State-of-the-Art-Review: Mechanisms of Action of SGLT2 Inhibitors and Clinical Implications

Abstract

BACKGROUND

Inhibitors of the Na⁺-coupled glucose transporter SGLT2 (SGLT2i) primarily shift the reabsorption of large amounts of glucose from the kidney’s early proximal tubule to downstream tubular segments expressing SGLT1, and the non-reabsorbed glucose is spilled into the urine together with some osmotic diuresis. How can this protect the kidneys and heart from failing as observed in individuals with and without type 2 diabetes?

GOAL

Mediation analyses identified clinical phenotypes of SGLT2i associated with improved kidney and heart outcome, including a reduction of plasma volume or increase in hematocrit, and lowering of serum urate levels and albuminuria. This review outlines how primary effects of SGLT2i on the early proximal tubule can explain these phenotypes.

RESULTS

The physiology of tubule-glomerular communication provides the basis for acute lowering of GFR and glomerular capillary pressure, which contributes to lowering of albuminuria but also to long term preservation of GFR, at least in part by reducing kidney cortex oxygen demand. Functional co-regulation of SGLT2 with other sodium and metabolite transporters in the early proximal tubule explains why SGLT2i initially excrete more sodium than expected and are uricosuric, thereby reducing plasma volume and serum urate. Inhibition of SGLT2 reduces early proximal tubule gluco-toxicity and by shifting transport downstream may simulate “systemic hypoxia”, and the resulting increase in erythropoiesis, together with the osmotic diuresis, enhances hematocrit and improves blood oxygen delivery. Cardio-renal protection by SGLT2i is also provided by a fasting-like and insulin-sparing metabolic phenotype and, potentially, by off-target effects on the heart and microbiotic formation of uremic toxins.

The sodium-dependent glucose transporter SGLT2 in the early proximal tubule of the kidney is the primary mechanism for the reabsorption of glucose filtered by the kidney glomeruli (see ref. ¹ for review). As a consequence, inhibitors of SGLT2 (SGLT2i) spill glucose into the urine. This formed the primary rationale for the development of these compounds as blood glucose-lowering drugs in individuals with type 2 diabetes mellitus (T2DM). However, large placebo-controlled trials demonstrated that SGLT2i not only improve blood glucose control but, as a class effect, can protect the kidneys and heart from failing. When given on top of standard of care (including renin–angiotensin system blockade) in people at high cardiovascular (CV) risk with T2DM or chronic kidney disease (CKD), these drugs safely reduce the risk of acute kidney injury by ~25% and of kidney disease progression by ~40%.² Notably, similar kidney protective effects were reported for SGLT2i independent of T2DM,² but the latter and/or hyperglycemia may further enhance the efficacy.³ In addition, SGLT2i improve CV outcomes, particularly hospitalization for heart failure.² These CV benefits are most prominent in patients with preexisting heart failure, including heart failure with reduced or preserved ejection, and, likewise, occur largely independent of T2DM.^2,4

What is the magic of inhibiting a glucose transporter in the early proximal tubule of the kidney? Due to a high glomerular filtration rate (GFR), the kidneys filter large amounts of glucose (~180 g/day or ~1/3 of the body caloric expenditure).¹ The kidneys are programmed to retain this valuable energy substrate, which is maladaptive and maintains hyperglycemia in individuals with diabetes mellitus. In euglycemia, most of the filtered glucose (~97%) is being reabsorbed by high-capacity SGLT2 in the “early” proximal tubule (S1/S2 segments).^1,5 Glucose is typically not metabolized in these segments and exits passively through basolateral GLUT2¹ (Figure 1). The glucose that escapes SGLT2 is taken up downstream by SGLT1 in the “late” proximal tubule (S2/S3 segments) and thick ascending limb (Figure 2).¹ Under physiological conditions, tubular SGLT1 only reabsorbs ~3% of filtered glucose.⁶ This changes dramatically, however, when SGLT2i enhance the glucose delivered to the late proximal tubule and unmask a glucose transport capacity of SGLT1 of ~40%–50% of the filtered glucose under euglycemic conditions.⁷ In a nutshell, SGLT2i shift the reabsorption of large amounts of glucose from the early proximal tubule to the downstream late proximal tubule and thick ascending limb, and spill glucose into the urine which modestly lowers hyperglycemia. Moreover, since SGLT2 transports 1 Na⁺ per glucose and SGLT1 transports 2 Na⁺ per glucose,¹ this shift, viewed in isolation, may even cause Na⁺ retention. No obvious magic detected so far.

Figure 1.

