WASH Regulates Glucose Homeostasis by Facilitating Glut2 Receptor Recycling in Pancreatic β-Cells

Diabetes is a term used to describe a metabolic disorder of multiple etiology characterized by high blood glucose levels resulting from insulin secretion defects (type 1 diabetes), insulin action failure (type 2 diabetes), or both (1–5). Insulin release involves a sequence of well-controlled events in β-cells that start from environmental stimulations (sensing) and end with releasing of secretory granules containing insulin (action). Glucose is known to be the strongest stimulator for insulin release in pancreatic β-cells (6,7). There are 14 facilitative diffusion glucose transporters (Glut) encoded by the solute carrier family 2 (SLC2A) genes (8). Glut2 is well established as the principal membrane Glut with low affinity in rodent pancreatic β-cells (9,10), and previous studies using a transgenic mouse model showed that Glut2-null mice generated by homologous recombination provoked severe glycosuria and died at around the weaning period with a diabetic phenotype (11). Importantly, pancreatic-specific expression of Glut2 in Glut2-null mice restored normal glucose-stimulated insulin secretion (GSIS) and glucose-stimulated insulin biosynthesis (12). In addition, Glut2 protein levels in pancreatic islets are strongly reduced with loss of GSIS in numerous animal models of diabetes (13–17). Although the mechanism for Glut2 protein expression (18), posttranscriptional modification (19), in vitro trafficking (20), and in vivo subcellular translocation (21) has been identified, the in vivo regulation of Glut2 in pancreatic islets is still unclear.

WASH (Wiskott-Aldrich syndrome protein and SCAR homolog) (22) is a member of the Wiskott-Aldrich syndrome protein (WASP) family that promotes branched F-actin generation through activation of the Arp2/3 complex (23). WASH forms a multiprotein complex with FAM21, SWIP, strumpellin, and CCDC53 that is targeted to endosomes through an interaction of FAM21 with VPS35, a component of the endosomal coat complex known as the retromer (24–26). Several in vitro studies have demonstrated an important role for WASH in the recycling of plasma membrane receptors through the endosomal system in a manner dependent on the generation of branched F-actin by the Arp2/3 complex. These include, for example, receptors such as integrins, growth factor receptors, lipid transporters, and solute carriers (27–31). The mechanism by which this diverse cadre of receptors is trafficked into WASH-dependent sorting domains depends on their interaction with sorting nexin 27 (SNX27), which binds to the cytoplasmic tails of receptors via its postsynaptic density 95/discs large/zonus occludens-1 (PDZ) domain and directly couples to the retromer subunit VPS26 and the WASH complex member FAM21 (31) or via an interaction of SNX17-receptor complexes with the retriever (32).

We recently showed that patients with mutations in CCDC22 fail to appropriately traffic LDLR and ATP7A resulting in substantially elevated levels of serum cholesterol/LDL and copper, respectively. Significantly, patients with mutations in strumpellin were also found to have high levels of circulating cholesterol and LDL (29). Thus, defective trafficking of receptors through WASH endosomal sorting domains can have a physiological impact beyond the intellectual disability associated with mutations in CCDC22 and strumpellin. Using our previously described WASH conditional knockout (cKO) mice (30), we asked whether WASH might be involved in pancreas development or function using a pancreas-specific Cre mouse model. Interestingly, WASH deletion did not affect body weight, fasting blood glucose, or pancreas tissue development compared with wild-type (WT) animals. However, WASH cKO mice showed decreased insulin release and delayed glucose clearance. Significantly, total and plasma membrane Glut2 levels were significantly reduced in cKO compared with WT mice leading to diminished glucose uptake. Taken together, these results identify that WASH plays an important and unique physiological role in pancreatic β-cell glucose sensing and insulin secretion through trafficking of Glut2.

