Pharmacological HIF- PHD inhibition reduces renovascular resistance and increases glomerular filtration by stimulating nitric oxide generation

Aim: Hypoxia- inducible factors (HIFs) are O 2 - sensitive transcription factors that regulate multiple biological processes which are essential for cellular adaptation to hypoxia. Small molecule inhibitors of HIF- prolyl hydroxylase domain (PHD) dioxy-genases (HIF- PHIs) activate HIF- dependent transcriptional programs and have broad clinical potential. HIF- PHIs are currently in global late- stage clinical development for the treatment of anaemia associated with chronic kidney disease. Although the effects of hypoxia on renal haemodynamics and function have been studied in animal models and in humans living at high altitude, the effects of pharmacological HIF activation on renal haemodynamics, O 2 metabolism and metabolic efficiency are not well understood. Methods: Using a cross- sectional study design, we investigated renal haemodynamics, O 2 metabolism, gene expression and NO production in healthy rats treated with different doses of HIF- PHIs roxadustat or molidustat compared to vehicle control. Results: Systemic administration of roxadustat or molidustat resulted in a dose-dependent reduction in renovascular resistance (RVR). This was associated with increased glomerular filtration rate (GFR), urine flow and tubular sodium transport rate (T Na ). Although both total O 2 delivery and T Na were increased, more O 2 was extracted per transported sodium in rats treated with high- doses of HIF- PHIs, suggesting a reduction in metabolic efficiency. Changes in RVR and GFR were associated with increased nitric oxide (NO) generation and substantially suppressed by pharmacological inhibition of NO synthesis. Conclusions: Our data provide mechanistic insights into dose- dependent effects of short- term pharmacological HIF activation on renal haemodynamics, glomerular filtration and O 2 metabolism and identify NO as a major mediator of these effects.


| INTRODUCTION
Small molecule inhibitors of hypoxia-inducible factor (HIF)prolyl hydroxylase domain (PHD) O 2 sensors (HIF-PHIs) are a promising new class of orally administered drugs that mimic hypoxia responses and have broad therapeutic potential. 1,2The HIF/PHD O 2 -sensing pathway regulates multiple cellular responses to hypoxia and is involved in numerous disease states including ischemia-reperfusion injury, pulmonary arterial hypertension, cancer and abnormal erythropoiesis. 3n the context of anaemia therapy, HIF-PHIs promote erythropoiesis through an increase in HIF activity, which stimulates the production of endogenous erythropoietin (EPO) in kidney and liver and promotes iron uptake and mobilization. 1owever, given the breadth of HIF-regulated biological processes, HIF-PHI therapy is expected to have clinical effects beyond erythropoiesis that may be beneficial for patients with chronic kidney disease (CKD) or could potentially cause harm. 2,4][8][9] Under normal O 2 conditions, the hydroxylation of specific proline residues within the continuously synthesized HIF-α subunit initiates its rapid proteasomal degradation via ubiquitylation by the von Hippel-Lindau (VHL)-E3 ubiquitin ligase complex, resulting in very low or non-detectable cellular HIF-α levels.Under hypoxia or following the administration of HIF-PHIs, the activity of HIF-PHDs and the rate of HIF-α degradation are reduced.This results in cellular HIF-α accumulation, the formation of HIF-α/β heterodimers in the nucleus and the increased transcription of O 2 -regulated genes, such as EPO, vascular endothelial growth factor A (VEGFA), phosphoglycerate kinase 1 (PGK1), lactate dehydrogenase (LDH) and genes involved in the regulation of vascular tone such as nitric oxide synthase (NOS). 8,10he HIF system plays a major role in the regulation of vascular tone.In the pulmonary vasculature, hypoxia and HIF activation promote the development of pulmonary arterial hypertension through increased expression of vasoconstrictive factors, ion channels and ion transporters, 11 whereas hypoxia and HIF activation in other vascular beds are associated with vasodilation, e.g., in the muscle or skin, promoting blood flow. 12Little is known about the effects of systemic pharmacological HIF activation on renal haemodynamics, O 2 metabolism and tubular transport efficiency.In this crosssectional study, we investigated two HIF-PHIs, roxadustat (FG-4592) and molidustat (BAY 85-3934), which have been approved for the treatment of renal anaemia in Japan. 1 Our data provide novel insights into the effects of short-term systemic HIF activation on renal haemodynamics and O 2 metabolism and identify nitric oxide (NO) as a major mediator of these effects.

