VAMP7 is a SNARE protein that mediates specific membrane fusions in intracellular trafficking and was recently reported to regulate autophagosome formation. However, its function in pancreatic β-cells is largely unknown. To elucidate the physiological role of VAMP7 in β-cells, we generated pancreatic β-cell–specific VAMP7 knockout (Vamp7flox/Y;Cre) mice. VAMP7 deletion impaired glucose-stimulated ATP production and insulin secretion, though VAMP7 was not localized to insulin granules. VAMP7-deficient β-cells showed defective autophagosome formation and reduced mitochondrial function. p62/SQSTM1, a marker protein for defective autophagy, was selectively accumulated on mitochondria in VAMP7-deficient β-cells. These findings suggest that accumulation of dysfunctional mitochondria that are degraded by autophagy caused impairment of glucose-stimulated ATP production and insulin secretion in Vamp7flox/Y;Cre β-cells. Feeding a high-fat diet to Vamp7flox/Y;Cre mice exacerbated mitochondrial dysfunction, further decreased ATP production and insulin secretion, and consequently induced glucose intolerance. Moreover, we found upregulated VAMP7 expression in wild-type mice fed a high-fat diet and in db/db mice, a model for diabetes. Thus our data indicate that VAMP7 regulates autophagy to maintain mitochondrial quality and insulin secretion in response to pathological stress in β-cells.
Intracellular membrane fusion events within eukaryotic cells are mediated by members of the soluble N-ethylmaleimide-sensitive fusion protein attachment protein receptor (SNARE) protein family. All SNAREs contain a characteristic coiled-coil SNARE motif, through which SNAREs interact with one another. Functionally, SNAREs can be divided into vesicular (v-SNARE) and target membrane SNAREs, which are associated with intracellular trafficking vesicles and their target compartments, respectively. The formation of a functional complex between cognate v-SNAREs and target membrane SNAREs brings the opposing membranes close together and eventually catalyzes membrane fusion. Because each SNARE protein shows distinct tissue expression and intracellular localization patterns, the assembly of a functional SNARE complex is remarkably important for both specifying membrane fusion and coordinating different membrane trafficking pathways (1,2).
Vesicle-associated membrane proteins (VAMPs) function as v-SNAREs. In studies with pancreatic β-cells, among seven VAMP family proteins (VAMP1–5, 7, and 8), VAMP2, 3, 4, and 8 were expressed. VAMP2 and VAMP3/cellubrevin were localized to insulin granules and synaptic-like microvesicles and regulated glucose-stimulated insulin secretion (3,4). VAMP4 was expressed in β-cell–derived Min6 cells (5) and was implicated in post-Golgi vesicular trafficking (6,7). VAMP8/endobrevin was localized to endosomes, insulin granules, and synaptic-like microvesicles and was reported to regulate insulin and γ-aminobutyric acid secretion in pancreatic β-cells (8). VAMP8 was also recently found to participate in β-cell proliferation by regulating mitosis (9). VAMP7, however, has not yet been studied in pancreatic β-cells.
VAMP7 was originally identified as a tetanus neurotoxin–insensitive VAMP regulating vesicular transport to apical membranes in polarized epithelial cells (10,11). Growing evidence has demonstrated that VAMP7 is involved in synaptic transmission, neurite extension, and intracellular vesicular trafficking (12). It was proposed that VAMP7 could be involved in autophagy (13–16), in which unnecessary protein aggregates and damaged organelles are engulfed by a double-membrane structure and transported to lysosomes, where their contents are degraded by lysosomal enzymes (17). VAMP7 global knockout mice were recently established, and the lack of VAMP7 in these mice did not cause any striking developmental or neurological defects (18,19). However, the role of VAMP7 in glucose homeostasis has not been investigated.
In this study we show that VAMP7 is expressed in pancreatic β-cells and functions to regulate autophagy. Our data demonstrate that VAMP7 regulates autophagy to maintain mitochondrial homeostasis and glucose-stimulated biphasic insulin secretion.