Antibodies to human WASH, mouse WASH, and FAM21 have previously been described (24–26). Antibody to insulin was obtained from Cell Signaling Technologies (Beverly, MA); antibody to β-actin and GFP were from Sigma-Aldrich (St. Louis, MO); antibody to Glut2 was from Proteintech Group (Rosemont, IL) and Abcam (Cambridge, MA); antibody to glucagon-like peptide 1 receptor (GLP-1R) was from Developmental Studies Hybridoma Bank (University of Iowa, Iowa), Proteintech Group, and Santa Cruz Biotechnology (Dallas, TX); antibody to Glut1 was from Abcam and Cell Signaling Technologies; and antibody to Lamp1 (CD107a) was from BD Pharmingen (San Jose, CA). For immunohistochemical and immunofluorescence staining, the following primary antibodies were used: rabbit anti-human WASH (1:500), rabbit anti-mouse WASH (1:500), mouse anti-insulin (1:300), mouse anti–chromogranin A (1:200), rabbit anti-Glut2 (1:100), rat anti-mouse Lamp1 (1:100), rabbit anti–GLP-1R (1:50), and rabbit anti-Glut1 (1:100). For immunofluorescence staining, Alexa Fluor 488 donkey anti-mouse IgG, Alexa Fluor 568 donkey anti-rabbit IgG (Life Technologies), and Alexa Fluor 633 donkey anti-rat IgG (Jackson ImmunoResearch Laboratories) were used as secondary antibodies at a 1:300 dilution.

WASH^flox/flox and Pdx1-Cre mice have previously been described (28,33). Pancreas-specific WASH cKO mice were generated by crossing WASH^flox/flox mice with Pdx1-Cre mice to produce Pdx1-Cre;WASH^flox/flox animals. These animals were crossed with WASH^flox/flox mice. Unless otherwise indicated, WASH^flox/flox mice are classified as WT mice and Pdx1-Cre;WASH^flox/flox mice are classified as cKO. Control experiments were performed using littermate WASH^flox/flox WT animals. Mice were housed in a 12 h–12 h light-dark cycle barrier facility. All procedures were approved by the Mayo Clinic Institutional Animal Care and Use Committee.

Islet isolation was performed following an established protocol (34). Briefly, islets were isolated by intraductal collagenase (Sigma-Aldrich) perfusion and digestion. Islets were handpicked using dithizone (Sigma-Aldrich) detection of zinc granules. After isolation, islets were placed in RPMI plus 10% FBS and cultured at 37°C and 5% CO₂ for future experiments.

Oral and intraperitoneal glucose tolerance tests (OGTTs and ipGTTs) were performed on mice, which had fasted for 12 h (8:00 p.m. to 8:00 a.m.). Blood glucose levels were measured at 0, 15, 30, 60, 90, and 120 min after oral or intraperitoneal administration of glucose (2 g/kg body wt). Blood samples from the tail vein were collected simultaneously in the presence of aprotinin (2 μg/mL) and EDTA (1 mg/mL). Serum was harvested and stored at −70°C. For the insulin tolerance test, mice were fasted for 4 h (8:00 a.m. to 12:00 a.m.) and injected with 1 IU/kg body wt human crystalline insulin (Eli Lilly, Indianapolis, IN). Blood glucose levels were determined by use of a Glucometer (Bayer Contour) with blood collected from the tail vein. Levels of plasma insulin were measured using an ELISA (cat. no. EZRMI-13K for insulin; Millipore). For measurement of pancreatic insulin content, the pancreas tail was isolated, homogenized in acid alcohol, and extracted overnight at −20°C. The solution was centrifuged to remove debris and neutralized and insulin content was determined by ELISA.

All the chemicals were obtained from Sigma-Aldrich unless otherwise specified. INS-1 cells were a gift from Dr. Weizhen Zhang (University of Michigan, Ann Arbor, MI) and cultured in RPMI-1640 containing 11.1 mmol/L glucose and supplemented with 10% FBS, 1 mmol/L pyruvate, 10 mmol/L HEPES, 50 μmol/L 2-mercaptoethanol, 100 units penicillin/mL, and 100 g streptomycin/mL. WASH^−/− mouse embryonic fibroblasts (WASHout MEFs) have previously been described (28). Stealth siRNAs (sequences provided in Supplementary Table 1) were purchased from Thermo Fisher Scientific and transfected with Lipofectamine RNAiMAX reagent (Thermo Fisher Scientific) according to the manufacturer’s protocol. A set of two siRNAs was used per target. Glut2-GFP plasmid was a gift from Dr. Jeffrey E. Pessin (Albert Einstein College of Medicine, Bronx, New York) and transfected using Lipofactamine 2000 according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA). For 2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-2-deoxyglucose (2-NBDG) uptake, the INS-1 cells transfected with control siRNA (siControl) or target siRNAs were resuspended in glucose-free KRBH buffer for 30 min and then supplemented with 30 μmol/L concentration of the glucose analog 2-NBDG (Thermo Fisher Scientific) for 1, 2.5, and 5 min and washed and run on a FACSCanto II flow cytometer (BD Bioscience), and the data were analyzed using FlowJo (TreeStar). The mean fluorescence intensity was defined as the geometric mean of the given fluorescent probe.

