Specific expression of heme oxygenase-1 by myeloid cells modulates renal ischemia-reperfusion injury

Renal ischemia-reperfusion injury (IRI) is a major risk factor for delayed graft function in renal transplantation. Compelling evidence exists that the stress-responsive enzyme, heme oxygenase-1 (HO-1) mediates protection against IRI. However, the role of myeloid HO-1 during IRI remains poorly characterized. Mice with myeloid-restricted deletion of HO-1 (HO-1M-KO), littermate (LT), and wild-type (WT) mice were subjected to renal IRI or sham procedures and sacrificed after 24 hours or 7 days. In comparison to LT, HO-1M-KO exhibited significant renal histological damage, pro-inflammatory responses and oxidative stress 24 hours after reperfusion. HO-1M-KO mice also displayed impaired tubular repair and increased renal fibrosis 7 days after IRI. In WT mice, HO-1 induction with hemin specifically upregulated HO-1 within the CD11b+ F4/80lo subset of the renal myeloid cells. Prior administration of hemin to renal IRI was associated with significant increase of the renal HO-1+ CD11b+ F4/80lo myeloid cells in comparison to control mice. In contrast, this hemin-mediated protection was abolished in HO-1M-KO mice. In conclusion, myeloid HO-1 appears as a critical protective pathway against renal IRI and could be an interesting therapeutic target in renal transplantation.

Generation of bone marrow-derived macrophages (BMDMs) and culture. Bone marrow cells were isolated from femurs and tibias of WT, LT and HO-1 M-KO mice, and cultured in Petri dishes (Greiner Bio-one). Bone marrow cells were incubated at 37 °C in a 5% CO 2 atmosphere. For generation of bone marrow-derived macrophages (BMDMs), bone marrow cells were grown in Dulbecco modified Eagle medium (DMEM) supplemented with L-Glutamine, 4.5 g/l glucose (Lonza), 10% heat-inactivated fetal calf serum (FCS), nonessential amino acids, sodium pyruvate, penicillin/streptomycin, β-mercaptoethanol, and 20% supernatant derived from macrophage colony-stimulating factor (M-CSF)-producing L929 cells. At day 3 of culture, 5 ml of complete medium containing 60% L929 cells supernatant was added to each dish. BMDMs were allowed to grow until day 7 after isolation. The purity of BMDMs in culture was over 97% as confirmed by a Fluorescence-activated cell sorting (FACS) analysis of CD11b and F4/80 expression. BMDMs were then collected and cultured in resting and stimulated conditions either with lipopolysaccharide (LPS) 100 ng/ml (Sigma-Aldrich) or hemin 15 μM (Sigma-Aldrich) by using complete medium supplemented with 2% L929 cells supernatant for 24 hours at 37 °C in a 5% CO 2 atmosphere.
Renal function assessment and histopathology. Renal function was evaluated by measuring creatinine in plasma samples as described 20 . Kidneys were fixed in 4% formaldehyde, embedded in paraffin, sectioned at 5-μm thickness, and stained with Periodic acid-Schiff-diastase and Masson Trichrome. Renal damage was assessed in a blinded manner by the Tubular Injury Score 21 . Briefly, the percentage of damaged tubules were assessed in the corticomedullary junction according to observation of necrosis, tubular dilatation, brush border loss, and cast deposition in 10 nonoverlapping fields for each sample (x400 magnification). A five-point scale was used: 0 = no damage, 1 = <10% of tubules injured, 2 = 10-25%, 3 = 25-50%, 4 = 50-75%, 5 = >75%. Interstitial fibrosis was quantified using NIH ImageJ software.
