Loss of the voltage-gated proton channel Hv1 decreases insulin secretion and leads to hyperglycemia and glucose intolerance in mice

Insulin secretion by pancreatic islet β-cells is regulated by glucose levels and is accompanied by proton generation. The voltage-gated proton channel Hv1 is present in pancreatic β-cells and extremely selective for protons. However, whether Hv1 is involved in insulin secretion is unclear. Here we demonstrate that Hv1 promotes insulin secretion of pancreatic β-cells and glucose homeostasis. Hv1-deficient mice displayed hyperglycemia and glucose intolerance because of reduced insulin secretion but retained normal peripheral insulin sensitivity. Moreover, Hv1 loss contributed much more to severe glucose intolerance as the mice got older. Islets of Hv1-deficient and heterozygous mice were markedly deficient in glucose- and K+-induced insulin secretion. In perifusion assays, Hv1 deletion dramatically reduced the first and second phase of glucose-stimulated insulin secretion. Islet insulin and proinsulin content was reduced, and histological analysis of pancreas slices revealed an accompanying modest reduction of β-cell mass in Hv1 knockout mice. EM observations also indicated a reduction in insulin granule size, but not granule number or granule docking, in Hv1-deficient mice. Mechanistically, Hv1 loss limited the capacity for glucose-induced membrane depolarization, accompanied by a reduced ability of glucose to raise Ca2+ levels in islets, as evidenced by decreased durations of individual calcium oscillations. Moreover, Hv1 expression was significantly reduced in pancreatic β-cells from streptozotocin-induced diabetic mice, indicating that Hv1 deficiency is associated with β-cell dysfunction and diabetes. We conclude that Hv1 regulates insulin secretion and glucose homeostasis through a mechanism that depends on intracellular Ca2+ levels and membrane depolarization.

Insulin secretion by pancreatic ␤-cells is precisely regulated by glucose homeostasis. Defective insulin secretion because of ␤-cell loss or dysfunction causes type 1 and type 2 diabetes (T2D), 3 respectively (1). In the presence of high concentrations of glucose, ␤-cells metabolize glucose and generate ATP, which closes ATP-dependent K ϩ channels and results in membrane depolarization and the subsequently increased intracellular Ca 2ϩ that triggers insulin granule release (1).
There are two different time stages in insulin secretion in humans (2) and rodents (3,4), including a fast transient first phase and a slow sustained second phase. In patients with T2D, the first phase has almost disappeared, and the second phase is markedly decreased (5). Multiple factors influence the biphasic nature of GSIS, including intracellular Ca 2ϩ levels, distinct pools of insulin granules, and metabolic signaling (6 -8). A decrease in insulin secretion may be caused by a single-or multiple-step defect in the trafficking cycle of insulin granules, including granule biogenesis from the trans-Golgi network, subsequent maturation, recruitment to the plasma membrane, exocytosis, and endocytosis.
Glucose metabolism by pancreatic ␤-cells accompanies proton generation, which implies a mechanism of intracellular pHregulation behind insulin release stimulated by the sugar (9). Manipulating intracellular as well as extracellular pH could affect the insulin secretion process, which is associated with changes in membrane potential, ionic flux, and insulin release (9 -11). Barg et al. (12) showed that the acidic pH in insulin granules might regulate priming of the granules for secretion, a process involving pairing of SNAREs on vesicles and target membranes to establish fusion competence.
The voltage-gated proton channel Hv1 is extremely selective for protons and has no detectable permeability to other cations (13,14). Hv1 is activated at depolarizing voltages and sensitive to the membrane pH gradient (13,14). Hv1 sustains calcium entry in neutrophils and maintains intracellular alkalization to support the activity of human spermatozoa (15,16). In our previous study, we identified that Hv1 is present in human and rodent pancreatic islet ␤-cells as well as ␤-cell lines (17). How-ever, the regulatory mechanism of Hv1 for insulin secretion of pancreatic islet ␤-cells is not known.
In this study, we discovered a regulatory mechanism for Hv1 in the modulation of ␤-cell insulin secretory function. Our in vivo and in vitro studies demonstrated that Hv1-deficient mice exhibit hyperglycemia and glucose intolerance because of abnormally decreased insulin secretion. These data provide direct genetic evidence that Hv1 regulates biphasic insulin secretion and glucose homeostasis in mice, implying a potential link between Hv1 and the pathogenesis of diabetes mellitus.
To evaluate the impact of Hv1 on disposal of a glucose load, i.p. glucose tolerance tests (IPGTTs) were performed. Compared with WT mice, KO and heterozygous mice 4 months of age showed significantly higher glucose levels following an i.p. glucose load (2 g/kg of body weight) (Fig. 1, G and H). The corresponding serum insulin levels were significantly lower in Hv1-deficient mice throughout the IPGTT after glucose challenge compared with WT mice, providing evidence for an insulin secretion defect in response to glucose (Fig. 1, I and J). Thus, Hv1-deficient mice exhibit impairment in their ability to dispose of a glucose load because of an insulin secretion defect.
To explore the possibility that the observed glucose intolerance was a result of peripheral insulin resistance, we performed i.p. insulin tolerance tests (IPITTs) in KO mice 4 months of age. We found that insulin administration lowered blood glucose levels in WT and Hv1-deficient mice to a similar extent, indicating that Hv1 deficiency does not impair peripheral insulin sensitivity (Fig. 1K). Taken together, these data are compatible with the notion that loss of Hv1 results in impaired glucose tolerance because of an insulin secretion defect in vivo.

