Rab27a mediates the tight docking of insulin granules onto the plasma membrane during glucose stimulation (original) (raw)
Expression of Rab27a, Rab27b, and granuphilin in pancreas. We first examined the expression of Rab27a and its related molecules in pancreatic islets (Figure 1A). As described previously (9), Rab27a protein was highly expressed in pancreatic islets of control C3H/He mice. By contrast, it was not detected in the islets of ashen mice, which is consistent with previous reports showing the lack of detectable Rab27a protein in _ashen_-derived cytotoxic T lymphocytes (13, 14) and platelets (15). Because the Rab27a protein is truncated due to a splice site mutation in ashen mice (8), the mutant protein either is unstable or loses its immunoreactivity. In either case, the mutant Rab27a protein is probably nonfunctional because it lacks some of the GTP-binding pockets. Rab27b, on the other hand, was barely detectable in the islets but was significantly expressed in the pituitary of control mice, as reported previously (16). The expression level of granuphilin, a Rab27a effector in pancreatic β cells (9), was not altered in ashen islets, in contrast to another Rab27a effector, melanophilin, which is severely downregulated in ashen melanocytes (17).
Expression of Rab27a, Rab27b, and granuphilin in pancreatic islets. (A) An equal amount of protein (20 μg) from the pancreatic islets of 17-week-old male C3H/He (lane 1) and ashen mice (lane 2) was separated by electrophoresis for immunoblotting with anti-Rab27a, anti-Rab27b, or anti-granuphilin (αGrp-aC) antibodies. The expression levels of α-tubulin were also examined for normalization. For the immunoblotting with anti-Rab27b and anti–α-tubulin antibodies, 20 μg of protein from the pituitary of C3H/He mice were loaded on lane 3 for the reference. Numbers to the left of each panel are molecular masses in kDa. (B) The pancreas organs of 17-week-old male C3H/He (upper) or ashen mice (lower) were immunostained with anti-granuphilin antibodies (αGrp-N). Granuphilin is distinguishably concentrated along the plasma membrane in ashen β cells (arrowheads) compared with control β cells, although the expression levels are similar. Scale bar: 20 μm.
Immunohistochemical analysis of mouse pancreas specimens confirmed that Rab27a was expressed in insulin-positive β cells of control C3H/He mice, as reported previously (9), but not in those of ashen mice (data not shown). By contrast, Rab27b was expressed in exocrine and peripheral islet cells, but not in β cells (H. Gomi, S. Zhao, and T. Izumi, unpublished observations). The islet architecture of ashen mice was normal, including a mantle structure between β and non-β cells, and the expression levels of pancreatic hormones were not significantly reduced in ashen islets (data not shown, but see Figure 1B). While the lack of Rab27a did not affect the expression level of its effector granuphilin (Figure 1A), its intracellular location was significantly affected. The distribution of granuphilin was distinguishably shifted to the plasma membrane in ashen β cells compared with its diffuse cytoplasmic distribution in control β cells (Figure 1B). Alteration in the subcellular localization in Rab27a-deficient β cells strongly supports our previous proposals that granuphilin is associated with secretory granules through Rab27a and that it is an effector of Rab27a, but not of Rab3a, in vivo, despite its affinities to both Rab proteins in vitro (9).
Glucose tolerance and related in vivo phenotypes. Because Rab27b is coexpressed and functionally redundant with Rab27a in platelets (15), the lack of Rab27b in the pancreatic β cells of ashen mice makes them an excellent subject of study for the evaluation of Rab27a function. We first examined in vivo phenotypes that might be affected by potential insulin secretion defects in ashen β cells. Body weight (Figure 2A) and blood glucose levels examined in mice fed ad libitum (data not shown) or in fasting mice (Figure 2B) did not differ between ashen and C3H/He mice. Blood glucose levels after a glucose load, however, were significantly higher in ashen mice (Figure 2B). Although it is generally difficult to measure small changes in plasma insulin levels accurately from a limited amount of samples in mice, unless they have marked hyperinsulinemia (see the examples in ref. 18), plasma insulin levels at fasting or during a glucose tolerance test were not significantly different, except at 120 minutes after the glucose load, at which point the levels were slightly higher in ashen mice (data not shown). Although the hyperglycemia with normoinsulinemia or slight hyperinsulinemia upon glucose challenge might suggest the presence of insulin resistance, insulin tolerance tests (0.75 U human insulin/kg body weight) revealed comparable insulin sensitivity between ashen and control mice (Figure 2C). The tests with smaller doses of insulin (0.1–0.5 U/kg) after overnight fasting also showed similar insulin sensitivity (data not shown). Insulin content in total pancreas was increased by 1.3-fold in ashen mice compared with controls (Figure 2D), which may reflect an impairment of insulin secretion. These findings indicate that ashen mice have glucose intolerance without signs of insulin resistance in peripheral tissues or absolute insulin deficiency in pancreatic β cells.
