Trem2⁺ Macrophages Alleviate Renal Tubule Lipid Accumulation and Ferroptosis in Diabetic Nephropathy by Repressing IL-1β–Mediated CD36 Expression

Abstract

The presence of macrophages surrounding lipotoxic tubular epithelial cells (TECs) is a hallmark of diabetic nephropathy (DN). Nevertheless, the mechanisms of communication between these cell types are not well understood. Previous studies have revealed a unique subset of macrophages that express triggering receptor expressed on myeloid cells 2 (Trem2) in kidneys of human patients and mice with DN. Here, we explored the characteristics and the function of Trem2⁺ macrophages in the progress of DN. RNA-sequencing of macrophages in kidneys of Trem2 knockout (KO) mice fed a high-fat diet plus streptozotocin (HFD/STZ) revealed functional enrichment of metabolic processes, cytokine production, positive regulation of extracellular signal-regulated kinase (ERK) cascades, and the regulation of phagocytosis. In vivo studies demonstrated that Trem2⁺ macrophages reduced lipid accumulation and mitigated ferroptosis of TECs in diabetic mice. Mechanistically, Trem2-deficient macrophages amplified the production of interleukin-1β (IL-1β) through activating the ERK signaling pathway. Furthermore, IL-1β triggered CD36 expression via the transcription factor NF-κB. Bioinformatics and functional assays showed NF-κB binds the CD36 promoter, which directly bound to the promoters of CD36 to facilitate its transcription. Inhibition of NF-κB blocked IL-1β–induced CD36 production. This mechanism is exacerbated in Trem2-deficient macrophages, which release excess IL-1β to activate NF-κB in tubular cells, promoting CD36-dependent lipid uptake and ferroptosis. Additionally, we found Trem2 plays a role in enhancing the phagocytosis and clearance of ferroptotic cells by bone marrow-derived macrophages. Altogether, our results suggest Trem2⁺ macrophages maintain homeostasis of the renal microenvironment and exert a protective function in DN.

Article Highlights

• Levels of triggering receptor expressed on myeloid cells 2 (Trem2) in macrophages are increased in human patients and in mice with diabetic nephropathy.

• Trem2 suppresses the extracellular signal-regulated kinase signaling pathways, thereby inhibiting IL-1β production in macrophages.

• Macrophage Trem2 deficiency exacerbates tubular cell lipid deposition and ferroptosis by increasing CD36 expression in an IL-1β–dependent manner

Graphical Abstract

Introduction

Diabetic nephropathy (DN) is a significant kidney complication associated with diabetes mellitus and has emerged as the leading cause of end-stage renal disease (1,2). It is increasingly believed that inflammation plays a significant role in the development of DN (3,4). DN is marked by the death of renal tubular epithelial cells (TECs) due to lipotoxicity and the infiltration of inflammatory cells (5,6). Lipid accumulation is a common phenomenon in patients with DN and is mainly concentrated in tubules (7,8). Moorhead et al. first proposed the nephrotoxicity hypothesis in 1982 (9). Lipid accumulation toxicity–mediated tubular damage may be a major cause of DN, and increasing evidence supports the hypothesis that lipotoxicity causes TEC damage and promotes renal disease progression (10–12). The presence of immune cell infiltration (primarily macrophages) is frequently noted in the glomeruli and interstitial areas of renal biopsy specimens across all stages of DN (13).

Single-cell RNA sequencing of CD45⁺ immune cells from diabetic mouse kidneys revealed that macrophages occupying the renal microenvironment were heterogeneous, including distinct macrophage subsets. Among these various macrophage populations, both Mrc1^hi Mac and Trem2^hi Mac displayed increased expression of genes associated with M2 polarization and the resolution of inflammation. Trem2^hi Mac also expressed genes such as Trem2, Cd9, Spp1, and Lgals3, which are linked to the attenuation of macrophage activation, suggesting anti-inflammatory function of these macrophages in early DN (4). Trem2 is a unique transmembrane protein that is part of the super Ig receptor family and is involved in metabolic homeostasis of myeloid cells (14). Trem2 was highly expressed in lipid-associated macrophages, Kupffer cells in nonalcoholic fatty liver disease, and aortic macrophages in atherosclerosis, showing a close and complex relationship with metabolic syndrome (14–16). Subramanian et al. (17) demonstrated the involvement of Trem2⁺ macrophages in lipid metabolism and the injury of TECs. However, little is known about the function of Trem2 macrophages on the lipid metabolism and injury of TECs in DN.

In this study, we demonstrated elevated Trem2 expression in both patients with DN and mice with diabetes induced by a high-fat diet and streptozotocin (HFD/STZ). Trem2^−/− mice fed an HFD exhibited more intensive kidney injury and lipid deposition in TECs. Trem2^−/− macrophages released more IL-1β with exposure to high glucose and fatty acid, which, in turn, promoted the expression of CD36 via the transcription factor NF-κB in TECs, resulting in enhanced lipid deposition and cell injury.

Research Design and Methods

Animal Model Establishment, Drug Administration, and Grouping Design

Male C57BL/6J mice (aged 6–8 weeks, weight 18–22 g) and C57BL/6J Smoc-Trem2em1Smoc (Trem2 KO) mice were obtained from Model Organisms. Male mice were randomly assigned to a standard-fat diet (SFD) (10% fat) or HFD (18.1% protein, 61.6% fat, 20.3% carbohydrates; Research Diets, catalog no. D12492). After 4 weeks of HFD, mice underwent unilateral nephrectomy under isoflurane anesthesia. HFD was continued for 2 more weeks, after which HFD mice received intraperitoneal STZ (50 mg/kg, Sigma-Aldrich, catalog no. S0130-500MG) for 5 consecutive days, while controls got citrate-phosphate buffer. Tail vein blood glucose was measured at 72 h and 7 days post-STZ; mice with stable glucose over 7 days were considered successful diabetic models. After 16 weeks of HFD, mice were fasted for 6 h and sacrificed.

