Na⁺-d-glucose Cotransporter SGLT1 is Pivotal for Intestinal Glucose Absorption and Glucose-Dependent Incretin Secretion

Mice were handled in compliance with Institutional guidelines and German, U.K., and U.S. laws. Comparison between Sglt1^−/− and wild-type mice was performed between the 8th and 12th generation of backcrossing of Sglt1^−/− (129/OLA/C57BL/6 background) with wild-type (C57BL/6 background). Animals were kept in a temperature-controlled environment with a 12-h–light/12-h–dark cycle.

Standard maintenance chow (Ssniff V1534–000 R/M-H, 10 mm) was obtained from Spezialdiäten GmbH, Soest, Germany. It contained 12.8 MJ kg⁻¹ metabolizable energy and was composed of 36.4% starch, 19% protein, 4.9% fiber, 4.7% mono- and disaccharides, and 3.3% fat, minerals, and vitamins. The glucose-galactose–free diet was prepared by Altromin Spezialfutter GmbH (Lage, Germany) and contained 13.2 MJ kg⁻¹ metabolizable energy. It was composed of 33.8% protein, 30.7% fiber, and 20.5% fat, minerals, and vitamins.

The antibodies used are described in Table 1.

Immunofluorescence histochemistry was performed with cryo-sections of p-formaldehyde–fixed tissue as described (22). For double staining sections were first incubated with GIP-Ab or GLP-1-Ab and respective secondary antibody and washed. Thereafter sections were incubated overnight at 4°C with SGLT1-Ab, rinsed with PBS, incubated with GAR-AF555 for 60 min at room temperature, and washed. Laser scanning fluorescence microscopy was performed with a confocal laser-scanning imaging system, LSM-510 using the argon laser for green fluorochromes (λ_ex = 488 nm, LP505 nm), and the helium-neon laser for red fluorochromes (λ_ex = 543 nm, BP560–615 nm) (Zeiss, Jena, Germany). The Zeiss LSM-510 software 2.5 SP2 stack, multi track, 8 bit scan mode was used.

After a 16- to 18-h fasting period mice were gavaged with a 10% d-glucose solution (2 mg/g body wt) in PBS, a 40% d-glucose solution (6 mg/g body wt) in PBS, or with olive oil (10 µL/g body wt). Control animals were gavaged with PBS.

For determination of d-glucose and insulin blood was collected from the tail. Concentrations of d-glucose were determined with an Ascensia CONTOUR Meter using Ascensia MICROFIL Test Strips (Bayer Vial GmbH, Leverkusen). Insulin levels were measured with the Insulin (Mouse) Ultrasensitive EIA from ALPCO Diagnostics (Salem, NH). For determination of enterohormones, mice were anesthetized with isoflurane and blood was collected after heart tap. For GLP-1 determination blood was immediately mixed with a dipeptidyl peptidase (DPP)-IV inhibitor yielding a final concentration of 100 μmol/L (Millipore Corporation, Billerica, MA). Active GLP-1 was determined by an ELISA kit supplied by Millipore Corporation (GLP-1(7–36) active, kit EGLP-35 K) (Fig. 5C). Total GLP-1 was determined by an ELISA kit provided by MesoScale Discovery (Gaithersburg, MD; kit K150JVC-4; Fig. 5D and Supplementary Fig. 5B). GIP was determined by an ELISA kit obtained from Millipore Corporation that is specific for both active GIP (1–42) and inactivated GIP (3–42) of rat and mouse (kit EZRMGIP-55 K).

Mice were killed 30 min after gavage. BBM vesicles (BBMVs) were isolated by magnesium precipitation followed by differential centrifugation. Small intestines of three mice were homogenized in 35 mL of homogenization buffer (100 mmol/L mannitol, 2 mmol/L HEPES/Tris [pH 7.1]). MgCl₂ solution (1 mol/L) was added to a concentration of 10 mmol/L, and the suspension was kept on ice for 20 min. After 15 min centrifugation at 3,000 g, the pellet containing nuclei and mitochondria was discarded. The supernatant was centrifuged for 30 min at 27,000 g, and the resulting pellet was suspended in 35 mL vesicle buffer (100 mmol/L mannitol, 20 mmol/L HEPES/Tris [pH 7.4]). After 30 min centrifugation at 27,000 g, the pellet was suspended in 300 µL vesicle buffer. The BBMVs were snap-frozen in liquid nitrogen and stored at −70°C until use.

