Obesity-induced kidney injury is attenuated by amelioration of aberrant PHD2 activation in proximal tubules

The involvement of tissue ischemia in obesity-induced kidney injury remains to be elucidated. Compared with low fat diet (LFD)-mice, high fat diet (HFD)-fed mice became obese with tubular enlargement, glomerulomegaly and peritubular capillary rarefaction, and exhibited both tubular and glomerular damages. In HFD-fed mice, despite the increase in renal pimonidazole-positive areas, the expressions of the hypoxia-responsive genes such as Prolyl-hydroxylase PHD2, a dominant oxygen sensor, and VEGFA were unchanged indicating impaired hypoxic response. Tamoxifen inducible proximal tubules (PT)-specific Phd2 knockout (Phd2-cKO) mice and their littermate control mice (Control) were created and fed HFD or LFD. Control mice on HFD (Control HFD) exhibited renal damages and renal ischemia with impaired hypoxic response compared with those on LFD. After tamoxifen treatment, HFD-fed knockout mice (Phd2-cKO HFD) had increased peritubular capillaries and the increased expressions of hypoxia responsive genes compared to Control HFD mice. Phd2-cKO HFD also exhibited the mitigation of tubular damages, albuminuria and glomerulomegaly. In human PT cells, the increased expressions of hypoxia-inducible genes in hypoxic condition were attenuated by free fatty acids. Thus, aberrant hypoxic responses due to dysfunction of PHD2 caused both glomerular and tubular damages in HFD-induced obese mice. Phd2-inactivation provides a novel strategy against obesity-induced kidney injury.


Results
Renal morphological changes and the hypoxic state in HFD-fed mice. In mice fed HFD, body weight increased but blood pressure did not change (Table 1). Though fasting blood glucose did not differ between the two groups, serum levels of triglycerides, FFA, and insulin in fasted mice were higher in HFD-fed mice (Table 1). HFD-induced obese mice exhibited renal damage as indicated by albuminuria and excretion of the proximal tubular injury markers, neutrophil gelatinase-associated lipocalin (NGAL) and cystatin C (Fig. 1A) though serum creatinine levels were not changed (Table 1). These biochemical data were consistent with renal damage in obesity 5 . In histology, kidneys from HFD-fed mice exhibited marked mesangial hypercellularity and enlarged glomerular size. In addition, PT cellular size was increased in mice fed HFD, compared with in those fed LFD (Fig. 1B). These hypertrophic changes suggested a spatial sparse of peritubular capillary (PTC) beds. CD34positive cell counts revealed that PTC density in the kidney was significantly decreased in HFD compared with LFD-fed mice (Fig. 1C). Because of PT enlargement and PTC rarefaction, we hypothesized that HFD-fed mice had PT hypoxia and performed pimonidazole staining to examine this alteration. Compared with LFD-fed mice, HFD-fed mice showed significantly increased pimonidazole-positive areas (Fig. 1D).
Hypoxic response in kidneys of HFD-fed mice. To confirm molecular changes responsive to hypoxia, expressions of the downstream hypoxia-responsive molecules, including PHD2 and VEGF, were measured. Though the pimonidazole-positive areas in HFD-fed mice were significantly larger than in LFD-fed mice, expressions of either PHD2 ( Fig. 2A) or genes downstream to PHD, including, Vegfa (VEGF-A), Pgk1 (PGK1), Slc2a1 (Glut-1), and Ldha (LDHA) (Fig. 2B) were not different between mice fed two diets. These data indicated that, in obese mice, renal tissues showed abnormal hypoxic responses. Generation of tamoxifen-inducible PT-specific PHD2 knockout mice. To explore the role of an impaired hypoxic response in obese mice, we created Tam-inducible PT-specific Phd2 knockout mice (Phd2-cKO) by crossing Phd2-floxed mice (Phd2 F/F ) 16 with knock-in mice harboring Tam-inducible Cre-recombinase gene driven by a PT-specific N-myc downstream regulated gene 1 (Ndrg1) promoter 17 (Fig. 3A). With Tam treatment, PHD2 expression was significantly decreased in the proximal tubules of Phd2-cKO mice on either diet group, confirming that Phd2-inactivation by the Cre-loxP system was effective in Phd2-cKO mice (Fig. 3B, columns 4 and 8). Tam treatment had no effect on PHD2 expression in Control mice (Fig. 3B, comparing columns 1 and 3 and columns 5 and 7). Immunostaining revealed that PHD2 expression was markedly decreased in Tam-treated Phd2-cKO mice (Fig. 3C). Expressions of downstream genes, Vegfa (VEGF-A), Pgk1 (PGK1), Slc2a1 (Glut-1), and Ldha (LDHA), were increased in Phd2-cKO mice, suggesting that hypoxia responsible gene downstream of PHD2 was subsequently induced by inactivating PHD2 (Fig. 3D). Immunostaining also showed the increase in VEGF-A  expression in PT of Phd2-cKO mice (Fig. 3E). These data indicated that the hypoxic responses are functionally activated in Phd2-cKO mice on HFD as in those on LFD.

