SNAP23 depletion enables more SNAP25/calcium channel excitosome formation to increase insulin exocytosis in type 2 diabetes

Abstract

SNAP23 is the ubiquitous SNAP25 isoform that mediates secretion in non-neuronal cells, similar to SNAP25 in neurons. However, some secretory cells like pancreatic islet β cells contain an abundance of both SNAP25 and SNAP23, where SNAP23 is believed to play a redundant role to SNAP25. We show that SNAP23, when depleted in mouse β cells in vivo and human β cells (normal and type 2 diabetes [T2D] patients) in vitro, paradoxically increased biphasic glucose-stimulated insulin secretion corresponding to increased exocytosis of predocked and newcomer insulin granules. Such effects on T2D Goto-Kakizaki rats improved glucose homeostasis that was superior to conventional treatment with sulfonylurea glybenclamide. SNAP23, although fusion competent in slower secretory cells, in the context of β cells acts as a weak partial fusion agonist or inhibitory SNARE. Here, SNAP23 depletion promotes SNAP25 to bind calcium channels more quickly and longer where granule fusion occurs to increase exocytosis efficiency. β Cell SNAP23 antagonism is a strategy to treat diabetes.

Keywords: Beta cells; Diabetes; Endocrinology; Insulin; Metabolism.

Figure 1. Generation of a mouse with β cell–specific deletion of SNAP23.

(A) Whole islets from SNAP23^fl/fl mice (24) show that SNAP23 is abundant in β cells (confocal imaging), shown more clearly in the enlarged box 1. Scale bars: 100 μm. Quantitation can be found in Supplemental Figure 1A, bottom left. (B) Single islet β cells from human (first row) and mouse (second row) show that SNAP23 is abundant in the insulin SGs. SNAP23 is partly colocalized with Stx-1A (third row) and SNAP25 (fourth row) on the PM. Scale bars: 5 μm. Quantitation can be found in Supplemental Figure 1A, bottom middle. SNAP23 antibody controls (without primary antibody) are shown in Supplemental Figure 1A, top left, and A, top right. (C) SNAP23 is also present in glucagon-containing α cells located in the periphery of the mouse islet. Scale bars: 10 μm. Quantitation in Supplemental Figure 1A, bottom right. (D) AAV8-RIP1-Cre drives Cre expression only in β cells (top) and not in α cells (bottom). Scale bars: 100 μm. Note the cytosolic insulin surrounding the nuclear Cre shown clearly in the enlarged box 2. (E) Efficient knockdown of islet β cell SNAP23 expression is shown in (top) where Cre-positive cells are SNAP23-negative, indicating SNAP23 deletion in those cells, and (bottom) the majority of insulin-positive cells are SNAP23-negative. The few SNAP23-positive cells are likely α cells. Scale bars: 100 μm. (F) Western blots of islets from SNAP23^fl/fl mice injected with the AAV8 (βSNAP23-KO) or not (Control) showed reduction of SNAP23 but not other exocytotic proteins. SNAP23 and VAMP8 are not abundant in mouse brain. Blots are representative from 3 independent experiments; analysis of n = 3 in Supplemental Figure 1B. (G) βSNAP23-KO versus SNAP23^fl/fl (Control) mice show SNAP23 to be reduced only in islets but not in fat, muscle, or liver, wherein SNAP23 and cognate Munc18c and Stx-4 are putative exocytotic proteins. Shown are representative of 3 independent experiments; analysis of n = 3 in Supplemental Figure 1C.

Figure 2. β Cell–specific deletion of SNAP23 in mice improves glucose homeostasis resulting from increased GSIS from pancreatic islets.

(A) IPGTT performed in the same SNAP23^fl/fl mice before AAV8-RIP1-Cre treatment and at 2 weeks after virus treatment showed improved glucose homeostasis (left) resulting from increased insulin secretion (right). Right panels show the respective AUC analyses of n = 11. (B) IPITT performed on the same mice before and after AAV8-RIP1-Cre treatment showed no significant changes in blood glucose levels, which indicates no effect on insulin sensitivity. Right panel shows the AUC analysis of n = 11. (C) SNAP23-KO did not affect body weight. n = 11. (D) Insulin-immunostained pancreatic sections (scale bars: 1000 μm), from which we conducted morphometric analysis (E), including β cell area per pancreatic area ratios, islet number per total pancreatic area, and islet size (n = 12 for each group). *P < 0.05; ***P < 0.001. Statistical significance was assessed by 2-tailed Student’s t test.

