Glycemic control by the SGLT2 inhibitor empagliflozin decreases aortic stiffness, renal resistivity index and kidney injury

Diabetic kidney disease (DKD) is the most common cause of end stage renal disease and the mechanisms underlying its development are complex and still not well understood [1‐3]. In this regard, arterial stiffness is an independent and robust predictor of chronic kidney and cardiovascular diseases [4, 5]. Measurement of aortic pulse wave velocity (PWV) is the reference standard technique for assessment of central aortic stiffness [4‐6]. Although aortic stiffness is considered a disease of aging, stiffening is accelerated during obesity and diabetes, and can be present at earlier ages [7]. Additionally, renal microvasculature stiffening, indicated by increased renal resistivity index (RRI), is associated with albuminuria, thereby suggesting a role for stiffening of both the macro- and micro-vasculature in the development of kidney injury [5, 7‐9]. Indeed, accumulating evidence indicates that increased PWV predicts adverse macrovascular and microvascular outcomes in diabetics with albuminuria and kidney disease [4, 5, 10]. In addition, renal tubulointerstitial fibrosis also contributes to renal microvascular stiffness, underscoring the intricate relationship between renal pathology and microvascular complications [6, 7, 11].

Current drugs targeting progression of DKD include anti-hyperglycemic agents, and inhibitors of inappropriate RAAS (renin–angiotensin–aldosterone system) activation, hypertension and hyperlipidemia [3]. Despite widespread use of these therapeutics, more than 30% of patients with type 2 diabetes mellitus (T2DM) develop DKD [3]. Recently, inhibition of the sodium-glucose co-transporter 2 (SGLT2) in kidneys, a more recent and novel therapeutic approach to lower hyperglycemia, has been shown to slow progression of DKD in preclinical models and clinical studies, either as stand-alone or combination therapy, including DPP-4 inhibitors, GLP-1 receptor antagonist, or ACEi/ARBs [12‐14]; however the mechanisms are not entirely known [15‐21]. SGLT2 inhibitors induce glycosuria by reducing reabsorption of glucose in proximal tubules, and evidence from clinical trials indicates that they also exert cardiovascular protection [17‐26]. Indeed, in a phase III clinical trial, empagliflozin (EMPA), a potent and highly selective SGLT2 inhibitor [27], was shown to reduce blood pressure (BP), but also markers of arterial stiffness [28]. EMPA-mediated improvements in central hemodynamics, vascular function and arterial stiffness were further corroborated in a recent double-blind, randomized, placebo-controlled crossover, interventional study involving 76 T2DM patients [29]. However, its effects on renal fibrosis, the sine qua non of chronic kidney disease, are not fully understood.

Several studies support a role for reversion-inducing-cysteine-rich protein with Kazal motifs (RECK), a membrane-anchored matrix metalloproteinase (MMP) regulator, in suppressing pro-fibrotic responses through inhibition of MMP activation [30, 31]. Although the role of RECK deficiency in promoting angiogenesis [32] and cardiac fibrosis [30] has been reported previously, the influence of low RECK expression on renal injury and fibrosis in diabetes has not been investigated. Therefore, we hypothesized that SGLT2 inhibition by EMPA blunts diabetic kidney injury and fibrosis by suppressing macrovascular/microvascular stiffness and upregulating anti-fibrotic RECK expression in the kidney. We further posited that the renovascular protective effects of EMPA are attributable to improved glycemic control, enhanced activation of endothelial nitric oxide synthase (eNOS), and suppressed oxidative stress. Herein we report that improved glycemic control by EMPA ameliorates kidney injury in female diabetic (db/db) mice by reducing systemic and renal artery stiffness, and restoring RECK expression.

Here we show for the first time that SGLT2 inhibition attenuates diabetes-induced systemic macrovascular and renal microvascular stiffness, as well as kidney injury and fibrosis. The improved glycemic control by EMPA is associated with enhanced renal eNOS activation and suppressed nitrosative stress. Furthermore, EMPA rescued hyperglycemia-induced suppression of the anti-fibrotic factor, RECK, in the diabetic kidney. Collectively, these data suggest that both vascular stiffening and renal fibrosis contribute to progression of diabetic nephropathy which can be mitigated by EMPA (Fig. 6).