SGLT2i reduce early proximal tubule hyperreabsorption and glucotoxicity. (1) Glycemia and glomerular filtration rate (GFR) determine the amount of filtered glucose that is reabsorbed together with Na⁺ by SGLT2, and exits the cells through basolateral GLUT2 (2). (3, 4) In diabetes or heart failure, or in hyperfiltering remnant nephrons (RN) in CKD, SGLT2 expression and activity can be increased by tubular growth, hyperglycemia, hyperinsulinemia, or increased angiotensin II (Ang II) and sympathetic tone (norepinephrine, NE). These settings also stimulate basolateral Na-K-ATPase, which is physiologically energized by adenosine triphosphate (ATP)-generating tricarboxylic acid (TCA) cycle and oxidative phosphorylation (OxPhos), and drives enhanced apical transport via SGLT2, NHE3, and URAT1, which are functionally coupled via neuro-hormonal regulation and scaffolding proteins (4, 5). (6) Hypoxia due to ischemia and injury, potentiated by excessive hyperreabsorption, activates HIF-1α, which limits hyperreabsorption and induces a metabolic shift to glycolysis in order to reduce reactive oxygen species formation from oxidative phosphorylation, but the involved basolateral glucose uptake (via GLUT1) is also linked to induction of TGF-β1 and tubular growth. (7) Excessive cytosolic glucose stimulates oxidative stress via myo-inositol oxygenase (MIOX) and inhibits autophagy through mammalian target of rapamycin complex 1 (mTORC1) activation (8). TGF-β1 enhances cyclin-dependent kinase inhibitors p21 and p27 and together with mTORC1 promotes senescence, which is linked to inflammation and fibrosis. (9) SGLT2i attenuate the described deleterious effects (see blue arrows) by reducing hyperinsulinemia, hyperglycemia, excessive intracellular glucose, and tubular hyperreabsorption (10). (11) SGLT2i enhance the delivery of fatty acids and ketone bodies as energy substrates. (12) SGLT2i may enhance gluconeogenesis (GNG) by lowering hyperinsulinemia, cytosolic glucose, and tubular injury, or due to a mild reduction in HCO₃⁻ following partial NHE3 inhibition). GNG is linked to tubular health, maybe in part by removing TCA cycle intermediates (cataplerosis) thereby facilitating the feeding of fatty acids and ketones into the TCA cycle (anaplerosis). Abbreviations: AA, amino acid; CKD, chronic kidney disease; HIF-1α, hypoxia-inducible factor 1 alpha; HNF, hepatic nuclear factor; NHE3, Na-H-exchanger 3; TGF-β1, transforming growth factor β1; TMEM27, scaffolding protein collectrin; URAT1, urate transporter 1. Adapted with permission from ref. ⁹².

Figure 2.

Primary effects of SGLT2i on the early proximal tubule can explain phenotypes associated with kidney and heart protection. The protective phenotypes identified in mediation analyses include a reduction of plasma volume (PV) or increase in hematocrit (Hct), and a lowering of serum urate levels and albuminuria. SGLT2i increase delivery of NaCl and K ([Na-Cl-K]_MD) and fluid (V) to the macula densa, which lowers GFR through the physiology of tubuloglomerular feedback (TGF) (1) and by increasing hydrostatic pressure in Bowman’s space (P_Bow) (2). SGLT2i lower GFR and glomerular capillary pressure (P_GC) by afferent arteriole constriction and by efferent arteriole dilation. Lowering P_GC and GFR (3) and hyperglycemia (4) protect glomerular and tubular function and reduce albuminuria by lessening physical stress and filtration of albumin and other tubule-toxic compounds, by reducing tubular growth and inflammation, and by lowering tubular transport work (5). The latter reduces cortical oxygen demand ($Q_{O_2}$) (6) and increases cortical oxygen availability ($P_{O_2}$) (7). Early proximal tubule transport work is further reduced by cellular SGLT2 blockade and the associated partial inhibition of other apical transporters including Na-H-exchanger NHE3 (8). SGLT2i shift glucose reabsorption to downstream SGLT1, the latter limiting glucosuria and hypoglycemia risk (9). Shifting the reabsorption of Na⁺, glucose and other metabolites to late proximal tubule and medullary thick ascending limb (mTAL) improves distribution of transport along the nephron but can increase $Q_{O_2}$ (10) and lower $P_{O_2}$ in the outer medulla (OM) (7), where it may activate hypoxia-inducible factor (HIF) and enhance erythropoietin (EPO) release (11). The increase in Hct (12) improves O₂ delivery to the kidney medulla and cortex (13) and other organs (14). More delivery of NaCl and fluid downstream of early proximal tubules facilitates responsiveness to an atrial natriuretic peptide (ANP) and diuretics as well as kaliuresis (15). The induced diuresis further increases Hct (16) and reduces PV and interstitial (ISV) volume and blood pressure (17). These effects can help protect the failing heart (18), while excessive fluid loss is prevented by compensatory transport upregulation downstream of the early proximal tubule, in part via vasopressin (19). Reduced cellular stress and improved metabolism in the early proximal tubule promote the integrity of the tubular and endothelial system and preserve tubular transport capacity and thereby GFR in the long term (20). The glucosuric effect induces a fasting-like response including lower therapeutic and/or endogenous insulin levels and increases in glucagon and FGF21 (21). This induces compensatory lipolysis, ketogenesis, and gluconeogenesis. SGLT2 inhibitors are uricosuric; involving URAT1 inhibition, potentially in part by lowering insulin (22). These metabolic adaptations reduce urate levels, the hypoglycemia risk, and body and organ fat mass, which together with the resulting mild ketosis have the potential to further protect the kidney and heart (18, 23). Abbreviations: U Na-Cl-K, urinary salt, and K excretion; UV, urinary flow rate. Adapted with permission from ref. ⁹².