BBMVs were thawed in a water bath, and uptake of radiolabeled substrates was determined at 22°C using the rapid filtration technique (23). To measure sodium-dependent and phlorizin-inhibitable uptake of α-methyl-d-glucoside (AMG), vesicles were incubated with vesicle buffer containing 0.1 mmol/L [¹⁴C]AMG plus either 100 mmol/L NaSCN or 100 mmol/L NaSCN plus 0.2 mmol/L phlorizin or 100 mmol/L KSCN. Uptake was stopped with ice-cold vesicle buffer containing 100 mmol/L NaSCN plus 0.2 mmol/L phlorizin (stop solution), and vesicles were washed on nitrocellulose filters with stop solution. The radioactivity on the filters was measured and phlorizin-inhibitable or sodium-dependent AMG uptake was calculated. To measure GLUT2-mediated d-glucose uptake, which is inhibited by glucosamine (24), BBMVs were incubated for 1 min at 22°C with 20 mmol/L HEPES/Tris [pH 7.4] containing 100 mmol/L [³H]-d-glucose plus 100 mmol/L mannitol or plus 100 mmol/L d-glucosamine. Uptake was stopped with ice-cold 20 mmol/L HEPES/Tris [pH 7.4] containing 100 mmol/L mannitol plus 100 mmol/L d-glucosamine, and vesicles were washed on nitrocellulose filters.

Everted rings of duodenum and jejunum were incubated for 2 min in Krebs-Ringer buffer containing 7 μmol/L [¹⁴C]AMG without and with 0.2 mmol/L phlorizin. Uptake was stopped with ice-cold Ringer buffer containing 0.2 mmol/L phlorizin. The length of individual segments was measured under a microscope, the segments were solubilized, and radioactivity was analyzed.

SGLT1 was expressed in oocytes of Xenopus laevis, and measurements of AMG uptake were performed as described (25). The cDNA of mouse Sglt1 (accession number BC003845) was cloned into the vector pRSSP, and m7G(5′)G-capped cRNA was prepared. Oocytes were injected with 5 ng of mouse Sglt1 cRNA and incubated 2 days for expression. For transport measurements the oocytes were incubated for 20 min at room temperature with [¹⁴C]AMG or [³H]d-glucose in the absence and presence of 100 μmol/L phlorizin, and phlorizin-inhibited uptake was analyzed.

Glomerular filtration rate (GFR) and renal glucose reabsorption were determined in awake and anesthetized mice as described previously (20,26,27).

The strategy to generate mice lacking part of the Sglt1 promotor and the first exon of Sglt1 is outlined in Supplementary Fig. 1A. Sglt1^+/− mice and Sglt1^−/− mice were identified by Southern blots and PCR (Supplementary Fig. 1B and C). To verify effects on glucose transport, we expressed murine Sglt1 or murine Sglt1 mutant without exon 1 in Xenopus laevis oocytes and measured phlorizin (100 μmol/L)-inhibitable uptake of 50 μmol/L AMG, a SGLT-specific substrate that is not transported by GLUT transporters. Whereas a high transport activity was observed after expression of SGLT1 wild type, no significant AMG uptake was observed with the SGLT1 mutant (Supplementary Fig. 2). Breeding of Sglt1 heterozygote mice demonstrated Mendelian type inheritance (from 531 outcomes 51% were heterozygotes, 27% wild type, and 22% Sglt1^−/−), indicating embryonic survival of Sglt1^−/− mice. Whereas preweaning Sglt1^−/− mice appeared to be healthy, Sglt1^−/− mice died within 2 days after weaning when they received standard diet. In contrast, weaned Sglt1^−/− mice survived well and were fertile when they were fed a diet that did not contain free monosaccharides and disaccharides (Fig. 1A). To determine whether the inactivation of Sglt1 in mice induced a GGM syndrome as in humans (19), two-month-old mice fed with glucose-galactose–free diet were transitioned to standard diet (Fig. 1B). Whereas wild-type mice did not show a significant change in weight, the Sglt1^−/− mice appeared unwell after 2 days, became weaker, lost body weight, and died within 7 to 12 days after the dietary change (Fig. 1B). Laparatomy 2 days after reapplication of standard diet revealed distended small and large intestines in the Sglt1^−/− mice (Fig. 1C). Absence of Sglt1 mRNA and SGLT1 protein in the small intestine of Sglt1^−/− mice was verified by Northern blotting, Western blotting, and immunohistochemistry (Fig. 1D–F).