Endogenous peroxidase activity was first quenched by H 2 O 2 peroxidase blocking reagent (DakoCytomation). Macrophages and neutrophils were detected by using an anti-F4/80 antibody (1:50; eBioscience) and an anti-Ly-6G antibody (1:50; BD Biosciences), respectively; for 30 minutes at room temperature (RT). Sections were then washed and incubated with a secondary biotinylated goat anti-rabbit antibody (1:500; Jackson Immunoresearch) for 30 minutes at RT. Streptavidin-HRP was added and coloration was revealed using diaminobenzidine (DAB) with the substrate chromogen system from Dakocytomation. The number of F4/80 + and Ly-6G + cells was counted in 10 non-overlapping fields (x400 magnification). Nitrotyrosine staining was performed using an anti-nitrotyrosine antibody (1:400; Abcam) and the OptiView DAB IHC Detection Kit (Ventana Medical Systems) according to manufacturer's instructions. Nitrotyrosine intensity in the renal cortex was quantified using NIH ImageJ software.
Kidney homogenates preparation and Enzyme-Linked Immunosorbent Assay. Kidneys were diluted in lysis solution containing RIPA buffer (Sigma-Aldrich), protease inhibitor cocktail (Sigma-Aldrich), and phosphatase inhibitor cocktail PhosSTOP (Roche Life Science). Kidneys were homogenized with the MagNa Lyser (Roche Diagnostics). Homogenates were subsequently centrifuged at 12,000 rpm for 20 minutes at 4 °C, and the supernatants were stored at −80 °C until use. HO-1 and p62 enzyme-linked immunosorbent assay (ELISA) kits were purchased from Enzo Life Sciences. Interleukin-6 (IL-6), KC, and MCP-1 ELISA kits were purchased from R&D Systems. Renal cytokines were measured according to manufacturer's instructions. Values were corrected for the amount of renal proteins using Micro BCA Protein Assay Kit (Thermo Scientific).
Kidney tissue digestion protocol for flow cytometry. Kidneys were harvested and washed in RPMI 1640 medium supplemented with L-Glutamine and 25 mM Hepes (Lonza). Kidneys were dissociated in 5 ml digestion buffer containing collagenase IV (Worthington Biochemical) and DNase I (Roche) using the mouse lung dissociation program 1 on gentleMACS Dissociator (Miltenyi Biotec). Then, samples were incubated for 20 minutes at 37 °C and agitated every 15 minutes. Complete tissue dissociation was achieved using the mouse spleen dissociation program 4 on gentleMACS Dissociator (Miltenyi Biotec). Cell suspension was passed through a 70 μm cell strainer (BD Biosciences) and was adjusted to a 30 ml volume with Phosphate Buffer Saline (PBS)/ Bovine Serum Albumin (BSA) 0.5%/Ethylenediaminetetraacetic acid (EDTA) 2 mM. Samples were centrifuged at 300 g for 5 minutes at 4 °C. Cell pellets were resuspended in 1 ml PBS-BSA 0.5%-EDTA 2 mM and suitable for flow cytometry staining protocol.
Statistical analysis. All data are expressed as mean ± standard error of the mean (SEM). A two-tailed nonparametric Mann-Whitney U test was used; P-values < 0.05 were considered to represent statistical significance. All graphs and statistical analyses were performed using GraphPad Prism 6.00 for Mac OS X (GraphPad Software, La Jolla California USA, www.graphpad.com).

Renal myeloid cells upregulate HO-1 upon renal IRI. Based on CD11b and F4/80 expression by flow
Scientific RepoRts | 7: 197 | DOI:10.1038/s41598-017-00220-w Myeloid HO-1 protects kidney from IRI. HO-1 M-KO mice were used to determine the role of myeloid HO-1 during IRI. Myeloid-restricted deletion of HO-1 was confirmed in both resting and LPS-stimulated BMDMs ( Fig. 2A,B). We identified that HO-1 M-KO mice exhibit increased susceptibility to renal IRI as demonstrated by the significant loss of renal function and more severe tubular injuries 24 hours after renal IRI in comparison to LT mice ( Fig. 3A-C). Interestingly, after renal IRI, HO-1 expression within the whole kidney was identical between LT and HO-1 M-KO mice (Fig. 4A,B), suggesting that, despite representing a minor source of HO-1 from a quantitative point of view, myeloid HO-1 mediates significant renal protection during IRI.