To learn more about SGLT2i, mediation analyses asked which clinical phenotypes of SGLT2i are associated with improved kidney and heart outcomes? This kind of analysis does not prove cause and effect, but it is remarkable that the strongest predictors of organ protection are basically identical for the kidney and heart, and include a reduction of plasma volume or increase in hematocrit, and lowering of serum urate levels and albuminuria.^8–11 As discussed in the following, all these clinical phenotypes can be explained directly or indirectly by the primary effect of these drugs on their target in the early proximal tubule of the kidney (Figure 2). This review outlines the primary effects and consequences of inhibiting SGLT2 in the proximal tubule and how this can protect the kidney and the heart, but also considers the emerging evidence for potential off-target effects of SGLT2i.

SGLT2I INDUCE A FASTING-LIKE METABOLIC PHENOTYPE THAT IS INDEPENDENT OF DIABETES AND INSULIN SPARING

By spilling glucose and calories into the urine, SGLT2i induce a compensatory and fasting-like metabolic response^1,12 that, due to extensive refinement during evolution, should be for the benefit of the organism and independent of diabetes.¹ This affects a key aspect and sequala of Western lifestyle: the urinary glucose loss by SGLT2i improves insulin resistance and lowers hyperinsulinemia and/or therapeutic insulin dosing. This promotes a shift in substrate utilization from carbohydrates to lipids, thereby reducing body fat, including visceral and subcutaneous fat¹³ (Figure 2). The released free fatty acids are in part used for the hepatic formation of ketone bodies, which provide additional energy for many organs, including the heart and kidneys^14,15 (Figures 1 and 2). On the flip side, excess ketone formation in response to SGLT2i is a concern in conditions of low food intake and insulin levels, including in patients with type 1 diabetes mellitus (T1DM).¹⁶

In contrast to insulin therapy, SGLT2i largely leave the hypoglycemia risk unchanged and lower body weight,¹³ and thus do not carry these deleterious effects of insulin that negatively affect CV risk.¹⁷ Exogenous insulin and hyperinsulinemia also enhance the CV risk by promoting kidney salt and urate retention (Figure 1). In other words, SGLT2i carry organ-protective potential in part by lowering insulin dosing or hyperinsulinemia, the latter being often found in obesity and outside of T2DM.

With regard to the hypoglycemia risk, SGLT2i stop lowering blood glucose once the drop in blood glucose (~60–70 mg/dl) allows all the filtered glucose to be recovered by downstream SGLT1 (~80 g/day)¹ (Figure 2). In addition, compensatory upregulation of gluconeogenesis in response to SGLT2i in the liver, and potentially the kidney, provide critical glucose for organs, like the brain, and also keep the hypoglycemia risk low¹³ (Figures 1 and 2). Thus, in individuals with T2DM, SGLT2i protect the renal and CV system from dangerous blood glucose highs and lows, i.e., blood glucose remains more on target, with only modest net effects expected on HbA1C.

SGLT2I LOWER GLUCOTOXICITY IN THE EARLY PROXIMAL TUBULE

SGLT2i reduce the apical glucose load into the cells of the early proximal tubule. In addition and as suggested by data in nondiabetic mice, SGLT2i can reduce kidney GLUT2 protein expression, which may attenuate basolateral back leak of glucose when apical SGLT2 is blocked.¹⁸ This is associated with reduced protein expression of proximal tubular enzymes involved in the handling of glucose, oxidative stress, and cellular detoxification, including myo-inositol oxygenase (MIOX, Figure 1) and glutathione peroxidase 3 (GPX3).¹⁸ In rodent proximal tubules, SGLT2i induce starvation-like responses including sirtuin (SIRT)/5' AMP-activated protein kinase (AMPK) activation and inhibition of the protein kinase B (AKT)/mTORC1 pathway, thereby counteracting the primary pathophysiology of diabetes and overnutrition and inducing protective autophagy while inhibiting deleterious senescence, which is linked to inflammation and fibrosis¹ (Figure 1). Suppression of proximal tubular mTORC1 by SGLT2i has been confirmed by single-cell RNA sequencing studies in kidney biopsies of young persons with T2DM.¹⁹

In patients with T2DM and albuminuria, SGLT2i increased urinary metabolites linked to mitochondrial metabolism, potentially indicating improvement of mitochondrial function and activity.²⁰ Similarly, single-cell RNA sequencing of proximal tubules in db/db mice indicated that SGLT2i restored mitochondrial function and fatty acid oxidation in proximal tubules.²¹ In nondiabetic mice SGLT2i caused distinct effects on kidney metabolism reflecting responses to partial NHE3 inhibition (see below) as well as urinary loss of glucose and NaCl; this included upregulated renal gluconeogenesis and using tubular secretion of the tricarboxylic acid cycle intermediate, alpha-ketoglutarate, to communicate to the distal nephron the need for compensatory NaCl reabsorption.²² Upregulation of gluconeogenesis could be of particular interest, because its activity is positively linked to proximal tubule mitochondrial function and kidney protection^23,24 (for more details see Figure 1).