Research Design and Methods
Pancreatic β-cell–specific VAMP7 knockout (Vamp7flox/Y;Cre) mice were generated by crossing VAMP7 floxed mice (Vamp7flox/Y) (18) with RIP-Cre (Cre) mice (20). All mice were housed on a 12-h light/12-h dark cycle in climate-controlled facilities and fed either standard (EC-2; Clea Japan, Inc., Tokyo, Japan) or 60% fat rodent food (D12492; Research Diets). Male db/misty and db/db mice were purchased from Sankyo Labo Service Corp. Male mice were fasted for 16 h before being subjected to oral glucose tolerance tests with 1.25 g glucose/kg body weight. Blood samples were collected from the tail vein, and blood glucose concentrations were measured by Glutest R (Sanwa Kagaku Kenkyusho Co., Aichi, Japan). Animal experiments were approved by the Kyorin University Animal Care Committee (permission no. 65-3).
Plasmids and Recombinant Adenoviruses
The plasmids encoding the floxed stop sequence and 4mtD3cpv were gifts from B. Sauer (Addgene plasmid no. 11952) and from A. Palmer and R. Tsien (Addgene plasmid no. 36324), respectively. The cDNA fragment encoding mouse VAMP7 was amplified from mouse cerebrum total RNA by RT-PCR. The floxed stop sequences followed by hemagglutinin-tagged VAMP7 (HA-VAMP7), 4mtD3cpv, or human growth hormone (hGH) were subcloned into a pENTR1A vector (Invitrogen). To generate expression vectors, the VAMP7 sequence or the floxed stop sequence, followed by hGH, were introduced into the pEF-DEST51 or pcDNA3.2/V5-DEST vector (Invitrogen), respectively, by site-specific recombination. To generate adenovirus vectors, the floxed stop sequence followed by HA-VAMP7 or 4mtD3cpv were introduced into the pAd/CMV/V5-DEST vector (Invitrogen) by site-specific recombination. Recombinant adenoviruses were generated according to the manufacturer’s instructions. All constructs were sequenced to confirm their identities with the designed sequences.
Pancreatic Islets and β-Cell Preparation and Secretion Assays
Pancreatic islets of Langerhans were isolated from male mice by collagenase digestion, as described previously (20). Islets were transfected by electroporation, as described previously (20), with a plasmid encoding a floxed stop sequence followed by hGH to express hGH specifically in Cre-expressing β-cells. Insulin and hGH secretion experiments were performed at 37°C in Krebs-Ringer buffer (KRB) containing 110 mmol/L NaCl, 4.4 mmol/L KCl, 1.45 mmol/L KH2PO4, 1.2 mmol/L MgSO4, 2.3 mmol/L calcium gluconate, 4.8 mmol/L NaHCO3, 2.2 mmol/L glucose, 10 mmol/L HEPES (pH 7.4), and 0.3% BSA, as described previously (20).
Cytosolic Ca2+ dynamics were monitored using Fura-2 acetoxymethyl ester (Fura-2, AM; Invitrogen), as described previously (20). For mitochondrial Ca2+ imaging, pancreatic β-cells cultured overnight were infected with recombinant adenovirus to express 4mtD3cpv. Two days after infection, cells were mounted on an open chamber and incubated for 15 min with KRB containing 2.2 mmol/L glucose. Stimulation with glucose was achieved by adding KRB containing 52 mmol/L glucose into the chamber, for a final concentration of 22 mmol/L glucose. Fluorescent signals derived from 4mtD3cpv were collected with an array-scanning laser confocal system (VT-Infinity3; VisiTech International, Sunderland, U.K.) operated with Metamorph (Universal Imaging). We used a 461-nm laser for excitation. Cyan fluorescent protein and yellow fluorescent protein signals were split by a DV2 multichannel imaging system (Photometrics) and projected onto an electron-multiplying charge-coupled device camera (DU-897E; Andor Technology, Belfast, U.K.). Images were acquired every 5 s.
