MAFB in Macrophages Regulates Prostaglandin E₂–Mediated Lipid Mediator Class Switch through ALOX15 in Ischemic Acute Kidney Injury

Ischemic acute kidney injury (AKI) is characterized by a rapid decline in renal function, occurring within a few hours to several days due to reduced blood flow into the kidneys, due to conditions such as sepsis or hypotension. The incidence of AKI has been increasing steadily annually. AKI is linked to severe outcomes: 10–15% of AKI patients succumb to the associated acute illness, 3–15% advance to end-stage renal disease, and 20–50% develop chronic kidney disease (CKD) (1, 2). Consequently, AKI significantly decreases the quality of life of patients. However, the only available treatment strategy for AKI focuses on restoring adequate blood flow to the kidneys. Moreover, no targeted therapies have been designed to suppress the underlying inflammatory processes specifically. Therefore, an in-depth understanding of the molecular mechanisms of AKI is essential for advancing AKI management.

Notably, the number of macrophages in the kidneys increases within 24 h of AKI onset. During the initial stage of AKI, macrophages promote tubular injury, whereas in later stages, they promote tubular proliferation, aiding the normal repair (3–5). Consequently, macrophages play a critical role in both the injury and recovery phases of AKI. Therefore, elucidating the specific functions of macrophages in promoting and resolving inflammation in AKI is crucial for understanding its pathology; however, the mechanisms remain unclear.

MAFB (V-maf musculoaponeurotic fibrosarcoma oncogene homolog B), a transcription factor belonging to the large Maf family, forms a dimer via a leucine zipper and binds to the Maf recognition element (MARE) in the DNA promoter region, thereby regulating the expression of downstream target genes (6, 7). MAFB plays an essential role in the differentiation and functional maintenance of various tissues and cells, including kidney glomerulus podocytes, pancreatic cells, and the parathyroid gland (8–10). In hematopoietic cell lines, MAFB is expressed specifically on monocytes and macrophages but not any other mature myeloid or lymphoid lineage cells (8, 11). Several studies have reported the function of MAFB in macrophages. MAFB promotes atherosclerosis development by regulating the expression of apoptotic inhibitors of macrophages (AIMs), in turn preventing foam cell apoptosis (12). MAFB in macrophages also plays a vital role in efferocytosis by regulating the expression of complement component C1q, as C1q binds to apoptotic cells to promote the recognition of apoptotic cells by macrophages (11). In addition, MAFB in macrophages promotes clearance of damage-associated molecular patterns (DAMPs) by regulating the expression of macrophage scavenger receptor 1 (MSR1) in ischemic stroke, and recently it has been reported that MAFB in macrophages promotes skin wound healing and cold-induced neuronal density in brown adipose tissue (13–15). Therefore, MAFB in macrophages is assumed to be crucial for homeostasis and inflammation suppression. Recently, it has been reported that MAFB expression is induced via the cyclooxygenase-2 (COX-2)/PGE₂/PGE2 receptor 4 (EP4) pathway in macrophages infiltrating the ischemic AKI-induced kidney. The induced MAFB promotes recovery from ischemic AKI by polarizing renal myeloid cells into an anti-inflammatory and proresolving phenotype (16). However, the precise molecular mechanisms, such as target genes of MAFB, remain unknown.

Arachidonate 15-lypoxygenase (ALOX15) is a lipoxygenase that catalyzes the oxidation of essential fatty acids. ALOX15 in mice has been reported to play a role as a suppressor of inflammation and cancer (17). ALOX15 expressed in macrophages produces precursors of inflammation-suppressing specialized proresolving mediators (SPMs) by oxidizing essential fatty acids at the site of inflammation (17–19). SPMs have anti-inflammatory and proresolving functions in immune cells, and thereby they play a crucial role in orchestrating the resolution of tissue inflammation (20). PGE₂, known as a proinflammatory lipid mediator (LM), exhibits diverse biological activities depending on the local inflammatory microenvironment and the receptor subtype on which it acts. Therefore, PGE₂ also has immunosuppressive functions. For example, PGE₂ inhibits inflammatory chemokine production by inhibiting NF-κB in a c-AMP–dependent manner via the EP4 receptor (21). In addition, PGE₂ plays an important role in LM class switch from proinflammatory LMs to SPMs by inducing ALOX15 expression (22).

Mouse ALOX15 shares 74–81% amino acid identity with human ALOX15 and is considered an ortholog of human ALOX15. Usually, orthologs of enzymes perform similar functions between different species. Therefore, mouse ALOX15 is considered a functional homolog of human ALOX15, albeit with slightly different reaction specificity in the oxidation of arachidonic acid (23).

In the current study, the prognosis of ischemic AKI was observed to be worse in macrophage-specific Mafb-deficient mice. The finding was similar in the mice-transplanted fetal liver cells of mice carrying the human-type MAFB p.Leu239Pro mutant, which is a loss-of-function mutation of MAFB. Notably, ALOX15 expression decreased at both the mRNA and protein levels in Mafb-deficient macrophages infiltrating the kidney 24 h after ischemic AKI induction. Furthermore, our in vitro assay showed that MAFB regulates Alox15 expression under the COX-2/PGE₂/EP4 pathway in macrophages. The results suggest that MAFB plays a crucial role for PGE₂-mediated LM class switch by regulating the expression of ALOX15 under acute inflammatory conditions.

