Tubular effects of sodium–glucose cotransporter 2 inhibitors: intended and unintended consequences

Abstract

Purpose of review

Sodium–glucose cotransporter 2 (SGLT2) inhibitors are antihyperglycemic drugs that act by inhibiting renal sodium–glucose cotransport. Here we present new insights into ‘off target’, or indirect, effects of SGLT2 inhibitors.

Recent findings

SGLT2 inhibition causes an acute increase in urinary glucose excretion. In addition to lowering blood glucose, there are several other effects that contribute to the overall beneficial renal and cardiovascular effects. Reabsorption of about 66% of sodium is accomplished in the proximal tubule and dependent on the sodium–hydrogen exchanger isoform 3 (NHE3). SGLT2 colocalizes with NHE3, and high glucose levels reduce NHE3 activity. The proximal tubule is also responsible for the majority of phosphate (P_i) reabsorption. SGLT2 inhibition is associated with increases in plasma P_i, fibroblast growth factor 23 and parathyroid hormone levels in nondiabetics and type 2 diabetes mellitus. Studies in humans identified a urate-lowering effect by SGLT2 inhibition which is possibly mediated by urate transporter 1 (URAT1) and/or glucose transporter member 9 in the proximal tubule. Of note, magnesium levels were also found to increase under SGLT2 inhibition, an effect that was preserved in nondiabetic patients with hypomagnesemia.

Summary

Cardiorenal effects of SGLT2 inhibition might involve, in addition to direct effects on glucose homeostasis, effects on NHE3, phosphate, urate, and magnesium homeostasis.

INTRODUCTION

Diabetes mellitus, as a chronic disease, has become the most common pandemic affecting more than 400 million adults worldwide [1]. The two major types of diabetes mellitus are type 1 diabetes mellitus (T1DM) which is caused by immunological responses leading to β-cell destruction in the pancreas, and type 2 diabetes mellitus (T2DM) which is caused by progressive insulin deficit on the background of insulin resistance [2]. The hallmark of diabetes mellitus is hyperglycemia, with the leading cause of death being related to cardiovascular complications. Prolonged exposure to glucose can impose damage on several cell types, particularly the vascular endothelium, and glucose toxicity is strongly correlated with diabetes mellitus-related microvascular and macrovascular complications. Various classes of drugs are currently utilized in the treatment of diabetes mellitus, including but not limited to insulin, metformin, sulphonylureas, glitazones, thiazolidinediones, glucagon-like peptide-1 receptor agonists, dipeptidyl peptidase-4 inhibitors and gliflozins. The latter are a novel class of antidiabetic drugs [3] that have been approved for the treatment of diabetes mellitus by the US Food and Drug Administration. Sodium-glucose cotransporter 2 (SGLT2) inhibitors mediate glucose-lowering effects through inhibition of sodium and glucose reabsorption in the kidney, thereby inducing glucosuria (see below). Data from three completed cardiovascular outcome trials (EMPA-REG OUTCOME [4], CANVAS Program [5], and DECLARE-TIMI 58 [6^■■]) demonstrate cardioprotective effects of SGLT2 inhibition. A recently published renal outcome trial (CREDENCE [7^■■]) showed renal protective effects of SGLT2 inhibitors in patients with T2DM and chronic kidney disease (CKD). Based on these positive findings, the American Diabetes Association, the European Association for the Study of Diabetes [8], and the European Society of Cardiology [9] updated their guidelines recommending SGLT2 inhibitors in diabetes mellitus to reduce the risk of cardiovascular disease and slow the progression of CKD. Aside from the glucose-lowering effects, the cardiovascular and renal protective effects of SGLT2 inhibitors are multifactorial, including diuretic and natriuretic effects, suppression of tubular hypertrophy, normalization of tubuloglomerular feedback (TGF), improvement of oxygenation, reductions in body weight, reductions in blood pressure, and reductions in urate levels and others. In this review, we provide a summary of recent findings on ‘off target’ effects of SGLT2 inhibitors, including effects on sodium/hydrogen exchanger isoform 3 (NHE3), phosphate, urate, and magnesium homeostasis.

