New insight in understanding the contribution of SGLT1 in cardiac glucose uptake: evidence for a truncated form in mice and humans

Abstract

Although sodium glucose cotransporter 1 (SGLT1) has been identified as one of the major SGLT isoforms expressed in the heart, its exact role remains elusive. Evidence using phlorizin, the most common inhibitor of SGLTs, has suggested its role in glucose transport. However, phlorizin could also affect classical facilitated diffusion via glucose transporters (GLUTs), bringing into question the relevance of SGLT1 in overall cardiac glucose uptake. Accordingly, we assessed the contribution of SGLT1 in cardiac glucose uptake using the SGLT1 knockout mouse model, which lacks exon 1. Glucose uptake was similar in cardiomyocytes isolated from SGLT1-knockout (^Δex1KO) and control littermate (WT) mice either under basal state, insulin, or hyperglycemia. Similarly, in vivo basal and insulin-stimulated cardiac glucose transport measured by micro-PET scan technology did not differ between WT and ^Δex1KO mice. Micromolar concentrations of phlorizin had no impact on glucose uptake in either isolated WT or ^Δex1KO-derived cardiomyocytes. However, higher concentrations (1 mM) completely inhibited insulin-stimulated glucose transport without affecting insulin signaling nor GLUT4 translocation independently from cardiomyocyte genotype. Interestingly, we discovered that mouse and human hearts expressed a shorter slc5a1 transcript, leading to SGLT1 protein lacking transmembrane domains and residues involved in glucose and sodium bindings. In conclusion, cardiac SGLT1 does not contribute to overall glucose uptake, probably due to the expression of slc5a1 transcript variant. The inhibitory effect of phlorizin on cardiac glucose uptake is SGLT1-independent and can be explained by GLUT transporter inhibition. These data open new perspectives in understanding the role of SGLT1 in the heart.

NEW & NOTEWORTHY Ever since the discovery of its expression in the heart, SGLT1 has been considered as similar as the intestine and a potential contributor to cardiac glucose transport. For the first time, we have demonstrated that a slc5a1 transcript variant is present in the heart that has no significant impact on cardiac glucose handling.

INTRODUCTION

To sustain its incessant work, the heart requires a constant supply of various metabolic substrates. Although it is widely known that fatty acids constitute the primary substrates for the heart, cardiac metabolism reliance on glucose becomes crucial in response to insulin, during ischemia, or in hypertrophic hearts. As glucose is hydrophilic, it needs transporters to cross the lipophilic membrane and enter into cells. Cardiac glucose uptake is usually thought to be mediated exclusively by facilitated glucose transporters (GLUTs; slc2 gene family). GLUT1 and GLUT4 are the two major isoforms expressed in the heart (1). GLUT1 is highly expressed in fetal hearts and declines after birth; it mediates mainly basal glucose uptake. Meanwhile, GLUT4 is more expressed in adult hearts and can be translocated to plasma membrane to favor glucose transport under certain stimuli, i.e., insulin or ischemia (1).

In addition to GLUT transporters, a second family of glucose transporters, namely sodium glucose cotransporters (SGLTs; slc5 gene family), has been described. These transporters rely on sodium gradient to bring sugar into cells (2). The SGLT family comprises seven isoforms: SGLT1, SGLT2, SGLT3, SGLT4, SGLT5, SGLT6 (sodium myo-inositol cotransporter 2, SMIT2), and SMIT1 (2). These isoforms differ in substrate specificity, but they all transport glucose with different affinities. SGLT1 exhibits the highest affinity for glucose (K_m ∼ 0.5 mM) among all SGLT isoforms (2). It is abundantly expressed in small intestine and kidneys, where it is responsible for glucose absorption and reabsorption, respectively (2). A high expression of SGLT1 along with SMIT1 has been demonstrated in human, rat, and murine hearts (3–5).

In humans, two transcript variants of SLC5A1 (SGLT1) have been described, variant A and variant B [NM_000343.4 and NM_001256314.2, respectively; National Center for Biotechnology Information (NCBI)]. These variants differ in their 5′ extremity. Variant A contains 15 exons, whereas variant B only has 14 exons, and their sequences present 98% of identity. Variant A encodes a 664-amino acid protein, whereas variant B codes for a shorter protein of 537 amino acids lacking the first three transmembrane domains (2, 6, 7). In mice, such transcript variants have never been described. Murine slc5a1 transcript encodes a 15-exon-long mRNA producing a 666-amino acid protein (8).

Several lines of evidence suggest that SGLT1 could participate in cardiac glucose handling, mainly based on data showing that phlorizin, a SGLT inhibitor, reduces cardiomyocyte glucose uptake (3). However, to date, no studies have confirmed this hypothesis by means of genetic invalidation tools.

Here, we have demonstrated that the SGLT1 ^Δex1KO murine model does not exhibit significant difference in cardiac glucose transport in vitro and in vivo. Moreover, we have revealed that phlorizin, the most common SGLT inhibitor, can inhibit glucose transport independently of SGLT1 when used at an elevated but commonly employed concentration. Furthermore, for the first time, we have demonstrated the existence of slc5a1 transcript variants both in mouse and human hearts, probably resulting in SGLT1-truncated proteins.

MATERIALS AND METHODS

Animal Handling

Animal handling was approved by local authorities (Comité d’éthique Facultaire pour l’Expérimentation Animale, 2016/UCL/MD/027) and performed in agreement with guidelines on animal experimentation at our institution. This study conformed to the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH Publication No. 85-23, revised 1996). SGLT1-deficient mice were generated as described elsewhere (9). Gorboulev et al. (9) performed genetic manipulations that led to a deletion of exon 1 and part of the promoter from slc5a1 murine gene. In this paper, these mice and their control littermates are called SGLT1-knockout (^Δex1KO) and wild type (WT), respectively. SGLT1 ^Δex1KO animals die shortly after weaning from glucose/galactose malabsorption syndrome (9). To avoid the development of this syndrome, the ^Δex1KO mice were fed using a glucose-/galactose-free diet. A WT littermate group fed with a glucose-/galactose-free diet was used as control.