The beneficial effect of EMPA on progression of kidney disease in the EMPA-REG OUTCOME TRIAL and pre-clinical studies has been ascribed to local hemodynamic changes, whereby SGLT2 inhibition results in decreased sodium reabsorption with consequent increased sodium delivery to the macula densa, increased afferent arterial tone and decreased estimated glomerular filtration rate (GFR), indicative of reduced hyperfiltration [17, 21, 45, 46]. With EMPA, this initial decrease in GFR stabilizes, and over a period of time, improves compared to control subjects. In addition, the resulting reductions in post-glomerular perfusion would predict an increase in acute kidney injury (when GFR < 60 ml/min), however EMPA suppressed kidney injury [45]. Taken together, the effects unrelated to increased afferent tone might have contributed to the beneficial effects of EMPA in slowing progression of renal injury in diabetic subjects. In this regard, hyperfiltration can be affected by worsening endothelial dysfunction and aortic stiffness [47]. We hypothesized that suppression of macrovascular and microvascular stiffness may also be contributing to amelioration of microalbuminuria due to effective glycemic control by EMPA. The improvement of vascular stiffness, along with improved endothelial function by EMPA in this study may contribute, in part, to hyperfiltration, as well as, direct effects on vascular function in improving microalbuminuria (Fig. 6).

A number of clinical studies support the existence of relationships between increased aortic stiffness, albuminuria, and kidney injury [7‐9, 48]. However, the mechanisms behind these interactions are not well understood. In this regard, aortic stiffening has been shown to propagate excessive pulsatile (kinetic) energy into peripheral organs, such as the kidney, where there is high flow but low precapillary resistance [5, 7, 8, 11]. In aging humans, increases in resistivity and pulsatility in the kidney vasculature have been linked to reductions in glomerular filtration rate, albuminuria and chronic kidney disease [7]. However, the application of resistivity and pulsatility indexes to the exploration of mechanisms of renal vascular dysfunction in preclinical models has been limited [7, 9]. In this study, we utilized a novel non-invasive ultrasound technique to evaluate indices of kidney microvascular health (RRI and PI) in a preclinical model of diabetic nephropathy, as has been previously reported [49].

Stiffening of the aorta and microvasculature has been linked to kidney parenchymal injury and fibrosis [50, 51]. In this study, we observed significant increases in periarterial and tubulointerstitial fibrosis with concomitant increases in RRI and PI. Importantly, EMPA abrogated these structural and functional abnormalities. These findings support the notion that increases in cortical tissue periarterial and tubulointerstitial fibrosis, along with aortic stiffness, contribute to renal vascular stiffness and albuminuria in diabetes. One of the pathways that regulates kidney fibrosis is impairment in eNOS-dependent NO signaling [52]. In fact, we observed differential phosphorylation of eNOS (reduced S1177 phosphorylation and increased T495 phosphorylation) favoring its inactivation, and presumably reduced bioavailable NO, in diabetic kidneys. Notably, EMPA treatment restored eNOS activation, possibly contributing to reduced kidney injury and improved renal vascular function. In support of this notion, it has been previously shown that the reno-protective effects of EMPA were lost in type 1 diabetic mice deficient in eNOS [53]. Of note, hyperglycemia and increased AGE levels have also been shown to reduce bioavailable NO [42, 54]. In the present study, we showed that diabetic mice exhibit increased AGE in kidney cortical tissue along with reductions in eNOS activation and increases in periarteriolar and tubulointerstitial fibrosis. Therefore, effective control of hyperglycemia resulting in suppression of AGE expression, decreased nitrosative stress and increased eNOS activation collectively may contribute to improvement in kidney structure and function.

RECK is a membrane-anchored protein that suppresses a pro-fibrotic response through inhibition of MMP activation [30, 31]. However, its role in diabetes-induced renal injury and fibrosis has not been investigated. We previously reported that activation of the RAAS induces cardiac fibrosis with associated suppression in RECK expression in the heart [30]. However, it is not known whether hyperglycemia alone suppresses RECK expression in the kidney. Our data show that normal kidneys express high levels of RECK (Fig. 5a) that is predominantly localized to cortical proximal tubule cells and to a lesser extent in glomerular ECs. Notably, RECK expression was significantly downregulated in diabetic kidneys, but partially restored by EMPA. Supporting these in vivo studies, high glucose suppressed RECK expression in cultured proximal tubule epithelial cells and this suppression was prevented by EMPA. To our knowledge, this is the first report describing RECK downregulation in diabetic kidneys and its rescue by EMPA. Our future studies will determine the molecular mechanisms underlying RECK regulation by hyperglycemia/EMPA.

