Studies implicating sodium–glucose cotransporter 2 (SGLT2) inhibitors in glucagon secretion by pancreatic α-cells reported controversial results. We hypothesized that interindividual heterogeneity in SGLT2 expression and regulation may affect glucagon secretion by human α-cells in response to SGLT2 inhibitors. An unbiased RNA-sequencing analysis of 207 donors revealed an unprecedented level of heterogeneity of SLC5A2 expression. To determine heterogeneity of SGLT2 expression at the protein level, the anti-SGLT2 antibody was first rigorously evaluated for specificity, followed by Western blot and immunofluorescence analysis on islets from 10 and 12 donors, respectively. The results revealed a high interdonor variability of SGLT2 protein expression. Quantitative analysis of 665 human islets showed a significant SGLT2 protein colocalization with glucagon but not with insulin or somatostatin. Moreover, glucagon secretion by islets from 31 donors at low glucose (1 mmol/L) was also heterogeneous and correlated with dapagliflozin-induced glucagon secretion at 6 mmol/L glucose. Intriguingly, islets from three donors did not secrete glucagon in response to either 1 mmol/L glucose or dapagliflozin, indicating a functional impairment of the islets of these donors to glucose sensing and SGLT2 inhibition. Collectively, these data suggest that heterogeneous expression of SGLT2 protein and variability in glucagon secretory responses contribute to interindividual differences in response to SGLT2 inhibitors.

Sodium–glucose cotransporter 2 (SGLT2) inhibitors are a recent class of oral antihyperglycemic agents (1). Initially developed to induce glycosuria as a treatment for type 2 diabetes, SGLT2 inhibitors have since been shown to improve metabolic and cardiac activities as well. Additional positive benefits of SGLT2 inhibition include reductions in body weight, arterial blood pressure, heart insufficiency, and the delay in the progression of kidney diseases (24). In addition to their use in type 2 diabetes, SGLT2 inhibitors have also shown some efficacy in obese subjects with and without type 1 diabetes (57). However, data from eight recent clinical trials reported that SGLT2 inhibitor therapy as adjunctive therapy to insulin in patients with type 1 diabetes increased the risk of diabetic ketoacidosis (8,9). SGLT2 inhibitors stimulate glucagon secretion and endogenous glucose production (10,11), at least in part, related to SGLT2 expression in human α-cells (12,13). Indeed, both selective SGLT2 inhibition by siRNA silencing and dapagliflozin treatment were found to directly trigger glucagon secretion in both human and rodent islet cultures (12,14,15). However, other reports have failed to observe such effects for unknown reasons (1618). Besides differences in experimental conditions, a potential explanation may be an interindividual heterogeneity in SGLT2 expression and regulation of glucagon secretion. To address this latter hypothesis, we used an unbiased approach by searching the Translational Human Pancreatic Islet Genotype Tissue-Expression Resource (TIGER) to explore the expression pattern of SLC5A2 mRNA (encoding SGLT2) in 207 donors. To determine SGLT2 heterogeneity at the protein level, we applied a set of well-known validation criteria based on the principles of antibody action and established with a reasonable degree of assurance that the SGLT2 antibody from Novus Biologicals targets its correct antigen (19). Subsequently, an extensive evaluation of SGLT2 localization in 665 human islets was performed using the Bitplane IMARIS software (Bitplane AG, Zürich, Switzerland). Finally, using static incubation experiments on 31 human islet preparations, we determined the interindividual SGLT2 heterogeneity and the variability in glucagon secretory responses to glucose and dapagliflozin treatment. We conclude that SGLT2 expression and function in α-cells are characterized by considerable interindividual variability, which may contribute to individual differences in clinical responses to SGLT2 inhibitors in patients with diabetes.

Whole-Islet RNA-Sequencing Analysis of the TIGER Database

The RNA expression raw data from 207 human islet preparations were obtained from TIGER (available online at http://tiger.bsc.es). The corresponding online public data of the Oxford (https://www.ebi.ac.uk/ega/datasets/EGAD00001001601) and Lund (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE50244) (20,21) cohorts were used in the analysis. The expression matrix of selected genes (Supplementary Table 1) was quantified using salmon software, version 12.0 (https://combine-lab.github.io/salmon/), as previously reported by Patro et al. (22).

Tissue Processing and Sectioning

Human pancreatic tissues were harvested from brain-dead adult donors in the context of the traceability requirements for our clinical islet transplantation program (ClinicalTrials.gov, NCT01123187, NCT00446264, and NCT01148680). Prior to islet isolation, the pancreas was preserved at 4°C for <24 h, in the University of Wisconsin solution, which contains no glucose. Pancreatic tissue was immediately fixed in 4% (w/v) paraformaldehyde and embedded in paraffin. Sections (10 µm) were cut on a Leica RM 2165 microtome (Leica Biosystems, Copenhagen, Denmark). The experimental design and protocols used in this study were approved in agreement with French regulations and our Institutional Ethical Committee of the University of Lille and CHU Lille (Lille, France). As requested by recent guidelines (23), an overview of the clinical characteristics of human islet donors used in this study is summarized in the Human Islet Checklist (Supplementary Table 2).