| Systemic pharmacological HIF activation increases RBF, GFR and tubular sodium transport
Roxadustat and molidustat are potent inhibitors of HIF-PHDs.They stabilize both, HIF-1α and HIF-2α and stimulate the production of endogenous EPO in patients with CKD. 1 To study the effects of short-term HIF-PHI administration on renal haemodynamics, different doses of roxadustat or molidustat were injected intraperitoneally (i.p.) into healthy male Sprague Dawley rats.Two injections were administered 48 hours apart, except for molidustat dosed at 1 mg kg −1 , which was given three times.4][15] In vivo experiments were performed 4 or 6 hours after the final HIF-PHI dose was administered (Figure 1).
We invasively measured blood pressure (BP) and heart rate (HR) in anaesthetized rats by femoral artery catheterization (Figure 2A).Treatment with both roxadustat or molidustat resulted in dose-dependent differences in systolic and diastolic BP without affecting HR.Mean arterial pressure (MAP) was 94 ± 2 mm Hg for roxadustat 30 mg kg −1 and 94 ± 2 mm Hg for molidustat 10 mg kg −1 vs 115 ± 2 mm Hg for vehicle control (Table 1 and Figure 2B).
We next investigated to what degree the HIF-PHI-induced reduction in MAP affected renal perfusion.We used Doppler ultrasonography to assess renal blood flow (RBF) and found significant dose-dependent differences in RBF for roxadustat dosed at 10 and 30 mg kg −1 and molidustat dosed at 10 mg kg −1 , indicating a change in renovascular resistance (RVR).RBF was 12.9 ± 0.4 mL min −1 per kidney for roxadustat dosed at 30 mg kg −1 and 11.9 ± 0.1 mL min −1 per kidney for molidustat dosed at 10 mg kg −1 vs 10.2 ± 0.2 mL min −1 per kidney for vehicle control.RVR decreased by 36% for roxadustat 30 mg kg −1 and by 30% for molidustat 10 mg kg −1 (Figure 2C).
Because increased GFR enhances sodium delivery to the nephron, we examined tubular sodium transport rate (T Na ).To calculate T Na , we determined the amount of filtered sodium and subtracted the amount of sodium that was excreted in the urine.T Na was tightly correlated with GFR across all treatment groups (Figure 3) and was increased in the highdose HIF-PHI cohorts compared to lower doses (Table 2).T Na was increased by ~30% for roxadustat 30 mg kg −1 and by ~20% for molidustat 10 mg kg −1 ; 263.4 ± 7.9 μmol min −1 per kidney and 240.3 ± 8.8 μmol min −1 per kidney, respectively, vs 196.2 ± 4.3 μmol min −1 per kidney for vehicle control.Increased T Na in the high-dose roxadustat cohort was furthermore associated with increased urine flow (Table 2).No significant differences between cohorts were found with regard to filtration fraction (FF) and blood urea nitrogen (BUN) levels (Figure 2C and Table 1).

| High-dose HIF-PHI administration increases O 2 consumption disproportionally
Most of the O 2 consumed by the kidney is used for sodium transport and is in linear relationship to tubular sodium reabsorption. 16,17Because HIF-PHI administration increased T Na , we asked whether and to what degree HIF-PHI treatment affected renal O 2 consumption (QO 2 ) and tubular sodium transport efficiency (T Na /O 2 ), the latter being a measure of metabolic efficiency, ie the amount of O 2 required to reabsorb a given amount sodium.We calculated O 2 content in arterial and renal vein blood arriving at O 2 delivery and QO 2 by multiplying with RBF.Not surprisingly, we found a dose-dependent increase in O 2 delivery; 2.5 ± 0.1 mL min −1 per kidney for roxadustat 30 mg kg −1 (~40% increase) and 2.1 ± 0.04 mL min −1 per kidney for molidustat 10 mg kg −1 (~18% increase) vs 1.8 ± 0.04 mL min −1 per kidney for vehicle control (Figure 3).Renal QO 2 was significantly increased for roxadustat 30 mg kg −1 and molidustat 10 mg kg −1 with 0.45 ± 0.06 mL min −1 for roxadustat 30 mg kg −1 and 0.34 ± 0.04 mL min −1 for molidustat 10 mg kg −1 vs 0.19 ± 0.01 mL min −1 for vehicle control but not for lower doses of roxadustat and molidustat (Figure 3).To assess metabolic efficiency, we calculated the T Na /QO 2 ratio.Although we did not detect significant differences in T Na /QO 2 for the roxadustat 1, 2 and 10 mg kg −1 and molidustat 1 mg kg −1 cohorts, T Na /QO 2 was decreased by ~40% for the roxadustat 30 mg kg −1 and by ~30% for the molidustat 10 mg kg −1 cohort, suggesting that metabolic efficiency of tubular sodium transport was reduced for high-dose but not for low-dose HIF-PHI treatment (Figure 3).