INS-1 cells or isolated islets were preincubated in KRBH buffer: 136 mmol/L NaCl, 4.7 mmol/L potassium chloride (KCl), 1.2 mmol/L KH₂PO₄, 1.2 mmol/L MgSO₄, 5 mmol/L NaHCO₃, 1 mmol/L CaCl₂, 10 mmol/L HEPES, and 0.5% BSA, pH 7.4) containing 3 mmol/L glucose at 37°C for 1 h. After aspiration of the buffer, INS-1 cells or isolated islets were incubated in fresh KRBH buffer supplemented with 3 mmol/L glucose, 25 mmol/L glucose, 20 mmol/L KCL, or 10 nmol/L glucagon-like peptide 1 (GLP-1) (Sigma-Aldrich), at 37°C for 30 min. The supernatant was collected and centrifuged for later use. The unreleased insulin within the INS-1 cells or isolated islets was extracted by the acid alcohol method. The insulin collected from the supernatant medium and remaining islets were first diluted and then measured by ELISA for calculating the percentage of insulin secretion.

Mice were anesthetized using isoflurane (Piramal Critical Care, Bethlehem, PA), followed by cervical dislocation. The whole pancreas was quickly removed and fixed overnight in 4% PFA with gentle shaking, embedded in paraffin, and cut into 5-μm-thick sections. Human and mouse pancreas sections were subjected to hematoxylin-eosin staining, immunohistochemistry, or immunofluorescence staining as previously described (35). This reference encompasses the Hoechst staining as well. Confocal images were collected with an LSM-710 laser scanning confocal microscope with a 63× water Plan-Apochromat objective lens using ZEN 2009 software (Carl Zeiss, Oberkochen, Germany). Islet cell size and number were assessed by hematoxylin-eosin staining of pancreas tissue section at 100-µm intervals, and β-cell mass was assessed by immunohistochemistry staining of insulin. Area of insulin-positive cells were measured and normalized to total pancreatic area using ImageJ (National Institutes of Health, Bethesda, MD). β-Cell mass is expressed in grams after normalization to total pancreas mass.

Snap frozen pancreata or isolated islets from mice of desired genotypes were homogenized or lysed in radioimmunoprecipitation assay buffer (Abcam) or islet lysis buffer (36). Protein extracts were prepared, separated by SDS-PAGE, transferred to polyvinylidene fluoride membrane, and immunoblotted as previously described (35). Protein bands of interest were quantified by calculating an integrated density value for each band using ImageJ. Pancreatic total RNA was isolated using Trizol and further purified with an RNeasy Mini Kit (Qiagen, Valencia, CA). Reverse transcription was performed with the Superscript III RT-PCR Kit (Invitrogen). Quantitative PCR was performed with the SYBR Green PCR Master Mix using the ABI StepOnePlus Sequence Detection System (Applied Biosystems, Carlsbad, CA). Gapdh and Rplp0 were used as housekeeping genes for normalization of gene expression. The double Δ Ct method was used to analyze gene expression. Experiments were performed in triplicate using three independent cDNAs. Primer sequences are provided in Supplementary Table 1.

Data are expressed as mean ± SEM and analyzed by repeated-measures ANOVA and unpaired Student t test using GraphPad Prism software (GraphPad Software, La Jolla, CA). A value of P <0.05 denotes statistical significance.

