Background
Sodium–glucose linked cotransporter-2 inhibitors (SGLT2i) increase urinary glucose excretion thereby off-loading energy, lowering plasma glucose and body weight as well as inducing a modest diuresis that reduces extracellular fluid volume and lowers blood pressure. In addition, two recent cardiovascular outcome studies with these agents demonstrated reductions in heart failure hospitalisation suggesting that this class of drug exerts cardioprotective effects beyond glucose and blood pressure lowering [
1,
2].
In addition to SGLT2, another glucose transporter, SGLT1, cotransports sodium and glucose in the distal segment of the proximal tubule with higher affinity and lower capacity than its more proximal counterpart. Unlike SGLT2, whose expression is confined to the kidney, SGLT1 is also abundantly present in the gut where it accounts for much of glucose (and galactose) absorption by enterocytes. Accordingly, combined SGLT1/2 inhibition offers the potential to not only increase glucosuria beyond that seen with SGLT2 inhibition alone but to reduce glucose absorption allowing the monosaccharide to stimulate release of glucagon-like peptide-1 (GLP-1) in the ileum. As such, dual SGLT1/2 inhibition presents an appealing strategy for glucose lowering in diabetes with drugs that do so currently in development [
3].
Beyond its expression in the kidney and small intestine, however, SGLT1 is also present in various other organs in humans including the lung, liver, pancreatic alpha cells, skeletal muscle and particularly in the heart where its abundance exceeds that of the kidney [
4‐
6]. Similar high expression levels have recently also been reported in the hearts of rats and mice [
7]. With its constant and high energy requirements and little storage capacity, the heart requires a constant supply of energy-generating substrates. This is especially important during ischaemia when the ability to generate more ATP per O
2 molecule consumed and to also generate ATP anaerobically by glycolysis render glucose preferable to fatty acids as a substrate for energy production [
8]. Although the facilitated glucose transporters GLUT1 and GLUT4 were previously thought to account entirely for glucose transport in the heart, more recent studies attest to the additional contribution of SGLT1 [
9‐
12].
Given their potential to enter the clinical arena, we sought to examine the effects of dual SGLT1/2 inhibition in experimental heart disease using the rat coronary artery ligation model that develops ischemia, infarction and heart failure. Accordingly, the primary objective of the study was to assess changes in cardiac function and secondarily changes in structure. To avoid the requirement for longterm parenteral administration with the poorly absorbed and rapidly metabolized phlorizin, we administered the readily absorbed SGLT1/2 dual inhibitor prodrug, T-1095, that is quickly converted to the active moiety, T-1095A following its entry into the systemic circulation [
13,
14].
Discussion
Unlike SGLT2 whose expression is almost exclusively confined to the kidney, SGLT1 is abundantly expressed in the heart. Blocking the SGLT2 transporter alone with dapagliflozin had no effect on cardiac function in either the control setting or after myocardial infarction. Dual blockade of SGLT 1 and 2 with T-1095, however, while not affecting function under physiological circumstances led to an exacerbation of impairment in both systole and diastole in the post myocardial infarction setting.
Ligation of the left anterior descending coronary artery leads to substantial infarction of the left ventricle, providing a well-established model of progressive heart failure in rodents [
15]. As expected, all animals in the current study that had undergone ligation displayed evidence of systolic dysfunction. Notably, the key features of left ventricular function such as ejection fraction and dP/dt
max were both worse in the group that had received the SGLT1/2 inhibitor, T-1095 when compared with those treated with either vehicle or dapagliflozin. Similarly, diastolic function was also abnormal 4 weeks following LAD ligation but worse still among those rats that had received T-1095. Moreover, abnormalities in both the early, active, energy-dependent phase of diastole as indicated by prolongation of the diastolic time constant, Tau and the later, more passive phase of diastole as reflected by an increase in the end-diastolic pressure volume relationship (EDPVR) were also more severely affected in T-1095-treated animals.
Following ischaemic necrosis of the myocardium the heart undergoes adaptive (in addition to adverse) remodelling in an attempt to compensate for the loss of contractile tissue. In this regard, an increase in left ventricular wall thickness as a consequence of myocyte hypertrophy reduces wall stress, as dictated by the Laplace law [
35]. Notably, the requirement for anabolic growth in addition to maintaining ATP generation for cardiac contractility necessitates an increase in glycolytic flux [
36]. In this setting, the role of SGLT1-mediated glucose entry into cardiac myocytes that can be inhibited by agents such as T-1095 becomes increasingly important [
11,
37,
38]. Indeed, inhibition of cardiac hypertrophy in the setting of increased wall stress, as demonstrated in the present study would be predicted to be detrimental [
39] as demonstrated by the cardiac dysfunction and reduced survival when hypertrophy is inhibited in experimental animals subjected to cardiac stress [
40‐
44]. In contrast to the aforementioned primary inhibition of myocyte hypertrophy, ACE inhibitors and β-blockers reduce preload and afterload so that the need for hypertrophy is reduced and reverse remodeling may occur.