Myeloid HO-1 mitigates both innate immune responses and oxidative stress upon renal IRI.
The impact of myeloid HO-1 on renal inflammation 24 hours after IRI was analyzed. HO-1 M-KO mice expressed increased levels of renal inflammatory mediators (i.e. IL-6, KC, and MCP-1) in comparison to LT (Fig. 5A). In addition, HO-1 M-KO mice also exhibited increased neutrophil and macrophage infiltrates surrounding necrotic tubular cells in the corticomedullary junction ( Fig. 5B-E). Finally, accumulation of tubular nitrotyrosine, a well-described marker of oxidative damage 22 , was exacerbated within the renal cortex of HO-1 M-KO mice in comparison to LT (Fig. 5F,G).
Myeloid HO-1 deficiency leads to impaired tubular repair and renal interstitial fibrosis. Tubular cell-cycle arrest represents a major event that promotes interstitial fibrosis and finally chronic kidney disease (CKD) upon IRI 5,23 . In order to investigate a possible role of myeloid HO-1 in the prevention of CKD, we analyzed the expression of the cell-cycle inhibitor p53 and its target p21. As compared to LT, HO-1 M-KO mice exhibited a significant increase in both p53 and p21 mRNA expression 4 hours and 24 hours after IRI, respectively (Fig. 6A,B). In addition, we noted a significant accumulation of renal p62 24 hours post IRI in the HO-1 M-KO animals as compared to LT, suggesting potential impaired autophagy, a phenomenon known to enhance interstitial fibrosis upon tubular stress 24 (Fig. 6C). Finally, 7 days after reperfusion, HO-1 M-KO mice exhibited more renal interstitial fibrosis in comparison to LT (Fig. 6D,E). Hemin-mediated protection against IRI depends on HO-1 upregulation by renal CD11b + F4/80 lo myeloid cells. To determine whether the protection against IRI conferred by hemin pretreatment involved HO-1 + myeloid cells, we treated HO-1 M-KO and WT mice either with hemin (5 mg/kg) or saline. Twenty-four hours after hemin administration, WT mice exhibited strong HO-1 expression within the renal P 2 cells, while neither the P 1 nor P 3 subsets increased their HO-1 expression (Fig. 7A,B). Interestingly, hemin also induced HO-1 expression within the same subset of splenic myeloid cells (i.e. P 2 ) (Supplemental Fig. S2). In contrast, no HO-1 induction in renal myeloid cells from HO-1 M-KO mice was noted (Fig. 7C,D). Twenty-four hours after renal IRI, the proportion of HO-1 + P 2 cells increased in the hemin-vs. the saline-treated WT mice while it remained unchanged in HO-1 M-KO mice in both conditions (Fig. 8A-C). Concurrently, hemin-treated WT mice displayed renal resistance against IRI (Fig. 8D, Supplemental Fig. S3A-C) while hemin did not preserve renal function in HO-1 M-KO mice (Fig. 8E). These results strongly suggest that CD11b + F4/80 lo cells (i.e. P 2 ) are the main protective myeloid source of HO-1 within the kidney upon IRI. We also observed that, in comparison to WT mice, LT did not benefit from the protection against renal IRI when they were treated with hemin 24 hours before IRI (Supplemental Fig. S4A). This could be explained by a hypomorphic HO-1 allele in LT that was due to the modification of the Hmox1 allele with loxP sites (Supplemental Fig. S4B-D). However, the HO-1 hypomorphism was not sufficient to provoke particular susceptibility to renal IRI in LT as attested by similar creatinine levels compared to WT mice (Supplemental Fig. S5).

