Protective Effects of Hif2 Inhibitor PT-2385 on a Neurological Disorder Induced by Deficiency of Irp2

Iron regulatory protein 2 (IRP2) deficiency in mice and humans causes microcytic anemia and neurodegeneration due to functional cellular iron depletion. Our previous in vitro data have demonstrated that Irp2 depletion upregulates hypoxia-inducible factor subunits Hif1α and Hif2α expression; inhibition of Hif2α rescues Irp2 ablation-induced mitochondrial dysfunction; and inhibition of Hif1α suppresses the overdose production of lactic acid derived from actively aerobic glycolysis. We wonder whether Hif1α and Hif2α are also elevated in vivo and play a similar role in neurological disorder of Irp2–/– mice. In this study, we confirmed the upregulation of Hif2α, not Hif1α, in tissues, particularly in the central nervous system including the mainly affected cerebellum and spinal cord of Irp2–/– mice. Consistent with this observation, inhibition of Hif2α by PT-2385, not Hif1α by PX-478, prevented neurodegenerative symptoms, which were proved by Purkinje cell arrangement from the shrunken and irregular to the full and regular array. PT-2385 treatment did not only modulate mitochondrial morphology and quality in vivo but also suppressed glycolysis. Consequently, the shift of energy metabolism from glycolysis to oxidative phosphorylation (OXPHOS) was reversed. Our results indicate that Irp2 depletion-induced Hif2α is, in vivo, in charge of the switch between OXPHOS and glycolysis, suggesting that, for the first time to our knowledge, Hif2α is a clinically potential target in the treatment of IRP2 deficiency-induced neurodegenerative syndrome.

