Suppression of NRF2 Activity by HIF-1α Promotes Fibrosis after Ischemic Acute Kidney Injury

Acute kidney injury (AKI) is a rapid decline in renal function and can occur after ischemia/reperfusion injury (IRI) to the tubular epithelia. The nuclear factor erythroid-2-related factor 2 (NRF2) pathway protects against AKI and AKI-to-chronic kidney disease (CKD) progression, but we previously demonstrated that severe IRI maladaptively reduced NRF2 activity in mice. To understand the mechanism of this response, we subjected C57BL/6J mice to unilateral kidney IRI with ischemia times that were titrated to induce mild to severe injury. Mild IRI increased NRF2 activity and was associated with renal recovery, whereas severe IRI decreased NRF2 activity and led to progressive CKD. Due to these effects of ischemia, we tested the hypothesis that hypoxia-inducible factor-1α (HIF-1α) mediates NRF2 activity. To mimic mild and severe ischemia, we activated HIF-1α in HK-2 cells in nutrient-replete or nutrient-deficient conditions. HIF-1α activation in nutrient-replete conditions enhanced NRF2 nuclear localization and activity. However, in nutrient-deficient conditions, HIF-1α activation suppressed NRF2 nuclear localization and activity. Nuclear localization was rescued with HIF-1α siRNA knockdown. Our results suggest that severe ischemic AKI leads to HIF-1α-mediated suppression of NRF2, leading to AKI-to-CKD progression.


Introduction
Acute kidney injury (AKI) is a rapid decline in renal function that is associated with high morbidity, mortality, and healthcare costs [1][2][3][4][5]. Survivors also have a higher risk of developing chronic kidney disease (CKD) and end-stage renal disease (ESRD) [6]. Treatments for AKI and AKI-to-CKD progression are urgently needed.
Numerous studies demonstrate that NRF2 activation is protective in IRI-AKI. NRF2 −/− mice are markedly sensitized to IRI whereas pharmacologic activation of NRF2 protects against injury [15,16]. We and others show that mice with constitutive NRF2 activation are protected against AKI-to-CKD progression [17,18]. Interestingly, wild-type mice exposed to severe IRI exhibit a potentially maladaptive decrease in kidney NRF2 activity [17]. Similar reductions in NRF2 activity are seen in folic acid-induced nephropathy in mice and humans [19]. The mechanisms for reduced NRF2 activity are unknown.
Because the activity of NRF2 and HIF-1α is simultaneously affected by IRI-AKI, we hypothesized that these pathways may crosstalk in the kidney. Studies show that HIF-1α reduces NRF2 activity or that NRF2 activates or inhibits HIF-1α [28][29][30][31][32][33][34][35], but these interactions have not been extensively studied in AKI. In this study, we demonstrate that the IRI severity determines whether NRF2 is activated or suppressed, and this effect is mediated by HIF-1α.

Animals
IACUC protocol approval was obtained at the University of Pittsburgh. Mice were afforded the ethical and scientific standards recommended by the Guide for the Care and Use of Laboratory Animals. The authors complied with the ARRIVE guidelines [36]. For the IRI study, 9-to 10-week-old (mean 24.1 g; range 21.7-27.3 g) male, wild-type C57BL/6J mice (#000664, Jackson Labs, Bar Harbor, ME, USA) were housed on a 12:12 h light-dark cycle with free access to water and Prolab ® IsoPro ® RMH 3000 5P75 diet. For the IOX2 study, 6-to 7-week-old (mean 23.8 g; range 21.9-25.7 g) male, wild-type C57BL/6J mice were provided the same conditions.

Ischemia/Reperfusion Injury
IRI was conducted according to modification of a published protocol [17]. Mice were randomly assigned to three groups of IRI severity: mild (15 min of ischemia), moderate (20 min of ischemia), and severe (30 min of ischemia). Briefly, mice were anesthetized with IP injection of ketamine (100 mg/kg) and xylazine (10 mg/kg), and surgeries were performed on a heating pad to maintain body temperature. After abdominal incision, the left renal pedicle was isolated, and an atraumatic surgical clamp placed (#RS-5459, Roboz, Gaithersburg, MD, USA). After confirming clamp placement, the kidney was observed for visible ischemia. The clamp remained for 15, 20, or 30 min before removal. Reperfusion was visually confirmed before closure. To assess renal function of the injured kidney, the contralateral kidney was removed 24 h prior to sacrifice or at day 10 for the day 28 endpoint. The mice were euthanized at 3 h or days 1, 3, 10, or 28 after clamp removal, at which time kidneys and sera were harvested. The total number of mice at time of sacrifice are as follows: Day 1, 15 min (3), 20 min (4), and 30 min (4); Day 3, 15 min (4), 20 min (4), and 30 min (4); Day 10, 15 min (3), 20 min (4), and 30 min (4); and Day 28, 15 min (2), 20 min (3) and 30 min (0). All three mice in the Day 28, 30 min group died after contralateral kidney removal on day 10.

