Hmox1 Deficiency Sensitizes Mice to Peroxynitrite Formation and Diabetic Glomerular Microvascular Injuries

Objective Indirect evidence suggests a role for heme oxygenase-1 (HO-1) in limiting diabetic vasculopathy. The goal of this study was to assess the role of HO-1 in the development of microvascular lesions within glomeruli during diabetes mellitus using a mouse model with specific alteration of the Hmox1 gene. Approach and Results The effects of Hmox1 haploinsufficiency were studied as a means of assessing the intrinsic contribution of HO-1 in the development of renal microvascular lesions during diabetes. Renal function and histology were analyzed 10 weeks after diabetes induction with streptozotocin. Diabetic Hmox1+/− mice showed higher levels of albuminuria and blood urea compared to their wild-type diabetic littermates. More severe glomerular microvascular lesions were also observed in the diabetic Hmox1+/− mice. This was associated with a renal increase in the expression of the oxidative stress marker, nitrotyrosine. Conclusions Genetic Hmox1 partial deficiency is sufficient to sensitize mice to the development of diabetic glomerular microvascular lesions. HO-1 exerts antioxidant effects in the kidney during diabetes mellitus. These have protective effects on the development of glomerular endothelial injury.


Background
Diabetic nephropathy (DN) is a major microvascular complication of diabetes [1]. Progressive loss of specialized pericytes, the podocytes, and an irreversible decrease in the density and function of intrarenal microvessels correlates with renal function decline in DN [2,3]. Furthermore, podocyte injury is important in the development of DN in both type 1 and type 2 diabetes mellitus (DM) [4][5][6].
Heme oxygenase (HO) is the rate-limiting enzyme that degrades heme to produce carbon monoxide, iron, and biliverdin [7,8]. HO is also anti-inflammatory, antioxidant, and antiapoptotic [9]. HO-1, the inducible form of HO, can be upregulated by many factors including oxidative stress [9]. As several reports have suggested that increased oxidative stress contributes to the development and progression of vascular complications in DM [10,11], we focused on the putative role of HO-1 in the development of renal microvascular lesions during DN development.

Materials and Methods
2.1. Animals. Hmox1 +/− deficient mice were kindly provided by Yet et al. [12]. Mice were on pure C57BL/6J genetic background. As Hmox1 deficient (Hmox1 −/− ) mice show a high level of embryonic lethality [13], the effects of Hmox1 haploinsufficiency were studied as means of assessing the intrinsic contribution of HO-1 in the development of renal microvascular lesions during diabetes mellitus. Deletion of one Hmox1 allele (Hmox1 +/− ) reduced Hmox1 mRNA expression by 70% in bone marrow and whole blood leukocytes.

Induction of Diabetes Mellitus with Streptozotocin (STZ).
Hmox1 +/+ (wild-type) or Hmox1 +/− mice were made diabetic by streptozotocin (STZ) injection as previously described [14]. All mice were given free access to water and standard chow. Ten-to twelve-week-old males were rendered diabetic by STZ (Sigma-Aldrich, number S-0130) (100 mg/kg in sodium citrate buffer pH = 4.5) intraperitoneal injection on two consecutive days. Control mice received citrate buffer alone. Mice with fasting glycemia above 300 mg/dL were considered diabetic. Mice were killed 10 weeks after the induction of diabetes.
2.3. Assessment of Renal Function and Albuminuria. Urinary creatinine and blood urea nitrogen (BUN) concentrations were quantified spectrophotometrically by colorimetric methods. Urinary albumin excretion was measured with a specific ELISA assay (Cusabio, number CSB-E13878m).