Immunostaining Techniques

Human pancreatic slices from 12 donors were deparaffinized following a standard protocol. Heat-mediated antigen retrieval was performed under specific conditions for each primary antibody (Supplementary Table 3). Human kidney HEK293 cells were fixed on coverslips with 4% paraformaldehyde at room temperature for 1 h. Blocking was performed using Protein Block Serum-free (catalog number X0909; Dako) for 20 min at room temperature. The primary antibodies were diluted in PBS (Supplementary Table 3) and incubated overnight at 4°C. Single- or double-immunofluorescence staining was performed using anti-SGLT2 antibody (Novus Biologicals) with or without its specific blocking peptide (Novus Biologicals) and costained with anti-insulin (Dako), anti-glucagon (GeneTex), anti-somatostatin (GeneTex), or an additional anti-somatostatin antibody (Millipore). Antibody and blocking peptide information are summarized in Supplementary Table 3, rows 1–6. Single- or double-immunofluorescence staining was performed using an additional anti-SGLT2 antibody (Santa Cruz Biotechnology) costained with anti-insulin (Abcam), anti-glucagon (Abcam), or anti-somatostatin (Dako). Antibody information is summarized in Supplementary Table 3, rows 6–9. Secondary Alexa Fluor antibodies (anti-rabbit, AF-594, catalog number 11012, Invitrogen; anti-rabbit, AF-488, catalog number 21206, Invitrogen; anti-mouse, AF-488, catalog number 11001, Invitrogen; anti-mouse, AF-594, catalog number 11032, Invitrogen; anti-rat, AF-594, catalog number 11007, Invitrogen; and anti–guinea pig, AF-488, catalog number ab150185, Abcam) were diluted 1/800 and applied for 1 h at room temperature. The nuclei were counterstained with DAPI shield and mounted with Dako Fluorescence Mounting Medium (catalog number 53023). For immunohistochemistry staining, human pancreatic slices were incubated overnight with the anti-SGLT2 antibody (Abcam) or anti-insulin antibody (Dako) with or without its specific SGLT2-blocking peptide (Abcam) at 4°C. Staining was performed with 1:1,000 dilution of 3,3′-diaminobenzidine from Dako (catalog number K3467) for 2 s, followed by nuclei staining with hematoxylin. Slices were mounted in Immu-Mount (catalog number 9990402; Thermo Fisher Scientific, Illkirch, France). Details of antibodies used in the immunhistochemistry analysis are presented in Supplementary Table 3.

Image Acquisition, Analysis, and Quantification

Images were acquired using the confocal Zeiss LSM 710 microscope with Airyscan superresolution module or the Cell Observer Spinning Disk (Zeiss, Jena, Germany). The images were obtained using a ×40 objective with immersion oil and 1.4 of the numerical aperture. For qualitative assays, the images were processed and adjusted using ImageJ, version 2.0.0-rc-43/1.50e (https://imagej.nih.gov/ij). For quantitative colocalization studies, the voxels of 8-μm–thick Z-stacks from 665 human islets were analyzed. Immunohistochemistry images were obtained with a DM-R microscope from Leica Microsystems (Nanterre, France). The background level was manually retracted for every islet, and the Pearson coefficient in the colocalized volume was calculated using the colocalization feature of the Bitplane IMARIS software between the SGLT2 and the glucagon, insulin, or somatostatin channel. The Pearson coefficient value ranges from 1 to −1 and indicates the level of correlation between the channels analyzed. A value close to 1 describes a strong correlation or overlaps, whereas a value close to 0 or negative value corresponds to a lower correlation between the two channels of interest.

Cell Culture

Homo sapiens embryonic kidney (HEK293) cells (CRL-1573; ATCC) were cultured in DMEM GlutaMAX 5.5 mmol/L glucose supplemented with 10% FBS and 100 U/mL penicillin/streptomycin (catalog number 15140122; Thermo Fisher Scientific) according to the manufacturer’s recommendations.

siRNA Transfections

HEK293 cells (3.5 × 105/well) were seeded in six-well plates with DMEM GlutaMAX 5.5 mmol/L glucose supplemented with 10% FBS and 100 U/mL penicillin/streptomycin media. After 24 h, cells were transfected with 50 nmol/L scrambled control (catalog number B-001810–10; Dharmacon, Cambridge, U.K.) and 50 nmol/L or 10 nmol/L human SGLT2 siRNA ON-TARGETplus (catalog number L-007590–00; Dharmacon), with Lipofectamine 3000 Transfection reagent (catalog number L3000001; Thermo Fisher Scientific), according to the manufacturer’s instructions. The cells were collected 48 h after transfection for protein extraction. All experiments were performed in duplicate.