Flow Cytometry Analysis

For cell surface staining, suspensions were incubated with antibodies at 4°C for 30 min. Intracellular staining involved fixation/permeabilization using the Foxp3/Transcription Factor Staining Buffer Set (eBioscience), followed by 30-min antibody incubation at 4°C. Samples were washed, acquired on a BD LSRFortessa cytometer, and analyzed with FlowJo V.10.0.7.

Most antibodies and isotype controls were obtained from BioLegend (BD), including the Zombie NIR Fixable Viability Kit (BD, catalog no. 423106), TruStain FcX CD16/32 (BD, catalog no. 101320), CD45-PE (BD, catalog no. 147712), F4/80-BV421 (BD, catalog no. 123132), CD11b-PerCp/Cy5.5 (BD, catalog no. 101228), CD4-FITC (BD, catalog no. 100405), CD19-BV421 (BD, catalog no. 115537), T-bet-PE/Cy7 (BD, catalog no. 644823), PD-1-APC (BD, catalog no. 135210), and CD45-PerCp/Cy5.5 (BD, catalog no. 550994); other reagents were sourced from different suppliers: Arg1 (R&D, catalog no. IC5868N), Trem2-FITC (Thermo Fisher Scientific, catalog no. MA5-28223), and Foxp3-PE (eBioscience, catalog no. 12–5773-82).

Cell Culture and Treatment

Primary TECs from the renal cortex of 8-week-old, male C57/BL6 mice were cultured in RPMI 1640 medium with 10% FBS and 1% penicillin/streptomycin under sterile conditions. Bone marrow-derived macrophages (BMDMs) from WT and Trem2^−/− mice were exposed to a high glucose (HG) concentration (30 mmol/mL) plus palmitate (PA) (0.3 mmol/L) or vehicle for 12 h. Supernatants were discarded and replaced with RPMI 1640 medium for 24 h, and the resulting conditioned medium (CM) was mixed with the HG plus PA combination (HGPA) to coincubate primary TECs for 24 h. TECs treated with BSA served as normal controls (NCs).

To explore if Trem2 deficiency promotes IL-1β via extracellular signal-regulated kinase (ERK) activation, BMDMs were treated with the ERK inhibitor PD98059. After 12 h of HGPA/vehicle exposure, supernatants were replaced with RPMI 1640 medium containing PD98059 for 24 h. CM from these was mixed with HGPA to stimulate TECs for 24 h.

For neutralization studies, BMDMs exposed to HGPA/vehicle for 12 h had supernatants replaced with RPMI 1640 medium plus IL-1β neutralizing antibody (5 μg/mL) for 24 h. The resulting CM mixed with HGPA-stimulated TECs for 24 h.

Cell Imaging

Oil Red O Staining

After the interventions, cultured renal TECs were washed thrice with 1× PBS, fixed in 4% paraformaldehyde for 20 min, then washed twice. They were stained with Oil Red O (ORO) reagents for 30 min, followed by thorough 1× PBS washes to remove excess dye. Intracellular lipid droplets were observed and photographed via inverted microscopy.

FerroOrange Staining

FerroOrange is a fluorescent probe that enables live-cell fluorescent imaging of intracellular iron ions. To summarize, TECs were treated with HBSS buffer solution containing 1 µmol/L FerroOrange reagent. After a 1-h incubation, intracellular Fe²⁺ was detected by flow cytometry.

Measurement of Lipid Peroxidation

Liperfluo staining involved incubating cells with 10 μmol/L Liperfluo at 37°C for 1 h, followed by trypsin digestion and immediate flow cytometric analysis via FITC channel for fluorescence intensity measurement.

Determination of Malondialdehyde and Glutathione Levels

Malondialdehyde (MDA) and glutathione (GSH) levels in cortical tissues and primary TECs were measured using Beyotime’s Lipid Peroxidation MDA Assay Kit (catalog no. S0131) and GSH Assay Kit (catalog no. S0053), respectively. Assays were conducted following manufacturer instructions, with concentrations determined via microplate reader at 532 nm (MDA) and 412 nm (GSH).

Primary TECs Treated With IL-1β

Primary TECs were extracted as indicated. Primary TECs from the second passage were cultured in serum-free medium for 24 h and then subjected to 1) medium containing 5 mmol/L glucose (NC group); 2) HGPA (HGPA group); or 3) HGPA plus IL-1β (10 ng/mL) (IL-1β group).

RNA Sequencing Analysis of Macrophages

CD45⁺F4/80⁺ macrophages in renal tissues of HFD/STZ mice were sorted using flow cytometry. Single cells were treated with lysis buffer, followed by quantification and purification, and cDNA libraries were constructed using Clontech’s SMART-Seq HT Kit according to the manufacturer’s instructions. Samples were subjected to 150 bp paired-end sequencing on an Illumina NovaSeq6000 sequencer. After filtering and quality control of raw reads, directional sequencing reads were mapped to the GRCh38 genome using the spliced mapping algorithm of Hisat2 (version 2.0.4). After genome alignment, gene expression levels were determined by calculating fragments per kilobase of exon model per million (FPKM) mapped reads using Perl. edgeR was used for differential gene analysis among samples.

P values were subjected to multiple hypothesis testing correction to obtain q values (with the false discovery rate [FDR] as the control threshold), and the fold-change was calculated based on FPKM values. Significant differentially expressed genes (DEGs) were filtered using the criteria of FDR <0.05 and fold-change >1.5. Subsequently, Gene Ontology(GO) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of DEGs were performed using the Database for Annotation, Visualization, and Integrated Discovery.

mRNA-Seq Analysis of Primary TECs Treated With WT or Trem2^−/− BMDM CM

Primary TECs from the renal cortex of 8-week-old, male C57/BL6 mice were cultured in RPMI 1640 with 10% FBS and 1% penicillin/streptomycin. BMDMs from WT and Trem2^−/− mice were exposed to HGPA or vehicle for 12 h, then cultured in RPMI 1640 for 24 h. The CM was collected, mixed with HGPA, and coincubated with primary TECs for 24 h. Primary TECs with BSA served as NCs. For mRNA-seq (n = 3), total RNA was extracted using the RNeasy Micro Kit. cDNA library quality was assessed by the Agilent 4200 bioanalyzer before sequencing with Illumina Novaseq6000. Library construction included RNA purification, fragmentation, cDNA synthesis, end repair, 3′ end addition, and ligation. Raw reads were filtered by Seqtk and mapped to genome using Hisat2 (version 2.0.4), and gene fragments counted by StringTie (version 1.3.3b) with normalization of trimmed mean of M values. DEGs were identified using edgeR with FDR <0.05 and fold change >2. Each group had three biological replicates.