Quantitative RT-PCR.
Total RNA extraction of mice renal cortex was performed by using Qiazol (Qiagen), according to manufacturer's recommendations. RNA was reverse transcripted by using the Quantitect Reverse Transcription kit (Qiagen) according to the manufacturer's protocol. The Maxima SYBR Green/Rox qPCR mix (Thermo Scientific Fermentas) was used to amplify cDNA for 40 cycles on an ABI PRISM thermocycler. The comparative method of relative quantification (2-DDCT) was used to calculate the expression level of each target gene, normalized to Ppia. The oligonucleotide sequences are available upon request. The data are presented as the fold change in gene expression.
2.5. Histopathology and Immunohistochemistry. Kidneys were immersed in 10% formalin and embedded in paraffin. Sections (4 μm thick) were stained with hematoxylin/ eosin or Masson's trichrome and processed for histopathology or immunohistochemistry. For immunohistochemistry, paraffin-embedded sections were stained with rabbit anti-nytrotyrosine primary antibody (Upstate, number 06284, 1 : 200), rabbit anti-CD3 antibody (DAKO, number A0452), and rat anti-F4/80 antibody (AbD Serotec, number MCA497R) then the staining was revealed with Histofine reagent (Nichirei Biosciences) and slides were counterstained with hematoxylin. For immunofluorescence, paraffin-embedded sections were stained with a guinea pig anti-nephrin antibody (Progen) and a rat anti-CD31 antibody (Dianova, number SZ31) then the stainings were revealed with a goat anti-guinea pig alexa568-coupled antibody and a donkey anti-rat alexa488-coupled antibody (Invitrogen) and nuclei were counterstained with DAPI. Photomicrographs were taken with an Axiophot Zeiss photomicroscope. Staining surface quantifications were performed with a macro designed on ImageJ.
2.6. Electron Microscopy. Small pieces of renal cortex were fixed in Trump's fixative (EMS, number 11750) and embedded in Araldite M (Sigma-Aldrich, number 10951). Ultrathin sections were counterstained with uranyl acetate and lead citrate and examined with a transmission electron microscope.

Statistical
Analyses. Data are expressed as mean ± SEM. Statistical analyses were calculated with GraphPad Prism (GraphPad software). Comparisons between 2 groups were performed with a 2-tailed Student t-test and between multiple groups were performed with one-way ANOVA followed by the Newman-Keuls test. A p value < 0.05 was considered statistically significant.
In keeping with increased oxidative stress, renal staining for nitrotyrosine was greater in Hmox1 +/− diabetic mice compared to WT (Figure 2(a)). Taken together, these results demonstrate that Hmox1 haploinsufficiency sensitizes mice to DN development and that HO-1 exerts protective renal effects, partly through its antioxidant effects.
Control WT and Hmox1 +/− mice showed no functional and histological differences at baseline or after 60 days. DN is strongly linked to loss of integrity of the glomerular filtration barrier, particularly podocyte foot process effacement and alterations in glomerular endothelial cells. Histological analysis revealed similar levels of glomerulosclerosis in both diabetic WT and Hmox1 +/− groups (Figure 2(a)) (32.4 ± 9.2% versus 37.3 ± 7.6% of glomeruli with mesangial expansion) and no difference in tubular injury and tubulointerstitial fibrosis. These results were confirmed by qPCR analysis showing similar levels of expression of mRNAs implicated in fibrosis such as Tgfb1, Col1a1, and Col3a1 and of the tubular injury marker Kim1 (Figure 2(b)).
As HO-1 also exerts anti-inflammatory effects, we explored renal inflammation in our model. We did not detect F4/80 + macrophages in the diabetic kidneys (not shown) and only few CD3 + lymphocytes, with no difference between WT and Hmox +/− diabetic mice (Figure 3), thus suggesting that HO-1 does not modulate inflammatory cells in that model. Correlating with such results, we observed similar mRNA expression of Ccl2, coding for the chemokine MCP-1 ( Figure 3). Other mRNAs like Trp53, coding for TNFα, Il10, and Il6, were undetectable in diabetic kidneys of both WT and Hmox +/− mice (not shown). Interestingly, IL-1β mRNA expression was increased in Hmox +/− diabetic mice when compared to WT diabetic mice (Figure 3).

