Inhibition of SNAT5 Induces Incretin-Responsive State From Incretin-Unresponsive State in Pancreatic β-Cells: Study of β-Cell Spheroid Clusters as a Model

The biology of pancreatic β-cells, including insulin secretion and cell signaling, is often studied in monolayer cultured cells (MCs), that is, in two-dimensional (2D) architecture, using various β-cell lines such as RINm5F, HIT, βTC, INS-1, and MIN6. However, native β-cells operate as a 3-dimensional (3D) structure in the islets of the pancreas. In addition, pancreatic islets comprise heterogenous populations of endocrine cells, including insulin-secreting β-cells, glucagon-secreting α-cells, and somatostatin-secreting δ-cells. Although MCs of clonal β-cells may show poor insulin secretory responses to various stimuli compared with native pancreatic islets, it is well established that formation of islet-like 3D structure (spheroid clusters [SCs]) by aggregation of the cells can restore β-cell morphology and insulin secretory properties (1–4). This demonstrates that homologous interactions among β-cells are necessary for normal regulation of insulin secretion. Several other studies have emphasized the importance of cell-cell interaction in maintaining normal insulin secretory potential in rodent and human β-cells (5–8).

Although glucose is physiologically the most important regulator of insulin secretion, neuronal and hormonal inputs to β-cells also are critical for maintaining blood glucose levels within the physiological range (9–13). The incretins glucagon-like peptide 1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP), which are secreted from enteroendocrine L cells and K cells, respectively, in response to meal ingestion, play critical roles in preventing postprandial hyperglycemia by potentiating insulin secretion through activation of cAMP signaling in β-cells (14,15). Using the advantage of the glucose concentration–dependent effect of incretins on insulin secretion, incretin-based antidiabetic drugs, such as dipeptidyl peptidase (DPP)-4 inhibitors and GLP-1 receptor agonists, are currently used for treatment of type 2 diabetes (16–18).

Using a metabolomics-based approach with incretin-responsive and -unresponsive MIN6 β-cell lines, designated MIN6-K8 and -K20, respectively (19), we recently found that glutamate produced by glucose in β-cells acts as a key cell signal in incretin-induced insulin secretion (IIIS) (20).Two important mechanisms are involved in this process: 1) cytosolic glutamate production by glucose stimulation through the malate-aspartate (MA) shuttle linked to glycolysis in β-cells and 2) transport of cytosolic glutamate into insulin granules by cAMP signaling. β-Cell glutamate thus links glucose metabolism to cAMP action in insulin secretion (20). In the course of our study of MIN6-K20, we found that aggregation of incretin-unresponsive K20-MC into 3D architecture (i.e., formation of K20-SC) drastically induced IIIS (19). K20-SC therefore provides a useful model for studying how an incretin-responsive state is induced from an incretin-unresponsive state.

In the current study, we aimed to elucidate the mechanisms by which incretin responsiveness is acquired in incretin-unresponsive β-cells. We compared insulin secretory properties, cAMP signaling, metabolomic profiling, and gene expression profiling between K20-MC and K20-SC. We also examined the pathophysiological role of one of the sodium-coupled neutral amino acid transporters (SNATs), SNAT5, a candidate molecule in the mechanism of induction of incretin responsiveness using animal models of obesity and type 2 diabetes.

All animal experiments were approved by the Kobe University Committee on Animal Experimentation, performed in accordance with the Guidelines for Animal Experimentation at Kobe University, and conformed to the Guide for the Care and Use of Laboratory Animals (Eighth Edition, 2011; http://grants.nih.gov/grants/olaw/guide-for-the-care-and-use-of-laboratory-animals.pdf). Male C57BL/6J and ob/ob (C57BL/6JHamSlc-ob/ob) mice (8 weeks old) were purchased from Japan SLC (Hamamatsu, Japan), and KK (KK/TaJcl) and KK-Ay (KK-Ay/TaJcl) mice were from CLEA Japan (Tokyo, Japan).