HIF-activating compounds increases the expression of HIF-regulated genes in a dose-dependent manner
To assess the degree of HIF activation, we examined serum EPO levels and HIF target gene expression in kidney and liver Data are represented as mean ± SEM; one-way ANOVA followed by Tukey's post-hoc analysis; n = 12 for vehicle control, n = 8 for all other cohorts; *P < .05,† P < .01 and ‡ P < .001compared with vehicle control.ABG, arterial blood gas (femoral artery); BP, blood pressure (femoral artery); moli, molidustat; VBG, venous blood gas (left renal vein); veh, vehicle from HIF-PHI-treated rats subjected to haemodynamic studies.We found dose-dependent differences in serum EPO levels compared to vehicle-treated rats, which was manifested in a ~6-fold increase in serum EPO for roxadustat dosed at 2 mg kg −1 and a ~130-fold increase for roxadustat dosed at 30 mg kg −1 .Approximately 5-fold and 70-fold greater serum EPO concentrations were measured in rats treated with molidustat dosed at 1 mg kg −1 and 10 mg kg −1 , respectively, compared with vehicle-treated controls.Differences in serum EPO levels were mirrored in kidney and liver Epo transcript levels (Figure 4A).Increased Epo expression was also found in other organs such as the lung and heart (Figure S1).Systemic administration of roxadustat resulted in stabilization of HIF-1α and HIF-2α in the kidney (Figure S2).In addition to renal Epo, which is regulated by HIF-2, 18 we detected significant increases in the expression of genes that are either predominantly HIF-1-depedent or co-regulated by HIF-1 and by HIF-2.These included adrenomedullin 1 (Adm1), BCL2 interacting protein 3 (Bnip3), Egln3, Ldh, pyruvate dehydrogenase kinase 1 (Pdk1), Pgk1 and Vegfa (Figure 4B, Figures S3 and S4). 19Systemic HIF activation was furthermore reflected in dose-dependent increases in the expression of HIF-regulated duodenal cytochrome B and divalent metal transporter 1 (Figure S5).

| High-dose HIF-PHI administration is associated with increased renal pO 2
In order to examine whether the decrease in T Na /QO 2 in rats treated with high-dose roxadustat was associated with a reduction in tissue pO 2 levels, we measured cortical and medullary renal pO 2 in independent rat cohorts.Cortical tissue pO 2 was measured with a Clark-type electrode at 1 mm depth and medullary pO 2 at 4 mm depth below the renal capsule in vehicle-treated rats and in roxadustat-treated rats dosed at 10 or 30 mg kg −1 .Whereas cortical and medullary tissue pO 2 in the 10 mg kg −1 cohort was comparable to vehicle-treated  rats, cortical and medullary pO2 in the 30 mg kg −1 cohort were increased to 45.0 ± 0.5 mm Hg and 36.7 ± 0.5 mm Hg, respectively, vs 42.2 ± 0.6 mm Hg and 33.0 ± 0.4 mm Hg in control (Figure 5).Elevated renal pO 2 in the roxadustat 30 mg kg −1 cohort was associated with an increase in haemoglobin (Figure 5).

| Systemic HIF-PHI administration induces NOS expression and stimulates NO generation
NO plays a critical role in the regulation of renal haemodynamics and metabolic efficiency. 20,21Since HIF-1 and HIF-2 have been shown to regulate nitric oxide synthesis, [22][23][24] we examined transcript levels of Nos1, Nos2, Nos3, arginase (Arg) 1 and Arg2 in kidney, liver, heart, thoracic aorta, lung and primary lung endothelial cells (EC).We found that Nos3 expression was significantly upregulated in kidneys from rats treated with high-dose roxadustat or molidustat.This was reflected in a comparable increase in renal NOS3 protein levels (Figure 6A).Although renal Arg1 expression did not change in roxadustat and molidustat-treated cohorts, Arg2 was upregulated by ~2-fold with molidustat 10 mg kg −1 but not with roxadustat (Figure 6A).To assess other tissues, we examined Nos and Arg transcript levels in lung, heart, liver and thoracic aorta.Nos3 was significantly upregulated in total lung tissue, liver, thoracic aorta and primary lung EC isolated from rats treated with roxadustat 30 mg kg −1 (Figure 6B).In contrast, Nos expression in the heart did not change compared to control despite HIF activation as indicated by a significant increase in Epo transcript levels (Figure 6B and Figure S1).
To determine whether upregulation of Nos in the kidney and other tissues resulted in a detectable increase in NO generation, we measured nitrate and nitrite (NO metabolites) and creatinine in urine and calculated the NO/creatinine ratio.The urine NO/creatinine ratio was increased by 2.7-and 1.7fold for roxadustat 30 mg kg −1 and molidustat 10 mg kg −1 , respectively, compared with vehicle control (Figure 7A).