Figure 3. β Cell exocytotic events that account for the increased GSIS in βSNAP23-KO mice.

(A) Islet perifusion assays showing that βSNAP23-KO mouse islets exhibited enhanced first- and second-phase GSIS compared with SNAP23^fl/fl (Control) mouse islets; AUC analysis shown in middle. Left: Total islet insulin content was not affected in the βSNAP23-KO mice islets (n = 5 for each group). (B) Patch-clamp Cm on dispersed single β cells of βSNAP23-KO versus Control mice. Top shows representative recordings of exocytosis evoked by a train of ten 500-ms depolarizations from –70 mV to 0 mV. Bottom left: Cumulative changes in cell capacitance normalized to basal cell membrane capacitance (fF/pF) in Control and βSNAP23-KO β cells, shown as (bottom right) summary graph (Control: n = 15 cells, βSNAP23-KO: n = 13 cells). (C) TIRF microscopy imaging of exocytosis of predock and newcomer SGs. Left: Histograms of different fusion events in first phase (first 6 minutes after 16.7 mM glucose stimulation) and second phase (6–14 minutes) in Control (top) versus βSNAP23-KO β cells (bottom). Data obtained from 3 independent experiments (5–6 cells from each experiment; WT = 15 cells, βSNAP23-KO = 16 cells). Right top: Summary of the 3 modes of fusion events in first (top) and second phases (bottom). This is the sum of the different types of SG fusion events occurring during the first phase (all frames assessed in the first 6 minutes) and second phase (6–14 minutes) normalized to the PM area. Right bottom: We also obtained TIRF microscopy images of docked insulin SGs before the stimulation (basal) above, and found no change in SG density (averaged number of SGs normalized to cell PM area on each TIRF microscopy imaging frame before stimulation) between Control versus βSNAP23-KO β cells (summary graph on right). Scale bars: 2 μm. *P < 0.05; **P < 0.01. Statistical significance was assessed by 2-tailed Student’s t test.

Figure 4. SNAP23 depletion increases GSIS in normal human islets.

(A) Ad-SNAP23 shRNA/mCherry treatment (48 hours) of human islets (see Supplemental Table 1 for details on human donors) effectively depleted SNAP23 (SNAP23-KD) as shown on confocal imaging (left) and islet protein levels (right). Scale bars: 100 μm. Left: SNAP23 is abundant in human islet β cells (top) and not affected by Ad-scrambled shRNA/mCherry (middle). SNAP23 was severely depleted by Ad-SNAP23 shRNA/mCherry treatment (bottom), wherein the few SNAP23-positive cells are mCherry-negative, allowing calculation of the ratio of SNAP23-positive area per islet area (right graph, n = 19 islets from each group). Right: Western blots of islets show reduction in SNAP23 protein levels by Ad-SNAP23 shRNA/mCherry but no effects on the other exocytotic proteins. Shown are representative of 3 experiments, with analysis of n = 3 in Supplemental Figure 4A. (B) SNAP23-KD of human islets increased first- and second-phase GSIS compared with Ad-scrambled shRNA/mCherry–treated (Control) islets. AUC analysis in bottom from n = 4 for each group, from 4 independent experiments. (C) These human islets were then dispersed into single cells for patch-clamp Cm measurements performed on mCherry-positive β cells, which showed that SNAP23-KD β cells exhibited increased exocytosis. Top: Representative recordings. Bottom left: Cumulative changes in cell capacitance normalized to basal cell membrane capacitance (fF/pF) in Control versus SNAP23-KD (n = 15 cells) β cells, and shown as (bottom right) summary graph (Control: n = 14 cells, SNAP23-KD: n = 15 cells). *P < 0.05; **P < 0.01; ***P < 0.001. Statistical significance was assessed by 2-tailed Student’s t test.

Figure 5. SNAP23 depletion increases GSIS in T2D human islets.

(A) Western blots of T2D human islets (see Supplemental Table 1 for T2D patients’ donor information) show reduction of Stx-1A and SNAP25 as previously reported (24), but with normal levels of SNAP23 and several other syntaxins. Shown are representative of 3 experiments with analysis of n = 3 in Supplemental Figure 4B. (B) Ad-SNAP23 shRNA/mCherry treatment to deplete SNAP23 in T2D human islets increased first- and second-phase GSIS compared with Ad-scrambled shRNA/mCherry (Control)-treated T2D islets. n = 4 for each group, from 4 independent experiments. *P < 0.05. Statistical significance was assessed by 2-tailed Student’s t test.

Figure 6. SNAP23 overexpression inhibits GSIS in normal human islets and β cells.