EFFECT OF SODIUM-GLUCOSE COTRANSPORTER 2 INHIBITORS ON URINARY GLUCOSE EXCRETION IN HEALTH AND DIABETES MELLITUS

Several excellent reviews have covered this topic [3,10^■,11–13], which is why we will only give a brief overview of the effects of renal SGLT1/2 on glucose homeostasis. SGLT2 is expressed in the apical membrane of the early proximal tubule (S1/S2 segments) with SGLT1 being expressed in the late proximal tubule (S3 segments). In healthy adults, approximately 180 g of glucose is filtered per day and completely reabsorbed by the kidneys. SGLT2 mediates about 97% of total renal glucose reabsorption [14], whereas the remaining 3% is accounted for by SGLT1 [15,16]. These findings are based on studies in mice lacking SGLT1 [14,15], SGLT2 [14] and SGLT1/2 [16]. The use of micropuncture and clearance studies helped to establish their respective contributions [14,15,17]. Of note, pharmacological inhibition or knockout of SGLT2 reduced fractional glucose reabsorption by about 50% and not as expected by about 97%, a finding that was later determined to be caused by significant upregulation of SGLT1 [16]. In diabetic patients, the blood glucose level can exceed the renal glucose reabsorption capacity (~11.1 mmol/l) and the excessive filtered glucose is excreted in the urine [3]. Renal growth is a typical feature in the early stage of diabetes mellitus and closely associated with higher expression of SGLT2 [18,19,20^■■]. Higher SGLT2 expression, as found in diabetes mellitus of humans [21] and rodents [18,19,22^■,23^■], enhances glucose reabsorption capacity, which exacerbates hyperglycemia in diabetes mellitus and could be considered maladaptive. Of note, streptozotocin-induced T1DM in mice caused a reduction in SGLT2 expression [19,24^■]. The reasons for these differences remain elusive but need to be considered when studying animal models. The reabsorptive capacity of SGLT1 is also substantially increased in T2DM [25], providing support to the hypothesis that dual SGLT1 and SGLT2 inhibition could have additional beneficial effects because of greater glucose lowering as well as other mechanisms (reviewed in [10^■]).

EFFECT OF SODIUM-GLUCOSE COTRANSPORTER 2 INHIBITORS ON GLOMERULAR FILTRATION RATE

Approximately 40% of patients with T2DM and 30% of patients with T1DM will develop CKD [26]. Glomerular hyperfiltration occurs in the early stages of diabetic kidney disease (DKD), which is associated with higher risk of developing diabetic nephropathy [27,28]. The tubular hypothesis postulates that the tubular growth and hypertrophy is critical for the onset and progression of glomerular hyperfiltration in DKD [20^■■]. Tubular hypertrophy is associated with increased SGLT2 expression. Although tubular growth contributes to a generalized increase in transporters [29], the reason for the increase in SGLT2 is likely multifactorial (i.e., diabetes-induced hyperglycemia, hyperinsulinemia). The consequence of increased SGLT2 expression is enhanced glucose, Na⁺ and fluid reabsorption in the proximal tubule, which results in reduced delivery to downstream nephron segments and lowers the TGF response via the macula densa [30]. Adenosine is the mediator of TGF [31] and mice lacking adenosine A1 receptors do not develop glomerular hyperfiltration [32,33]. A novel SGLT1-nitric oxide synthase 1 (NOS1)-pathway has been identified in mice as another contributing factor for glomerular hyperfiltration in diabetes mellitus [34^■■,35^■]. Here, the enhanced luminal glucose level in diabetes mellitus is sensed by the macula densa via SGLT1, consequently leading to NOS1 mediated nitric oxide generation. Nitric oxide counteracts the vasoconstriction mediated by adenosine and therefore increases glomerular filtration rate (GFR). Of note, SGLT1 has not been localized to the macula densa of humans [36].

The amount of glucosuria as a result of SGLT2 inhibition depends on the GFR, which determines the amount of filtered load of glucose. Consequently, the glucose-lowering effect of SGLT2 inhibitors becomes less effective in patients with diabetes mellitus and advanced CKD, because reduced GFR leads to a reduction in filtered glucose (~80 g daily). Nevertheless, in 2019 the first dedicated clinical trial (CREDENCE) to investigate canagliflozin primarily for renal protection in patients with T2DM and CKD was terminated early because of achieving prespecified efficacy criteria related to slowing the progression of CKD [7^■■]. During the first three weeks, canagliflozin treatment resulted in a greater reduction in the estimated GFR compared with the placebo control, consistent with activation of TGF. As a physiologic response, this effect is reversible. Consequently, the canagliflozin group showed a slower decline in estimated GFR compared with placebo. This was a big milestone considering that SGLT2 inhibitors are the first class of drugs to show renoprotective effects since the 1980s, where inhibitors of the renin–angiotensin–aldosterone system (RAAS) were introduced to slow down the progression of CKD.