Immunoblotting was performed as described previously (20). Antibodies against Glut2 and glucokinase (21), VAMP2 (22), SNAP23 (23), and phogrin (24) were described previously. Antibodies against VAMP3 and VAMP4 were raised against synthetic peptides (VAMP3: STGVPSGSSAATGSC, VAMP4: CDFFLRGPSGPRFGPRND) and purified as previously described (23). The following antibodies were purchased from commercial sources: VAMP7 (Abcam), α-tubulin, GAPDH, insulin (Sigma-Aldrich), Syntaxin1A, SNAP25 (Wako Pure Chemical Industries, Japan), Syntaxin3, Syntaxin4, VAMP8 (Synaptic Systems, Göttingen, Germany), Cre recombinase (Millipore), p62 (Progen), LC3 (Cell Signaling Technologies), HA-tag (Medical & Biological Laboratories Co., Japan), Bcl-2 (BD Transduction Laboratories), and LAMP1 (Developmental Studies Hybridoma Bank).
After overnight culture, 40 size-matched islets were preincubated with KRB containing 2.2 mmol/L glucose for 60 min at 37°C, followed by 0, 5, 20, or 30 min of incubation with KRB containing 16 mmol/L glucose. Cellular ATP was extracted using extraction buffer (Applied Medical Enzyme Research Institute, Co., Tokushima, Japan) and assayed by high-performance liquid chromatography (HPLC) on a reversed-phase SUPELCOSIL column (Sigma-Aldrich), as described previously (21). ATP levels were normalized to the amount of total cellular DNA measured by a PicoGreen dsDNA assay kit (Invitrogen).
ROS Measurements Using MitoSOX
Pancreatic β-cells cultured on coverslips for 48 h were stained with 5 μmol/L MitoSOX (Molecular Probes) for 20 min at 37°C. After three washes with KRB containing 2.2 mmol/L glucose, fluorescent signals derived from MitoSOX were detected using VT-Infinity3. The signal intensities per area within cellular regions were quantified using ImageJ software.
Preparation of the Islet Mitochondrial Fraction
Freshly isolated islets in homogenization buffer (250 mmol/L sucrose, 5 mmol/L MgCl2, 5 mmol/L EDTA, 1 mmol/L phenylmethylsulfonyl fluoride, and 80 mmol/L Tris [pH 6.8]) were homogenized in a Potter-Elvehjem homogenizer using 10 strokes of a close-fitting Teflon pestle and then centrifuged at 600g for 10 min to separate the nuclear and cytosolic fractions. The cytosolic fraction was further centrifuged at 8,000g for 15 min to pellet the mitochondria. The pellet was solubilized in an SDS sample buffer and subjected to immunoblotting.
Second-Phase Insulin Secretion Was Selectively Impaired in Vamp7flox/Y;Cre Islets
To explore the potential role of VAMP7 in pancreatic β-cells, we generated Vamp7flox/Y;Cre mice by crossing previously established Vamp7flox/Y mice (18) with mice expressing Cre recombinase in pancreatic β-cells (Cre) (25). VAMP7 was expressed in islets from Vamp7flox/Y mice. By contrast, islets from Vamp7flox/Y;Cre mice exhibited an approximately 70% reduction of VAMP7, with no changes in expression levels of GLUT2, glucokinase, and other SNARE proteins (Fig. 1). This result also confirmed that VAMP7 is expressed in pancreatic β-cells.
To investigate the physiological roles of VAMP7 in pancreatic β-cells, we first examined glucose-stimulated insulin secretion in isolated islets. As shown in Fig. 2A, glucose-stimulated insulin secretion was markedly decreased in Vamp7flox/Y;Cre islets compared with Vamp7flox/Y islets, whereas total insulin content was not altered (83.40 ± 10.52 and 88.98 ± 4.66 ng/islet for Vamp7flox/Y and Vamp7flox/Y;Cre islets, respectively). To examine whether the VAMP7 deletion was responsible for the impaired insulin secretion in Vamp7flox/Y;Cre islets, VAMP7 was transfected in Cre or Vamp7flox/Y;Cre islets by electroporation. Because of its low transfection efficacy, hGH was coexpressed as a reporter to evaluate its secretion from the transfected β-cells. As shown in Fig. 2B, transient expression of VAMP7 restored glucose-stimulated hGH secretion in Vamp7flox/Y;Cre islets.