Hv1 deletion mice display an age-dependent development in glucose intolerance
To examine the age-dependent effect of Hv1 on glucose tolerance and insulin resistance, we performed IPGTTs and IPITTs in 2-and 6 month-old WT and KO mice as described above. The fasting blood glucose levels in KO mice (9.7 Ϯ 0.31 mmol/liter, n ϭ 10, p Ͻ 0.001 for 2 month-old mice; 10.63 Ϯ 0.47 mmol/liter, n ϭ 10, p Ͻ 0.001 for 4-month-old mice) were also significantly higher than in WT mice (6.42 Ϯ 0.27 mmol/ liter, n ϭ 10 for 2 month-old mice; 6.3 Ϯ 0.2 mmol/liter, n ϭ 10 for 4 month-old mice) ( Fig. 2A). There was almost no difference in the blood insulin levels in the fasted state between 2-and 6-month-old KO mice (Fig. 2B). The HOMA parameters calculated from the corresponding blood glucose and insulin levels showed a potential tendency for insulin resistance in 2-and 6-month-old KO mice (Fig. 2, C and E) and significant dysfunction of ␤-cells (Fig. 2D). The IPGTT showed more serious glucose intolerance in 6-month-old KO mice than in 2-month-old KO mice (Fig. 2, F and G), whereas WT mice showed no glucose intolerance at different ages. Although serum insulin levels in response to glucose in both 2-and 6-month-old KO mice were markedly lower than in WT mice (Fig. 2H), the IPITT showed no obvious insulin resistance in 2-and 6-month-old

Loss of Hv1 inhibits insulin secretion
KO mice (Fig. 2I), indicating that hyperglycemia in Hv1deficient mice is not caused by abnormal insulin sensitivity or T2D.
with Fig. 3A. Insulin secretion in the time windows of 0 -10 min and 10 -20 min in Fig. 3B was calculated as the first phase and second phase (Fig. 3D). The same results also occurred in perifusion assays. GSIS in KO mice was markedly decreased in the first and second phase (Fig. 3E). Our results clearly demonstrate that loss of Hv1 inhibits insulin secretion in the first and second phase.
We also investigated the effect of Hv1 on insulin granule biogenesis. Insulin and proinsulin content in KO islets was reduced by 17% (n ϭ 8, p Ͻ 0.001) and 25% (n ϭ 8, p Ͻ 0.01), respectively, compared with WT islets (n ϭ 8) under basal conditions (2.8 mM glucose) (Fig. 3, F and G). The ratio of insulin to proinsulin content, however, was indistinguishable between KO and WT islets (Fig. 3H), suggesting that insulin synthesis, but not insulin maturation, is abnormal in KO islets. Proinsulin secretion was barely detectable under basal conditions (2.8 mM glucose) in WT, Hv1 ϩ/Ϫ , and KO islets (data not shown). In the presence of 16.7 mM glucose, proinsulin secretion was reduced by 59% (n ϭ 8, p Ͻ 0.001) and 78% (n ϭ 8, p Ͻ 0.001) in heterozygous and KO islets compared with WT islets (n ϭ 8) (Fig. 3I). However, the ratio of insulin to proinsulin secretion in Hv1 ϩ/Ϫ and KO islets in the presence of 16.7 mM glucose was not different from WT islets (Fig. 3J). The ratio of insulin secretion to insulin content was remarkably reduced in islets of Hv1 ϩ/Ϫ and KO mice compared with WT mice (Fig. 3K). These results suggest that no significant accumulation of proinsulin occurred in KO islets and that the significant reduction in insulin secretion is not completely caused by insulin granule biogenesis.

