Determinants of kidney oxygen consumption and their relationship to tissue oxygen tension in diabetes and hypertension

SUMMARY

1. The high renal oxygen (O₂) demand is associated primarily with tubular O₂ consumption (Q_O2) necessary for solute reabsorption. Increasing O₂ delivery relative to demand via increased blood flow results in augmented tubular electrolyte load following elevated glomerular filtration, which, in turn, increases metabolic demand. Consequently, elevated kidney metabolism results in decreased tissue oxygen tension.

2. The metabolic efficiency for solute transport (Q_O2/T_Na) varies not only between different nephron sites, but also under different conditions of fluid homeostasis and disease. Contributing mechanisms include the presence of different Na⁺ transporters, different levels of oxidative stress and segmental tubular dysfunction.

3. Sustained hyperglycaemia results in increased kidney Q_O2, partly due to mitochondrial dysfunction and reduced electrolyte transport efficiency. This results in intrarenal tissue hypoxia because the increased Q_O2 is not matched by a similar increase in O₂ delivery.

4. Hypertension leads to renal hypoxia, mediated by increased angiotensin receptor tonus and oxidative stress. Reduced uptake in the proximal tubule increases load to the thick ascending limb. There, the increased load is reabsorbed, but at greater O₂ cost. The combination of hypertension, angiotensin II and oxidative stress initiates events leading to renal damage and reduced function.

5. Tissue hypoxia is now recognized as a unifying pathway to chronic kidney disease. We have gained good knowledge about major changes in O₂ metabolism occurring in diabetic and hypertensive kidneys. However, further efforts are needed to elucidate how these alterations can be prevented or reversed before translation into clinical practice.

INTRODUCTION

Renal oxygenation is based on a balance between oxygen (O₂) supply and consumption (Q_O2). Under physiological steady state conditions the O₂ supply to the renal tissues is well in excess of the O₂ demand. Renal O₂ extraction in the healthy kidney is only 10–15%^1,2; in most other organs it is closer to 45%. Under pathological conditions the balance of O₂ supply compared with demand is disturbed due, in part, to the unique arrangement of the renal microvasculature and its diffusive shunting pathways.^3–5 The high O₂ demand is associated with the tubular Q_O2 necessary for solute exchange⁶ and the high rate of aerobic glycolysis.⁷ Even though the kidney is only 0.5% of total bodyweight, it uses approximately 7% of the O₂ consumed by the body.⁸

“High O₂ demand due to high solute exchange”

The vast majority of Q_O2 is due to reabsorption of approximately 99.5% of filtered sodium (Na⁺).⁹ This further drives the cellular and paracellular transport of solutes and water. The amount of filtered Na⁺ is the product of the glomerular filtration rate (GFR) and the plasma Na⁺ concentration. With approximations of GFR of 125 mL/min and plasma Na⁺ concentrations of 140 mmol/L, the amount of Na⁺ reabsorbed is in the order of 1 mol/h. Because there is little variation in plasma Na⁺ concentrations in healthy individuals, renal Q_O2 is related directly to GFR because the latter determines the sodium load. The estimated energy required to reabsorb 1 mol Na⁺ against an electric potential of −70 mV in the cytoplasm and the chemical gradient is approximately 7 kJ (using Faraday’s constant). To give further appreciation of the amount of energy required, this corresponds to lifting 1 mol Na⁺ (≈20 g) to a height of approximately 70 km. To perform such an effort it is obvious that a large amount of ATP is required continuously, meaning a high Q_O2. Most of the ATP produced by the kidney (some 95%) is through aerobic mechanisms,¹⁰ whereas some nephron segments, particularly in the medulla, can use anaerobic metabolism efficiently.⁷

“Q_O2 is a determinant for kidney tissue P_O2”

The ratio between the GFR (Q_O2 related to active transport) and renal blood flow (RBF; O₂ delivery) is the filtration fraction, which does not vary to any large extent in humans under normal physiological conditions.^{11, 12} If the filtration fraction does not vary significantly from day to day, then a near-constant tissue O₂ tension (P_O2) should prevail in the tissue. This infers that the kidney P_O2 will vary primarily based on Q_O2. A deranged Q_O2 with consequences for tissue P_O2 has been suggested as an important parameter leading to kidney dysfunction in several major disease conditions.^13–15 For example, increased kidney metabolism is associated with diabetic nephropathy¹⁶ and diabetes is associated with decreased kidney P_O2 in both animals and patients.^17–22 Fine et al. proposed that initial glomerular injury decreases blood flow through peritubular capillaries and results in decreased P_O2 of the kidney, promoting tubulointerstitial fibrosis and progression to kidney damage.²³ Importantly, chronic tubulointerstitial hypoxia is acknowledged as a common pathway to end-stage renal disease.^24–28 The two pathways may seem contradictory, however both scenarios are plausible and do not exclude each other. The metabolic changes in experimental diabetes occur before the structural changes appear and the loss of capillaries can follow. However, in any kidney damage scenario a loss of capillaries will also lead to tissue hypoxia.

In most tissues an increased demand for O₂ is followed by increased perfusion and delivery of O₂. However, in the kidney increased O₂ delivery (i.e. increased RBF) is likely to increase GFR and, thus, sodium load. This will increase the work load for reabsorption and the beneficial effect on oxygen homeostasis is unclear. In the kidney, both perfusion and tubular transport activity are governed mainly by a number of hormones and substances. As metabolic products, carbon dioxide (CO₂) and protons act as vasodilators in precapillary resistance vessels, thereby increasing perfusion in tissues like skeletal muscle during work. In the kidney, increased perfusion will not necessarily give rise to a similar increase in P_O2 as in most other organs. This is due to the fact that increased perfusion will also increase the filtered load of solutes, which, in turn, will increase the tubular workload (i.e. Q_O2). For example, giving a vasodilator like atrial natriuretic peptide (ANP) to rats does not increase renal oxygenation but, in contrast, reduces both cortical and medullary P_O2.²⁹ This occurs through an elevation in the filtered load of sodium, which, in turn, increases Q_O2 leading to reduced P_O2. Conversely, if Q_O2 is reduced in rats by, for example, giving the loop diuretic furosemide, which reduces sodium transport in the thick ascending limb of the loop of Henle (mTAL), tissue P_O2 increases.^29,30 Furthermore, acute hypotension paradoxically increases medullary P_O2, which can be abolished using furosemide,³¹ suggesting reduced filtered load as a plausible mechanism. This emphasizes the importance of Q_O2 as a determinant for kidney tissue P_O2.