The clinical observation that SGLT2i have a phosphate-sparing effect²⁵ may also relate to the interference with early proximal tubule SGLT2. An increase in luminal Na⁺ availability as a consequence of SGLT2i may facilitate proximal tubule Na⁺-coupled phosphate reabsorption. Moreover, SGLT2-dependent glucose uptake is needed for phosphate sensing by the proximal tubule, which involves glycolysis-driven formation and subsequent release of glycerol-3-phosphate to stimulate bone release of phosphaturic FGF23.²⁶ While the SGLT2i-induced increase in serum phosphate is associated with an increase in circulating FGF23,^27,28 the rise in FGF23 could be attenuated and insufficient due to the blunted glycerol-3-phosphate response.

SGLT2I CAUSE A BROAD RECONFIGURATION OF METABOLITE TRANSPORTERS IN THE EARLY PROXIMAL TUBULE

Recent data indicate an extensive physical and functional coupling of SGLT2 with other sodium and metabolite transporters in the luminal membrane of the early proximal tubule (Figure 1). The inhibitory interaction of SGLT2i with these transporters is thought to involve scaffolding proteins like MAP17, but may also relate to changes in transporter phosphorylation and the insulin-lowering effect of SGLT2i¹ (Figures 1 and 2). Insulin co-stimulates SGLT2, NHE3, the urate transporter URAT1, and potentially more transporters in the early proximal tubule¹ (Figure 1). This co-stimulation and interaction make sense in the post-prandial phase when the rise in GFR enhances the amounts of filtered sodium, glucose, bicarbonate, urate, and other substrates that are reabsorbed in the early proximal tubule. In hyperinsulinemic states like T2DM and obesity, co-stimulation facilitates renal glucose, NaCl, and urate retention. Finally and most importantly for this review, this co-stimulation and interaction make an array of transport processes in the early proximal tubule sensitive to SGLT2i¹ (Figures 1 and 2).

Studies in rodents first described that the inhibitory effect of SGLT2i on NHE3 contributes to both the acute natriuretic and the chronic volume depletion effect.^22,29 A single dose of empagliflozin, titrated to cause minimal glucosuria [i.e., shifting glucose reabsorption from SGLT2 (transporting 1 Na⁺/glucose) to SGLT1 (transporting 2 Na⁺/glucose)], increased urinary excretion of Na⁺ and bicarbonate and raised urine pH in wild-type mice but had an antinatriuretic effect in mice lacking tubular NHE3.²² In accordance, application of empagliflozin in patients with diabetes and heart failure resulted in a larger than expected decrease in proximal tubule Na⁺ reabsorption with several observations suggesting inhibition of NHE3,³⁰ and similar findings were made in healthy volunteers.³¹ Studies in knockout mice³² and pharmacological studies in humans³³ implicated URAT1 inhibition in the serum urate lowering effect of SGLT2i (Figures 1 and 2).

Recent mouse and human kidney SGLT2 interactome studies and kidney cortex proteomics/phospho-proteomics in nondiabetic mice confirmed the interaction of SGLT2 with multiple NHE-scaffolding proteins. In both mouse and human kidney tissue, SGLT2 interacted with MAP17, NHE-RF2, NHE-RF3 as well as the G-protein-coupled receptor GPRC5C,¹⁸ which has likewise been implicated in NHE3 regulation.³⁴ Dapagliflozin actually reduced MAP17 protein expression in mouse kidneys.¹⁸ Furthermore, evidence was presented that SGLT2 can physically interact with URAT1 as well as with other apical urate transporters (incl. GLUT9 and ABCG2), and dapagliflozin reduced URAT1 protein expression and phosphorylation in the mouse kidney cortex. Beyond these interactions, the studies demonstrated direct or indirect physical interactions of SGLT2 with multiple other apical metabolite transport proteins in the early proximal tubule, many of which are Na coupled, including potential interactions with early proximal tubule amino acid transport¹⁸ (Figure 1).

In summary, inhibition of SGLT2i appears to lower the early proximal tubule transport burden not only by inhibiting the cotransport of Na⁺ and glucose but, as a consequence of insulin co-stimulation and/or physical interaction, by partially inhibiting multiple other transporters at the same time, many of which are also Na⁺ coupled.¹⁸

SGLT2I SHIFT TRANSPORT LOAD AND TRANSPORT WORK DOWNSTREAM: DELETERIOUS OR BENEFICIAL?