Isolation and Culture of Mouse and Rat Cardiomyocytes

Adult mouse cardiomyocytes were isolated from SGLT1 breeding lineage and C57Bl/6N mice and cultured as described previously (10). Briefly, 2- to 4-month-old mice were anesthetized using a mixture of xylazine (10 mg/kg) and ketamine (80 mg/kg). Once asleep, mice were dislocated, and the heart was immediately excised. Hearts were then perfused for 6 min and subsequently digested with Liberase DH enzyme (0.020 mg/heart, 0501054001; Roche) and trypsin (2.1 mg/heart, 15090-046; Life Technologies) for another 25 min. Cells were purified, cultured, and plated on laminin-coated dishes and incubated for 1 h in fresh minimum essential medium (MEM) with Hank’s salts (11575-032; Life Technologies) supplemented with l-glutamine (2 mM), bovine serum albumin (BSA; 100 μg/mL), penicillin (100 U/mL), and streptomycin (100 μg/mL). Treatment details are provided in figure captions.

Rat adult cardiomyocytes were isolated and cultured, as described previously (11). Briefly, hearts from 250-g male Wistar rats were excised and retrogradely perfused via the aorta. After 10 min of perfusion with Ca²⁺-free Krebs-Henseleit buffer containing 5 mM glucose, 2 mM pyruvate, and 10 mM HEPES (pH 7.4), hearts were digested by adding 0.2 mM Ca²⁺, 1 mg/mL collagenase (Worthington), and 0.4% (wt/vol) BSA. Once digested, hearts were mechanically disrupted with scissors, and Ca²⁺ was added to reach a final concentration of 1 mM. Finally, cardiomyocytes were purified and washed by sedimentation. Cells were cultured in Medium 199 (supplemented with 2 mM carnitine, 5 mM creatine, 5 mM taurine, 10⁻¹⁰M triiodothyronine, 0.2% free fatty acid BSA, and antibiotics) for 1 h before treatment. Treatment details are provided in the figure captions.

Glucose Uptake in Cultured Mouse Cardiomyocytes

Glucose uptake was measured, as described previously (11, 12), by following the [2-³H]glucose detritiation rate, which occurs after glucose phosphorylation during rapid isomerization of hexose-6-phosphate catalyzed by phosphoglucose isomerase. Briefly, for the last 30 min of treatment, 0.2 µCi/ml of [2-³H]glucose were added to the MEM culture medium containing 5.5 mM of glucose. Tritiated water present in the supernatant was separated from tritiated glucose using chromatography on anion exchange resin (borate form) and measured by a scintillation counter.

When specified, glucose uptake was additionally measured by the entry of [³H]-2-deoxyglucose. Once within the cells, [³H]-2-deoxyglucose was phosphorylated by hexokinase and could not be further metabolized. After the cells were washed, the phosphorylated form of [³H]-2-deoxyglucose trapped within the cell was released by cellular lysis and measured using a scintillation counter.

Glycogen Content Measurement

Glycogen content was measured following the protocol described in Passonneau and Lauderdale (13). Hearts from 2- to 4-month-old fed mice were excised and rapidly snap-frozen. The frozen hearts were homogenized in 10 volumes (vol/wt) of 1 M KOH and extracted using an Ultra-Thurax. Samples were incubated at 80°C for 15 min to complete the extraction and then neutralized with 3.3 M acetic acid. Supernatant obtained after centrifugation (6,000 rpm for 6 min) was employed for the assay. Glycogen was converted to glucose via the enzyme amylo-α-1,4-α-1,6-glucosidase. Glycogen content assay was based on quantifying the NADPH content obtained from the NADP conversion during production of 6-phosphogluconic acid from glucose-6-phosphate by the glucose-6-phosphate dehydrogenase enzyme. Glucose-6-phosphate was obtained from glucose by adding the hexokinase enzyme. Absorbance of samples was measured at three different time points: sample and buffer for background absorbance (E0), after adding glucose-6-phosphate dehydrogenase to measure glucose-6-phosphate from sources other than glycogen (E1), and after adding ATP-MgCl₂ and hexokinase (E2). Glycogen content was calculated by subtracting E1 and blank from E2. Results were expressed as µmol/mg frozen tissue.

RNA Extraction, RT-qPCR, and RT-PCR Analysis

Total mRNA was extracted from whole hearts from SGLT1 WT and ^Δex1KO or C57Bl/6N cardiomyocytes via a chloroform/isopropanol procedure (Tripure Isolation Reagent, 11667165001; Roche) and subjected to on-column treatment using an RNAse-free DNase set (79254; Qiagen). RNA content was quantified using a nanodrop (Thermo Scientific); 1 µg of each RNA preparation was submitted to reverse transcription (RT) for 5 min at 35°C, followed by 30 min at 42°C and 5 min at 85°C using an iScript cDNA synthesis kit (1708891; Bio-Rad Laboratories). All PCRs contained 2 μL of RT reaction diluted five times and 0.25 μM of forward and reverse primers in 25 μL volume. PCR was performed with GoTaq G2 DNA polymerase (M7845; Promega), and reactions were heated to 95°C for 2 min, followed by 45 cycles of 95°C for 30 s, 59–60°C for 30 s, and 72°C for 1–2 min, and then heated to 72°C for 5 additional min in an MJ MiniTM Gradient Thermal Cycler (PTC-1148; Bio-Rad). The PCR products were separated by agarose gel electrophoresis (1 to 2% wt/vol) and stained with ethidium bromide. Primer sequences are listed in Supplemental Table S1 (Supplemental Material for this article can be found online at dx.doi.org/10.17504/protocols.io.bqr6mv9e). Quantitative (q)RT-PCR was performed with a CFX Connect Real-Time PCR detection system (Bio-Rad) using a qPCR core kit for SYBR green (Eurogentec). The mRNA level of each gene was normalized to the housekeeping gene ribosomal protein L32 after ΔΔC_T correction. Primer sequences are listed in Supplemental Table S2.