Western Blot Analysis

Human islets (2,000 islet equivalents) were harvested in lysis buffer containing 20 mmol/L Tris-acetate, 0.27 mol/L sucrose, 1% Triton X-100, 1 mmol/L EDTA, 1 mmol/L EGTA, 50 mmol/L sodium fluoride, and 10 mmol/L β-glycerophosphate. HEK293 cells were harvested with CellLytic M reagent (catalog number 2978; Sigma-Aldrich, Saint-Quentin-Fallavier, France). Lysis buffers were supplemented with proteinase inhibitors (catalog number P8340; Sigma-Aldrich) and phosphatase inhibitors from PhosSTOP (catalog number 4906837001; Roche, Basel, Switzerland). After sonication, insoluble material was removed by centrifugation at 12,000 rpm for 20 min at 4°C. Protein concentrations were determined using the Pierce BCA protein assay kit (catalog number 23225; Thermo Fisher Scientific). An equal amount of protein (35 µg) was supplemented with 10 mmol/L dithiothreitol (catalog number 707265ML; Thermo Fisher Scientific) and 4× Laemmli sample buffer (catalog number NP008; Thermo Fisher Scientific). Samples were denatured at 75°C for 10 min and separated with 4–12% SDS-PAGE; proteins were transferred to nitrocellulose membranes using the iBlot transfer system. The membranes were blocked with 5% BSA in TBS containing 0.1% (v/v) Tween 20 (TBST) for 60 min at room temperature and washed three times with TBST for 15 min. Membranes were incubated overnight at 4°C with the anti-SGLT2 antibody (Novus Biologicals), anti-SGLT2 antibody (Abcam), or anti-SGLT1 antibody (Merck Millipore) in 5% BSA-TBST (Supplementary Table 3). The membranes were washed three times for 15 min in TBST. The anti-rabbit horseradish peroxidase–conjugated antibody (catalog number NA934V; GE Healthcare Life Sciences, Amersham, U.K.) was diluted 1:10,000 in 5% BSA-TBST for 60 min at room temperature. β-Actin was used as a loading control (Supplementary Table 3) and anti-mouse horseradish peroxidase–conjugated antibody (catalog number NXA91; GE Healthcare Life Sciences) diluted 1:10,000 in 5% BSA-TBST for 60 min at room temperature. The membranes were washed three times for 15 min in TBST, and blots were developed with ECL Plus (cat. no. RPn2236; GE Healthcare Life Sciences) according to the manufacturer’s instructions. Digital images were taken and analyzed with the Amersham imager 680 system.

Human Islet Culture

Human islets were isolated as previously described (24). For hormone secretion experiments in response to glucose and drug treatments, we performed a sample size analysis, which predicted that for 80% power with 5% significance as a measure of sensitivity, the minimal sample size required to depict a statistical difference is five human islet donors (https://powerandsamplesize.com/Calculators/Test-1-Mean/1-Sample-Equality). Human islets were cultured in glucose-free RPMI 1640 medium (catalog number 11879020; Gibco, Life Technologies, Paris, France) supplemented with 0.625% human serum albumin, 1% penicillin/streptomycin, and 5 mmol/L glucose. Islets were exposed to 1 mmol/L glucose or 6 mmol/L glucose in the presence or absence of different concentrations of dapagliflozin (at concentrations of 10 nmol/L, 100 nmol/L, 1 μmol/L, and 12 μmol/L) (AstraZeneca, Mölndal, Sweden). After a 1-h incubation, supernatants and islets were collected, and hormone secretion was measured. Each condition was performed in quadruple for each donor islet preparation. Hormone secretion per hour was normalized and expressed as a percentage of intracellular hormone content per hour as recommended by an expert (25).

Hormone Measurements

Glucagon concentrations were measured using the Glucagon Quantikine ELISA Kit (catalog number DGCG0; R&D Systems Europe, Lille, France) or Glucagon ELISA (catalog number 10-12-71-01; Mercodia AB, Uppsala, Sweden) according to the manufacturer’s instructions. Insulin concentrations were measured using a DxI Access Immunoassay System (Beckman Coulter).

Statistical Analysis

Data are expressed as means ± SD and analyzed by two-way ANOVA, with appropriate corrections such as Tukey and Bonferroni post hoc tests for multiple comparisons. The linear relationships among normally distributed variables were measured by Pearson correlation coefficient (PCC). Statistical analyses were performed using GraphPad Prism 7.0 (GraphPad Software, La Jolla, CA). Differences were considered significant at P < 0.05.

Data and Resource Availability

All data generated or analyzed during this study are included in this published article (and its Supplementary Data).