Phagocytosis Assay

Ferroptosis was induced in primary TECs (seeded at 0.5 × 10⁶ cells/mL) using 0.5 µM (1S,3R)-RSL3. Induced ferroptotic cells were resuspended in FBS-free medium and stained with 20 ng/mL pHrodo iFL STP ester for 30 min. For phagocytosis assays, stained ferroptotic cells were resuspended in complete RPMI 1640 medium and cocultured with BMDMs for 1 h at a consistent phagocyte-to-target cell ratio of 1:5. Phagocytic engulfment rate was determined via flow cytometry as the percentage of PHrodo-positive cells within the F4/80⁺ population.

Measurement of FFA Uptake

To assess Transformed C3H Mouse Kidney-1 (TCMK-1) cells uptake of free fatty acid (FFA), cells were washed in PBS and loaded with the fluorescent probe boron dipyrromethene (BODIPY) FL C16. In vitro uptake was monitored via confocal microscopy until plateau, with fluorescence measured at 505 nm excitation and 510 nm emission.

Isolation and Culture of Primary TECs

Renal cortex tissues were obtained form 8-week-old male C57/BL6 mice under sterile conditions. All experiments were approved by the Animal Ethics Committee of the Chinese PLA General Hospital, Beijing, China. All efforts were exerted to minimize suffering and reduce the number of animals used. Tissues were digested with type I collagenase and passed through a screen filter to isolate primary renal proximal tubule epithelial cells. Cells from the second passage were cultured in serum-free medium for 24 h and then subjected to 1) medium containing 5 mmol/L glucose (NC group); 2) medium containing 5 mmol/L glucose + Ferrostatin-1 (NC-Fer-1 group); 3) High glucose (30 mmol/L) + palmitate (300 μmmol/L) medium (HGPA group); 4) HGPA medium + Fer-1 (HG+PA-Fer-1 group); and 5) 1S,3R (RSL3), 0.5 μmmol/L medium) (RSL3 group).

Statistical Analysis

Statistical analyses were conducted with GraphPad Prism 8. Results are expressed as mean ± SEM. Based on experimental design (detailed in figure legends), either two-tailed unpaired Student t test or one-way ANOVA was used for statistical assessment. P < 0.05 was considered statistically significant.

Ethics Statement

The studies involving human participants were reviewed and approved by the Ethics Committee of the Chinese People’s Liberation Army General Hospital, Beijing, China. Ethical approval was granted by the ethics committee (approval no. S2022-482-01). The patients/participants provided their written informed consent to participate in this study. All animals were housed in the specific pathogen-free laboratory animal center of Chinese People’s Liberation Army General Hospital, according to the guidelines of the Institutional Animal Care and Use Committee at the facility. The investigations were conducted in accordance with the principles of the Declaration of Helsinki and were approved by the Chinese People’s Liberation Army General Hospital.

Data and Resource Availability

The raw sequencing reads of macrophages generated in this study have been deposited in the Sequence Read Archive of the National Center for Biotechnology Information (Bioproject accession no. PRJNA1208585).

Results

Renal Tubular Lipid Deposition, Ferroptosis, and Abundant Trem2⁺ Macrophages in Diabetic Kidneys

We modeled DN in mice in a metabolic model of HFD/STZ (Fig. 1A). Compared with the SFD group, the HFD/STZ group exhibited significantly elevated levels of blood glucose (Fig. 1B) and urinary albumin to creatinine ratio (Fig. 1C). Measurement of renal function showed that the levels of blood urea nitrogen (BUN) (Fig. 1D) and serum creatinine (Scr) (Fig. 1E) were increased significantly in HFD/STZ mice. In addition, the cholesterol and triglyceride levels were elevated significantly in diabetic mice (Fig. 1F and G). Periodic acid-Schiff staining showed mesangial matrix expansion in diabetic kidney (Fig. 1H). Transmission electron microscopy showed that HFD/STZ mice had glomerular basement membrane thickening (Fig. 1I). This result confirmed that the DN mouse model was successfully established at week 16. Immunofluorescence staining and Western blot (WB) analyses showed that the level of kidney injury molecule-1 (KIM-1) was higher in DN mice compared with control mice (Fig. 1J and K). A previous study showed that lipid accumulation has direct toxic effects on renal cells by initiating endoplasmic reticulum stress and oxidative stress (18). Oxidative stress induces mitochondrial iron overload and ferroptosis in cardiomyocytes (19).

Figure 1.