To determine the contribution of SGLT1 to Na⁺-d glucose cotransport across BBM of small intestine we compared AMG uptake across the BBM of wild-type versus Sglt1^−/− mice. Comparison was performed in mice fed a glucose-galactose–free diet. Three methods were used: measurement of phlorizin-inhibitable AMG uptake into everted rings of small intestine (Fig. 1G), measurement of glucose-induced short circuit currents across small intestinal mucosa (Fig. 1H), and measurement of phlorizin-inhibitable AMG uptake into isolated BBMVs (Fig. 2C). The data show that absence of SGLT1 abolishes Na⁺-d-glucose cotransport across the small intestinal BBM. Mice express SGLT1 homologs SGLT3a and SGLT3b (30). Confirming recently reported data (30), we observed phlorizin-inhibitable uptake of AMG after expression of SGLT3b in oocytes of Xenopus laevis. By comparison, phlorizin-inhibitable AMG uptake could not be detected after expression of SGLT3a (Supplementary Fig. 3A). Using immunohistochemistry we observed immunoreactivity of an antibody raised against SGLT3b at the BBM of small intestinal enterocytes (Supplementary Fig. 3B). The functional data in BBMs, however, indicate that SGLT3b activity cannot substitute for SGLT1.

Kellet and coworkers raised the hypothesis that at high luminal glucose concentrations GLUT2 is incorporated into the BBM and is the dominant pathway for d-glucose translocation across the BBM under these conditions (5,6,8). These authors postulated that the incorporation of GLUT2 into the BBM is triggered by SGLT1 activity leading to activation of the Ca²⁺ channel Ca_v1.3 (5,8). However, in a more recent review it was proposed that Ca_v1.3 is activated by sweet taste receptors and endocrine and paracrine hormones including insulin (6).

We compared the increase of plasma d-glucose concentrations after gavage of a solution containing d-glucose (2 mg/g body wt) in wild-type versus Sglt1^−/− mice (Fig. 2A). In wild-type mice the increase of d-glucose concentrations in plasma was significantly higher than in Sglt1^−/− mice (3.3-fold vs. 1.8-fold after 13 min). The increase of plasma d-glucose after glucose gavage in Sglt1^−/− mice (1.8 ± 0.2-fold) was similar in magnitude to the increase observed in Sglt1^−/− mice (1.8 ± 0.1-fold) and wild-type mice (1.7 ± 0.1-fold) after gavage of solvent control (Fig. 2A) and may reflect stress (adrenalin)-induced release of d-glucose from the liver. These data provide evidence that SGLT1 is required for small intestinal mass absorption of d-glucose.

To determine whether d-glucose transport across the BBM after a glucose bolus is mediated by SGLT1 or GLUT2 we performed studies with BBMVs from wild-type and Sglt1^−/− mice. About 10-fold higher K_m values were obtained for SGLT1-mediated uptake of d-glucose/AMG across small intestinal BBMs of wild-type mice compared with values obtained from Xenopus oocytes expressing mouse SGLT1 (Fig. 3). BBMVs were prepared from mice gavaged with PBS or PBS containing d-glucose (6 mg/g body wt) and killed 30 min later. Whereas in BBMVs of wild-type mice, which received the glucose bolus, the K_m value of sodium-dependent AMG uptake was modestly (25%) reduced compared with wild-type mice receiving only PBS (Fig. 2B), the V_max value of AMG uptake was 4.4-fold increased (Fig. 2C). The corresponding amount of SGLT1 protein in BBMVs estimated from Western blots was increased 2.4-fold (Fig. 2D). In BBMVs of Sglt1^−/− mice no uptake of the SGLT specific substrate AMG was detectable without and with d-glucose gavage (Fig. 2C). The data indicate upregulation of SGLT1 expression in the BBM after gavage with d-glucose in mice.