[U-¹³C]glucose and l-glutamic acid γ-monohydroxamate (GluγHA) were purchased from Sigma-Aldrich (St. Louis, MO), 8-bromo-cAMP-acetoxymethyl ester (8-Br-cAMP-AM) was purchased from BIOLOG Life Science Institute (Bremen, Germany), and anti-CREB antibody and anti-phosphorylated CREB antibody were purchased from Cell Signaling Technology (Danvers, MA).

Cell culture was performed as described previously (21) using DMEM (25 mmol/L glucose and 4 mmol/L glutamine; Sigma-Aldrich). Functional experiments were performed on these cells at 80% of confluency.

Cells were seeded into ultra-low-attachment 10-cm dishes (Corning, Corning, NY) at a density of 1.5 × 10⁶ cells/dish and incubated for 7 days to form SCs. All of the studies using MCs and SCs were performed in parallel using nearly the same passage between 20 and 28.

Insulin secretion experiments were performed as described previously (21).

Transmission electron microscopy was performed as described previously (22).

Immunostaining of β-cell lines was performed as described previously (21,22) using anti-SNAT5 antibody (Abcam, Cambridge, U.K.), anti-insulin antibody (DAKO, Agilent, Santa Clara, CA), and anti-Ki67 antibody (DAKO). Immunostaining of mouse pancreata was performed by the polymer-based detection method using anti-SNAT5 antibody (Santa Cruz Biotechnology, Dallas, TX) and N-Histofine Simple Stain Mouse MAX PO (Nichirei Biosciences, Tokyo, Japan) according to the manufacturers’ instructions.

The ATP content was determined by an EnzyLight ATP assay kit (BioAssay Systems, Hayward, CA).

Measurement of cAMP content was performed as described previously (19). To prevent degradation of cAMP, all experiments were performed in the presence of 100 μmol/L 3-isobutyl-1-methylxanthine.

Quantitative RT-PCR was performed as described previously (21).

RNA sequencing was performed on 1 μg each of total RNA using an Illumina HiSeq 2500 system by Eurofins Genomics (Ebersberg, Germany). Sequence reads were cleaned using Trimmomatic (ver. 0.32), and then aligned to the mouse genome (GRCm38) using Bowtie 2 (ver. 2.2.2) and TopHat (ver. 2.0.11). Raw read counts and fragments per kilobase of transcript per million mapped reads were generated using Cufflinks (ver. 2.2.1). Transcripts were merged using Cuffmerge (ver. 2.2.1), and differential gene expression was determined using Cuffdiff (ver. 2.2.1). The RNA sequencing data have been deposited in the DDBJ Sequence Read Archive (DRA) with the accession number of DRA006332.

K20-MC and K20-SC were preincubated for 60 min in HEPES–Krebs-Ringer buffer with 2.8 mmol/L glucose (without glutamine), and then incubated for 30 min in HEPES–Krebs-Ringer buffer with 16.7 mmol/L glucose (without glutamine). Metabolome analysis was performed as described previously (20).

Measurement of phosphorylation of CREB was performed as described previously (23).

Knockdown experiments using siGENOME siRNA (Dharmacon, Lafayette, CO) were performed as previously described (20). Small hairpin RNA (shRNA) vectors were constructed using the BLOCK-iT adenoviral RNAi expression system (Invitrogen) according to the manufacturer’s instruction. The target sequences were designated as follows: nontarget shRNA top strand, 5′-CACCGTAAGGCTATGAAGAGATACCGAAGTATCTCTTCATAGCCTTA-3′, and bottom strand, 5′-AAAATAAGGCTATGAAGAGATACTTCGGTATCTCTTCATAGCCTTAC-3′; and Slc38a5 shRNA top strand, 5′-CACCGGGAGCTATGTTCATCATGTCGAAACATGATGAACATAGCTCC-3′, and bottom strand, 5′-AAAAGGAGCTATGTTCATCATGTTTCGACATGATGAACATAGCTCCC-3′. Isolated mouse islets were infected with shRNA adenovirus at a multiplicity of infection of 50. After 2 days of culture, insulin secretion experiments were performed.