Kidney perfusion and QO₂ are heterogeneous. From this follows that regions of the kidney may be susceptible to hypoxia. Only some 10–15% of renal perfusion transits the medullary region. At the same time, the mTAL is a major O₂-consuming site in this region owing to Na⁺–K⁺–2Cl⁻ transporter-mediated solute uptake. Furthermore, the countercurrent exchange system (i.e. the vasa recta) also contributes to reduced delivery through shunting between closely lying descending and ascending vessels and because they also contain a lower haematocrit than systemic blood.³² Together, these factors will result in a lower P_O2 in the medulla than in the cortex. This is supported by studies showing a P_O2 gradient from the cortex and along the medullary structures.^33–35 Furthermore, high expression of hypoxia inducible factor (HIF)-1α in the medullary region³⁶ is a marker of a low tissue P_O2 in this region.

In 2003, the first report demonstrating kidney tissue hypoxia in diabetic rats was published.³⁵ This finding has since been verified by several international laboratories in animals and humans.^17–22,37 We propose a hypothesis that will provide a mechanistic explanation for the development of chronic kidney disease (CKD) under conditions associated with increased risk of the progressive loss of kidney function (Fig. 1). Briefly, elevated intrarenal angiotensin (Ang) II activates NADPH oxidase via AT₁ receptors to produce superoxide radicals ($\dot{}{\mathrm{O}}_2^{-}$).³⁸ The resulting oxidative stress reduces electrolyte transport efficiency³⁹ and causes mitochondrial uncoupling via activation of uncoupling protein (UCP)-2.⁴⁰ Reduced tubular electrolyte transport efficiency is a result of an intricate interplay of several different mechanisms, including altered paracellular electrolyte permeability, direct effects on Na⁺/K⁺-ATPase, the shift of Na⁺ transport to less-efficient nephron segments and ‘slipping’ of the Na⁺ pumps. The mitochondrial uncoupling is manifested as increased kidney Q_O2. It can be speculated that the increase in Q_O2 serves to maintain adequate ATP production.⁴¹ Both these mechanisms are tightly regulated by oxidative stress and result in increased kidney Q_O2. Because the kidney, in contrast with the brain, for example, does not match increased metabolic demand (i.e. increased Q_O2) with increased O₂ delivery, the result is a mismatch between supply and demand that causes intrarenal tissue hypoxia.³⁵ Normally the cellular response to sustained tissue hypoxia includes a counteracting activation of the HIF system resulting in the transcription of genes involved in angiogenesis, cellular energy production and oxidative stress defence.⁴² However, for currently unknown reasons, HIF is not adequately activated in hypoxic CKD and the expression of HIF-regulated genes, such as erythropoietin, vascular endothelial growth factor or heme oxygenase-1, are not elevated in these kidneys.⁴³ Rather, the hypoxia induces elevated levels of several detrimental profibrotic and proinflammatory molecules such as transforming growth factor (TGF)-β and tumour necrosis factor (TNF)-α. Finally, sustained intrarenal tissue hypoxia results in proteinuria, tubulointerstitial fibrosis and infiltration of immune cells,^44,45 all hallmarks of developing CKD.

Fig. 1.

Unifying hypothesis for the development of chronic kidney disease (see text for details). AngII, angiotensin II; UCP, uncoupling protein; HIF, hypoxia-inducible factor; TGF-β, transforming growth factor-β; TNF-α, tumour necrosis factor-α.

The cascade of events described in Fig. 1 can be initiated at several levels. Hyperglycaemia, hypertension and low salt intake are associated with increased intrarenal AngII levels,^46–48 the uraemic toxin indoxyl sulphate induces $\dot{}{\mathrm{O}}_2^{-}$ production by NADPH oxidase,¹⁴ whereas reduced O₂ content in the inhaled air results directly in reduced intrarenal tissue P_O2.⁴⁹

“Mismatch between metabolic demand and O₂ delivery”

VARIABILITY IN METABOLIC EFFICIENCY DURING TUBULAR REABSORPTION

The kidney exhibits an extraordinarily efficient autoregulation relating to the myogenic and tubuloglomerular feedback, resulting in relative constancy of blood flow and glomerular ultrafiltration.^50–52 Experimental findings suggest that the tubuloglomerular feedback system normally provides exquisite coordination between: (i) kidney blood flow and the load of glomerular filtration; and (ii) tubular reabsorption, predominantly Na⁺, and an accompanying anion. This highly regulated process may maintain a balance of metabolic or O₂-related supply and demand, especially because the kidney P_O2 is low and the dominant tubular segment for reabsorption exhibits an obligate aerobic dependency for its energy supply.^53,54 In support of such a link between filtered load and tubular O₂ requirements for reabsorption, it has been observed that the oscillatory frequency of the tubuloglomerular feedback system is identical to the oscillations in P_O2 within the kidney cortex.⁵⁵ Therefore, it has been logical to describe the metabolic efficiency or Q_O2 as factored by the major ‘work’ of the kidney, namely Na⁺ reabsorption (T_Na).^56,57 Recent evidence shows that the slope of this relationship (Q_O2/T_Na), or the metabolic efficiency of the kidney, is variable among physiological and pathophysiological conditions and that it is affected by a variety of factors, including hormonal influences.^39,57 In fact, the normal index of metabolic efficiency of the kidney (Q_O2/T_Na) is not constant and can increase by over 100%, either acutely or chronically, indicating a marked reduction in the metabolic efficiency of the kidney and indicating a large increase in kidney Q_O2.^57,58 This is usually accompanied by a significant reduction in kidney tissue P_O2. However, this variability in Q_O2/T_Na under physiological and pathophysiological conditions can be caused by multiple mechanisms. It should be noted that in computation of Q_O2/T_Na the fixed cost (i.e. basal metabolism) should not be included. Otherwise Q_O2/T_Na approaches infinity as T_Na is reduced. Let us consider some possibilities prior to examining the evidence for variability in metabolic efficiency:

• a shift in the site of T_Na along the nephron

• altered efficiency of the process of T_Na, such as altered permeability of tight junctions in the epithelium

• loss of passive T_Na or the addition of active Na⁺ reabsorptive processes with the accompanying ATP and O₂ costs

• mitochondrial changes, such as the action of uncoupling proteins that increase Q_O2 without generation of ATP

• reabsorption of molecules other than Na⁺, such as filtered proteins reabsorbed by the proximal tubule

• induction of gluconeogenesis and glucose generation from predominantly lactate, glutamine or glycerol; this could occur under conditions of starvation or even insulin resistance, and possibly as a result of the actions of AngII

• activation of other oxidases, such as NADPH oxidase, which contributes to increases in Q_O2 independent of any changes in T_Na

• nitric oxide (NO) deficiency from a variety of causes

• reductions in phosphorylated AMP-activated protein kinase (AMPK) possibly related to insulin status

• a change in the balance of aerobic metabolism and glycolysis.

The simplest explanation for a major increase in Q_O2/T_Na is a shift in the site of T_Na to more distal sites within the nephron. However, the exact contribution of such a shift to increased O₂ requirements has not been firmly quantified. It is clear the Q_O2/T_Na increases as it moves along the nephron, with the proximal tubule being the most metabolically efficient segment. The exact computation of the metabolic costs for each nephron segment has varied over the years. Early studies estimated that the costs were 0.36 calories/mEq Na⁺ in the proximal tubule, 1.4 calories/mEq Na⁺ in the thick ascending limb (TAL), 2.7 calories/mEq Na⁺ in the distal connecting tubule and 4.6 calories/mEq Na⁺ in the collecting duct.⁵⁹ Later estimates of segmental ratios have been somewhat less varied, with ratios of 1 : 2 : 6 for the proximal : TAL : distal tubule.⁶⁰ There are at least two examples whereby an increase in Q_O2 and a reduction in metabolic efficiency could be explained. in part, by such a shift in the site of Na⁺ reabsorption. Studies by Brezis et al.⁶¹ and De Nicola et al.⁶² have examined the effects of non-selective NO synthase (NOS) blockade. These studies showed a reduction in proximal reabsorption with a shift into the loop of Henle and distal tubule. Since then, studies have clearly shown that NOS blockade markedly increases Q_O2 and increases Q_O2/T_Na. Brezis et al.⁶¹ demonstrated an increase in medullary Q_O2 with NOS blockade. Below, we discuss how NOS blockade exerts other effects that can increase Q_O2 and decrease metabolic efficiency. The contribution of this shift in reabsorption is potentially significant, but has not been quantified precisely. As discussed later, subtotal nephrectomy, a model of CKD, also demonstrates a shift in Na⁺ reabsorption from the more proximal to distal tubular segment and exhibits increased Q_O2, tissue hypoxia and a marked decrease in metabolic efficiency.^57,58 Potential contributors to increased Q_O2 or treatments that have been shown to normalize Q_O2, such as AngII blockade, do not shift the site of T_Na to more proximal nephron segments, making this mechanism less likely to be the sole cause of altered metabolic efficiency.⁵⁷

“Variability in metabolic efficiency along nephron”

Specific factors influencing reabsorptive efficiency

Changes in the structure of the tubule may also decrease the efficiency of T_Na. After periods of kidney ischaemia, an increase in Q_O2 has been documented.^63,64 This could be due to at least two reasons relating to changes in the tight junctions, particularly in the proximal tubule. Not only may back-leak of Na⁺ be increased, but Molitoris et al. have shown that the apical and basolateral locations of the transporters are significantly altered, thereby markedly decreasing the efficiency of vectorial NaCl reabsorption.⁶⁵ Any other process that alters the Na⁺ or anionic permeability of the tubule could potentially reduce the efficiency of reabsorption and increase Q_O2. Evidence for specific mechanisms contributing to changes in the metabolic efficiency of the kidney (that are usually reversible) has been accumulating. For example, circumstances under which there is loss of passive reabsorption of Na⁺ or additional active transport can also increase Q_O2/T_Na. Benzolamide is a carbonic anhydrase inhibitor that decreases proximal tubular reabsorption by approximately 50% and activates tubuloglomerular feedback in the rat.⁶⁶ This effect should shift reabsorption into the distal nephron, but major reductions in T_Na may decrease Q_O2. In fact, Q_O2 increased by >50% despite the major reductions in GFR and T_Na, and Q_O2/T_Na increased by >80% (Fig. 2).⁶⁷ Benzolamide causes a major reduction in proximal tubular luminal pH.⁶⁷ When we applied agents that inhibited proton secretion in the proximal tubule, 5-(N-ethyl-N-isopropyl)-amiloride (EIPA) and adenosine A₁ receptor antagonists, these agents normalized Q_O2 and Q_O2/T_Na. In vitro studies in isolated proximal tubules gave identical results, whereby benzolamide increased Q_O2 and this effect was prevented by inhibition of Na⁺/H⁺ exchanger isoform 3 and proton secretion.⁶⁷ Weinstein et al.⁶⁸ had observed similar findings earlier, but attributed the greater Q_O2 to elimination of the chloride gradients that promoted passive reabsorption. However, this did not explain fully the documented increase in Q_O2 that we observed. We documented an absolute increase in chloride reabsorption by induction of active transport that was due to the marked reduction in luminal pH, which led to activation of chloride/formate exchange and promoted the recycling of formate across the proximal tubular apical membrane.⁶⁹

Fig. 2.