Pancreatic tissue sections from humans or mice were stained for WASH protein expression. WASH expression is weakly cytoplasmic in the human and mouse exocrine pancreas but is highly concentrated in the islets (Fig. 1A). For further confirmation of this expression pattern of WASH, double immunofluorescent staining was performed to colocalize the signal of WASH with the endocrine cell marker chromogranin A (CgA) in the pancreas of humans and mice. As shown in Fig. 1B, WASH showed strong immunoreactivity in both human and mouse islets that also stained positive for CgA. These data indicate that WASH is expressed in the human and mouse pancreas but significantly accumulated in islets of Langerhans.

To delineate the physiological role of WASH in the pancreas, we generated pancreas-specific WASH cKO mice by crossing Pdx1-cre mice with WASH^flox/flox mice. Analysis of lysates from the pancreas of WT and cKO animals showed a substantial reduction of WASH protein, as well as the WASH complex member FAM21 (Fig. 2A and B). Consistent with the immunoblotting results, WASH staining was greatly reduced in CgA-positive endocrine cells in cKO mice compared with WT mice (Fig. 2C). Thus, we have generated pancreas-specific WASH knockout mice, allowing in vivo study of the physiological role of WASH in the pancreas.

Given the essential role of pancreas in body weight and glucose homeostasis regulation, we first examined the body weight and fasting blood glucose of 8- and 16-week-old WT and cKO mice. As shown in Fig. 3A and B, there was no difference in total body weight or fasting glucose between WT and cKO mice. Since there was no alteration of pancreas weight in cKO mice (data not shown), we next evaluated the expression of several mRNAs that are enriched in the pancreas including α-amylase, insulin, glucagon (Gcg), somatostatin, and cytokeratin-19 (CK19). As shown in Fig. 3C, deletion of WASH did not affect the expression level or ratio of the tested genes. WASH deletion also had no effect on islet size (Fig. 3D and E), β-cell mass, average islet size or islet number (Supplementary Fig. 1). Finally, we extracted pancreatic insulin and observed no significant alteration of insulin content between WT and cKO animals (Fig. 3F). Taken together, these data suggest that deletion of WASH during pancreas development does not overtly affect pancreas development or impact body weight.

As there is no apparent developmental deficiency in islets from cKO mice, we next tested pancreas function by performing an OGTT, ipGTT, and insulin tolerance test. Interestingly, we found glucose was cleared more slowly in cKO mice relative to WT mice in both OGTT and ipGTT (Fig. 4A–D). This observation indicates a significant impairment in glucose tolerance in WASH cKO mice. On the other hand, there was virtually no difference in insulin sensitivity in peripheral tissues between WT and cKO mice (Fig. 4E and F). In addition, serum insulin release during the glucose load remained significantly lower in cKO mice than in WT mice (Fig. 4G–J). Thus, defective insulin release in response to glucose loading in cKO mice may explain, at least in part, glucose intolerance in this model.

The widely expressed Glut isoform Glut2 is thought to have highest expression among all other Gluts in rodent pancreatic islet β-cells and is required for GSIS and glucose homeostasis (10,11,37,38). Previous studies in T cells showed that T cells lacking WASH had reduced Glut1 endosome-to-membrane recycling (30). We therefore investigated whether the impaired OGTT, ipGTT, and insulin responses to glucose are due to an alteration of Glut2 expression in cKO mice. Interestingly, total Glut2 protein levels were lower in islets obtained from cKO mice compared with WT mice (Fig. 5A and B). In addition, we found that the mRNA expression of Glut2 was also substantially reduced in cKO mice relative to WT mice (Fig. 5C). We further found, using immunohistochemistry and immunofluorescence staining, that a large proportion of Glut2 is expressed on the surface of insulin-positive β-cells within WT islets, whereas Glut2 levels were lower in cKO mice and less Glut2 was detected on the surface of the insulin-positive β-cells (Fig. 5D and E). Moreover, consistent with serum insulin level changes during OGTT in vivo, we observed a significant reduction of GSIS in islets of cKO mice relative to WT mice ex vivo (Fig. 5F). Finally, as insulin is also secreted in response to different secretagogues such as KCl and hormonal factors (e.g., glucagon, GLP-1) and the magnitude of this response is modulated by a variety of receptors and transporters (6), we next examined the contribution of WASH loss in pancreas to the mRNA, protein expression level, or membrane localization of several Gluts and GLP-1R. To this end, we found that only Glut2 mRNA was significantly decreased (Supplementary Fig. 2A). Additionally, only Glut2 protein levels were down in the WASH knockout islets, whereas Glut1 and GLP-1R were unaffected (Supplementary Fig. 2B–E). Furthermore, islets from WT and cKO mice secreted similar levels of insulin upon GLP-1 stimulation in the presence of either KCl or 3 mmol/L glucose (Supplementary Fig. 2F). Taken together, these data indicate that WASH deletion decreases overall Glut2 protein levels and appears to alter GSIS in islets as a result of reduced plasma membrane localization of Glut2, whereas other Gluts and GLP-1R were unaffected in either expression or function.