Macrophage lineage-specific Mafb-deficient mice (C57BL/6J) were generated by mating Mafb^f/f mice with LysM-Cre mice (The Jackson Laboratory, Bar Harbor, ME) under the control of the endogenous Lys2 promoter. Mafb^f/f and LysM-Cre mice were generated in our laboratory as described previously (24). Male 8- to 12-wk-old mice were used in the experiments.

Mafb GFP knock-in mice (Mafb heterozygous; Mafb^gfp/−) with a C57BL/6J background were also used in this study (8).

The mice were maintained under specific pathogen-free conditions at the Laboratory Animal Resource Center of the University of Tsukuba. All experiments complied with relevant Japanese and institutional laws and guidelines, and were approved by the University of Tsukuba Animal Ethics Committee (23-041).

Renal ischemia-reperfusion injury (IRI) was performed as previously described (25). Briefly, a retroperitoneal incision was made under isoflurane anesthesia. The bilateral renal arteriovenous tissues were clamped for 60 (Fig. 1A) and 30 min (for other than Fig. 1A) using micro clamps (18055-02; Fine Science Tools). After the clamp was removed, reperfusion of both kidneys was visually confirmed. All experiments were performed on a 37°C warm plate to maintain body temperature.

Blood samples were collected from the cheek veins of mice. Serum blood urea nitrogen (BUN) and creatinine concentrations were measured using DRI-CHEM (NX600; Fujifilm).

Kidney tissues were fixed in 4% paraformaldehyde and embedded in paraffin. Alternatively, for frozen sections, kidney tissues were fixed with 4% paraformaldehyde.

Periodic acid–Schiff (PAS) staining was performed on 3-µm sections of paraffin-embedded kidney blocks.

For the staining of MAFB, CD68, and F4/80, 5-µm sections of paraffin-embedded kidney blocks were incubated with a 1:100 dilution of rabbit monoclonal anti-MAFB Ab (E309X; Cell Signaling Technology), a 1:100 dilution of rabbit polyclonal anti-CD68 Ab (ab125212; Abcam), and a 1:200 dilution of rat monoclonal anti-F4/80 Ab (CL8940AP; Cedarlane). For staining of ALOX15 and GFP, 5-µm frozen sections were incubated with a 1:500 dilution of rabbit monoclonal anti–15-lipoxygenase 1 Ab (ab244205; Abcam) and a 1:500 dilution of chicken polyclonal anti-GFP Ab (Aves Lab). To stain MAFB and CD68, sections were incubated with Histofine simple mouse stain (rabbit; 414341; Nichirei). For the fluorescent staining of F4/80, ALOX15, and GFP, the sections were incubated with a 1:500 dilution of Alexa Fluor 647–conjugated donkey anti-rat IgG (ab150155; Abcam), a 1:500 dilution of Alexa Fluor 594–conjugated donkey anti-rabbit IgG (A21207; Invitrogen), and a 1:500 dilution of Alexa Fluor 488–conjugated goat anti-chicken IgG (A11039; Invitrogen). Nuclear staining was performed with Hoechst 33342 solution (346-07951; Dojindo), and sections were mounted using Fluoromount (K024; Diagnostic BioSystems). Specimens were analyzed using a Biolevo all-in-one fluorescence microscope (BZ-X810; Keyence).

The PAS-stained debris area at the corticomedullary junction was quantified using ImageJ software, and the percentage areas per whole section are presented. Six different fields (original magnification, ×100) were analyzed for each slide and the average value is displayed in the figure.

The acute tubular necrosis (ATN) score was based on the percentage of tubules that displayed cell necrosis, loss of brush border, cast formation, and tubule dilation. The scoring system ranged as follows: 0, none; 1, 10%; 2, 11–25%; 3, 26–45%; 4, 46–75%; and 5, >76% (26). Fifteen different fields (original magnification, ×200) were reviewed for each slide by a renal pathologist in a blinded manner.

Kidneys were digested with enzyme mixture (970 µl of HBSS, 20 µl of collagenase A [10103578001; Roche], 10 µl of DNase I). After digestion, the cells were incubated with anti-CD11b (clone M1/70; BioLegend), anti-Ly6C (clone HK1.4; BioLegend), and anti-Ly6G (clone RB6-8C5; GeneTex) Abs. Dead cells were stained by DAPI solution (340-07971; Cellstain). Flow cytometry analysis was performed using CytoFLEX (Beckman Coulter), and the results were analyzed using FlowJo software (Tree Star).

Kidneys were digested as described above and stained with Zombie Violet (BioLegend) at 100× dilution for 15 min at room temperature in the dark, followed by surface Abs for 25 min at 4°C in the dark. The cells were fixed with 1× fixation regent (10× fixation regent; 2.75 ml of formaldehyde in 47.25 ml of distilled water containing 0.32 g of NaH₂PO₄ and 0.2 g of Na₂HPO₄) for 30 min at room temperature in the dark. The cells were then permeabilized by washing with 0.1% saponin at 1000 rpm for 4 min at 25°C. The cells were incubated with anti–IL-6 Ab (BioLegend) for 30 min at room temperature in the dark, followed by washing with 0.1% saponin at 300 × g for 4 min at 25°C. Flow cytometry analysis was performed using CytoFLEX (Beckman Coulter), and the results were analyzed using FlowJo software (Tree Star). For intracellular staining of IL-6, mice were injected with 250 µg of brefeldin A (BioLegend) 6 h before the experiment, and all reagents for kidney digestion and staining contained brefeldin A 1× injection.