Evaluation of GLUT4 Translocation

Once plated on six-well plates, rat adult cardiomyocytes were infected with adenoviruses expressing a green fluorescent protein (GFP)-modified GLUT4 protein tagged with an hemagglutinin (HA)-tag at the NH₂-terminal part (extracellular part when translocated; HA-GLUT4-GFP), as previously described (14). Then, 48 h after infection, phlorizin (to reach a final concentration of 1 mM) was added 15 min before insulin stimulation, which lasted for 30 additional min. Cells were then fixed using 4% paraformaldehyde and blocked for 30 min with 5% PBS-BSA. Primary antibody [HA-tag (C29F4) Rabbit mAb; Cell Signaling Technology] and secondary antibody [Alexa Fluor 568 donkey anti-rabbit IgG (H + L), Invitrogen], both dissolved in PBS/BSA 1%, were incubated on the cells for 1 h each before mounting on slides. Staining and GFP fluorescence were visualized under a Zeiss Imager.Z1 microscope equipped with an ApoTome device. Red fluorescent dots (HA staining reflecting GLUT4 attached to the membrane) were quantified relative to GFP (total GLUT4 protein) by means of ImageJ software.

Immunoblotting Analysis

Protein content was measured via Bradford method with BSA as standard. Immunoblotting was performed with extracts separated on 8% or 10% SDS-PAGE, transferred to polyvinylidene difluoride membranes that were blocked with BSA 5%, then incubated with corresponding antibodies. Following incubation with an appropriate secondary antibody (anti-rabbit or anti-mouse), proteins were visualized using electrochemical luminescence (Pierce). GLUT detection was performed on whole-heart protein extract; eEF2 served as a loading control when investigating GLUT1, GLUT4, and S6 expression and phosphorylation state, while GAPDH or eEF2 was employed as a loading control when investigating AKT expression and phosphorylation. Primary antibodies used are listed in Supplemental Table S3.

Glucose Uptake in Entire Animals Using Positron Emission Tomography/Computed Tomography Imaging

In vivo cardiac glucose uptake was assessed using whole body positron emission tomography (PET) imaging performed on a dedicated small-animal PET scanner (Mosaic; Philips Medical Systems) with a spatial resolution of 2.7 mm (FWHM) (15). Prior to treatment, mice underwent overnight fasting. Animals were injected with insulin (0.5 U/kg) 5 min before a 2-deoxy-2-[¹⁸F]-d-glucose (¹⁸F-FDG) injection at a dose of 6.7–12 MBq/animal (Betaplus Pharma). For phlorizin treatment, 400 mg/kg (dissolved in 75% saline, 15% DMSO, and 10% ethanol) or vehicle was injected 30 min before insulin stimulation. All injections were performed intraperitoneally. Animals were kept under a heating lamp during the 1-h ¹⁸F-FDG uptake. Thereafter, they were anesthetized with 2% isoflurane in air at 2 L/min and placed on a heated animal bed, which provided anesthesia through a nose cone. A 10-min emission scan was performed 1 h after tracer injection and immediately followed by a 10-min transmission scan using a 370MBq ¹³⁷Cs source. Anesthetized mice were then transferred on the same bed to a μCT (NanoSPECT/CT Small Animal Imager, Bioscan, Inc.) for anatomic reference image acquisition. After correction for attenuation factors obtained from the transmission scan, PET images were reconstructed using a fully three-dimensional interactive algorithm (3D-RAMLA) in a 128 × 128 × 120 matrix, with a voxel of 1 mm³. Regions of interest were delineated on the reconstructed PET images using the PMOD software (PMOD, version 3.406; PMOD Technologies Ltd., Zurich, Switzerland). To discriminate hot pixels from neighboring tissues, PET/computed tomography (CT)-fused images were employed. Cardiac tracer uptake was assessed and expressed as standardized uptake value (SUV) max (normalized to body weight).

RNA Sequencing and Human Biopsies

RNA sequencing (RNA-Seq) data used for mouse studies have been obtained using other breeding lineages, SMIT1 WT and KO, available in our animal facility and described elsewhere (16). RNA-Seq data were aligned with SGLT1 mRNA sequence (NCBI RefSeq NM_019810.4) from mus musculus. For human data, we used three publicly available studies found in the GEO accession database GSE116250 (17), GSE126569 (18) and GSE123976 (19). From these studies, control or nonfailing heart data were aligned to variant A (NCBI RefSeq NM_000343.4) or variant B (NCBI RefSeq NM_001256314.2) SGLT1 mRNA sequences in nucleotide BLAST from NCBI.

Human heart tissues for PCR results were obtained from patients undergoing left ventricular assist device implantation, as described previously (5). This protocol was approved by the local ethics committee (Comité Éthique Hospitalo-Facultaire des Cliniques Universitaires St. Luc, Brussels, Belgium). All experimental procedures were performed in accordance with relevant guidelines and regulations. Intestine samples, used as positive controls, were residual material from surgery.

Statistical Analysis

All data were expressed as means ± SE. Statistical analyses were performed by a two-way ANOVA test, followed by Sidak’s multiple-comparison post hoc test for multiple group comparisons. Comparison of two groups was performed using an unpaired Student’s t test. Statistical significance was considered when the P value was <0.05. All analyses were performed using GraphPad Prism (GraphPad Software).

RESULTS

Cardiac Glucose Uptake in Isolated Cardiomyocytes from SGLT1 ^Δex1KO Mice

As introduced earlier, numerous studies have advocated a role of SGLT1 in cardiac glucose uptake (3, 4, 20–22). Therefore, we studied this process in a SGLT1-knockout mouse model generated by the Koepsell group. This KO has been produced by deleting exon 1 (the reason why we called it SGLT1 ^Δex1KO) and previously used for studies in various tissues, including the heart (9, 23). We first examined the contribution of SGLT1 in glucose uptake in isolated murine cardiomyocytes in culture by measuring the intracellular accumulation of [³H]-2-deoxyglucose (2DG). 2DG uptake did not differ between cardiomyocytes from SGLT1 WT and ^Δex1KO mice under normoglycemia (NG; 5 mM) (Fig. 1A). As previously described (24, 25), basal 2DG uptake was largely increased in the presence of hyperglycemia (HG; 21 mM, 3-fold) or 3 nM insulin (2-fold) in WT cardiomyocytes. 2DG uptake stimulation was similar in ^Δex1KO cardiomyocytes (Fig. 1A).

Figure 1.