Iron regulatory protein 2 (IRP2) deficiency in mice and humans causes microcytic anemia and neurodegeneration due to functional cellular iron depletion. Our previous in vitro data have demonstrated that Irp2 depletion upregulates hypoxia-inducible factor subunits Hif1α and Hif2α expression; inhibition of Hif2α rescues Irp2 ablationinduced mitochondrial dysfunction; and inhibition of Hif1α suppresses the overdose production of lactic acid derived from actively aerobic glycolysis. We wonder whether Hif1α and Hif2α are also elevated in vivo and play a similar role in neurological disorder of Irp2 −/− mice. In this study, we confirmed the upregulation of Hif2α, not Hif1α, in tissues, particularly in the central nervous system including the mainly affected cerebellum and spinal cord of Irp2 −/− mice. Consistent with this observation, inhibition of Hif2α by PT-2385, not Hif1α by PX-478, prevented neurodegenerative symptoms, which were proved by Purkinje cell arrangement from the shrunken and irregular to the full and regular array. PT-2385 treatment did not only modulate mitochondrial morphology and quality in vivo but also suppressed glycolysis. Consequently, the shift of energy metabolism from glycolysis to oxidative phosphorylation (OXPHOS) was reversed. Our results indicate that Irp2 depletion-induced Hif2α is, in vivo, in charge of the switch between OXPHOS and glycolysis, suggesting that, for the first time to our knowledge, Hif2α is a clinically potential target in the treatment of IRP2 deficiency-induced neurodegenerative syndrome.
IRP1 and IRP2 are both iron-regulatory RNA binding proteins that regulate the expression of a series of iron-related genes at the post-transcriptional levels (Hentze et al., 2010;Anderson et al., 2012). Under conditions of iron deficiency, the IRE in the target mRNA can be recognized and bound by IRPs, but the consequence of IRP binding depends on the position of the IRE on the mRNA of the target genes. If the IRE is in the 5untranslated region (UTR) of the target mRNA, the binding of IRPs may inhibit the translation of the genes, including L-and H-ferritin and FPN1; if the IRE is in the 3 -UTR, the binding of IRPs may stabilize the mRNA, such as that of transferrin receptor 1 (TfR1) (Casey et al., 1988;Müllner et al., 1989) and divalent metal transporter 1 (DMT1) (Tybl et al., 2020). When cellular iron is sufficient, IRP1 binds to a (4Fe-4S) cluster and, therefore, gains aconitase activity and loses the ability to bind IRE, whereas IRP2 is removed by iron and oxygen-mediated proteasome degradation (Salahudeen et al., 2009;Vashisht et al., 2009) to avoid the excessive iron uptake and to promptly store excess intracellular iron and/or export excess iron.
Studies in animal models have shown that IRPs also play an important role in the regulation of systemic iron homeostasis. It has been reported that Irp2 −/− mice suffer from microcytic anemia (Cooperman et al., 2005;Galy et al., 2005), neurologic defects (LaVaute et al., 2001;Jeong et al., 2011), and diabetes (Santos et al., 2020). The cause of these symptoms is considered to be lack of functional iron in erythroblast progenitors, the cells in central nervous system (CNS), and β cells of Irp2 −/− mice. The patient with bi-allelic loss-of-function variants in the gene IREB2 leading to an absence of IRP2 also presented neurological and hematological features (Costain et al., 2019), similar, but much more severely, to the observation in Irp2 −/− mice. The symptoms could be caused by the deficiency of Fe-S biogenesis, which further compromised the mitochondrial quality (Li et al., 2018(Li et al., , 2019. Recently, we found that mitochondrial dysfunction was closely associated with the reduced expression of a number of genes that are involved in Fe-S biogenesis and mitochondrial respiratory chain (Li et al., 2018). Further investigation revealed that Irp2 may function as a key to switch the metabolism between aerobic glycolysis and oxidative phosphorylation (OXPHOS), mediated by upregulation of hypoxia-inducible factor subunits Hif1α and Hif2α in mouse embryonic fibroblasts (MEFs) (Li et al., 2019). Hif1 and Hif2 are two important transcription factors that can regulate the expression of a series of genes. Active Hif is a heterodimer, composed of an α subunit (Hif1α or Hif2α) and a β subunit (Hif1β, also called ARNT), and can bind hypoxia-responsive element (HRE), which is a very important mechanism in intestinal iron absorption (Mastrogiannaki et al., 2009;Shah et al., 2009) and under other conditions, such as cancer (Keith et al., 2011) and ischemia (Kapitsinou et al., 2014;Barteczek et al., 2017). IRP1 can bind to the IRE in the 5 -UTR of Hif2α mRNA to regulate Hif2α at the post-transcriptional level (Sanchez et al., 2007;Zimmer et al., 2008). In Irp1 −/− mice, elevated Hif2α upregulates erythropoietin (EPO), causing the mice to develop polycythemia and pulmonary hypertension (Anderson et al., 2013;Ghosh et al., 2013;Wilkinson and Pantopoulos, 2013). Interestingly, in Irp2 −/− MEFs, we found that Hif1α and Hif2α were both upregulated, which switches the metabolism type from OXPHOS to glycolysis (Li et al., 2019). Inhibition of both Hif1α and Hif2α reversed the energy metabolism.
In this study, we confirmed the elevated Hif2α, not Hif1α, in Irp2 −/− mice. The upregulated glycolytic pathway-related proteins were also observed and associated with the enhanced glycolysis, while downregulated frataxin (Fxn) and iron-sulfur cluster scaffold protein (IscU), respectively, were observed and associated with deficiency of iron-sulfur clusters. Consequently, the expression of electron transport chain (ETC) subunits was reduced and OXPHOS was weakened. After the inhibition of Hif2α by PT-2385, the energy metabolism was shifted from glycolysis to OXPHOS in Irp2 −/− mice, the histological and behavioral indicators were restored, and neurodegenerative symptoms were alleviated. Our results indicate that the neurodegenerative disorder induced by the loss of Irp2 is, at least partially, mediated by the upregulated Hif2α.

Mice
The Irp2 +/− heterozygous mice were obtained by crossing the purchased Irp2 −/− mice (purchased from MMRRC at UC Davis, United States, Cat. No. 030490-MU) with wild-type (WT) C57/BL6 mice. Both Irp2 −/− and WT mice used in the experiment were the descendants of Irp2 +/− heterozygous mice. The mice were fed a standard rodent pellet diet (200 mg iron/kg) and maintained a constant 12-h light/dark cycle. All procedures were carried out according to the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Animal Experimentation Administration of Nanjing University. Both PT-2385 and PX-478 were dissolved in DMSO, diluted with normal saline, and injected intraperitoneally into 6-monthold male Irp2 −/− mice for 1 month. The injection doses of PT-2385 and PX-478 were 0.4 mg/kg body weight and 5 mg/kg body weight, respectively, and the injection was carried out every other day.