Amelioration of the hypoxic condition by restoration of PTC in Tam-inducible PT-specific
Phd2-cKO mice. Because it is hypothesized that hypoxic response does not work properly, thereby contributing to development of renal pathological lesions in obese mice, we examined effects of PHD2-suppression on renal damage using the Tam-inducible PT-specific Phd2-cKO mice. HFD increased body weights of both Control and Phd2-cKO mice, with no weight difference between Phd2-cKO and Control mice on either diet group (Fig. 4A). HFD increased serum FFA similarly in Phd2-cKO and Control mice (Fig. 4B). Albuminuria, the marker for glomerular damages, as well as the markers for proximal tubular injury including urinary excretion of NGAL and cystatin C were higher in Control HFD mice compared with Control LFD mice. These HFD-dependent changes were attenuated in Phd2-cKO mice with Tam treatment (Fig. 4C). To explore the mechanism whereby Phd2 gene silencing ameliorated renal damages in HFD-induced obese mice, we examined the tissue ischemic state. The number of CD34-positive cells was decreased in Control HFD mice compared with Control LFD mice. This capillary loss was restored in Phd2-cKO HFD mice with Tam treatment (Fig. 4D). Consistent with these results, the pimonidazole-positive area was increased in Control HFD mice. However, in Phd2-cKO HFD mice, pimonidazole-positive area was smaller, nearly the same as in Control LFD mice (Fig. 4E). In addition to these results, electron microscope showed PTC injuries in Control HFD mice. In HFD mice, endothelial cells of PTC was enlarged, basement membrane was thickening, slit structures collapsed, and lumen of PTC was narrowing, which were ameliorated in Phd2-cKO HFD mice (Fig. 4F). This indicated that, in Phd2-cKO mice, ischemia was ameliorated in the HFD-induced obese state. The histological abnormalities in Control HFD mice, including glomerulomegaly and enlarged PT cellular size, were completely prevented in Phd2-cKO HFD mice with Tam treatment, suggesting that the early intervention could suppress the progression to the renal pathological changes of both tubular and glomerular lesions in the HFD-induced obese state (Fig. 4G).

Dysregulation of hypoxic response induced by FFA.
To further examine the mechanism for the impaired hypoxic response observed in HFD-fed obese mice, a human PT cell line, HK-2 cells, were cultured in either normoxic or hypoxic (1% O 2 ) condition to mimic kidney environment in HFD-induced obese mice. The expressions of hypoxia-responsible genes downstream of PHD2-HIF pathway including glucose transporter 1 (GLUT1) and VEGFA were increased under hypoxia (Fig. 5A,B, lanes 1 and 2). We examined the effects of insulin or FFA, both of which were systemically elevated in HFD-induced obesity (Table 1). Insulin did not affect the expression of SLC2A1 (GLUT1. Fig. 5A, lanes 3 and 4) or VEGFA (Fig. 5B, lanes 3 and 4). However, FFA dampened the upregulation of these genes by hypoxia (Fig. 5A,B, lanes 5 and 6).