Cardiac glucose uptake in sodium glucose cotransporter 1 (SGLT1) wild-type (WT) and knockout (^Δex1KO mice). Glucose uptake in cardiomyocytes isolated from SGLT1 WT (n = 4-6) and ^Δex1KO (n = 6) animals was measured under normoglycemia (NG; 5 mM), under hyperglycemia (2 h, 21 mM), and after insulin stimulation (30 min, 3 nM). A and B: 2 radioactive analogues were used: [³H]-2-deoxyglucose (A) and [2-³H]glucose (B). C: glucose uptake was measured in isolated cardiomyocytes from SGLT1 WT (n = 6) and ^Δex1KO (n = 6) mice after insulin stimulation (30 min, 3 nM) or oligomycin (1 h, 1 µM) stimulation. In this experiment, the contribution of glucose transporter 4 (GLUT4) to glucose uptake under insulin or oligomycin was further evaluated in the presence of indinavir (50 µM, 30 min) used as a specific GLUT4 inhibitor. D: insulin-stimulated cardiac glucose uptake (SUV_max) was assessed in vivo in SGLT1 WT (n = 8) and ^Δex1KO (n = 7) animals using a positron emission tomography (PET) scan. Glycemia measured at the end of experiments were for SGLT1 WT: control, 137 ± 9 mg/dL; insulin, 38 ± 5 mg/dL; and for SGLT1 ^Δex1KO: control, 164 ± 10 mg/dL, insulin, 41 ± 4 mg/dL. E: glycogen content was evaluated in SGLT1 ^Δex1KO (n = 9) and WT (n = 8) total hearts. Data were expressed as means ± SE. Statistical analysis was conducted via a 2-way ANOVA (A–D) or t test (E). *Values statistically different from corresponding control, P < 0.05; ^$values statistically different from insulin alone, P < 0.05; #values statistically different from oligomycin alone, P < 0.05.

Although radiolabeled 2DG is employed to measure glucose uptake, electrophysiological studies have demonstrated that 2DG is a poor substrate for SGLT transporters (26–29). Thus, uptake measured using this approach primarily represents glucose transport through GLUT transporters, but not SGLT1. Therefore, we employed a [2-³H]glucose radiotracer to evaluate SGLT1-dependent glucose uptake. Similarly to 2DG, SGLT1 WT and ^Δex1KO cardiomyocytes exhibited the same level of [2-³H]glucose uptake under basal conditions as well as under HG or insulin (Fig. 1B).

It has recently been hypothesized that SGLT1 participates in ischemia-induced glucose utilization via the activation of AMP-activated protein kinase (AMPK); such a mechanism is proposed as a major process responsible for ischemic preconditioning-induced cardioprotection (30). Therefore, we evaluated the role of SGLT1 on glucose uptake following oligomycin-induced AMPK activation. Oligomycin, like ischemia, increases the AMP/ATP ratio, promoting AMPK activation and glucose uptake independently of insulin (11, 14). A fivefold increase in glucose uptake into WT cardiomyocytes was observed 1 h after adding 1 µM oligomycin (Fig. 1C). As observed with insulin and HG, these responses were not different in ^Δex1KO cardiomyocytes (Fig. 1C), indicating that SGLT1 does not contribute to cardiac AMPK-induced glucose uptake.

In Vivo Glucose Uptake and Glycogen Content in SGLT1 ^Δex1KO Mice

Contribution of SGLT1 to cardiac glucose uptake was further investigated in vivo on anesthetized animals via a micro-PET scan using radiotracer ¹⁸F-FDG, as previously described (31, 32). Maximal standard uptake values (SUV_max) values in hearts of SGLT1 WT and ^Δex1KO mice were similar in basal (0.72 ± 0.18 versus 0.67 ± 0.12, respectively) and insulin-stimulated states (2.76 ± 0.68 versus 2.62 ± 0.22, respectively) (Fig. 1D). As glucose transport, glycogen content, the main form of glucose storage, did not differ between SGLT1 ^Δex1KO and WT mice (Fig. 1E). In agreement with our in vitro experiments, these in vivo observations do not support a major role of SGLT1 in cardiac glucose handling.

GLUT Transporters in SGLT1 ^Δex1KO Mice

Lack of SGLT1’s effect on glucose transport could be due to a compensatory mechanism favoring facilitated glucose uptake in absence of SGLT1. To test this hypothesis, we analyzed GLUT transporters at three levels: expression, activity, and regulation. Initial analysis of cardiac GLUT1 and GLUT4 expression at the mRNA and protein levels revealed no significant differences between SGLT1 WT and ^Δex1KO mice (Fig. 2, A–D). Then, we investigated the contribution of GLUT4 to glucose uptake in SGLT1 WT and ^Δex1KO cardiomyocytes. To abolish GLUT4 contribution, we measured insulin- and oligomycin-stimulated [2-³H]glucose uptake in the presence of indinavir, a specific GLUT4 inhibitor (33). Insulin- and oligomycin-stimulated glucose uptake are due primarily to recruitment of GLUT4 to plasma membrane (14, 34, 35). An excess of GLUT4 translocation could potentially mask SGLT1 contribution in glucose uptake. Under basal conditions, treatment with 50 µM indinavir had no effect on either SGLT1 WT or ^Δex1KO cardiomyocytes (Fig. 1C). Moreover, insulin- and oligomycin-stimulated glucose uptakes were abolished to a similar extent, using indinavir in cardiomyocytes from both genotypes (Fig. 1C). These results indicate that GLUT4 does not participate in compensatory glucose uptake in SGLT1 ^Δex1KO cardiomyocytes.

Figure 2.

Glucose transporter (GLUT) expression and signaling in hearts from sodium glucose cotransporter 1 (SGLT1) wild-type (WT) and knockout (^Δex1KO) mice. GLUT1 (A) and GLUT4 (B) mRNA were determined in SGLT1 WT (n = 8) and ^Δex1KO (n = 8) total hearts by RT-quantitative (q)PCR. Protein contents of GLUT1 (C) and GLUT4 (D) were assessed by immunoblotting in SGLT1 WT (n = 8) and ^Δex1KO (n = 7–8) total heart lysates. Immunoblots in C and D present 2 examples of both groups. Phosphorylation of Akt/protein kinase B (PKB) (E) and S6 (F) was analyzed by immunoblotting in SGLT1 WT (n = 3) and ^Δex1KO (n = 3) cardiomyocytes treated with hyperglycemia (2 h, 21 mM) or insulin (30 min, 3 nM). G: AMP-activated protein kinase (AMPK) phosphorylation was assessed by immunoblotting on SGLT1 WT (n = 6) and ^Δex1KO (n = 6) cardiomyocytes treated with oligomycin (1 h, 1 µM). Data were expressed as means ± SE. Statistical analysis was performed via a t test (A–D) or 1-way ANOVA (E–G). *Values statistically different from corresponding control, P < 0.05.