Insulin secretion from pancreatic β-cells shows a characteristic biphasic pattern consisting of a transient first phase followed by a sustained second phase (26). We next investigated the effects of VAMP7 deletion on biphasic insulin secretion in isolated islets. Perifusion experiments revealed that glucose-stimulated second-phase secretion (>10 min after the onset of glucose stimulation) was selectively reduced in Vamp7flox/Y;Cre islets (Fig. 2D). In addition, insulin secretion evoked by 40 mmol/L KCl, which selectively elicits first-phase secretion (27), was not altered in Vamp7flox/Y;Cre islets (Fig. 2C). Glucose-stimulated cytosolic Ca2+ ([Ca2+]i) concentrations were not affected in Vamp7flox/Y;Cre β-cells (Fig. 2E).
Because we previously reported that second-phase secretion was composed of fusions originating from newcomer insulin granules (20,28), these results prompted us to hypothesize that VAMP7 was localized selectively to insulin granules, mediating their fusions. To investigate VAMP7 localization, we performed subcellular fractionation in Min6 cells (Fig. 3A). Phogrin, an insulin granule marker, together with insulin, were recovered in fraction 12 (24), but VAMP7 was not. A part of phogrin localized in the plasma membrane was recovered in fraction 4 that contains components of endolysosomes and plasma membrane particles (24). By contrast, most of the VAMP7 and some of the LAMP1, a lysosome marker protein, were recovered in fraction 4. Next, we expressed HA-VAMP7 to examine VAMP7 localization in β-cells because six VAMP7 antibodies we tested (two from commercial sources) had not been useful for immunostaining in pancreatic β-cells. As shown in Fig. 3B, insulin was not colocalized with HA-VAMP7. However, some HA-VAMP7 colocalized with LAMP1 (Fig. 3C), as previously reported in other cell types (11,12). Consistent with these results, ultrastructural analysis revealed that the distribution and morphology of insulin granules were not altered in Vamp7flox/Y;Cre β-cells (Supplementary Fig. 1). These results suggested that VAMP7 did not directly mediate exocytosis of insulin granules, but might participate in intracellular membrane trafficking routed to lysosomes.
Defects in Autophagosome Formation in Vamp7flox/Y;Cre β-Cells
Upon autophagy, unnecessary proteins and damaged organelles are sequestered into autophagosomes that subsequently fuse with lysosomes to enable the degradation of their contents. Recent studies demonstrated that VAMP7 depletion, using small interfering RNA, blocked autophagosome formation (13,14). To determine whether VAMP7 deletion was linked to defective autophagy, we next examined autophagosome formation in Vamp7flox/Y;Cre β-cells. LC3 is a mammalian autophagy protein, and induction of autophagy by starvation converts LC3-I, the cytosolic form of LC3, to lipidated LC3-II, which accumulates in autophagosome membranes. LC3-II is subsequently degraded when autophagosomes fuse with lysosomes. Thus the LC3-II–to–LC3-I ratio reflects the quantity of autophagosomes (29). Starvation for 3 h increased the LC3-II-to-LC3-I ratio in Vamp7flox/Y islets, but not in Vamp7flox/Y;Cre islets (Fig. 4A). The attenuated response of the LC3-II-to-LC3-I ratio in Vamp7flox/Y;Cre islets might indicate either a defect in autophagosome formation or increased autophagy degradation. To distinguish between these two possibilities, islets were treated with chloroquine, a lysosomal enzyme inhibitor, to inhibit LC3-II degradation. Chloroquine treatment did not increase the LC3-II-to-LC3-I ratio in starved Vamp7flox/Y;Cre islets, suggesting that VAMP7 deletion blocked autophagosome formation rather than promoting degradation. In agreement with this, the number of LC3 puncta induced by starvation in the presence of chloroquine, representing autophagosomes (29), was markedly reduced in Vamp7flox/Y;Cre β-cells and was restored by transient expression of HA-VAMP7 (Fig. 4B and C). p62/SQSTM1 is associated with unnecessary protein aggregates and damaged organelles to be degraded by autophagy. Thus p62 accumulation reflects levels of targets for autophagy and is widely used to detect autophagy defects (29). Under basal conditions, levels of p62 in Vamp7flox/Y;Cre islets were nearly equivalent to those in Vamp7flox/Y islets. Starvation alone did not increase p62 in Vamp7flox/Y islets, but starvation in the presence of chloroquine induced a significant p62 accumulation. By contrast, p62 levels in Vamp7flox/Y;Cre islets were significantly increased by starvation, and chloroquine treatment had no further effect (Fig. 4D). These results indicated that autophagosome formation in response to starvation was defective in Vamp7flox/Y;Cre β-cells.