Inhibition of proximal tubular reabsorption by benzolamide (BNZ) increased the cost of Na⁺ reabsorption (T_Na), as determined by the ratio of tubular O₂ consumption (Q_O2)/T_Na (●). Concurrent application of BNZ with either the adenosine A₁ receptor antagonist 1,3-dipropyl-8-cyclopentylxanthine (○) or the sodium/hydrogen exchange blocker 5-(N-ethyl-N-isopropyl)-amiloride (▼) prevented the BNZ-induced increase in Q_O2/T_Na, suggesting that the BNZ-induced increase in oxygen consumption requires proton secretion and luminal acidification. Data are the mean ± SEM. *P < 0.01 compared with control (10 min before drug administration). Reproduced with permission from Deng et al.⁶⁷

There are several other potential mechanisms for which there is even less evidence for a contribution to changes in metabolic efficiency. Uncoupling proteins are expressed in mitochondria and function as proton channels to allow proton leakage back across the inner mitochondrial membrane without creating ATP.⁷⁰ The uncoupling could be a defence system against oxidative stress because superoxide activates UCP-2.⁷¹ Uncoupling proteins could certainly increase Q_O2, possibly as a regulatory event that is completely separate from the function of T_Na. Leakage of protons through UCP results in elevated Q_O2. Although the sensing mechanisms for this remain unclear, it can be speculated that the initially reduced production of ATP elevates Q_O2 to sustain similar ATP production. There is more evidence that other oxidases contribute to Q_O2 sufficiently to contribute to reduced metabolic efficiency. For example, NADPH oxidase is greatly increased during subtotal nephrectomy, possibly due to increased AngII activity, and is normalized following combined therapy with angiotensin-converting enzyme (ACE) inhibitors and AngII receptor blockers (ARB).⁷² The exact molar contribution of NADPH oxidase to the overall increase in Q_O2 has not been quantified exactly, but it is likely that these oxidases make up a significant portion of the increase in Q_O2. Glomerular protein leakage and reabsorption of proteins and peptides by the proximal tubule may also require significant energy and O₂. However, in subtotal nephrectomy studies, we found that inhibition of tubular protein reabsorption did not significantly reduce Q_O2.⁷²

Metabolic changes independent of tubular reabsorption

Another potential contribution to increased Q_O2 and the apparent decrease in metabolic efficiency is the induction of gluconeogenesis.^59,60 This is not an easy contribution to quantify in the in vivo rat kidney except through indirect approaches using blockers of gluconeogenesis. Lactate is reabsorbed and secreted by the tubule, so quantification of lactate used to synthesise glucose requires complex in vivo analysis. Under certain conditions the kidney can rival the liver in its contribution of glucose to the circulation.^59,60 Major gluconeogenesis is usually not the case, but under conditions of starvation and with certain acid–base conditions, glucose is synthesised, usually from either lactate or glutamine, but at a significant cost of ATP and O₂. There are few data regarding the contribution of gluconeogenesis to increased Q_O2 and QO₂/T_Na under normal physiological conditions. We have examined the effects of acute insulin administration in the subtotal nephrectomy model and observed a tendency for Q_O2 to decrease towards normal values. This effect could be related to reductions in gluconeogenesis, but that has not yet been proven (RC Blantz, unpubl. obs., 2012). Although AngII blockade has no observable effect on kidney Q_O2 in the normal rat,⁵⁷ we have found that combined AngII blockade (ARB + ACE inhibitors or ARB + HIF-1 induction) does normalize Q_O2/T_Na in the subtotal nephrectomy model of CKD.^58,72 We have shown that AngII can produce a form of insulin resistance in the kidney⁷³ and that it is possible that under pathophysiological conditions glucose production via gluconeogenesis may be elevated as a by-product of AngII-induced insulin resistance.

“NO a critical modulator of Q_O2”

Nitric oxide and other factors influencing oxygen consumption

Nitric oxide is a critical modulator of tubuloglomerular feedback activity and for the temporal adaptation of tubuloglomerular feedback.⁷⁴ Given the multiple vascular, tubular and metabolic effects of NO, this substance has been a logical candidate for modulation of the metabolic efficiency of kidney function. Laycock et al.⁷⁵ have demonstrated in the dog that application of non-selective NOS inhibitors produces major increases in kidney Q_O2. Koivisto et al.⁷⁶ also demonstrated an effect of NO on Q_O2 in isolated kidney proximal tubules in vitro. We have recently demonstrated an important effect of NO derived from NOS-1 in both the in vivo kidney and freshly harvested isolated proximal tubules (Fig. 3).⁵⁷ It is of interest that NOS-1 also mediates much of the modulation of tubuloglomerular feedback function.⁷⁴ The effects of NOS-1 inhibition were not dependent on changes in kidney blood flow and were not influenced by an intermediary action of AngII.⁵⁷ In vitro studies in freshly harvested proximal tubules have shown that application of S-methyl-l-thiocitrulline (SMTC), an inhibitor of NOS-1, produced an immediate increase in Q_O2 (Fig. 3). The addition of NO donors immediately reverses this process and normalizes Q_O2.⁵⁷ Nitric oxide has several biochemical interactions that influence oxidative metabolism. For example, NO suppresses the citric acid cycle enzyme aconitase, as does superoxide, and inhibits mitochondrial pyruvate uptake. However, the major ‘braking’ effect of NO on oxidative metabolism is accomplished via the inhibition of cytochrome c oxidase.^77–81 Studies have shown that NO can inhibit mitochondrial respiration in vitro by up to 85% and that it prevents progressive loss of mitochondrial membrane potential and apoptosis.^77–81 It has been suggested that NO inhibits not only these systems, but also critical mitochondrial enzymes in complex I and complex II and cytochrome b.⁸¹ Moncada and Erusalimsky showed that, during sepsis, increased generation of NO acts to reduce Q_O2 in a variety of tissues.⁸¹ In subtotal nephrectomy, the increase in Q_O2 observed by several laboratories is accompanied by substantial evidence for NO deficiency in this model of CKD.^24,72–81 There are multiple reasons for the reduction in NO activity, including increased reactive oxygen species (ROS) and inactivation of NO, reduced NOS activity, increased NOS inhibitors, such as asymmetric dimethylarginine. We have observed decreased functional NOS activity in the subtotal nephrectomy model.⁷² Combined AngII blockade does normalize metabolic efficiency and kidney Q_O2 and, concurrently, NOS and NO functional activity are normalized, implying that reduced NO activity contributes to the increase in oxygen consumption and reduced metabolic efficiency.

“Metabolic efficiency changes in pathophysiology”

Fig. 3.