In vivo phenotypes of ashen and C3H/He mice. All the phenotypes were derived from male C3H/He (white bars or open circles) or ashen (black bars or filled circles) mice. Body weight (A), blood glucose concentrations during an intraperitoneal glucose tolerance test (B), and total insulin content in the pancreas (D) were measured at 15 weeks of age. An intraperitoneal insulin tolerance test (C) was performed at 16 weeks of age, and its results are expressed as a percentage of the initial blood glucose concentration. Values are mean ± SE (A and B, n = 8 for C3H/He, n = 7 for ashen; C, n = 5 for C3H/He and ashen; D, n = 8 for C3H/He and ashen). The statistical significance of differences between means was assessed by Student’s t test. *P < 0.05, **P < 0.005, ***P < 0.0005 vs. C3H/He mice.
Insulin secretion profiles in perifused islets. Because the overexpression of active forms of Rab27a has been shown to augment insulin secretion in the cultured β-cell line MIN6 (9), the lack of functional Rab27a may induce insulin secretion defects in ashen mice. We incubated pancreatic islets, isolated from C3H/He or ashen mice, overnight in culture medium to decrease potential environmental effects, such as differences in blood glucose levels, in vivo. They were then examined for the ability to secrete insulin in response to several secretagogues by perifusion analyses. Insulin secretion in response to a high glucose level (16.7 mM) was significantly reduced in both the first (approximately 5 minutes) and second phases in ashen islets (repeated measure ANOVA, P < 0.0001), and the accumulated glucose-induced insulin secretion over 30 minutes was diminished by approximately 50% compared with the control islets (Figure 3A). Surprisingly, the reduction in insulin release was specific to glucose stimulation. Insulin secretion in response to depolarization induced by high K+ concentration (60 mM) was normal in ashen islets (Figure 3B). Furthermore, both forskolin (activator of adenylate cyclase) and phorbol-12-myristate-13-acetate (PMA; protein kinase C activator) stimulated insulin secretion in the presence of high glucose levels indistinguishably between ashen and control islets (Figure 3, C and D). These findings indicate that the defect in insulin secretion in ashen islets does not simply reflect a general decrease in the releasable pool of granules. Consistent with this idea, insulin secretion was normal even after repeated exposure to high concentrations of K+ in ashen islets (Figure 3F). By contrast, insulin secretion was decreased at each of the repeated exposures to the glucose stimuli (Figure 3E; P < 0.0001), suggesting that Rab27a is specifically required for the replenishment of a pool of granules released by glucose.
Insulin secretion profiles of perifused islets. Insulin secretion was examined in islets isolated from age-matched (15–17 weeks of age) male C3H/He (open symbols) or ashen mice (filled symbols). The islets were stabilized by perifusion of standard low-glucose (2.8 mM) Krebs-Ringer buffer for 10 minutes, after which an appropriate secretagogue was applied. (A and B) Islets were stimulated by 16.7 mM glucose for 30 minutes (A; n = 9 for each strain) or with 60 mM KCl for 15 minutes (B; n = 6). (C and D) Islets were stimulated by 16.7 mM glucose for 30 minutes in the continuous presence of either forskolin (C, 10 mM; n = 8) or PMA (D, 0.5 mM; n = 8). (E and F) Islets were stimulated 4 times by either 16.7 mM glucose for 20 minutes (E; n = 3) or by 60 mM KCl for 10 minutes (F; n = 3), with 10-minute intervals of the standard buffer. (G) Islets were stimulated by 60 mM (squares), 30 mM (circles), or 20 mM KCl (triangles) for 15 minutes, followed by the standard buffer for 15 minutes (n = 3 for each condition and strain). (H) Islets were perifused with buffer containing 250 μM diazoxide and 16.7 mM glucose for 10 minutes. They were further perifused with the buffer containing 30 mM KCl for 30 minutes, in the continuous presence of diazoxide and glucose (n = 3 for each strain). Values are mean ± SE.