Renal tubular lipid deposition, ferroptosis, and abundant renal Trem2⁺ macrophages in diabetic kidneys. A: WT mice at 8 weeks old were induced into type 2 diabetes and fed an HFD. B: Blood glucose levels of two groups of mice. C: Urinary albumin to creatinine ratio was measured at 16 weeks. D and E: BUN and Scr of two groups mice at 16 weeks. F: Serum cholesterol. G: Serum triglycerides. H: Periodic acid-Schiff (PAS) staining of mouse kidney sections in two group, Quantification of mesangial matrix expansion (%). Scale bar = 20 μm. I: Representative transmission electron microscopy (TEM) images. Scale bar = 1 μm. J: Representative images and quantitative data of KIM-1 by immunofluorescent staining. Scale bar = 20 μm. K: Expression of KIM-1 in diabetic mice kidney measured by WB. L: Representative images and quantitative data of BODIPY by immunofluorescent staining. Scale bar = 20 μm. The iron (M), MDA (N), and GSH (O) contents in mouse kidney tissue lysates. P: Expression of 4-HNE, GPX4, and ACSL4 in diabetic mouse kidney measured by WB. Q: The statistical results of detecting the 4-HNE, GPX4, and ACSL4 proteins by WB. R: Representative dual-immunofluorescence staining of F4/80 (green) and Trem2 (red) in murine kidney. White arrows indicate F4/80 and Trem2 copositive cells. Scale bar = 20 μm. S: Percentage of Trem2⁺ macrophages in diabetic kidney detected by flow cytometry. T: The statistical results of detecting the Trem2⁺ macrophages in diabetic kidney. U: Expression of Trem2 protein level in diabetic mice kidneys measured by WB. V: Relative mRNA expression of Trem2 in diabetic mice kidney measured by RT-qPCR. Data are presented as the mean ± SD; n = 6. *P < 0.05, **P < 0.01, ***P < 0.001.

To characterize the lipid metabolism and cell injury in tubule cells of HFD-induced diabetic mice, lipid deposition and markers of ferroptosis were detected in kidney tissues. The area positive for BODIPY staining in the HFD/STZ group was significantly greater than that in SFD group (Fig. 1L), indicating tubular lipid accumulation in DN. Compared with the SFD group, the HFD/STZ group had increased renal iron content, indicating iron accumulation in diabetic kidneys (Fig. 1M). MDA and 4-hydroxynonenal (4-HNE) are two important markers of lipid oxidative injury in the cell membrane. The HFD/STZ mice manifested increased 4-HNE expression and MDA level in the kidney tissue (Fig. 1N, P, and Q), suggesting lipid oxidative injury in the kidney.

Next, we determined the expression of the ferroptosis-related proteins acyl-CoA synthetase long-chain 4 (ACSL4) and glutathione peroxidase 4 (GPX4) in renal tissues. The expression of GPX4 was significantly decreased and the expression of ACSL4 was significantly increased in the kidney tissues of HFD/STZ mice (Fig. 1P and Q), suggesting ferroptosis in HFD/STZ mice. In line with the reduced GPX4, levels of the cellular antioxidant GSH also were significantly lower in the HFD/STZ kidney (Fig. 1O), indicating that the diminished antioxidant capacity further exacerbated ferroptosis.

The accumulation of Trem2⁺ macrophages were observed in the kidneys of HFD/STZ mice (Fig. 1R). Furthermore, flow cytometry revealed Trem2⁺ macrophages accounted for 35.36% of F4/80⁺ macrophages in HFD/STZ versus 4.45% in SFD kidneys (Fig. 1S and T). The expression of Trem2 was significantly upregulated in the kidneys of HFD/STZ mice (Fig. 1U and V).

Next, we examined the accumulation of Trem2⁺ macrophages in patients with established DN and glomerular minor lesion (GML) (Supplementary Table 1 lists the clinical characteristics). There was no difference in glomerular filtration rate between patients with DN and those with GML (Supplementary Fig. 1A), although the levels of Scr were increased significantly in patients with DN (Supplementary Fig. 1B). Patients with DN had dysregulated lipid metabolism, as evidenced by elevated serum levels of triglycerides, total cholesterol, and LDL (Supplementary Fig. 1C–E). Immunostaining showed that the number of Trem2⁺ macrophages in patients with DN was significantly increased compared with those in patients with GML (Fig. 1F), implicating the involvement of Trem2⁺ macrophages in DN progression.

Trem2 Deficiency Exacerbates Renal Lipid Accumulation and Ferroptosis in HFD/STZ-Induced Diabetic Mice

Recent studies showed that Trem2⁺ macrophages in adipose tissues might regulate lipid metabolism and protect against HFD-induced adipocyte hypertrophy (14). To explore their roles in tubular lipid metabolism and cell injury in diabetic kidneys, we generated Trem2 KO (Trem2^−/−) mice and analyzed renal lipid accumulation and ferroptosis after HFD/STZ treatment (Fig. 2A). As anticipated, the HFD/STZ-induced diabetic mice had elevated blood glucose levels. However, no significant differences in blood glucose level were observed between Trem2^−/− and WT mice subjected to HFD/STZ (Fig. 2B). The levels of Scr (Fig. 2C) and BUN (Fig. 2D) were significantly increased in diabetic Trem2^−/− mice compared with WT mice. Notably, compared with WT mice, Trem2^−/− mice had aggravated lipid metabolism disorders, as evidenced by an increase of serum triglyceride and cholesterol levels after 16 weeks of HFD (Fig. 2E and F). BODIPY staining revealed increased lipid accumulation in TECs of diabetic Trem2^−/− mice compared with WT mice (Fig. 2G), indicating Trem2 deficiency aggravated lipid metabolism disorders in the TECs (Fig. 2H). Moreover, significant upregulation of KIM-1 expression was observed in the tubular cells of diabetic Trem2^−/− mice compared with WT mice (Fig. 2I and M). The iron content in kidneys of Trem2^−/−-HFD/STZ mice was increased compared with the WT-HFD/STZ mice (Fig. 2J). Notably, the MDA level was higher (Fig. 2K) and GSH level was lower in Trem2^−/− kidneys (Fig. 2L). In Trem2^−/− kidneys, oxidative stress markers were elevated, as evidenced by increased 4-HNE and ACSL4 expression (Fig. 2M and N), which was accompanied by decreased expression of the GPX4 (Fig. 2M, P, and Q). Collectively, these data demonstrate that Trem2 deficiency exacerbates lipid-induced oxidative injury and ferroptosis in renal tubules.

Figure 2.