Hang Tests
Hang test was performed to assess the grip strength. In the hang test, mice were allowed to grip a wire mesh square that was then inverted. The latency time that mice could hang on to an inverted wire mesh square before falling was measured, and each mouse was tested for three times with an interval of 5 min (n = 10).

Rotarod Tests
The motor functions of balance and coordination were assessed using an accelerating rotarod (Jiangsu SANS Technology Co., Ltd.). The staying time of mice on the rotating rod (the rotating rod accelerated from 4 to 40 rpm within 5 min) was recorded, and each mouse was tested for three consecutive times (n = 10).

Hematoxylin-Eosin Staining
In hematoxylin-eosin (H&E) staining, tissue sections were dealt with the following steps: dewaxed for 10 min in xylene twice, hydrated for 5 min in each 100-50% ethanol gradient buffers, rinsed for 5 min in running water at room temperature, and stained with hematoxylin for 10 min and then in Eosin Y for 10 min. Slides were dehydrated through gradual ethyl alcohol solutions for imaging (n = 3).

Electron Microscopy
The cerebellum and spinal cord were separated in the size of rice grains, placed in a mixed solution of 2% paraformaldehyde and 0.1 M cacodylate for 30 min at room temperature, and then stored at a constant temperature of 4 • C (inside a refrigerator). The samples were rinsed once or twice, dehydrated through a series of ethanol from 50 to 100%, and then propylene oxide was used instead of ethanol. The samples were stored in 50% propylene oxide and 50% EPON resin (1:1 mix) for 1 h and then placed in pure EPON. The samples were transferred to fresh EPON in molds or BEEM Embedding Capsules, which were filled carefully to avoid air bubbles, and kept at 60 • C for at least 24 h. Samples were observed and photographed by using HT7800 electron microscope at 80 keV, and electron micrographs were commented by Hitachi TEM system (n = 3).

Western Blot Analysis
The total protein of each entire tissue was extracted and analyzed (25-35 µg total protein/lane) by 7.5-12.5% SDS-PAGE at 100 V, transferred onto nitrocellulose membrane at 250 mA for 1.5 h, and analyzed by immunoblotting. The information of the primary antibody is as follows: anti-ferritin light chain (cat# 69090), Hif2α (cat# 109616), NcoA4 (cat# ab86707), and SdhB (cat# 178423) from Abcam (Cambridge, MA, United States); anti-TfR (cat# 136800) from Zymed (San Francisco, CA, United States); anti-beta-actin (cat# BM0627) from Boster (Wuhan, China); anti-Hk2 (cat# 22029-1-AP), Glut1 (cat# 21829-1-AP), LdhA (cat# 19987-1-AP), Ndufs1 (cat# 12444-1-AP), and Uqcrfs1 (cat# 18443-1-AP) from Proteintech Group Inc. (Chicago, IN, United States); anti-Hif1α (cat# 14179) from Cell Signaling Technology Inc. (Shanghai, China); anti-ferritin heavy chain (cat# AJ1290b) from ABGENT (San Diego, CA, United States), and anti-Fxn, IscU, Irp1, and Irp2 (polyclonal, self-made, raised from rabbits). All the self-made antibodies were validated in previous studies (Li et al., 2018(Li et al., , 2019. When it is necessary to detect multiple proteins in one blot and the molecular weight of the protein is different, we cut the blotted nitrocellulose membrane according to the molecular weight and then incubate with different antibodies. When the molecular weights are very close, we ran multiple gels with the same prepared total protein samples, transfer them to nitrocellulose membranes, cut them according to molecular weights, and then incubate them with different antibodies. We used Tanon Science and Technology Co., Ltd. (Shanghai, China) ECL-plus reagent to visualize the detected proteins. The intensity of the western blot band was quantified by ImageJ software. Each experiment was repeated at least three times independently, and biological replicates were performed in parallel each time. The average intensity of the bands from replicate samples was first normalized to an internal control (actin) and then normalized to a WT control, with the value set to 1. The final value was the average value from at least three independent experiments (n = 3/4).