The early proximal tubule has evolved in a way such that it reabsorbs a large fraction of the glomerular filtrate, which comes with a large demand for oxygen. This setup works fine in a healthy kidney, but may make the early proximal tubule particularly vulnerable during kidney injury. As discussed above, SGLT2i shifts some of the glucose, NaCl, urate, amino acid, and fluid reabsorption downstream, and thereby more equally distributes the transport burden along the tubular system, which may help to preserve tubular function in the long term.¹ This transport shift in response to SGLT2i is associated with a shift in energy and oxygen demand from early to late proximal tubule and thick ascending limb as proposed by mathematical modeling^35–37 and, more recently, by changes in tricarboxylic acid and glycolysis-related gene expression in single-cell RNA sequencing of kidney biopsies of young persons with T2D.¹⁹ In the nondiabetic setting, the transport shift by SGLT2i is associated with an increase in kidney weight³⁸ including hypertrophy of late proximal tubules/S3 segments³⁹ as well as downregulation of renal membrane SGLT1 protein expression, potentially to limit tubular glucose overload in those segments.³⁸ Thus, SGLT2i more equally distribute the transport burden along the nephron but put stress on these downstream segments and the vulnerable region of the outer medulla, which may dampen the overall kidney benefits of SGLT2i. The transport shift-induced stress by SGLT2i is attenuated by the associated early reduction in GFR (see below) and potentially by metabolic use of the delivered glucose by downstream segments.¹

The transport shift may explain the upregulation of stress markers, like hemoxygenase-1^38,40; on the flip side, this way SGLT2i may drive a tubular pre-conditioning-like adaptation¹ that could contribute to the clinically observed protection from acute kidney injury.^38,40–42 Moreover and based on mathematical modeling, the transport shift has been proposed to “simulate” systemic hypoxia at the oxygen sensor in the deep cortex and outer medulla of the kidney³⁷ (Figure 2): by shifting transport work from the early proximal tubule to tubular segments in the deep cortex and outer medulla, including proximal tubule S3 segments and medullary thick ascending limbs, SGLT2i increase the oxygen consumption at these sites, which is then sensed as reduced oxygen availability by the erythropoietin secreting interstitial cells found primarily in this region of the kidney. Thus, independent of systemic hypoxia, SGLT2i trick these cells to sense hypoxia.³⁷ Supporting this hypothesis, treatment of individuals with T2DM with an SGLT2i for 32 weeks reduced kidney medullary oxygenation and this was associated with increased erythropoietin levels and hematocrit.⁴³ The latter is further enhanced by the diuretic effect of SGLT2i (see below). The resulting improvement in oxygenation of the kidney (particular importance of kidney cortex oxygenation for CKD progression^44,45) and heart may explain why the increase in hematocrit and hemoglobin are among the strongest predictors of improved kidney and CV outcome by SGLT2i^8–11 (Figure 2).

THE BIPHASIC EFFECT OF SGLT2I ON GFR

The difficulty to protect the kidneys is in part due to their structural complexity. Are some structures, segments, or cells of the kidney more important to protect than others in order to conserve kidney function? Does this include the early proximal tubule? A more general approach could be to manipulate the kidneys’ energy and O₂ needs. Since basically all filtered NaCl and fluid (and many other filtered solutes) need to be reabsorbed along the tubular system to match urinary excretion to daily intake, and since this transport work costs energy and oxygen, the GFR is the primary determinant of kidney energy and oxygen need. Metabolic control of heart function includes workload-induced coronary vasodilation to adjust oxygen supply to demand; this involves cellular adenosine release and activation of adenosine A₂ receptors. Metabolic control of kidney function also involves adenosine release in response to excessive transport work or energy imbalance, and adenosine primarily reduces GFR.⁴⁶ Also the diseased kidney may in part reduce GFR to limit the kidneys’ energy and oxygen need.⁴⁷ These responses point to a general strategy of kidney protection, namely a functional and modest lowering of GFR. A functional reduction in GFR and glomerular capillary pressure (P_GC) has been implicated in kidney protection by renin–angiotensin system blockade^48,49 and now by SGLT2 inhibition,⁵⁰ and the clinical trials provided strong evidence that the 2 strategies are additive.⁵⁰

Considering that GFR is typically ~100-fold greater than urine flow rate, there is a need for mechanisms that coordinate glomerular filtration and tubular transport capacity. SGLT2i functionally lower GFR by manipulating one of these mechanisms, namely the tubuloglomerular feedback (TGF). The TGF describes an inverse relationship between the NaCl and fluid delivery to the macula densa (MD) and the GFR of the same nephron (Figure 3), which serves to stabilize fluid and electrolyte delivery to the MD thereby facilitating fine regulation of excretion in downstream distal segments.⁵⁰ SGLT2i reduce proximal tubule reabsorption and enhance the delivery of NaCl and fluid to the MD; the physiology of the TGF then aims to normalize MD NaCl and fluid delivery by reducing GFR (Figures 2 and 3). This has been demonstrated by micropuncture studies in hyperfiltering diabetic rats,^51,52 and the GFR lowering effect of short-term SGLT2 inhibition was subsequently confirmed in diabetic mice and humans with T1DM and T2DM.⁵⁰ Some GFR reduction was also observed in nondiabetic mice and humans.⁵⁰

Figure 3.

