Critical Reanalysis of the Mechanisms Underlying the Cardiorenal Benefits of SGLT2 Inhibitors and Reaffirmation of the Nutrient Deprivation Signaling/Autophagy Hypothesis

Abstract

SGLT2 (sodium-glucose cotransporter 2) inhibitors produce a distinctive pattern of benefits on the evolution and progression of cardiomyopathy and nephropathy, which is characterized by a reduction in oxidative and endoplasmic reticulum stress, restoration of mitochondrial health and enhanced mitochondrial biogenesis, a decrease in proinflammatory and profibrotic pathways, and preservation of cellular and organ integrity and viability. A substantial body of evidence indicates that this characteristic pattern of responses can be explained by the action of SGLT2 inhibitors to promote cellular housekeeping by enhancing autophagic flux, an effect that may be related to the action of these drugs to produce simultaneous upregulation of nutrient deprivation signaling and downregulation of nutrient surplus signaling, as manifested by an increase in the expression and activity of AMPK (adenosine monophosphate–activated protein kinase), SIRT1 (sirtuin 1), SIRT3 (sirtuin 3), SIRT6 (sirtuin 6), and PGC1-α (peroxisome proliferator–activated receptor γ coactivator 1-α) and decreased activation of mTOR (mammalian target of rapamycin). The distinctive pattern of cardioprotective and renoprotective effects of SGLT2 inhibitors is abolished by specific inhibition or knockdown of autophagy, AMPK, and sirtuins. In the clinical setting, the pattern of differentially increased proteins identified in proteomics analyses of blood collected in randomized trials is consistent with these findings. Clinical studies have also shown that SGLT2 inhibitors promote gluconeogenesis, ketogenesis, and erythrocytosis and reduce uricemia, the hallmarks of nutrient deprivation signaling and the principal statistical mediators of the ability of SGLT2 inhibitors to reduce the risk of heart failure and serious renal events. The action of SGLT2 inhibitors to augment autophagic flux is seen in isolated cells and tissues that do not express SGLT2 and are not exposed to changes in environmental glucose or ketones and may be related to an ability of these drugs to bind directly to sirtuins or mTOR. Changes in renal or cardiovascular physiology or metabolism cannot explain the benefits of SGLT2 inhibitors either experimentally or clinically. The direct molecular effects of SGLT2 inhibitors in isolated cells are consistent with the concept that SGLT2 acts as a nutrient surplus sensor, and thus, its inhibition causes enhanced nutrient deprivation signaling and its attendant cytoprotective effects, which can be abolished by specific inhibition or knockdown of AMPK, sirtuins, and autophagic flux.

SGLT2 (sodium-glucose cotransporter 2) inhibitors slow the evolution and progression of heart failure (HF), whether they are administered to high-risk patients without symptoms or to symptomatic patients across a broad range of ejection fractions. In 12 large-scale trials involving >70 000 patients, long-term SGLT2 inhibition produced a consistent 20% to 25% reduction in the combined risk of cardiovascular death or hospitalizations for HF.¹ SGLT2 inhibitors are now considered 1 of the 4 foundational drugs for the management of HF in patients with a reduced ejection fraction.

Potential Actions of SGLT2 Inhibitors as Nephrocentric Neurohormonal Antagonists and as Osmotic Diuretics

The other foundational drugs for HF—angiotensin receptor neprilysin inhibitors, β-blockers, and mineralocorticoid receptor antagonists—interfere with a neurohormonal mechanism whose activation leads to deleterious effects on the myocardium (ie, angiotensin II, catecholamines, aldosterone, and neprilysin). However, an important neurohormonal mechanism that is not fully antagonized by current foundational drugs is the marked increase in sympathetic nerve traffic to the kidneys. Renal sympathetic tone is strikingly enhanced in experimental and clinical HF and adversely influences prognosis.^2,3 An increase in renal sympathetic nerve traffic adversely affects the heart by increasing the release of angiotensin II and neprilysin from the kidneys, both acting to promote adverse cardiac remodeling and fibrosis as well as sodium retention and volume expansion.^4,5 Enhanced renal sympathetic nerve traffic has also been implicated in the progression of chronic kidney disease, in part related to the stimulation of intrarenal angiotensin II and in part related to the activation of α-2 adrenoreceptors, which may play a role in the pathogenesis of renal interstitial inflammation and fibrosis.^6,7 Renal denervation produces favorable effects in experimental HF and renal disease and improves cardiac function in the clinical setting.^4–9 Of note, although they are generally characterized as inhibitors of the sympathetic nervous system, β-blockers do not reduce renal sympathetic nerve traffic.¹⁰

Potential Antagonism of Renal Sympathetic Hyperactivity by SGLT2 Inhibitors

Enhanced renal sympathetic nerve traffic would be expected to promote sodium retention by the kidney. The antinatriuretic effect of renal sympathetic activation is particularly pronounced in the proximal renal tubule¹¹ and proximal tubular hyperreabsorption of sodium is abrogated by renal denervation.¹² The interplay of NHE3 (sodium-hydrogen exchanger isoform 3) and SGLT2 plays a key role in the reabsorption of sodium in the proximal tubule, and increases in renal sympathetic nerve traffic augments the expression of both NHE3 and SGLT2,^13,14 accounting for their increased activity in HF.^14–16

Most diuretics do not address sodium reabsorption in the proximal renal tubule. It is therefore noteworthy that SGLT2 inhibitors interfere with both SGLT2 and NHE3 in the proximal tubule^16,17 and this action yields short-term increases in the fractional excretion of sodium.¹⁸ HF potentiates the natriuretic effects of SGLT2 inhibition (as would be expected by upregulation of SGLT2 and NHE3 in this disorder)^14,16 and renal NHE3 inhibition by SGLT2 inhibitors promotes euvolemia in rats with HF.¹⁶ In addition, SGLT2 inhibition attenuates renal sympathetic activity and reduces renal norepinephrine content in states of experimental nutrient excess.^19,20 Renal denervation attenuates the magnitude of the responses to SGLT2 inhibition in the kidney in animals with (but not without) HF.¹⁴ Taken collectively, these observations position SGLT2 inhibitors as functional antagonists of renal sympathetic nerve hyperactivity in HF. However, any postulated sympatholytic action is likely to be highly selective for the renal nerves, because SGLT2 inhibition does not reduce cardiac sympathetic activity²¹ and has not been shown to decrease central sympathetic outflow.