Finally, SGLT1 could influence signaling pathways controlling GLUT4 translocation and, particularly, insulin and AMPK signaling pathways. To test this hypothesis, we analyzed phosphorylation of two downstream insulin-signaling targets, namely Akt/protein kinase B (PKB) and S6. We observed no difference in Akt/PKB and S6 phosphorylation under basal, HG, or insulin conditions in either SGLT1 ^Δex1KO or WT cardiomyocytes (Fig. 2, E and F). In addition, we assessed the impact of SGLT1 on AMPK activation following oligomycin stimulation. Immunoblotting for phospho-AMPK did not exhibit any changes in AMPK activation in cardiomyocytes isolated from SGLT1 ^Δex1KO mice compared with WT cardiomyocytes (Fig. 2G). Altogether, these results suggest that there is no compensation by GLUT expression, function, or regulation in SGLT1 ^Δex1KO mice.

Impact of Phlorizin on Glucose Uptake in SGLT1 ^Δex1KO Mice

The role of cardiac SGLT1 in physiological and pathological conditions has been investigated using phlorizin. However, concerns remain with respect to potential phlorizin off-target effects on GLUT-dependent sugar transport. To address this issue, we first evaluated the impact of increasing phlorizin concentrations on glucose uptake in isolated mouse cardiomyocytes; 10 µM and 100 µM phlorizin scarcely affected basal cardiomyocyte [2-³H]glucose uptake (Fig. 3A), whereas 1 mM of phlorizin significantly reduced it. More importantly, phlorizin inhibited insulin-stimulated cardiomyocyte glucose uptake in a dose-dependent fashion, with insulin response being blunted at 1 mM (Fig. 3A). To evaluate the role of SGLT1 on phlorizin effects, these experiments were additionally conducted on SGLT1 ^Δex1KO cardiomyocytes. In both basal and insulin-stimulated states, phlorizin response was similar between the two genotypes (Fig. 3A). These results were confirmed in vivo by measuring cardiac ¹⁸F-FDG uptake in SGLT1 WT and ^Δex1KO mice via a micro-PET scan. In presence of insulin stimulation, we observed an inhibition of cardiac glucose uptake from WT mice when a standard concentration (400 mg/kg) of phlorizin was employed (Fig. 3B). Interestingly, this inhibitory effect of phlorizin on insulin-stimulated cardiac glucose transport was similarly present in SGLT1 ^Δex1KO mice (Fig. 3B). Two hours after insulin injection, glucose level in the blood was similar despite the genotype or phlorizin treatment, although hypoglycemia tended to be more pronounced in presence of phlorizin (Fig. 3B). It is worthy to notice that phlorizin treatment per se induced a significant decrease in glycemia, which was identical in SGLT1 WT and ^Δex1KO animals (for SGLT1 WT: control, 141 ± 5 mg/dL without phlorizin, 81 ± 4 mg/dL with phlorizin, P < 0.05; for SGLT1 ^Δex1KO: control, 121 ± 8 mg/dL without phlorizin, 71 ± 7 mg/dL with phlorizin, P < 0.05).

Figure 3.

Impact of phlorizin on cardiac glucose uptake in vitro and in vivo on sodium glucose cotransporter 1 (SGLT1) wild-type (WT) and knockout (^Δex1KO) animals. A: influence of increasing concentrations of phlorizin on basal glucose uptake was measured in cardiomyocytes isolated from SGLT1 WT (n ≥ 7) and ^Δex1KO (n ≥ 6) mice, using [2-³H]glucose as a radiotracer. Increasing concentrations of phlorizin were added to insulin stimulation, and glucose uptake was assessed with [2-³H]glucose in cardiomyocytes isolated from SGLT1 WT (n = 7) and ^Δex1KO (n = 6). A comparison of control and insulin-stimulated (30 min, 3 nM) conditions was performed at 3 concentrations of phlorizin (45 min): 10 µM, 100 µM, and 1 mM. Data are expressed as means ± SE. Statistical analysis was carried out via a 2-way ANOVA. *Values statistically different from control, without insulin, P < 0.05. $Values statistically different from insulin, without phlorizin, P < 0.05. B: in vivo cardiac glucose uptake (SUV_max) was measured by a micro-positron emission tomography (PET) scan in SGLT1 WT (n = 5) and ^Δex1KO (n = 4) animals treated with insulin (0.5 U/kg) along with phlorizin (400 mg/kg). Glycemia, measured at baseline and at the end of the experiment 2 h after insulin stimulation, is presented below the corresponding pictures. Data are expressed as means ± SE. Statistical analysis was carried out via a 2-way ANOVA. *Values statistically different from corresponding control, insulin without phlorizin, P < 0.05; #values statistically different from baseline, P < 0.05.

Next, we ruled out the possibility that high phlorizin concentrations do block the insulin-signaling pathway or GLUT4 translocation to plasma membrane. We assessed the effect of 1 mM phlorizin on insulin-induced Akt/PKB phosphorylation (on Ser⁴⁷³) and S6 phosphorylation in isolated mouse cardiomyocytes in culture and GLUT4 translocation using a model overexpressing HA-GLUT4-GFP fusion protein, as described previously (14). Rather than reduce it, phlorizin actually enhanced insulin-induced Akt/PKB and S6 phosphorylation (Fig. 4, A and B). Moreover, phlorizin’s effect on the insulin pathway was similar in SGLT1 ^Δex1KO cardiomyocytes (Fig. 4, A and B). We also analyzed Akt/PKB phosphorylation in vivo on hearts from SGLT1 WT mice treated with phlorizin and/or insulin. In accordance to our in vitro data, we showed no inhibition and even a trend toward an enhanced insulin signaling in the presence of phlorizin (Fig. 4C). In agreement with Akt/PKB phosphorylation data, we observed that phlorizin tends to further favor GLUT4 translocation to plasma membrane in response to insulin (Fig. 5). Taken together, these results demonstrate that phlorizin directly interferes with GLUT activity so as to reduce glucose transport without affecting its translocation.

