably reflecting apoptotic cells (Figure 2, C and D; cytokines effects not shown). These findings imply that modulation of ER stress per se does not necessarily entail modified ER morphology.

Figure 2.

Effect of ER stress on ER cisternal expansion. (A) Effect of palmitate on UPR activity. INS-1E β-cells were treated at 11.1 mmol/L glucose with control media or 0.5 mmol/L palmitate for different periods of time. CHOP expression, eIF2α, JNK, and c-JUN phosphorylation analyzed by Western blot (n = 4–9). Spliced Xbp1 was analyzed by real-time PCR (n = 5–6). (B–D) Effect of tunicamycin on ER morphology. INS-1E β-cells expressing GFP-ER were treated at 11.1 mmol/L glucose without and with 1 μg/mL tunicamycin for 6 and 24 hours followed by confocal microscope analysis of ER morphology. (B) Tunicamycin effects on the expression and phosphorylation of different UPR markers. (C) Representative confocal imaging of the ER of tunicamycin-treated β-cells. (D) The percentage of cells with tubular, expanded, or fragmented ER is shown. Results are expressed as means ± SEM of three to five separate experiments analyzing 46 to 76 cells and compared to control cells at the same time point. (E) Confocal microscopy of WT, IRE1α, or Xbp1 deficient MEFs expressing eYFP-ER and treated with control media or 0.5 mmol/L palmitate for 6 hours. White square indicates zoom area. (F, G) Quantification of ER morphology in untreated and palmitate-treated MEFs. Results are expressed as means ± SEM of five to six separate experiments analyzing 15 to 46 cells per treatment. *, P < .05; **, P < .01.

IRE1α has been implicated in the regulation of ER morphology (24). To study its role in stressed cells, we used MEFs derived from IRE1α and Xbp1 knockout mice. These contained fewer flat cisternae-like structures than wild-type (WT) MEFs (Figure 2, E and F), suggesting that spliced Xbp1 regulates ER morphology in nonstressed cells. Similar to β-cells, palmitate induced ER expansion in WT MEFs; strikingly, the morphological response to palmitate in IRE1α and Xbp1 KO cells was similar to that in WT cells (Figure 2, E and G). Collectively, these findings indicate that stimulation of the UPR is not sufficient, nor necessary, for stress-induced ER enlargement.

DRP1 is an ER resident and regulates ER morphology in stressed β-cells

DRP1 is a mitochondrial fission protein; its inhibition prevents palmitate-induced mitochondrial fission in β- cells (22). DRP1 is recruited from a cytosolic pool to mitochondrial fission sites, which are often present in areas of ER-mitochondria interactions (25); therefore, we studied the intracellular localization of DRP1 and its role in regulating ER morphology in stressed β-cells. First, we labeled INS-1E β-cells with GFP-DRP1, the ER marker turquoise-ER and the mitochondrial dye TMRE. As expected, we observed intense GFP-DRP1 punctae on mitochondria, indicative of mitochondrial fission sites (Figure 3A). In addition, there was considerable colocalization of the DRP1-GFP and the ER marker, indicating that DRP1 is also localized to the ER (Figure 3, A and B). The overlap and correlation coefficients were 0.83 ± 0.02 and 0.76 ± 0.03, respectively. To confirm this, we performed subcellular fractionation and isolated cytosol, mitochondria, and ER from nontransfected INS-1E cells (Figure 3B). The purity of the fractions was confirmed by Western blot for specific cytosol and mitochondria markers and by analyzing cytochrome C reductase activity for ER purity (Figure 3, C and D). DRP1 was expressed in both the cytoplasm and the ER and to a much lesser extent mitochondria. Thus, we conclude that DRP1 is a bona fide ER resident in β-cells.

Figure 3.