Maximal transport activity (V_max) of GLUT2 across the BBM was determined by measuring the uptake of 100 mmol/L d-glucose (partially [³H]-labeled) that could be blocked by the GLUT2 specific inhibitor glucosamine (100 mmol/L) (24). In oocytes expressing mouse SGLT1 we verified that glucosamine does not inhibit SGLT1 (Supplementary Fig. 4). V_max of GLUT2-mediated d-glucose uptake into BBMVs in wild-type mice without glucose gavage was similar to Sglt1^−/− mice without or with glucose gavage (43–70 pmol × mg⁻¹ × s⁻¹; Fig. 2E). Gavage of d-glucose increased V_max of GLUT2-mediated d-glucose uptake into BBMVs of wild-type mice 2.7-fold (168 ± 31 pmol × mg⁻¹ × s⁻¹). Western blots using antibodies against GLUT2 from human (GLUT-Ab) that cross-react with mouse GLUT2 (21) showed a similar magnitude in the increase of GLUT2 protein expression in BBMVs of wild-type mice after the d-glucose bolus (Fig. 2F). The increase in both the GLUT2 protein expression and the GLUT2-mediated d-glucose uptake in response to the glucose bolus were blunted in Sglt1^−/− mice. These findings are consistent with a role of SGLT1 in the upregulation of GLUT2 activity in the BBM by high glucose concentration.

Without d-glucose bolus the V_max of AMG uptake via SGLT1 was 5.5 times higher than maximal d-glucose uptake via GLUT2, and after glucose gavage the V_max of AMG uptake via SGLT1 was 8.7 times higher compared with d-glucose uptake via GLUT2 (compare Fig. 2C and E). This indicates that d-glucose uptake via the luminal membrane of small intestinal enterocytes is predominantly (>80%) mediated via SGLT1.

We investigated the expression of SGLT1 in enteroendocrine L-cells and K-cells, which are known to secrete GLP-1 and GIP, respectively (Fig. 4). Immunohistochemistry was performed using double staining with specific primary antibodies raised in rabbit or goat and fluorophore-labeled secondary antibodies raised in chicken. Fluorescence was visualized by laser scanning microscopy using an optical section thickness of 0.1 µm. Coexpression of SGLT1 with GLP-1 and GIP was observed in individual cells of jejunal crypts. SGLT1-Ab immunoreactivity was observed at the BBM of enteroendocrine cells, whereas immunoreactivities for antibodies against GIP or GLP-1 were predominantly observed in intracellular compartments (Fig. 4).

Oral application of d-glucose stimulates the pancreatic secretion of insulin more than intravenous application. In humans, this surplus of insulin secretion is a result of the combined effects of GIP and GLP-1 (12,31). After exposure of the small intestine to high glucose the concentrations of these enterohormones increase in systemic blood.

Thirteen minutes after glucose gavage (2 mg/g body wt), plasma insulin was approximately two times higher in wild-type mice compared with Sglt1^−/− mice (Fig. 5A). At the same time, GIP was increased ∼10-fold in the systemic blood of wild-type mice but was not changed in Sglt1^−/− mice (Fig. 5B). By comparison 13 min after gavage with olive oil (10 µL/g body wt), GIP increased four- to fivefold in the systemic blood of both wild-type and Sglt1^−/− mice. Thirteen minutes after glucose gavage the active GLP-1 concentration in systemic blood was not increased in wild-type or Sglt1^−/− mice, whereas application of oil induced a SGLT1-independent twofold increase in active GLP-1 (Fig. 5C). Because the concentration of active GLP-1 in systemic blood may be reduced by degradation we also measured total GLP-1 in the systemic blood 5 min after gavage of a greater d-glucose load (6 mg/g body wt; Fig. 5D). Under these conditions a significant glucose-dependent increase of GLP-1 concentration was observed in wild-type mice but not in Sglt1^−/− mice. Measuring glucose-dependent stimulation of GIP and GLP-1 secretion from primary intestinal cultures prepared from Sglt1^+/− and Sglt^−/− mice, we confirmed that SGLT1 is required for glucose-dependent stimulation of these enterohormones (Supplementary Fig. 5).