Figure 4.

Impact of phlorizin at higher concentrations on insulin signaling. The effect of 1 mM phlorizin on Akt/protein kinase B (PKB; A) and S6 (B) phosphorylation was evaluated by immunoblotting in sodium glucose cotransporter 1 (SGLT1) wild-type (WT; n = 5–6) and knockout (^Δex1KO; n = 5–6) cardiomyocytes. C: Akt/PKB phosphorylation was assessed by immunoblotting in hearts from SGLT1 WT (n = 5) mice treated with insulin (0.5 U/kg) along with phlorizin (400 mg/kg). Data are expressed as means ± SE. Statistical analysis was carried out via a 2-way ANOVA. *Values statistically different from corresponding control, without treatment, P < 0.05; #values statistically different from corresponding control, phlorizin without insulin, P < 0.05; ^$values statistically different from insulin without phlorizin, P < 0.05.

Figure 7.

SLC5A1 transcripts found in human hearts. A and B: representation of SLC5A1 variants in human. Variant A (A) and variant B (B) amplifications were performed in human hearts (n=5) with the help of specific primers deciphering between the 2 isoforms. C and D: data from RNA sequencing (RNA-Seq) of human hearts from 3 independent studies. Study 1 (17), study 2 (18), and study 3 (19) were aligned with sequence of SLC5A1 variant A (C) or variant B (D). Red dotted boxes discriminate specific sequences of the 2 variants. One representative heart of each study is represented.

Figure 5.

Impact of phlorizin at higher concentrations on glucose transporter 4 (GLUT4) translocation. GLUT4 translocation was analyzed in adult rat cardiomyocytes (n = 4) treated with 1 mM phlorizin (45 min) or 3 nM insulin (30 min). The scale of these pictures represents 50 µm. Data are expressed as means ± SE. Statistical analysis was performed via a 2-way ANOVA. #Values statistically different from control, phlorizin without insulin, P < 0.05. GFP, green fluorescent protein; HA, hemagglutinin.

Based on these data, we assume that high phlorizin concentrations affect cardiac glucose transport through a mechanism that is unrelated to SGLT1.

Transcript Variant of slc5a1 in Mouse Hearts

The lack of cardiac phenotype observed in SGLT1 ^Δex1KO mice, without any impact on cardiac glucose uptake, on cardiac glycogen content, or on phlorizin inhibitory effect, could be considered as intriguing. Therefore, we further investigated SGLT1 expression in our experimental model. As expected (9), we confirmed the deletion of exon 1 and part of the promotor in heart genomic DNA (Fig. 6A). We next analyzed SGLT1 deletion at the mRNA level (Fig. 6B). As expected, the lack of exon 1 was confirmed in intestine SGLT1 mRNA in ^Δex1KO mice. By contrast to the intestine, no expression of exon 1-containing transcripts of slc5a1 could be detected in SGLT1 WT and ^Δex1KO hearts (Fig. 6C). Puzzled by such results, we further characterized cardiac SGLT1 mRNA via a systematic PCR amplification of each exon. Interestingly, PCR results demonstrated the presence of a cardiac-truncated SGLT1 mRNA containing exon 9 to exon 15 in both SGLT1 WT and ^Δex1KO hearts as well as C57Bl/6N isolated cardiomyocytes (Fig. 6, C–E and Supplemental Fig. S1; dx.doi.org/10.17504/protocols.io.bqr7mv9n). Those transcripts were quantified via quantitative PCR (qPCR), showing no differences between WT and ^Δex1KO (data not shown).

Figure 6.

Mouse slc5a1 transcripts in wild-type (WT) and knockout (^Δex1KO) hearts. A: genomic DNA has been extracted from sodium glucose cotransporter 1 (SGLT1) WT (n=5) and ^Δex1KO (n=5) hearts, and genotyping was performed as described elsewhere (19). B: SGLT1 mRNA exons and coding sequence (CDS) are represented. C–E: amplifications of exon 1 to exon 15 (C), exon 8 to exon 15 (D), and exon 9 to exon 15 (E) were performed by PCR in SGLT1 WT (n=3) and ^Δex1KO (n=3) hearts and intestines as well as cardiomyocytes isolated from C57Bl/6N (n=5). Results of amplification of the other exons of SGLT1 are resumed in Supplemental Fig. S1. F: data from RNA sequencing (RNA-Seq) on SMIT1 WT and KO cardiac mRNA were aligned with SGLT1 sequence. Each dark gray line represents a read matching SGLT1 sequence, and vertical dotted lines highlight the beginning of exon 9. Three representative mice are represented.

With the help of RNA sequencing (RNA-Seq) data obtained from hearts of control mice, we performed a coverage analysis of SGLT1 mRNA sequence. We found that cardiac SGLT1 mRNA expression was low since reads matching SGLT1 sequence were rare. Moreover, we observed reads covering SGLT1 mRNA sequence starting from ∼1,200 bp, corresponding to exon 9 (Fig. 6F). These data confirmed the presence of a transcript variant for slc5a1 in mouse hearts that lacks a large 5′ portion (exons 1 to 8). To our knowledge, it is the first report of a transcript variant of slc5a1 in murine heart.

Expression of Variant B, the Shortest SLC5A1 Transcript Variant in Human Hearts

Our discovery of a truncated SGLT1 mRNA in murine hearts questioned cardiac SGLT1 mRNA expression in other species, particularly in humans. As stated in the introduction section, two variants of SLC5A1 have been detected in human tissues. Variant A was not expressed in the heart, whereas variant B of SLC5A1 could be detected and amplified (Fig. 7, A and B). To confirm our results, we carried out an in silico analysis of RNA-Seq data from human hearts of three different studies that are publicly available in GEO accession database (17–19). RNA-Seq data were aligned with the known sequences of variant A or variant B. These alignments, as well as a representation of SLC5A1 variants in humans, are shown in Fig. 7, C and D. We observed specific RNA seq reads for variant B and not for variant A (Fig. 7, C and D). Altogether, these data show for the first time that human heart expresses SLC5A1 transcript variant B, encoding a shorter protein that loses three transmembrane domains as well as residues involved in glucose and sodium bindings. Further investigations are needed to understand the role and potential activity of this protein isoform in the human heart.