Trem2 deficiency exacerbates renal lipid accumulation and ferroptosis in HFD mouse model. A: The Trem2^−/− and WT mice at 8 weeks old were induced into type 2 diabetes and fed an HFD. B: blood glucose levels were measured every week. C and D: Serum Cr and BUN of four groups mice at 16 weeks. E and F: The serum concentrations of cholesterol (E) and triglycerides (F) were measured. G: Representative images and quantitative data of BODIPY by immunofluorescent staining. Scale bar = 20 μm. H: Representative image of PAS staining of kidney cortex and Quantification of tubulointerstitial injury index. Scale bar = 20 μm. I: Representative images and quantitative data of KIM-1 by immunofluorescent staining. Scale bar = 20 μm. J–L: The iron (J), MDA (K), and GSH (L) contents in WT and Trem2^−/− mouse kidney tissue lysates. M: Expression of KIM-1, 4-HNE, GPX4, and ACSL4 in diabetic mice kidney measured by WB. N–Q: The statistical results of detecting the KIM-1 (N), 4-HNE (O), GPX4 (P), and ACSL4 (Q) proteins by WB. R: The expression of interleukin 1β (IL-1β) and tumor necrosis factor-α (TNF-α) in diabetic mice kidney measured by WB. S and T: The statistical results of detecting the IL-1β (S) and TNF-α (T) protein by WB. U–Z: Relative mRNA expression of inflammatory cytokines detected by RT-qPCR. Data are presented as the mean ± SD; n = 6. *P < 0.05, **P < 0.01, ***P < 0.001.

To further assess the effect of Trem2 deficiency on inflammation in DN kidneys, expression of inflammatory cytokines was measured. Compared with WT diabetic kidney, Trem2^−/− kidneys had significantly higher levels of proinflammatory cytokines IL-1β, TNF-α, CCL2, and CXCL2 (Fig. 2R–X). In contrast, no differences were observed in IL-6 and the anti-inflammatory cytokine IL-10 between the two groups (Fig. 2Y and Z). To further interrogate how Trem2 deficiency modified the immune response to DN, we detected immune cell populations by using flow cytometry in the kidneys of WT and Trem2^−/− mice. Trem2⁺ macrophages accounted for 48.6% of the F4/80^high CD11b^hi macrophages in WT diabetic mice and the percentage was significantly reduced in Trem2^−/− mice (Supplementary Fig. 2A). Arg1⁺ macrophages, representing an M2 phenotype, are recognized as major contributors to inflammation suppression and tissue repair (20,21). The percentage of Arg1⁺ macrophages in F4/80^high CD11^high macrophages was 20.2% in diabetic WT mice, which was reduced to 0.06% in Trem2^−/− mice (Supplementary Fig. 2B), indicating that Trem2 might be involved in the inflammation-resolving function of macrophages. We also found an increase of CD4⁺ T cells, Foxp3⁺ T cells, PD-1⁺ T cells, T-bet⁺ T cells, and CD19⁺ B cells in diabetic Trem2^−/− mice compared with WT mice (Supplementary Fig. 2C–G), further indicating unleashed inflammation in diabetic Trem2^−/− mice kidneys. These changes may be influenced by multiple factors (e.g., macrophage-derived cytokines, lipid metabolites).

Trem2-Deficient Macrophages Enhanced Production of Proinflammatory Cytokines by Activating the ERK Pathway

To investigate the mechanisms underlying the impaired lipid metabolism and tubular injury observed in diabetic Trem2^−/− mice, we performed RNA-seq on macrophages isolated from Trem2^−/− and WT mice kidneys and the DEGs were identified (P < 0.05; log₂ fold-change > 1.5). In diabetic Trem2^−/− mice, 425 genes were significantly upregulated, including the proinflammatory mediators IL-1β, CCL2, Mink1, and Klf2, and 2,269 genes were downregulated, including the anti-inflammatory mediators Arg1, Trem2, SPP1, and Fabp3 (Fig. 3A). Functional GO enrichment of DEGs highlighted pathways of metabolic process, cell migration, lymphocyte activation, cytokine production, and ERK1 and ERK2 cascade (Fig. 3B). KEGG pathway analysis showed enrichment in the chemokine signaling pathway, MAPK signaling pathway, and cytokine-cytokine receptor interaction (Fig. 3C). To validate the impact of Trem2 on macrophage-mediated inflammation, WT and Trem2^−/− BMDMs were treated with HGPA, and inflammatory cytokine production was measured. The efficiency of Trem2 deletion in BMDMs was validated (Fig. 3E and F). As shown in Fig. 3G and H, the mRNA levels of IL-1β, TNF-α, and iNOS were upregulated in Trem2^−/− BMDMs treated with HGPA compared with WT BMDMs, supporting unleashed production of inflammatory mediators due to Trem2 deficiency.

Figure 3.

Trem2-deficient macrophages enhanced production of IL-1β by activating the ERK pathway. The RNA-seq analysis of macrophages from diabetic Trem2^−/− or WT diabetic mice for 16 weeks. A: Volcano plot of the inflammatory genes with significantly different levels in macrophages. B and C: Representative GO (B) and KEGG (C) enrichment pathways enriched in DEGs in macrophages. D: GO enrichment of Trem2-related differential gene pathways. E: The protein expression of Trem2 in WT BMDMs and Trem2^−/− BMDMs treated with HGPA was detected by WB. F: The Trem2 mRNA was detected in WT BMDMs and Trem2^−/− BMDMs by RT-qPCR. G: Relative mRNA expression levels of IL-1β, TNF-α, and INOS in WT and Trem2^−/− BMDMs were measured by RT-qPCR. H: ELISA analysis of the protein abundance of IL-1β and TNF-α in the cultural media of BMDMs with or without HGPA treatment. I: The levels of phosphorylated ERK by WB in Trem2^−/− BMDMs and WT BMDMs treated with HGPA or an equal volume of BSA (vehicle). J: RT-qPCR analysis of the mRNA of IL-1β, TNF-α, and INOS of BMDMs with or without PD98059. K: ELISA assay the protein abundance of IL-1β and TNF-α in the cultural media of PD98059-treated BMDMs. Data are presented as the mean ± SD; n = 3. *P < 0.05, *P < 0.01, ***P < 0.001.