Discussion
Obesity and metabolic syndrome are complicated with renal damages in which multiple factors, including systemic hypertension, dyslipidemia and abnormal glucose metabolism are involved. We previously reported obesity-induced histological changes in the kidney, including glomerular hypercellularity, macrophage infiltration and vacuolization in PT 5 . In the present study, we demonstrated that hypoxic tissue insults are present in obese kidney, which is caused by PTC rarefaction through the impaired hypoxic response. We also demonstrated that the restoration of oxygen supply by induced PT-specific PHD2 deletion mitigated not only in tubular damages but also glomerular damages. This restoration also ameliorated renal histological changes including both tubular and glomerular hypertrophy. Our data provide the evidence for a crucial role of an improper hypoxic response in PT to hypoxia in the development of renal lesions of both tubules and glomeruli in the obese condition. It is also demonstrated how the aberrant PT function propagate to interstitial vascular disorganization or glomerular damages to establish the obesity-induced renal damages. In response to hypoxic stress, cellular adaptive processes are regulated by HIF-1 and HIF-2 in various tissues including kidney 18 . Under hypoxia, PHD2 is inactivated since its activity is dependent on oxygen molecules, and the hydroxylation of HIF-1α is inhibited, causing HIF-1α to accumulate and inducing genes downstream to HIF, including Vegfa, Pgk1, Slc2a1, and Ldha. However, in the obese mice, even under hypoxic conditions, the expression of these hypoxia-responsible genes were not induced (Fig. 2), suggesting the impaired hypoxic response. Using an inducible gene-engineered mouse model, we have now shown that these lesions were almost fully eliminated by PHD2 downregulation in PT, suggesting that obesity-induced renal damage is attributed mainly to the lack of an inactivation of PHD2 in PT. This reactivation of HIF-VEGF pathway by deletion of the Phd2 gene in PT increased number of PTCs and subsequently prevented hypoxic renal injuries. It was recently reported that tubulo-vascular crosstalk involving VEGF is essential to maintain PTC networks in the kidney 19,20 . Our results support this concept, providing new evidence for intercellular communication between PT cells and PTC cells and for the pathological relevance of this communication in obesity-induced kidney disease. Moreover, we further implicated FFA as one of the candidate factors for impairing the hypoxic response of PT cells although the detailed mechanism has not been elucidated. It was demonstrated that the activation of nuclear receptor, peroxisome proliferator-activated receptors γ (PPARγ ) induces PHD up-regulation in adipocytes 21 . Since certain kinds of FFA and their derivatives including palmitate and oleate that were used in the present in vitro study have been shown to activate PPARγ 22,23 . HFD-induced increase in serum FFA concentration might maintain the PHD2 activity and inhibit the inactivation of PHD2 in spite of hypoxic condition. Our data provide mechanistic clues that dyslipidemia is closely associated with CKD progression 24,25 and controlling hyperlipidemia could augment HIF-VEGF signaling and correct an impaired hypoxic response.
The cellular enlargement of PT, described as histological findings in obesity 4,5 , would be expected to contribute to the tissue hypoxic condition through the rarefaction of PTCs. However, the mechanism of cellular enlargement of PT is not clear. In previous reports, a variety of factors, including insulin or tubular cell autophagy, could also induce tubular cell enlargement. Insulin is a well-known inducer of cellular hypertrophy by activating mammalian-target of rapamycin (mTOR)/S6 kinase pathway 26,27 . It has been also reported that autophagy-deficient renal tubular cells in Atg6-deficient mice accumulated deformed mitochondria and cytoplasmic inclusions, leading to cellular hypertrophy 28 and that, in obesity-induced renal damages, autophagy insufficiency in PT cells exacerbates proteinuria-induced tubulointerstitial lesions 29,30 . In our study, PT enlargement was alleviated in HFD-fed Phd2-cKO mice, supporting the idea that tubular hypertrophy in obesity-induced kidney injury may be caused by insufficient hypoxic response in PT. Cellular hypertrophy is often associated with cell growth arrest 31,32 and hypoxia is known to be one environmental factor that halts the cell cycle and cell proliferation 33,34 . Hypoxia was reported to upregulate the cyclin dependent kinase inhibitors p21 Cip1 and p27 Kip1 , which block cell cycle progression 35 . Hypoxia may also induce cell cycle arrest by inhibiting c-Myc transcriptional activity 36 . Another molecular mechanism for the relationship between hypoxic condition and cellular hypertrophy is the AMP-activated protein kinase (AMPK) signaling pathway that is activated by increased ratio cellular AMP/ATP ratio in hypoxic condition 37 . AMPK activation attenuated cellular hypertrophy 38,39 by inhibiting protein synthesis through the mTOR pathway 40 . Using models for renal disease, Li et al. showed that HIF-1α and AMPK were linked at a molecular level during the response to hypoxic stress in the pathophysiology of CKD. AMPK activation was decreased in the subtotal nephrectomy model and was markedly restored by HIF-1α activation 41 . In our study, we speculate that HIF-1α did not function normally and AMPK may not have been fully restored by HIF-1α to facilitate cellular adaptation to hypoxia, resulting in cellular hypertrophy.