Ferrozine Iron Assay
The serum, intestinal, cerebellum, and spinal cord iron contents were detected by the ferrozine iron assay. We took 50 µl of lysate or serum (we took double volume for lysis buffer as control) and added 11 µl concentrated HCl (11.6 M). All the tubes were then placed on a heating block for 20 min at 95 • C, centrifuged at the highest speed for 10 min, and then removed very gently from the centrifuge. Very carefully, 45 µl of supernatant was removed. To each tube, 18 µl of ascorbate (75 mM) was added; ascorbate acts as a reductant, moving Fe from 3 + state to the 2 + state. The tubes were vortex-quick spun and incubated for 2 min, and 18 µl ferrozine (10 mM) was added to each tube. Ascorbate acts as an oxidant, taking Fe from the 2 + state to the 3 + state. The tubes were incubated for another 2 min. Then, 36 µl of saturate ammonium acetate (NH 4 Ac) was added to each tube and the tubes were incubated for 2 min. Samples were read at 562 nm using a multifunctional fluorescent microplate reader (n = 3-5).

Enzymatic Activities
The activities of complexes I and II were measured according to the manufacturer's protocols, respectively. Both kits were purchased from Comin Biotechnology Co. (Suzhou, Jiangsu, China; n = 5).

Determination of ATP Content
The levels of ATP in tissues were detected by using an ATP determination kit (Beyotime Biotechnology). The reading is measured by GloMax TM 96-well plate luminescence detector (E6521, n = 5).

Lactic Acid Production
The tissue lysates were collected and assayed according to the lactic acid production detection kit (Nanjing Jiancheng Bioengineering Institute). The assay was detected by a multifunctional fluorescent microplate reader at 530 nm (n = 5).

Statistical Analysis
Student's t-test or one-way analysis of variance (ANOVA) was carried out using GraphPad Prism 8. The measurement was expressed as the mean ± SD; all the experiments were repeated more than three times independently. Significance was considered at P < 0.05.