Glomerular hemodynamic response to SGLT2i. (a) SGLT2i increase Na-Cl-K delivery to the macula densa (MD), which is sensed by an NKCC2-dependent mechanism and enhances basolateral release of ATP. ATP is converted by endonucleotidases CD73/39 to adenosine (ADO), which activates adenosine A1 receptor (A1AR) on vascular smooth muscle cells of the afferent arteriole to increase cytosolic Ca2⁺ and induce vasoconstriction. ADO can also activate adenosine A2 receptors (A2AR) on the efferent arteriole to reduce cytosolic Ca2⁺ and induce vasodilation. Both effects contribute to the TGF response and lower GFR and P_GC. Efferent dilation can enhance GFR by increasing renal blood flow (RBF), thus causing a stronger effect on P_GC than GFR. SGLT2i increase glucose delivery to the MD, which is sensed by MD SGLT1 and triggers NOS1-mediated NO formation, which can dilate the afferent arteriole but potentially also the efferent arteriole. Volume depletion (e.g., following SGLT2i) can inhibit MD NOS1 activity. (b) Glomerular hemodynamic effects of SGLT2i superimposed on a model of TGF, which incorporates effects on afferent and efferent arteriolar resistances. Idealized TGF curves are shown for GFR and P_GC. The P_GC curve lies to the right of the GFR curve owing to differences in the way that pre- and postglomerular resistances react across the range of inputs, as shown in the bottom portion of the figure. Note the efferent arteriolar dilation for high [NaCl] at the macula densa. Depending on the position of the operating point, identical increases in macula densa delivery by SGLT2i can induce different relative decreases in GFR vs. P_GC; e.g., in CKD and due to hyperfiltration of remnant nephrons, the TGF may operate closer to the elbow of the TGF curve (blue arrow), resulting in a smaller decline in GFR but a larger decline in P_GC. Abbreviations: CKD, chronic kidney disease; NO, nitric oxide; NOS1, NO synthase; P_GC, glomerular capillary pressure. Adapted with permission from ref. ^53,92.

When MD cells sense an increase in luminal NaCl, the resulting TGF-induced basolateral release of ATP promotes local formation of adenosine, which primarily constricts the afferent arteriole (via adenosine A1 receptors) but can also dilate the efferent arteriole (via adenosine A2 receptors)⁵⁰ (Figures 2 and 3). Both effects lower P_GC as was directly demonstrated in response to SGLT2i by micropuncture in rats with T1DM_.⁵³ Those studies further indicated that SGLT2 inhibition can induce a robust reduction in P_GC even when GFR decreases only slightly and vice versa. This argued against a sole effect of SGLT2i on the afferent arteriole, and implied combined constriction of the afferent arteriole and dilation of the efferent arteriole⁵³ (Figure 3), which could explain the observed reduction in filtration fraction by SGLT2i in some of the clinical studies.^54,55 This may have implications in advanced CKD (GFR <30 ml/(min × 1.73 m²)), where the initial GFR drop in response to SGLT2 inhibition can be small, but the kidney protective effect is preserved.⁵⁶ We hypothesize that compensatory hyperfiltration in remaining nephrons shifts the operating point along the TGF curve, such that the increase in MD NaCl by SGLT2i (and/or glucose, see below) induces a predominant reduction in P_GC⁵³ (Figure 3).

The MD expresses SGLT1 in its apical membrane and thereby senses luminal glucose to stimulate local nitric oxide (NO) formation via neuronal NO synthase (NOS1).^57–59 The SGLT2i-induced rise in MD glucose delivery activates the MD–SGLT1–NOS1 pathway⁵⁸ (Figure 3). Considering the close proximity of the MD to its afferent and efferent arteriole, it seems possible that the SGLT2i-induced activation of the MD–SGLT1–NOS1 pathway dilates not only the afferent but also the efferent arteriole (the latter effect reducing P_GC), particularly in settings of CKD where endothelial integrity and efferent arteriolar NO tone may be reduced.

Most importantly, SGLT2i induce a biphasic GFR response in patients with T2DM and normal or reduced kidney function: the described initial reduction in GFR is followed by long-term GFR preservation.⁵⁰ Moreover, eGFR recovered to baseline when long-term SGLT2i treatment was discontinued,⁵⁰ indicating that the early rise in plasma creatinine in response to an SGLT2 inhibitor reflects a “functional and reversible” reduction in GFR, and is part of the therapeutic effect. Indeed, SGLT2i decreased urinary excretion of markers of tubular injury in patients with T2DM,⁶⁰ and meta-analyses of clinical studies showed a ~25% reduced incidence of acute kidney injury.^41,42 Furthermore, the application of SGLT2i in individuals with T2DM and acute kidney injury was associated with a lower risk of mortality (by 31%), and major kidney (by 38%) and adverse cardiovascular events (by 25%) compared with nonuse.⁶¹

By reducing GFR and P_GC (and increasing hydrostatic back pressure in Bowman space due to delivering more fluid to the high resistance distal tubule), SGLT2i reduce the physical stress on glomerular capillaries and diminish glomerular filtration of tubulo-toxic factors (e.g., albumin with or without fatty acids bound to it, growth hormones, advance glycation end products).⁵⁰ The interaction of these factors with the tubular system requires energy and promotes hypoxia, impairs autophagy, and triggers tubular remodeling, dedifferentiation, senescence, renal oxidative stress, inflammation, and fibrosis, and thereby the development and progression of diabetic and nondiabetic kidney disease^50,62 (Figure 2).