HIF-PHI-induced haemodynamic changes
To examine the contribution of NO to HIF-PHI-induced changes in renal haemodynamic parameters, we administered NOS inhibitor Nω-nitro-L-arginine methyl ester (L-NAME) to rats pretreated with roxadustat dosed at 2 mg kg −1 , roxadustat dosed at 30 mg kg −1 or molidustat dosed at 10 mg kg −1 .L-NAME was administered i.p. following the final HIF-PHI injection.L-NAME treatment significantly decreased urine NO excretion in all HIF-PHI-treated cohorts, reaching low levels that were comparable to L-NAME-treated rats, which did not receive HIF-PHIs (Figure 7A).Because NO has been shown to inhibit the catalytic activity of HIF-PHDs, 8 we examined whether L-NAME had any effects on HIF-regulated gene expression.We found that the expression levels of representative HIF-regulated genes induced by HIF-PHI administration were not significantly different between L-NAME-treated rats and rats that did not receive L-NAME (Figure 7A), suggesting that L-NAME did not impact HIF-PHI-induced transcriptional responses in our model.
We next examined the effects of L-NAME on BP, RVR, RBF and GFR and urine parameters (Figure 7B and Figure S6).Although L-NAME reversed the haemodynamic effects induced by HIF-PHI administration in all treatment groups, reversal was only partial as differences in MAP, RVR, RBF and GFR were still detectable between the L-NAME-treated F I G U R E 5 Systemic administration of high-dose HIF-PHI increases renal tissue pO 2 .Left panels, tissue pO 2 measurements in renal cortex at 1mm depth and in renal medulla at 4 mm depth.Right panels, corresponding haemoglobin (hb) and hematocrit (hct) values.Hb and hct values for R30 are significantly different from vehicle control when analysed by Student's t test.Data are represented as mean ± SEM; one-way ANOVA followed by Tukey's post-hoc analysis; n = 8-9; † P < .01compared with vehicle control.R10, roxadustat 10 mg kg -1 ; R30 roxadustat 30 mg kg -1 ; veh, vehicle vehicle group and the L-NAME-treated HIF-PHI groups; GFR was 1.39 ± 0.07 mL min −1 per kidney for roxadustat dosed at 30 mg kg −1 , 1.29 ± 0.02 mL min −1 per kidney for molidustat 10 mg kg −1 and 1.14 ± 0.02 mL min −1 per kidney for vehicle-treated rats with P <.001 and <0.05, respectively, by one-way ANOVA followed by Dunnett's multiple comparison test (Figure 7B).