Potential Benefits of SGLT2 Inhibitors Acting as Osmotic Diuretics

The possibility that SGLT2 inhibitors might treat HF by functioning as antagonists of renal sympathetic nerve hyperactivity or by ameliorating sodium retention is appealing (Figure 1). However, it has been difficult to show that SGLT2 inhibitors exert important natriuretic effects in patients with HF. In this setting, SGLT2 inhibition exerts a short-term osmotic diuretic effect with little increase in urinary sodium excretion,^22,23 suggesting that any increase in urine volume is related to augmented glucose (and not sodium) excretion. However, early increases in urinary water excretion are not sustained because of the activation of adaptive mechanisms that decrease free water clearance,²⁴ explaining why long-term SGLT2 inhibition does not change serum sodium concentration.²⁵

Figure 1.

Proposed framework by which SGLT2 (sodium-glucose cotransporter 2) inhibitors might exert cardioprotective and nephroprotective effects by acting to mute renal sympathetic nerve activity and promote natriuresis and osmotic diuresis. NHE3 indicates sodium-hydrogen exchanger isoform 3.

Additional lines of evidence raise further doubts that a sustained diuretic effect can account for the benefits of SGLT2 inhibitors in HF. SGLT2 inhibitors are not more effective in patients with HF who have volume overload, as compared with those who are euvolemic.²⁵ Although these drugs produce a modest decrease in body weight, this appears to be related to the urinary caloric loss and the shrinkage of fat depots rather than to changes in fluid status^26,27; changes in body weight are poorly correlated with changes in natriuretic peptides.²⁵ The increase in hematocrit seen in clinical trials is a delayed effect that is related to erythrocytosis rather than hemoconcentration.²⁸ Although SGLT2 inhibitors can decrease plasma volume, this represents a compensatory mechanism that is triggered by the increase in red blood cell mass²⁹; if plasma volume were not reduced, erythrocytosis would lead to intolerable hypervolemia.

It has been proposed that the osmotic diuretic effect of SGLT2 inhibitors leads primarily to decreases in interstitial fluid (rather than plasma volume), but this hypothesis has been on the basis of modeling rather than direct measurements.³⁰ Indeed, at a time when the diuretic effects of SGLT2 inhibitors are most prominent (ie, during the first weeks of treatment), changes in cardiac filling pressures and natriuretic peptides are very small.^31–33 A transient diuresis cannot explain the meaningful changes in cardiac geometry³⁴ or the striking reduction in the risk of major adverse renal events seen in patients with HF with a reduced ejection fraction.³⁵ Any improvement stemming from an osmotic diuretic action of these drugs would be expected to be potentiated in individuals with diabetes (who exhibit the most marked glycosuria), but patients with HF are not more likely to respond to SGLT2 inhibitors if they are diabetic.^35,36 Conversely, preexisting renal impairment would be expected to attenuate the glycosuric and osmotic diuretic actions of SGLT2 inhibitors; however, chronic kidney disease does not diminish the effect of SGLT2 inhibitors to reduce hospitalizations for HF.^26,37

Enthusiasm for the hypothesis that SGLT2 inhibitors act primarily as nephrocentric neurohormonal antagonists or as diuretics has been driven by the belief that the efficacy of these drugs should be linked to their actions within the kidney, because SGLT2 is expressed primarily in the proximal renal tubule. A diuretic effect of SGLT2 inhibitors may underlie some of the short-term benefits of these drugs, and the enhanced distal delivery of sodium may also promote the exchange of sodium for potassium, thus enhancing kaliuresis and mitigating the risk of hyperkalemia.³⁸ Aside from these actions, renal sympatholysis or increases in urinary volume do not appear to contribute importantly to the long-term ability of SGLT2 inhibitors to reduce the risk of HF or renal events.

Potential Importance of Mitigation of Glomerular Hypertension by SGLT2 Inhibitors

Glomerular hyperfiltration has long been regarded as the mechanism driving the progression of kidney disease in patients with diabetes. The benefits of angiotensin receptor blockers and mineralocorticoid receptor antagonists on the evolution of diabetic nephropathy have been attributed to their effect to decrease intraglomerular pressures. However, the role of glomerular hyperfiltration in the progression of kidney disease in HF has been questioned. Angiotensin-converting enzyme inhibitors and mineralocorticoid receptor antagonists do not slow the loss of nephrons in patients with HF.^39,40 Furthermore, neprilysin inhibition reduces the risk of major renal events in patients with HF, even though the afferent arteriolar vasodilation produced by enhanced cyclic guanosine monophosphate signaling increases intraglomerular pressures and albuminuria.⁴¹

Doubts about the role of glomerular hyperfiltration in the progression of renal disease in patients with HF have clouded efforts to understand the nephroprotective effects of SGLT2 inhibitors in this disorder. By inhibiting proximal tubular sodium hyperreabsorption, SGLT2 inhibitors can rapidly reduce intraglomerular pressures, although this may not be related to enhanced tubuloglomerular feedback.⁴² SGLT2 inhibition produces an immediate decrease in glomerular filtration rate in patients with HF.³⁵ The magnitude of the early decline in glomerular filtration is attenuated in patients with renal impairment⁴³; however, the ability of SGLT2 inhibitors to slow the progression of renal disease is not diminished.³⁷ It is therefore noteworthy that experimental knockout of SGLT2 is sufficient to attenuate glomerular hyperfiltration, but it does not prevent renal injury, inflammation, or fibrosis in experimental diabetes or ischemia.^44,45 An action of SGLT2 inhibitors to reduce intraglomerular pressures may not be relevant to their renoprotective effects in HF.