We next explored whether Trem2 regulates macrophage polarization in BMDMs. Flow cytometric analysis indicated that the percentage of F4/80⁺CD206⁺ cells was decreased in Trem2^−/− BMDMs treated with HGPA compared with WT BMDMs (Supplementary Fig. 3A), indicating that HGPA treatment induced Trem2^−/− BMDM polarization to M1.

GO enrichment of Trem2-related differential genes (Fig. 3D) indicated that the ERK pathways are involved in Trem2 signaling. ERK1 and ERK2 are major MAPKs that regulate a wide variety of cellular processes and are critical regulators of inflammation (22). We examined the activation of the ERK pathway in WT and Trem2^−/− BMDMs treated with HGPA. Phosphorylation of ERK was enhanced in Trem2^−/− BMDMs compared with WT BMDMs after HGPA treatment (Fig. 3I), which is in accordance with the increased expression of inflammatory cytokines. Notably, the upregulation of inflammatory cytokines in Trem2^−/− BMDMs was significantly suppressed by the ERK pathway antagonist PD98059 (Fig. 3J and K), indicating that Trem2 deficiency resulted in an exacerbated proinflammatory response to HGPA through the ERK signaling pathway.

Notably, GO enrichment of Trem2-related DEGs underlined the process of phagocytosis (Fig. 3D). The process of clearing dying cells by macrophages (efferocytosis) is a critical step in inflammation resolution and tissue repair (23,24). We sought to identify the function of Trem2 in the clearance of ferroptotic tubule cells by macrophages. The Phrodo-labeled ferroptotic cells were cocultured with WT and Trem2^−/− BMDMs. The phagocytosis efficiency of BMDMs was evaluated (Supplementary Fig. 4A). The results showed that Trem2 deficiency attenuated phagocytosis capacity of macrophages to clear dead TECs (Supplementary Fig. 4B and C), suggesting that Trem2 might regulate the capacity of macrophages to remove tubule cell debris, which may also be an underlying mechanism for resolving tubular inflammation in DN.

Trem2-Deficient Macrophages Promote Lipid Accumulation and Ferroptosis of TECs via Enhanced IL-1β Production

To further validate the tubule cell lipid metabolism disorders and ferroptosis, primary TECs were isolated and stimulated with HGPA for 24 h. ORO staining revealed that primary TECs treated with HGPA had significantly increased lipid accumulation (Supplementary Fig. 5A), and excessive accumulation of iron was revealed by FerroOrange staining (Supplementary Fig. 5B). Consistent with in vivo data, primary TECs treated with HGPA had significantly reduced GPX4 expression and increased ACSL4 compared with the NC group (Supplementary Fig. 5C), a finding that was consistent with that induced by ferroptosis inducer RSL3 (0.5 μmol/L). Additionally, levels of the cellular antioxidant GSH were notably lower in the HGPA group (Supplementary Fig. 5D), suggesting that diminished antioxidant capacity further exacerbated ferroptosis. HGPA treatment also increased MDA and lipid peroxidation intensity (Supplementary Fig. 5E and F). Moreover, the cell death rate was significantly higher in the HGPA group than in the NC group, indicating that HGPA can induce TEC cell death (Supplementary Fig. 5G). Furthermore, treatment with ferrostatin-1 (100 µM), an inhibitor of the ferroptosis, rescued the HGPA-induced ferroptosis in TECs (Supplementary Fig. 5C–G). These results suggest lipid accumulation has a direct toxic effect on TECs by inducing oxidative stress, thereby inducing iron overload and ferroptosis.

To decipher the mechanisms of Trem2⁺ macrophages in regulating lipid accumulation and ferroptosis in TECs, we treated WT BMDMs or Trem2^−/− BMDMs with HGPA, and the CM was used to treat primary TECs (Fig. 4A). TECs exposed to CM from HGPA-challenged Trem2^−/− BMDMs had significantly higher levels of intracellular lipid accumulation (Fig. 4B and C). The CM from HGPA-challenged Trem2^−/− BMDMs decreased GPX4 expression and increased ACSL4 expression in TECs (Fig. 4D). Concomitantly, the CM from HGPA-challenged Trem2^−/− BMDMs also led to a decrease in GSH in TECs (Fig. 4E). The CM from HGPA-challenged Trem2^−/− BMDMs enhanced lipid peroxidation intensity (Fig. 4F) and increased the MDA in TECs (Fig. 4G). When comparing the ratio of TEC death between different groups, it was found that the CM from HGPA-treated Trem2^−/− BMDMs significantly enhanced TEC cell death. Specifically, flow cytometry analysis using propidium iodide staining showed that the death rate of TECs treated with CM from HGPA-challenged Trem2^−/− BMDMs reached 86.65%, whereas a lower TEC death rate of 74.44% was observed in the group treated with CM from HGPA-challenged WT BMDMs. (Fig. 4H).

Figure 4.

Trem2 ablation macrophages promote lipid accumulation and ferroptosis of TECs. A: BMDMs harvested from WT mice and Trem2^−/− mice were exposed to HGPA or vehicle for 12 h, then the cultural medium was changed with serum-free medium. Twenty-four hours later, CM was harvested from BMDMs and mixed with HGPA-treated mouse primary TECs. B: Lipid accumulation in TECs stimulated CM from HGPA-challenged BMDMs. Representative photomicrographs and quantitative assessment of ORO staining. Scale bars = 20 μm. C: Statistical analysis of ORO staining. D: WBs of GPX4 and ACSL4 in primary TECs after stimulation by CM from HGPA-challenged BMDMs. E: The GSH contents in TECs lysates. F: TEC-stimulated CM and assessed peroxidation levels by flow cytometry. G: The MDA contents in TEC lysates. H: TEC-stimulated CM stained with propidium iodide and the cell death rates detected by flow cytometry. n = 3. PE-A, phycoerythrin-area. Data are presented as the mean ± SD. &P < 0.05 versus NC group; #P < 0.05, ##P < 0.01, ###P < 0.001 versus HGPA group; $$P < 0.01, $$$P < 0.001 versus WT-vehicle group; ^^P < 0.01, ^^^P < 0.001 versus Trem2^−/−-vehicle group; *P < 0.05 versus WT-HGPA group.