Histology
Kidney tissue was fixed in 10% buffered formalin, paraffin embedded, and sectioned at 3 µm and stained with Periodic Acid Schiff (PAS) (MilliporeSigma) or Masson's Trichrome Stain (MTS) (ThermoFisher Scientific, Waltham, WA, USA). The PAS images were subjected to pathologic scoring by an investigator (RJT) masked to group assignment according to the following scale: 0, no injury; 1, 1 to 25% of parenchyma affected by injury; 2, 26 to 50% involvement; 3, 51 to 75% involvement; and 4, 76 to 100% involvement. Injury was defined as tubular dilation, presence of casts, cell sloughing, or loss of brush borders.

Immunohistochemistry
Paraffin-embedded sections were hydrated through a series of xylene and graded alcohols. To the deparaffinized tissues, citrate-based antigen unmasking (Vector Laboratories, Burlingame, CA, USA) was performed followed by blocking for 1 h in 10% normal donkey serum before overnight incubation in primary antibody at 4 • C. Tissue was incubated in biotinylated secondary antibody (Jackson ImmunoResearch Laboratories, West Grove, PA, USA) for 2 h followed by a 1 h incubation in ABC reagent and developed with AEC reagent (both from Vector Laboratories). Slides were developed in the same batch and for the same amount of time.
For HIF-1α knockdown experiments, cells were transfected with Lipofectamine ® RNAiMAX Reagent (#13778-075, Invitrogen, Carlsbad, CA, USA) and HIF-1α siRNA (#sc-35561, Santa Cruz Biotechnology, Dallas, TX, USA) or control siRNA-A (#sc-37007, Santa Cruz Biotechnology) in Opti-MEM ® medium (Gibco Laboratories) at a final siRNA concentration of 30 pM per well. The siRNA-lipid complex was added directly to the cell suspension. The cells were plated, and the media changed after 6 h. Twenty-four hours later, the cells were exposed to 200 µM CoCl 2 in nutrient deficient conditions for 16 h before lysis.

Immunofluorescence
Cells were fixed in 2% paraformaldehyde for 30 min before quenching with 0.2 M NH 4 Cl for 10 min and permeabilization with 10% BSA in 0.1% Triton X-100 for 10 min. In a humidified chamber, cells were incubated overnight in primary antibody at 4 • C. Cells were incubated in fluorescent-labeled secondary antibody for 1 h at 4 • C. Imaging was performed by using a Leica TCS SP5 STED CW confocal microscope (Leica Microsystems Inc., Buffalo Grove, IL, USA).

Immunoblot
Tissue homogenates were prepared by douncing in pre-chilled radioimmunoprecipitation buffer (RIPA) supplemented with 1× Halt™ Protease & Phosphatase Single-Use Inhibitor Cocktail (ThermoScientific, Rockford, IL, USA) and centrifuged at 16,000× g for 15 min at 4 • C to recover the supernatant.
Protein concentrations were determined with the Pierce™ BCA Protein Assay Kit (ThermoScientific, Rockford, IL, USA). Samples were boiled in Laemmli sample buffer for 10 min. Equivalent protein concentrations were subjected to SDS-PAGE prior to transfer to PVDF membrane. The membrane was blocked in 5% nonfat milk and incubated overnight in primary antibody at 4 • C. The membrane was incubated in horseradish peroxidaseconjugated secondary antibody for 1 h before detection with Pierce™ SuperSignal ® West-Pico Chemiluminescent Substrate (ThermoScientific).

Co-Immunoprecipitation
The cell extracts were centrifuged at 8000× g for 10 min at 4 • C. The supernatant was precleared by incubation for 20 min with 40 µL of 50% slurry of Protein G-conjugated to Sepharose (Invitrogen, Carlsbad, CA, USA) on ice before centrifugation at 15,300× g at 4 • C. The supernatant was incubated overnight in primary antibody (HIF-1α) at 4 • C. Protein G-conjugated to Sepharose (30 µL 50% slurry) was added with end-over-end mixing for 2 h at 4 • C. The immunoprecipitate on beads was washed once with 1 mL ice cold HEPES-buffered saline (150 mM NaCl; 10 mM HEPES, pH 7.4). Protein was eluted with 30 µL NuPAGE LDS 2x sample buffer (Invitrogen) including 10% reducing agent for 5 min at 90 • C. Cell extracts (1.5%) or immunoprecipitants were immunoblotted and developed with Bio-Rad Clarity ECL (Bio-Rad Laboratories, Hercules, CA, USA) and a ChemiDoc Imaging System (Bio-Rad Laboratories).