To further examine the role of WASH in Glut2 trafficking and its physiological function as part of glucose sensing in vitro, we used RNA interference–mediated suppression of WASH and Glut2 in INS-1 cells. Consistent with our in vivo cKO data, knockdown of WASH in INS-1 cells reduced Glut2 protein levels similar to those observed in Glut2 siRNA (siGlut2)-treated cells (Fig. 6A and B), and IF showed that WASH was dramatically reduced in insulin-positive INS-1 cells (Fig. 6C). Consistent with the notion that decreased Glut2 total or plasma membrane protein levels in WASH cKO mice might be the reason for decreased glucose sensitivity and insulin release in vivo, siWASH-treated INS-1 cells had a significant decrease in glucose uptake and subsequent GSIS comparable with siGlut2-treated cells (Fig. 6D and E). Moreover, transfection of a Glut2-GFP plasmid into WASH-depleted cells showed a substantial loss of Glut2-GFP in the plasma membrane (Fig. 6F and G). These findings further support the idea that WASH plays an important role in glucose homeostasis through endosome-to-membrane trafficking of Glut2.

We have previously shown that T cells derived from WASH knockout mice accumulated Glut1 in lysosomes due to defective endosome–to–plasma membrane recycling (30). Since our results above showed that depletion of WASH resulted in reduced cell surface and total levels of Glut2 (Fig. 6A and F), we next investigated whether Glut2 is degraded and accumulates in lysosomes in the absence of WASH. To test this, we used INS-1 cells treated with siControl or siWASH and transfected them with a Glut2-GFP expression vector. The cells were subsequently treated with cycloheximide for 2 or 4 h to block new protein synthesis, and Glut2 levels were measured by immunoblotting for GFP. Although Glut2-GFP levels started lower in the siWASH-treated cells, Glut2-GFP exhibited faster degradation kinetics than control cells (Fig. 7A and B). To determine whether Glut2 accumulated in lysosomes, we transfected Glut2-GFP into WT and WASHout MEFs (Fig. 7C). Significantly, while Glut2-GFP accumulated in the plasma membrane of WT MEFs, we found an increased incidence of Glut2-GFP colocalization with LAMP1 in WASH knockout MEFs (Fig. 7D and E). Taken together, these data indicate that WASH is critical for the appropriate trafficking of Glut2 and that in the absence of WASH, Glut2 is aberrantly trafficked to the lysosome and degraded.

Prior cell line studies have linked WASH to the recycling of several cell surface receptors through its interaction with the CCC retriever and retromer complexes, which direct the trafficking of SNX17- and SNX27-dependent cargo, respectively (23,39). Herein we show that WASH is enriched in the islets of Langerhans where it plays an important role in pancreatic β-cell regulation of Glut2 trafficking and whole-body glucose homeostasis. Mechanistically we found that WASH spares Glut2 from lysosomal degradation and promotes its endosome–to–plasma membrane recycling. In the absence of WASH, Glut2 becomes associated with the lysosome and is degraded following internalization leading to decreased GSIS and ultimately defective glucose homeostasis.