Accumulation of Dysfunctional Mitochondria in Vamp7flox/Y;Cre Islets
Next we examined whether the defective autophagy occurring with VAMP7 deletion was accompanied by impairment of glucose-induced second-phase secretion. Autophagy plays an important role in mitochondrial quality control (30). Interestingly, mitochondrial Ca2+ ([Ca2+]mt) increases were shown to augment mitochondrial metabolism (31) and enhance second-phase secretion (32). Therefore we hypothesized that the defective autophagosome formation caused by VAMP7 deletion would lead to the accumulation of dysfunctional mitochondria, which in turn would affect [Ca2+]mt increases and second-phase secretion. To test this, we examined endogenous reactive oxygen species (ROS) levels in Vamp7flox/Y;Cre islets because mitochondrial dysfunction often leads to excessive ROS production (33). Pancreatic β-cells cultured on coverslips were stained with MitoSOX, a mitochondrial superoxide indicator. As shown in Fig. 5A, the MitoSOX signal intensities in Vamp7flox/Y;Cre β-cells were modestly but significantly higher than in Vamp7flox/Y β-cells. In addition, H2O2 production, reflecting the intracellular ROS levels, was also significantly increased in Vamp7flox/Y;Cre islets (Supplementary Fig. 2). Then we visualized functional mitochondria using Mitotracker Red CM-H2XRos (MTR), which is converted to a fluorescently active state only when incorporated into functional mitochondria. To identify mitochondria in β-cells, we expressed enhanced green fluorescent protein tagged with the mitochondria targeting sequence (2mt-EGFP) and then stained cells with MTR. As shown in Figs. 5B and C, the MTR signal intensity in the mitochondrial regions was significantly lower in Vamp7flox/Y;Cre β-cells compared with Vamp7flox/Y β-cells. Ultrastructural analysis revealed that mitochondria in Vamp7flox/Y;Cre β-cells were swollen and their crista exhibited abnormal structures (Figs. 5D and E), reminiscent of β-cells with dysfunctional mitochondria (34,35). These results suggested that damaged mitochondria were accumulating in Vamp7flox/Y;Cre β-cells.
We next examined whether p62 selectively accumulated in mitochondria from Vamp7flox/Y;Cre β-cells, though we had already found that p62 accumulation was not consistently detected in whole islets from Vamp7flox/Y;Cre mice under basal conditions (Fig. 4D). Mitochondria-enriched fractions containing Bcl-2 were prepared from isolated islets and amounts of their p62 concentrations were determined. Amounts of p62 were in fact increased in mitochondria-enriched fractions prepared from Vamp7flox/Y;Cre islets compared with those from Vamp7flox/Y islets (Fig. 5F). This result indicated that dysfunctional mitochondria destined for degradation by autophagy accumulated in Vamp7flox/Y;Cre β-cells.