The synthesized cDNA of mouse Slc38a5 was purchased from Eurofins Genomics. Recombinant adenovirus carrying Slc38a5 tagged with DYKDDDDK (Slc38a5-DYK) was generated using the ViraPower adenoviral expression system (Invitrogen) according to the manufacturer’s instruction. MIN6-K8 cells and isolated mouse islets were infected with adenovirus carrying Slc38a5-DYK or mCherry at a multiplicity of infection of 50. After 2 days of culture, insulin secretion experiments were performed.

Western blotting was performed as described previously (21) using anti-SNAT5 antibody (Abcam) and anti-actin antibody (formerly Calbiochem, Merck KGaA, Darmstadt, Germany).

Pancreatic islets were isolated from mice (10–12 weeks of age) as described previously (24).

The results are presented as mean ± SEM. Differences among the groups were analyzed with the Dunnett method or Welch t test, as indicated in the figure legends. P < 0.05 was regarded as statistically significant.

We previously reported that SCs formed from incretin-unresponsive MIN6-K20 cells using gelatin-coated dishes exhibited incretin responsiveness in insulin secretion with a size and shape similar to those of native mouse pancreatic islets (19). In this study, we used ultra-low-attachment dishes to form SCs and confirmed that formation of SCs endows incretin-unresponsive MIN6-K20 cells with incretin responsiveness (Fig. 1A). On one hand, glucose induced insulin secretion in a concentration-dependent manner in K20-MC and K20-SC (Fig. 1B and C). On the other hand, neither GLP-1 nor GIP potentiated insulin secretion in K20-MC, whereas GLP-1 and GIP both potentiated insulin secretion in a glucose concentration–dependent manner in K20-SC (Fig. 1B and C). GLP-1 potentiated insulin secretion at ≥0.1 nmol/L, whereas GIP did so at ≥1 nmol/L (Fig. 1D and E). In MIN6-K8 cells, which already exhibit IIIS in MCs to some extent, incretin responsiveness is further enhanced by formation of SCs (Supplementary Fig. 1).

The size and shape of K20-SC were similar to those of native mouse pancreatic islets (Fig. 2A). Electron microscopic observation revealed that the number of insulin granules was increased, and the mitochondria were well developed in K20-SC (Fig. 2B). K20-SC showed a lower proliferative state than that of K20-MC, as assessed by Ki67 staining (Supplementary Fig. 2). Insulin content was significantly increased in K20-SC to a level comparable to that in native mouse islets (Fig. 2C), and ATP content was also markedly increased in K20-SC (Fig. 2D). These data indicate that incretin-unresponsive MIN6-K20 cells acquire morphological and functional properties similar to those of native pancreatic islets when they form SCs.

We then attempted to clarify whether induction of incretin responsiveness in insulin secretion from SCs could be due to enhanced incretin/cAMP signaling. We first examined expressions of incretin receptors in K20-SC by quantitative PCR. There was no difference in the expression level of GLP-1 receptor between K20-MC and K20-SC. The expression of GIP receptor tended to be increased in K20-SC, but not significantly, compared with that in K20-MC (Fig. 3A). We found that expressions of the adenylyl cyclases Adcy1 and Adcy7 were upregulated and Adcy5 and Adcy6 were downregulated (Fig. 3B), whereas those of the phosphodiesterases Pde1c, Pde3b, Pde4b, and Pde10a were downregulated in K20-SC (Fig. 3C), suggesting decreased hydrolysis of cAMP. In fact, cAMP content in K20-SC was markedly increased by GLP-1 treatment compared with that in K20-MC (Fig. 3D).