| DISCUSSION
In this cross-sectional study, we investigated the effects of short-term pharmacological HIF activation on renal haemodynamics, O 2 metabolism and metabolic efficiency in healthy rats.We found that treatment with roxadustat or molidustat, two compounds which have completed phase 3 studies for F I G U R E 6 Systemic HIF-PHI administration induces nitric oxide synthase in multiple tissues.(A) Relative mRNA levels for nitric oxide synthase (Nos) 1, Nos2, Nos3, arginase (Arg)1 and Arg2 as determined by qPCR in total kidney homogenates; n = 4 for all groups.Lower panel shows NOS3 and β-actin levels in whole renal cortex extracts by immunoblot for; n = 5 for vehicle-and roxadustat-treated groups.(B) Relative transcript levels of Nos1, Nos2, Nos3, Arg1, Arg2 and vascular endothelial growth factor A (Vegfa) in total lung homogenates, primary lung endothelial cells (EC), heart, liver and thoracic aorta isolated from HIF-PHI-treated rats compared with vehicle-treated control; n = 4-5.Arg2 transcripts were not detected in tissue homogenates from heart and liver.Data are represented as mean ± SEM; one-way ANOVA followed by Tukey's post-hoc analysis or Student's two-tailed t test used for NOS3/β-actin ratios and for mRNA expression panels lung EC R30, heart R30, liver R30 and aorta R30; *P < .05,† P < .01 and ‡ P < .001compared with vehicle control.M1, molidustat 1 mg kg -1 ; M10, molidustat 10 mg kg -1 ; moli, molidustat; R1, roxadustat 1 mg kg -1 ; R2 roxadustat 2 mg kg -1 ; R10, roxadustat 10 mg kg -1 ; R20, roxadustat, 20mg kg -1 ; R30, roxadustat 30 mg kg - 1 ; veh, vehicle the treatment of renal anaemia, resulted in a dose-dependent increase in GFR, renal O 2 delivery and tubular sodium transport.These effects were partially mediated by the stimulation of NO synthesis.Larger HIF-PHI doses were associated with disproportionally higher rates of O 2 consumption, suggesting that pharmacological HIF activation beyond a certain level might be disadvantageous and attenuate metabolic efficiency in the kidney.
We demonstrate that short-term administration of HIF-PHIs roxadustat or molidustat lowered RVR and increased RBF and GFR in the absence of atmospheric hypoxia.6][27][28] Exposure of rats to chronic hypobaric hypoxia resulted in polycythemia and increased RBF and decreased RVR, 29,30 which correlated with the duration of hypoxia exposure and degree of polycythemia 30 ; GFR remained relatively normal in these studies, despite the decrease in RPF.Of interest in this context are studies in a model of recombinant EPO-induced polycythemia, which raised the possibility that increased RBF under chronic hypoxia may be because of enhanced endothelial NO generation from endothelial shear stress. 31An increase in RBF was also found in rats exposed to acute hypoxia. 32However, the kidney's ability to autoregulate regional blood flow at low perfusion pressures was reported to be diminished under these conditions and was associated with a decrease in medullary perfusion compared to normoxic conditions. 32Whether and to what degree treatment with HIF-PHIs impacts on the autoregulation of regional blood flow in the kidney is unclear and warrants further investigation.In this regard, a better understanding of HIF-PHI actions on renal resistance vessels would be of importance, especially for patients with CKD who are at increased risk for developing acute kidney injury due to of hypotension.
In contrast with other organs, such as heart, muscle or brain, kidneys are relatively limited in their ability to match F I G U R E 7 Pharmacological inhibition of NO synthesis suppresses HIF-PHI-induced hemodynamic changes.(A) Left panel, NO to creatinine ratio in urine from vehicle-treated control and HIF-PHI-treated rats in the absence (−) or presence (+) of L-NAME; n = 5.Right panel, HIF target gene expression in kidneys from HIF-PHI-treated rats compared with vehicle-treated rats in the absence (−) or presence (+) of L-NAME; n = 4. (B) Mean arterial pressure (MAP), renovascular resistance (RVR), renal blood flow (RBF) and glomerular filtration rate (GFR) for each treatment cohort in the absence (−) or presence (+) of L-NAME.L-NAME-treated cohorts were compared to the same experimental groups used in Figure 2; n = 12 for vehicle control, n = 4-5 for all cohorts treated with L-NAME, n = 8 for all other groups.Data are represented as mean ± SEM; two-way ANOVA followed by Tukey's post-hoc analysis; *P < .05,† P < .01 and ‡ P < .001compared to vehicle in the corresponding L-NAME group (− or +).Epo, erythropoietin; i.p., intraperitoneal; Ldh, lactate dehydrogenase; L-NAME, N ω -Nitro-L-arginine methyl ester; M10, molidustat 10 mg kg -1 ; Nos3, nitric oxide synthase 3; R2, roxadustat 2 mg kg -1 ; R30, roxadustat 30 mg kg -1 ; veh, vehicle metabolic demand by increasing O 2 delivery.A rise in RBF is associated with increased sodium filtration and tubular reabsorption and thus increased O 2 demand.The renal O 2 consumption rate QO 2 is an indirect measure of the kidney's metabolic activity, i.e., metabolic demand, and is in linear relationship with tubular sodium transport. 16We found that QO 2 increased disproportionally compared with T Na when high doses of roxadustat were administered (38% decrease in T Na /QO 2 ), suggesting reduced metabolic efficiency.The reasons for this are not clear and may involve HIF-regulated shifts in tubular epithelial energy metabolism towards glycolysis. 33,34HIF has also been shown to regulate mitochondrial function and the expression of electron transport chain (ETC) complexes, such as cytochrome c oxidase subunit 4 (COX4). 34,35However, we did not find that short-term treatment with higher HIF-PHI doses affected the expression of COX4 or transcript levels of genes encoding ETC components (Figure S4).
A reduction in metabolic efficiency can lead to renal hypoxia. 36Although, T Na /QO 2 was decreased in rats treated with high-dose roxadustat, we found that renal tissue pO 2 was increased.This was most likely because of increased haemoglobin, i.e., O 2 -carrying capacity, in roxadustat-treated rats.However, it is plausible that high-dose HIF-PHI treatment may cause or exacerbate renal hypoxia in the presence of anaemia and/or pathological conditions that decrease metabolic efficiency in the kidney, such as relative NO deficiency and oxidative stress, 20,[36][37][38][39] abnormalities in mitochondrial metabolism, 40 and cellular and molecular alterations in transport processes shifting the sites of sodium reabsorption towards less energy-efficient distal nephron segments. 41ur data suggest that increased NO generation is a significant contributor to HIF-PHI-induced changes in renal haemodynamic parameters.3][44][45][46] We found that pharmacological NOS blockade with L-NAME reversed HIF-PHI-mediated haemodynamic effects substantially but not completely, suggesting that additional signalling pathways must have contributed to HIF-PHI-mediated changes in renal haemodynamics.This notion is supported by results from genome-wide mRNA expression analysis of kidneys from HIF-PHI-treated rats, which identified 322 differentially regulated genes involved the regulation of BP and vascular tone (Table S2).Adrenomedullin and VEGF are HIF-induced oxygen-sensitive vasodilatory proteins, 47,48 which have been shown to regulate renal haemodynamics.and Adm1 transcript levels were creased kidneys from HIF-PHI-treated rats and may have contributed HIF-PH-mediated haemodynamic effects directly or indirectly.Adrenomedullin increases RBF, enhances diuresis and promotes natriuresis 49,51 and VEGF has been shown to increase RBF. 50][54] In our studies, lower HIF-PHI doses were sufficient for the activation of renal Epo transcription, whereas the induction of other HIF-regulated genes appeared to require higher HIF-PHI doses.These findings are in line with the notion that the thresholds for HIF target gene induction are genedependent, with EPO being particularly sensitive to HIF activation under hypoxic conditions. 55A 6-to 22-fold increase in serum EPO was reported for patients following the oral administration of 1 and 2 mg kg −1 of roxadustat, which is comparable to the 6 to 19-fold increase in serum EPO that we observed in rats treated with 2 and 10 mg kg −1 of roxadustat, respectively. 56Our study results are therefore relevant to patients with CKD anaemia treated with HIF-PHIs.Preliminary data from pooled phase 3 studies in patients not on dialysis have suggested that roxadustat may increase GFR and delay time to initiation of renal replacement therapy. 1 Although preliminary, these data would be consistent with findings in our studies.
In conclusion, our studies provide novel insights into the effects of pharmacological HIF activation on renal haemodynamics and identify NO as a major mediator of these effects.Our findings provide strong rational for additional investigations into the role of systemic HIF activation in renal energy homeostasis and O 2 metabolism and for clinical studies that assess renal haemodynamics and NO metabolism in patients treated with HIF-PHIs.