Quantitative, Real-Time Reverse Transcriptase Polymerase Chain Reaction
Total RNA was extracted with TRIzol ® Reagent (Ambion ® , Carlsbad, CA, USA), and cDNA was generated with the RevertAid Reverse Transcriptase Kit (ThermoFisher Scientific, Pittsburgh, PA, USA). Reaction was performed by using iTAQ™ Universal SYBR ® Green Supermix (Bio-Rad Laboratories) and the CFX Connect™ Real-Time Sys-  [46]. Results were normalized to β-Actin and gene expression was determined with the 2 −∆∆Ct method [47]. Melt curve was performed to ensure specificity for a single product.

Statistical Analysis
A Shapiro-Wilk's W normality test was performed. For two group comparisons, an unpaired, two-tailed Student's t-test was performed. For comparisons between three or more groups, a one-way ANOVA was performed, followed by a Tukey's multiple comparisons test. To assess between-group differences, a two-way ANOVA was performed followed by a Tukey's or Dunnett's multiple comparisons test. Results are reported as mean ± S.E.M. GraphPad Prism 7.03 software (GraphPad Software Inc., La Jolla, CA, USA) was used. The threshold for significance was p < 0.05.

Severe Ischemia Promotes AKI-to-CKD Progression
C57BL/6J mice were subjected to unilateral IRI to evaluate AKI-to-CKD progression [17,48]. Clamp times of 15, 20, or 30 min were selected to induce mild, moderate, or severe IRI, respectively. Mild and moderate IRI led to a transient increase in serum creatinine and blood urea nitrogen (BUN) peaking at day 1. Severe IRI led to an inexorable rise in serum creatinine and BUN as well as death after contralateral kidney removal on day 10 ( Figure 1A,B). At day 10, the mRNA expression of tubular injury markers, kidney injury molecule 1 (Kim1) and neutrophil gelatinase-associated lipocalin (Ngal), also increased with longer ischemia times ( Figure 1C,D). Histology revealed a graded increase in tubular injury with longer ischemia times and was characterized by loss of brush borders, flattening of the epithelia, and cell death ( Figure 1E).
Antioxidants 2022, 11, x FOR PEER REVIEW 6 of molecule 1 (Kim1) and neutrophil gelatinase-associated lipocalin (Ngal), also increas with longer ischemia times ( Figure 1C,D). Histology revealed a graded increase in tubu injury with longer ischemia times and was characterized by loss of brush borders, flatte ing of the epithelia, and cell death ( Figure 1E).  Staining reveals increasing amounts of tubular damage with longer ischemia times (arrowheads). Contralateral kidney from day 1 is used as the control. Bar = 100 µm. Enlargement of the boxed area is shown to the right of each image. Included are tubular injury score at days 1 and 3. Bars are mean ± S.E.M. One-way ANOVA. * p < 0.05, ** p < 0.01, and *** p < 0.001.
To assess AKI-to-CKD progression, we examined the development of renal fibrosis. Mice exposed to mild IRI did not exhibit fibrotic lesions on Masson's trichrome stain, while longer ischemia times led to fibrosis ( Figure 2A). As noted above, none of the mice exposed to severe IRI survived to day 28 after contralateral kidney removal on day 10. In agreement with our histology results, fibronectin (FN1) protein expression was increased with longer ischemia times on day 10 ( Figure 2B, Supplemental Figure S1). Overall, mild IRI led to reversible AKI with minimal to no chronic disease, while moderate and severe IRI led to tubulointerstitial fibrosis, indicative of AKI-to-CKD progression.
at day 10. Kim1 and Ngal increase in the 20-and 30-min groups compared to control. The contralateral kidney from the 15-min group is used as the control. Bars are mean ± S.E.M. One-way ANOVA. * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001. (E) Representative PAS-stained images are shown. Staining reveals increasing amounts of tubular damage with longer ischemia times (arrowheads). Contralateral kidney from day 1 is used as the control. Bar = 100 µm. Enlargement of the boxed area is shown to the right of each image. Included are tubular injury score at days 1 and 3. Bars are mean ± S.E.M. One-way ANOVA. * p < 0.05, ** p < 0.01, and *** p < 0.001.
To assess AKI-to-CKD progression, we examined the development of renal fibrosis. Mice exposed to mild IRI did not exhibit fibrotic lesions on Masson's trichrome stain, while longer ischemia times led to fibrosis ( Figure 2A). As noted above, none of the mice exposed to severe IRI survived to day 28 after contralateral kidney removal on day 10. In agreement with our histology results, fibronectin (FN1) protein expression was increased with longer ischemia times on day 10 ( Figure 2B, Supplemental Figure S1). Overall, mild IRI led to reversible AKI with minimal to no chronic disease, while moderate and severe IRI led to tubulointerstitial fibrosis, indicative of AKI-to-CKD progression. Thirty-minute ischemia leads to animal death by day 28. Kidney sections from the 30-min group display extensive collagen deposition (blue staining, arrowheads). Contralateral kidney from day 3 is used as the control. Bar = 100 µm. Enlargement of the boxed area is shown to the right of each image. (B) Immunoblot of FN1 protein expression at day 10. FN1 expression increased with longer ischemia times.