To determine whether signaling downstream of cAMP is involved in the induction of incretin responsiveness in K20-SC, we examined the effect of 8-Br-cAMP-AM, a membrane permeable cAMP analog, on phosphorylation of CREB, a target of protein kinase A (PKA), as well as insulin secretion. We found 8-Br-cAMP-AM markedly potentiated insulin secretion in K20-SC compared with K20-MC (Fig. 3E) and also found a significant increase in CREB phosphorylation by 8-Br-cAMP-AM in K20-SC (Fig. 3F). These results suggest that signaling downstream of cAMP also is enhanced in K20-SC.

As described above, the mitochondria have become well developed in K20-SC, suggesting enhancement of cellular metabolism. We therefore performed metabolome analysis on K20-MC and K20-SC under glucose (16.7 mmol/L)-stimulated condition (Supplementary Table 1). The score plot of principal component analysis revealed that samples of K20-MC and K20-SC were separated into distinct groups (Fig. 4A), suggesting a difference in the metabolomic profile between K20-MC and K20-SC. The loading plot showed that various metabolites contribute to the difference (Fig. 4B). In K20-SC, there were significant increases in metabolites in glycolysis (G6P, F6P, FBP, and pyruvate) and the tricarboxylic acid cycle (fumarate, malate, and succinate), amino acids (proline, aspartate, and glutamate), and nucleotides (AMP, ADP, ATP, GTP, and cAMP) (Fig. 4C). These data indicate that cellular metabolism is generally enhanced in K20-SC.

We recently found that glutamate produced through the MA shuttle is critical for IIIS (20) and therefore investigated the role of intracellular glutamate in the induction of incretin responsiveness in K20-SC. When cells were treated with [U-¹³C]glucose, a stable isotope-labeled glucose, not only M and M + 1 glutamate isotopomers (no or one ¹³C substitutions for ¹²C), which were naturally existing, but also M + 2 to M + 5 glutamate isotopomers (two to five ¹³C substitutions for ¹²C), which were derived from [U-¹³C]glucose, were increased significantly in K20-SC (Fig. 5A). The latter glutamate isotopomers are produced through the MA shuttle (20). To clarify the role of the MA shuttle in IIIS and cAMP production directly, we used aspartate aminotransferase 1 (AST1, gene symbol Got1), a critical enzyme in the shuttle, knockout (KO) clonal β-cells, which we recently established (21). GLP-1– and 8-Br-cAMP-AM–induced insulin secretions were significantly reduced in Got1-KO-SC compared with those in K8-SC (Fig. 5B and C). There was no difference in cAMP production by glucose or GLP-1 between K8-SC and Got1-KO-SC (Fig. 5D), indicating that inhibition of the MA shuttle does not affect cAMP production. These data strongly suggest that an increase in cellular glutamate production through the MA shuttle is essential for incretin responsiveness in K20-SC.