Potential Actions of SGLT2 Inhibitors to Promote Energy Substrate Delivery

In the absence of a nephrocentric explanation for the benefits of SGLT2 inhibitors, investigators have focused on 2 prominent physiologic features of these drugs: ketogenesis and erythrocytosis. Both ketonemia and an increased red blood cell mass might augment delivery of efficient fuels and oxygen, both of which may have favorable effects on the energy state of hearts and kidneys under stress (Figure 2).

Figure 2.

Proposed framework by which SGLT2 (sodium-glucose cotransporter 2) inhibitors might act to increase delivery of substrates that could lead to enhanced synthesis of ATP (adenosine triphosphate).

Potential Role of Ketone Bodies Acting as a Fuel for the Heart and Kidneys

By promoting glycosuria, SGLT2 inhibitors trigger a state of starvation mimicry that is characterized by the production of ketone bodies—particularly β-hydroxybutyrate—in the liver.⁴⁶ Although the infusion of large doses of β-hydroxybutyrate exerts positive inotropic and chronotropic effects,⁴⁷ SGLT2 inhibition does not produce these hemodynamic effects in patients or in isolated cardiomyocytes.^33,34,48 Furthermore, circulating ketone bodies are increased in patients with HF,⁴⁹ and the failing heart already uses ketone bodies as a preferred fuel.⁵⁰ In patients without diabetes, the degree of ketonemia is muted,^36,51 but the HF benefits are not.³⁵

Furthermore, under experimental conditions, SGLT2 inhibitors have not consistently enhanced ketone body consumption by the myocardium,^52–54 and increases in ATP (adenosine triphosphate) production seen after SGLT2 inhibition are not related to increased ketone body metabolism.⁵² Augmentation of ketone body oxidation is not beneficial in experimental HF,⁵⁵ and clinically, ketogenesis after SGLT2 inhibition does not consistently yield increases in myocardial ATP.^54,56 Similarly, the nephroprotective actions of SGLT2 inhibitors cannot be directly linked to ketogenesis-related increases in ATP.⁵⁷

Potential Role of Erythrocytosis to Increase Tissue Oxygen Delivery

It has been hypothesized that the increased tubular workload related to increased distal sodium delivery after SGLT2 inhibition might produce renal medullary hypoxia,⁵⁸ thus triggering the synthesis of erythropoietin and an increase in hemoglobin. However, magnetic resonance imaging has not discerned evidence for the induction or alleviation of renal hypoxia in patients treated with SGLT2 inhibitors,⁵⁹ suggesting that the increase in erythropoietin is likely related to an effect on hypoxia-inducible factors.⁶⁰

Might the increase in red blood cell mass improve tissue oxygenation and thereby reduce HF or renal events? Oxygen utilization is not typically impaired in hemodynamically stressed hearts, but oxygen delivery still might be important in patients with coronary artery disease. However, the benefits of SGLT2 inhibitors in HF are not enhanced by the presence of ischemic heart disease.³⁵ Furthermore, when red blood cell mass is increased by means of erythropoietin-mimetic agents in large-scale trials, the risk of major HF events or of progression of renal disease is not diminished, even though patients were anemic at the start of treatment.^61,62 Therefore, the totality of evidence does not suggest that the cardioprotective or renoprotective effects of SGLT2 inhibitors can be explained by enhanced delivery of energy substrates.

Potential Actions of SGLT2 Inhibitors to Inhibit Sodium-Hydrogen Exchange

The function of sodium-glucose cotransporters and sodium-hydrogen exchangers are closely intertwined in several organs. The actions of SGLT1 (sodium-glucose cotransporter 1) and NHE3 are linked in the intestinal mucosa⁶³ and SGLT2 and NHE3 are colocalized in the proximal renal tubule,⁶⁴ such that inhibition of SGLT2 also impairs the actions of NHE3.^16,17 It is therefore possible that SGLT2 inhibitors might interfere with other sodium-hydrogen exchangers (eg, NHE1 [sodium-hydrogen exchanger isoform 1]). Experimental activation of NHE1 in the heart can promote hypertrophy, cell death, and the development of HF,⁶⁵ making it an attractive therapeutic target.

Baartscheer and colleagues^66,67 first proposed that SGLT2 inhibitors might function as NHE1 antagonists in the heart; they postulated that these drugs could exert direct benefits on cardiomyocytes, independent of renal glycosuria or any interaction with SGLT2, because SGLT2 is not expressed in the myocardium. Incubation of rat cardiomyocytes with SGLT2 inhibitors delayed pH recovery after an ammonium ion (NH₄+) pulse and reduced intracellular sodium concentrations, actions consistent with NHE1 inhibition, and the investigators reported a docking site for SGLT2 inhibitors on the NHE1 protein. The effects on cardiomyocytes were attenuated by cariporide, a canonical NHE1 antagonist.

Despite these observations, a role for NHE1 inhibition in mediating the cardiovascular benefits of SGLT2 inhibitors remains in doubt. Chung et al.⁶⁹ reported no effect of SGLT2 inhibitors to delay pH recovery after a NH₄+ pulse or to reduce intracellular sodium concentrations, thus challenging the findings of Baartscheer and colleagues.^66,67 The published interchange between the 2 groups revealed that the data of Chung et al.⁶⁹ were derived from a larger sample size.^68,69 Osaka et al.⁷⁰ claimed that SGLT2 inhibition inhibited NHE1, but the effect was substantially less than that seen with cariporide. Li et al.⁴⁸ failed to show meaningful docking of empagliflozin to NHE1 and reported that cariporide did not mimic the action of the SGLT2 inhibitor on isolated cardiomyocytes. Therefore, if there is an effect of SGLT2 inhibitors to blunt increases in intracellular sodium, it may be secondary to their effects to reduce cellular stress or to an action on the late sodium current.^60,71

Actions of SGLT2 Inhibitors to Promote Nutrient Deprivation Signaling and Autophagy to Reduce Cellular Stress and Promote Cellular Survival