The stimulation with high K+, forskolin, or PMA induced much higher insulin secretion compared with the 16.7 mM glucose stimulation (note differences in the ordinates of each panel in Figure 3). Stronger stimuli might have obscured subtle functional differences between the 2 kinds of islets. The lower concentrations of KCl (20 mM or 30 mM) induced an insulin secretion at the peak that was equivalent to that induced by 16.7 mM glucose during perifusion of the control islets (Figure 3, A and G). The same treatment in ashen islets evoked an indistinguishable amount and manner of insulin secretion (Figure 3G). Therefore, ashen β cells showed specific defects in insulin secretion in response to glucose, but not to high K+, independent of the strength of the depolarization stimuli. These findings indicate qualitative differences in the mode of exocytosis induced by glucose and that induced by other, nonphysiological secretagogues.
Although high K+ stimulation physically induces membrane depolarization and a subsequent rise of intracellular Ca2+ concentration ([Ca2+]i), glucose stimulates intracellular metabolism and ATP production, which is thought to induce additional effects as well as triggering depolarization by the closure of ATP-sensitive K+ (KATP) channel. This amplifying effect by glucose has classically been evaluated by the amount of insulin secreted in the presence of diazoxide, an opener of KATP channels (19, 20). As expected, 250 μM diazoxide inhibited insulin secretion induced by 16.7 mM glucose in both ashen and control islets (see the insulin secretion during the first 10 minutes in Figure 3H). To induce insulin secretion, the islets were depolarized by 30 mM KCl in the continuous presence of diazoxide and 16.7 mM glucose. The amount of insulin secreted was not significantly different, though slightly reduced, in the ashen islets compared with the control islets (Figure 3H). These findings suggest that the amplifying pathway by glucose is largely intact in ashen β cells.
Analysis of the exocytosis of insulin granules by two-photon excitation imaging. To explore the mechanism responsible for the decrease in glucose-induced insulin secretion, we directly examined the exocytosis of granules by 2-photon excitation imaging (21). This newly innovated morphological technique enables us to precisely monitor individual exocytotic events per the constant area, as well as other parameters relevant to exocytosis, in isolated islets. Both the amplitude and time course of the glucose-induced rise in Ca2+ concentration measured using Fura-2 was not different between ashen and C3H/He mice (Figure 4 and Table 1), which suggests normal glucose metabolism and ATP production in ashen β cells. Fusion pore dynamics examined by measuring latency for the onset of staining with fluorescent markers of different sizes was also unchanged in ashen islets (Table 1). However, the exocytotic events during the first and second 5 minutes were significantly reduced (Table 1), which is consistent with the findings of the perifusion assays (Figure 3A). These results independently demonstrated the reduction of glucose-induced fusion events and excluded the possibility of changes in the dynamics of [Ca2+]i or fusion pore opening in ashen islets.
The rise in cytosolic Ca2+ concentration in response to glucose stimulation. The cytosolic Ca2+ concentration was measured using Fura-2 acetoxymethyl ester by two-photon excitation imaging in pancreatic β cells of either C3H/He (gray) or ashen (black) mice and was represented by (F_0–_F)/_F_0, where _F_0 and F stand for resting and fluorescence and fluorescence after 20 mM glucose stimulation, respectively. Mean values are shown (n = 4).
Analysis of the exocytosis of insulin granules by 2-photon excitation imaging
Analysis of the exocytosis of insulin granules by total internal reflection fluorescence microscopy. We next utilized total internal reflection fluorescence microscopy (TIRFM) to investigate the exocytotic events occurring just beneath the plasma membrane in isolated pancreatic β cells. TIRFM detects cytoplasmic events within 100 nm of the plasma membrane (22). We first immunostained endogenous insulin in fixed pancreatic β cells. Because fluorescent imaging by TIRFM can depict the single insulin granules closely associated with the plasma membrane (23), we counted the number of morphologically docked granules (Figure 5A). The ashen β cells showed a marked decline in the number (130 ± 34 per 200 μm2 in ashen β cells vs. 219 ± 34 per 200 μm2 in C3H/He β cells; n = 24 for ashen and n = 21 for C3H/He) to 59% of normal levels (P < 0.0001). These findings suggest that granules docked onto the plasma membrane are reduced in ashen β cells.