To further investigate the mechanism by which incretin responsiveness is induced in K20-SC, we performed RNA-sequencing analyses of K20-MC and K20-SC. Monolayer cultured MIN6-K8 (K8-MC), K20-MC, K20-SC, and mouse pancreatic islets exhibited distinct gene expression profiles (Supplementary Fig. 3). However, expression of the genes involved in glucose-induced insulin secretion (GIIS) was similar in K20-MC and K20-SC (Supplementary Fig. 4), except for Slc2a2 (GLUT2), the expression of which was slightly higher in K20-SC, suggesting a possible enhancement of glucose use by K20-SC. Because the insulin secretory response to incretins becomes stronger in the order of pancreatic islets, K20-SC, K8-MC, and K20-MC (Fig. 1A) (19), we focused on the genes with expression levels that are correlated or inversely correlated with the incretin responsiveness. We detected five genes with expression levels that were relatively high and inversely correlated with incretin responsiveness (Supplementary Fig. 5A). By knockdown experiments on each of these five genes, a significant enhancement of IIIS was found only when Slc38a5 was knocked down (Supplementary Fig. 5B). We confirmed by TaqMan assay that the expression level of Slc38a5 was highest in K20-MC, followed by K8-MC, followed by K20-SC, and lowest in mouse pancreatic islets, showing a rigorous, inverse correlation between Slc38a5 (SNAT5) expression level and incretin responsiveness (Fig. 6A). SNATs play an important role in glutamine influx/efflux, thereby regulating the cellular glutamate concentration in exocytotic cells such as neurons (25–28). No other SNAT genes exhibited correlation between expression levels and incretin responsiveness (Supplementary Fig. 6A). In addition, knockdown of Slc38a2 (SNAT2) or Slc38a4 (SNAT4), the expression levels of which were relatively high in K20-MC, by siRNA in K20-MC did not endow the cells with incretin responsiveness (Supplementary Fig. 6B). Immunostaining of K8-MC and K20-MC cells revealed SNAT5 (Slc38a5) was expressed in β-cells (Supplementary Fig. 7A and B). We therefore explored the possible involvement of SNAT5 in the induction of incretin responsiveness. GluγHA, a SNAT5 inhibitor (29), restored IIIS in K20-MC (Fig. 6B), with no significant effect on ATP or cAMP content (Supplementary Fig. 8A and B). In addition, siRNA knockdown of Slc38a5 in K20-MC induced incretin responsiveness (Fig. 6C and Supplementary Fig. 8C), with no significant effect on gene expression of insulin or incretin receptors (Supplementary Fig. 8D). We also examined the effects of overexpression of Slc38a5 on incretin responsiveness in K8-MC and found that it did not affect GIIS but decreased IIIS (Fig. 6D). Metabolic flux analysis using [U-¹³C]glucose showed that the cellular glutamate content in K20-MC treated with GluγHA was elevated significantly compared with that without treatment, apparently due to an increase in the glutamate M isotopomer level (Fig. 6E), indicating that cellular glutamate content was increased mainly from a source other than glucose. Because GluγHA may inhibit glutamine efflux from cells through SNAT5 under the condition of glutamine-free buffer, cellular glutamate content is increased probably due to conversion from accumulated glutamine, thereby contributing to induction of incretin responsiveness.

To clarify the pathophysiological relevance of SNAT5, we examined mRNA expression levels of Slc38a5 in pancreatic islets of mouse models of obesity and type 2 diabetes and also evaluated the effect of SNAT5 inhibition on insulin secretion from these islets. Slc38a5 expression was increased more than threefold in ob/ob mice, a model of obesity, compared with control wild-type mice (Fig. 7A). Treatment of the islets isolated from ob/ob mice with GluγHA did not affect GIIS but enhanced IIIS (Fig. 7B). Similarly, Slc38a5 expression was increased more than fivefold in KK-Ay mice, a model of obese type 2 diabetes, compared with control KK mice (Fig. 7C). Immunostaining confirmed the higher expression of SNAT5 in the islets of KK-Ay mice and also clarified that although SNAT5 is predominantly present at the plasma membrane in exocrine cells of the pancreas, it is diffusely present in islet cells (Supplementary Fig. 7C and D). Treatment of the islets isolated from KK-Ay mice with GluγHA did not affect GIIS but enhanced IIIS (Fig. 7D). Interestingly, there was no significant enhancement of IIIS from GluγHA-treated islets of wild-type or KK mice (Supplementary Fig. 9). Similar to the findings obtained with K8-MC and K20-MC, overexpression of Slc38a5 impaired incretin responsiveness in control B6 mouse islets (Fig. 7E), and knockdown of Slc38a5 improved the responsiveness in KK-Ay mouse islets (Fig. 7F). These findings indicate that upregulation of SNAT5 in pancreatic β-cells is associated with impaired IIIS in obesity and type 2 diabetes.

In the current study, we aimed to elucidate the mechanisms of induction of incretin responsiveness from incretin unresponsiveness in pancreatic β-cells by using SCs as a model system. Pancreatic islets exist anatomically as a 3D structure. The normal secretory function of β-cells in response to a variety of stimuli depends on this architecture, which allows for modulation of intercellular communication through direct cell-cell contacts via gap, tight, or adherent junctions and paracrine mechanisms (1,30,31). SCs constituted from MIN6-K20 cells (K20-SC) are similar to native islets in size and structure and have well-developed mitochondria and a greater number of insulin granules. However, these structures are made of nonprimary cells, are composed of insulin-secreting cells only, and are neither vascularized nor innervated. By metabolome analysis, we noticed that most metabolites of glycolysis and the tricarboxylic acid cycle, amino acids, and nucleotides were upregulated in K20-SC. These metabolites provide various metabolic signals that are important for triggering and potentiating insulin secretion (13,32–34).