The key clinical biomarkers of the action of SGLT2 inhibitors—ketogenesis and erythrocytosis—reflect the typical responses to nutrient and oxygen deprivation.⁶⁰ SGLT2 inhibitors also reduce serum uric acid, an indicator of oxidative stress in patients with HF.⁷² How are these 3 biomarkers related to each other? Enhanced nutrient and oxygen deprivation signaling can promote erythropoietin synthesis and reduces oxidative stress in cardiomyocytes and renal parenchymal cells.⁶⁰ It is therefore noteworthy that mediation analyses of large-scale outcomes trials in patients with type 2 diabetes have consistently identified increases in hemoglobin and decreases in uric acid as statistical determinants of the ability of SGLT2 inhibitors to reduce the risk of HF hospitalizations and major adverse renal events.^73–75 These clinical findings, taken together with a wealth of data from experimental studies, have led to the hypothesis that the cardiorenal benefits of SGLT2 inhibitors are related to the activation of nutrient deprivation signaling, with its attendant effects to promote autophagy and mitochondrial health, reduce the generation of reactive oxygen species, mute inflammation and fibrosis, and enhance the viability of cardiomyocytes and renal parenchymal cells.⁶⁰ Although this response may be triggered by an action of SGLT2 inhibitors to induce a state of starvation mimicry secondary to the urinary loss of calories,⁷⁶ numerous studies published during the past 2 years have shown that the adaptive cellular reprogramming produced by these drugs is seen in isolated cell cultures, indicating that SGLT2 inhibitors have direct glycosuria-independent actions to reduce cellular stress and promote cellular survival.

Nutrient Sensors and Cellular Signaling in the Heart and Kidney in Health and Disease

SGLT2 protein functions as an energy sensor, discerning an excess nutrient state when there is excessive glucose in the proximal renal tubules, and changes in SGLT2 parallel changes in other energy sensors in response to shifts in environmental nutrients.⁷⁷ When faced with changes in extracellular glucose and amino acids, the interplay of several master switches adapts the cell to promote its growth or survival. These include mTOR (mammalian target of rapamycin); sirtuins (SIRT1 [sirtuin 1], SIRT3 [sirtuin 3], and SIRT6 [sirtuin 6]); and AMPK (adenosine monophosphate–activated protein kinase; Figure 3).

Figure 3.

Effect of nutrient deprivation and nutrient surplus signaling on the evolution and progression of cardiomyopathy and nephropathy in experimental and clinical settings. Akt indicates protein kinase B; AMPK, adenosine monophosphate–activated protein kinase; mTOR, mammalian target of rapamycin; PGC-1α, peroxisome proliferator–activated receptor γ coactivator 1-α; SIRT1, sirtuin 1; SIRT3, sirtuin 3; and SIRT6, sirtuin 6.

Mammalian Target of Rapamycin

mTOR is a serine/threonine protein kinase that is activated by a surplus of environmental amino acids and functions to promote cell growth and proliferation. mTOR exists in 2 complexes—mTOR complex 1 (mTORC1) and mTOR complex 2—and Akt (protein kinase B) potentiates the activation of mTORC1, which is preferentially inhibited by rapamycin. Akt/mTORC1 signaling influences hundreds of downstream effectors to promote anabolic pathways, driving the mitochondrial production of reactive oxygen species to facilitate cellular replication, proinflammatory pathways, and innate immunity, and enhancing the expression of the senescence-associated secretory phenotype that is essential to the cellular disposal required for effective organ growth.⁷⁸ The action of mTOR activity to promote anabolic pathways is required for cardiomyocyte replication during fetal development and adaptive hypertrophy during acute massive pressure overload,^79,80 but it leads to maladaptive cardiac remodeling when activated in hearts that are injured in adulthood. In experimental models, cardiac-specific overactivation of the Akt/mTOR pathway induces HF,⁸¹ whereas suppression of Akt or mTOR signaling ameliorates the development of cardiomyopathy.^82,83 Similarly, Akt/mTOR signaling is hyperactivated in the kidney in nutrient surplus states, thus triggering proinflammatory pathways and promoting renal injury,^84–86 whereas inhibition of mTOR can ameliorate fibrosis and the development of chronic kidney disease.^86,87

In the clinical setting, patients with dilated cardiomyopathy show aberrant myocardial activation of mTORC1, the intensity of which is associated with the severity of cardiac fibrosis and a poor prognosis.⁸⁸ Activation of Akt in the human myocardium characterizes the transition from well-compensated left ventricular hypertrophy to decompensated HF.⁸⁹ mTOR-activating sequence variations can cause clinical cardiomyopathy and kidney tubulopathy⁹⁰ and renal mTORC1 activation is associated with disease activity and prognosis in patients with nondiabetic kidney disease.⁹¹

Sirtuin 1 and Other Sirtuins

Sirtuins are a family of redox-sensitive nicotinamide adenine dinucleotide (NAD)–dependent deacetylases that catalyze the posttranslational modification of hundreds of proteins that are involved in metabolism and cellular homeostasis. They serve as a redox rheostat and represent the primary cellular response to glucose deprivation. SIRT1, SIRT3, and SIRT6 share similar functions, with SIRT1 and SIRT6 being localized to the nucleus and SIRT3 to the mitochondria. Cardiac-specific deletion or suppression of SIRT1, SIRT3, and SIRT6 augments production of reactive oxygen species, enhances endoplasmic reticulum stress, and sensitizes the heart to injury, leading to cardiomyopathy.^92–94 Conversely, SIRT1, SIRT3, and SIRT6 enrichment or activation alleviates oxidative and endoplasmic reticulum stress, disposes of dysfunctional mitochondria and promotes mitochondrial biogenesis, mitigates cell senescence and death, ameliorates fibrosis and remodeling, and preserves cardiac function.^95–98 Deficiencies in SIRT1, SIRT3, and SIRT6 exacerbate glomerular injury and glomerulosclerosis after renal stress,^99–101 and SIRT1, SIRT3, and SIRT6 activation is accompanied by reduction in oxidative stress, tubulointerstitial inflammation, and fibrosis; amelioration of glomerular and tubular damage; and maintenance of renal function.^102–104 The adaptive effects of sirtuins on organellar health and cellular stress are potentiated by its downstream effector (eg, PGC-1α [peroxisome proliferator–activated receptor γ coactivator 1-α]), which plays a key role in mitochondrial biogenesis, and PGC-1α has been implicated in the pathogenesis of experimental cardiomyopathy and nephropathy.^105,106