TIRFM analysis of the exocytosis of insulin granules. (A) Pancreatic β cells from either C3H/He (left) or ashen (right) mice were fixed, immunostained with anti-insulin antibodies, and observed by TIRFM. The surrounding lines represent the outline of cells that are attached to the cover glass. Scale bar: 5 μm. (B) Pancreatic β cells of either C3H/He (upper) or ashen (lower) mice were infected with adenoviruses encoding insulin-GFP. Evanescent images in live cells were acquired every 300 milliseconds after glucose stimulation. The fusion events per 200 μm2 were manually counted. The histograms show the number of fusion events at 60-second intervals after high-glucose (16.7 mM) stimulation. The black bars show the fusion from previously docked granules, whereas the white bars represent that from newly recruited granules. Values are mean ± SE (n = 24 for ashen and n = 21 for C3H/He mice). The statistical significance of differences between means were assessed by repeated-measures ANOVA. (C) The number of morphologically docked granules was counted on the TIRFM images of pancreatic β cells that had been infected with adenoviruses encoding insulin-GFP. The number after a glucose stimulation (16.7 mM, 15 minutes) was normalized to that prior to the stimulation in the identical area (200 μm2) of C3H/He (open circles) or ashen (filled circles) β cells. Values are mean ± SE (n = 4). The statistical significance of differences between means was assessed by Student’s t test. *P < 0.0005 vs. C3H/He mice.
We then observed fusion events of fluorescence-labeled granules by TIRFM in live β cells that had been infected with an adenovirus encoding GFP-tagged insulin (23). The fusion events in response to 30 mM KCl occurred predominantly during the first 5 minutes after the stimulation and were not significantly different between ashen and control β cells (40.1 ± 5.9 in 0–5 minutes in ashen β cells vs. 56.7 ± 6.9 in 0–5 minutes for C3H/H3 β cells, n = 4 each). There were no significant differences in the exocytosis of previously docked granules (4.5 ± 2.1 in ashen β cells vs. 7.5 ± 2.0 in C3H/He β cells) or in that of newly recruited granules (34.4 ± 4.4 in ashen β cells vs. 47.7 ± 6.9 in C3H/He β cells). The fusion events in response to 16.7 mM glucose, however, were significantly decreased (Figure 5B; 15.4 ± 2.5 in 0–5 minutes in ashen β cells vs. 34.5 ± 5.6 in 0–5 minutes in C3H/He β cells, P < 0.05; and 32.5 ± 4.2 in 5–16 minutes vs. 72.4 ± 11.5 in 5–16 minutes, P < 0.0001; n = 24 for ashen and n = 21 for C3H/He). Remarkably, few fusion events occurred from previously docked granules in ashen β cells (5.8 ± 1.3 in 0–16 minutes in ashen β cells vs. 26.3 ± 4.3 in 0–16 minutes in C3H/He β cells). The exocytosis of newly recruited granules was relatively modestly affected, although it was also significantly reduced (42.1 ± 4.2 in 0–16 minutes in ashen β cells vs. 80.6 ± 11.2 in 0–16 minutes in C3H/He β cells). We then analyzed the changes in the total number of docked granules during glucose stimulation using the TIRFM images. Comparison of the numbers of docked granules in the identical area before and after glucose stimulation should eliminate the effect of potential differences in infection efficiency between cells. The number of docked granules after 15 minutes of glucose stimulation remained 80.3% ± 2.1% of the initial number in the control β cells (Figure 5C). By contrast, it was markedly decreased to 32.1% ± 6.6% in ashen β cells, indicating that refilling of the pool of docked granules is defective.
Analysis of insulin granules by EM. We next performed the EM examination of pancreatic β cells from fixative-perfused mice. It did not reveal any obvious alterations in the number, distribution, or appearance (dense-core formation) of granules in ashen β cells (data not shown, but see Figure 6A), except that the average diameter was slightly decreased (298 ± 7 nm in ashen β cells vs. 322 ± 7 nm in C3H/He β cells, P < 0.05; n = 24 for ashen and n = 25 for C3H/He; note that the values represent the mean of the profile diameter in sections but not that of the maximum diameter). To determine whether any of the changes in glucose-stimulated exocytosis identified here are correlated with the finding of EM, islets were first isolated from mice and incubated with either 2.8 mM or 25 mM glucose, and then chemically fixed and subjected to morphometric analysis (Figure 6A). Granules were divided according to the distance from their center to the plasma membrane, and the density of granules in concentric shells below the plasma membrane was calculated relative to the average density of all cytoplasmic granules. Although the density of granules whose centers were located 300–500 nm from the plasma membrane was not significantly different from the average, the density of granules whose centers were located within 100 nm of the plasma membrane was much lower in both nonstimulated and stimulated control β cells (Figure 6B, left). Given that the diameter of insulin granules is 300–350 nm, which is equivalent to the previously reported value (24), the latter finding is not surprising because the center of granules should not approach the plasma membrane within their radius. By contrast, the density of granules whose centers reside at 100–300 nm significantly increased above the average, which may indicate the accumulation of granules docked onto the plasma membrane. This fraction is thought to correspond to the granules visualized by TIRFM considering the TIRFM coverage area (within 100 nm from the plasma membrane) and the mean diameter of granules (300–350 nm). Interestingly, the density of this fraction was significantly augmented by the exposure to 25 mM glucose in the control β cells. By contrast, the glucose-dependent accumulation of granules beneath the plasma membrane was lacking in ashen β cells, although the densities of other fractions were not significantly different from those in the control β cells (Figure 6B, right). These findings suggest that glucose has the ability to replenish a pool of docked granules and that this glucose-driven refilling process is impaired in ashen β cells.