RNA-Sequencing Analysis Reveals SLC5A2 Expression in Human Pancreatic Islets With a High Interindividual Heterogeneity

Recent single-cell RNA-sequencing (RNA-seq) analysis reported SLC5A2 gene transcripts to be almost absent in human α-, β-, and δ-cells (18). Because single-cell RNA-seq analysis primarily detects highly expressed transcripts, we analyzed deep RNA-seq data sets of 207 intact human islets from the TIGER resource (available at http://tiger.bsc.es/) (20,21). We first confirmed that the SLC5A2 gene is expressed in human islets (median 1.255 transcripts per million [TPM]; coefficient of variation calculated as the ratio of the SD to the mean of 1.24). As expected, SLC5A2 was detected at lower abundance than insulin, glucagon, or somatostatin but at levels similar to those for other α-cell–specific transcripts, such as HPA2 (26) (median 0.072 TPM; coefficient of variation 0.81) or aristaless-related-homeobox (ARX) (27) (median 3.753 TPM; coefficient of variation 1.02), whereas reference genes showed less variation (coefficient of variation for GAPDH 0.66 and for ACTB 0.49). Intriguingly, this analysis also unmasked an unprecedented interindividual heterogeneity of SLC5A2 mRNA levels in the 207 donors (Fig. 1).

Rigorous Antibody Validation Confirms the Specificity of the Novus Biologicals Anti-SGLT2 Antibody

To study the expression of SLC5A2 at the protein level, experiments were first performed to further validate the specificity of the anti-SGLT2 antibody from Novus Biologicals previously used (12,28). Using a recombinant protein antigen corresponding to human SGLT2 (Novus Biologicals) as a blocking agent, we demonstrate the specificity of the SGLT2 antibody for its antigen by a clear reduction in the SGLT2 signal, while neither glucagon nor insulin signals were affected (Fig. 2A). These findings were confirmed in HEK293 cells by immunofluorescence (Fig. 2B) and Western blot analysis (Fig. 2C). Furthermore, siRNA-mediated gene silencing effectively reduced the intensity of only the specific band for SGLT2 (∼62 kDa) in HEK293 cells, compared with nontransfected cells or scrambled controls by Western blot analysis (Fig. 2D) or the SGLT2 signal by immunofluorescence (Fig. 2E). To determine if the intracellular localization of SGLT2 is dependent on prevailing glucose concentrations, HEK293 cells were cultured in 2 mmol/L, 11 mmol/L, or 25 mmol/L glucose concentrations for 24 h. SGLT2 localized in the cytoplasm of HEK293 cells when cultured at 2 mmol/L and 11 mmol/L, while at the high glucose levels of 25 mmol/L, SGLT2 was predominantly present on the membrane (Fig. 2F–H).

SGLT2 Protein Expression in Human Islets Is Highly Heterogeneous Among Donors

Consistent with the RNA-seq data, Western blot analysis revealed that SGLT2 protein expression was also highly heterogeneous in islet preparations (purity >80% endocrine: exocrine) obtained from 10 donors (Fig. 3A). Given the homology between SGLT1 and SGLT2, we also assessed whether the anti-SGLT2 antibody (Novus Biologicals) is specific for SGLT2 and not SGLT1. To do this, islets from seven donors were simultaneously probed with SGLT2 and SGLT1 antibodies. Although both antibodies detected a band at ∼62 kDa, the anti-SGLT1 antibody revealed an additional band at ∼70 kDa, which most likely corresponds to glycosylated SGLT1 (29) (Fig. 3B). Densitometric analysis indicated distinct interislet protein expression patterns between SGLT2 and SGLT1 (Fig. 3C), suggesting that the anti-SGLT2 antibody (Novus Biologicals) does not target SGLT1. Finally, using a second anti-SGLT2 antibody (Abcam) further confirmed SGLT2 expression and heterogeneity in four additional donor islet preparations (Fig. 3D).

PCC Analysis Reveals Colocalization of SGLT2 With Glucagon but Not With Insulin or Somatostatin

Using the technically validated SGLT2 antibody (Novus Biologicals), SGLT2 was found to colocalize with glucagon in human α-cells, while no colocalization was observed between SGLT2 and insulin or somatostatin (Fig. 4A–C). To quantify the proximity of SGLT2 with glucagon, Z-stacks of 665 human pancreatic islet images, derived from 12 human pancreata, were analyzed. Z-projection of the Z-stacks shows the overlapping area in white (Fig. 4D, top panel), whereas all voxels of the Z-stacks are displayed in a dot plot (Fig. 4D, bottom panel). The intensity of colocalization is expressed as the PCC. A value close to 1 indicates a strong correlation, and a negative or close to 0 value describes a weak or no correlation. This colocalization analysis revealed a strong correlation, and thus the proximity, of SGLT2 with glucagon expression (PCC = 0.8399 ± 0.074) as opposed to insulin (PCC = −0.1408 ± 0.157) or somatostatin (PCC = −0.1302 ± 0.100) (Fig. 4E).

Confirmation of SGLT2 Localization in α-Cells Using a Second Antibody

Using a second anti-SGLT2 antibody (Santa Cruz Biotechnology) previously used to demonstrate colocalization with somatostatin in a few single islet cells (30), we confirmed that SGLT2 colocalizes predominantly with glucagon in human α-cells, but not with insulin in β-cells (12) or somatostatin in δ-cells (Supplementary Fig. 1A). Using a third anti-SGLT2 antibody from Abcam, we failed to detect a signal for SGLT2 in human pancreatic tissue by immunofluorescence, as previously reported (17). However, immunohistochemical analysis using this Abcam anti-SGLT2 antibody revealed an SGLT2-specific signal in human islets, which was decreased in the presence of the corresponding blocking peptide (Abcam), while the insulin signal remained intact (Supplementary Fig. 1B).