We further investigated glucose-stimulated mitochondrial function in Vamp7flox/Y;Cre β-cells. Glucose-stimulated augmentation of mitochondrial function was examined using tetramethylrhodamine ethyl ester, a mitochondrial membrane potential (ΔΨmt)–sensitive dye (21). Vamp7flox/Y;Cre β-cells exhibited a marked reduction in ΔΨmt in response to glucose stimulation (Fig. 6A). Because impaired ΔΨmt response is accompanied by a reduction in glucose-stimulated [Ca2+]mt responses (36), we next studied glucose-stimulated [Ca2+]mt dynamics in β-cells. The calcium-sensitive probe D3cpv (37) tagged with a mitochondrial targeting sequence, denoted as 4mtD3cpv, was expressed in β-cells, and changes in 4mtD3cpv signal intensity were evaluated. As shown in Fig. 6B, glucose-stimulated [Ca2+]mt signaling was attenuated in Vamp7flox/Y;Cre β-cells. Increased [Ca2+]mt activates Ca2+-dependent enzymes in mitochondria to augment the synthesis of mitochondrial products, including ATP (31). Therefore we examined cellular ATP content in isolated islets. In Vamp7flox/Y;Cre islets ATP concentrations were not changed under basal conditions, but were significantly lower compared with in Vamp7flox/Y islets after glucose stimulation for >20 min, conditions corresponding to second-phase secretion (Fig. 6C). Taken together, our data indicate that defective autophagy in Vamp7flox/Y;Cre β-cells caused the accumulation of dysfunctional mitochondria, reduced [Ca2+]mt responses, and decreased ATP production during the second phase. These findings suggest that VAMP7 regulates autophagy to eliminate dysfunctional mitochondria and to control second-phase secretion.
A High-Fat Diet Reduced Both First- and Second-Phase Insulin Secretion and Caused Glucose Intolerance in Vamp7flox/Y;Cre Mice
Our results indicated that VAMP7 deletion caused defective autophagy and impaired insulin secretion in pancreatic β-cells, prompting us to examine its effects on glucose homeostasis. At 10 weeks of age, Vamp7flox/Y;Cre mice had normal glucose tolerance, as detected after oral glucose administration (Fig. 7A). This was expected because VAMP7 deletion did not affect first-phase secretion (Fig. 2D), the major contributor to maintaining postprandial blood glucose homeostasis (38). Feeding mice a high-fat diet (HFD) was reported to enhance autophagy in β-cells promoting the elimination of damaged organelles and protecting cells from pathological stresses (39). Because Vamp7flox/Y;Cre β-cells could not properly induce autophagy (Fig. 4), we hypothesized that feeding an HFD to Vamp7flox/Y;Cre mice would induce further accumulation of dysfunctional mitochondria, which would be increasingly detrimental to the cells and lead to impaired insulin secretion and diabetes (40). To test our hypothesis, mice were fed an HFD for 10 weeks beginning at 5 weeks of age. Blood glucose concentrations at 30 and 60 min after glucose administration were significantly higher in Vamp7flox/Y;Cre mice than in Vamp7flox/Y mice (Fig. 7B), whereas the body weights of HFD-fed Vamp7flox/Y and Vamp7flox/Y;Cre mice were not significantly different (33.9 ± 1.0 and 33.7 ± 1.5 g, respectively). By contrast, fasting blood glucose concentrations in Vamp7flox/Y;Cre mice were not different from those in Vamp7flox/Y mice even after being fed an HFD for >10 weeks (Supplementary Fig. 3). Thus Vamp7flox/Y;Cre mice displayed an impaired glucose tolerance but not marked hyperglycemia.
To examine whether glucose intolerance in HFD-fed Vamp7flox/Y;Cre mice was accompanied by reduced first-phase secretion, perifusion experiments were performed in isolated islets. As shown in Fig. 7C, both first- and second-phase secretion were markedly reduced in islets isolated from HFD-fed Vamp7flox/Y;Cre mice compared with HFD-fed Vamp7flox/Y mice, whereas total insulin content was not altered (213.6 ± 10.5 and 214.5 ± 10.0 ng/islet for Vamp7flox/Y and Vamp7flox/Y;Cre islets, respectively). Glucose-stimulated [Ca2+]i was also lower in β-cells isolated from HFD-fed Vamp7flox/Y;Cre mice (Fig. 7D). We next examined whether the impaired insulin secretion in islets isolated from HFD-fed Vamp7flox/Y;Cre mice was accompanied by further accumulation of dysfunctional mitochondria. Whole islets isolated from HFD-fed Vamp7flox/Y;Cre mice showed marked accumulation of p62 (Fig. 7E). In addition, parkin, a protein known to accumulate on dysfunctional mitochondria (33), was also markedly accumulated in these islets (Fig. 7F). Consistent with these findings, ATP concentrations in islets isolated from HFD-fed Vamp7flox/Y;Cre mice were significantly lower compared with those from HFD-fed Vamp7flox/Y mice under both basal and glucose-stimulated conditions (Fig. 7G). These results suggested that pathological stress facilitated the accumulation of dysfunctional mitochondria in Vamp7flox/Y;Cre β-cells, decreasing overall mitochondrial function and causing impaired glucose-stimulated biphasic insulin secretion and, therefore, glucose intolerance.