SGLT2 inhibition improves albuminuria in humans [55] and in preclinical models of diabetes and kidney fibrosis [56, 57]. Similarly, we observed decreased albuminuria and kidney fibrosis in EMPA-treated female diabetic mice. However, in a comparable study using a similar dose of EMPA, no improvement in albuminuria was observed in male db/db mice of similar age, despite similar histopathological changes in the kidney [58]. Interestingly, the magnitude of glycemic control needed to reduce albuminuria appears to be a glucose concentration of 15 mmol/l or less [21]. In the present study, we observed improvement in albuminuria in concert with blood glucose levels below 15 mmol/l. In the other study that employed male mice [58], blood glucose levels remained above 15 mmol/l with no improvement in albuminuria. Thus, these comparable studies performed in male and female db/db mice offer insights into the notion that the magnitude of glycemic control may be an underlying factor for the responsiveness of kidneys to progressive renal injury by SGLT2 inhibition [19].

Aortic and renal vascular stiffness are commonly associated with high BP, and EMPA has been shown to decrease BP and arterial stiffness in humans [28, 29]. However, in our model, reductions in aortic and microvascular stiffness by EMPA were not associated with improvements in BP [40], indicating that additional mechanisms, independent of BP reduction, might have contributed to the reno-protective effects of EMPA. Indeed, our BP results are consistent with one other study reporting no differences in BP following 10 weeks of EMPA treatment in db/db mice [59]. In this regard, our data suggest that reduced EC stiffening and enhanced eNOS activation might have contributed to the beneficial vascular effects of EMPA. In fact, we and others have previously demonstrated an association between impaired vascular NO signaling and arterial stiffening in a diet-induced obesity model, independent of changes in BP [60, 61].

It is noteworthy that despite improved glycemia, EMPA was not associated with improvement in plasma lipids, fat pad mass or body weight. As far as we are aware, EMPA is not reported to lower total cholesterol, free fatty acids or body weight in rodents [33, 56, 58], including db/db mice [58]. However, though EMPA treatment is associated with weight loss in diabetic patients, it either elevated or had no effect on serum lipids in humans (versus placebo) [62‐64]. It has also been suggested that there may be species-dependent effects of SGLT2 inhibition on body weight, and that other factors such as activity level and food intake, might contribute to divergent responses in these outcomes [33].

A limitation of the present study is that we did not examine the direct effects of EMPA treatment on cultured ECs or vascular rings. One recent study demonstrated pleiotropic effects of EMPA in cultured ECs [56], while another suggested that EMPA does not affect vasoreactivity of aortic rings from mice or rats [65]. Interestingly, a more recent investigation indicated a robust increase in SGLT1 and SGLT2 expression in rat aortas, as well as high glucose-treated porcine ECs [66]. Thus, it is possible that glycemic control by EMPA, via inhibition of SGLT2 in kidney proximal tubules, as well as in the vascular wall, contributes to improvement in vascular stiffness and kidney function. Further work is needed to elucidate the role of endothelial SGLT2 expression in modulating vascular health. Although recent studies show renoprotection with other SGLT2 inhibitors, suggesting possible class effects [67, 68], the individual underlying drug effects and their specific mechanisms remain to be determined. Further, completion of several ongoing clinical trials, including, EMPEROR-Reduced, EMPEROR-Preserved, DAPA-HF, DAPA-DKD and CREDENCE, will shed insights into the CV and renoprotective effects of SGLT2 inhibitors [68].

In summary, we demonstrate an association between aortic stiffness, endothelial dysfunction, renal vascular stiffness and kidney injury in diabetic mice, that is improved with SGLT2 inhibition by EMPA, independent of changes in BP or body weight. We also report evidence of RECK deficiency in diabetic kidneys, and its rescue by EMPA. Together, these data indicate that glycemic control by EMPA exerts vascular- and reno-protective effects in diabetic mice, resulting in reduced kidney fibrosis and injury.