IRI Severity Determines NRF2 and HIF-1α Activity
Our previous study revealed a maladaptive decrease in NRF2 activity in wild-type mice during AKI, which we measured as a decrease in NQO1 protein expression [17]. We evaluated Nqo1 expression over time in response to differing IRI severities. Supporting our prior results, immunoblot and immunohistochemistry analyses revealed decreased NQO1 protein expression at day 1 for all ischemia times. Although NQO1 expression recovered by day 10 in mild IRI, it declined in severe IRI ( Figure 3A,C). Thirty-minute ischemia leads to animal death by day 28. Kidney sections from the 30-min group display extensive collagen deposition (blue staining, arrowheads). Contralateral kidney from day 3 is used as the control. Bar = 100 µm. Enlargement of the boxed area is shown to the right of each image. (B) Immunoblot of FN1 protein expression at day 10. FN1 expression increased with longer ischemia times.

IRI Severity Determines NRF2 and HIF-1α Activity
Our previous study revealed a maladaptive decrease in NRF2 activity in wild-type mice during AKI, which we measured as a decrease in NQO1 protein expression [17]. We evaluated Nqo1 expression over time in response to differing IRI severities. Supporting our prior results, immunoblot and immunohistochemistry analyses revealed decreased NQO1 protein expression at day 1 for all ischemia times. Although NQO1 expression recovered by day 10 in mild IRI, it declined in severe IRI ( Figure 3A,C). To determine the relationship of NRF2 and HIF-1α in progressive kidney injury, we examined mRNA expression of the NRF2 and HIF-1α targets genes of Nqo1 and Vegf, respectively, over time in our mild, moderate, and severe IRI groups. HIF-1α and NRF2 activity are inversely correlated during AKI. Mild IRI led to an early (day 1) and sustained increase of Nqo1 expression over the 10-day time course. Conversely, longer ischemia times led to delayed and transient increases in expression ( Figure 3D). Vegf expression decreased in mild and moderate IRI at day 1 while Nqo1 expression increased. However, To determine the relationship of NRF2 and HIF-1α in progressive kidney injury, we examined mRNA expression of the NRF2 and HIF-1α targets genes of Nqo1 and Vegf, respectively, over time in our mild, moderate, and severe IRI groups. HIF-1α and NRF2 activity are inversely correlated during AKI. Mild IRI led to an early (day 1) and sustained increase of Nqo1 expression over the 10-day time course. Conversely, longer ischemia times led to delayed and transient increases in expression ( Figure 3D). Vegf expression decreased in mild and moderate IRI at day 1 while Nqo1 expression increased. However, severe IRI increased Vegf expression at day 1 while Nqo1 expression decreased. At later time points, longer ischemia times led to decreased Vegf expression while Nqo1 expression increased ( Figure 3E). At an earlier time point, 3-h post-reperfusion, moderate IRI decreased Nqo1 expression and Vegf expression increased ( Figure 3F). Collectively, mild IRI led to rapid and sustained NRF2 activity and eventual restoration of NQO1 protein expression. However, in severe IRI, NRF2 activity is suppressed and associated with loss of NQO1 protein expression. HIF-1α activity could be responsible for the suppression of NRF2 activity.