Because IIIS is mediated by cAMP signaling (35–37), we first focused on how cAMP signaling is changed by formation of SCs. No significant difference in the expression of the GIP receptor or the GLP-1 receptor between K20-MC and K20-SC was observed in our data. Downstream of these receptors in the cAMP signaling pathway, adenylyl cyclases (Adcys) and phosphodiesterases (Pdes) are involved in the production and hydrolysis of cAMP, respectively. There are nine isoforms of transmembrane Adcys with different regulatory properties (38,39). Most of them are expressed in pancreatic islets and β-cell lines (40–42). Adcy1 and Adcy7 were significantly upregulated, whereas Adcy5 and Adcy6 were significantly downregulated in K20-SC. The enhanced cellular metabolism in K20-SC may well affect the activity of Adcy, because recent data indicate that β-cell Adcy activity can be directly regulated by cellular metabolism, probably via ATP (43). Especially, Adcy1 and Adcy8, both of which are thought to be important for GIIS in β-cells, are stimulated by the intracellular Ca²⁺ level (41,44,45), whereas Adcy5 and Adcy6 are inhibited by PKA-mediated phosphorylation (46,47). The Pdes important for the regulation of cAMP levels in β-cells, Pde1c, Pde3b, Pde4b, and Pde10a, were downregulated in K20-SC. Pharmacological or genetic inhibition of these Pdes has been shown to potentiate insulin secretion (48–51). Thus, upregulation of Adcys and downregulation of Pdes found in K20-SC is likely to contribute to the elevated cAMP levels, which then enhance the activity of PKA, its downstream target. These findings together clearly suggest that cAMP signaling is enhanced by formation of SCs.

Our findings that glutamate production after glucose stimulation is significantly enhanced in K20-SC and that genetic disruption of the MA shuttle diminishes IIIS without impairment of cAMP production indicate the central role of glutamate signaling in the induction of incretin responsiveness. RNA sequencing analysis revealed that of all the SNATs, only the expression of Slc38a5 (SNAT5) was correlated with incretin responsiveness; that is, inversely in pancreatic islets, K20-SC, K8-MC, and K20-MC, in that order. We therefore focused on SNAT5 and its role in IIIS. Interestingly, inhibition of SNAT5 induced incretin responsiveness in K20-MC, which was accompanied by an increase in cellular glutamate content. Although there is no direct evidence for the direction of glutamine transport through SNAT5 in β-cells at present, our findings indicate that SNAT5 may function as an exporter of glutamine under the condition in which extracellular glutamine is absent or very low. Furthermore, SNAT5 expression was upregulated in mouse models of obesity and diabetes, both of which exhibited impaired IIIS, and inhibition of SNAT5 improved IIIS in islets isolated from these mice, probably by increasing cellular glutamate in β-cells. To our knowledge, there are no findings on the relationship between SNAT5 and insulin secretion in β-cells. Misiewicz et al (52) reported in 2013 that Slc38a5 has an endoplasmic reticulum (ER) stress element–like element, CCAAT-N26-CCACG, in the promoter and that induction of ER stress increased the expression in cultured primary human neurons, suggesting that transcription of Slc38a5 in β-cells is also regulated by ER stress.