Ultrastructure of the pancreatic β cells. (A) Electron micrographs of β cells were taken in nonstimulated (upper) or glucose-stimulated (lower) islets of C3H/He (left) and ashen (right) mice. Scale bar: 1 μm. (B) Relative density of granules below the plasma membrane in nonstimulated (white bars) or glucose-stimulated (black bars) β cells of C3H/He (left) and ashen (right) mice is shown as a function of the distance from granule center to the plasma membrane (nm). Data are represented as a percentage of the granule density in each concentric shell below the plasma membrane relative to the average density in cytoplasm (100% = number of total granules per the area of cytoplasm, that is, the cell area minus the nuclear area). Values are mean ± SE (n = 10). The statistical significance of differences between means was assessed by Student’s t test. *P < 0.05.
Complex formation between granuphilin and syntaxin 1a. We previously demonstrated that the Rab27a effector granuphilin directly interacts with syntaxin 1a, a SNARE protein on the plasma membrane (11), and that wild-type granuphilin, but not its mutants that are defective in binding to either Rab27a or syntaxin 1a, promotes the plasma membrane targeting of insulin granules (12). These findings suggest that granuphilin tethers insulin granules and the plasma membrane through interactions with Rab27a and syntaxin 1a, respectively. To examine the complex formation between granuphilin and syntaxin 1a, coimmunoprecipitation experiments were performed with lysates of isolated islets. Although the expression levels of syntaxin 1a and granuphilin were similar between ashen and control islets, the amount of syntaxin 1a coprecipitated with anti-granuphilin antibodies was significantly decreased in ashen islets (Figure 7A, lower panel). The reduction was specific, because the amount of another complex containing syntaxin 1a with Munc18-1 (25) was comparable between ashen and control islets (Figure 7A, upper panel). This finding is compatible with our previous finding that active Rab27a promotes complex formation between granuphilin and syntaxin 1a (11) and may explain the decreased docking of insulin granules with the plasma membrane in ashen β cells. Quantification in independent experiments revealed that the amount of granuphilin/syntaxin 1a complex was reduced to approximately 60% compared with the control (Figure 7B); this value is roughly equivalent to the degree of decrease in the number of docked granules (Figure 5A). The decreased complex formation between syntaxin 1a and granuphilin also indicated that the redistribution of granuphilin to the peripheral plasma membrane region in ashen β cells (Figure 1B) is not due to the increased association with syntaxin 1a.
Protein interactions of syntaxin 1a in pancreatic islets. (A) Protein interactions of syntaxin 1a with either Munc18-1 (upper) or granuphilin-a (lower) were analyzed in extracts of C3H/He or ashen islets. The amount of Munc18-1, granuphilin-a, and syntaxin 1a in each lysate (5 μg, left) and 30–40% of the immunoprecipitates of Munc18-1 or granuphilin-a (right) were examined by immunoblotting with the antibodies indicated. (B) Results of coimmunoprecipitation experiments independently performed as described in A were gathered for statistics. Relative intensities of syntaxin 1a signals in ashen mice (black bars) to those in C3H/He mice (white bars) were calculated and expressed as mean ± SE from 7 (vs. Munc18-1) and 6 (vs. granuphilin-a) immunoblot preparations. When analyzed by a Wilcoxon signed-ranks test, ashen islets showed significantly reduced interaction of syntaxin 1a with granuphilin-a (*P = 0.027), but not with Munc18-1.