Intracellular localization of DRP1. (A) Colocalization of DRP1 and ER in INS-1E β-cells. DRP1 and ER are visualized by expression of GFP-DRP1 and turquoise-ER and mitochondria by staining with the dye TMRE. (B) Scatter plots of GFP-DRP1 vs turquoise-ER (of images in [A]). (C) Subcellular localization of DRP1 by Western blotting. Cytosol, mitochondria, and ER were isolated from INS-1E β-cells, followed by Western blotting for DRP1; GAPDH used as cytosol marker and complex II and cytochrome C for mitochondria. (D) Cytochrome C reductase activity in cytosolic, mitochondrial, and ER fractions.

Next, we inhibited DRP1 action by adenoviral expression of DRP1 DN (Ad-K38A) (Figure 4). Expression of DRP1 DN did not affect the morphology of the ER, whereas it induced mitochondrial hyperfusion as expected (Figure 4B), although mitochondrial function reflected by oxygen consumption was not affected (not shown), suggesting that DRP1 has a minor role in the regulation of ER morphology and mitochondrial function in nonstressed β-cells. By contrast, DRP1 DN markedly reduced ER expansion, as well as mitochondrial fragmentation, induced by palmitate (Figure 4, B and D). DRP1 was present in mitochondrial fission sites (Figure 4E) and the number of GFP-DRP1 punctae per cell decreased in cells where mitochondria were fragmented (Figure 4F). Collectively, these findings suggest that DRP1 mediates palmitate-induced ER expansion and mitochondrial fragmentation, followed by dissociation from mitochondrial fission complexes and departure from these sites.

Figure 4.

Effect of DRP1 inhibition on palmitate-induced ER expansion and mitochondrial fragmentation. INS-1E β-cells were transduced with an adenovirus vector expressing DN DRP1 or with photoactivatable GFP used as control (Ctrl). (A) DRP1 expression analyzed by Western blotting. (B) INS-1E β-cells expressing eYFP-ER were treated at 11.1 mmol/L glucose with control media or 0.5 mmol/L palmitate for 6 hours and then stained with the mitochondrial dye TMRE. White square indicates zoom area. (C, D) Quantification of palmitate-induced ER expansion and mitochondrial fragmentation in INS-1E β-cells expressing DRP1 DN and control cells. Results are expressed as means ± SEM of five to six separate experiments analyzing 11 to 66 cells per treatment. (E) DRP1 is present in palmitate-induced mitochondrial fission sites. Cells expressing GFP-DRP1 and stained with the mitochondrial dye TMRE. (F) Quantification of GFP-DRP1 punctae on mitochondria. Results are means ± SEM of three separate experiments analyzing 6 to 23 cells. *, P < .05; **, P < .01.

Regulation of DRP1 phosphorylation by β-cell stressors and AMPK

DRP1 function is regulated by several posttranslational modifications including phosphorylation (26). Phosphorylation at S616 is associated with increased activity of DRP1 (profission), whereas phosphorylation at S637 is linked to reduced activity (antifission) (27). This was confirmed in β- cells by treatment of INS-1E cells with the mitochondrial uncoupler FCCP, which induces mitochondrial fragmentation (Figure 5A). As expected, FCCP markedly decreased S637 phosphorylation (Figure 5B). Palmitate mildly decreased S616 and S637 phosphorylation (Figure 5, C and D), which may suggest that FFA stimulation of DRP1 is not mediated via modulation of DRP1 phosphorylation. By contrast, high glucose and cytokines that did not expand the ER increased S637 phosphorylation and S637/S616 phosphorylation ratio (Supplemental Figure 3). Hence, we hypothesized that DRP1 S637 regulates ER morphology, its phosphorylation preventing ER expansion. To test this hypothesis, we attempted to modulate DRP1 S637 phosphorylation and studied the impact on palmitate-induced ER expansion. AMPK plays an important role in the regulation of nutrient stress through its kinase activity (28); thus, we hypothesized that AMPK may affect DRP1 phosphorylation, thereby modulating ER morphology. Pharmacological activation of AMPK by AICAR increased phosphorylation at S637 in both the absence and the presence of palmitate, whereas S616 phosphorylation and total DRP1 expression were unaffected (Figure 5, C and D). We further studied whether AMPK directly phosphorylates the DRP1 S637 site by an in vitro phosphorylation assay using human AMPKα1β1γ1 and DRP1 proteins. The AMPK substrate SAMS was used as a positive control and showed robust phosphorylation by AMPK (Figure 5E). In contrast, DRP1 was not phosphorylated by AMPK in vitro, indicating that its cellular effect on S637 phosphorylation is indirect.