Kidneys were digested using the method described above and incubated with an anti-CD11b Ab conjugated to allophycocyanin. These cells were then incubated with anti-allophycocyanin microbeads (130-090-855; Miltenyi Biotec), applied to the mass spectrometry (MS) columns (130-042-201; Miltenyi Biotec), and passed through the magnetic column. The MS column was then removed from the magnet, and 1 ml of MACS buffer via the plunger was used for flushing out the microbead-labeled cells from the column, initially retained by the magnetic field. The microbead-labeled cells were incubated with anti-CD11b, anti-Ly6C, and anti-Ly6G Abs. Ly6G^lowLy6C⁺CD11b⁺cells were sorted into Isogen (311-02501; Nippon Gene) using a MoFlo XDP cell sorter (Beckman Coulter).

Total RNA was extracted from Ly6G^lowLy6C⁺CD11b⁺cells using TRIzol reagent (15596026; Thermo Fisher Scientific). The RNA sequencing (RNA-seq) library was prepared using the NEBNext Ultra directional RNA library prep kit (New England Biolabs, Ipswich, MA) after rRNA depletion (NEBNext rRNA depletion kit; New England Biolabs). Paired-end (2 × 36 bases) sequencing was performed using NextSeq 500 (Illumina). The FASTQ files were imported into the CLC Genomics Workbench (version 10.1.1; Qiagen). The sequence reads were mapped to the mouse reference genome (mm10). Gene expression levels were calculated as total read counts and normalized using the quantile method. Genes with zero counts in all samples were excluded, and differential expression was analyzed using the empirical analysis of DGE tool (edgeR test) within the CLC Main Workbench (version 7.7.3; Qiagen). Differentially expressed genes were extracted using false discovery rate–corrected p < 0.05.

Total RNA was extracted using Isogen (311-02501; Nippon Gene). cDNA was synthesized using the QuantiTect reverse transcription kit (205313; Qiagen). The mRNA levels of the mouse genes of interest were measured using SYBR Green PCR master mix (QPS-201; Takara Bio). mRNA levels were normalized to mouse Hprt mRNA levels. The primer sequences for quantitative RT-PCR (qRT-PCR) were as follows: Hprt forward (5′-CAAACTTTGCTTTCCCTGGT-3′) and reverse (5′-CAAGGGCATATCCAACAACA-3′); Mafb forward (5′-TGAATTTGCTGGCACTGCTG-3′) and reverse (5′-AAGCACCATGCGGTTCATACA-3′); Alox15 forward (5′-CGTGGTTGAAGACTCTCAAGG-3′) and reverse (5′-CGAAATCGCTGGTCTACAGG-3′); C1qa forward (5′-GGATGGGGCTCCAGGAAATC-3′) and reverse (5′-CTGATATTGCCTGGATTGCC-3′); C1qb forward (5′-TGGCTCTGATGGCCAACCAG-3′) and reverse (5′-GACTTTCTGTGTAGCCCCGT-3′); C1qc forward (5′-AGGACGGGCATGATGGACTC-3′) and reverse (5′-TGAATACCGACTGGTCTTC-3′); Msr1 forward (5′-TGGAGGAGAGAATCGAAAGCA-3′) and reverse (5′-CTGGACTGACGAAATCAAGGAA-3′); and Ptgs2 forward (5′-GTATCAGAACCGCATTGCCTCTGA-3′) and reverse (5′-CGGCTTCCAGTATTGAGGAGAACAGAT-3′).

Liquid chromatography–tandem MS (LC-MS/MS)–based targeted lipidomics was performed as described previously (27). Briefly, frozen kidneys in ice-cold methanol (Wako) were homogenized by Zirconia beads of 3.0 and 5.0 mmφ (TOMY) using a Precellys 24 homogenizer (Bertin Technologies) and kept at −30°C overnight. The methanolic supernatant was applied to MonoSpin C18-AX cartridges (GL Science, Tokyo, Japan) to extract lipid metabolites in the presence of deuterated internal standards: 1 ng of arachidonic acid–d8, 15-hydroxyeicosatetraenoic acid–d8, LTB₄-d4, LTD₄-d5, and PGE₂-d4 (Cayman Chemical, Ann Arbor, MI) for monitoring recovery rates during sample preparation. For LC-MS/MS analysis, a triple-quadrupole linear ion-trap mass spectrometer (QTRAP 4500, AB Sciex, Foster City, CA) equipped with an Acquity UPLC BEH C18 column (1.0 × 150 mm, 1.7-µm particle size; Waters, Milford, MA) was used. MS/MS analysis was performed in the negative ion mode, and metabolites were identified and quantified by multiple reaction monitoring. Calibration curves between 1 and 1000 pg and the LC retention times for each compound were established using synthetic standards. Raw data were analyzed using MultiQuant software (Sciex). PD1, PDX, and their stereoisomers were not separatable under the LC-MS/MS setting above.

The mouse macrophage RAW264.7 cell line was cultured in DMEM supplemented with 10% FBS and 1% l-glutamine.