VAMP7 Was Upregulated by the Diabetic Condition
Our results suggested that VAMP7 is required for autophagy to maintain mitochondrial homeostasis in islets isolated from HFD-fed mice. Because LC3 and p62 were reported to be upregulated by HFD feeding to address increased autophagy flux in β-cells (41), we examined VAMP7 expression in islets isolated from HFD-fed wild-type mice. We found that VAMP7 expression was increased in islets isolated from the HFD-fed mice. VAMP7 upregulation was also observed in a diabetic mouse model (db/db mice) (Fig. 8). These results suggest that upregulation of VAMP7 is required to counteract pathological stress occurring under diabetic conditions.
In this study we found that VAMP7 plays an important role in mitochondrial quality control by regulating autophagy in pancreatic β-cells. The VAMP7 deletion disrupted autophagy flux and resulted in the accumulation of dysfunctional mitochondria and the impairment of glucose-induced [Ca2+]mt signaling, ATP production, and second-phase insulin secretion. In addition, feeding an HFD to Vamp7flox/Y;Cre mice exacerbated mitochondrial dysfunction, which caused the reduction of both first- and second-phase secretion and, consequently, glucose intolerance.
We found that VAMP7 deletion attenuated starvation-induced autophagosome formation. This is consistent with previous reports using cells depleted of VAMP7 with small interfering RNA (13,14). However, other studies suggested that VAMP7 could mediate the fusion of autophagosomes with lysosomes (15,16). Thus we have not ruled out the possibility that VAMP7 might mediate the two distinct fusion steps in the autophagy process: autophagosome formation and the subsequent fusion of autophagosomes with lysosomes. Because these two events occur sequentially, we could detect only a defect in autophagosome formation in Vamp7flox/Y;Cre β-cells.
Genetic ablation of Atg7, an essential factor for autophagy, caused massive accumulation of p62 and increased apoptotic cell death in pancreatic β-cells (42,43). Though Vamp7flox/Y;Cre β-cells exhibited defective autophagosome formation, we did not observe a detectable loss of insulin content or an increase in caspase-3/7 or caspase-9 activity (Supplementary Fig. 4). We also observed no gross morphological abnormalities in Vamp7flox/Y;Cre islets, further indicating that apoptotic cell death did not occur in Vamp7flox/Y;Cre β-cells. Thus, in contrast to Atg7 deletion, autophagy was not completely blocked in Vamp7flox/Y;Cre β-cells, probably because other proteins might compensate for VAMP7 function in the autophagy process. The longin domain at the N-terminal of VAMP7 was implicated as playing an essential role in autophagosome formation (13,14). Sec22 and Ykt6, both SNARE proteins with longin domains (12), were required for autophagosome formation in yeast (44). Thus mammalian homologs of yeast Sec22 and Ykt6 may also participate in autophagosome formation in β-cells, but further studies are required.