Beyond the effects on myocyte hypertrophy, we also observed increased collagen deposition in animals that had received T-1095. As described above, T-1095 mediated inhibition of SGLT1 leads to a diminution in glucose entry that reduces ATP generation, consistent with the reduction in systolic function (EF, dp/dt
max) and the impairment in the active phase of diastolic relaxation (Tau) reported in the present study. In addition to impaired function, however, the resultant energy depletion would also provide a profibrotic stimulus [
45] that would lead to the increase in interstitial collagen.
Phlorizin, the prototypical SGLT1/2 inhibitor is poorly absorbed from the gut so that in addition to a low plasma concentration following oral dosing, its high luminal concentration induces diarrhoea by inhibiting enterocyte SGLT1. To overcome these limitations, we took advantage of T-1095, a phlorizin derivative prodrug that is readily absorbed from small intestine prior to being converted to its active form, T-1095A, in the systemic circulation [
13]. Similar to phlorizin that is a near-equipotent inhibitor of SGLT1 and 2, T-1095A has only a fourfold selectivity for SGLT2 over SGLT1 [
14]. In contrast, the three widely-marketed SGLT2 inhibitors, canagliflozin, dapagliflozin and empagliflozin that have SGLT2:SGLT1 selectivity ratios in the order of > 250, > 1200 and > 2500, respectively [
46] and while canagliflozin may inhibit gut SGLT1 in clinically used doses [
47], its plasma concentration would not be expected to reach levels high enough to inhibit cardiac SGLT1. Beyond canagliflozin, however, drugs with even less selectivity are in development, seeking to capitalise on the theoretical additional glucose lowering that may be achieved by inhibiting SGLT1 in the distal segments of the proximal tubule as well as SGLT2 in the more proximal segments [
48]. Notably, two drugs designed to inhibit SGLT1 have already entered clinical trials. Sotagloflozin (LX4211) with an IC
50 ratio of 20:1 in favour of SGLT2 has completed phase 3 while others such as GSK-1614235 (Glaxo Smith-Kline, PA), designed to preferentially inhibit SGLT1 with an IC
50 that is 300-fold less for SGLT1 versus SGLT2, has recently undergone phase 1 clinical testing [
49].
The present study has several limitations. First and foremost, while a commonly used model, the abrupt induction of ischemic necrosis in a rodent heart has only superficial resemblance to the myocardial infarction in humans that occurs in the setting of longstanding atherosclerotic disease, collateral development and in whom a vast array of cardioprotective measures are now employed. Notably, clinical studies of SGLT2 inhibitors have not shown evidence of benefit in being able to reduce the risk of myocardial infarction. As such, the absence of significant improvement in cardiac parameters with dapagliflozin should not be interpreted as an indication that this compound does not prevent hospitalization for heart failure in the clinical setting especially considering the positive findings of its phase 2/3 program [
50], post-marketing studies [
51] and the effects of other members of the SGLT2 inhibitory class [
52] along with supporting basic and translational research studies [
53‐
55]. Secondly, the compound used to inhibit SGLT1/2, T-1095, is not one that is in current use and while the drug was used as a pharmacological probe, extrapolations to other chemical entities may not be warranted. In particular, the combined SGLT1/2 inhibitor, sotagliflozin, as noted previously, has a 20:1 selectivity in favour of SGLT2, contrasting the 4:1 ratio for T-1095.
Although T-1095A’s ability to inhibit SGLT1-mediated glucose uptake has been well-documented [
13,
14,
16], we did not directly assess glucose transport into the myocardium in the current study. Several other groups have explored the effects of SGLT1 inhibition in heart in cells, atrial strips and using the ex vivo Langendorff technique with T-1095 and phlorizin [
10,
11,
38,
56]. These studies demonstrated the importance of SGLT1-mediated glucose transport in the stress setting whereby its inhibition exacerbates ischemia–reperfusion injury [
38,
56]. The novelty and potential clinical relevance of the current study, however, centers on it being the first to examine the effects of SGLT1 inhibition following cardiac injury in the in vivo setting. Notably, given SGLT-mediated effects on preload and afterload [
57], we used conductance catheterization that unlike echocardiography provides load-independent measurements of cardiac function in intact animals [
25]. While the plasma concentration of T-1095A was not measured in the current study, its effects on SGLT-1 mediated glucose transport can be inferred from the IC
50 for rat SGLT1 of < 1 µM [
14] and the plasma concentration of ~ 10 µM that, according to published pharmacokinetic for the compound, would have been attained with the doses used [
58]. Finally, although a direct effect of T-1095 on cardiac SGLT1 provides a cogent explanation for the observed effects, it is conceivable that altered glucose transport in other cells that express SGLT1 might have contributed, including the cardiac capillary endothelium [
4].
Authors’ contributions
KAC conceived of the study, analyzed the echocardiographic and conductance catheter data and reviewed and approved the final version of the manuscript. KT performed molecular and histological analyses and reviewed and approved the final version of the manuscript. YZ performed histological and statistical analyses and reviewed and approved the final version of the manuscript. J-FD, performed animal surgery, echocardiography and conductance catheterization and reviewed and approved the final version of the manuscript, REG conceived of the study and wrote the manuscript and reviewed. All authors read and approved the final manuscript.