In the clinical setting, patients with cardiomyopathy show suppressed expression of SIRT1, SIRT3, and PGC-1α^107,108 and point sequence variations in the genes for SIRT3 and PGC-1α.^109,110 Loss-of-function polymorphisms in the SIRT1 gene increase the risk of chronic kidney disease in patients with type 2 diabetes.¹¹¹ In patients with diabetes, the glomerular expression of SIRT6 is suppressed¹⁰¹; serum SIRT1 levels are decreased, with a decline that parallels the development of albuminuria¹¹²; and there is an inverse relationship between the expression of SIRT1 and SGLT2 in the human diabetic kidney.⁷⁷

Adenosine Monophosphate–Activated Protein Kinase

AMPK is a serine/threonine kinase that discerns the balance between cytosolic levels of ATP and AMP (adenosine monophosphate). AMPK is activated when the ATP-to-AMP ratio is low and phosphorylates downstream proteins that promote increased catabolism and decrease anabolism, thereby augmenting ATP production.¹⁹⁹ Knockout or suppression of AMPK diminishes the heart’s ability to respond to stress, promotes cardiac aging, and leads to cardiomyopathy.^113,114 Activation of AMPK ameliorates oxidative stress, mitochondrial dysfunction, proinflammatory pathways, fibrosis, and apoptosis and preserves ventricular function during cardiac stresses produced by ischemia, diabetes, pressure overload, or cardiotoxic agents.^115–117 AMPK activity is impaired in the diabetic kidney, which contributes to the development of nephropathy.¹¹⁸ Activation of AMPK promotes glomerular health and podocyte survival and reduces tubular injury, apoptosis, and fibrosis triggered by metabolic stresses, ischemia, and nephrotoxic drugs.^119–121

In the clinical setting, patients with sequence variations in the PPKAG2 gene (which codes for an AMPK subunit) develop cardiomyopathy with atrial and ventricular arrhythmias.¹²² Sequence variants of PPKAG2 are also linked to accelerated decline in glomerular filtration rate and the development of kidney disease in the general population and in patients with diabetes.^123,124

These 3 master regulators—mTOR, sirtuins, and AMPK—modulate changes in cellular biology that allow organisms to adapt to environmental opportunities and challenges. When nutrients are plentiful, organisms prioritize the use of fuels to expand cell mass and mTOR signaling is central to this process. In contrast, when nutrients are depleted, organisms mute anabolic pathways and adopt a sheltered set of biological conditions to preserve the structural and functional integrity of existing cells; the sirtuins, PGC-1α, and AMPK are critical to this response. The set point for the interplay of these master switches is determined by the level of nutrients and the redox state.¹²⁵

Of note, SIRT1/PGC-1α and Akt/mTOR pathways are highly interconnected at a molecular level. SIRT1 and PGC-1α can negatively regulate the transcription of Akt and directly interfere with the actions of Akt and mTOR.^126,127 Conversely, upregulation of Akt leads to suppression of PGC-1α, whereas inhibition of mTOR by rapamycin or Akt downregulation leads to activation of SIRT1 and PGC-1α.^128,129 The effect of AMPK to promote NAD+ leads to SIRT1 activation¹³⁰ and AMPK can activate PGC-1α by phosphorylation.¹³¹ Conversely, AMPK can inhibit mTOR by an action on Akt as well as through a direct effect on the mTORC1 complex.¹³²

Action of Nutrient Sensors to Influence Cellular Stress by Modulation of Autophagy

The interplay of the actions of the mTOR, sirtuin, and AMPK master switches to reduce cellular stress and promote cellular survival may be primarily mediated by their actions to influence the cellular housekeeping process of autophagy. Autophagy is an evolutionarily conserved intracellular degradative pathway, which involves the encircling of unwanted or dangerous cellular components by a double-membrane vesicle (the autophagosome), whose fusion with the lysosome allows degradative enzymes to destroy the contents of the vesicle.^133,134 The primordial stimulus to autophagy is nutrient deprivation, and when activated nonselectively, autophagy allows for recycling of cellular constituents, which generates ATP for energy-starved cells.¹³⁵ However, autophagic flux also can be activated selectively in response to a broad range of cellular stresses, including oxidative and endoplasmic reticulum stress. The most important sources of oxidative stress are dysfunctional mitochondria and peroxisomes and their clearance by autophagy is referred to as mitophagy and pexophagy. Endoplasmic reticulum stress is typically caused by the accumulation of misfolded proteins or glucose or fatty acid intermediates that may result from a variety of cellular injuries. Autophagic clearance of these damaged cellular constituents markedly reduces cellular stress and autophagic flux also directly acts to mitigate proinflammatory and profibrotic responses.¹³⁶ The overall effect of enhanced autophagic flux is to preserve cellular integrity and prevent apoptotic cell death, thus maintaining and restoring organ structure and function.

Autophagic flux is driven by the interplay of nutrient deprivation and surplus sensors. The formation of the autophagosome is initiated by the concerted interaction of the ULK1 (Unc-51-like kinase 1) complex with the Beclin 1–PI3KC3 (class III phosphatidylinositol 3-kinase) complex.^134,135 AMPK activates ULK1, but ULK1 phosphorylation by mTORC1 prevents its activation. Beclin 1 is activated when deacetylated by SIRT1¹³⁷ and TLR9 (Toll-like receptor 9) interacts with Beclin 1 to regulate the assembly of the PI3KC3 complex.¹³⁸ Elucidation of the numerous influences of mTOR, sirtuins, and AMPK on the components of the autophagic machinery is beyond the scope of this review.