Figure 5.

AMPK regulation of DRP1 phosphorylation. (A) Mitochondrial fragmentation induced by FCCP. INS-1E β-cells stained with Mitotracker red. (B) Effect of mitochondrial fragmentation by FCCP on DRP1 phosphorylation. INS-1E β-cells were treated with FCCP for 60 minutes followed by Western blotting for total DRP1 and DRP1 S616 and S637 phosphorylation. (C) AMPK activation by AICAR increases DRP1 S637 phosphorylation. INS-1E β-cells were treated at 11.1 mmol/L glucose with control medium or 0.5 mmol/L palmitate with and without 1 mmol/L AICAR for different periods of time, followed by Western blotting for total and phospho-S616 and S637 DRP1. (D) Quantification of S616 and S637 DRP1 phosphorylation corrected for total DRP1. Results were then normalized to untreated control cells at each time point. Results are means ± SEM of eight separate experiments. (E) In vitro protein phosphorylation assay analyzing phosphorylation of human DRP1 by AMPKα1β1γ1; SAMS peptide used as a positive control. *, P < .05; **, P < .01.

AMPK activation prevents palmitate-induced ER enlargement in β-cells

We further examined the effects of the AMPK activators AICAR and metformin on ER changes induced by palmitate in INS-1E cells and human islets. Both drugs markedly decreased palmitate-induced ER enlargement (Figure 6). To exclude off-target effects of AICAR, we used an AMPK DN construct (Figure 6, F and G). Expressing AMPK DN in β- cells abolished the rescue by AICAR of palmitate-induced ER enlargement, indicating that the AICAR effect is indeed mediated via AMPK. Next, we studied whether AMPK activation by AICAR rescued ER and mitochondria morphology by DRP1 phosphorylation. We transduced β-cells with the rat DRP1 phosphorylation mutant (GFP-DRP1 S656A). The rat S656 parallels the S637 site in man (27). Strikingly, as compared to WT GFP-DRP1, GFP-DRP1 S656A attenuated the rescue effect of AMPK on ER and mitochondria morphology, although the effect on the mitochondria was not statistically significant (Figure 6, H and I). In summary, AMPK phosphorylation of DRP1 at S637 in man/S656 in rat mediates its effects on ER and mitochondria morphology.

Figure 6.

Effect of AMPK activation on palmitate-induced ER expansion. (A) AMPK activation by AICAR and metformin. INS-1E β-cells were treated as indicated at 11.1 mmol/L glucose for 6 hours (C, control; P, palmitate [0.5 mmol/L], A, AICAR [1 mmol/L], M, metformin [1 mmol/L]). AMPK activity analyzed by Western blotting for phosphorylated (Thr172) AMPK and phosphorylated (Ser79) ACC. (B–E) Effect of AMPK activation on palmitate-induced ER expansion in INS-1E β-cells (B, C) and human islets (D, E). INS-1E β-cells and human islets expressing eYFP-ER were treated as indicated incubated at 11.1 mmol/L glucose for 6 hours followed by confocal microscopy (Ctrl, control; Palm, palmitate; A, AICAR; Met, metformin). White square indicates zoom area. (C, E) Quantification of ER expansion in INS-1E β cells and human islets. Results are means ± SEM of six separate experiments analyzing at least 40 cells per treatment (C), and of three to four separate experiments analyzing at least 30 cells per treatment (E). (F) Effect of AMPK DN on AMPK activity. INS-1E β-cells were transduced with an AMPK DN adenoviral vector followed by Western blotting for total and phospho-ACC (EV, empty vector; DN, dominant negative). (G) Quantification of ER expansion in INS-1E β-cells transduced with empty vector (EV) or AMPK DN and treated with or without palmitate and/or AICAR at 11.1 mmol/L glucose for 6 hours. Results are means ± SEM of three to four separate experiments analyzing at least 40 cells per treatment. (H, I) Quantification of ER and mitochondrial morphology in cells expressing WT and phosphorylation incompetent form (S656A) of GFP-DRP1. Cells were transfected with the ER marker eYFP-ER and stained with the mitochondrial dye TMRE. Results are means ± SEM of three separate experiments analyzing 12 to 34 cells per treatment. (J) EM images of WT and LKB1 knockout (KO) mouse β cells in vivo. Arrows point to ER. Asterisk in EM image indicates insulin granule. Size bar, 200 nm. *, P < .05; **, P < .01.