Discussion
We have shown that myeloid HO-1 mediates protection against renal IRI and that its preemptive induction by hemin confers significant resistance against renal IRI. These findings contrast with the common belief that only epithelial (i.e. tubular cells) and endothelial cells are the critical source of HO-1 during renal IRI. This hypothesis was mainly supported by the intense susceptibility of fully HO-1-deficient to renal IRI 10,25,26 . However, Ferenbach DA et al. already demonstrated that genetically modified or hemin-induced HO-1 + macrophages, a subset of myeloid cells, mediate protection against renal IRI 27,28 . Our results strongly support this observation. In addition to these previous studies, we have observed that, in response to IRI, naturally occurring myeloid HO-1 may already modulate the severity of AKI. Indeed, even if we showed that the global expression of HO-1 in the whole kidney was not affected, the absence of myeloid HO-1 was critical in the outcome of renal IRI. More recently, Hull et al. showed that HO-1 is a critical regulator of the trafficking of myeloid cells in AKI. However they did not report a difference between LT and HO-1 M-KO mice in term of renal function (i.e. plasma creatinine) or in tubular damage 1 day after reperfusion 9 . In contrast, our results pointed out the myeloid HO-1 as a critical regulator of the earliest phases of IRI (i.e. lower plasma creatinine, tubular damage, and renal inflammation) that may mitigate the risk of severe AKI upon IRI. By inference, as severe IRI in a renal transplant also increases its immunogenicity, we may postulate that, by downsizing the early inflammatory response, the induction of myeloid HO-1 could be a regulatory mechanism decreasing transplant alloreactivity.
A cell-cycle arrest at the G2/M phase is associated with maladaptive repair and subsequent fibrosis in renal IRI 5,23 . In our experiments, HO-1 M-KO mice exhibited impaired renal repair upon IRI as suggested by the upregulation of cell-cycle regulatory proteins (i.e. p53, p21), potential disruption of autophagy (i.e. p62 accumulation) and early interstitial fibrosis, a central marker of CKD. The roles of cell-cycle inhibitors p53/p21 in the pathogenesis of AKI are complex and remain a matter of debate. For instance, if p53 expression by leukocytes has  been identified as protective against AKI, its induction in tubular cells was associated with worsened AKI and subsequent interstitial fibrosis [29][30][31] . In line with a previous study 32 , the cell-cycle inhibitor p21 has been shown protective in AKI by mediating a cell-cycle arrest in G1 phase and therefore allowing repair of DNA-damage 33 . However, the authors observed that p21 was not involved in protection against renal fibrosis 33 . Herein, we focus on the link between p21 and the progression of AKI to fibrosis. Interestingly, even if p21 expression may confer early resistance to AKI, p21 expression by tubular cells also reflects their cellular senescence that is associated with limited regenerative ability of tubular cells and interstitial fibrosis following AKI suggesting an additional role for p21 34,35 . Indeed, these different data are consistent with the recent concept that p21 has different effects during AKI or CKD/fibrosis progression 36 .
Altogether, these observations sustain the hypothesis that the upregulation of both p53 and p21 upon IRI in HO-1 M-KO kidneys may encourage fibrotic processes in HO-1 M-KO mice. Interestingly, fully HO-1-deficient mice exhibit more severe interstitial fibrosis upon AKI 37 and renal fibrosis was also reported in HO-1 M-KO mice 1week after IRI 9 . Further, our results suggest a link between myeloid HO-1 deficiency and renal fibrosis because of maladaptive repair.
Even if the role of myeloid HO-1 as a critical modulator of the late fibrotic processes remains a matter of debate, it has been shown that AKI could lead to severe CKD. Indeed, this is supported by the recent key concept of cell-cycle arrest and maladaptive repair of tubular cells that is induced by AKI itself 5,23 . Briefly, a clear correlation between the severity of the renal lesions and the risk of maladaptive repair has been established in several experimental models, leading to interstitial fibrosis and subsequent CKD 5,23 . Even if the loss of myeloid HO-1 expression before IRI was responsible for the observed kidney injury, a specific role for HO-1 during the later phases of renal repair after IRI is not excluded. For instance, HO-1 may decrease TGF-β expression at the tubular level, thereby preventing renal fibrosis 38 . HO-1 was also shown to modulate tubular autophagy, a major event involved in cell repair after insult 39 . Of interest, we have observed that HO-1 M-KO mice exhibited p62 accumulation upon renal IRI which may be seen as a surrogate marker of impaired autophagy 40 .