Glucose-induced second-phase secretion was selectively impaired in Vamp7flox/Y;Cre islets. Some studies have addressed the molecular mechanisms of second-phase secretion. Genetic deletion of Pak1 (45) and the R-type calcium channel (46) each selectively reduced second-phase secretion. However, we found that amounts of Pak1 (data not shown) and glucose-induced [Ca2+]i dynamics (Fig. 2E) were not affected in Vamp7flox/Y;Cre β-cells. Wiederkehr et al. (32) recently demonstrated that glucose-induced [Ca2+]mt dynamics were important for second-phase secretion. Mitochondrial Ca2+ activates Ca2+-dependent enzymes involved in mitochondrial processes (31), for example, augmenting ATP production required for insulin secretion (26). In addition, ouabain, an inhibitor of Na+/K+-ATPase, was reported to generate ROS, suppress mitochondrial ATP production, and selectively inhibit second-phase secretion (47). Consistent with these reports, we found that glucose-stimulated augmentation of mitochondrial function was impaired in Vamp7flox/Y;Cre β-cells. Autophagy was not induced by acute glucose stimulation (42), suggesting that the autophagic elimination of dysfunctional mitochondria derived from mitochondrial turnover was not induced by 30 min of glucose stimulation. Moreover, the expression levels of proteins required for mitochondrial fusion/fission were not altered in Vamp7flox/Y;Cre islets (Supplementary Fig. 5), implying that not the mitochondrial turnover but instead the capacity of mitochondrial metabolism augmented by glucose stimulation may be critical for second-phase insulin secretion in vivo. Glucose-stimulated first-phase secretion and [Ca2+]i dynamics were not affected in Vamp7flox/Y;Cre β-cells. Wiederkehr et al. also showed that inhibition of [Ca2+]mt did not affect [Ca2+]i dynamics and first-phase secretion. Glucose-stimulated augmentation of mitochondrial metabolism in β-cells can be divided into Ca2+-dependent and Ca2+-independent mechanisms, and first-phase secretion is triggered by Ca2+-independent augmentation of mitochondrial metabolism (48). Because we did not detect apoptotic cell death in Vamp7flox/Y;Cre islets, some mitochondrial function seems to have been preserved in Vamp7flox/Y;Cre β-cells, and these functionally preserved mitochondria could account for the first-phase insulin secretion.
Our results suggest that the defective autophagy in Vamp7flox/Y;Cre β-cells selectively disrupted mitochondrial function, potentially accounting for defective insulin secretion. One remaining question is why mitochondrial function was selectively disrupted in Vamp7flox/Y;Cre β-cells. A recent report suggested that mild oxidative stress selectively induces mitophagy (49). The defective autophagy in Vamp7flox/Y;Cre β-cells should have caused a gradual accumulation of spontaneously generated dysfunctional mitochondria that would increase cellular ROS concentrations. We did not detect apoptotic cell death in Vamp7flox/Y;Cre islets, though apoptosis is often associated with marked ROS elevations. This implies that VAMP7 deletion only moderately increased ROS concentrations. In fact, ROS concentrations in Vamp7flox/Y;Cre islets were mildly, but significantly, increased (Fig. 5A and Supplementary Fig. 2). Thus mild ROS elevations in Vamp7flox/Y;Cre β-cells might selectively induce p62 accumulation in mitochondria (Fig. 5F), leading to mitophagy. Feeding an HFD to Vamp7flox/Y;Cre mice exacerbated this mitochondrial dysfunction and impaired insulin secretion, further supporting an important role of VAMP7 in the elimination of dysfunctional mitochondria through autophagy.
In conclusion, we demonstrated a physiological role of VAMP7 in pancreatic β-cells. Our findings in Vamp7flox/Y;Cre mice suggest that VAMP7 regulates autophagosome formation to maintain mitochondrial homeostasis and insulin secretion. Further studies are necessary to elucidate the molecular mechanism of VAMP7-dependent autophagy and the protective role of VAMP7 against pathological stress and diabetes.
Acknowledgments. The authors thank H. Ohnishi (Gunma University) and T. Yamamoto (Kyorin University) for helpful discussions.
Funding. This work was supported by grants from the Ministry of Education, Culture, Sports, Science, and Technology in Japan (grant no. 25860756 [to K.A.])/Japan Society for the Promotion of Science KAKENHI (grant nos. 15K19525 [to K.A.] and 26460396 [to M.O.-I.]); the Joint Research Program of the Institute for Molecular and Cellular Regulation, Gunma University (grant no. 15010 [to K.A.]); the Japan Diabetes Foundation (to M.O.-I.); and the Science Research Promotion Fund from the Promotion and Mutual Aid Corporation for Private Schools of Japan (to S.N.).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. K.A., M.O.-I., A.H., M.T., and S.N. designed the study. K.A., M.O.-I., S.T., Y.A., C.N., and Y.N. performed the experiments. K.A., M.O.-I., M.I., S.T., T.K., H.K., M.T., and S.N. analyzed data. K.A., M.O.-I., and S.N. wrote the manuscript. S.N. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.