The adaptive actions of autophagy are particularly important in the heart and kidneys. Both the heart and kidneys are replete with mitochondria and peroxisomes, which underlie their enormous capacity to consume oxygen and generate reactive oxygen species. Amelioration of oxidative and endoplasmic reticulum stress is particularly important to cardiomyocytes, because nonproliferating cells cannot replace cells that have died. Chronic HF and chronic kidney disease is characterized by the accumulation of intracellular debris and deleterious metabolic intermediates, a marked increase in oxidative stress, and the activation of proinflammatory and profibrotic signals.⁶⁰ Autophagic flux represents a major adaptive and restorative response; impairment of autophagy dramatically heightens the likelihood that an injurious event will lead to cardiomyopathy and nephropathy.^139,140 Although excessive stimulation of autophagy may exert deleterious effects on the heart (particularly after acutely imposed massive stresses in the absence of HF),^141,142 measured augmentation of selective autophagy protects the heart against pressure overload, hypoxia, and injury produced by cardiotoxic agents,^143,144 and shields podocytes and renal tubular cells from damage caused by diabetes, ischemia, and nephrotoxic drugs.^145,146

Autophagic vacuoles are increased in patients who show reverse cardiac remodeling or receive mechanical unloading,^147,148 whereas persistent impairment in autophagy flux in the failing human heart is a poor prognostic sign.^149,150 Similarly, autophagic function is blunted in renal tubular cells of patients with diabetic nephropathy¹⁵¹ and loss of autophagosomes is a marker of chronic graft dysfunction in renal transplant patients.¹⁵² By contrast, increased autophagosome density in podocytes presages the preservation of glomerular function and a reduced risk of progression in patients with kidney disease.^153,154

Effects of SGLT2 Inhibitors on Nutrient Deprivation and Surplus Signaling, Autophagic Flux, and Cellular Stress in Experimental Cardiomyopathy and Nephropathy

SGLT2 inhibitors produce a highly distinctive pattern of favorable effects on the evolution and progression of cardiomyopathy and nephropathy in experimental animals (Figure 4). First, SGLT2 inhibitors consistently enhance autophagic flux (including mitophagy) in the heart and kidney; the action to augment autophagy is seen in tissue harvested from chronically treated animals as well as in isolated cell cultures perfused with SGLT2 inhibitors (Table 1).^155–173 Second, SGLT2 inhibitors mute the generation of reactive oxygen species, enhance antioxidant mechanisms, and mitigate endoplasmic reticulum stress.^{165,167,169,174–176} Third, SGLT2 inhibitors accelerate the disposal of injured mitochondria, restore healthy mitochondrial function, and promote mitochondrial biogenesis^167,169; the increase in mitochondrial mass is a characteristic feature of the action of these drugs on electron microscopy.¹⁵⁶ Fourth, SGLT2 inhibitors reduce inflammation and fibrosis by interfering with the activation of NF-κB (nuclear factor kappa B) and the NLRP3 (NLR family pyrin domain containing 3) inflammasome, thus reducing the synthesis of proinflammatory cytokines; they also mute profibrotic pathways, fibroblast proliferation, and the deposition of collagen.^177,178 Fifth, SGLT2 inhibitors preserve cellular functions and integrity, prevent the loss of cells by apoptosis and senescence, and act to maintain normal tissue architecture, leading to the prevention of adverse structural remodeling and the restoration of organ function.^{34,48,54,167,179} This characteristic pattern of cellular and tissue responses has been observed with diverse SGLT2 inhibitors in experimental cardiomyopathy and nephropathy, regardless of the triggering mechanism.

Figure 4.

Proposed framework by which SGLT2 (sodium-glucose cotransporter 2) inhibitors can modulate nutrient deprivation signaling and thereby enhance autophagic flux and reduce cellular stress. AMPK indicates adenosine monophosphate–activated protein kinase; mTOR, mammalian target of rapamycin; and PGC-1α, peroxisome proliferator–activated receptor γ coactivator 1-α.

Table 1.

Experimental Studies on the Effect of SGLT2 Inhibitors on Autophagic Flux

The distinctive pattern of cellular and tissue benefits of SGLT2 inhibitors in the heart and kidney is consistently accompanied by simultaneous upregulation of nutrient deprivation signaling and downregulation of nutrient surplus signaling. The effect of SGLT2 inhibitors to reduce oxidative stress, enhance mitochondrial integrity, mute inflammatory pathways, and preserve cell viability is paralleled by an increase in expression or activity of AMPK, SIRT1, SIRT3, SIRT6, and PGC-1α and decreased activation of mTOR in diverse tissues under stress, particularly in experimentally induced cardiomyopathy and nephropathy (Figure 4 and Table 2).^{48,54,57,76,77,87,110,157,158,161–171,174–219} The increase in AMPK and autophagic flux also explains the increase in ATP production that is seen in the heart and kidneys, both experimentally and clinically, after SGLT2 inhibition.^{52,54,57,181,220}

Table 2.

Experimental Studies on the Effect of SGLT2 Inhibitors on Nutrient Deprivation and Surplus Signaling

When the concept of SGLT2-enhanced nutrient deprivation signaling and autophagic flux was first proposed,⁶⁰ there were only scattered supportive observations. However, during the past 2 years, nearly 60 additional studies have been published, using a broad range of experimental conditions and methods, and all have confirmed the original hypothesis (Tables 1 and 2). Changes in nutrient deprivation and surplus signaling produced by SGLT2 inhibitors have been noted both in harvested whole organs and in isolated cell cultures and have been seen in tissues that express SGLT2 (eg, proximal renal tubules) and in those that do not express SGLT2 (eg, heart). They also have been demonstrated in the endothelium, liver, and adipose tissue. Most importantly, in a large proportion of these reports, when the effects of SGLT2 inhibitors to promote activation of AMPK, SIRT1, SIRT3, and SIRT6 were attenuated by specific pharmacologic inhibition or knockdown of nutrient deprivation signaling pathways or when their ability to promote autophagy has been genetically or pharmacologically blocked, the beneficial effects of SGLT2 inhibitors to mute oxidative stress, promote mitochondrial health, attenuate inflammation and fibrosis, and maintain cell viability and organ integrity have been abolished (Tables 1 and 2). These experimental observations, taken together, provide compelling evidence that the cardioprotective and nephroprotective effects of SGLT2 inhibitors can be explained by their actions to modulate nutrient and energy sensors and their downstream effect to promote autophagy.