EFFECT OF SODIUM-GLUCOSE COTRANSPORTER 2 INHIBITORS ON ELECTROLYTE, MINERAL, AND URATE HOMEOSTASIS

Sodium-glucose cotransporter 2–sodium–hydrogen exchanger isoform 3 interactions

The NHE3 expressed in the proximal tubule is directly and indirectly responsible for the reabsorption of two-thirds of the filtered sodium [37,38]. Mathematical modeling studies calculated that the SGLT2 component of proximal tubule sodium reabsorption was about 8%, about 16%, and about 19% in healthy, controlled T2DM, and uncontrolled T2DM virtual patients, respectively [39^■]. Another study determined that under normal conditions, about 25% of sodium reabsorption in the proximal tubule is coupled to bicarbonate, which is the major driver of net proximal reabsorption. In conditions of glucose transport saturation, up to about 80% of sodium reabsorption could be coupled to bicarbonate [13]. Elegant functional studies using stationary microperfusion identified that low (5 mmol/l) glucose concentrations stimulated versus high (40 mmol/l) glucose concentrations inhibited NHE3-dependent bicarbonate reabsorption, respectively [40]. The NHE3 inhibitor S3226 inhibited these NHE3-dependent effects. Anatomically, NHE3 and SGLT2 show colocalization [40]. Taken together, these studies provide evidence for a functional interaction between NHE3/SGLT2 (Fig. 1). Mechanistically, treatment with empagliflozin identified that phosphorylation of NHE3 at serine residues 552 and 605 was increased, both are inhibitory sites of NHE3 [41]. It is plausible that, at least partially, the effects of SGLT2 inhibitors on plasma volume are mediated by their inhibitory action on NHE3 as well as osmotic diuresis. The diuretic and natriuretic effects of SGLT2 inhibitors cause plasma volume contraction. Of note, these changes appear to be temporary because no permanent water loss occurred in patients with T2DM treated for six months with empagliflozin or dapagliflozin [42^■]. Rodent models of T2DM treated with ipragliflozin showed that body fluid volume, measured by bioimpedance spectroscopy, was not different compared with vehicle-treated control animals [22^■,23^■]. Several compensatory mechanisms to maintain body fluid volume are activated, such as RAAS [42^■], water intake [22^■], water reabsorption via arginine-vasopressin/aquaporin 2 system [23^■] and expression of urea transporter A1 [43]. Further studies are needed to better understand the regulatory mechanisms and signaling pathways between NHE3 and SGLT2.

FIGURE 1.

Summary of effects of SGLT2 inhibitors (SGLT2-I) on NHE3, urate, phosphate, and magnesium. (1) In the kidney, SGLT2 mediates 97% and SGLT1 mediates 3% of fractional renal glucose reabsorption under euglycemic conditions. When SGLT2 is inhibited, SGLT1 can account for about 50% of fractional renal glucose reabsorption because of a significant compensatory capacity. On the basolateral side of the S1/2 segment, glucose exists via GLUT2. (2) SGLT2 and NHE3 show functional interaction and colocalization. High glucose levels inhibit NHE3-dependent bicarbonate reabsorption. It is possible that SGLT2-I increase glucose concentrations to levels where this interaction becomes physiologically relevant. (3) SGLT2-I increase phosphate (P_i) reabsorption, causing blood P_i levels to increase; consequently, PTH and FGF-23 levels increase and 1,25(OH)2D levels are reduced. The sodium P_i transporter Npt2 is expressed in the proximal tubule and mediates P_i reabsorption. The identity of the basolateral exit transporter(s) for P_i is unknown. In response to SGLT2-I, an increase in renal urate excretion is observed. Renal urate transport is a complex process of filtration, reabsorption, and secretion, and not all involved mechanisms are illustrated. In addition, species-specific differences exist. In mice, GLUT9 is expressed in the proximal tubule (at extremely low levels) and distal tubule, where it is highly expressed in the apical and basolateral membrane. In humans, no significant portion of urate transport is ascribed to the distal tubule, in contrast to the proximal tubule. On the basolateral side, urate exits via GLUT9 and other unknown transporter(s). (4) URAT1 is responsible for the effects of SGLT2-I on urate excretion in mice. (5 and 6) Of note, GLUT9 seems dispensable for the uricosuric effect of SGLT2-I, independent of its localization. (7) In the distal tubule, magnesium homeostasis is maintained via TRPM6. Treatment with SGLT2-I results in an increase in plasma magnesium levels. The exact mechanism(s) how SGLT2-I increase magnesium reabsorption remain speculative but might involve SGLT2-I induced improvements in insulin resistance and elevations in glucagon; the latter activates a G-protein-coupled receptor, increases cyclic adenosine monophosphate (cAMP) via adenylyl cyclase, and consequently activates protein kinase A which stimulates TRPM6 apical membrane trafficking.