Immunofluorescence Colocalization Studies Reveal Considerable Interdonor Variability of SGLT2 Protein Expression

Immunofluorescence analysis unmasked a high interindividual heterogeneity in the expression pattern of SGLT2. While the majority of glucagon-positive cells displayed complete or partial colocalization with SGLT2, there were a few other islet cells, which were SGLT2 positive but glucagon negative (Fig. 5A). The identity of these SGLT2-positive islet cells remains elusive because SGLT2 did not colocalize with insulin (Fig. 5B) or somatostatin (Fig. 5C) (using anti-somatostatin antibodies from Genetex or Merck Millipore [Supplementary Fig. 1C]).

SGLT2 Protein Displays Intradonor Islet Heterogeneity of Expression

Using pancreatic sections from five donors, qualitative immunofluorescence analysis on ∼30 islets from each donor revealed a large intradonor heterogeneity in islet size. Representative images of nine islets per donor are shown in Fig. 6A–C. The number of SGLT2-positive cells varied among all intradonor islets (Fig. 6A–C), and the quantitative level of SGLT2 colocalization with glucagon (but not with insulin or somatostatin) was significantly different among donors (P < 0.0001) (Fig. 7A–C).

Glucagon Secretion Triggered by Low-Glucose Stimulation or Dapagliflozin Is Highly Heterogeneous Among Individuals

Given the interindividual variation of SGLT2 protein expression, we next investigated whether dapagliflozin-induced glucagon secretion was also heterogeneous. Analysis of the response of a large number of human islet donor preparations (n = 31) showed that glucagon secretion markedly increased upon culturing in 1 mmol/L glucose (median 4.68 ± 1.53% of content; range 39.94–0.28%) compared with in 6 mmol/L glucose (median 2.12 ± 0.97% of content; range 24.54–0.23%). As previously described (12), treatment with dapagliflozin in the presence of 6 mmol/L glucose resulted in a similar increase of glucagon secretion (median 6.83 ± 1.63% of content; range 29.05–0.43%). However, the magnitude of glucagon response to both glucose and drug was highly variable; the glucagon secretion was 100-fold different among donors (Fig. 8A). This interindividual variability remained evident when the data are presented as fold change to the glucagon secretion at 6 mmol/L glucose (Fig. 8B). Moreover, the glucagon response of human islets exposed to 6 mmol/L glucose and dapagliflozin correlated with the glucagon response after glucose deprivation (1 mmol/L glucose) (r = 0.772; P < 0.0001) (Fig. 8C), indicating that SGLT2 inhibition by dapagliflozin mimics the effect of lowering glucose levels. The corresponding absolute values for glucagon secretion and intracellular content are shown in Supplementary Fig. 2A and B. Noteworthy, islets from three donors were nonresponsive to glucose and/or drug stimulation (Fig. 8D). Notably, high glucose–stimulated insulin secretion and low glucose–stimulated glucagon secretion were only weakly correlated in 31 donors analyzed (Fig. 8E), suggesting that the α-cell has its own glucose-sensing machinery, independent of that of the β-cell. Insulin secretion measured in parallel in 15 donor preparations did not show any effect of dapagliflozin on insulin secretion (Supplementary Fig. 2C) or any correlation with glucagon secretion (Supplementary Fig. 2D). To derive the functional relationship between the dose of dapagliflozin and response, human islet preparations (n = 6 donors) were incubated in the presence of 6 mmol/L glucose with vehicle or dapagliflozin at different concentrations (10 nmol/L, 100 nmol/L, 1 μmol/L, or 12 μmol/L). Dapagliflozin treatment showed a linear trend in increasing glucagon secretion (r = 0.98; P < 0.05), with the maximal induction of glucagon being observed at 12 μmol/L concentration (P < 0.0001) (Supplementary Fig. 2E). These results are in line with a recent study by Perry et al. (31), who have reported a significant increase in glucagon secretion in rat islets by using 10 μmol/L canagliflozin, at 5 mmol/L glucose concentration. In human, for a 10-mg dose of dapagliflozin, the peak of plasma concentrations is ∼150 ng/mL (∼400 nmol/L) (32). Because dapagliflozin is highly bound to albumin (>95%), at the steady state, the free fraction is only 4% in man (33). As the CMRL-1066 media used to culture islets contains 0.65% albumin, the effect of dapagliflozin-induced glucagon secretion at the concentration of 12 μmol/L corresponds to a free fraction of 480 nmol/L.