| Animal handling and general procedure overview
Eight to 12-week-old male Sprague Dawley rats weighing 300-350 g were purchased from Charles River, Germany.Rats had free access to standard chow (Lantmannen, Kimstad, Sweden) and water.For surgical procedures, rats were anaesthetized with 120 mg kg −1 of thiobutabarbital (Inactin ® ; Sigma-Aldrich, Steinheim, Germany).All surgical procedures were performed on a temperature-controlled operating station to maintain body temperature at 37°C.Spontaneous breathing was facilitated by placement of a tracheostomy tube.Rats were euthanatized by intravenous injection of saturated KCl solution and kidneys, liver, heart and lung were harvested for further analyses.
All animal handling and procedures were performed in accordance with European Guidelines for the Care and Use of Laboratory Animals and were reviewed, approved and monitored by the regional Animal Care and Ethics Committee responsible for Uppsala University (5.8.18-06724/2018).

| Drug administration
Roxadustat or molidustat were dissolved in vehicle [10% ethanol (70% grade) and 90% corn oil] and administered i.p.; roxadustat (FG-4592, Cayman Chemical, Ann Arbor, MI, USA) at a dose of 1, 2, 10, 20 or 30 mg kg −1 ; molidustat (BAY 85-3934, Cayman Chemical, MI Ann Arbor, USA) at a dose of 1 or 10 mg kg −1 .Given its longer half-life, two doses of roxadustat were administered 48 hours apart, with the second dose given 6 hours prior to tissue harvest or initiation of surgical procedures.Three doses of 1 mg kg −1 of molidustat were administered 24 hours apart and two doses of 10 mg kg −1 of molidustat were given 48 hours apart, with the final doses given 4 hours prior to tissue harvest or initiation of surgical procedures.L-NAME (Sigma-Aldrich, Steinheim, Germany) was dissolved in the same vehicle and was injected i.p. at a dose of 10 mg kg −1 on the day of surgery at the time of final HIF-PHI administration.