In the present study, we observed in detail PTC injuries in EM. Of note, in HFD mice, endothelial cells of PTC was enlarged, basement membrane was thickening, slit structures collapsed, and lumen of PTC was narrowing, which are new findings in obesity induced renal injuries and firstly reported. In addition to PTC rarefaction, these PTC injuries may contribute to hypoxic tissue insults in obese kidney. Although previous studies of obesity induced renal histological changes in human focused primarily on the glomerular area and obesity-related glomerulopathy has been characterized by glomerulomegaly with or without focal segmental glomerulosclerosis 4 , whether new findings in the present study are observed also in human needs to be evaluated in the future study.
One salient finding in our study is that albuminuria and glomerular hypertrophy in histology were also remitted by improving the hypoxic condition through an intervention affecting PHD2. We speculate that the activation of HIF-VEGF pathway by PHD2 deletion in PT might lead to the restoration of post-glomerular PTC network, resulting in the reduction of glomerular afterload and the amelioration of glomerular hypertension and hypertrophy. These alterations cumulated in the reduction of albuminuria in obese mice. Alternatively, this result is related to a novel pathological link between tubular lesion and glomerular lesion in diabetic nephropathy we recently reported 42 . In this study, the decreased expression of NAD + -dependent deacetylase Sirt1 in PT initiates diabetic albuminuria through the downregulation of Sirt1 and the upregulation of claudin-1, a tight junction component in podocytes. The present study provides evidence for this renal tubule-glomeruli communication and through this communication, the recovery from tubular damages leads to the recovery from glomerular damages and albuminuria. Therefore, the manipulation targeting PHD2-dependent hypoxic response has a promise for inhibiting the progression of glomerular damage in obese-induced kidney disease.
Several limitations of the present study merit comments. The present study demonstrated the results with short diet duration and there may be different findings with longer duration. In addition, it is difficult to show the activity of PHD2 in vivo and also to show the expression of HIF-1α by lacking of high quality antibodies. Instead, we have demonstrated that hypoxic state is detected by pimonidazole staining and shown the expression of mRNAs or proteins of downstream genes, but not HIF-1α . Although it has been described that the increase in glomerular basement membrane thickness may be seen after the onset of diabetes, we have firstly reported that basement membrane of tubules is already thickened in obesity before the onset of diabetes. However, it is needed to make sure of whether the same histological changes are observed in obesity in human. Further studies are required to clarify these limitations. In conclusion, this is the first study to demonstrate that a hypoxic condition due to inadequate hypoxic response is a pathophysiological effect associated with obesity-induced renal injury. Normalization of hypoxic response by the downregulation of PHD2 in PT ameliorated hypoxic damage not only of PT lesions also in glomerular lesions. An early intervention targeting PHD2, specifically in the proximal area, may represent a novel therapeutic strategy against the progression of obesity-induced kidney injury.