Preservation of cortical oxygenation is associated with the preservation of kidney function in patients with CKD.⁴⁵ By lowering GFR and SGLT2-dependent transport processes, SGLT2i reduce oxygen-consuming transport work in the early proximal tubule as predicted by mathematical modeling³⁷ and confirmed on the kidney cortex level in albuminuric patients with T1DM.⁴⁴ Particularly in the more chronic setting, not all clinical studies find an increase in cortical O₂ pressure with SGLT2i, which may reflect improved oxidative phosphorylation and oxygen utilization by the kidney cortex. Thus, by lowering GFR and the oxygen-consuming transport work in the kidney cortex, SGLT2i have the potential to preserve the integrity of the remaining nephrons and overall kidney function in the long term⁵⁰ (Figure 2).

SGLT2I REDUCE FLUID OVERLOAD AND IMPROVE HEART FAILURE OUTCOME

The improvement in heart failure outcome by SGLT2i occurs within weeks of treatment and is independent of hyperglycemia.^2,4 The immediate diuresis and natriuresis, the described kidney protective mechanisms, as well as the diabetes-independent metabolic resetting and the improvement of insulin sensitivity, all have the potential for SGLT2i-mediated heart protection.^13,63 This includes reducing volume overload and hypertension, preserving or increasing hematocrit and hemoglobin and thereby oxygen supply, inducing mild ketosis and thereby providing more energy substrate, lowering urate levels, inhibition of the sympathetic nervous system and NLRP3 inflammasome, and preserving GFR and limiting hyperkalemia, thereby improving tolerance to renin–angiotensin-aldosterone system (RAAS) blockade¹³ (Figure 2).

SGLT2i promote mild osmotic diuresis and natriuresis that reduce fluid overload, whole-body sodium content, and blood pressure, which may afford rapid protection against heart failure.^13,64 In heart failure, upregulation of proximal tubule NHE3, driven by angiotensin II or sympathetic nervous system, may further sensitize early proximal tubule Na⁺ reabsorption to SGLT2i.^22,29,30

The SGLT2i-induced natriuresis is immediately limited and increasingly attenuated by downstream SGLT1, by NKCC2-mediated NaCl reabsorption in the thick ascending limb^52,65 being stimulated by greater NaCl delivery³⁷ and vasopressin levels,^66,67 as well as by the modest lowering in arterial blood pressure. Additional compensation may include growth of the S3 segment and collecting duct³⁹ and α-ketoglutarate-driven NaCl transport in intercalated cells.²² Thus, the compensation reduces the natriuretic response within several days to weeks of continuous SGLT2i application. On the flip side, the compensation lowers the risk of severe volume depletion by SGLT2i and the clinically observed Mg2⁺ retention²⁵ may be due to increased thick ascending limb reabsorption.

The SGLT2i-induced reduction in fluid overload is independent of diabetes or kidney function,^68,69 and has been implicated in the initial reduction in systolic blood pressure, while the (mechanistically less clear) reduction in sympathetic nervous system activity may contribute to the long-term reduction.⁶⁸ The natriuretic effect of SGLT2i, which is associated with lowering plasma volume and increasing hematocrit,⁷⁰ appears more sustained and induces a relatively greater reduction in interstitial vs. intravascular volume than the loop diuretic bumetanide, potentially due to the osmotic diuresis causing a greater electrolyte-free water clearance⁷¹ (Figure 2). Moreover, the SGLT2i-induced increase in NaCl and fluid delivery downstream of the early proximal tubule is expected to enhance the responsiveness to loop and thiazide diuretics and potentially also endogenous atrial natriuretic peptide (Figure 2).

SGLT2 gene knockout or inhibition facilitates urinary K⁺ excretion.^5,13 The above-described adaptations reduce kidney K⁺ loss in response to SGLT2i by limiting the rise in distal tubule fluid and Na⁺ delivery and thereby flow-induced K⁺ secretion and principal cell Na⁺/K⁺ exchange. As a net effect, SGLT2i either cause no effect or a slight reduction in plasma K⁺ concentration in individuals with normal kidney function, but lower the hyperkalemia risk due to RAAS blockade and protect against hyperkalemia in individuals with CKD^13,72 (Figure 2).