Are nutrient deprivation pathways activated when SGLT2 inhibitors are used in the clinical setting? AMPK, SIRT1, SIRT3, and PGC-1α are the cornerstone of the body’s response to starvation, and thus, their synthesis in the liver is enhanced when there is substantial loss of calories in the urine (as with SGLT2 inhibition).^76,210 After the induction of glycosuria, the activation of SIRT1, SIRT6, and PGC-1α supports blood glucose by promoting gluconeogenesis^76,221—a distinctive metabolic signature of SGLT2 inhibitors (demonstrated clinically), which limits their antihyperglycemic response.^201,222 Activation of AMPK and PGC-1α promotes fatty acid oxidation and increased synthesis of SIRT1 and SIRT3 stimulates 3-hydroxy-3-methylglutaryl CoA synthase, the rate-limiting enzyme for ketone body formation.²²³ As a result, AMPK, SIRT1, SIRT3, and PGC-1α act in concert to promote ketogenesis in the liver and the clinical finding of ketogenesis during starvation and after SGLT2 inhibition reflects enhanced nutrient deprivation signaling. In addition, SIRT1 activates HIF-2α (hypoxia-inducible factor 2α),²²⁴ the master regulator of the production of erythropoietin and stimulus to erythrocytosis. The action of SIRT1 to reduce oxidative stress can also contribute to the decrease in uric acid seen in patients treated with SGLT2 inhibitors⁷² and SIRT1 can directly promote uric acid secretion.²²⁵ The simultaneous finding of ketogenesis, erythrocytosis, and reduced uricemia in clinical trials with SGLT2 inhibitors suggests that nutrient deprivation signaling is upregulated in patients who are treated with these drugs. Increased AMPK and SIRT1 activity may also directly enhance tubuloglomerular feedback,^226,227 thereby linking changes in these energy sensors to the effect of SGLT2 inhibitors to rapidly reduce glomerular filtration rate in the clinical setting.³⁵

A recent proteomics analysis of blood samples collected from >1100 patients with HF before and after treatment with placebo or empagliflozin identified a limited number of proteins whose serum levels were significantly changed as a result of SGLT2 inhibition.²²⁸ Of the 21 differentially increased proteins with known effects on the heart and kidneys, 6 were linked to the promotion of autophagy in the heart, kidneys, or endothelium, and 4 proteins (IGF-1 [insulin-like growth factor binding protein 1], TfR1 [transferrin receptor protein 1], erythropoietin, and AFABP [adipocyte fatty acid binding protein 4]) are known to be upregulated by enhanced SIRT1 signaling.^{224,229–231} The latter proteins were also noted to be differentially increased by Ferrannini et al.²³² in a second proteomics analysis of blood collected from patients with type 2 diabetes. Neither of the 2 proteomics analyses performed direct measurements of SIRT1, SIRT3, SIRT6, PGC-1α, or phosphorylated AMPK, but it is not clear that changes in blood levels of these proteins have the capacity to reflect the importance of their intracellular actions accurately.

Mechanism of Cellular Benefits of SGLT2 Inhibitors in Organs Without SGLT2 Expression

How can SGLT2 inhibitors promote nutrient deprivation signaling and autophagy with their attendant favorable effects on cellular stress, homeostasis, and survival in tissues that do not express SGLT2? There are 3 possibilities: glycosuric caloric loss, ketogenesis-induced autophagic flux, and direct nutrient-independent cellular effects.

Glycosuric Caloric Loss

SGLT2 inhibitors induce a substantial loss of calories in the urine and the resulting glucose depletion can stimulate a system-wide upregulation of nutrient deprivation signaling in organs throughout the body, a shift that does not depend on the presence or absence of SGLT2 in a specific tissue.^76,210 Because the action of SGLT2 inhibitors to promote gluconeogenesis through SIRT1 limits the reduction in blood glucose, glycosuria serves to shift the set point between SIRT1 signaling and glycemia, and as a result, the upregulation of nutrient deprivation signaling with SGLT2 inhibitors can occur even when blood glucose is only modestly diminished.⁷⁶ This hypothesis may explain the reported relationship between the magnitude of glycosuria and increases in erythropoietin in patients treated with these drugs.²³³ However, this hypothesis is difficult to reconcile with the fact that chronic kidney disease meaningfully impairs the glycosuric response to SGLT2 inhibitors, but it does not diminish the ability of these drugs to reduce HF and major renal events.^26,37

Ketone Body–Induced Nutrient Deprivation Signaling and Autophagic Flux

SGLT2 inhibitors induce a state of starvation mimicry characterized by ketogenesis^46,51 and the delivery of ketone bodies to organs can exert a direct effect to increase AMPK and decrease mTOR phosphorylation, to increase the expression of PGC-1α, and to promote autophagic flux.^57,234–238 Potentially acting through these mediators, ketonemia has been shown to reduce oxidative stress, mute cellular inflammation, enhance mitochondrial biogenesis, and preserve cardiac and renal function after diverse stresses.^57,234–240 Ketone body supplementation inhibits phenylephrine-induced hypertrophy in experiments²³⁸ and interruption of ketogenesis negates the benefits of SGLT2 inhibitors on the kidney.⁵⁷ These observations suggest that the ketonemia seen after SGLT2 inhibitors use may exert cardioprotective and nephroprotective effects as a result of a direct action to stimulate nutrient deprivation signaling rather than because of an ability to serve as an efficient fuel for the generation of ATP. However, the presence of diabetes attenuates the magnitude of ketonemia, but it does not influence the cardiorenal benefits of SGLT2 inhibitors.^35,36,51 Furthermore, as noted in Tables 1 and 2, SGLT2 inhibitors exert direct benefits on the heart and kidney in isolated cell preparations, which are not subject to fluctuations in environmental ketone bodies.