AMPK regulation of ER morphology in vivo was studied in mice with conditional deletion of the AMPK upstream regulator LKB1 in β-cells; these β-cells express exceedingly low AMPK activity (29, 30). EM of LKB1 null β-cells revealed marked enlargement of the ER (Figure 6J), suggesting that LKB1/AMPK regulates ER morphology in vivo even in the absence of exogenous stress. Stimulation of AMPK impacts several metabolic pathways that may affect ER configuration. We tested this by using different pharmacological modulators of AMPK downstream targets and of FFA metabolism, including fumonisin B1 (ceramide synthase inhibitor), simvastatin (HMG-CoA reductase inhibitor), rapamycin (mTORC1 inhibitor), and etomoxir (CPT1 inhibitor); however, none of these compounds modified the palmitate or AICAR effects on ER morphology.

Impact of AMPK activation on stressed β-cell function

In β-cells, proper function of the ER is essential for proinsulin transport to the Golgi and effective processing and channeling to the secretory pathway. We therefore performed single-cell analyses of proinsulin retention in the ER in β-cells that were transfected with GFP-proinsulin and the ER marker RFP-ER and then treated with palmitate and/or AICAR for 6 hours. In palmitate-treated β-cells, the retention of proinsulin-GFP in the ER was increased in cells presenting ER expansion, evident by greater colocalization of GFP-proinsulin and RFP-ER compared to cells lacking ER expansion (Figure 7A). Treatment with AICAR decreased the number of cells with ER expansion and resulted in reduced proinsulin-GFP retention in the ER. Finally, we found that palmitate decreased islet insulin content and increased the proinsulin/insulin ratio in mouse islet extracts, whereas AICAR prevented these changes (Figure 7, B and C). Treatment with palmitate significantly increased basal mitochondrial respiration and insulin secretion and decreased glucose-stimulated insulin secretion (Supplemental Figure 4). By contrast, AICAR inhibited basal oxygen consumption in control and palmitate-treated cells and prevented the palmitate effects on insulin secretion. Despite the suppressive effect of AICAR on insulin secretion, the response to glucose was augmented twofold (Figure 7D and Supplemental Figure 4). Insulin secretion of islets treated with palmitate and AICAR together was similar to that of the AICAR alone group. FCCP-stimulated maximal respiration was not affected by palmitate and/or AICAR (Supplemental Figure 4); thus, their effects on insulin secretion cannot be attributed to modulation of mitochondrial metabolic capacity. In summary, palmitate treatment impaired, whereas stimulation of AMPK ameliorated, palmitate-induced β-cell dysfunction.

Figure 7.