Effects of nitric oxide (NO) and the NO synthase 1 blocker S-methyl-l-thiocitrulline (SMTC) on oxygen consumption (Q_O2) in freshly harvested isolated proximal tubules. (a) Profile of declining O₂% in a metabolic chamber containing proximal tubules. When the NOS-1 inhibitor SMTC is added, Q_O2 increases (as evidenced by the increase in the slope of decline). When the NO donor NONOate is applied, Q_O2 decreases significantly. (b) Absolute Q_O2 is depicted in control tubules and in tubules after the addition of SMTC alone or with NONOate. The increase in Q_O2 after NOS-1 blockade is totally reversed by application of the NO donor, suggesting NO specificity to the phenomenon. These data suggest a major role for intracellular NOS activity in the regulation of Q_O2. Data are the mean ± SEM. *P < 0.05 compared with control; ^†P < 0.05 compared with SMTC. Figure partly based on data in Deng et al.⁵⁷

An understudied area relates to factors that determine the shift between aerobic or oxidative metabolism and glycolytic metabolism in the kidney.^59,60 Obviously shifts in the site of T_Na will influence this shift in metabolism, but factors such as HIF-1 and phosphorylation of AMPK may also exert an impact.⁵⁸ We have noted that, in subtotal nephrectomy, phosphorylation of AMPK is reduced and therapies that restore normal haemodynamics and Q_O2 tend to increase AMPK phosphorylation (RC Blantz, unpubl. obs., 2012). At this stage, there are many metabolic influences, including insulin resistance, that have the potential to change the metabolic efficiency of the kidney or, more specifically, Q_O2/T_Na. Some of the designated causal influences may not be additive or independent variables, but may be acting in series. However, further investigations are required to determine the exact relationships among the multiple mechanisms influencing the metabolic efficiency of T_Na.

In summary, recent studies have made it clear that the metabolic efficiency of the kidney can change significantly and that under pathophysiological conditions, such as CKD, this may make an important contribution to the progressive decline in kidney function and fibrosis that characterizes this important disease.

OXYGEN CONSUMPTION IN DIABETES AND ITS RELATIONSHIP TO TISSUE OXYGEN AVAILABILITY

Today, diabetes mellitus is the most common cause of end-stage renal disease. The prevalence of diabetes mellitus in the industrialized world continues to increase and is presently approximately 7%. Type 1 diabetes accounts for 5–10% of all diagnosed cases of diabetes (see http://diabetes.niddk.nih.gov, accessed 10 Dec 2012). Despite intense research, the mechanisms underlying the development of diabetic nephropathy remain largely unknown. An early theory ascribed diabetic nephropathy to sustained barotraumas of the glomerular capillaries due to a combination of preglomerular vasodilatation, which increased glomerular capillary pressure, and to glomerular hypertrophy, which increased capillary wall tension.⁸² A competing theory states that critical changes in the cytokine and growth factor environment in the glomerular capillary result in kidney damage.⁸³ Neither theory alone is presently fully compatible with the available evidence. Recently, focus has increased on the involvement of diabetes-induced formation of ROS, increased Q_O2 and altered renal energy metabolism.⁸⁴ Diabetes is primarily a disease of metabolism and diabetes-induced changes in metabolism are important in the development of other secondary complications of diabetes, such as retinopathy and neuropathy.⁸⁵ In accordance with the hypothesis that diabetes-induced changes in renal O₂ metabolism mediate the development of altered kidney function, we and others have found severe alterations in O₂ metabolism that have been related to increased oxidative stress.^35,86 It is well known that mitochondrial function is deranged in diabetes due, at least in part, to increased oxidative stress, but it has also been proposed that the altered mitochondrial function per se increases $\dot{}{\mathrm{O}}_2^{-}$ formation and augments the progression of diabetic nephropathy via increasing Q_O2.^27,87

Fat, carbohydrates and proteins are metabolised through β-oxidation, glycolysis and deamination to enter the Krebs’ cycle located in the mitochondrial matrix. This creates NADH and FADH₂, which are used by the mitochondria to produce energy (ATP) via oxidative phosphorylation. Oxidative phosphorylation takes place in the mitochondrial electron transport chain (ETC), which consists of four complexes located in the mitochondrial inner membrane. During oxidation of NADH and FADH₂, electrons are transported via these complexes and donated to O₂, the ultimate electron acceptor. Protons (H⁺) are pumped across the inner membrane to the intermembrane space during this process, which creates an electrochemical gradient, commonly referred to as the membrane potential. Thereafter, protons are released back across the membrane via channels referred to as ATP synthase because this process produces ATP from ADP and inorganic phosphate (P_i). The mitochondrial membrane potential is also used to transport ions and metabolites in and out of the mitochondria.

The ETC. complexes are regulated by their redox status (i.e. by the electrochemical membrane). During the creation and maintenance of a high and stable mitochondrial electrochemical membrane potential, $\dot{}{\mathrm{O}}_2^{-}$ is produced. Different complexes react differently to changes in membrane potential, which results in a complex regulation of radical production. The vast majority of radicals are formed by complex I and complex III (Fig. 4).⁸⁸ However, $\dot{}{\mathrm{O}}_2^{-}$ is rapidly converted to hydrogen peroxide by superoxide dismutase under normal physiological conditions. Catalase further metabolises H₂O₂ to water and O₂.⁸⁵ During excessive substrate load to the ETC. (e.g. intracellular hyperglycaemia), the mitochondrial membrane potential is elevated, which induces excessive formation of $\dot{}{\mathrm{O}}_2^{-}$.⁸⁹ This will result in increased damage by the radicals formed, including the formation of protein carbonyls, lipid peroxidation and direct DNA damage. The role of mitochondrial $\dot{}{\mathrm{O}}_2^{-}$ formation in the development of diabetes-related pathologies has been clearly demonstrated by Brownlee.⁹⁰ Inhibition of mitochondrial $\dot{}{\mathrm{O}}_2^{-}$ formation totally inhibits all the classical pathways known to result in diabetes-related diseases.⁸⁹

“Diabetes-induced formation of ROS increases Q_O2”

Fig. 4.