Animal 2: Inducible PT-specific PHD2 knockout mice. We generated PT-specific conditional Phd2 knockout (Phd2-cKO) mice by crossing the Phd2 F/F mice from Dr. Yoji Andrew Minamishima (Keio University) 16 and the Ndrg1-Cre ERT2/+ mice with a PT-specific Ndrg1 gene promoter from Prof. Motoko Yanagita (Kyoto University) 17 . To obtain Phd2-cKO mice, Phd2 F/F mice were crossed with Ndrg1-Cre mice to generate Phd2 +/F ; Ndrg1-Cre mice. Phd2 +/F ; Ndrg1-Cre mice were then crossed with Phd2 +/F mice to generate Phd2 F/F ; Ndrg1-Cre (Phd2-cKO) mice and Phd2 +/+ ; Ndrg1-Cre (Control) as littermate controls. Genomic DNA was isolated from tail biopsies at 4 week of age using a DNeasy kit (Qiagen Inc., Valencia, CA, USA) and genomic DNA samples were screened by polymerase chain reaction using the transgene-specific oligonucleotide primers shown in Table 2. The primers used to amplify the Phd2 F/F allele are also shown in Table 2 Horseradish peroxidase-conjugated anti-rabbit IgG antibodies (Dako, Glostrup, Denmark) were used as secondary antibodies. Staining was visualized with a diaminobenzidine (DAB) chromogen, followed by counterstaining with hematoxylin. The extent of histochemical and immunohistochemical staining were quantified using computer-assisted image analysis 46 . Sections incubated with normal rabbit serum instead of the primary antiserum served as negative controls. Fifty glomeruli and fifty proximal tubular cells were counted in one kidney section from each mouse. The areas of the glomeruli and PT were measured by image analysis of high magnification photographs. Immunostaining was assessed at 100x, 200x or 400x magnification using 20 randomly selected fields for each mouse. These morphological evaluations were conducted in a blinded manner by two independent observers.
Transmission electron microscopic (TEM) analysis. TEM analysis was performed as previously reported 47 . Briefly, the kidneys were dissected out without any perfusion and were fixed with 2.5% glutaraldehyde in 100 mM phosphate buffer (pH 7.4) for 12 hours at 4 °C. After the 2 hours of post-fixation with 1% osmium  Hypoxic exposure and stimulation by insulin or FFA. Early passage (passages 3-4) HK-2 cells were grown to 80% confluence and made quiescent by serum starvation for 24 h. Cells were then exposed to hypoxia (1% O 2 , 5% CO 2 , 94% nitrogen gas) for 24 h in a hypoxia workstation (Hirasawa Works, Tokyo, Japan) or remained in the CO 2 incubator used for routine culture (normoxia). An oxygen sensor was used to ensure that the oxygen concentration inside the workstation was maintained at 1% throughout the experiments (MC-8G-S, Iijima electronics corporation, Aichi, Japan). Cells were further supplemented with 1 μ M human insulin (Sigma Aldrich) or 0.3 mM FFA. The FFA stock solution consisted of 6.35 mM sodium palmitate (Sigma Aldrich) and 12.7 mM sodium oleate (Sigma Aldrich) in FFA-free bovine serum albumin (BSA) (1.8 mM) as previously described 48,49 . Cells were divided into six groups: (1) cells without insulin or FFA in normoxia (normal O 2 concentration of 20%), (2) cells without insulin or FFA in hypoxia, (3) cells with insulin in normoxia, (4) cells with insulin in hypoxia, (5) cells with FFA in normoxia and (6) cells with FFA in hypoxia. After incubation for 24 h, cells were collected using TRIzol reagent (Invitrogen, Carlsbad, CA) to perform real-time RT-PCR.
Real-time RT-PCR. Equal amount (1 μ g) of total RNA from each sample was subjected to reverse transcription in a 20 μ L reaction mixture containing random primers and Superscript II enzyme (Invitrogen). Real-time PCR was performed using an ABI Step One Plus sequence detector (Applied Biosystems, Foster City, CA) 50 . mRNA levels were normalized to those of RPLP0 (36B4) genes. The sequences of primers used are shown in Table 3.
Statistical Analysis. Data are expressed as means ± standard error of mean. One-way analysis of variance was used to determine significant differences among groups. In the overall analysis of variance, the Kruskal-Wallis test for multiple comparisons was used to assess individual group differences. P < 0.05 was considered statistically significant.
Ethics Statement. This study was performed in accordance with the Institutional Guidelines on Animal Experimentation at Keio University. All methods were carried out in accordance with the guidelines for animal experiments of the Ministry of Education, Culture, Sports, Science and Technology, Japan. The experimental protocols were approved by the Animal Care and Experimentation Committee in Keio University (ID; 09119-(3)). All surgery was performed under sodium pentobarbital anesthesia by intraperitoneal injection, and all efforts were made to minimize suffering. At the end of the experiments, the mice were euthanized by intraperitoneal injection of sodium pentobarbital.