Glycolysis-Related Gene Expression Is Enhanced and Oxidative
Phosphorylation-Related Gene Expression Is Weakened in the Central Nervous System Tissues of Irp2 −/− Mice Our previous in vitro study demonstrated that metabolism switch took place from OXPHOS to glycolysis in Irp2 −/− MEFs (Li et al., 2019). We wonder if it is the case in vivo. First, we detected the expression levels of iron-related proteins in the CNS tissues (cerebrum, cerebellum, brainstem, and spinal cord) of Irp2 −/− mice. Compared with that in WT mice, ferritin expression was increased while TfR1 was decreased in Irp2 −/− mice (Figures 1A,B), which was in line with a previous study (Jeong et al., 2011). Next, we detected the expression levels of Hif1α and glycolysis-related proteins and did not find the same elevation of Hif1α as we observed in Irp2 −/− MEF (Li et al., 2019). However, glycolysis-related proteins, including LdhA, Glut1, and Hk2, were upregulated, compared with that in WT mice (Figures 1C,D), although these genes are also thought to be the members of Hif1 regulon. Then, we detected the expression levels of Hif2α, the Fe-S biogenesis-related proteins (IscU and Fxn), and mitochondrial respiratory complex subunits (Ndufs1, SdhB, and Uqcrfs1). We found that the levels of Hif2α increased by around 50% in general; the levels of IscU and Fxn reduced to 50-70%; and the levels of subunits of complex I (Ndufs1), II (SdhB), and III (Uqcrfs1) also reduced to 50-60% in CNS tissues of Irp2 −/− mice, compared with WT mice (Figures 1E,F), suggesting a reduction of OXPHOS. Taken together, our results confirmed the biochemical changes in vivo related to energy metabolism, OXPHOS, and glycolysis, in the CNS tissues of Irp2 −/− mice.
Administration of PT-2385 Significantly Improves the Behavioral Performance and Anemia of Irp2 −/− Mice As presented above, we only observed the upregulation of Hif2α in central nervous tissues of Irp2 −/− mice, but both inhibitors, PX-478 (5 mg/kg body weight) for Hif1α and PT-2385 (0.4 mg/kg) for Hif2α, were still injected into Irp2 −/− mice intraperitoneally every other day for 1 month, individually. During the 1 month, the weight of the mice was monitored and was found to increase normally without difference compared with the vehicle treatment, indicating the safety of the drug and its dosage (Figure 2A). In terms of behavioral performance, the latency time of Irp2 −/− mice on the rotating rod and the hanging time on the wire mesh square were significantly shorter than those of the WT mice. However, it significantly recovered by 50% in Irp2 −/− mice after administration of PT-2385, while no efficacy was observed after PX-478 treatment, which is in agreement with Hif1α levels that did not change in the CNS tissues of mutant mice (Figures 2B,C). Through inhibition of Irp2 deficiency-induced Hif2α, we confirmed the effect of PT-2385 in Irp2 −/− mice, and we also treated the WT mice with PT-2385. The rotarod and hang tests did not show a significant difference between the vehicle and PT-2385 treatment (not shown). These data proved the critical role of Hif2α in CNS of Irp2 −/− mice. Since then, our work focused on Hif2α inhibition by PT-2385.
hemoglobin, and hematocrit, but not the MCV (Figures 2D-G). Then, we measured more parameters to evaluate the iron status, including the EPO mRNA in the kidney, serum EPO, and iron. Surprisingly, the iron status globally improved (Figures 2H-J).
To understand how the PT-2385 treatment corrected the iron insufficiency anemia of Irp2 −/− mice, we assessed the iron content in the intestine and liver, HAMP mRNA level in the liver, and serum IL-6 to determine whether the serum Frontiers in Neuroscience | www.frontiersin.org iron resulted from intestinal uptake or iron release from the liver. The results showed that HAMP mRNA levels in the liver, serum IL-6, and iron content in the intestine and liver all reduced to be comparable with that in WT after PT-2385 treatment (Figures 2K-N), suggesting that both ways, intestinal uptake and iron release from the liver, contributed to the elevation of the serum iron. This assumption was further supported by the increased NcoA4 (Figures 2O,P), which is involved in ferritinophagy for ferritin degradation to release iron (Mancias et al., 2014).

The Administration of PT-2385 Protects the Histological Morphology and Mitochondrial Ultrastructure in the Spinal Cord and Cerebellum of Irp2 −/− Mice
Previous studies have demonstrated that misregulation of iron metabolism from loss of Irp2 causes neuronal degeneration and mitochondrial dysmorphology (Jeong et al., 2011). The cerebellum and spinal cord are among the most severely affected tissues, so we examined them hereinafter. The H&E staining revealed that cerebellar Purkinje cells were full, intact, and tightly arranged in line in WT mice, while they were severely shrunk and/or missing in Irp2 −/− mice. However, PT-2385 treatment significantly suppressed the cerebellar Purkinje cells from degeneration or loss in Irp2 −/− mice ( Figure 3A). The results from the electron microscopy showed that the density of the mitochondria in mutant cerebellum is lightened, a phenomenon which was much more severe in the spinal cord of Irp2 −/− mice than that in WT mice. More affectedly, the morphology of the mitochondria in the spinal cord of Irp2 −/− mice became swollen, vacuolated, and internal-cristae damaged. Then again, PT-2385 treatment significantly alleviated the poor presentation, including the deformed mitochondria and Wallerian and segmental demyelination (Figures 3B-D), suggesting the beneficial effect of PT-2385 against motor neurodegeneration.