Electron transport chain and the proposed mechanisms of mitochondrial uncoupling via adenosine nucleotide translocase and uncoupling proteins. e⁻, electrons. Reproduced with permission from O’Rourke et al.⁸⁸

A pivotal prerequisite for the proposed hypothesis (Fig. 1) is that intrarenal AngII is elevated in diabetes. An early study by Machimura demonstrated beneficial effects on serum creatinine levels after ACE inhibition in patients with diabetic nephropathy.⁹² Furthermore, Onozato et al.⁴⁶ showed that streptozotocin-diabetic rats have elevated intrarenal AngII levels. Importantly, elevated AngII levels increase mitochondrial H₂O₂ production and reduce the respiratory control ratio.⁹² Notably, NADPH oxidase activity is increased by AngII, but all the AngII-induced alterations are prevented by treatment with the mitochondria-targeting anti-oxidant mitoTempo,⁹² including increased NADPH oxidase activity. Indeed, chronic treatment with apocynin to inhibit NADPH oxidase activity during the entire 4-week duration of diabetes in rats resulted in improved kidney P_O2,⁹³ establishing a link between increased NADPH oxidase activity and deranged O₂ metabolism in the diabetic kidney. This further supports the hypothesis of an intimate interplay between NADPH oxidase and the mitochondria.

It has been shown recently that acute inhibition of NADPH oxidase significantly improves intrarenal P_O2 via inhibition of active tubular Na⁺ transport.⁹⁴ Because the diabetes-induced glomerular hyperfiltration was unaffected by both chronic and acute apocynin treatment,^93,94 it can be concluded that changes in GFR merely play a minor role in increased kidney Q_O2 and subsequent tissue hypoxia in the diabetic kidney. This is in good agreement with our previous report showing that anti-oxidant treatment in diabetic rats normalizes kidney P_O2 independent of any effects on GFR.³⁵

Role of UCP-2

To decrease mitochondrial membrane potential under conditions of excess substrate availability, it has been hypothesized (and shown from in vitro studies) that cells can increase the H⁺ leak across the mitochondrial membrane by synthesising and incorporating ion channels.⁹⁵ Activation of UCP results in H⁺ leak across the inner membrane that reduces the mitochondrial membrane production and thus limits oxidative stress, but also uncouples Q_O2 from the production of ATP.⁹⁶ This results in increased Q_O2 for fixed ATP production. This alters the redox status of the ETC. complexes and several studies have shown that this results in decreased $\dot{}{\mathrm{O}}_2^{-}$ production,⁴⁰ which implies an important regulatory function of UCPs. Uncoupling protein-1 is present in brown adipose tissue, where it has a solely thermogenic effect, whereas UCP-2 is present in tissues with significant levels of ATP production, including the kidneys.⁹⁷ Furthermore, increases have been demonstrated in UCP-2 and UCP-3 when oxidative stress and lipid peroxidation increase.⁹⁸ It has been hypothesized that altered regulation of UCPs could be a significant part of the pathology of arteriosclerosis, hypertension and diabetes.^99,100 The reported increased expression of UCP-2 during hyperglycaemia¹⁰¹ could be viewed as a cellular defence against excessive $\dot{}{\mathrm{O}}_2^{-}$ formation, but with a significant increase in Q_O2 as a noticeable side-effect (Fig. 4). Indeed, we have shown recently that UCP-2 expression and activity are increased in diabetic kidneys.^41,97,102

“Activation of UCP-2 results in increased Q_O2”

It should be noted that the mitochondrial membrane potential is normal as long as UCP-2 function is maintained, but increased when GDP is present.⁹⁶ However, administration of short interference (si) RNA against UCP-2, which reduced UCP-2 protein expression by 62%, paradoxically increased uncoupling, evident from increased glutamate-stimulated Q_O2 and significantly lower membrane potential.⁹⁶ Importantly, GDP did not affect the uncoupling in mitochondria from rats administered siRNA against UCP-2, indicating no involvement of UCPs.⁹⁶ The addition of ADP in the absence of ATP and during blockade of ATP synthase by oligomycin reduced Q_O2,⁹⁶ indicating a pivotal role of the adenosine nucleotide transporter (ANT). It is known that, under certain conditions, ANT has uncoupling properties and it was therefore proposed that this is a backup mechanism controlling mitochondrial membrane potential during UCP-2 dysfunction.⁹⁶ The ANT-mediated uncoupling was far more potent in reducing both mitochondrial membrane potential and kidney tissue oxidative stress, but resulted in a further accelerated Q_O2. Thus, it seems that the system favours the regulation of mitochondrial membrane potential and radical formation at the expense of elevated Q_O2 and the risk of intrarenal tissue hypoxia (Fig. 5).

Fig. 5.

Simplified hypothesis for how diabetes, via mitochondrial uncoupling mediated by either uncoupling protein (UCP)-2 and adenosine nucleotide translocase (ANT) increases kidney oxygen consumption (Q_O2), which results in kidney tissue hypoxia and the development of diabetic nephropathy. So far, it has only been demonstrated that mitochondrial uncoupling via ANT occurs when normal UCP-2 function is reduced.¹⁰² AngII, angiotensin II; ROS, reactive oxygen species. Modified from Welch et al.¹¹⁴ and Welch et al.¹¹⁷

Link between increased Q_O2 and diabetic nephropathy involves intrarenal tissue hypoxia