SGLT2i have clinically documented antiarrhythmic effects on the heart^73–75 but the involved mechanisms remain unclear. Potentially relevant in this regard is evidence for SGLT2-independent cardiac effects of SGLT2i. These “off-target effects” of SGLT2i have been proposed to affect pro-inflammatory and oxidative stress processes, ion transport mechanisms (like NHE1) controlling cellular sodium and calcium homeostasis, and cardiac metabolic and mitochondrial pathways⁷⁶ but more studies are needed.⁷⁷ In addition, SGLT2i have been proposed to directly interact with and inhibit the intestinal microbiota fermentation of amino acids to uremic toxins (Figure 4). This lowers circulating levels of the uremic toxin p-cresol sulfate, as documented in mice (including Sglt2 KO mice), rats as well as patients with diabetes or decompensated heart failure, and bears potential to lower cardiac formation and release of GDF15 and thereby improve cardiac contractility and rhythm.¹⁸ GDF15 has been linked to cardiac atrophy⁷⁸ as well as higher mortality risk in hemodialysis patients.⁷⁹ Uremic toxins positively regulate their own kidney secretion and excretion by stimulating basolateral organic anion transporters (OAT1/OAT3) in the proximal tubule. As a consequence, by reducing the formation and circulating concentrations of uremic toxins, SGLT2i can lower the activity of these kidney transporters, which has 2 consequences: first, transporter substrates can be retained in circulation, including amino acid metabolism-derived compounds with CV benefits (e.g., lowering blood pressure or inflammation^18,80) and, second, proximal tubular exposure to organic anion toxins can be attenuated by SGLT2i¹⁸ (Figure 4) as recently proposed for aristolochic acid.⁸¹

Figure 4.

SGLT2i may protect the kidney and heart by inhibition of microbiotic formation of uremic toxins. Chemical structures of SGLT2 inhibitors dapagliflozin and empagliflozin and p-cresol are shown high-lighting similarities in red, which may provide a mechanistic basis. SGLT2i lower circulating levels of uremic toxins including p-cresol/p-cresol sulfate, thereby lowering cardiac formation and deleterious effects of GDF15 on the heart and kidney. Uremic toxins positively regulate their own kidney excretion by stimulating basolateral organic anion transporters (OAT1/OAT3) in the proximal tubule. By reducing circulating levels of uremic toxins, SGLT2i can lower the activity of these kidney transporters which can retain body-derived OAT substrates, including amino acid metabolism-derived compounds with cardiovascular benefits. On the other hand, proximal tubular exposure to uremic and other organic anion toxins would be attenuated. As suggested by proteomics in mice and human kidney interactome studies, SGLT2i may enhance kidney expression and physically interact with the environment-susceptible P-glycoprotein (P-gp, ABCB1), a primary proximal tubule pathway involved in secretion of xenobiotics and toxins.¹⁸ Abbreviations: aKG, alpha-ketoglutarate; OAT1/3, organic anion transporters 1 and 3. Adapted with permission from ref. ¹⁸.

OUTLOOK

Insights from multi-omic studies are needed at the cellular level of the early proximal tubule, where SGLT2 is targeted, but also in downstream segments exposed to greater amounts of glucose, NaCl, and multiple other substrates. This includes the S3 segment of the proximal tubule but also the MD, which has been implicated in broader kidney remodeling, in part through glucose sensing.^50,58,82 How do SGLT2i interfere with the efferent arteriole as well as the erythropoietin system? Why is sympathetic nervous system activation attenuated in response to SGLT2i,¹³ and what is the quantitative relevance of off-target effects of SGLT2i on the heart, kidney, and other organs. Is there therapeutic potential for SGLT2i in type 1 diabetes^83,84? What is the role of SGLT1 in dual SGLT2/SGLT1 inhibitors^85,86? Are SGLT2i synergistic with other kidney protective strategies, beyond renin-angiotensin system blockade, like GLP1 agonists⁸⁷ and aldosterone antagonists⁸⁸? Finally, can we use what we learn from SGLT2 and its inhibition about the inner workings of the kidney for the identification of new therapeutic targets? Table 1 summarizes the characteristics of SGLT2 and its inhibition. The Na-coupled neutral amino transporter SLC6A19 fulfills these characteristics, and the first studies document metabolic benefits⁸⁹ and kidney protection⁹⁰ of the knockout in mice and highlight a potential creatinine phenotype of human mutations in the gene.⁹¹

Table 1.

Characteristic of SGLT2 and its inhibition

- Sodium-coupled transporter

- Kidney expression restricted to the early proximal tubule

- Functional coupling with other apical transporters in the early proximal tubule

- Excess of substrate transport and uptake is deleterious to early proximal tubule cells

- Inhibition can lower GFR

- Inhibition causes a shift in transport to downstream tubular segments

- Inhibition induces a fasting-like metabolic response

FUNDING

The author was supported by National Institutes of Health (NIH) Grants R01 DK112042, RF1 AG061296, R01 DK132690, R01 DK 134616, and the University of Alabama at Birmingham/University of California-San Diego O’Brien Center of Acute Kidney Injury NIH Grant P54 DK137307, and the Department of Veterans Affairs.

CONFLICT OF INTEREST

Over the past 24 months, V.V. has served as a speaker and received honoraria from Astra-Zeneca and Boehringer-Ingelheim, and received grant support for investigator-initiated research from Boehringer-Ingelheim, Gilead, Lexicon, Novo-Nordisk, and Maze Therapeutics.

AUTHOR CONTRIBUTIONS

V.V. wrote review.

DATA AVAILABILITY

Not applicable.