Palmitate effects on proinsulin retention in the ER and β-cell function. (A) INS-1E β cells expressing GFP-proinsulin and ER-RFP were treated as indicated at 11.1 mmol/L glucose. GFP-proinsulin/ER-RFP colocalization, indicative of proinsulin retention in the ER, was analyzed by confocal microscopy. (B) Quantification of GFP-proinsulin and ER-RFP colocalization in tubular and expanded ER. Results are means ± SEM of five separate experiments analyzing 47 cells altogether. (C–E) Effects of palmitate and AICAR on proinsulin/insulin ratio (C), insulin content (D), and secretion (E). Isolated mouse islets were treated as indicated at 11.1 mmol/L glucose for 24 hours. Proinsulin and insulin were measured in islet extracts by proinsulin ELISA and insulin RIA. Insulin secretion was assessed by static incubations at 3.3 mmol/L (basal) and 16.7 mmol/L (stimulated) glucose concentrations. The ratio of stimulated/basal insulin secretion is shown. *, P < .05; **, P < .01.

Impact of stimulating AMPK on mitochondrial fragmentation and apoptosis in palmitate-stressed β-cells

Because AMPK activation inhibited DRP1, we studied its effect on mitochondrial morphology and function. AICAR did not affect mitochondrial morphology in control cells but markedly decreased mitochondrial fragmentation induced by palmitate (Figure 8, A and B). We further examined the effect of AMPK activation on apoptosis. Intriguingly, AICAR alone increased cleaved caspase-3, measured by Western blotting, and nucleosome content, measured by ELISA (Figure 7, C and D). However, when added to palmitate, AICAR significantly reduced β-cell apoptosis measured by these parameters (Figure 8, C and D).

Figure 8.

AICAR effects on palmitate-induced mitochondrial fragmentation and apoptosis. (A) INS-1E β-cells expressing eYFP-ER were treated as indicated at 11.1 mmol/L glucose (C, control; P, palmitate; A, AICAR) and then stained with the mitochondrial dye TMRE and mitochondrial fragmentation analyzed by confocal microscopy. (B) Quantification of mitochondrial fragmentation. Results are means ± SEM of five separate experiments analyzing 155 to 325 cells imaged per treatment. (C, D) Western blot for cleaved and total caspase-3 in INS-1E β-cells treated as indicated. A representative blot and quantification of cleaved caspase-3 at 24 hours, normalized to total caspase 3 is shown. Results are means ± SEM of three to eight separate experiments. (E) Apoptosis ELISA assay quantifying nucleosome content. Results are means ± SEM of four to five separate experiments. *, P < .05; **, P < .01.

Discussion

In this study we describe a novel signaling pathway that we believe plays a central role in transmitting the effect of increased FFA on β-cell dysfunction. We found that the mitochondrial fission protein DRP1, an ER resident in β-cells, is controlled by AMPK and is implicated in the ER shape and function modifications that occur under lipotoxic conditions.

Organelle morphology and interorganelle interactions may have robust effects on cellular function and survival. The current study sheds new light on the regulation of ER morphology in metabolic stress and its impact on β-cell function. Previous studies in other mammalian cells and yeast showed that ER stress may induce varying morphological changes in the ER (31–33). Enlargement of the ER was reported in β-cells exposed to palmitate, as well as to other inducers of ER stress (34, 35). However, most studies lacked comprehensive assessment of ER morphology. In the present study, detailed qualitative and quantitative analyses of ER morphology showed that in nonstressed β- cells the ER consists mainly of narrow and long tubular structures. Stressing β-cells or MEFs with palmitate or thapsigargin transformed the tubular ER into larger sack-like cisternal structures. It could be expected that stress-induced ER expansion is mediated via the UPR; however, several lines of evidence indicate that this is not the case. High glucose and cytokines that are involved in diabetes pathophysiology and activate β-cell stress did not expand the ER. Also, tunicamycin, a potent inducer of ER stress and the UPR, did not expand the ER. Finally, IRE1α and Xbp1 deficiency did not affect the alteration of ER morphology by palmitate or thapsigargin.