NRF2 Activity Is Responsive to HIF-1α Activation and Nutrient Availability
To examine the relationship between NRF2 and HIF-1α, we induced HIF-1α activation in mice with IOX2, a potent inhibitor of PHD2 [37], and assessed mRNA expression of HIF-1α and NRF2 target genes. IOX2-mediated HIF-1α activation increased mRNA expression of HIF-1α target genes while simultaneously decreasing expression of NRF2 target genes ( Figure 4A,B). These results are consistent with HIF-1α mediating NRF2 activity. time points, longer ischemia times led to decreased Vegf expression while Nqo1 expression increased ( Figure 3E). At an earlier time point, 3-h post-reperfusion, moderate IRI de creased Nqo1 expression and Vegf expression increased ( Figure 3F). Collectively, mild IR led to rapid and sustained NRF2 activity and eventual restoration of NQO1 protein ex pression. However, in severe IRI, NRF2 activity is suppressed and associated with loss o NQO1 protein expression. HIF-1α activity could be responsible for the suppression o NRF2 activity.

NRF2 Activity Is Responsive to HIF-1α Activation and Nutrient Availability
To examine the relationship between NRF2 and HIF-1α, we induced HIF-1α activa tion in mice with IOX2, a potent inhibitor of PHD2 [37], and assessed mRNA expression of HIF-1α and NRF2 target genes. IOX2-mediated HIF-1α activation increased mRNA ex pression of HIF-1α target genes while simultaneously decreasing expression of NRF2 tar get genes ( Figure 4A,B). These results are consistent with HIF-1α mediating NRF2 activity To more fully understand how NRF2 and HIF-1α activity is affected by IRI severity in our model, we utilized an in vitro system that could mimic mild and severe IRI. Because ischemia impairs nutrient and oxygen delivery, we exposed HK-2 cells to either nutrien replete (complete media) or nutrient deficient (Hank's balanced salt solution) conditions [26,41,42] in the presence or absence of HIF-1α activators, CoCl2 and IOX2 [37][38][39][40]. We postulated that HIF-1α activation in nutrient-deficient conditions would simulate a more severe injury compared to nutrient-replete conditions. In nutrient-replete conditions CoCl2 dose-dependently increased both VEGF and NQO1 mRNA and protein expression ( Figure 5A,C). Interestingly, in nutrient deficient conditions, CoCl2 dose-dependently in creased VEGF mRNA and protein expression but suppressed NQO1 mRNA and protein expression ( Figure 5B,D). These results are consistent with NRF2 activity being increased in mild IRI and decreased in severe IRI. This appears to be contingent on HIF-1α activation and nutrient availability. To more fully understand how NRF2 and HIF-1α activity is affected by IRI severity in our model, we utilized an in vitro system that could mimic mild and severe IRI. Because ischemia impairs nutrient and oxygen delivery, we exposed HK-2 cells to either nutrient replete (complete media) or nutrient deficient (Hank's balanced salt solution) conditions [26,41,42] in the presence or absence of HIF-1α activators, CoCl 2 and IOX2 [37][38][39][40]. We postulated that HIF-1α activation in nutrient-deficient conditions would simulate a more severe injury compared to nutrient-replete conditions. In nutrientreplete conditions, CoCl 2 dose-dependently increased both VEGF and NQO1 mRNA and protein expression ( Figure 5A,C). Interestingly, in nutrient deficient conditions, CoCl 2 dose-dependently increased VEGF mRNA and protein expression but suppressed NQO1 mRNA and protein expression ( Figure 5B,D). These results are consistent with NRF2 activity being increased in mild IRI and decreased in severe IRI. This appears to be contingent on HIF-1α activation and nutrient availability. To further explore the effect of HIF-1α activation and nutrient availability on NRF2 activity, we assessed NRF2 nuclear localization in the presence or absence of HIF-1α activators, IOX2 and CoCl2. As expected, both IOX2 and CoCl2 increased HIF-1α nuclear localization regardless of nutrient availability. In replete conditions, IOX2-and CoCl2-mediated HIF-1α activation increased NRF2 nuclear localization. However, although nutrient-deficient conditions alone increased NRF2 nuclear localization, this was suppressed in the presence of IOX2-and CoCl2-mediated HIF-1α activation ( Figure 6A,B). Cytosolic levels of NRF2 remained unchanged in either nutrient condition in the presence or absence of HIF-1α activation (data not shown). To support our immunoblot findings, immunofluorescence staining for NRF2 revealed its nuclear localization occurring in replete conditions with CoCl2-mediated HIF-1α activation. Conversely, in deficient conditions, CoCl2-mediated HIF-1α activation suppressed NRF2 nuclear localization ( Figure 6C). Collectively, these results are consistent with the likelihood that HIF-1α-mediated suppression of NRF2 activity in deficient or severe conditions is due to the lack of NRF2 nuclear localization. One-way ANOVA. * p < 0.05, ** p < 0.01, and **** p < 0.0001 compared to condition without CoCl 2 . (C,D) Representative immunoblots of whole-cell lysates of HIF-1α, VEGF, and NQO1 protein expression after 16 h exposure to increasing concentrations of CoCl 2 in either nutrient-replete or nutrient-deficient conditions. In nutrient-replete conditions, CoCl 2 dose-dependently increases VEGF and NQO1 expression. However, in nutrient-deficient conditions, CoCl 2 dose-dependently increases VEGF expression but not NQO1 expression.
To further explore the effect of HIF-1α activation and nutrient availability on NRF2 activity, we assessed NRF2 nuclear localization in the presence or absence of HIF-1α activators, IOX2 and CoCl 2 . As expected, both IOX2 and CoCl 2 increased HIF-1α nuclear localization regardless of nutrient availability. In replete conditions, IOX2-and CoCl 2 -mediated HIF-1α activation increased NRF2 nuclear localization. However, although nutrient-deficient conditions alone increased NRF2 nuclear localization, this was suppressed in the presence of IOX2-and CoCl 2 -mediated HIF-1α activation ( Figure 6A,B). Cytosolic levels of NRF2 remained unchanged in either nutrient condition in the presence or absence of HIF-1α activation (data not shown). To support our immunoblot findings, immunofluorescence staining for NRF2 revealed its nuclear localization occurring in replete conditions with CoCl 2 -mediated HIF-1α activation. Conversely, in deficient conditions, CoCl 2 -mediated HIF-1α activation suppressed NRF2 nuclear localization ( Figure 6C). Collectively, these results are consistent with the likelihood that HIF-1α-mediated suppression of NRF2 activity in deficient or severe conditions is due to the lack of NRF2 nuclear localization.  To substantiate that HIF-1α mediates the suppression of NRF2 nuclear localization and activity in nutrient-deficient conditions, we performed siRNA-mediated HIF-1α knockdown prior to exposure to deficient conditions and CoCl 2 . Our method achieved greater than 90% knockdown efficiency ( Figure 7A). Knocking down HIF-1α prior to exposure to deficient conditions and CoCl 2 increased NQO1 mRNA expression ( Figure 7B) and restored NRF2 nuclear localization ( Figure 7C, Supplemental Figure S2) suggesting HIF-1α mediates the suppression of NRF2 nuclear localization and activity in nutrient-deficient or severe conditions. nutrient-replete or nutrient-deficient conditions in the presence or absence of 200 µM CoCl2. In replete conditions, NRF2 nuclear localization increases with CoCl2-mediated HIF-1α activation. However, in deficient conditions, CoCl2-mediated HIF-1α activation suppresses NRF2 nuclear localization.
To substantiate that HIF-1α mediates the suppression of NRF2 nuclear localization and activity in nutrient-deficient conditions, we performed siRNA-mediated HIF-1α knockdown prior to exposure to deficient conditions and CoCl2. Our method achieved greater than 90% knockdown efficiency ( Figure 7A). Knocking down HIF-1α prior to exposure to deficient conditions and CoCl2 increased NQO1 mRNA expression ( Figure 7B) and restored NRF2 nuclear localization ( Figure 7C, Supplemental Figure S2) suggesting HIF-1α mediates the suppression of NRF2 nuclear localization and activity in nutrientdeficient or severe conditions. The effect of HIF-1α was not due to direct interaction because we could not co-immunoprecipitate HIF-1α and NRF2 in cells exposed to either nutrient condition in the presence or absence of CoCl2 ( Figure 8A). Overall, our results demonstrate that HIF-1α The effect of HIF-1α was not due to direct interaction because we could not coimmunoprecipitate HIF-1α and NRF2 in cells exposed to either nutrient condition in the presence or absence of CoCl 2 ( Figure 8A). Overall, our results demonstrate that HIF-1α influences NRF2 nuclear localization and target gene transcription in a manner that is dependent on nutrient availability. influences NRF2 nuclear localization and target gene transcription in a manner that is dependent on nutrient availability. Figure 8. HIF-1α and NRF2 do not directly interact in cells exposed to experimental conditions. HIF-1α is immunoprecipitated and immunoblots for NRF2 and HIF-1α were performed after exposure to replete or deficient conditions in the presence or absence of 200 µM CoCl2. In all cell culture conditions, there was no indication of co-immunoprecipitation of HIF-1α and NRF2. Cell lysates without immunoprecipitation are shown as a control.