To collect peritoneal macrophages, a thioglycolate medium was injected into the abdominal cavities of the mice. Four days after the injection, PBS was injected into the abdominal cavities of the mice. The PBS was then collected and centrifuged at 1000 rpm for 4 min at 4°C. The cell pellet was plated in macrophage serum-free medium (1265074; Life Technologies) supplemented with 1% penicillin/streptomycin for 3 h to remove neutrophils (nonadherent cells).

Macrophages were stimulated with LPS (500 ng/ml) for 3 d and PGE₂ (Cayman Chemical) (1000 ng/ml) for 2 d.

PGE₂ was from Cayman Chemical, and L161982 (EP4 antagonist) was from Tocris Bioscience for in vitro study.

The promoter region of the Alox15 gene was examined using the UCSC Genome Browser (http://genome.ucsc.edu/index.html). Fragments of 200 or 500 bp of the Alox15 promoter were extracted from C57BL/6 mouse genomic DNA and cloned into a pGL4.10 luciferase vector lacking a promoter (Promega). Site-specific mutations were introduced into the putative MARE sequence at site 2 using a QuickChange site-directed mutagenesis kit (Agilent Technologies) with the primer sequences as follows: Alox15 mut2-MARE forward (5′-GAGGCCCCGCCCCTTCGCGAGTCTAGTTTAGGGTTGAGC-3′) and reverse (5′-GCTCAACCCTAAACTAGACTCGCGAAGGGGCGGGGCCTC-3′). In addition, a mutated version of the putative MARE sequence at site 1 was synthesized by GenScript.

To investigate the potential activation of the Alox15 promoter by MAFB, vectors containing the Alox15 promoter were cotransfected with a Mafb expression vector into RAW264.7 cells. A pRL-TK vector (Promega) was also introduced into the RAW264.7 cells for normalization purposes. Transfection was carried out using FuGENE HD transfection reagent (Promega), followed by a 48-h cell culture. Luciferase activity was assessed using a Dual-Luciferase reporter assay system (Promega). Briefly, after transfection for 48 h, the cells were lysed, and 20 µl of cell lysate was transferred to 100 µl of luciferase assay reagent II. The samples were then measured using a luminometer. To halt the reaction, 100 µl of Stop&Glo reagent was added, and readings were taken again. A pRL-TK vector expressing Renilla reniformis luciferase was used to normalize the transfection efficiency.

Data are presented as mean ± SEM. Statistical significance for the difference in mean derived from several groups was determined using a Welch t test.

RNA-seq data are deposited in the DNA Databank of Japan (https://ddbj.nig.ac.jp/resource/sra-submission/DRA017447).

To elucidate the molecular mechanisms by which MAFB in myeloid cells alleviates AKI in detail, Mafb^f/f::LysM-Cre mice, in which the Mafb gene was deleted specifically in the myeloid cell lineage, were used, as previously reported (11). AKI was induced in Mafb^f/f and Mafb^f/f::LysM-Cre mice using the IRI model and the prognosis was compared. Our results showed that with a 60-min clamp, Mafb^f/f::LysM-Cre mice began to die on day 2, and the survival rate was ∼17% on day 6, although none of the Mafb^f/f mice died after IRI (Fig. 1A). In addition, serum BUN and creatinine levels were higher in Mafb^f/f::LysM-Cre mice (Supplemental Fig. 1A). Then, we shortened the clamping time to 30 min and conducted all subsequent experiments under these conditions. Although both Mafb^f/f and Mafb^f/f::LysM-Cre mice did not die by a 30-min clamp, the levels of serum BUN and creatinine were significantly higher in Mafb^f/f::LysM-Cre mice (Fig. 1B). Because Mafb^f/f::LysM-Cre mice started to die on day 2 after IRI with the 60-min clamp, and serum BUN and creatinine levels were significantly higher in Mafb^f/f::LysM-Cre mice on day 1 after IRI in the 30-min clamp, MAFB in macrophages may play a significant role in suppressing inflammation from a very early stage. PAS staining of kidney sections revealed significantly higher amounts of debris in the kidneys of Mafb^f/f::LysM-Cre mice, as well as impaired tissue repair, as indicated by the ATN score at 7 d post-IRI (Fig. 1C–E). It has been reported that the corticomedullary junction of the kidney is mainly injured in ischemic AKI (28). Consistently, we observed debris and injury mainly in the corticomedullary junction (Fig. 1C). These results indicated a worse prognosis for AKI in Mafb^f/f::LysM-Cre mice than in Mafb^f/f mice.

Whether similar results could be obtained in other MAFB-dysfunctional mouse models was tested. Previously, we have reported that human MAFB p.Leu239Pro mutant mice exhibit a phenotype similar to Mafb-deficient mice (29). To generate mice with this mutation only in myeloid cells, we transplanted fetal liver cells from MAFB p.Leu239Pro mutant (Mafb^mt/mt) mice into 10 Gy–irradiated CD45.1 wild-type (WT) recipient mice (Fig. 1F). Chimerism in recipient mice was confirmed by FACS analysis, and mice with >95% chimerism were used for further experiments (Supplemental Fig. 1B). Ischemic AKI was induced in fetal liver cell–transplanted mice. Serum BUN and creatinine levels were significantly higher in Mafb^mt/mt cell–transplanted mice than those in WT cell–transplanted mice (Fig. 1G). PAS staining of kidney sections revealed impaired tissue repair in Mafb^mt/mt cell–transplanted mice, as indicated by the ATN score at 7 d post-IRI (Fig. 1H, 1I).