The islet α-cell plays a critical role in the regulation of glucose homeostasis. When blood glucose levels are low, glucagon is released to promote hepatic gluconeogenesis. Several lines of evidence indicate that glucagon secretion is dysregulated in type 1 and type 2 diabetes; in particular, postprandial glucagon is no longer suppressed by glucose and/or insulin, and hyperglucagonemia persists (34). SGLT2 inhibitors have recently been shown to decrease glycemia by inducing glycosuria (10,11) and to reduce the risk of heart failure and major adverse cardiovascular events (4). Due to their efficacy in the treatment of type 2 diabetes, several clinical trials have been performed to assess the efficacy and safety of SGLT2 inhibitors in combination with insulin therapy in patients with type 1 diabetes. Although data on the long-term glycemic control are not available, Taylor et al. (8,35) reported that SGLT2 inhibitor therapy provokes serious adverse events, including an approximately sixfold increased risk of diabetic ketoacidosis. These new data have brought to light the importance of the understanding of the regulation of glucagon and β-hydroxybutyrate levels elicited by the use of SGLT2 inhibitors.

Type 1 diabetes is characterized by considerable interindividual variability with respect to pathology, disease progression, and efficacy of therapeutic interventions (36). This variability in clinical outcomes may be reflected by the heterogeneous phenotype of human islets. While recent studies have revealed that the β-cell is not a homogeneous cell population (37), glucagon-secreting α-cells have gained less attention. In this study, we report that the α-cell is also a heterogeneous cell type based on gene transcript analysis with a particular emphasis on SLC5A2 and SGLT2 protein morphological readouts, which directly regulates glucagon secretion at physiological glucose concentrations (12,15). Of note, previous single-cell RNA-seq studies failed to detect SLC5A2 gene transcripts in a few thousand islet cells (i.e., <10 human islets) obtained from few individuals (<10) (18,38). However, we have shown in this study that the SLC5A2 gene is expressed in the whole islet and that its expression is highly heterogeneous among the 207 donors analyzed. The reasons why SLC5A2 levels diverge between the whole-islet versus single-islet cells remain elusive, but may be related to detection limitation of lower expressed transcripts in the single-cell analysis (39). Using a large number of human pancreatic tissues (665 islets), we confirmed our previous findings (12) that SGLT2 is predominantly localized in α-cells, but now also show that its expression is highly heterogeneous among different donors or between islets from the same donor. Due to the large number of donor islets analyzed, we were able to scrutinize the high interindividual variation in low glucose–stimulated glucagon secretion and found it to be closely correlated to dapagliflozin-induced glucagon secretion in the presence of basal glucose levels. Our data suggest that under both conditions, similar downstream signaling pathways mediate the induction of glucagon secretion, and therefore, further studies are warranted to elucidate the underlining mechanisms. In this context, human islet heterogeneity in SGLT2 expression and function is an important parameter to consider for study design and interpretation of results. Notably, islets from three donors did not respond to low glucose or dapagliflozin with respect to glucagon secretion, thus suggesting a functional impairment of these islets to glucose sensing and SGLT2 inhibition. It is tempting to speculate that these findings in α-cells may have clinical relevance, because they may partially explain the heterogeneous plasma glucagon levels found in patients (40) and their variable response to pharmacological treatment with SGLT2 inhibitors (41,42).

C.S., M.M., and A.A.-M. contributed equally to this work.

See accompanying article, p. 864.

Acknowledgments. The authors thank Dr. Meryem Tardivel and Antonino Bongiovanni of the BioImaging Center Lille (Campus Hospitalo-Universitaire, Lille, France) for expert technical advice.

Funding. This study was funded by Conseil Régional Nord-Pas-de-Calais, Société Francophone du Diabète, the European Commission (FEDER 12003944), the European Genomic Institute for Diabetes (ANR-10-LABX-46), and the European Consortium for Islet Transplantation funded by JDRF. C.S. was supported by the European Foundation for the Study of Diabetes/Lilly. A.A.-M. was supported by I-SITE Université Lille Nord-Europe. TIGER was developed within the European Union’s Horizon 2020 research and innovation program project T2DSystems, under grant agreement number 667191 (to M.C.). B.S. is a recipient of an Advanced European Research Council grant (694717).