Hv1 deficiency reduces ␤-cell mass and pancreatic insulin content
To determine whether Hv1 deletion affects islet development, we conducted immunohistological studies. Morphometric analysis of pancreatic sections from WT, Hv1 ϩ/Ϫ , and KO mice at 4 months of age showed relatively normal islet architecture in each case, with ␤-cells concentrated in the core and ␣-cells located mainly in the periphery (Fig. 4A), whereas the morphology of isolated islets cultured overnight from KO mice was not overtly different from WT islets (data not shown). On the other hand, the number of the isolated islets per pancreas was not significantly different between WT and KO mice (data not shown), which is consistent with the result from the immunohistochemical analysis (Fig. 4B). However, the average islet size, calculated from isolated islets (Fig. 4C), and islet area to total pancreas area ( Fig. 4D), analyzed by immunohistochemistry of pancreatic sections, were decreased in KO mice compared with WT and Hv1 ϩ/Ϫ mice.
Quantification of total ␤-cell mass displayed a genotype-dependent difference. ␤-Cell mass was decreased by 13% (n ϭ 6, p Ͻ 0.05) in KO mice compared with WT mice (n ϭ 6), as measured by morphometric analysis of insulin-positive islet cells (Fig. 4E). Total pancreatic insulin content in KO mice was also decreased by 11% (n ϭ 6, p Ͻ 0.05) (Fig. 4F), the same as the result obtained from isolated islets (Fig. 3F). These results show that KO mice have sufficient ␤-cells and insulin, indicating that the in vivo phenotype is not due to gross developmental defects.

Hv1 deficiency affects the size of insulin granules but not the number and docking of vesicles
Insulin is stored in large dense-core secretory granules in ␤-cells and released via granule exocytosis upon stimulation. The correct size, number, and docking to the cell membrane of insulin granules are necessary for insulin secretion in ␤-cells (19,20). We used TEM to investigate whether loss of Hv1 disturbed vesicle distribution in ␤-cells. The ratios of small vesicles (Ͻ300 nm) and large vesicles (Ͼ400 nm) were increased by 4-fold and decreased by 47% in KO mice compared with WT mice, respectively, but the ratio of the medium vesicles (300 -400 nm) showed no significant difference (Fig. 4, G and H). However, the total number of vesicles in ␤-cells in KO mice was the same as that in WT mice (Fig. 4I). This result might explain why there is a difference in insulin content between KO and WT mice (Fig. 3F). Docking of insulin granules plays an important role in regulating insulin secretion (20). A detailed quantitative electron microscopy analysis of ␤-cells was performed to study docked granules. The number of secretory granules close to (Ͻ100 nm) the cell membrane in ␤-cells in KO mice was similar to that in WT mice (Fig. 4J), indicating that loss of Hv1 does not affect granule docking in ␤-cells.
We further investigated the expression of syntaxin1A (Stx1a), vesicle-associated membrane protein 2 (VAMP2), and synaptotagmin-7 (Syt7), proteins related to insulin secretion, in isolated islets from WT and KO mice (21,22). Their expression levels were slightly down-regulated, as determined by Western blotting (Fig. 4K). The reduced membrane fusion protein suggests that lack of Hv1 may affect fusion of insulin granules with the plasma membrane.