These results suggest that the loss of MAFB in myeloid cells leads to impaired kidney tissue repair after IRI, highlighting the pivotal role of MAFB in alleviating AKI in macrophages. After the onset of ischemic AKI, macrophages infiltrate damaged sites in the kidney (3). Therefore, we next examined MAFB expression in the macrophages, using Mafb^gfp/+ mice in which the Mafb gene locus is heterozygously replaced by the gfp gene (8). Mafb^gfp/+ mice were subjected to IRI and dissected 5, 12, or 24 h after IRI, and whole kidneys were extracted for FACS analysis. The gating strategy is presented in Supplemental Fig. 1C. We defined Ly6G^lowLy6C⁺CD11b⁺cells as macrophages, and consistent with a previous study (3), their proportion among all live cells was significantly increased until at least 24 h after IRI (Fig. 2A, 2B). To examine whether these macrophages express MAFB, we measured the percentage of GFP⁺ macrophages among all macrophages. Before IRI, the percentage of GFP⁺ macrophages was ∼50%. However, following IRI, the percentage of GFP⁺ macrophages significantly increased at 5 h and was maintained at a high level for at least 24 h post-IRI (Fig. 2A, 2C). These results indicated that the number of MAFB⁺ macrophages increased in the kidney after AKI induction.

Next, to directly examine the expression of MAFB protein, we performed immunostaining of WT kidneys with an anti-MAFB Ab. In the kidney without IRI, there were only slight MAFB⁺ signals; however, 5 h after IRI, the MAFB⁺ signals increased, mainly at the corticomedullary junction of the kidney (Fig. 2D). The results suggest that the MAFB⁺ macrophages are recruited to the damaged site in the kidney after AKI onset.

Several research groups have suggested that myeloid cell infiltration affects the prognosis of ischemic AKI significantly (30, 31). Therefore, we examined the degree of myeloid cell infiltration into the kidneys of Mafb^f/f and Mafb^f/f::LysM-Cre mice 24 h after IRI using FACS analysis. CD11b⁺ cells were isolated from whole kidneys using MACS, and FACS analysis was performed. The absolute numbers of CD11b⁺ cells, neutrophils (Ly6G^highCD11b⁺ cells), and macrophages (Ly6C⁺Ly6G^lowCD11b⁺ cells) were significantly higher in Mafb^f/f::LysM-Cre mice (Fig. 2E, 2F), suggesting severe inflammation in Mafb^f/f::LysM-Cre mice.

Proper migration to the injury site is essential for macrophages to perform their functions (3). To examine whether MAFB is involved in the proper migration of macrophages to the site of injury, immunostaining for macrophage markers CD68 and F4/80 was performed on the kidneys of Mafb^f/f and Mafb^f/f::LysM-Cre mice at 24 h post-IRI. According to PAS staining results, the corticomedullary junction of the kidney was primarily damaged by IRI (Fig. 1C, 1H). Consistently, CD68⁺ and F4/80⁺ cells were observed at the damaged corticomedullary junctions in the kidneys of both Mafb^f/f and Mafb^f/f::LysM-Cre mice (Supplemental Fig. 1E, 1F). Our results suggest that the infiltration ability of macrophages is not impaired by Mafb deficiency.

Mafb deficiency does not affect macrophage infiltrative capacity, suggesting that MAFB is involved in other functions of macrophages. Clements et al. (32) reported that macrophages infiltrating the kidney after IRI are divided into three populations according to the expression level of Ly6C: Ly6C^hiCD11b⁺ proinflammatory population, Ly6C^intCD11b⁺ wound healing population, and Ly6C^lowCD11b⁺ profibrotic population. We examined whether the deficiency of Mafb altered the composition of these macrophage populations. We performed FACS analysis of kidneys from Mafb^f/f and Mafb^f/f::LysM-Cre mice after IRI. The gating strategy is shown in Supplemental Fig. 1D. The results showed that, without IRI, there was no difference in the composition of Ly6C populations between Mafb^f/f and Mafb^f/f::LysM-Cre mice (Supplemental Fig. 1G, 1H); however, at 24 h after AKI induction, the Ly6C^int macrophage population was increased in the kidney of Mafb^f/f mice (Fig. 2G, 2H). The results suggested that MAFB shifts the macrophage phenotype toward an anti-inflammatory state in an IRI-dependent manner. To further analyze the feature of these Ly6C macrophage populations in detail, IL-6 expression in each population was measured by FACS. Ly6C exacerbates inflammation in the acute phase of AKI (33). Moreover, patients with higher IL-6 expression have a worse prognosis for AKI (34). FACS analysis showed significantly higher IL-6 expression in the Ly6C^high population compared with in the Ly6C^int and Ly6C^low populations (Fig. 2I, 2J). The results reinforce previous reports that Ly6C^high is an inflammatory population.