Obesity and type 2 diabetes have reached epidemic proportions in Western societies (40). The identification and characterization of the genes involved in type 2 diabetes have added to our understanding of the pathways regulating β-cell function. Recently, genome-wide association studies and biochemical studies have identified a member of the vacuolar protein sorting-10 (VPS10) family of receptors, SORCS1, and the retromer component VPS26 as risk factors for type 2 diabetes (41–43). Interestingly, both SNX27 and retromer were found to maintain the total and cell surface levels of Glut1 in HeLa cells through the interaction of the SNX27 PDZ binding domain with VPS26 and the Glut1 PDZ binding motif found at its COOH terminus. Significantly, loss of either SNX27 or the retromer subunit VPS35 results in Glut1 endosomal trapping and lysosomal degradation (31). In addition, using WASH knockout T cells we previously demonstrated that WASH regulates Glut1 protein levels, glucose uptake, and T-cell proliferation (30). We now provide evidence that Glut2 preferentially accumulates intracellularly in the absence of WASH in several model systems. In addition, we show that in the absence of WASH, Glut2 protein levels are reduced and that an exogenously expressed Glut2-GFP construct is more quickly degraded following the inhibition of protein synthesis. Moreover, the Glut2-GFP construct becomes localized with Lamp1⁺ vesicles within the cytoplasm of MEFs lacking WASH suggesting that the increased degradation is a result of defective endosome–to–plasma membrane retrieval Glut2. It remains to be determined whether Glut2 contains a COOH-terminal PDZ binding motif like Glut1, but our data would be consistent with a similar mechanism of endosomal retrieval of Glut2 that involves WASH, retromer, and sorting nexin 27.

While Glut2 is no doubt the most abundant Glut in rodent islets (10–12), its role in human β-cell development and function seems more complicated (10,44). In mice lacking Glut2, there are major defects in GSIS and mouse pups die within 2–3 weeks owing to severe diabetes. In human beings, some mutations in SLC2A2 encoding Glut2 are responsible for the Fanconi-Bickel syndrome, an autosomal recessive disorder associated with defective carbohydrate metabolism (45). In contrast to mice, patients with this rare disorder are not reported to require insulin treatment but, rather, display severe glycosuria due to deficient glucose reabsorption by the kidney (46). This discrepancy between humans and rodents is related to Glut2 tissue expression. Glut2 is the predominant Glut in rodent islets, while Glut1 and Glut3 are more prevalent in human β-cells (10,47). Surprisingly, we note that only Glut2 mRNA and protein expression was decreased in islets from WASH cKO mice, while other Gluts and GLP-1R remained unchanged. As a result, we detected impaired insulin secretion upon high glucose stimulation but not GLP-1, which is consistent with a previous study using WASH RNA interference (48). Although the WASH knockout islets do not lose their insulin secretion response to KCl, the contribution of WASH to the trafficking of K_ATP channel Kir6.2 needs further investigation. One possible explanation for the reduction of Glut2 mRNA is the feedback response of disrupted GSIS in WASH cKO mice, since it is known that impaired GSIS is associated with markedly reduced expression of Glut2 (14,15). Lastly, Glut1 and Glut3 mRNA expression in the mouse pancreas is barely detectable compared with Glut2 (10); thus, it is not surprising that Glut1 and Glut3 mRNA as well as Glut1 protein expression levels are not affected in islets from WASH deletion mice. Whether Glut1 and Glut3 expression, localization, or function is similarly affected in patients carrying mutations that impact WASH activity is of interest, as they might have defective GSIS and be more prone to develop glucose intolerance.

In summary, our study provides evidence that WASH is expressed in human and mouse pancreas and highly concentrated in pancreatic endocrine islets. In addition, WASH contributes to glucose uptake and insulin release through its critical role in the trafficking of the membrane Glut Glut2 both in vivo and in vitro. Interesting, although mutations in the genes encoding the CCC complex member CCDC22 and strumpellin are associated with intellectual disability syndrome, we have previously demonstrated that these patients also have undiagnosed pathologies including elevated levels of circulating cholesterol and LDL, which could make them prone to heart disease. Thus, patients carrying mutations in the WASH complex components strumpellin or SWIP, or the CCC complex member CCDC22, might be at a higher risk for developing diabetes.

Funding. This work was supported by the Mayo Foundation for Medical Education and Research. D.D.B. was supported by National Institutes of Health grant R01DK107733.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. L.D., L.H., and J.D. contributed to the study concept and design; data acquisition, analysis, and interpretation; drafting the manuscript; and critical revision of the manuscript for important intellectual content. D.D.B. contributed to the study concept and design; data acquisition, analysis, and interpretation; drafting the manuscript; critical revision of the manuscript for important intellectual content; technical support; and study supervision. D.D.B. 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.