Specific deletion of myeloid HO-1 was associated with sustained oxidative damage within the corticomedullary junction upon IRI. It is well known that HO-1 limits the heme-and hydrogen peroxide-mediated oxidative stress that heavily damages DNA and proteins and leads to cell death 41 . However, the fact that HO-1 M-KO mice exhibit intense tubular damage due to excessive oxidative stress strongly suggests potential cross-talk between the tubular environment and the HO-1 + myeloid cells. Accordingly, the release of by-products of heme degradation by HO-1 + myeloid cells (i.e. carbon-monoxide, IL-10) may inhibit the apoptosis of the surrounding tubular cells, thereby promoting their survival 42 . Interestingly, HO-1 M-KO mice exhibited sustained p53 expression in the whole kidney upon IRI. As tubular cells are major components of kidney extract, we may hypothesize that myeloid HO-1 may regulate some of their important functions upon IRI, decreasing the risk of both tubular apoptosis and cell-cycle arrest.
It has been shown that HO-1 expression was associated with CD11b + F4/80 lo macrophages that exhibit regulatory properties (i.e. "M2" macrophages) 43 . These M2 macrophages are potential modulators of inflammatory responses induced by renal IRI, thereby promoting renal repair after insult 16 . In this condition, HO-1 + myeloid cells may promote the development of a microenvironment dominated by regulatory macrophages that efficiently dampen early inflammatory responses and delay the onset of interstitial fibrosis 44 . However, the molecular mechanism of macrophage polarization mediated by HO-1 remains to be elucidated 44   suggesting a positive-feed-forward loop between these two anti-inflammatory factors, favoring the acquisition of an IL-10-producing, M2-like phenotype. The macrophage-derived molecules that promote kidney repair are Results are expressed as the mean ± SEM, ★ p < 0.05; ★★ p < 0.01; ★★★ p < 0.001. n = 7-9 for IRI groups and n = 5 for sham groups. poorly known 47 . However, both the macrophage-secreted Wnt7b and chitinase-like protein BRP-39 have been shown to promote kidney regeneration after IRI 48,49 . Accordingly, we may hypothesize that our HO-1 + CD11b + F4/80 lo anti-inflammatory macrophages protect the renal parenchyma upon IRI by secretion of some mediators which may interact with renal epithelial cells.
Even if we did not demonstrate it in our experiments, the intense inflammatory response that we observed in HO-1 M-KO kidneys upon IRI might be explained through a phenotypic polarization toward "M1" macrophages. Indeed, in absence of HO-1, a lack of M2 macrophages was observed with an excess of inflammatory "M1" macrophages 50 . Then, these macrophages secrete pro-inflammatory mediators that amplify intrarenal inflammation and injury through interaction with kidney resident cells 16,47 . Therefore, we assume that HO-1, in addition to its protective role against oxidative stress at the cellular level, also plays a critical role in the regulation of early innate immune responses induced by renal IRI.
The origin of HO-1 + myeloid cells that protect the injured kidney remains largely unknown. Theoretically, renoprotection might be provided by both resident and infiltrating HO-1 + myeloid cells in the kidney. In line with previous report 28 , we showed that, even in absence of IRI, hemin upregulated the HO-1 expression within CD11b + F4/80 lo myeloid cells in the kidney. This observation suggests that tissue-resident myeloid cells might be involved in the earliest phase of renal IRI. The recruitment of splenic macrophages that protect against IRI was already observed 51,52 . We also noted that hemin induced HO-1 within spleen CD11b + F4/80 lo myeloid cells suggesting that extra-renal HO-1 + myeloid cells might represent a reservoir that can be recruited in the injured kidney after renal IRI. Indeed, one day after reperfusion, we observed a higher proportion of CD11b + F4/80 lo myeloid cells within the kidney of hemin-treated WT mice suggesting that these protective cells could have been recruited from extra-renal sites such as the spleen.