Previously, we showed that diabetic rats have decreased P_O2 throughout their kidney tissue due to increased Q_O2.^35,86 Lower oxygen levels in diabetic kidneys have also been found in patients.^22,37 Functional impairment of the kidney is well correlated with the degree of tubulointerstitial damage and this pathological finding has led to the broad recognition that the final common pathway of progression of kidney failure operates principally in the tubulointerstitium.^103,104 The chronic hypoxia hypothesis emphasizes chronic hypoxic damage in the tubulointerstitium as a final common pathway in end-stage renal disease and has been intensively investigated by many.^105,106 It should be noted that the decreased intrarenal P_O2 was independent of GFR (i.e. it occurred under conditions of both hyperfiltration³⁵ and normal GFR⁸⁶, that is independent of altered T_Na). Isolated proximal tubular cells from diabetic rats exhibit increased Q_O2 compared with controls, which is totally prevented by intense insulin treatment.⁴¹ This highlights sustained hyperglycaemia as a pivotal factor for the development of the increased Q_O2 in diabetic animals. Furthermore, anti-oxidant treatment throughout the course of diabetes totally prevented the increase in Q_O2, suggesting a close relationship between increased radical formation and elevated Q_O2 in the kidneys of diabetic rats.³⁵ The increased Q_O2 resulted in decreased intrarenal tissue P_O2, which has been proposed to be involved in the development of diabetes-induced hypoxic kidney injury.^27,87 Until recently, the mechanisms mediating the altered O₂ metabolism in the diabetic kidney have been largely unknown, but we have now described in detail at least one mechanism responsible for the diabetes-induced increase in Q_O2.⁴¹ Increased UCP-2 levels in the diabetic mitochondria result in uncoupling, which was demonstrated as glutamate-stimulated Q_O2 during complete blockade of ATP production by oligomycin Q_O2.⁴¹ The specific involvement of UCP-2 was inferred by the abolished mitochondrial uncoupling either in the presence of the specific UCP inhibitor GDP or removal of the free fatty acids that are required for normal UCP function. Because only UCP-2 has been identified in rat kidneys,⁹⁷ the GDP-induced abolishment of uncoupling strongly suggests involvement of UCP-2.

Echtay et al., using kidney mitochondria from both rat and mouse, have shown that UCP-2 is activated by $\dot{}{\mathrm{O}}_2^{-}$.¹⁰⁷ Importantly, the absolute requirement for UCP-2 for mediating the H⁺ leak and reducing mitochondrial membrane potential was demonstrated using kidney mitochondria from UCP-2^−/− mice. In contrast with the study of Echtay et al., in which an exogenous $\dot{}{\mathrm{O}}_2^{-}$ source was used, Krauss et al. demonstrated that endogenous $\dot{}{\mathrm{O}}_2^{-}$ activates UCP-2 in a similar way.¹⁰⁸ Indeed, in a recent study, we confirmed that db/db mice, a model of Type 2 diabetes, also exhibit pronounced mitochondrial uncoupling in the kidney, which is prevented by anti-oxidant treatment with coenzyme Q10. Importantly, coenzyme Q10 not only prevented diabetes-induced oxidative stress, but also normalized mitochondrial morphology and function and normalized glomerular filtration and urinary protein leakage.¹⁰⁹ These findings expand the hypothesis beyond insulinopenic diabetes to also include Type 2 diabetes, the prevalence of which is increasing rapidly and which is far more frequent than Type 1 diabetes.

“Link between increased Q_O2 and diabetic nephropathy”

In summary, sustained hyperglycaemia increases intrarenal AngII levels, resulting in increased oxidative stress due to activation of NADPH oxidase. This results in increased mitochondrial uncoupling via UCP-2 and decreased electrolyte transport efficiency. The commonly occurring diabetes-induced glomerular hyperfiltration, with subsequent increased T_Na, together with reduced tubular electrolyte transport efficiency and increased mitochondrial uncoupling result in increased Q_O2. These events result in intrarenal tissue hypoxia and the development of clinical hallmarks of diabetic nephropathy, because no compensatory increase in oxygen delivery (i.e. RBF) occurs. However, these findings from experimental animal models need to be verified in the clinical setting before they can be translated into new therapeutic approaches for diabetic patients.

OXYGEN CONSUMPTION AND TISSUE OXYGENATION IN HYPERTENSION

Welch et al.³⁹ were the first to show that P_O2 was lower in both the cortex and medulla of kidneys from spontaneously hypertensive rats (SHR) compared with their normotensive genetic control, Wistar-Kyoto rats. In the outer cortex of kidneys from SHR, P_O2 was 8–10 mmHg lower than in normotensive rats. The differences were fractionally higher in the inner cortex and outer medulla, with P_O2 as low as 12 mmHg in kidneys from SHR. The GFR in the SHR was similar to that in the appropriate control rats, suggesting disruption of the normal relationship between O₂ and GFR. The ARB candesartan reduced mean arterial blood pressure (MAP) and normalized P_O2 in SHR (Fig. 6). Although the antioxidant tempol also normalized P_O2 in SHR, it lowered, but did not normalize, MAP. Because AngII receptor activation is linked to the generation of superoxide,¹¹⁰ the normalizing effects of the ARB could be partially linked to oxidative stress. Supporting the role of AngII and oxidant pathways, the combination of hydralazine, hydrochlorothiazide and reserpine (HHR), which reduces MAP, but does not block AngII or reduce oxidative stress,¹¹¹ had no effect on renal P_O2.³⁹ However, Dworkin showed that HHR was just as effective in preventing renal damage as other antihypertensive agents, although that study included uninephrectomized SHR, which may have complicated matters.¹¹² Subsequently, Welch et al.¹¹³ showed that renal P_O2 was lower in animals made hypertensive by AngII infusion. Renal cortical and medullary P_O2 was lower in AngII-infused rats, which raised blood pressure by 20–25 mmHg, compared with vehicle-infused rats. This reduction was normalized by the anti-oxidant tempol.¹¹³

“P_O2 lower in kidneys from SHR”

Fig. 6.

Renal oxygen tension was lower in spontaneously hypertensive rats (SHR) compared with Wistar-Kyoto (WKY) rats and was normalized by the angiotensin receptor blocker candesartan and the anti-oxidant tempol, but not by the antihypertensive combination of hydralazine, hydrochlorothiazide and reserpine (HHR).^113,117 Data are the mean ± SEM. **P < 0.01, ***P < 0.001 compared with WKY rats.

Recent studies using pimonidazole have confirmed these Clarke electrode studies. For example, AngII increased the number of pimonidazole-positive cells in kidneys of AngII-infused rats compared with control in two studies.^114,115 Renal hypoxia in humans has been difficult to assess, but in a recent study using blood oxygen level-dependent magnetic resonance imaging (BOLD-MRI),¹¹⁶ renal medullary P_O2 was lower in hypertensive African American subjects compared with normotensive controls. Typically these patients have elevated markers of oxidative stress.¹¹⁶ The hypoxic medulla was normalized by acute treatment with a loop diuretic, suggesting an overly active transport system in the mTAL in these patients and confirming P_O2 in this section reflects active Na⁺ transport.