The mechanism by which Trem2^−/− BMDMs promoted lipid deposition and ferroptosis in TECs was explored. According to previous studies, IL-1β can induce lipid accumulation in podocytes of DN (25). Based on our data, IL-1β is the most upregulated inflammatory mediator in Trem2^−/−-macrophages compared with WT macrophages (Fig. 3G). To validate whether IL-1β derived from macrophages promoted lipid deposition in TECs, primary TECs were cotreated with HGPA and IL-1β, followed by assessments of lipid deposition and ferroptosis markers. As expected, IL-1β enhanced the lipid deposits in TECs exposed to HGPA (Supplementary Fig. 6A). IL-1β treatment downregulated GPX4 and increased ACSL4 expression (Supplementary Fig. 6B) and reduced cellular GSH level (Supplementary Fig. 6C). IL-1β also increased lipid peroxidation intensity (Supplementary Fig. 6D), elevated MDA level (Supplementary Fig. 6E), and induced significant cell death in TECs (Supplementary Fig. 6F).

To investigate the role of macrophage-derived IL-1β in TEC lipid dysregulation, primary TECs were treated with CM from HGPA-stimulated WT BMDMs or Trem2^−/− BMDMs, with or without IL-1β neutralizing antibodies, and lipid deposition and ferroptosis in TECs were then analyzed. As shown in Fig. 5, IL-1β blockade significantly reduced lipid accumulation and ferroptosis of TECs exposed to HGPA-challenged Trem2^−/− BMDMs, demonstrated by decreased ORO staining (Fig. 5A and B). IL-1β blockade also significantly increased cellular GSH (Fig. 5C) and decreased MDA (Fig. 5D) levels, and attenuated lipid ferroptosis in TECs exposed in CMs from HGPA-stimulated Trem2^−/− BMDMs, as evidenced by upregulated GPX4 expression, decreased ACSL4 expression, lipid peroxidation, and reduced cell death (Fig. 5E–G). These results indicate Trem2-deficient macrophages exacerbated TECs lipid deposition and ferroptosis in an IL-1β–dependent manner.

Figure 5.

Trem2 ablation macrophages promote lipid accumulation and ferroptosis of TECs via IL-1β. A: Lipid accumulation in TEC-stimulated CM from IL-1β–antibody-treated (IL-1β-Ab) macrophages. Representative photomicrographs and quantitative assessment of ORO staining. Scale bars = 20 μm. B: Statistical analysis of ORO staining. C: The GSH contents in TECs lysates. D: The MDA contents in TECs lysates. E: WB of GPX4 and ACSL4 in primary TECs after stimulation by CM IL-1β-Ab -treated macrophages. F: Primary TECs stimulated with CM from IL-1β–Ab-treated macrophages were stained with Liperfluo, and peroxidation levels were quantified using flow cytometry. G: Primary TECs stimulated with CM from IL-1β–Ab-treated macrophages were stained with propidium iodide, and the cell death rates were detected by flow cytometry. Data are presented as the mean ± SD, n = 3. *P < 0.05, **P < 0.01, ***P < 0.001.

To further investigate the role of the ERK signaling pathway in the expression of IL-1β during lipid deposition and ferroptosis in macrophages and tubular cells, primary TECs were exposed to CM from HGPA-stimulated WT or Trem2^−/− BMDMs pretreated with ERK inhibitor PD98059. The results showed that PD98059 suppressed IL-1β in BMDMs, especially in Trem2^−/− BMDMs. Compared with CM from Trem2^−/− BMDMs, treatment with CM from PD98059-treated Trem2^−/− BMDMs decreased lipid accumulation and MDA, whereas the GSH level in TECs was increased (Supplementary Fig. 7A–D). Compared with CM from Trem2^−/− BMDMs, treatment with CM from PD98059-treated Trem2^−/− BMDMs increased GPX4 expression, decreased ACSL4 expression, and reduced the intensity of lipid peroxidation and cell death in TECs (Supplementary Fig. 7E–G), indicating that ERK-mediated IL-1β production is critical for the interplay between macrophage and renal tubule cells.

Macrophage-Derived IL-1β Promotes CD36 Expression in Renal TECs

To elucidate the molecular mechanism underlying the tubule cell injury in the cross-talk with Trem2⁺ macrophages, we performed RNA-seq analysis on primary TECs treated with CM from HGPA-challenged Trem2^−/− BMDMs or WT BMDMs, and the DEFs were identified (q < 0.05; log₂ fold-change > 2). TECs stimulated with Trem2^−/− BMDMs CM had 1,347 upregulated genes and 2,871 downregulated genes compared with those exposed to WT BMDM CM (Fig. 6A). The downregulated genes, including GPX3, GPX6, and GPX7, are antioxidant enzymes that play important roles in combating oxidative stress and maintaining redox balance. KEGG pathway analysis enriched pathways related to fatty acid metabolism, GSH metabolism, and ferroptosis (Fig. 6B). GO term analysis demonstrated pathways related to long-chain fatty acid (LCFA) import, including genes controlling lipid uptake, such as CD36 and SPX (Fig. 6C).

Figure 6.