In kidney the low affinity Na⁺-d-glucose cotransporter SGLT2 is located in BBMs of S1 segments of proximal tubules, whereas the high affinity Na⁺-d-glucose cotransporter SGLT1 is located in BBMs of the S2 and S3 segments (20,22). We confirmed that no expression of SGLT1 was observed in kidneys of Sglt1^−/− mice (Supplementary Figs. 6A and 7A) and determined at mRNA and protein levels that the expression of SGLT2 was not changed after removal of SGLT1 (Supplementary Figs. 6B and 7B and C). In wild-type and Sglt1^−/− mice similar mRNA and protein concentrations of the passive glucose transporters GLUT1, GLUT2, and GLUT5 were determined (Supplementary Fig. 7B and C).

Spontaneous urine collections in awake male mice revealed that urinary glucose concentrations were significantly increased in Sglt1^−/− compared with wild-type mice, whereas blood glucose levels were not different (Fig. 6A). Fractional glucose reabsorption was 100% in male wild-type and 97% in male Sglt1^−/− mice (Fig. 6B). In male mice, GFR measured in awake and anesthetized animals, blood pressure, heart rate, plasma urea concentrations, food intake and fluid intake, and renal excretion of fluid, Na⁺, and Cl⁻ were not changed after removal of SGLT1. Details and differences in food intake and fluid intake between males and females are described in the Supplementary Data.

Quantitative free-flow collections of tubular fluid were performed from the last surface loop of proximal convoluted tubules. Mean values for single nephron GFR and the amount of glucose filtered per nephron were not different between male Sglt1^−/− and wild-type mice (Fig. 7A and B). However, the concentration and delivery of glucose were increased in Sglt1^−/− versus wild-type mice, whereas fractional glucose reabsorption was reduced (Fig. 7C and D). In contrast with fractional reabsorption of glucose, fractional reabsorption of fluid and chloride was not significantly different between male Sglt1^−/− and wild-type mice (fluid: 0.48 ± 0.02 vs. 0.45 ± 0.03; chloride: 0.47 ± 0.03 vs. 0.45 ± 0.03; n = 17–25 nephrons in each group, not significant).

In the current study the role of SGLT1 for small intestinal glucose absorption and incretin secretion was investigated comparing nondiabetic mice without and with expression of SGLT1. Evidence is presented that SGLT1 mediates the majority of d-glucose transport across the BBM of enterocytes independent of the glucose load. SGLT3b, a second functional Na⁺-d-glucose cotransporter in the BBM of mice, does not contribute significantly. After high glucose load some GLUT2 is incorporated into the BBM, but this has minor impact on small intestinal glucose absorption. It is noteworthy that this translocation of GLUT2 was not observed in Sglt1^−/− mice, indicating that SGLT1 rather than sweet taste receptors is pivotal in this upregulation (5,6,8). Without and with d-glucose load the capacity of SGLT1 versus GLUT2 for d-glucose transport across the BBMs was 5.5 and 8.7 times higher, respectively. In response to continuous feeding of sugar-rich diets and during diabetes, however, the contribution of GLUT2 to small intestinal glucose absorption may become more relevant. Under these conditions the expression of SGLT1 and GLUT2 is increased, and an increase of GLUT2 in the BBM was observed after gavage with glucose or fructose (32–35).

Removal of SGLT1 in mice resulted in GGM syndrome (OMIM 182380) that could be cured by a glucose-galactose–reduced diet as has been observed in patients with loss-of-function mutations in SGLT1 (SLC5A1) (18,19). GGM is characterized by severe diarrhea and dehydration as a result of osmotic loss by retention of nonabsorbed d-glucose, d-galactose, and sodium in the small intestinal lumen. In contrast with humans, where severe diarrhea and dehydration are observed in suckling neonates, suckling Sglt1^−/− mice showed no obvious symptoms. This species difference may be explained by the three to four times higher concentrations of lactose in human milk compared with murine milk (36,37). In the small intestine lactose is degraded by lactase-isomaltase and provides the only relevant source of d-glucose and d-galactose in suckling neonates.