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

Author Contributions. C.S., M.M., A.A.-M., and A.P. performed most experiments, analyzed the data, and drafted and revised the manuscript. V.G., N.D., E.M., J.T., G.P., and A.C. performed specific experiments. A.P., M.C., B.S., and F.P. analyzed and interpreted data and wrote and edited the manuscript. J.K.-C. supervised human islet isolation, participated in the experimental design, and analyzed and interpreted data. C.B. and F.P. supervised the project, analyzed data, and wrote and edited the manuscript. C.B. and F.P. are the guarantors of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Riser Taylor
S
,
Harris
KB
.
The clinical efficacy and safety of sodium glucose cotransporter-2 inhibitors in adults with type 2 diabetes mellitus
.
Pharmacotherapy
2013
;
33
:
984
999
2.
Wiviott
SD
,
Raz
I
,
Sabatine
MS
.
Dapagliflozin and cardiovascular outcomes in type 2 diabetes. Reply
.
N Engl J Med
2019
;
380
:
1881
1882
3.
Akinci
B
.
Dapagliflozin and cardiovascular outcomes in type 2 diabetes
.
N Engl J Med
2019
;
380
:
1881
4.
Neal
B
,
Perkovic
V
,
Mahaffey
KW
, et al.;
CANVAS Program Collaborative Group
.
Canagliflozin and cardiovascular and renal events in type 2 diabetes
.
N Engl J Med
2017
;
377
:
644
657
5.
Gastaldelli
A
,
Repetto
E
,
Guja
C
, et al
.
Exenatide and dapagliflozin combination improves markers of liver steatosis and fibrosis in patients with type 2 diabetes
.
Diabetes Obes Metab
2020
;
22
:
393
403
6.
McCrimmon
RJ
,
Henry
RR
.
SGLT inhibitor adjunct therapy in type 1 diabetes
.
Diabetologia
2018
;
61
:
2126
2133
7.
Petrie
JR
.
SGLT2 inhibitors in type 1 diabetes: knocked down, but up again
?
Lancet Diabetes Endocrinol
2017
;
5
:
841
843
8.
Taylor
SI
,
Blau
JE
,
Rother
KI
.
SGLT2 inhibitors may predispose to ketoacidosis
.
J Clin Endocrinol Metab
2015
;
100
:
2849
2852
9.
U.S. Food and Drug Administration
.
FDA Briefing Document Endocrinologic and Metabolic Drugs Advisory Committee Meeting [Internet], 2019. Available from https://www.fda.gov/downloads/AdvisoryCommittees/CommitteesMeetingMaterials/Drugs/EndocrinologicandMetabolicDrugsAdvisoryCommittee/UCM629485.pdf. Accessed 15 May 2018
10.
Ferrannini
E
,
Muscelli
E
,
Frascerra
S
, et al
.
Metabolic response to sodium-glucose cotransporter 2 inhibition in type 2 diabetic patients
.
J Clin Invest
2014
;
124
:
499
508
11.
Merovci
A
,
Solis-Herrera
C
,
Daniele
G
, et al
.
Dapagliflozin improves muscle insulin sensitivity but enhances endogenous glucose production
.
J Clin Invest
2014
;
124
:
509
514
12.
Bonner
C
,
Kerr-Conte
J
,
Gmyr
V
, et al
.
Inhibition of the glucose transporter SGLT2 with dapagliflozin in pancreatic alpha cells triggers glucagon secretion
.
Nat Med
2015
;
21
:
512
517
13.
Haliyur
R
,
Tong
X
,
Sanyoura
M
, et al
.
Human islets expressing HNF1A variant have defective β cell transcriptional regulatory networks
.
J Clin Invest
2019
;
129
:
246
251
14.
Pedersen
MG
,
Ahlstedt
I
,
El Hachmane
MF
,
Göpel
SO
.
Dapagliflozin stimulates glucagon secretion at high glucose: experiments and mathematical simulations of human A-cells
.
Sci Rep
2016
;
6
:
31214
15.
Saponaro
C
,
Gmyr
V
,
Thévenet
J
, et al
.
The GLP1R agonist liraglutide reduces hyperglucagonemia induced by the SGLT2 inhibitor dapagliflozin via somatostatin release
.
Cell Rep
2019
;
28
:
1447
1454.e4
16.
Solini
A
,
Sebastiani
G
,
Nigi
L
,
Santini
E
,
Rossi
C
,
Dotta
F
.
Dapagliflozin modulates glucagon secretion in an SGLT2-independent manner in murine alpha cells
.
Diabetes Metab
2017
;
43
:
512
520
17.
Suga
T
,
Kikuchi
O
,
Kobayashi
M
, et al
.
SGLT1 in pancreatic α cells regulates glucagon secretion in mice, possibly explaining the distinct effects of SGLT2 inhibitors on plasma glucagon levels
.
Mol Metab
2019
;
19
:
1
12
18.
Kuhre
RE
,
Ghiasi
SM
,
Adriaenssens
AE
, et al
.
No direct effect of SGLT2 activity on glucagon secretion
.
Diabetologia
2019
;
62
:
1011
1023
19.
Bordeaux
J
,
Welsh
A
,
Agarwal
S
, et al
.
Antibody validation
.
Biotechniques
2010
;
48
:
197
209
20.
van de Bunt
M
,
Manning Fox
JE
,
Dai
X
, et al
.