The relationship between Q_O2 and T_Na has been used to define O₂ metabolism in multiple models. We reported that tubular transport efficiency was altered in hypertensive rats, with higher levels of Q_O2 per Na⁺ transported, suggesting inefficient renal O₂ usage in SHR.¹¹⁷ For a given amount of Na⁺ transported, nearly twice as much O₂ was consumed. This suggests that the major energy-requiring function of the kidney is not limited by O₂ delivery. This is consistent with previous suggestions that the kidney is unique in that function is not tied to O₂ delivery.⁵⁶ Because these rats were in steady state and excreted the same levels of Na⁺ as the control rats, we speculated that the kidney maintains normal Na⁺ uptake requirements, primarily by maintaining local O₂. Therefore, the excess use of O₂ in hypertensive rats suggests that tissue O₂ will be lower, reflecting this presumed inefficiency. However, as these data suggest, this may be part of the kidney’s ability to maintain sufficient Na⁺ reabsorption and should not be considered a lack of efficiency, per se.

What are the consequences of the reduced efficiency found in hypertensive kidneys? The immediate and obvious effect is the lower P_O2 levels observed in SHR and other models of hypertension,^{113,114,118,119} as well as in some hypertensive patients.^3,116 Renal tissue hypoxia is linked to tubulointerstitial fibrosis.¹²⁰ The hypoxia, along with barotrauma of high blood pressure on preglomerular vasculature, may be considered as the precursor to the renal damage that accompanies hypertension. In addition, systemic hypertension is accompanied by the release of multiple vasoconstrictors that also impact on the effects of hypoxia on renal damage. These include higher levels of the renin–angiotensin–aldosterone system^121,122 and higher levels of constrictor prostaglandins¹²³ and endothelin.¹²⁴ However, as these studies suggest, oxidants released during hypertension have major effects on blood pressure.

OXIDATIVE STRESS AND HYPERTENSION

Oxidative stress occurs when pro-oxidants systems, such as enzymes and mitochondria, generate levels of ROS that are not quenched by endogenous anti-oxidant systems. Multiple markers of ROS are elevated in SHR,^15,125,126 AngII-induced hypertension,¹¹⁴ renovascular hypertension^127,128 and hypertension induced by blockade of NO.¹²⁹ Renal levels of ROS are obviously elevated during hypertension.¹³⁰ Blockade of AngII receptors corrected the hypoxia and the inefficient use of O₂ in SHR, whereas lowering of blood pressure alone has only minor effects on these parameters.¹³¹ Furthermore, the common pathway seemed to be elevated ROS, because exogenous anti-oxidants were as effective as ARBs. Reduction of p22^phox, the critical component of NADPH oxidases, also corrected hypoxia and the associated renal dysfunction. Laycock et al.⁷⁵ have reported that NO inhibition, which increases superoxide, reduced tubular electrolyte transport efficiency even in normotensive animals. This suggests that regulation of tubular transport efficiency is closely linked to superoxide.

Activation of AngII receptors stimulates oxidant pathways by activating NADPH oxidase and mitochondria, and many of the prohypertensive effects of AngII may be linked to oxidative stress.¹²⁹ Elevated ROS are associated with end-organ damage not only in hypertensive animals and humans, but also in experimental models of and patients with diabetes, as well as in ageing humans.¹⁹ Therefore, we propose that increased oxidants during hypertension, combined with tissue hypoxia, are critical elements in the early stages of tubulointerstitial fibrosis and renal damage.

The absence of endogenous anti-oxidant enzymes, such as superoxide dismutase (SOD), can also affect oxidative stress and renal O₂ balance. Multiple markers of oxidative stress, including superoxide, are elevated in extracellular superoxide dismutase (EC-SOD) knockout mice.¹³² These mice also have lower renal P_O2 and lower tubular electrolyte transport efficiency compared with wild-type controls.¹³² Both renal hypoxia and O₂ efficiency are normalized by tempol treatment.¹³²

Hypoxia generated by shifts in Na⁺ uptake and the inefficient use of O₂ is part of the overall dysfunction that occurs in hypertensive kidneys. The resulting hypoxia can lead to tissue damage in both vascular and tubular cells. The increased levels of superoxide and related oxidants, such as H₂O₂, peroxynitrite and hydroxyl radicals, exacerbate the effects of hypoxia. Therefore, prolonged exposure to hypoxia, oxidative stress and increased levels of vasoconstrictor hormones lead to renal injury, which is often seen in older SHR.¹³³ Expression of the prooxidant enzyme NADPH oxidase is higher in kidneys of older compared with young SHR.¹³⁴ Conversely, expression of the anti-oxidant enzymes SOD1 and SOD3 is lower in older SHR. The loss of the anti-oxidant NO, generated by NOS isoforms, is also associated with renal injury in SHR. Long-term treatment with ARBs has been shown to enhance renal NOS and reduce renal injury in SHR.¹³⁵ Furthermore, long-term treatment with exogenous anti-oxidants reduces blood pressure¹³⁶ and improves GFR¹³⁷ in SHR. These results, combined with observations that ARBs and tempol both restore renal hypoxia and the inefficient use of O₂ in SHR and other forms of hypertension,^39,113,117 suggest that renal injuries induced by hypertension are preceded by hypoxia as a contributing factor.

In summary, the kidney is relatively hypoxic in experimental models of hypertension, which may exacerbate renal damage associated with the disease. The changes in oxygen use along the nephron may contribute to renal tissue hypoxia. The kidney adjusts to the resulting hypoxia and maintains sodium balance.

“AngII receptors stimulates oxidant pathways”

CONCLUSION

An increasing number of experimental studies are providing evidence that intrarenal hypoxia is a unifying pathway to CKD. To be able to find new treatment modalities in, for example, diabetic nephropathy and hypertension, it is first necessary to verify the findings of deranged oxygen metabolism in the clinical setting. The need for non-invasive methodology is crucial for such a translation and, in the paper by Francis et al. in this series, such methodology is presented. The use of magnetic resonance imaging for the approximation of tissue oxygenation and blood flow is already in use in patients and is promising for the verification of experimental data.