To investigate how Mafb deficiency causes a shift in the macrophage phenotype toward a proinflammatory state, we isolated macrophages from the kidneys of Mafb^f/f and Mafb^f/f::LysM-Cre mice at 24 h post-IRI and compared gene expression by RNA-seq analysis. For the isolation of macrophages, CD11b⁺ cells were isolated from the kidneys by MACS, and Ly6G^lowLy6C⁺CD11b⁺ cells were sorted from these CD11b⁺ cells as macrophages (Supplemental Fig. 2A). RNA-seq analysis showed that gene expressions were altered in macrophages by the deficiency of Mafb, as shown by a heatmap (Supplemental Fig. 2B). Pathway analysis of these differentially expressed genes showed that the deficiency of Mafb caused abnormalities in the inflammatory response of macrophages (Supplemental Fig. 2C). These results suggest that the deficiency of Mafb in macrophages shifts the macrophage phenotype toward a proinflammatory state.

The volcano plot of the RNA-seq results showed that the expression of Alox15 and Syne1 was extremely downregulated in Mafb-deficient macrophages (Fig. 3A, 3B). Because ALOX15 is known to play a significant role in suppressing acute inflammation (17, 18), we hypothesized that MAFB promotes inflammation resolution in ischemic AKI by regulating the expression of Alox15 in macrophages. ALOX15 is a lipoxygenase that catalyzes the oxidation of essential fatty acids. ALOX15 expressed in macrophages produces precursors of inflammation-suppressing SPMs by oxidizing essential fatty acids at the site of inflammation (18, 19). SPMs have anti-inflammatory and proresolving functions in immune cells; therefore, they play essential roles in orchestrating the resolution of tissue inflammation (20).

To confirm the reduced ALOX15 expression in macrophages in Mafb^f/f::LysM-Cre mice at the protein level, immunostaining of ALOX15 and the macrophage marker F4/80 was performed on the kidneys of Mafb^f/f and Mafb^f/f::LysM-Cre mice before AKI induction and at 12 and 24 h post-AKI. The results showed no expression of ALOX15 in the kidney before IRI in both Mafb^f/f and Mafb^f/f::LysM-Cre mice (Fig. 3C, 3D, top). ALOX15⁺F4/80⁺ cells were already present in the kidneys of Mafb^f/f mice at 12 h post-IRI and their number increased further at 24 h; however, the number of ALOX15⁺F4/80⁺ cells was significantly lower in the kidneys of Mafb^f/f::LysM-Cre mice than in those of Mafb^f/f mice at both time points (Fig. 3D, 3D, middle, bottom). The results suggest that MAFB induces ALOX15 expression in macrophages in an AKI-dependent manner. Subsequently, LC-MS analysis for LMs was performed on the kidneys of Mafb^f/f mice before and at 12 and 24 h after AKI induction to verify whether the production of SPMs by ALOX15 actually takes place in the AKI-induced kidneys (Supplemental Fig. 3A). The results showed that 17-HDoHE, a precursor of the resolvin D series of SPMs, increased after 12 h. ALOX15 is essential for the production of 17-HDoHE from docosahexaenoic acid (DHA). After 24 h, 17-HDoHE was reduced, possibly due to conversion of 17-HDoHE to the resolvin D series. Previous studies have shown that the resolvin D series is protective against ischemic AKI (35, 36). Resolvins D1 and D2 were not detected in the current study; however, according to a previous study, the amount of the resolvin D series is not largely altered unless DHA is administrated before AKI induction (35). In the current study, the analysis was performed under natural conditions without DHA administration; therefore, resolvins would have been below detection sensitivity. Conversely, not much change was observed in the mediators produced from arachidonic acid by ALOX15. Indeed, there are reports of improved AKI prognosis in dogs and mice fed DHA rather thanarachidonic acid (37–39). PGE₂, PGF_2a, and TxB2, a metabolite of TxA2, was reduced at 24 h, suggesting that LM class switch from inflammatory LMs to SPMs would have occurred between 12 and 24 h after AKI induction.

To confirm that MAFB⁺ cells express ALOX15, immunostaining was performed on the kidneys of Mafb^gfp/+ mice 24 h after IRI. The results showed that ALOX15 colocalized with GFP (Fig. 3E). However, not all GFP⁺ cells colocalized with ALOX15, suggesting that a limited macrophage population expresses ALOX15.

Several downstream target genes of MAFB in macrophages have been identified, such as C1q genes, AIM, and Msr1 (11) (12, 13). We measured the expression of these MAFB target genes in macrophages infiltrating the kidney at 24 h after IRI, using qRT-PCR. The expression of Msr1 and C1q genes was decreased significantly; however, the expression of AIM did not change in macrophages from Mafb^f/f::LysM-Cre mice (Supplemental Fig. 3B). Because MSR1 and C1q are involved in the clearance of DAMPs and efferocytosis, respectively (11, 13), MAFB in macrophages may promote the suppression of inflammation by regulating the expression of the genes in ischemic AKI. Further studies are required to elucidate the involvement of MSR1 and C1q.

The Alox15 gene is expressed in M2-type macrophages following treatment with IL-4, IL-13, and RXR/LXR agonists (19). However, its expression is also induced in M1-type macrophages activated by LPS (40).