Transcript expression data from human islets links regulatory signals from genome-wide association studies for type 2 diabetes and glycemic traits to their downstream effectors
.
PLoS Genet
2015
;
11
:
e1005694
21.
Fadista
J
,
Vikman
P
,
Laakso
EO
, et al
.
Global genomic and transcriptomic analysis of human pancreatic islets reveals novel genes influencing glucose metabolism
.
Proc Natl Acad Sci U S A
2014
;
111
:
13924
13929
22.
Patro
R
,
Duggal
G
,
Love
MI
,
Irizarry
RA
,
Kingsford
C
.
Salmon provides fast and bias-aware quantification of transcript expression
.
Nat Methods
2017
;
14
:
417
419
23.
Hart
NJ
,
Powers
AC
.
Use of human islets to understand islet biology and diabetes: progress, challenges and suggestions
.
Diabetologia
2019
;
62
:
212
222
24.
Kerr-Conte
J
,
Vandewalle
B
,
Moerman
E
, et al
.
Upgrading pretransplant human islet culture technology requires human serum combined with media renewal
.
Transplantation
2010
;
89
:
1154
1160
25.
Henquin
JC
.
The challenge of correctly reporting hormones content and secretion in isolated human islets
.
Mol Metab
2019
;
30
:
230
239
26.
Dorrell
C
,
Schug
J
,
Lin
CF
, et al
.
Transcriptomes of the major human pancreatic cell types
.
Diabetologia
2011
;
54
:
2832
2844
27.
Collombat
P
,
Mansouri
A
,
Hecksher-Sorensen
J
, et al
.
Opposing actions of Arx and Pax4 in endocrine pancreas development
.
Genes Dev
2003
;
17
:
2591
2603
28.
Scafoglio
CR
,
Villegas
B
,
Abdelhady
G
, et al
.
Sodium-glucose transporter 2 is a diagnostic and therapeutic target for early-stage lung adenocarcinoma
.
Sci Transl Med
2018
;
10
:
eaat5933
29.
Balen
D
,
Ljubojevic
M
,
Breljak
D
, et al
.
Revised immunolocalization of the Na+-D-glucose cotransporter SGLT1 in rat organs with an improved antibody
.
Am J Physiol Cell Physiol
2008
;
295
:
C475
C489
30.
Vergari
E
,
Knudsen
JG
,
Ramracheya
R
, et al
.
Insulin inhibits glucagon release by SGLT2-induced stimulation of somatostatin secretion
.
Nat Commun
2019
;
10
:
139
31.
Perry
RJ
,
Rabin-Court
A
,
Song
JD
, et al
.
Dehydration and insulinopenia are necessary and sufficient for euglycemic ketoacidosis in SGLT2 inhibitor-treated rats
.
Nat Commun
2019
;
10
:
548
32.
Kasichayanula
S
,
Liu
X
,
Lacreta
F
,
Griffen
SC
,
Boulton
DW
.
Clinical pharmacokinetics and pharmacodynamics of dapagliflozin, a selective inhibitor of sodium-glucose co-transporter type 2
.
Clin Pharmacokinet
2014
;
53
:
17
27
33.
Meng
W
,
Ellsworth
BA
,
Nirschl
AA
, et al
.
Discovery of dapagliflozin: a potent, selective renal sodium-dependent glucose cotransporter 2 (SGLT2) inhibitor for the treatment of type 2 diabetes
.
J Med Chem
2008
;
51
:
1145
1149
34.
Unger
RH
,
Orci
L
.
Paracrinology of islets and the paracrinopathy of diabetes
.
Proc Natl Acad Sci U S A
2010
;
107
:
16009
16012
35.
Taylor
SI
,
Blau
JE
,
Rother
KI
,
Beitelshees
AL
.
SGLT2 inhibitors as adjunctive therapy for type 1 diabetes: balancing benefits and risks
.
Lancet Diabetes Endocrinol
2019
;
7
:
949
958
36.
Woittiez
NJ
,
Roep
BO
.
Impact of disease heterogeneity on treatment efficacy of immunotherapy in type 1 diabetes: different shades of gray
.
Immunotherapy
2015
;
7
:
163
174
37.
Nasteska
D
,
Hodson
DJ
.
The role of beta cell heterogeneity in islet function and insulin release
.
J Mol Endocrinol
2018
;
61
:
R43
R60
38.
Segerstolpe
Å
,
Palasantza
A
,
Eliasson
P
, et al
.
Single-cell transcriptome profiling of human pancreatic islets in health and type 2 diabetes
.
Cell Metab
2016
;
24
:
593
607
39.
Mawla
AM
,
Huising
MO
.
Navigating the depths and avoiding the shallows of pancreatic islet cell transcriptomes
.
Diabetes
2019
;
68
:
1380
1393
40.
Wewer Albrechtsen
NJ
,
Kuhre
RE
,
Pedersen
J
,
Knop
FK
,
Holst
JJ
.
The biology of glucagon and the consequences of hyperglucagonemia
.
Biomarkers Med
2016
;
10
:
1141
1151
41.
Ferrannini
E
.
Sodium-glucose co-transporters and their inhibition: clinical physiology
.
Cell Metab
2017
;
26
:
27
38
42.
Wanner
C
,
Marx
N
.
SGLT2 inhibitors: the future for treatment of type 2 diabetes mellitus and other chronic diseases
.
Diabetologia
2018
;
61
:
2134
2139
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/content/license.

Supplementary data