Hv1 deficiency affects glucose-induced Ca 2؉ oscillation and limits membrane depolarization
Glucose stimulates insulin secretion by induction of Ca 2ϩdependent electrical activity, which triggers exocytosis of insulin granules. Glucose-stimulated calcium (GSCa) signaling and GSIS show similar trajectories (23). Thus, GSCa can be used to assess the physiological response of islets to glucose stimulation. Ca 2ϩ imaging is advantageous because it provides high temporal precision of real-time changes in response to stimuli at the level of the individual islet (24). The data presented in Fig.  5, A and B, are representative curves and average graphs of GSCa over time for islets isolated from WT and KO mice, respectively. The baseline, amplitude, timing, and duration of the biphasic response to glucose stimulation were calculated according to Fig. 5A. As shown in Fig. 5C, the mean value of cellular Ca 2ϩ levels for baseline under basal conditions was increased by 12% in KO islets compared with the WT, suggesting that elevated [Ca 2ϩ ] i under low glucose does not lead to increased insulin release. After stimulation, both WT and KO islets displayed a large spike in calcium influx, and the mean values for peaks showed difference between KO and WT islets (Fig. 5D). However, the mean duration was decreased by 32% in KO islets compared with the WT (Fig.  5E), and the time to peak was decreased by 31% (Fig. 5F),

Loss of Hv1 inhibits insulin secretion
suggesting that loss of Hv1 results in dissociation between calcium signaling and GSIS.
In pancreatic ␤-cells, the increase in [Ca 2ϩ ] i occurs with Ca 2ϩ entry across voltage-sensitive Ca 2ϩ channels activated by membrane depolarization (6). To confirm whether the change in calcium signaling by knockout of Hv1 is coupled to membrane polarization, the membrane potential difference was monitored with DiBAC4(3) fluorescence. As shown in Fig. 5, G and H, isolated islet ␤-cells from WT mice were depolarized significantly more than isolated islet ␤-cells from KO mice after glucose stimulation, indicating that Hv1 deficiency impairs glucose-induced membrane depolarization. Thus, the reduction in insulin secretion by Hv1 deficiency is involve in electrical activity and [Ca 2ϩ ] i signaling.

Loss of Hv1 inhibits insulin secretion Effect of Hv1 on cytosolic pH in pancreatic ␤-cells
To assess the effect of Hv1 on cytosolic pH in ␤-cells, cytosolic pH was evaluated by BCECF fluorescence. As shown in Fig. 6A, the pH in KO ␤-cells was lower about 0.1 than that in WT ␤-cells at both 2.8 and 16.7 mM glucose conditions. When we changed the glucose concentration from 2.8 to 16.7 mM, there was a slight synchronous reduction in pH in WT and KO ␤-cells (Fig. 6A). The pH under steady-state basal (2.8 mM) and high-glucose (16.7 mM) conditions in KO ␤-cells was also slightly lower compared with the corresponding WT ␤-cells, but there were no significant differences between them (Fig.  6B), indicating that knockout of Hv1 has no a significant effect on cytosolic pH in ␤-cells stimulated by glucose.
PMA is a well-known activator of Hv1 and has been used extensively for Hv1 function studies (25,26). We measured PMA-induced insulin secretion in isolated islets from WT and KO mice. As shown in Fig. 6E, in the presence of 5 M PMA, insulin secretion in isolated islets from WT mice increased 6.3fold compared with the absence of PMA, whereas insulin secretion in KO islets was significantly decreased by 61% (p Ͻ 0.001) compared with WT islets.
To estimate the effect of PMA on cytosolic pH, the change in fluorescence intensity at 488 nm excitement was observed with addition of PMA solution. Interestingly, the fluorescence intensity for KO ␤-cells was significantly reduced compared with that of the WT (Fig. 6, C and D), indicating that PMA results in a decrease in cytosolic pH.

Hv1 is down-regulated in pancreatic ␤-cells in STZ-induced diabetic mice
To further determine the relationship between Hv1 and insulin secretion, WT and KO mice were treated with STZ to induce diabetes. We found that fasted blood glucose levels after STZ injection in KO mice were always higher than in WT mice (Fig. 7A), and the IPGTT also showed higher blood glucose levels in STZ-treated KO mice than in WT mice (Fig. 7B). These data show that STZ-treated KO mice seem to be more sensitive to diabetes induction.
The insulin content at 2.8 mM glucose was decreased by 91% in ␤-cells of STZ-treated WT mice (Fig. 7C), whereas insulin secretion stimulated by 16.7 mM glucose was reduced by 97% (Fig. 7D), which is a serious insulin deficiency. Immunohistochemical analyses of islets in control and STZ-treated WT mice using anti-insulin and anti-Hv1 antibodies also showed that the mean optical density of insulin-and Hv1-positive areas in STZtreated WT mice were decreased by 53% and 60%, respectively, compared with control mice (Fig. 7E), demonstrating that the expression levels of insulin and Hv1 are significantly decreased in ␤-cells of STZ-treated diabetes mice, as shown in Fig. 7E, b and d. The lower expression of Hv1 in ␤-cells of diabetic mice

Loss of Hv1 inhibits insulin secretion
might be due to the decrease in insulin secretion, which is consistent with the result that insulin secretion is significantly reduced in islets of Hv1 ϩ/Ϫ mice (Fig. 3A). These data suggest that Hv1 is closely related to insulin secretion.