Therefore, we investigated whether MAFB regulates the expression of Alox15 in M1- or M2-type macrophages using in vitro–cultured macrophages. For this purpose, we first treated bone marrow–derived macrophages from Mafb^f/f and Mafb^f/f::LysM-Cre mice with the Th2 cytokines IL-4, IL-13, and 9cRA/T0901317 (an RXR/LXR agonist) and compared the expression of Alox15 by qRT-PCR. The expression of Alox15 was also measured in thioglycolate-elicited peritoneal macrophages from Mafb^f/f and Mafb^f/f::LysM-Cre mice because such macrophages are polarized to the M2 type (41). The results showed that in bone marrow–derived macrophages treated with Th2 cytokines, RXR/LXR agonists, and thioglycollate-elicited peritoneal macrophages, the expression of Alox15 was increased significantly in macrophages from Mafb^f/f::LysM-Cre mice (Supplemental Fig. 3C, 3D). The results suggest that, in M2-type macrophages, MAFB negatively regulates Alox15 expression. M2 macrophages have an important profibrotic role in chronic inflammatory conditions (42). Under chronic inflammatory conditions, such as CKD, ALOX15 has been reported to worsen prognosis by promoting fibrosis (43–45). Therefore, in M2 macrophages, MAFB would downregulate ALOX15 expression to suppress fibrosis. This regulation differs from the response observed during acute inflammation, such as AKI. Further analyses are required to clarify the molecular mechanism by which MAFB suppresses ALOX15 expression in M2 macrophages.

Next, we examined the regulation of Alox15 expression by MAFB in M1-type macrophages. For this purpose, thioglycollate-elicited peritoneal macrophages from Mafb^f/f and Mafb^f/f::LysM-Cre mice were primed with LPS, and Alox15 expression was compared using qRT-PCR. In contrast to in the M2-type macrophages, Alox15 expression was decreased significantly in Mafb-deficient macrophages 3 d after LPS treatment (Fig. 4A). The trend was consistent with the in vivo results (Fig. 3B, 3D). Because MAFB expression is regulated by the COX-2/PGE₂/EP4 pathway in macrophages infiltrating the kidney after IRI, and LPS induces COX-2 expression (16, 46), we hypothesized that MAFB regulates Alox15 expression under the COX-2/PGE₂/EP4 pathway in macrophages. EP4 induced MAFB expression via the cAMP/PKA/CREB pathway (16). The expression of Ptgs2, a gene encoding COX-2, was induced after LPS treatment in both Mafb^f/f and Mafb^f/f::LysM-Cre mice, and induction of Mafb expression was observed in Mafb^f/f-derived macrophages after 2 d of treatment (Fig. 4B). The results suggest that MAFB regulates Alox15 expression under the COX-2/PGE₂/EP4 pathway. To further confirm that MAFB regulates Alox15 expression under the COX-2/PGE₂/EP4 pathway, we primed peritoneal macrophages from Mafb^f/f and Mafb^f/f::LysM-Cre mice with PGE₂ and performed qRT-PCR. The results showed that Mafb expression increased in Mafb^f/f-derived macrophages significantly from day 1 after treatment (Fig. 4C). In addition, Alox15 expression was significantly lower in Mafb-deficient macrophages on day 2 after PGE₂ treatment (Fig. 4D). As PGE₂ exerts its effects at the picomolar to nanomolar levels in vivo, we investigated the PGE₂ concentration required for MAFB-mediated Alox15 induction in macrophages. The result showed that Mafb expression was induced in Mafb^f/f-derived macrophages from a PGE₂ concentration of 10 nM (Fig. 4E, left). Moreover, Mafb induction led to Alox15 induction in Mafb^f/f-derived macrophages starting at a PGE₂ concentration of 10 nM, whereas no Alox15 induction was observed in Mafb^f/f::LysM-Cre-derived macrophages (Fig. 4R, right). Furthermore, we examined whether administration of an EP4 antagonist (L161982) abolishes the expression of Mafb and Alox15 downstream of COX-2/PGE₂/EP4. As expected, LPS-induced Mafb and Alox15 expression was suppressed by inhibition of EP4 receptors (Fig. 4F). With regard to Alox15 mRNA, two out of five L161982-administrated samples were excluded because their dissociation curves showed abnormal shapes. The findings suggest that Alox15 was not amplified properly in these two samples. The results indicate that MAFB regulates the expression of Alox15 under the COX-2/PGE₂/EP4 pathway in macrophages.

Next, to determine MAFB-responsive regions in the Alox15 promoter, we searched for putative transcription factor–binding sites in the promoter of the Alox15 gene using the UCSC Genome Browser (http://genome.ucsc.edu/index.html). Of note, a potential half of the MARE (half-MARE) sites were identified as −267 bp (site 1) and −58 bp (site 2) upstream of the transcription initiation site of Alox15. The half-MARE sequence in site 1 is relatively conserved among multiple mammalian species but not in site 2 (Fig. 5A). To evaluate the potential of MAFB to interact with these half-MAREs, we constructed luciferase reporter genes linked to the 500 or 200 bp of the Alox15 promoter. These constructs were cotransfected with an MAFB-expressing vector into RAW264.7 cell line. The results showed that the MAFB-expressing vector activated only 500 bp of the Alox15 promoter but not 200 bp (Fig. 5B), suggesting that MAFB regulates the expression of Alox15 through the half-MARE at site 1. To confirm this result, a mutation was inserted at the site 1 and site 2 half-MARE site (Fig. 5B). As expected, the Alox15 promoter was deactivated when the mutation was inserted at site 1 but not at site 2 (Fig. 5B). These results suggest that MAFB regulates the Alox15 gene through a half-MARE site −267 bp upstream of the transcriptional initiation site of Alox15.