Confirming previous results (15,16) we showed that SGLT1 is expressed in the luminal membrane of enteroendocrine cells staining positive for the incretin hormones, GLP-1 and GIP. Active forms of GLP-1 and GIP are detectable in the systemic blood of human subjects after glucose ingestion (38,39), but are rapidly inactivated in the circulation by DPP-IV (10). Both GIP and GLP-1 trigger glucose-dependent insulin secretion, and although the effectiveness of GIP is impaired in people with type 2 diabetes (40), the preserved insulinotropic effect of GLP-1 (40) has led to the successful development of GLP-1 mimetics and DPP-IV inhibitors as new classes of antidiabetic agent (41). Earlier in vitro studies have demonstrated that SGLT1-mediated monosaccharide transport triggers incretin secretion via the electrogenic coupled influx of Na⁺, which results in membrane depolarization, electrical activity, voltage-gated Ca²⁺ entry, and exocytosis (16,42), although the contribution of this pathway to glucose-mediated secretion in vivo has remained controversial. Using the SGLT1^−/− mice we now demonstrate the pivotal role of SGLT1 for both GLP-1 and GIP secretory responses triggered by oral glucose in vivo. This is consistent with the reported phlorizin sensitivity of glucose-induced incretin increases in the portal vein (13) showing the involvement of a phlorizin-sensitive SGLT-type transporter or channel (3,30,43), as well as with the finding that incretin secretion is triggered in vivo by substrates of SGLT1 (44), but not by other sweet tasting compounds (14). The results suggest that secretion of these hormones would be initiated after food ingestion by the early arrival of even small glucose loads in the small intestine, consistent with previous reports that GLP-1 release is initiated in humans when glucose infusion rates exceed the absorption capacity of the duodenum (45).

Whereas glucose is nearly completely reabsorbed by the kidneys in wild-type mice, Sglt1^−/− mice lose ∼3% of the filtered glucose in the urine, leading to a urinary d-glucose concentration of ∼20 mmol/L. Notably Sglt2 null mice lose only ∼60% of the filtered glucose in the urine although the expression of SGLT1 was downregulated by 50% (20). This suggests that wild-type mice do not use the maximal transport capacity of SGLT1 at normoglycemic conditions. This is different when the glucose load to the SGLT1-expressing S2 and S3 segments is increased. In diabetic patients in which early proximal tubule glucose transport capacity is overwhelmed by high d-glucose concentrations, and even more in diabetic patients treated with an SGLT2 specific inhibitor, SGLT1 may operate at full transport capacity (17).

The Sglt1 null mice provide a valuable tool for future studies of SGLT1 functions under pathological conditions as well as the therapeutic potential of SGLT1 inhibition.

The work was supported by the Deutsche Forschungsgemeinschaft (Grant SFB 487/C1 to H.K., Grant Nutrasense DA 190/10-1 to H.K., and Grant SFB 487/C4 to M.S.); the National Institutes of Health (Grants R01DK56248, R01HL094728, R01DK28602, and P30DK079337 to V.V.); the American Heart Association (GRNT3440038 to V.V. and 10SDG2610034 to T.R.); a Carl Gottschalk Research Grant of the American Society of Nephrology (to T.R.); the Department of Veterans Affairs (to V.V., T.R., and R.C.); the Ministry for Science, Education and Sports of the Republic of Croatia (Grant 022-0222148-2146 to I.S.); and the Wellcome Trust (Grants WT088357 to F.M.G., WT084210 to F.R.). GLP-1 assays in Cambridge were performed with the support of the MRC Centre for Obesity and Related Metabolic Diseases.

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

V.G., A.S., V.V., F.M.G., and I.S. researched data, contributed to the discussion, and reviewed the manuscript. H.Ki., A.J., F.R., F.L., and M.S. researched data and contributed to the discussion. D.K., A.F., S.S., T.R., R.C., M.V.-W., A.S., D.Ba., D.Br., R.R., H.E.P., S.W. researched data. H.Ko. designed the study, researched data, wrote the manuscript, and takes full responsibility for the article and its originality.

The authors thank Dr. Bettina Holtmann from the Clinical Neurobiology of the University of Würzburg (Germany) for germ line transfection and Ursula Roth from the Institute of Anatomy and Cell Biology of the University of Würzburg (Germany) for expert technical assistance.