Hv1 is down-regulated in high-glucose-induced dysfunctional ␤-cells
Chronic hyperglycemia can cause loss of GSIS (27). To confirm whether Hv1 is also down-regulated in dysfunctional ␤-cells, INS-1 (832/13) cells were incubated in 11 or 25 mM glucose, respectively, for 48 h. Following chronic incubation, insulin secretion was measured at two different concentrations of glucose (2.8 and 16.7 mM), and Hv1 expression levels were detected by immunofluorescence.
Insulin secretion in INS-1 (832/13) cells incubated chronically at 11 mM glucose was increased in response to glucose stimulation but completely unchanged with 25 mM glucose chronic incubation (Fig. 7F). These data are consistent with another report (27) showing that chronic high glucose causes complete loss of glucose responsiveness in INS-1 (832/13) cells. Immunofluorescence analyses of cells under 11 or 25 mM glu-cose chronic incubation using anti-Hv1 antibody showed that the mean fluorescence intensity in 25 mM glucose chronic incubation cells was decreased by 33.5% compared with 11 mM (Fig.  7G). These data demonstrate that the Hv1 expression level is also significantly decreased in the dysfunctional ␤-cells, which is in accordance with the STZ-induced diabetes model.

Discussion
Here we show in vivo that Hv1 KO and heterozygous mice display hyperglycemia and glucose intolerance because of markedly decreased insulin secretion. In vitro, Hv1 deficiency causes a remarkable defect in glucose-and K ϩ -induced insulin secretion, limiting the capacity for glucose-induced membrane depolarization that accompanies the reduced ability of glucose to raise the level of Ca 2ϩ in islets, as evidenced by the decreased duration of individual calcium oscillations. These data indicate that Hv1 is required for insulin secretion and maintenance of glucose homeostasis and reveal a significant role of the proton channel in modulation of pancreatic ␤-cell function.
Hv1 is extremely selective for protons and has no detectable permeability to other cations (13,14). Gating of Hv1 strongly depends on membrane depolarization and intracellular and extracellular pH (13)(14)(15)26). The proton channel also sustains Ca 2ϩ influx and enables neutrophils to generate calcium signals in response to chemoattractants, in addition to maintenance of normal cytosolic pH and membrane potential during the respiratory burst (15). Our previous work demonstrated that Hv1 mainly localizes to the insulin-containing granule membrane (17). Considering fusion of insulin-containing granules with the plasma membrane during exocytosis, Hv1 should also be in the plasma membrane in pancreatic islet ␤-cells. Therefore, Hv1 might be involved in regulating insulin-containing granules, cytosolic pH, membrane potential, and Ca 2ϩ influx in pancreatic ␤-cells.
Manipulation of extracellular and intracellular pH in pancreatic islets or ␤-cells is associated with changes in membrane potential, ionic flux, and insulin release (9 -11). Glucose-induced changes in cytosolic pH in islet ␤-cells are coupled to glucose metabolism and associated with triggering insulin release (10,11). Glucose-induced priming of insulin secretion, which is thought to be mediated by the amplifying pathway (28), has been proposed to be linked to pH changes in ␤-cells (29). It has been suggested for decades that the proton gradients in insulin-containing granules might be involved in fusion of secretory vesicles to the target membrane (30, 31), which might affect triggering of vesicles for secretion, a process involving pairing of SNARE proteins on vesicles and target membranes to establish fusion competence (12). Dependence of insulin secretion on Hv1 activity reflects that insulin-containing granules and cytosolic pH are directly involved in secretion of secretory granules. A change in membrane potential in Hv1-deficient pancreatic ␤-cells under glucose stimulation condition also reflects the pH dependence of the secretory machinery as well as altered sensitivity of the secretory machinery to [Ca 2ϩ ] i .
The fact that heterozygous mice also have hyperglycemia with a low insulin level illustrates that Hv1 is at an important control point in the metabolic pathway regulating insulin secretion and that relatively small changes in Hv1 activity are likely to have important effects on insulin secretion. Similar effects