DISCUSSION

To date, no clear function of SGLT1 has been identified within the heart. Previous studies, including from our group, have shown that SGLT1 and SMIT1 are expressed in human, mouse, and rat hearts (3–5). The main findings of this study are that no significant changes in murine cardiac glucose handling are observed in the classical genetic model of SGLT1 invalidation and that high phlorizin concentrations lower cardiac glucose uptake independently of SGLT1. Importantly, we showed that murine and human hearts express slc5a1 alternative transcripts encoding for truncated proteins unlikely to be involved in glucose transport (Fig. 8).

Figure 8.

Putative sodium glucose cotransporter 1 (SGLT1) proteins expressed in the heart. A: schematic structure of SGLT1 protein is represented, and main residues involved in glucose and sodium bindings are highlighted in red, adapted from Ref. 2. B and C: representation of the hypothetical truncated SGLT1 proteins expressed in mouse (B) and human (C) hearts.

In contrast to GLUT-dependent facilitated glucose transport, SGLTs employ the sodium gradient maintained by the Na⁺/K⁺-ATPase (2, 26) to actively uptake glucose. SGLTs exist under several isoforms that all differ in their affinity toward glucose (Barbeau, no. 148) and alternative monosaccharides (2, 36). Several studies highlighted monosaccharide structural requirements, including a pyranose ring in d-configuration, to be transported through SGLTs. d-Glucose is recognized by SGLTs through its C2-hydroxyl group and C5-methyl or substituted methyl group (37–39). Therefore, SGLT-dependent glucose transport theoretically cannot be estimated by measuring intracellular accumulation of radiolabeled glucose analogs such as 2DG or 2FDG, because these tracers lack the structural specificities mentioned above. Thus, it is not surprising that no differences could be observed in 2DG or 2FDG uptake between SGLT1 WT and ^Δex1KO mice in vitro and in vivo. Furthermore, we obtained similar results with a [2-³H]glucose radiotracer transported by SGLTs, suggesting that SGLT1 contribution to cardiac glucose uptake is, at best, marginal. In addition, it should be noted that SGLT1 ^Δex1KO does not affect facilitated glucose transport.

When interpreting the passive or active nature of glucose transport, a warning applies to the specificity of inhibitors used to discriminate glucose transporters, typically phloretin (GLUT inhibitor) and phlorizin (SGLT inhibitor). Phlorizin is considered a SGLT1 inhibitor, although it actually inhibits most SGLT isoforms with different affinities (K_i ≈ 140 nM for SGLT1, K_i ≈ 11 nM for SGLT2, K_i ≈ 120µM for SGLT3, K_i ≈ 10µM for SGLT4, K_i ≈ 1.7 µM for SGLT5, K_i ≈ 76µM for SGLT6, K_i ≈ 60µM for SMIT1) (40–44). Phlorizin barely affects glucose handling at concentrations close to its K_i for SGLT1. However, our data clearly indicate that 1 mM of phlorizin can resume phloretin effects on glucose uptake and, therefore, behave as an inhibitor of facilitated transport. Moreover, our data using 2FDG in PET scan further confirm the indirect effect of 400 mg/kg phlorizin on GLUT transporters since SGLTs do not transport 2FDG. These essential results question the role of SGLT1 demonstrated via a pharmacological approach with high phlorizin doses. Indeed, a previous study has shown that SGLT1 mediates insulin- and leptin-induced glucose uptake in mouse heart. The use of high in vivo concentrations of phlorizin (400 mg/kg) and 2-deoxy-d-[¹⁴C]-glucose as glucose tracer indicate that these results rather reflect an unspecific effect of phlorizin on GLUT transporters. Because phlorizin differs from phloretin by its β-d-glucopyranosyl residue attached at position 2, along with a tendency to be hydrolyzed to its aglycone form (45), a high phlorizin concentration can alter facilitated transport either directly or through its conversion into phloretin.

SGLT1 expression in mouse heart remains controversial. Of importance, we report for the first time the existence of a slc5a1 transcript variant in mouse heart. Indeed, cardiac SGLT1 mRNA starts from exon 9 to exon 15, making it undetectable with primers targeting exons 1 to 8. One may wonder whether this particular expression could not be specific to our mouse breeding lineage. The absence of full length has been recapitulated in all strains that we have studied, ruling out such hypothesis. Our discovery explains apparent discrepancies regarding SGLT1 mRNA detection in the heart. Madunić et al. (46) have observed no cardiac SGLT1 mRNA using primers amplifying exons 1 to 4 of slc5a1 mRNA, although most other studies have used primers specific for the 3′ extremity (3, 23, 30, 47), which is present in mouse heart. The slc5a1 cardiac transcript variant is detectable in both WT and SGLT1 ^Δex1KO hearts. The expected mouse cardiac SGLT1 isoform is a shorter protein, lacking a large NH²-terminal extremity that contains residues involved in glucose and sodium bindings as well as eight transmembrane domains (2, 6, 7), all required for glucose transport (Fig. 8, A and B). Accordingly, a role of SGLT1 in cardiac glucose uptake is unlikely. In line with our data, a recent study involving transgenic mice with cardiomyocyte-specific RNA interference knockdown of SGLT1 (TG^SGLT1-DOWN) showed that SGLT1 reduction barely affects ex vivo cardiac glucose consumption, although glucose uptake was not directly measured using radioactive tracers (30). One may notice that the SGLT1 siRNA sequence they used targets the mouse cardiac variant of slc5a1 and potentially knocks down all putative cardiac SGLT1 proteins. Although there were no differences between the WT and ^Δex1KO in terms of slc5a1 transcript in the heart, we have much evidence still supporting the lack of SGLT1-dependent glucose transport. Indeed, both genotypes show 1) similar responses to different glucose tracers, 2) similar impact of indinavir, and 3) absence of key regions for glucose binding and transport in the sequence expressed in the heart.