In conclusion, MAFB expressed on macrophages would promote the resolution of inflammation in ischemic AKI by inducing ALOX15 expression to produce SPMs under the COX-2/PGE₂/EP4 pathway (Fig. 6).

A previous study has reported participation of MAFB downstream of the COX-2/PGE₂/EP4 pathway in ischemic AKI (16). In macrophage-specific Mafb-deficient mice, renal function deteriorates during AKI onset, and the absence of MAFB leads to a decline in anti-inflammatory gene expression (16). However, the target genes of MAFB in the pathway remain unknown. According to the findings of the current study, MAFB regulates Alox15 expression in macrophages under the COX-2/PGE₂/EP4 pathway. ALOX15 is essential for production of SPMs. SPMs play a crucial role in the resolution of inflammation in ischemic AKI within 24 h by diminishing leukocyte infiltration and other associated inflammatory responses (35, 47). In addition, administration of DHA, the source of SPMs, reportedly improved the prognosis of AKI in mice and dogs within 24 h after the onset of AKI (37–39). Our data clearly demonstrated that MAFB regulates ALOX15 expression in macrophages infiltrated into the kidney 24 h after IRI (Fig. 3). Therefore, MAFB expressed on macrophages would promote the resolution of inflammation in ischemic AKI by inducing ALOX15 expression to produce SPMs (Fig. 6). PGE₂ plays a pivotal role in the LM class switch from inflammatory LMs to SPMs; therefore, MAFB in macrophages would play a crucial role in the PGE₂-mediated LM class switch.

A key limitation of the current study is that there is no direct evidence that ALOX15 promotes SPM production and improves AKI prognosis in vivo, which is a subject for future work. In addition, our luciferase assay clearly demonstrated that the half-MARE site at −267 bp upstream of the transcriptional initiation site of Alox15 is important for the MAFB-mediated regulation of Alox15 expression. However, further analyses such as chromatin immunoprecipitation assays are required to provide direct evidence for the direct regulation of Alox15 by MAFB.

We also observed that MAFB in macrophages regulates C1q and Msr1 expression in ischemic AKI. C1q induces efferocytosis, and MSR1 accelerates the clearance of DAMPs (11, 13). Therefore, MAFB may also promote AKI alleviation by regulating these genes. Further analyses are required to elucidate the involvement of C1q and MSR1 in the MAFB-mediated alleviation of ischemic AKI.

Recently, the mechanism by which PGE₂ induces ALOX15 expression in human amniotic fibroblasts has been reported (48, 49). In human amniotic fibroblasts, PGE₂ activates the cAMP/PKA pathway by binding to EP2 receptors, followed by the phosphorylation of CREB and STAT3. Phosphorylated CREB and STAT3 interact with p300 and glucocorticoid receptor (GR) at the ALOX15 gene promoter to induce ALOX15 expression. In ischemic AKI, PGE₂ activates the cAMP/PKA/CREB pathway by binding to the EP4 receptor, increasing Mafb expression in macrophages (16). Our results showed that MAFB regulates Alox15 expression under this condition. Therefore, phosphorylated CREB, STAT3, p300, and GR may regulate Alox15 expression in macrophages by interacting with MAFB. In fact, MAFB has been reported to interact with GR and p300 (50, 51). Although there is no direct evidence for a relationship between MAFB and CREB or STAT3, they may interact to regulate Alox15 expression.

Previous studies have reported that MAFB in macrophages is activated in response to body abnormalities, such as autoimmune diseases, ischemic stroke, wound injury, obesity, and cold exposure; therefore, MAFB in macrophages is essential for homeostasis in the body (11, 13–15, 52). MAFB is upregulated by various nuclear receptor agonists, including PPARs, LXRs, RARs, and GRs (10, 11). Such agonists are potential candidates for therapeutic agents targeting MAFB under various pathological conditions. However, in the presence of an LXR/RXR agonist, MAFB downregulated Alox15 expression (Supplemental Fig. 3B); this is because in macrophages infiltrated into AKI-induced kidney, MAFB expression is regulated by the COX-2/PGE₂/EP4 pathway, which is quite different from the regulation by nuclear receptors. Conversely, under chronic inflammatory conditions, such as CKD induced by unilateral ureteral obstruction or the 5/6 nephrectomized model, ALOX15 promotes fibrosis and worsens prognosis (43–45). In such CKD and in the chronic phase of AKI, MAFB would suppress ALOX15 expression, and LXR/RXR agonists could be therapeutic agents. Further analyses are required to clarify how MAFB regulation is altered by the surrounding environment.

In conclusion, the current study identified Alox15 as a target gene of MAFB in macrophages under the COX-2/PGE₂/EP4 pathway. MAFB would play a crucial role in the PGE₂-mediated LM class switch by inducing the expression of ALOX15, which catalyzes SPM production. Our findings provide new insights into the molecular mechanisms of LM regulation under acute inflammatory conditions and pave the way for the development of novel therapies for acute inflammatory diseases.

The authors have no financial conflicts of interest.

We acknowledge the School of Integrative and Global Majors (SIGMA) Ph.D. program in Human Biolog, University of Tsukuba, for student support. We thank Masami Ojima for providing technical support. We also thank Yukiyo Ida for assisting in purchasing the materials required for the experiments. We also acknowledge Yuri Inoue, Omar Samir, and Hibiki Ueno for supporting Mafb^f/f::LysM-Cre mice colony management.