Loss of Hv1 inhibits insulin secretion
have been observed for glucokinase (32). Hv1 expression levels are significantly decreased in ␤-cells of STZ-treated diabetes mice and mice with chronic hyperglycemia-induced dysfunction. These data further confirm that Hv1 is closely related to insulin secretion. The findings of this study clearly demonstrate that Hv1 plays an important role in positively regulating GSIS.
␤-Cell failure is associated with not only decreased ␤-cell insulin secretory function but also reduced overall ␤-cell mass (33). In this study, there was no difference in islet morphology between KO and WT mice and only a very modest decrease in islet size and ␤-cell mass. The smaller size of Hv1-deficient pancreatic islets may be related to the decrease in ␤-cell mass and result from impaired insulin secretion function. It is impor-tant to note that, in vitro, siRNA-mediated knockdown of Hv1 in isolated islets and INS-1 (832/13) cells caused decreased GSIS (17), suggesting that the in vivo decrease in insulin secretion in KO mice was not due to an in vivo ␤-cell developmental defect.
Glucose metabolism generates ATP and closes ATP-regulated K ϩ channels, leading to membrane depolarization. Membrane depolarization of ␤-cells leads to activation of voltagesensitive L-type Ca 2ϩ channels with a subsequent rise in intracellular Ca 2ϩ , which then drives vesicular exocytosis (34). Sulfonylureas, such as tolbutamide and glibenclamide, bind to the SUR subunit of ATP-sensitive K ϩ channels and induce channel closure (35). Knockout of Hv1 inhibits tolbutamide-

Loss of Hv1 inhibits insulin secretion
and glibenclamide-induced insulin secretion (data not shown), suggesting that the effect of Hv1 on insulin secretion is downstream of ATP-sensitive K ϩ channel closure but upstream of the final exocytotic event.
PMA is a well-known Hv1 activator (13,14). Our results show that loss of Hv1 strikingly inhibits PMA-stimulated insulin secretion, demonstrating that PMA regulates insulin secretion by activating Hv1 activity. PMA treatment results in a decrease in cytosolic pH in Hv1-deficient ␤-cells, suggesting that Hv1 is involved in regulating cytosolic pH. It is noteworthy that Hv1-deficiet neutrophils were acidified upon stimulation with PMA, which indicates that Hv1 extrudes the acid generated in the cytosol during activation of neutrophils by PMA (15). It is possible that a similar mechanism exists in islets. Glucose metabolism accompanies proton generation, and therefore, ␤-cells must have a dynamic intracellular pH-regulatory system (9). The requirement of Hv1 for insulin secretion in islet ␤-cells suggests that pH regulation is a link in stimulussecretion coupling in ␤-cells. pH regulation in pancreatic islets and ␤-cells in response to glucose might be mainly due to activation of Hv1.
The proposed mechanism of Hv1 in signal transduction pathways relevant to glucose-and PMA-induced insulin secretion is shown in Fig. 8. Glucose metabolism generates ATP and results in depolarization of the plasma membrane because of closure of ATP-sensitive K ϩ channels (K ATP channels) (36). Depolarization leads to Ca 2ϩ influx through voltage-dependent Ca 2ϩ channels (6), opens the Hv1 channel that affects the oscillations in membrane potential (V m ) and cytosolic Ca 2ϩ levels in the feedback path, and then triggers insulin exocytosis. ATP is also converted into cAMP and activates PKA and also leads to closure of K ATP channels. PMA, as an Hv1 activator, activates PKC (26) and opens Hv1, resulting insulin secretion. The important defect that ablates the Hv1 gene could be predicted from the reduced depolarization in ␤-cells and the reduced ability of glucose to raise the level of Ca 2ϩ in islets, as we observed in this work. Without activation of Hv1, depolarization by closure of the K ATP channel might be weakened a lot, and these distal effects could not occur.
Blood glucose control is a complex metabolic process involving numerous tissues and organs. Global knockout of Hv1 might affect glucoregulation in many tissues and organs. Insulin is the only hypoglycemic hormone secreted from pancreatic ␤-cells. Hv1 is highly expressed in pancreatic ␤-cells (17). Knockout of Hv1 remarkably reduces glucose-stimulated insulin secretion in isolated islets and serum insulin levels in mice, suggesting that hyperglycemia and glucose intolerance in Hv1KO mice are due to decreased insulin secretion in pancreatic ␤-cells. The decreased insulin and Hv1 expression in pancreatic islets in diabetes model mice imply a link of Hv1 with insulin secretion and diabetes. Taken together, the data in this study demonstrate that Hv1-deficient mice exhibit hyperglycemia and impaired glucose tolerance because of reduced insulin secretion. This finding provides direct evidence of a functional role of the proton channel in maintenance of glucose homeostasis. This study describes a novel pathway regulating ␤-cell secretory function in which Hv1 sustains entry of calcium ions into ␤-cells by regulating membrane depolarization to promote increased insulin secretion responses.