Previous studies showed that IRE1α regulates ER biogenesis and expansion in various non-β-cell types (36–38). This was mediated via the endoribonuclease activity of IRE1α and its product spliced Xbp1 (36, 39). Consistent with these findings, we found that MEFs lacking IRE1α or Xbp1 contained fewer cisternae-like structures, suggesting that in nonstressed cells, spliced Xbp1 participates in the regulation of ER morphology. In yeast, the IRE1α orthologue Ino2/4 regulates ER expansion in response to stress (33); however, we could not confirm this finding in mammalian cells. Indeed, the ER in Xbp1-deficient cells reacted to stress identically as in WT cells. Thus, although we cannot exclude UPR implication in the regulation of ER morphology, its role in stress-induced ER shape changes, if at all present, must be quite minor.

DRP1 regulation of ER and mitochondria morphology

Palmitate induces mitochondrial fragmentation in β- cells (22). We noticed a temporal correlation between mitochondrial fragmentation and ER expansion; this led us to postulate that common proteins coordinate morphology in these organelles. Overexpression of the mitochondrial fission protein DRP1 was reported to induce ER stress in β-cells (40); hence, we studied its implication in ER morphology. The current view is that DRP1 is primarily a cytosolic protein that is recruited to mitochondrial fission sites (25). A few studies reported that DRP1 is also found in the nucleus (41), peroxisomes (42), and the ER (43). However, the role of DRP1 in the regulation of altered ER morphology in response to stress was not reported. We found that in β-cells DRP1 is localized abundantly in the ER, as well as in mitochondrial fission sites. Of note, a recent report demonstrated that fission occurs at points where ER tubules wrap around the mitochondria and mark mitochondrial division sites (11). Herein, we show that inhibition of DRP1 attenuated both mitochondria fragmentation and ER expansion induced by FFA, suggesting that these processes are coordinated, rather than being stochastic. The GTP hydrolysis-incompetent mutant (K38A) sequesters endogenous DRP1 into aggregated structures consisting of membrane tubules, thus inhibiting DRP1 localization to mitochondrial fission sites and acting as a DN (44). It is thus tempting to postulate that DRP1 located at fission sites coordinates ER and mitochondria morphology and their interorganelle cross-talk. However, we cannot exclude the possibility that the effects of the K38A DRP1 mutant on ER morphology are mediated via inhibition of DRP1 in other ER subdomains. Better characterization of DRP1 assembly in the ER and recruitment to ER-mitochondria interaction sites is required to distinguish between these possibilities.

AMPK regulation of DRP1 phosphorylation and ER morphology

We identified AMPK as a major regulator of ER and mitochondria morphology in FFA-induced stress. AMPK is a cellular energy sensor that promotes ATP production by increasing the activity or expression of proteins involved in catabolism, while switching off biosynthetic pathways to conserve ATP (28). Hitherto, the role of AMPK in the regulation of ER morphology in response to stress was unknown. We show here for the first time that AMPK regulates ER morphology: AICAR and the common T2D drug metformin, which activate AMPK (45), prevented FFA-induced ER expansion; genetic inhibition of AMPK verified that this was indeed AMPK-dependent. Moreover, the β-cell ER of mice deficient in the upstream AMPK regulator LKB1 appeared as large cisternae-like structures, suggesting that a basal AMPK activity is necessary to maintain normal ER structure in vivo. Importantly, we found that AMPK effects on ER and mitochondria morphology are mediated, at least in part, via DRP1.