Under basal state, SGLT1 ^Δex1KO and TG^SGLT1-DOWN mice exhibit no specific cardiac phenotype (5, 30). However, a role of cardiac SGLT1 has been advocated in several pathological conditions. First, SGLT1 has been implicated in the pathophysiology of cardiomyopathy linked to PRKAG2 (gene expressing AMPK regulatory subunit) mutation. It can mediate the increase in glycogen storage induced by the γ₂-subunit mutation of AMPK. SGLT1 knockdown was shown to decrease cardiac glycogen content and improve cardiac function in a model of PRKAG2 cardiomyopathy (47). Second, SGLT1’s absence has been shown to counteract transaortic constriction-induced cardiac hypertrophy, resulting in well-preserved left ventricular function under pressure overload (23). Finally, other lines of evidence suggest a contribution of SGLT1 in cardiac ischemia-reperfusion injury. Indeed, SGLT1 expression is increased following coronary ligation in mice (3, 30), and SGLT1 knockdown protects the heart from ischemia-reperfusion injury by reducing protein kinase C (PKC), NADPH oxidase (NOX2) activation, and reactive oxygen species production (30). These results seem to be in conflict with the observations of Kashiwagi et al. (4), who showed that SGLT inhibition using phlorizin is deleterious upon ischemia-reperfusion. These contradictory results could be explained by phlorizin concentrations used that impact GLUT-mediated glucose uptake, indicating that phlorizin should be employed with caution. In all of these pathological situations, mechanisms beyond cardiac SGLT1 remain elusive and need further investigation. One may hypothesize that cardiac SGLT1, even truncated, modifies signaling pathways involved in pathophysiology.

Finally, our data also demonstrate the presence of an alternative transcript of SLC5A1 in the human heart. Several studies have shown a significant cardiac expression of SGLT1 in humans (3, 4, 48–52), but we are the first to specify that transcript variant B is the one expressed, at least in all analyzed samples, leading to the expression of a shorter protein lacking residues implicated in glucose and sodium bindings as well as three transmembrane domains (Fig. 8C). This cardio-specific variant of SGLT1 might exhibit impaired glucose affinity and transport capacity, as well as sensitivity to inhibitors, compared with other organs (2, 6, 7).

Interestingly, SGLT2 inhibitors, widely used for type 2 diabetes mellitus treatment, have shown cardioprotective properties independently of their blood glucose-lowering effect (53–55). However, several studies have highlighted the lack of SGLT2 expression in the heart (5, 48, 49, 51, 56), questioning the possibility of an indirect effect of these inhibitors on cardiac function. Therefore, the beneficial cardiovascular effect of SGLT2 inhibitors is intriguing and currently being investigated (57). The affinity of these inhibitors for SGLT1 cardiac isoform remains to be defined. Moreover, given SGLT1 functions in glucose renal reabsorption and intestinal absorption (2), dual SGLT1/SGLT2 inhibitors have been recently developed in the context of diabetes treatment (58, 59), which also conferred the ability to dramatically reduce risk of cardiovascular death or hospitalization for heart failure in diabetic patients (60). Long-term results will help determine the potential impact of SGLT1 inhibition on various organs in which it is expressed, including the heart. Understanding the role of SGLT1 outside intestine or kidney is of clinical relevance in this context (61).

Several limitations of our study have to be pointed out. Although our cardiac glycogen measurements showed no changes in SGLT1 WT and ^Δex1KO animals at the basal state, the latter has not been verified in stimulated conditions and pathology. Moreover, technical issues concerning GLUT4 translocation to the plasma membrane in SGLT1 ^Δex1KO model would have to be addressed to rule out an effect of SGLT1 in this process. Further investigation will be conducted to decipher SGLT1 implication in glycogen and glucose homeostasis in stimulated states and pathology in the heart. Finally, we were not able to assess the impact of the cardiac SGLT1 mRNA variant on its protein expression due to the lack of specific available antibodies. New tools and further investigations are needed to address SGLT1 protein expression in the heart in an appropriate manner.

CONCLUSIONS

In this study, we have demonstrated that SGLT1 ^Δex1KO murine models exhibit normal glucose handling and cardiac glucose transport in isolated cardiomyocytes or in vivo. This could be explained by the expression of a SGLT1 mRNA lacking exons 1 to 8, leading to a truncated protein in mouse hearts. A different slc5a1 transcript variant is also expressed in human heart. Further investigations are required to elucidate the physiological relevance of this transporter within the heart. Another essential study conclusion is that using high doses of the SGLT inhibitor phlorizin exerts a drastic effect on cardiac glucose transport independent of SGLT1. Therefore, extreme caution appears to be warranted when employing this inhibitor to investigate SGLTs or, more specifically, the function of SGLT1.

GRANTS

This work was funded by grants from the Fonds National de la Recherche Scientifique (FNRS) et Médicale, Belgium (T.0241.16, 2016-2018, T.0011.19, 2019-2020), and Action de Recherche Concertée de la Communauté Wallonie-Bruxelles, Belgium (ARC 18/23–094).

L.F. was supported by the Fund for Scientific Research in Industry and Agriculture (FRIA); S.H. is the research associate, and L.B. is the director of research of FNRS, Belgium.

This work was supported by unrestricted grants form Astra Zeneca.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

L.F., L.D., S.H., L.B., C.B., A.V. and H.K. conceived and designed research; L.F., C.B., A.M., S.B., L.B., A.V., A.B., J.C. and A.G. performed experiments; L.F., L.D., L.H., S.H., L.B., C.B., S.B. and A.B. analyzed data; L.F., L.H., S.H., L.B., C.B., S.B. and A.B. interpreted results of experiments; L.F., C.B., A.M., L.B. and A.B. prepared figures; L.F., C.B. and S.B. drafted manuscript; L.F., L.H., S.H., L.B., C.B., A.M., L.B. and H.K. edited and revised manuscript; L.F., L.D., L.H., S.H., L.B., C.B., A.M., S.B., L.B., A.V., A.B., J.C., A.G. and H.K. approved final version of manuscript.