(A) Ad-SNAP23/mCherry treatment (48 hours) of human islets effectively overexpressed SNAP23 as shown on Western blots with no effects on other exocytotic proteins. Shown are representative of 3 experiments with analysis of the n = 3 in Supplemental Figure 4C. (B) Islet perifusion assays showing that SNAP23 overexpression in human islets decreased first- and second-phase GSIS. AUC analysis of n = 3 from 3 independent experiments shown in bottom. (C) TIRF microscopy imaging of exocytosis of predock and newcomer SGs. Top: Histograms of different fusion events in first-phase (first 6 minutes after 16.7 mM glucose stimulation) and second-phase (6–14 minutes) in Control versus SNAP23 overexpressing β cells. Data obtained from 3 independent experiments (5–6 cells from each experiment; control = 14 cells, SNAP23 overexpression = 15 cells). Bottom: Summary of the 3 modes of fusion events in first and second phases. *P < 0.05. Statistical significance was assessed by 2-tailed Student’s t test.

Figure 7. Pancreatic SNAP23 depletion in type-2 diabetic GK rats rescues insulin secretory deficiency and restores glucose homeostasis.

Ad-SNAP23 shRNA/mCherry (6.6 × 10⁹ PFU) versus Ad-mCherry (same dose) was infused via pancreatic duct into GK rats. IPGTTs (blood glucose and insulin levels obtained) were performed after infusion at 1 (A), 2 (B), 4 (C), and 8 weeks (D). Glybenclamide treatment of age- and sex-matched GK rats as conventional treatment for T2D was also performed; IPGTT was then performed at the same time points. Graphs on the right show AUCs encompassing 180 minutes of the IPGTTs. Glybenclamide group: n = 6 and Ad-SNAP23 shRNA/mCherry group: n = 6 for 1, 2, and 4 weeks after infusion, and n = 5 for 8 weeks after infusion. Ad-mCherry control groups: n = 6 for 1 and 2 weeks after infusion, and n = 5 for 4 and 8 weeks after infusion. We also performed vehicle control in GK rats compared with glybenclamide treatment, and conducted IPGTTs as shown in Supplemental Figure 6. *P < 0.05; **P < 0.01; ***P < 0.001. Statistical significance was assessed by repeated-measures ANOVA.

Figure 8. SNAP23 acts as inhibitory SNARE that reduces SNAP25 binding to Stx-1A and Stx-3 from forming SNARE complexes, whereby these SNAP25 but not SNAP23 SNARE complexes could form more fusion-competent machineries with Ca_v, priming proteins and Ca²⁺ sensors.

Coimmunoprecipitation (Co-IP) studies were performed on INS-832/13 cells before stimulation (basal) and after maximal stimulation with 16.7 mM glucose plus 10 nM GLP-1. (A) SNAP23 forms similar SM/SNARE complexes as those previously reported with SNAP25. Co-IP was performed with SNAP23 antibody. Shown are representative of 3 experiments with analysis of the 3 experiments shown in Supplemental Figure 7A. (B) SNAP23 depletion in INS-832/13 cells enables endogenous SNAP25 to promote formation of more SM/SNARE complexes to mediate the increased exocytosis of predocked SGs (Munc18a, Stx-1A, VAMP2) and newcomer SGs (Munc18b, Stx-3, VAMP8) that underlie the increased first- and second-phase GSIS shown in Supplemental Figures 2–4. INS-832/13 cells were first treated with Ad-SNAP23 shRNA/mCherry to deplete SNAP23 versus Ad-scrambled shRNA/mCherry (Control). Co-IP was performed with SNAP25 antibody. Shown are representative of 3 experiments, with analysis of the 3 experiments shown in Supplemental Figure 7B. (C) SNAP23 blocks SNAP25 (and vice versa) binding to Stx-1A and Stx-3. HEK293 cells transfected with Stx-1A, Stx-3, SNAP23, or SNAP25 were subjected to pulldown with GST-SNAP23 (in the presence or absence of expressed SNAP25) or GST-SNAP25 (in the presence or absence of expressed SNAP23) to assess how much Stx-1A (top) or Stx-3 (bottom) could be precipitated. GST was used as control. Shown are representative of 4 experiments, with analysis of the 4 experiments shown in Supplemental Figure 7C. SNAP23 SNARE complexes (D) are less able than SNAP25 complexes (E) to pull down Ca_vs (Ca_v1.2, -1.3, and -2.3) synaptotagmins (Syt-1 and -7) and priming proteins (RIM2, Munc13-1). Note that 16.7 mM glucose plus GLP-1 increased the binding of some of these proteins in SNAP25 but not SNAP23 complexes. This is a representative of 3 separate experiments with analysis shown in Supplemental Figure 8. (F) Control (Ad-scrambled shRNA/mCherry) or SNAP23-knockdown (Ad-SNAP23 shRNA/mCherry) human β cells were stimulated with 16.7 mM glucose plus 10 nM GLP-1, immunolabeled with SNAP25 and Stx-1A or Stx-3 antibodies, and then assessed by structured illumination microscopy imaging. SNAP23 depletion increased SNAP25 colocalization with Stx-1A (top; Pearson’s correlation coefficient [PCC], Control: –0.42 ± 0.09; KD: 0.36 ± 0.04, n = 11 from 2 independent experiments) and Stx-3 (bottom; PCC, Control: –0.74 ± 0.06: KD: 0.14 ± 0.02, n = 12 from 2 independent experiments). Scale bars: 2 μm. ***P < 0.001. Statistical significance was assessed by 2-tailed Student’s t test.