Regulation of DRP1 and/or its interacting proteins by phosphorylation is important for the intracellular cycling of DRP1 (46). Phosphorylation of DRP1 S637 plays an important role in this context. As an example, protein kinase A phosphorylation of DRP1 S637 was previously shown to mediate cAMP-induced mitochondrial elongation under nutrient starvation conditions (47). Conversely, calcineurin dephosphorylates DRP1 S637 and stimulates mitochondrial fragmentation (48). We show here that DRP1 is regulated by AMPK through phosphorylation of this antifission site. In line with this it was reported in lung adenocarcinoma cell lines that AMPK knockdown decreased general DRP1 phosphorylation in response to hypoxia (41). Importantly, prevention of DRP1 S637 phosphorylation partially prevented the rescuing effect of AMPK on mitochondrial fragmentation and on ER expansion. We expected that AMPK would directly phosphorylate DRP1 S637; however, at least under the in vitro test conditions used, this was not the case. Future studies will show whether AMPK regulates DRP1 phosphorylation through modifications of its interacting proteins. Of note, the prevention of AICAR effects on ER morphology by DRP1 DN or the S637 phosphorylation mutant was partial, indicating that additional mechanisms are involved in AMPK regulation of ER and mitochondria morphology. It is unlikely that AMPK's effect on ER morphology is mediated via modulation of metabolism as it did not affect mitochondrial respiration in cells treated with palmitate (Supplemental Figure 5). Furthermore, inhibition of FFA oxidation by etomoxir did not affect the ER expansion in response to palmitate, nor did it abolish its prevention by AICAR. Similarly, inhibition of ceramide and cholesterol synthesis also failed to affect ER morphology. Of note, treatment with AICAR did not affect triglyceride levels in β-cells exposed to hyperglycemia (49), further supporting our claim that the beneficial effects of AMPK are not mediated via effects on lipid handling in the β-cell. Further studies are required to unravel DRP1-independent pathways regulating organelle morphology in stressed β-cells.

AMPK regulation of β-cell function and survival

To understand the functional impact of the expanded cisternal ER shape on β-cell function better, we developed an assay that allows assessment of proinsulin retention in the ER of single cells. We observed ER proinsulin retention in palmitate-treated cells that presented expanded ER, but not in those that retained tubular ER (Figure 7); this was associated with increased proinsulin/insulin ratio in islets. These findings are consistent with previous studies showing that palmitate inhibited ER-Golgi trafficking and proinsulin processing (50, 51). The tight correlation between ER expansion and proinsulin retention suggests a cause-and-effect relationship between morphology and function.

We found that AMPK stimulation by AICAR decreased basal and glucose-stimulated mitochondrial respiration and insulin secretion, and increased apoptosis. This is consistent with previous studies showing that pharmacological stimulation of AMPK increased apoptosis (52), whereas AMPK deficiency in β-cells led to augmented insulin secretion and increased β-cell mass (29, 30), The mechanisms by which AMPK stimulation generates stress and impairs β-cell function might be related to mitochondrial dysfunction and increased production of reactive oxygen species (52), although the precise mechanism is unknown. By contrast, AICAR prevented the deleterious effects of palmitate on β-cell function. In addition, AICAR decreased palmitate-induced apoptosis, in agreement with previous reports on its protective effects against lipotoxicity (53, 54). The prevention of palmitate effects on mitochondrial fragmentation and respiration, as well as on ER expansion and the associated derangements in β-cell function and survival, emphasize that AMPK rescue of β-cells is specific to metabolic stress elicited by FFA. DRP1 was recently shown to mediate FFA-induced β-cell apoptosis (55). Our findings that AMPK inhibits DRP1 and prevents FFA-induced mitochondrial fragmentation and β-cell apoptosis strongly suggest that DRP1 mediates AMPK protection of β-cells in lipotoxicity.

We conclude that the AMPK-DRP1 connection delineates a novel pathway that links metabolic stress, organelle morphology, and function. AMPK preserves ER and mitochondrial structure and function in part through inhibition of DRP1, which contributes to improved β-cell function and survival. These findings have important implications for understanding the regulation of ER morphology and for exploring novel treatments of diabetes.

Funding Statement

This study was supported by a grant from the Israel Science Foundation to G.L. and by grants from the Golda Meir foundation and the European Foundation for the Study of Diabetes to J.D.W. We thank the generous colleagues who provided plasmids and Dr Orian Shirihai, Boston University, for providing Drp1 DN adenovirus.