The Administration of PT-2385 Effectively
Rescues the Weakened Oxidative Phosphorylation in the Cerebellum and Spinal Cord of Irp2 −/− Mice According to the results in Figure 1 and previous studies (Li et al., 2018(Li et al., , 2019, Irp2 depletion induced mitochondrial dysfunction via reduction of Fe-S biogenesis. We evaluated the levels of Fxn and IscU and of complex subunits, Ndufs1, SdhB, and Uqcrfs1, in the cerebellum and spinal cord tissues of Irp2 −/− mice. Again, Hif2α was found increased, and the expression of mitochondrial proteins, including Fxn, IscU, and complex subunits, was decreased in Irp2 −/− mice compared with WT. However, PT-2385 treatment inhibited all these biochemical changes compared with Irp2 −/− mice (Figures 4A-D). In line with these results, the activities of mitochondrial complexes I and II were also significantly restored in both tissues (Figures 4E,F). The exception is the coupled ETC product ATP. In the cerebellum, ATP content was lower in Irp2 −/− mice than that in WT mice, and PT-2385 administration increased it, which is correlated with the ETC activities ( Figure 4G). However, in the spinal cord, ATP content increased significantly in Irp2 −/− mice (Figure 4H), which is consistent with the previous studies (Li et al., 2018(Li et al., , 2019 in MEFs, though the ETC-related proteins and enzymatic activities were lower in Irp2 −/− mice than those in WT mice (Figures 4C,D,F). The reason will be discussed further in section "Discussion." Very surprisingly, compared with Irp2 −/− mice, more ATP was produced after PT-2385 administration ( Figure 4H). Overall, the inhibition of Hif2α by PT-2385 rescues the weakened OXPHOS in Irp2 −/− mice to provide more ATP to fulfill the energy need.
The Administration of PT-2385 Attenuates Enhanced Glycolysis in the Cerebellum and Spinal Cord of Irp2 −/− Mice Though Hif1α was not found to be upregulated in vivo, we observed the enhanced glycolysis-related gene expression (Figure 1B), which is similar to the previous results in Irp2 −/− MEFs (Li et al., 2019). We still used the cerebellum and spinal cord tissues to check the effect of PT-2385 on the expression of LdhA, Glut1, Hk2, and Edn1, genes that are considered to be the members of Hif regulon. The results showed that the expression of these genes increased by 100-150% in the cerebellum and by 170-300% in the spinal cord of Irp2 −/− mice and reduced to the WT levels after PT-2385 treatment (Figures 5A,B), confirming the action of PT-2385 on Hif2α and the regulon relationship of Hif2α to the tested genes. The protein levels of glycolysis-related genes including LdhA, Glut1, and Hk2 were also significantly reduced after PT-2385 treatment (Figures 5C-F). Among them, Glut1 showed extraordinary change in the spinal cord. Accordingly, the lactic acid levels were significantly higher in both the cerebellum and spinal cord of Irp2 −/− mice than those in WT, and PT-2385 treatment significantly lowered the levels in both tissues of Irp2 −/− mice (Figures 5G,H). Comparing the two tissues, the cerebellum and spinal cord, the upregulated Hif-targeted genes seemingly responded stronger in the spinal cord than in the cerebellum since their mRNA levels were more elevated in the spinal cord ( Figure 5B) than in the cerebellum (Figure 5A), particularly for Hk2 and Edn1, after Irp2 depletion. The protein levels of LdhA and Glut1 in the spinal cord increased about twofold and fivefold, respectively (Figures 5E,F), vs. 1.3-and 2-fold in the cerebellum (Figures 5C,D) after Irp2 depletion. The results suggest that the spinal cord might suffer more from the upregulated Hif2α and active glycolysis.