| In vivo studies
Eight to 12 rats were randomly assigned to experimental cohorts and treated with either vehicle or HIF-PHI.GFR was assessed in anaesthetized rats prior to abdominal surgery by means of measuring FITC-sinistrin clearance utilizing the MediBeacon transdermal detection system, which was placed on the shaved chest (MediBeacon, Mannheim, Germany).Background measurements were obtained over a period of 5 minutes prior to tail vein injection of FITC-sinistrin (5 mg 100 g −1 body weight).Blood FITC-sinistrin concentrations were recorded transdermally over 30-45 minutes and FITCsinistrin clearance was calculated according to the manufacturer's instructions using software provided by MediBeacon.
After completion of GFR measurements, a polyethylene catheter (AgnTho's, Stockholm, Sweden) was inserted into the left femoral artery for invasive BP and HR measurements (ADInstruments, Sydney, Australia) and for obtaining arterial blood samples.Standard Ringer's solution (5 mL kg −1 hr −1 ) was infused into the left femoral vein.After catheterization of the urinary bladder for urine drainage, the left kidney was exposed by flank incision, immobilized in a plastic cup and covered with a saline-soaked cotton pad.The left ureter was catheterized for urine collections.A Transonic Doppler ultrasound probe (Transonic Systems, Ithaca, NY, USA) was placed on the renal artery to assess RBF.Rats were allowed to recover from abdominal surgery for 45 minutes before initiation of haemodynamic measurements.Prior to study completion the left renal vein was catheterized for blood collections.Tissue was harvested for gene expression analysis after the study was completed.
Arterial BP and RBF were measured over a period of 30 minutes.Urine was continuously collected over this time period, and urine volume was determined gravimetrically.Arterial and venous blood samples were obtained after the completion of haemodynamic measurements.Blood gases, haemoglobin concentrations and serum chemistries were analyzed utilizing the iSTAT system (Abbott Laboratories, Green Oaks, IL, USA).Hematocrit was determined by capillary centrifugation using a Fresco 21 centrifuge (ThermoFisher Scientific, Waltham, MA, USA).Urine electrolyte concentrations were determined by flame photometry (IL543, Instrumentation Lab, Milan, Italy).Serum iron levels, UIBC, TIBC, total cholesterol, triglycerides and glucose were measured in renal vein blood.Measurements were performed by the Clinical Pathology Laboratory of the University Animal Hospital, Uppsala, Sweden.Serum EPO levels were measured as described below in blood collected at the end of studies.

| Calculations of in vivo parameters
Arterial and renal venous O 2 content (O 2 ct) was determined by means of the following standard equation: O 2 ct (mL dL −1 ) = [Hb × 1.34 × (O 2 sat × 0.01)] (pO 2 × 0.023), where Hb is the haemoglobin concentration in g dL −1 , O 2 sat is the percentage saturation of haemoglobin with O 2 , pO 2 is partial pressure of O 2 in kPa, 1.34 represents the Hüfner coefficient in mL g −1 and 0.023 represents the O 2 solubility coefficient in mL kPa −1 dL −1 .Renal O 2 delivery was estimated from the arterial O 2 ct multiplied by RBF (mL min −1 ) and renal QO 2 was estimated from the arterio-venous O 2 ct difference multiplied by RBF.T Na expressed in μmol min −1 was assessed by calculating (P Na × GFR) − (U Na × U⩒), where P Na represents serum sodium concentration in mmol L −1 , GFR represents glomerular filtration rate in mL min −1 , U Na represents urine sodium concentration in mmοl L −1 and U⩒ represents urine flow rate in μL min −1 .FE Na was estimated as the ratio of sodium clearance to GFR, expressed as percentage, [(U Na × U⩒/ P Na )/GFR] × 100.RVR (mm Hg min mL −1 ) was calculated as MAP (mm Hg) divided by RBF (mL min −1 ).FF was estimated as GFR/[RBF × (1 − hematocrit)], where haematocrit is expressed as a decimal notation.

| Measurement of renal pO 2
Kidney pO 2 was measured in a dedicated cohort of vehicle or HIF-PHI-treated male Sprague Dawley rats (n = 8-9) using a Clark-type O 2 microelectrode with a tip diameter of 10 µm (Unisense, Aarhus, Denmark) as previously decribed. 57In brief, the electrodes were two-point calibrated at 37°C in water saturated with either Na 2 S 2 O 5 or air.Microelectrodes were inserted into the renal tissue with the aid of a micromanipulator.A linear correlation was obtained between O 2 | 11 of 14 BURMAKIN et Al.
tension and the electric current.O 2 tension measurements were carried out at 1 mm (cortex) and 4 mm (medulla) distance from the renal surface.This procedure was repeated more than or equal to five times in each left kidney.The averages for each respective depth from one animal were then considered as one experiment in the statistical analysis.