SODIUM-GLUCOSE COTRANSPORTER 2 INHIBITORS – PHOSPHATE

In the proximal tubule, the majority of phosphate (P_i) is reabsorbed under the control of parathyroid hormone (PTH), fibroblast growth factor 23 (FGF-23) and α-klotho [44]. The sodium/P_i cotransporters Npt2a and Npt2c mediate the majority of renal P_i uptake. CKD patients are prone to disturbances in P_i homeostasis and show, dependent on CKD stage, elevated FGF-23, PTH, and hyperphosphatemia [45]. Phloridzin, an unselective SGLT inhibitor, administration to healthy volunteers resulted in about a four-fold reduction in fractional P_i excretion compared with the control period [46]. In a randomized crossover study in healthy volunteers, canagliflozin treatment increased serum P_i by 16%, plasma FGF-23 by 20%, and PTH by 25%, whereas 1,25-dihydroxyvitamin D decreased by 10% [47^■]. Of note, at the end of the five-day experimental period all differences, except for the increase in PTH, disappeared. Along with increased PTH, a tendency of increased urinary calcium excretion was observed. In another study in patients with T2DM, which by itself already causes diabetic bone disease [48], dapagliflozin treatment showed that serum P_i increased by 9%, PTH increased by 16%, FGF-23 increased by 19%, and 1,25-dihydroxyvitamin D decreased by 12% that persisted over 6 weeks of treatment [49^■]. Consistent with changes in mineral homeostasis, clinical studies identified an increased risk of fractures in patients treated with gliflozins [50]. Mice with the Jimbee mutation (exhibiting similar features like SGLT2 knockout) show reduced bone mineralization; however, fracture resistance was unaffected [51^■■]. At this point, there are still many unknowns about the mechanism(s) how SGLT2 inhibitors affect P_i homeostasis. One theory would be that SGLT2 inhibition increases the electrochemical gradient for Na⁺, which promotes P_i reabsorption via Npt2, leading to hyperphosphatemia [52] and consequently the observed changes in hormones involved in P_i homeostasis (Fig. 1). Further studies are needed to understand the interaction between SGLT2 and Npt2a/c. Will the renoprotective effects of SGLT2 inhibitors balance out the fracture risk? Is the fracture risk a class effect or a compound-specific effect and what will be the effects of long-term treatment? In addition, a better subgroup analysis of patients at risk is warranted.

SODIUM-GLUCOSE COTRANSPORTER 2 INHIBITORS – URATE

Multiple clinical trials described a urate-lowering effect in response to SGLT2 inhibition [53–56]. In 1970, a study in five male volunteers injected with phloridzin showed a doubling of the fractional excretion of urate compared with the control period, being consistent with an effect on tubular transport because the phloridzin effect was greater compared with mannitol or glucose. The latter two indicate that neither osmotic diuresis nor a direct effect on renal urate transport can explain the finding [46]. Consistent with this, another study showed that SGLTs do not seem to transport urate [57]. However, the underlying mechanism for the effect on urate handling has not been fully elucidated but we are beginning to understand the involved mechanisms. In addition to URAT1 [58], the facilitated glucose transporter member 9 (GLUT9) has been identified as a major regulator of urate homeostasis [59,60^■]. Patients with loss-of-function mutations in GLUT9 develop familial renal hypouricemia type 2. Symptoms include hypouricemia, renal urate wasting, and kidney stones [61]. First evidence for the interdependence of these mechanisms comes from SGLT1, SGLT2, inducible tubule-specific GLUT9 and URAT1 knockout mice [62^■■]. These elegant studies provided several lines of evidence that lack or inhibition of SGLT2 increases fractional excretion of urate; lack of SGLT1 enhances the uricosuric effects of canagliflozin; at least in mice, GLUT9 is not required for the uricosuric effects of canagliflozin; and URAT1 mediates the uricosuric effects of canagliflozin (Fig. 1). At this point, it is unclear if SGLT2 inhibitors directly act on URAT1. Interestingly, there seems to be a biphasic effect of SGLT2 inhibitors on urate levels depending on baseline urate levels inasmuch that high urate levels decreased, and low urate levels increased under canagliflozin treatment [63^■]. High serum urate levels were associated with β-cell dysfunction and insulin is known to decrease urinary urate excretion [64,65]. Further, hyperuricemia is associated with an increased risk of CKD [66,67], metabolic syndrome [68,69], hypertension [70,71], and cardiovascular events [72,73]. Whether the beneficial effects of SGLT2 inhibitors on cardiovascular events are a compound of glucose and urate level improvements still needs to be determined.