DISCUSSION
In this study, we found that Irp2 ablation induced Hif2α upregulation associated with the metabolism switch from OXPHOS to glycolysis in vivo. The protective effect of Hif2 inhibitor PT-2385 suggested that Hif2α could be a potential target therapeutically in the treatment of IRP2 mutant-caused neurodegenerative syndrome.
Irp2 depletion causes the deficiency of cellular functional iron by decreasing transferrin receptor and increasing ferritin expression (Jeong et al., 2011), which compromise the heme biosynthesis and finally trigger microcytic anemia in mice (Cooperman et al., 2005;Galy et al., 2005) and in humans (Cooper et al., 2019;Costain et al., 2019). Both the functional iron deficiency and anemia-induced low oxygen tension/hypoxia could upregulate Hif2α expression in vivo. In our previous study, we did find that Irp2 ablation induced the increase of Hif1α and Hif2α in Irp2 −/− MEFs (Li et al., 2019). However, we did not find an increase of Hif1α, but of Hif2α, in Irp2 −/− mice. Inhibition of Hif2α by PT-2385 greatly alleviated the progress of neurodegeneration accompanied with the enhancement of mitochondrial Fe-S biogenesis and suppression of glycolytic pathway-related proteins. In agreement with previous studies in vitro (Li et al., 2018(Li et al., , 2019, the loss of Irp2 in mice Values represented the mean ± SD n = 3-5, and the ANOVA was used for statistics to evaluate the group differences. In panels (B,D-H), *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, Irp2 −/− DMSO vs. WT DMSO; # P < 0.05, ## P < 0.01, ### P < 0.001, Irp2 −/− PT-2385 vs. Irp2 −/− DMSO. downregulated the expression of Fxn and IscU, two important core components in Fe-S biogenesis machinery. Increase of Fxn and IscU in Irp2 −/− cells, either by overexpression of the two genes (Li et al., 2018) or by inhibition of Hif2α (Li et al., 2019;this study), augments the mitochondrial complex activities and ATP content. At the same time, inhibition of Hif2α weakened glycolysis to avoid the toxicity of high levels of lactic acid through suppression of LdhA, Glut1, and Hk2. Therefore, PT-2385 administration dramatically protected from the progressive neurodegeneration as modeled in Figure 6.
iron content in the liver of Irp2 mutant reduced and the increased ferritin dropped back to WT levels. More profoundly, we found that NcoA4 expression increased in the liver after Hif2 inhibition, suggesting the important role of NcoA4 for iron release from ferritin, very likely, through ferritinophagy.
Both Hif1α and Hif2α are regulated by oxygen and iron (Majmundar et al., 2010;Prabhakar and Semenza, 2012). Both the functional iron deficiency and induced anemia could upregulate Hif1α expression as well. However, we did not see similar responses of Hif1α with Hif2α in this study. We found that elevated Hif2α alone, not together with Hif1α, contributes to the FIGURE 6 | A working model to illustrate the potential target, Hif2α, in the therapeutic application of Irp2 loss-triggered neurodegenerative syndrome. Irp2 depletion causes the deficiency of cellular functional iron by decreased transferrin receptor and increased ferritin, which compromise the heme biosynthesis and finally trigger microcytic anemia in mice. Both the iron deficiency and anemia/hypoxia upregulate Hif2α expression in vivo. The elevated Hif2α upregulates glycolytic pathway-related genes to enhance glycolysis, while downregulates iron-sulfur cluster biogenesis-related genes, Fxn and IscU, to weaken mitochondrial function. After the inhibition of Hif2α by PT-2385, the energy metabolism is shifted from glycolysis to oxidative phosphorylation in Irp2 −/− mice and neurodegenerative symptoms are alleviated. switch of energy metabolism from OXPHOS to glycolysis. The rationale for Hif2α to regulate IscU has been demonstrated in which IscU is a member of the miR-210 regulon (Chan et al., 2009) and the promoter of miR-210 contains a HRE for Hif1/2 binding (Kulshreshtha et al., 2007). Therefore, downregulation of IscU may be explained to be through the miR-210-Hif2 axis in Irp2 −/− mice. The accompanied co-regulation of Fxn with IscU was often observed (Ferecatu et al., 2018;Li et al., 2018Li et al., , 2019, and the detailed regulation mechanism of Hif2α on Fxn remains to be explored. If mouse Fxn was regulated by Hif2α as reported (Oktay et al., 2007), it would be expected that Fxn expression should be increased. Indeed, Irp2 depletion induced the downregulation of Fxn, the expression of which was reversed after inhibition of Hif2α by PT-2385. The co-regulation of Fxn and IscU could be the key to response to PT-2385 treatment since the interaction of Fxn with IscU is important to facilitate Fe-S biogenesis (Fox et al., 2019;Gervason et al., 2019) to cure mitochondrial dysfunction. A similar work has been reported in which neuronal Hif1α and Hif2α deficiency improves neuronal survival and sensorimotor function in the early acute phase after ischemic stroke (Barteczek et al., 2017).