Figure 9. Live-cell single-molecule localization and tracking of SNAP25-mScarlet binding to endogenous Ca_v1.2. (A) SNAP23-knockout (SNAP23-KO) INS-832/13 cells.

(B) SNAP23-overexpressing (OE) INS-832/13 cells. Representative live-cell single-molecule image of SNAP25-mScarlet and Ca_v-mNeonGreen in SNAP23-KO (A, top left) and SNAP23-OE (B, top left) INS-832/13 cells; these images are representative of 4 independent experiments. (A, top left) Endogenous Ca_v1.2 in SNAP23-KO cells (mTagBFP2) was labeled with mNeonGreen by an MMEJ-mediated knockin method as previously described (54) and shown in detail in Supplemental Figure 9 and described briefly in the Methods. Right: The trajectories of SNAP25-mScarlet were overlaid onto the locations of the endogenous Ca_v1.2 clusters (indicated in red circles). (A, top right) Representative intensity traces of single SNAP25-mScarlet molecules binding to Ca_v1.2 clusters (from top to bottom, white circles 1–3 indicated in A, top left). (A, bottom left) Examples of sequential images of the mobility of single SNAP25-mScarlet coming from the cytoplasm (top: corresponds to white circle indicated as event 4 in A, top left) or along the PM (bottom: corresponds to white circle indicated as event 1 in A, top left) traveling to bind a Ca_v1.2 cluster; the latter trajectory shown in (A, bottom right). Blue rectangle indicates the start, red circle indicates the location of Ca_v1.2. (B, top left) Ca_v1.2-knockin INS-832/13 cells were transfected with SNAP23-SNAPf for 24 hours and then incubated with far-red fluorescent substrate, 647-SiR, for imaging. SNAP25-mScarlet was transfected again for 3 to 6 hours before imaging. Trajectories of SNAP25-mScarlet were overlaid onto the locations of endogenous Ca_v (indicated as red circles). (B, top right) Representative intensity traces (from top to bottom, white circles 1–3 indicated in B, top left). (B, bottom left) Two examples of sequential images showing the mobility of single SNAP25-mScarlet molecules coming from the cytoplasm (top: corresponds to white circle indicated as event 2 in B, top left) or along the PM (bottom: corresponds to white circle indicated as event 4 in B, top left) traveling to bind Ca_v1.2 clusters; the latter trajectory shown in B, bottom right. (C) Comparison between SNAP23-KO versus WT versus SNAP23-OE. Left: Cumulative dwell time of SNAP25 binding to Ca_v1.2 per Ca_v1.2 clusters. Middle: Cumulative frequency of SNAP25 binding to Ca_v1.2 per Ca_v1.2 clusters. Right: Travel time histogram of single events of SNAP25 binding to Ca_v1.2 (n = 10 each from 4 independent experiments). *P < 0.05. Statistical significance was assessed by repeated-measures ANOVA.

Figure 10. Individual insulin SG exocytosis at Ca_v1.2 clusters.

(A) Control INS-832/13 cells. (C) SNAP23-knockout cells. Data are representative of 3 independent experiments. Scale bars: 2 μm. (B) Representative montage of SG exocytosis at Ca_v1.2 clusters. (D) Cumulative exocytosis events per Ca_v1.2 cluster along the different time points during stimulation. (E) Average exocytosis events per Ca_v1.2 cluster per minute (12 cells each from 3 independent experiments). *P < 0.05. Statistical significance was assessed by 2-tailed Student’s t test.