Discussion
AKI remains a significant problem affecting hospitalized patients and contributes to the long-term risks of developing CKD and ESRD [6,49,50]. The development of CKD is often associated with the severity of AKI [51]. Prior investigations have identified tubular epithelial cell cycle arrest, microvascular rarefaction, and the development of interstitial fibrosis as contributors, but they are unlikely to be the only causes [52][53][54][55][56]. In this regard, we chose to study the relationship between NRF2 and HIF-1α, two well-known pathways involved in AKI-to-CKD pathogenesis.
HIF-1α is upregulated during IRI-AKI and activates pathways involved in vasculogenesis and glucose uptake and metabolism. HIF-1α expression is induced throughout the nephron including [57] in proximal tubular epithelial cells during the acute reperfusion phase of ischemic AKI [26,27,58]. As NRF2 has also been shown to be expressed in proximal tubules and other cortical nephron segments [59], it is possible that the HIF-1α and NRF2 pathways interact. Although it has been shown that short-term preischemic activation of HIF-1α reduces IRI severity, chronic activation may promote chronic kidney injury through inflammation, cell proliferation, and the development of fibrosis [60][61][62][63]. The downstream mediators of these opposing acute and chronic effects were not fully understood.
NRF2 is a cytoprotective pathway known to be protective in IRI-AKI. Genetic and pharmacologic NRF2 activation reduce AKI severity and prevent AKI-to-CKD progression [16][17][18]. Previously, we reported a maladaptive reduction in NRF2 activity in wildtype mice with severe IRI leading to progression to CKD [17]. Based on this result, we hypothesized that severity of IRI-AKI affects the expression of protective NRF2, which in turn affects long-term kidney outcomes. Figure 8. HIF-1α and NRF2 do not directly interact in cells exposed to experimental conditions. HIF-1α is immunoprecipitated and immunoblots for NRF2 and HIF-1α were performed after exposure to replete or deficient conditions in the presence or absence of 200 µM CoCl 2 . In all cell culture conditions, there was no indication of co-immunoprecipitation of HIF-1α and NRF2. Cell lysates without immunoprecipitation are shown as a control.