Though the 5 -UTR of Hif2α mRNA contains an IRE element for Irp1 binding (Sanchez et al., 2007;Barteczek et al., 2017), Irp1-IRE binding activity keeps constant in Irp2-depleted tissues (Zumbrennen-Bullough et al., 2014). Hif2α is, likely, upregulated by iron deficiency and anemia-induced hypoxia in Irp2 −/− mice as discussed above. Another regulation is also possible as XIAP and Ubc13-dependent Lys63-linked polyubiquitination promotes Hif1α nuclear retention leading to an increase in the expression of Hif1-responsive genes (Park et al., 2017). We do not know if a similar pathway exists for Hif2α stabilization. Interestingly, Irp1 −/− -induced elevated Hif2α upregulates EPO, causing the mice to develop polycythemia and pulmonary hypertension (Anderson et al., 2013;Ghosh et al., 2013;Wilkinson and Pantopoulos, 2013). However, the upregulated EPO expression in Irp2 −/− mice is probably invalid due to the iron limit in the bone marrow (Cooperman et al., 2005). Moreover, the increased Hif2α endowed glycolysis-related genes, such as LdhA, Glut1, and Hk2, in the cerebellum and spinal cord of Irp2 −/− mice. Therefore, inhibition of Hif2 by PT-2385 did not only increase the expression of Fxn and IscU to strengthen mitochondrial function but also decrease the expression of LdhA, Glut1, and Hk2 to weaken glycolysis to avoid the toxicity of high level of lactic acid. Although according to the astrocyteneuron lactate shuttle hypothesis, lactic acid can be used as an energy metabolism substrate for neurons, and Irp2 deficiencyinduced mitochondrial dysfunction is insufficient to meet the energy needs in neurons. Intriguingly, elevated lactic acid was observed more in the spinal cord than in the cerebellum, which is consistent with more severity in motor than in other behaviors (Jeong et al., 2011). The enhanced glycolysis could be the reason why the ATP production was slightly, but significantly, more in the spinal cord than in the cerebellum after Irp2 ablation.
The patient with the absence of IRP2 shows functional iron deficiency and mitochondrial dysfunction that emulate Irp2 −/− mice (Costain et al., 2019). The complete loss of IRP2 in patientderived lymphoblasts also induces the decreased expression of complex subunits and activities of mitochondrial complexes I and II (Costain et al., 2019), although the expression levels of Hifs, Fxn, IscU, and glycolytic pathway-related proteins are not detected. We expect that Hif2α is upregulated in the tissues of the CNS in patients as we observed in Irp2 −/− mice, and the inhibition of Hif2 may be a therapeutic option for the IRP2loss patients. On the other hand, genetical knockdown of Hif2α in Irp2 −/− mice needs to be carried out to firmly validate the effects of Hif2 inhibition. The mechanism of action of PT-2385 is thought to selectively antagonize Hif2 heterodimerization, and DNA-binding activity and has no effect on Hif1 and no significant off-target activity (Yu et al., 2017).

CONCLUSION
In summary, we have demonstrated that Irp2 ablation induces the expression of Hif2α, not Hif1α, in the tissues of CNS of Irp2 −/− mice, and Hif2 inhibitor PT-2385 dramatically protects the mice from the neurodegenerative disorder accompanied with the energy metabolism shift from glycolysis to OXPHOS, indicating that Hif2α is a potential target for neurodegenerative syndrome caused by loss of IRP2.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

ETHICS STATEMENT
The animal study was reviewed and approved by the Animal Experimentation Administration of Nanjing University.

AUTHOR CONTRIBUTIONS
JS, EH, TQ, and KL conceptualized the study. JS conducted the formal analysis and performed the visualization. TQ and KL acquired the funding. JS, LX, YXL, WD, YTL, HZ, and TX conducted the investigation. JS, LX, JC, TX, and KL performed the methodology. JS and KL were involved in project administration, wrote the original draft, and reviewed and edited the manuscript. YC gathered the resources. JS, YXL, and YTL performed the validation. All authors contributed to the article and approved the submitted version.

FUNDING
This study was supported by grants from the National Basic Research Program of China (Grant No. 2015CB856300) and by the National Natural Science Foundation of China (Grants Nos. 31871201 and 31571218).