Discussion
As only indirect evidence suggests that HO-1 activity may limit progression of DN, we sought to investigate the specific role of this enzyme in experimental DN using a model of genetic Hmox1 deficiency. Various levels of HO-1 expression and activity have been reported in tissues from diabetic animals [15] with no definitive conclusion.
Administration of HO-1 inducers (hemin, zinc protoporphyrin, cobalt protoporphyrin, and tert-butylhydroquinone) protects from renal damage, particularly glomerular lesions during DN development [16][17][18][19]. Meanwhile, the specificity of these compounds has been challenged [20,21]. Thus, direct and specific alteration of HO-1 such as the one achieved with ablation of the Hmox1 gene is needed. The importance of HO-1 in vascular protection is demonstrated by the spontaneous severe endothelial damage observed in human cases of HO-1 deficiency and in Hmox1 −/− mice [22,23]. These studies suggest that HO-1 might exert its protective effects through its antioxidant, antiapoptotic, and anti-inflammatory properties.
Recently, Zheng and colleagues demonstrated that NRF2 stimulation limits the progression of DN by repressing ECM production and P21 expression [24]. Here, we found that HO-1 does not modulate tubulointerstitial fibrosis nor glomerulosclerosis during diabetes. As NRF2 induces HO-1 but also other proteins such as NQO1 and GST-1P, the benefit induced by NRF2 stimulation on ECM production could be independent on HO-1. These observations suggest that NRF2 and HO-1 could exert renoprotection through different targets and/or pathways and that stimulating both NRF2 and HO-1 during diabetes could provide complementary beneficial effects on both kidney tubulointerstitial compartments and glomerular microvasculature.
We have shown that HO-1 protects the glomerular microcirculation during diabetes. HO-1 haploinsufficiency sensitized mice to the development of glomerular injuryespecially the glomerular endothelial cells-during DN. Although moderate improvement in insulin sensitivity has been shown in diabetic mice following HO-1 induction [25][26][27], we found no difference in fasting glucose levels in HO-1 deficient or control animals supporting direct vascular actions of HO-1. Interestingly, IL-1β mRNA expression was increased in Hmox +/− diabetic mice. (i) Few ancient studies showed that the main source of IL-1β in renal pathology is the glomerulus itself (probably podocytes) [28,29]. (ii) IL-1β, but also IL-18, a cytokine secreted by the inflammasome concomitantly to IL-1β, exerts several deleterious effects on endothelial cells such as increased vascular permeability for IL-1β [30][31][32] and induction of apoptosis and of adhesion molecule for IL-18 [33,34]. While the deleterious role of IL-1β in renal cells during diabetes is not clearly established, IL-18 deleterious role is clearly demonstrated [35]. Thus, diabetic glomerular microvascular injuries induced by HO-1 partial deficiency might be due, at least partly, to increased inflammasome activation. Supporting this hypothesis, a recent study showed that hemin inhibits NLRP3 inflammasome activation in a model of lung injury, through HO-1 activation [36]. Investigating the mechanisms for this should be the area of future research. Peroxynitrite is a potent oxidant produced by rapid interaction between superoxide anion and nitric oxide radicals and induces oxidative stress and cell death. Beckman and colleagues proposed that the production of peroxynitrite may play a role in endothelial damage in vivo [37]. Since then, robust evidence supports the possibility of the in vivo formation of this molecule [38]. Peroxynitrite contributes to endothelial damage through a mechanism involving ER stress [39].
We found markedly less nitrotyrosine staining in kidneys from wild-type diabetic mice than in Hmox1 +/− diabetic animals, suggesting that either superoxide anion production and/or nitric oxide radical formation was blunted by HO-1 activity. Nitrosylation of tyrosine moieties of cellular proteins in the kidneys of patients with DM has been demonstrated. Moreover, the absence of nitrotyrosine in other types of nephropathies indicates that this pathway is specific for DN [40]. HO-1 activity may both prevent peroxynitrite formations and may also display cytoprotective action through generation of carbon monoxide that protects cells from peroxynitrite-induced apoptotic death [41]. Interestingly, HO-1 induction by peroxynitrite is also a defense mechanism [42]. Although one cannot rule out potential nonenzymatic effects of HO-1 [43], a potent antioxidant role is likely.
HO-1 produces biliverdin/bilirubin and CO. As all these products have been found to exert anti-inflammatory, antiapoptotic, and antioxidant effects, we cannot, in the present study, determine if the observed changes in our Hmox +/− diabetic mice are specifically due to one of this product, if it is a combination of all of them, or if it is an independent protective effect of HO-1.

Conclusion
In summary, using for the first time a specific Hmox1 genetargeting approach, we highlight a role of HO-1 in protecting glomerular endothelial cells and podocytes against progressive DN. We show that mice with a heterozygous deletion of Hmox1 alleles are prone to accelerated diabetes-induced structural and ultrastructural damage of the glomerular capillary with accentuated microalbuminuria. Interestingly, we also found the first evidence that HO-1 drives protection from nitrative stress. Our results support the hypothesis that HO-1 is an important modulator of microvascular endothelial injury during DN and thus an interesting target for DN therapy.