Animals and treatments
Mice bearing a targeted disruption in the VSOP/Hv1 (VSOP/ Hv1 Ϫ/Ϫ , backcrossed eight times, kindly provided by Dr. Y. Okamura, School of Medicine, Osaka University), as described previously (37). WT mice (VSOP/Hv1 ϩ/ϩ ) were of the same genetic background (C57BL/6J). Animals were kept in a pathogen-free facility under a 12-h light/dark cycle with access to water and a standard mouse diet (Lillico Biotechnology). Genotyping was performed by PCR as described by Ramsey et al. (25). Experiments were performed with 2-to 4-month-old male mice unless indicated otherwise. All animal husbandry and experiments were approved by and performed in accordance with guidelines from the Animal Research Committee of Nankai University.

Blood glucose and insulin determination
Blood glucose levels were measured from blood obtained from the tail vein after fasting for 6 h using an automated glucometer (One Touch, Johnson & Johnson). For glucose tolerance tests, mice 4 months of age were fasted for 6 h before i.p. injection of glucose (2 g/kg of body weight), and blood glucose concentration was measured 0, 15, 30, 60, and 120 min after injection. For insulin tolerance tests, mice were fasted for 6 h before intraperitoneal injection with insulin (1.0 units/kg of body weight), and blood glucose concentration was measured at 0, 15, 30, 60, and 120 min.
For serum insulin measurements, glucose (2 g/kg of body weight) was injected i.p. Venous blood was collected at 0, 2, 5,  15, and 30 min in chilled heparinized tubes and immediately centrifuged, and the serum was stored at Ϫ80°C. Insulin levels were measured by ELISA (Mercodia).

Insulin secretion assay
Insulin secretion from isolated islets was measured as described previously (17). Briefly, after overnight culture, the islets were aliquoted into microtubes (20 size-matched islets/ tube) and preincubated for 1 h in KRBH-BSA buffer containing 2.8 mM glucose at 37°C. Then the islets were incubated with different stimuli in KRBH-BSA buffer for 1.5 h. To measure biphasic insulin secretion, islets were first stabilized for 30 min in KRBH-BSA buffer containing 2.8 mM glucose at 37°C. Following establishment of a baseline with 2.8 mM glucose in KRBH-BSA buffer for 10 min, islets were picked by hand and plated into KRBH-BSA buffer containing 16.7 mM glucose for 10 min and then in another dish for 20 min. Insulin levels in fractions were measured. Islet perifusion was performed at 37°C using a system designed in-house. Forty equal size islets were paired between WT and KO mice in each experiment. Islets were perifused with KRBH-BSA buffer. The flow rate was controlled by a perfusion pump at 500 l/min, and fractions were collected in microtubes for insulin determination. Total insulin and proinsulin content in isolated islets was extracted with acidic ethanol and determined using rat/mouse proinsulin and mouse and rat insulin ELISA (Mercodia) according to the manufacturer's protocol.

Immunohistochemistry
Immunohistochemistry was performed as described previously (17).

Transmission EM (TEM)
Islets isolated from six 4-month-old animals were fixed in 2.5% glutaraldehyde for ultrathin (70 nm) sectioning and imaging by TEM (Hitachi HT7700). For quantification of vesicles, images were captured using TEM at magnification ϫ10,000 -15,000. Docked vesicles were counted, with vesicles whose outer surface was within 100 nm of the plasma membrane considered docked granules. At least 50 random sections were used to capture microscopic fields from six mice per group. Then about 50 cells of each genotype were analyzed before identifying their genotype.