We identified that the intensity of myeloid HO-1 expression was an important determinant of efficient hemin-mediated protection against renal IRI. Indeed, we found that the floxation of the Hmox1 allele with loxP sites in LT significantly decreased both native and induced HO-1 expression in comparison to WT mice. This "hypomorphism" induced by allele floxation has been described in others Cre-LoxP knockout systems 53,54 . Interestingly, even if the HO-1 hypomorphism was insufficient to cause particular susceptibility to renal IRI in LT compared to WT mice, it leaded to the loss of the hemin-mediated protection in LT. Therefore, we may postulate that myeloid HO-1 expression induced by hemin requires a critical level to be effective and that this minimal threshold in HO-1 expression was not reached in LT. By inference, the length polymorphism of guanosine thymidine dinucleotide (GT) n repeats in the promoter region of Hmox1 was associated with lower HO-1 activity 55 , mimicking the effect of a hypomorphic allele. Interestingly, longer (GT) n repeats in the Hmox1 promoter were associated with the occurrence of CKD and decreased renal function after cardiac surgery or kidney transplantation in humans 11,[56][57][58] . Also, it has been largely reported that hemin preconditioning significantly mitigates IRI-induced AKI 14,15 . A recent study showed that hemin safely induces HO-1 in deceased donor renal  WT and HO-1 M-KO mice were treated with hemin (5 mg/kg, grey bars) or saline (white bars) 24 hours before renal IRI. At day 1 of reperfusion, mice were sacrificed. (A) Representative dot plots of saline-and hemintreated WT IRI mice showing the myeloid cell populations in the kidney according to the expression of CD11b and F4/80 surface markers (i.e., CD11b − F4/80 + (P 1 ), CD11b + F4/80 lo (P 2 ), and CD11b hi F4/80 − (P 3 )). (B) Quantification of renal myeloid cells upon IRI, presented as a proportion of the renal CD45 + cells extracted from saline-and hemin-treated WT IRI mice. Results are expressed as the mean ± SEM, ★ p < 0.05. n = 5-6 per group in WT mice. (C) Quantification of renal myeloid cells upon IRI, presented as a proportion of the renal CD45 + cells extracted from saline-and hemin-treated HO-1 M-KO IRI mice. Results are expressed as the mean ± SEM. n = 3-5 per group in HO-1 M-KO mice. Plasma creatinine levels in saline-and hemin-treated WT (D) and HO-1 M-KO (E) mice subjected to sham surgery or 24 hours of reperfusion after renal IRI. Results are expressed as the mean ± SEM, ★★★ p < 0.001. In WT mice, n = 16-20 for IRI groups and n = 6 for sham groups. In HO-1 M-KO mice, n = 9 for IRI groups and n = 5 for sham groups.
Interestingly, myeloid HO-1 has been involved in the suppression of alloreactivity [60][61][62] . This suggests that myeloid HO-1 could have additional regulatory properties that may contribute to both better survival and tolerance of the allograft after IRI. However, further studies are needed to confirm the benefit of the HO-1 induction in renal transplantation.
In conclusion, our results strongly support the importance of HO-1 + myeloid cells for conferring resistance against renal IRI and are in line with other recent published works 9,27,28 . Myeloid HO-1 appears as a critical modulator of the interstitial inflammation and subsequent fibrosis that both follow renal IRI. Pharmacological HO-1 induction by hemin before renal IRI could be an efficient preventive strategy for limiting kidney damage in many situations such as renal transplantation.