Direct Glucose- and Ketone-Independent Cellular Effects

SGLT2 inhibitors exert direct cellular effects that are independent of changes in glucose induced by loss of glucose by the kidney or ketone body production in the liver. Numerous studies (Tables 1 and 2) have shown that SGLT2 inhibitors can enhance nutrient deprivation signaling, promote autophagy, maintain mitochondrial health, mute cellular stresses, and mitigate proinflammatory and profibrotic pathways in isolated organs and in cell cultures—experimental conditions where the level of environmental glucose or ketone bodies is held constant. Studies demonstrating the direct effects of these drugs on isolated cardiomyocytes, podocytes, and proximal renal tubular cells have also established that the cardioprotective and renoprotective benefits of these drugs are not dependent on changes in cardiovascular physiology (eg, changes in blood pressure or renal sympathetic tone); on changes proximal renal tubular function (glycosuria or natriuresis); or on changes in the level of a blood or plasma constituent (eg, hemoglobin, glucose, uric acid, or ketones).

How do SGLT2 inhibitors exert direct effects on cardiac and renal parenchymal cells to preserve homeostasis and survival? Many studies have been performed in cells that express SGLT2, either clinically or experimentally (eg, proximal renal tubular cells, endothelial cells, and hepatocytes), but SGLT2 inhibitors exert direct cytoprotective effects in cardiomyocytes that do not express SGLT2 (Tables 1 and 2). It is possible that the concentrations of SGLT2 inhibitors used in certain experiments might interfere with other glucose transporters. Kondo et al.¹⁸⁴ suggested that the cardioprotection produced by canagliflozin was related to SGLT1 inhibition,²⁴¹ but empagliflozin (which does not inhibit SGLT1) produces cytoprotective effects on isolated cardiomyocytes. Li et al.⁴⁸ have proposed that empagliflozin can dock with GLUT1 (glucose transporter 1) and GLUT4 (glucose transporter 4) receptors to inhibit the cellular uptake of glucose, which can trigger the nutrient deprivation signaling and autophagy. An inhibitory effect of SGLT2 inhibitors on GLUT receptors has been proposed by others,^208,242 but has yet to be confirmed. Potential effects of SGLT2 inhibitors on ion transport^67,71 or Ca/calmodulin-dependent kinase²⁴³ cannot explain their effects on autophagy or nutrient deprivation signaling.²⁴⁴

Because SGLT2 functions as a nutrient surplus sensor, it is noteworthy that there exists an inverse relationship between SGLT2 and nutrient deprivation signaling in individual cells.⁷⁷ This observation raises the possibility that SGLT2 inhibitors might bind directly to nutrient sensors to influence their function. Sun et al.¹⁸⁷ reported that canagliflozin binds to mTOR in the same structural domain used by rapamycin. Ying et al.²⁴⁵ demonstrated that phloretin (an inhibitor of SGLT1 and SGLT2) ameliorates cardiomyocyte injury by upregulating SIRT1 and showed that phloretin docks directly with SIRT1 to form a stable complex. Wang et al.¹¹⁰ proposed that empagliflozin acts directly on SIRT3 to promote the formation of a complex of SIRT3, Beclin 1, and TLR9, which enhances autophagic flux. In that study, experimental knockout of either SIRT3 or TLR9 abolished the ability of SGLT2 inhibition to protect the heart from injury by doxorubicin. Wang et al.¹¹⁰ identified 3 patients with dilated cardiomyopathy who underwent myocardial biopsy before and after treatment with empagliflozin for 28 days. Two patients without a SIRT3 variation exhibited the expected enhancement of mitochondrial respiration and expression of TLR9 after treatment with the SGLT2 inhibitor; however, these changes were not seen in 1 patient who had a loss-of-function variation in SIRT3 in the myocardium.¹¹⁰

Conclusions

SGLT2 inhibitors produce a distinctive pattern of favorable effects on the evolution and progression of cardiomyopathy and nephropathy, which is characterized by a reduction in oxidative and endoplasmic reticulum stress, restoration of healthy mitochondrial function and enhanced mitochondrial biogenesis, a decrease in proinflammatory and profibrotic pathways, preservation of cellular integrity and viability, and maintenance of normal organ structure and function. A substantial body of evidence indicates that this characteristic pattern of responses can be explained by the action of SGLT2 inhibitors to promote cellular housekeeping by autophagic flux, a process that is triggered by the effect of these drugs to produce simultaneous upregulation of nutrient deprivation signaling and downregulation of nutrient surplus signaling, as manifested by an increase in the expression and activity of AMPK, SIRT1, SIRT3, SIRT6, and PGC-1α and decreased activation of mTOR in diverse tissues under stress. The distinctive pattern of cardioprotective and renoprotective effects of SGLT2 inhibitors is abolished by specific inhibition or knockdown of autophagy, AMPK, and sirtuins. In the clinical setting, proteomics analyses of blood collected from participants in randomized trials show a pattern of differentially increased proteins that is consistent with enhanced autophagy and SIRT1 activity. Patients treated with SGLT2 inhibitors exhibit augmented gluconeogenesis, ketogenesis, and erythrocytosis and reduced uricemia, the hallmarks of nutrient deprivation signaling and the principal statistical mediators of the ability of SGLT2 inhibitors to reduce the risk of HF hospitalizations and serious renal events in large-scale trials. The action of SGLT2 inhibitor to promote autophagic flux is seen in isolated cells and tissues that do not express SGLT2 and are not exposed to changes in environmental glucose or ketones and may be related to an ability of these drugs to bind directly to sirtuins or mTOR. Changes in renal or cardiovascular physiology, energy use, or metabolism cannot explain the benefits of SGLT2 inhibitors. Substantial experimental work on the molecular and cellular actions of these drugs, when taken together with highly supportive observations in the clinical setting, has provided critical insights into the mechanisms that underlie their remarkable ability to preserve cardiac and renal function in patients whose hearts and kidneys are under stress.

Article Information

Sources of Funding

None.

Disclosures

Dr Packer has served as a consultant for AbbVie, Altimmune, Amarin, Amgen, Ardelyx, AstraZeneca, Boehringer Ingelheim, Caladrius, Casana, CSL Behring, Cytokinetics, Imara, Lilly, Moderna, Novartis, Reata, Relypsa, and Salamandra. The author reports no current or planned financial relationships related to SGLT2 inhibitors or neprilysin inhibition.