Measurements of intracellular Ca 2؉
Isolated islets were incubated in KRBH buffer with basal glucose containing 2 M Fura-2/AM dye (Invitrogen) for 30 min at 37°C. The glass coverslips were placed on the stage of an inverted microscope (Eclipse Ti, Nikon). Epifluorescence mode with a ϫ10 objective was used, stepping to 340 nm excitation and 380 nm excitation for 100 ms every 3 s. Regions of interest were outlined and monitored simultaneously during the whole experimental procedure using MetaFluor software. The resultant fluorescent signals of Fura-2 were successfully monitored at a wavelength of 535 nm with the help of a digital CMOS camera (Zyla 4.2 Plus, Undor). Individual images and intensity values were recorded along with the emission data. After data acquisition, an area of the coverslip without cells was measured as background. Background fluorescence was then subtracted, and the ratio of emission with 340 nm excitation to emission with 380 nm excitation was calculated, showing the F340/F380 ratio for real-time intensity measurements.

Measurements of membrane potential
The voltage-sensitive bisoxonol fluorescent dye DiBAC4(3) was used to study the membrane potential changes of dispersed islet ␤-cells. Freshly isolated mouse islets were dispersed at room temperature into single cells by addition of 0.1% trypsin-EDTA solution for 2 min with gentle pipetting. The dispersed cells were washed with RPMI 1640 medium containing 10% FBS and placed on 0.01% poly-L-lysine-precoated glass coverslips for 12 h before experiments. The cells were incubated with 1 M DiBAC4(3) for 30 min at 37°C in basal glucose in KRBH buffer prior to fluorescence measurement. Data were expressed as an average of three experiments (80 -100 cells/experiment).

Measurements of cytosolic pH
Cytosolic pH (pH c ) was assessed using BCECF/AM (Invitrogen) as described previously (40). ␤-Cells seeded on glass coverslips were loaded with 2 g/ml of BCECF-AM (Molecular Probes) at 37°C for 30 min, inserted in a thermostatic chamber, and imaged through a ϫ10 objective. To determine the cytosolic pH, fluorescence at excitation wavelengths of 436 and 495 nm was recorded at an emission wavelength of 535 nm. Calibration of fluorescence versus pH was performed by equilibration of external and internal pH with nigericin (10 M) in a high K ϩ buffer with a pH range of 5.5 to 8.0. High-K ϩ buffer contained 145 mM KCl, 2.8 mM glucose, 1 mM CaCl 2 , 1 mM MgCl 2 , and 20 mM HEPES (or MES). The relative fluorescence ratio values were plotted against corresponding pH c values, which allowed determination of the unknown pH c . The data were expressed as an average of three experiments (50 -100 cells/ experiment). To observe the change in cytosolic pH induced by PMA (an activator of Hv1), fluorescence at an excitation wavelength of 488 nm was recorded at an emission wavelength of 535 nm. Analysis was performed with MetaMorph software. To exclude any contribution of the Na ϩ /H ϩ antiporter, experi-

Loss of Hv1 inhibits insulin secretion
ments were performed in Na ϩ -free medium containing 121 mM N-Methyl-D-glucaMine-Cl, 5.6 mM KCl, 1.2 mM MgCl 2 , 2.6 mM CaCl 2 , 5 mM HEPES, and 5 mM glucose (pH 7.4). The changes in cytosolic pH were estimated as the relative changes of the initial fluorescence intensity (F/F 0 ). The data were expressed as an average of three experiments (20 -30 cells/experiment).

Streptozotocin (STZ) diabetes model
To induce diabetes, WT and Hv1 Ϫ/Ϫ mice were treated with STZ (50 mg/kg of body weight i.p. for 5 consecutive days). Additionally, some WT and Hv1 Ϫ/Ϫ mice were treated with saline as vehicle controls. Blood glucose levels were measured once every week with fasting for 6 h after the final injection of STZ. At the end of the fourth week, an IPGTT was performed, using an i.p. injection of glucose at 2 g/kg of body weight after 6 h of fasting. Blood glucose was analyzed 0, 15, 30, 60, and 120 min after introducing glucose. Diabetic hyperglycemia was defined as a fasting blood glucose concentration of more than 11.1 mM for two or more consecutive tests.

Statistical analysis
All statistics were performed using SPSS20.0 software. Comparison of the mean between groups was performed by t test. p Ͻ 0.05 was considered significant.