Discussion
AKI remains a significant problem affecting hospitalized patients and contributes to the long-term risks of developing CKD and ESRD [6,49,50]. The development of CKD is often associated with the severity of AKI [51]. Prior investigations have identified tubular epithelial cell cycle arrest, microvascular rarefaction, and the development of interstitial fibrosis as contributors, but they are unlikely to be the only causes [52][53][54][55][56]. In this regard, we chose to study the relationship between NRF2 and HIF-1α, two well-known pathways involved in AKI-to-CKD pathogenesis.
HIF-1α is upregulated during IRI-AKI and activates pathways involved in vasculogenesis and glucose uptake and metabolism. HIF-1α expression is induced throughout the nephron including [57] in proximal tubular epithelial cells during the acute reperfusion phase of ischemic AKI [26,27,58]. As NRF2 has also been shown to be expressed in proximal tubules and other cortical nephron segments [59], it is possible that the HIF-1α and NRF2 pathways interact. Although it has been shown that short-term preischemic activation of HIF-1α reduces IRI severity, chronic activation may promote chronic kidney injury through inflammation, cell proliferation, and the development of fibrosis [60][61][62][63]. The downstream mediators of these opposing acute and chronic effects were not fully understood.
NRF2 is a cytoprotective pathway known to be protective in IRI-AKI. Genetic and pharmacologic NRF2 activation reduce AKI severity and prevent AKI-to-CKD progression [16][17][18]. Previously, we reported a maladaptive reduction in NRF2 activity in wild-type mice with severe IRI leading to progression to CKD [17]. Based on this result, we hypothesized that severity of IRI-AKI affects the expression of protective NRF2, which in turn affects long-term kidney outcomes.
To test this, we determined in our murine model system a threshold of IRI severity that leads to progressive kidney disease. While moderate and severe IRI had similar increases in serum creatinine and BUN at day 1, only the severe IRI mice exhibited persistent elevation in these functional markers ( Figure 1A,B). As expected, the severe IRI mice also exhibited increased tubular injury markers Kim1 and Ngal ( Figure 1C,D) and renal fibrosis (Figure 2A,B).
We then examined how injury severity affected NRF2 activity. When assessing NQO1 protein, we found that all levels of injury acutely suppressed NQO1 at day 1. NQO1 eventually returned to baseline levels in mild IRI but continued to decline in severe IRI ( Figure 3A,C). When we measured NQO1 expression, we found that mild IRI led to an early and sustained increase of NRF2 activity. Conversely, moderate and severe IRI led to delayed and transient increases of NRF2 activity ( Figure 3D). Importantly, at day 1, severe IRI was associated with increased HIF-1α activity and decreased NRF2 activity ( Figure 3D,E). This finding suggests that in severe IRI HIF-1α activation has a suppressive effect on NRF2 activity. Our results also indicate that mild IRI is associated with a more rapid recovery of NRF2 activity which may protect against AKI-to-CKD progression [17,18].
Our findings help to clarify existing studies that examine NRF2 in AKI. In Swiss mice, bilateral IRI increases Nqo1 gene expression, but in CD-1 mice there was no change [15,64]. Whereas the latter study identified other increased NRF2 targets, Nqo1 is more specific for NRF2 [13]. In C57BL/6 mice given unilateral ischemia for 23 min, NRF2 mRNA and protein expression increased at 4 and 8 h but not at 24 h after ischemia [65]. Decreased NRF2 activity was found during AKI-to-CKD progression in folic acid nephropathy in C57BL/6 mice and after kidney injury in humans [19]. Similarly, aristolochic acid-induced AKI in C57BL/6 mice showed impaired NRF2 activation and expression of target genes [66]. Considering these different responses, it is possible that the type of AKI and severity of injury affects overall NRF2 responses.
It is known that the IRI model is affected by ischemia time, clamp type, surgical technique, mouse strain, surgeon/laboratory, and whether the injury is unilateral or bilateral [48]. In our unilateral IRI model, we varied only ischemia time to titrate the injury while keeping other variables constant. It is possible that prior studies showing NRF2 activation in AKI utilized mild injury, whereas studies showing decreases in NRF2 activity utilized more severe injury. Our results would suggest this is the case.
To examine the relationship between NRF2 and HIF-1α, we induced HIF-1α activation in mice with IOX2. IOX2-mediated HIF-1α activation decreased NRF2 target genes suggesting that HIF-1α mediates NRF2 activity ( Figure 4A,B). In our IRI model, we showed an inverse relationship between NRF2 and HIF-1α activities during recovery from different IRI severities. To further investigate this, we mimicked mild and severe ischemia conditions in vitro in HK-2 cells by using nutrient-replete or nutrient-deficient conditions, respectively. We then activated HIF-1α with CoCl 2 or IOX2. In nutrient-replete conditions, HIF-1α activation enhanced NRF2 nuclear localization and target gene transcription, but in nutrient-deficient conditions, NRF2 nuclear localization and activity was suppressed ( Figures 5A-E and 6A-C). This latter effect was directly dependent on HIF-1α because HIF-1α siRNA knockdown restored NRF2 nuclear localization and target gene transcription ( Figure 7B,C). A limitation of our study is that we did not perform our experiments in primary tubular cell lines. Nonetheless, our current results in HK-2 cells help to explain the results of our animal studies.
We also did not assess the von Hippel-Lindau (VHL) protein, which participates in the degradation of HIF-1α. It has been found that VHL deletion (and increased HIF1-α activity) in tubular cells can provide protection against AKI due to IRI and rhabdomyolysis [57,67]. Similar to the results we report above, we would predict that nutrient environment, or severity of kidney injury, would alter the effect of VHL deletion on HIF-1α-mediated regulation of NRF2 in our model. Notably, the prior studies in animals do not specifically assess injury severity.
The timing of NRF2 and HIF-1α activation are critical to the overall renoprotective effect. Prior studies show that preconditioning or early activation of the NRF2 or HIF-1α response ameliorates injury, whereas activation at later timepoints does not afford protection [16,18,62]. We demonstrate that more severe IRI in mice leads to higher HIF-1α and lower NRF2 activity at day 1. Early stages of AKI injury are critical time points at which NRF2 is required to protect against further injury [16]. The suppression of NRF2 could be a contributor to the AKI-to-CKD progression observed in severe injuries.

Conclusions
We demonstrate that severe AKI leading to CKD and fibrosis is associated with a maladaptive suppression of the protective NRF2 response. In vivo, this was associated with stronger HIF-1α activation. We show that in kidney cells in vitro, HIF-1α suppresses NRF2 activity under stringent nutrient-deficient conditions that would mimic severe AKI. We propose that the higher HIF-1α activity induced by severe injury leads to the suppression of NRF2, which predisposes the kidney to chronic fibrosis.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antiox11091810/s1, Figure S1: Longer ischemia times lead to more fibronectin accumulation. Mice were exposed to the indicated ischemia times and euthanized at 10 days. Fibronectin was assessed by western blot in Figure 2, with densitometry performed and presented.; Figure S2: HIF-1alpha knockdown increases NRF2 nuclear localization. Shown is densitometry from western blots assessing nuclear localization of NRF2 in HK-2 cells exposed to nutrient-deficient conditions and CoCl 2 . Two independent experiments were combined to generate this figure.