| RNA, DNA, protein analysis
Total RNA was extracted from whole tissue homogenates or from primary cells and purified utilizing the RNeasy Plus kit according to the manufacturer's instructions (Qiagen, Germany).RNA concentrations were measured with a NanoDrop microvolume spectrophotometer (ThermoFisher Scientific, Waltham, MA, USA) and subjected to reverse transcription using iScript™ cDNA Synthesis Kit (Bio-Rad Laboratories, Stockholm, Sweden).qPCR was performed on a Bio-Rad CFX96 platform using iTaq™ Universal SYBR ® Green Supermix.The comparative Ct-method was used for the analysis of relative mRNA expression levels according to the manufacturer's instructions (Bio-Rad Laboratories, Stockholm, Sweden).Relative mRNA levels were expressed as a fold-change over control.18S ribosomal RNA was used as internal control.Primer sequences are listed in Table S1.
RNA sequencing analysis was performed with whole renal cortex extract obtained from a dedicated cohort of rats treated with roxadustat 2 mg kg −1 or 30 mg kg −1 , molidustat 10 mg kg −1 or vehicle on an Illumina 2000 NGS platform (Illumina, San Diego, CA, USA).
For HIF immunohistochemistry, formalin-fixed, paraffinembedded kidney sections from rats and mice were stained with anti-HIF-1α polyclonal antibody (cat.# 10006421, Cayman Chemical Co., Ann Arbor, MI, USA) or anti-HIF2-α PM8 antiserum, a gift from Peter Ratcliffe (University of Oxford, Oxford, UK) as described previously. 58For RNA fluorescent in situ hybridization, Epo transcripts were detected in formalin-fixed, paraffin-embedded kidney sections using Rn-Epo (cat.# 455901, Advanced Cell Diagnostics, Hayward, CA, USA) and the RNAscope ® multiplex fluorescent kit according to the manufacturer's instructions.

| Statistical analysis
Statistical analysis was performed by either Student's twotailed t test for two-group comparisons and by one or twoway ANOVA followed by Tukey's post-hoc analysis for multi-group comparisons; in addition, one-way ANOVA followed by Dunnett's multiple comparison test was performed (Prism 9, GraphPad Software, La Jolla, CA, USA); a P value of <.05 was considered statistically significant.All data are presented as mean ± SEM.

F I G U R E 1
HIF-PHI administration schedule.Schematic illustrating the timing of drug administration (arrows) in relation to the time point of the in vivo analysis or tissue harvest (red asterisk).Blue arrows denote drug or vehicle administration for all cohorts, the orange arrow indicates that an additional dose was given for the molidustat 1 mg kg −1 cohort at the -24-h time point.i.p., intraperitoneal; moli, molidustat; roxa, roxadustat | 3 of 14 BURMAKIN et Al.

F I G U R E 2
Systemic HIF-PHI administration increases renal blood flow and glomerular filtration.(A) Schematic illustrating the anatomic sites of haemodynamic measurements and blood or urine collections; aorta, arteries are shown in red, inferior vena cava and renal veins in blue, ureter and renal pelvis are shown in yellow.(B) Mean arterial pressure (MAP) readings from femoral artery in vehicle-treated control and HIF-PHI-treated cohorts.(C) Renal blood flow (RBF) per kidney, renal vascular resistance (RVR), glomerular filtration rate (GFR) assessed by means of FITC sinistrin clearance and filtration fraction in vehicle-treated and HIF-PHI-treated cohorts.

T A B L E 1
Haemodynamic and blood parameters in vehicle-and HIF-PHI-treated cohorts Note: Data are represented as mean ± SEM; one-way ANOVA followed by Tukey's post-hoc analysis.Abbreviations: BP, blood pressure; BUN, blood urea nitrogen; Hb, haemoglobin; HR, heart rate; paO 2 , partial arterial O 2 pressure (femoral artery); SaO 2 , percentage arterial haemoglobin saturation with O 2 .High-dose HIF-PHI treatment increases renal O 2 consumption disproportionally and reduces metabolic efficiency.Upper panels, O 2 delivery to the left kidney expressed as volume of O 2 (mL) delivered per minute per kidney (ki) and renal O 2 consumption rate (QO 2 ) expressed as volume of consumed O 2 (mL) per minute per kidney.Left lower panel, tubular sodium transport rate (T Na ) in relation to GFR per kidney; shown are the data for all animals in the study.Colour scheme used here corresponds to the colour schemes used in the other panels.Right lower panel, tubular transport efficiency assessed as transported sodium per consumed O 2 (T Na /QO 2 ).For ratio calculations, QO 2 was converted from mL min −1 to μmol min −1 .Data are represented as mean ± SEM; one-way ANOVA followed by Tukey's post-hoc analysis; n = 12 for vehicle control, n = 8 for all other groups; † P < .01 and ‡ P < .001compared with vehicle control.moli, molidustat; veh, vehicle † P < .01. ‡ P < .001compared with vehicle control.F I G U R E 3 | 5 of 14 BURMAKIN et Al.
Urine parameters per kidney in vehicle-and HIF-PHI-treated cohorts T A B L E 2Note: Data are represented as mean ± SEM; one-way ANOVA followed by Tukey's post-hoc analysis.Abbreviations: FE Na , fractional excretion of sodium (calculated per kidney); P Na , serum sodium concentration; T Na , tubular sodium transport per kidney; U⩒, urine flow rate per kidney; U Na , urine sodium concentration (left ureter).† P < .01. ‡ P < .001compared with vehicle control.