SODIUM-GLUCOSE COTRANSPORTER 2 INHIBITORS – MAGNESIUM

Hypomagnesemia is closely linked to the development and progression of diabetes mellitus and associated with an increased risk of cardiovascular complications [74^■,75,76]. It was proposed that insulin [77] and glucagon [78] stimulate magnesium reabsorption; of note, PTH was also able to increase magnesium uptake [77] by a cAMP-dependent mechanism via transient receptor potential channel melastatin subtype 6 (TRPM6) [79]. Vice versa, two single-nucleotide polymorphisms (SNPs) in TRPM6 rendered insulin inactive in activating TRPM6. Women carrying these SNPs have a higher likelihood of developing gestational diabetes mellitus [80]. With regards to SGLT2 inhibitors, data from a metaanalysis of randomized controlled trials including 15,309 T2DM patients showed that canagliflozin, dapagliflozin, empagliflozin, and ipragliflozin were associated with an increase in serum magnesium levels, ranging from 0.04 to 0.10 mmol/l [81]. The effects were larger in magnitude in patients with reduced renal function [82]. The possible mechanisms for the increased magnesium reabsorption in response to SGLT2 inhibition include improvement of insulin resistance, volume depletion with possible magnesium reabsorption in the proximal tubule/thick ascending limb, increased glucagon levels (inconsistent finding in clinical studies), and insulin-induced shift from the intravascular to the intracellular space (Fig. 1). Reduced insulin levels in response to SGLT2 inhibition might reverse this shift. A study in three patients with refractory hypomagnesemia showed that canagliflozin, empagliflozin, and dapagliflozin were able to increase serum magnesium levels and reduce fractional excretion of magnesium, possibly indicating that SGLT2 inhibitors could be a novel treatment option for this intractable disease [83^■]. Further studies are required to better define the mechanisms how SGLT2 inhibitors affect magnesium reabsorption and if the small increase in magnesium levels contributes to reductions in cardiovascular events in patients with diabetes mellitus treated with SGLT2 inhibitor.

CONCLUSION

SGLT2 inhibitors reduce renal glucose reabsorption in the proximal tubule. The cardiorenal protective effects of SGLT2 inhibitors are consistently observed across large clinical outcome trials. We are beginning to recognize that in addition to glucosuria there are several accessory effects that might be indirectly related to changes in inhibition of sodium and glucose reabsorption. Notably, the renal handling of phosphate, urate, and magnesium are affected. The uricosuria and elevation in serum magnesium level are proposed as potential causal mechanisms for cardiovascular and renal protection by SGLT2 inhibitor treatment. In contrast, hyperphosphatemia, elevation of FGF-23 and PTH levels, especially in the setting of CKD, would be considered a potential negative effect. Further studies are also needed to determine the functional interaction of SGLT2 and NHE3. Of note, NHE3 might be a ‘master regulator’ of other transport proteins in the proximal tubule. In addition to the described functional interaction with SGLT2, kidney-specific NHE3 knockout diminished Ntp2c expression and Npt2a expression was about 30% lower compared with control mice [37]. Clearly, further studies are needed to determine these functional interactions.

KEY POINTS.

• SGLT2 inhibitors are the new mainstay in the treatment of diabetes mellitus.

• In addition to the inhibition of sodium and glucose transport, the actual characteristics of the transporter, several others transport mechanisms are affected, for example, phosphate, magnesium, and urate.

• In addition to the effects on glucose homeostasis, effects on magnesium and urate might provide additional benefits and possibly contribute to the improved cardiovascular risk reported in clinical outcome trials.

• Additional studies are needed to better understand the mechanisms behind these ‘off target’ effects and if these effects are maintained in populations without diabetes mellitus.