SNATs are classified as system A (SNAT1, SNAT2, and SNAT4) or system N (SNAT3 and SNAT5) (25–28). In neurons, SNAT1-5 are important in regulating cellular glutamate content by controlling influx or efflux of glutamine, the immediate precursor of glutamate (53). System A SNATs are mainly involved in influx of glutamine, and system N SNATs are mainly involved in efflux of glutamine (28,54). In tumor cells, glutamine transport into cells through SNAT5 is suitable because of the provision of glutamine for the tumor cell–specific metabolic pathways and the intracellular alkalization (55). It is noteworthy that Slc38a5 is a transcriptional target for the oncogene c-Myc (56), suggesting that some tumor cells may upregulate this transporter as a part of their tumor-promoting gene expression program. Studies have recently reported that inhibition of glucagon signaling in the liver leads to α-cell proliferation and increased glucagon secretion, which is mediated by increased glutamine levels in the circulation and upregulation of SNAT5 in α-cells (57,58), suggesting that SNAT5 functions as an importer of glutamine in α-cells. Efflux or influx of glutamine through SNAT5 may depend on glutamine concentration gradients between the inside and outside of cells. The role of SNAT5 in the interaction between α-cells and β-cells in normal and pathological conditions, such as obesity and diabetes, remains to be elucidated.

More recently, Cfap126 (also known as Fltp) has been reported to be increased when endocrine cells aggregate into 3D architecture; the 3D structure and Wnt ligands are sufficient to trigger β-cell maturation (59). Despite being highly expressed in mouse islets, we found no significant difference in the expression of Cfap126 between K20-MC and K20-SC (Supplementary Fig. 10). Interestingly, expression of Wnt5a was significantly decreased, whereas that of Wnt5b was increased in K20-SC compared with that in K20-MC. Especially, expression levels of Wnt5b were positively correlated with the incretin responsiveness. A strong association of one SNP in the WNT5B gene with type 2 diabetes has previously been reported (60), but the mechanism involved remains unknown.

Taken together, our present study of β-cell SCs indicates that activation of cAMP signaling and an increase in cellular glutamate production participates in the induction of incretin responsiveness from incretin-unresponsive β-cells. The importance of cell-cell interaction in incretin responsiveness was further confirmed by the experiments on dispersed islets in terms of profiles of gene expression and insulin secretion (Supplementary Fig. 11). The present findings also support our model of the mechanism of IIIS, in which cellular glutamate production by glucose and activation of cAMP signaling, which promotes glutamate transport into insulin granules, are both required for the induction of incretin responsiveness. Because inhibition of SNAT5 improves IIIS in obese mice and obese diabetic mice and impairment in β-cell glutamate production is linked with impaired IIIS (20,61), clarification of the mechanism of SNAT5 regulation may provide an attractive avenue for better treatment of diabetes as well as further understanding of incretin unresponsiveness in diabetes and obesity.

Acknowledgments. The authors thank Yasuhiro Minami (Kobe University) for his support and encouragement throughout this study; Ritsuko Hoshikawa, Chihiro Seki, Akiko Kanagawa, Niina Ota, Ayako Kawabata, and Haruyo Maeda (Kobe University) for their excellent technical assistance in this study; and Christopher M. Carmean (Kobe University) for his valuable comments and suggestions.

Funding. This study was supported by grants-in-aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan, MSD K.K., and the Kobayashi Foundation. M.H. was supported by a scholarship from Honjo International Scholarship Foundation and by a student fellowship from Novo Nordisk A/S.

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

Author Contributions. M.H., N.Y., H.T., G.G., O.S.O., T.H., N.M., S.H., K.M., A.M., and S.S. reviewed the manuscript and participated in discussion of the results. M.H., N.Y., G.G., O.S.O., T.H., N.M., S.H., and A.M. performed experiments. M.H., N.Y., and S.S. designed the study, analyzed data, discussed the data, and wrote the manuscript. H.T. and K.M. contributed to analysis and discussion of the data and edited the manuscript. S.S. conceived the project. S.S. 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.

Prior Presentation. Parts of this study were presented in abstract form at the 52nd Annual Meeting of the European Association for the Study of Diabetes, Munich, Germany, 12–16 September 2016, and at the 53rd Annual Meeting of the European Association for the Study of Diabetes, Lisbon, Portugal, 11–15 September 2017.