Macrophage-derived IL-1β promotes CD36 expression in renal TECs. RNA-seq analysis using primary TECs treated with CM from HGPA-challenged Trem2^−/− BMDMs or WT BMDMs. A: Volcano plot of the genes with significantly different levels in TECs. B: KEGG pathway analysis indicated alterations in pathways. C: GO term analysis of differentially regulated genes demonstrated alterations in pathways. D: Representative images and quantitative data of CD36 in primary TECs after stimulation by CM of BMDMs by immunofluorescent staining. Scale bar = 20 μm. E: Statistical analysis of CD36 immunofluorescence in TECs. F: WB of CD36 in primary TECs after stimulation by CM of BMDMs. G: Representative images and quantitative data of CD36 in WT and Trem2^−/− mice kidneys. Scale bars = 20 μm. H: The expression of CD36 in WT and Trem2^−/− mice kidneys measured by WB. In the cell experiment, n = 3. Data are presented as the mean ± SD. $$P < 0.01 vs. the WT-vehicle group; ^^P < 0.01, ^^^P < 0.001 vs. Trem2^−/−-vehicle group; #P < 0.05, ##P < 0.01 vs. Trem2^−/−-HGPA group; *P < 0.05, ***P < 0.001 vs. the WT-HGPA group. In the animal experiment, n = 6. Data are presented as the mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. AGE, advanced glycation end-product; ECM, extracellular matrix; IL-1β-Ab, IL-1β–antibody; RAGE, cell-bound receptor of AGE.

CD36 is the main receptor for the uptake of FFAs by several cell types. Based on the RNA-seq data and previous studies, we hypothesized that CD36 might be the key molecule mediating lipid deposition in TECs. We stimulated primary TECs with IL-1β and the results showed that CD36 expression was significantly upregulated (Supplementary Fig. 8A and B). Immunostaining revealed that TECs treated with CM from HGPA-challenged Trem2^−/− BMDMs had a higher level of CD36 expression on the cell membrane than those treated with CM from HGPA-challenged WT BMDMs (Fig. 6D and E), which was in accordance with the enhanced lipid accumulation in TECs. WB results confirmed the expression tendency of CD36 (Fig. 6F).

To further validate the role of macrophage-derived IL-1β in the CD36 expression of TECs, primary TECs were incubated with CM from HGPA-challenged WT BMDMs or Trem2^−/− BMDMs in the presence of IL-1β neutralizing antibodies. Blocking IL-1β significantly reduced CD36 expression on TECs exposed to CM from HGPA-challenged Trem2^−/− BMDMs (Fig. 6D–F). These findings indicate that Trem2^−/− BMDMs enhanced CD36 expression in TECs via IL-1β. Furthermore, compared with CM from Trem2^−/− BMDMs, treatment with CM from Trem2^−/− BMDMs pretreated with the ERK pathway antagonist PD98059 decreased CD36 expression in TECs (Supplementary Fig. 9A–C). These results suggest that Trem2-deficient macrophages regulate CD36 expression on TECs through the ERK-IL-1β axis. In the kidneys of diabetic mouse models, CD36 expression was higher in renal tubule cells of diabetic Trem2^−/− mice than of WT mice (Fig. 6G and H), consistent with the in vitro data, further demonstrating that Trem2^−/− macrophages promote CD36 expression in TECs.

CD36 Silencing Inhibited Lipotoxicity and Ferroptosis Induced by Macrophage-Released IL-1β

Recent studies showed that HFD-fed mice with chronic inflammation had increased CD36 expression and significant severe glomerular and tubular damage (26). CD36 regulates the production of reactive oxygen species (ROS) and mediates lipid homeostasis (27), but little is known about the relationship between CD36 and ferroptosis, especially in DN. To investigate the function of CD36 in lipid accumulation and ferroptosis in TECs, we silenced the expression of CD36 by transient small interfering RNA (siRNA) transfection (Fig. 7A). To investigate whether macrophage-derived IL-1β exacerbates tubular cell lipid deposition and ferroptosis in a CD36-dependent manner, CM was harvested from HGPA-challenged WT or Trem2^−/− BMDMs and used to treat TECs with CD36 silencing (Fig. 7B). CD36 deficiency significantly attenuated lipid accumulation in TECs treated with the CM of HGPA-challenged Trem2^−/− BMDMs (Fig. 7C and D). Additionally, CD36 deficiency led to significant increase in cellular GSH and decrease in MDA (Fig. 7E and F). Correspondingly, when exposed to CM from HGPA-challenged Trem2^−/− BMDMs, CD36-deficient TECs had significantly elevated GPX4 expression, along with increased ACSL4 (Fig. 7G). CD36 silencing also eliminated the increased lipid peroxidation levels (Fig. 7H) and reduced cell death in TECs treated with CM-derived Trem2^−/− BMDMs (Fig. 7I). These findings suggest lipid-induced injury of TECs is dependent on CD36 expression, and Trem2-deficient macrophages exacerbated tubular cell lipid deposition and ferroptosis in a CD36-dependent way. Furthermore, CD36 deficiency also reversed the lipid accumulation and ferroptosis in TECs induced by IL-1β (Supplementary Fig. 10A–F).

Figure 7.

CD36 silencing inhibited lipotoxicity and ferroptosis induced by macrophage-released IL-1β. A: CD36 expression was measured in siRNA-transfected TECs. B: BMDMs harvested from WT mice and Trem2^−/− mice were exposed to HGPA or vehicle, then the cultural medium was changed with serum-free medium 24 h later, CM was harvested from BMDMs and mixed with HGPA-treated CD36 siRNA-transfected TECs. C: ORO staining was used to investigate the impact of CM derived from BMDMs challenged with HGPA on lipid deposition in CD36 siRNA-transfected TECs. Scale bars = 20 μm. D: Statistical analysis of ORO staining. E and F: The levels of GSH and MDA in knocking down CD36 in TECs treated with the CM from HGPA-challenged BMDMs. G: WB analysis was conducted to evaluate GPX4 and ACSL4 levels in primary TECs that were knocked down using CD36 siRNA and subsequently stimulated with the CM from BMDM challenged with HGPA. H: CD36 siRNA knockdown TECs that were stimulated with the CM from HGPA-challenged BMDMs were stained with Liperfluo and detected by flow cytometry. I: CD36 siRNA knockdown TECs were treated with the CM from HGPA-challenged BMDMs, followed by staining with propidium iodide, and the cell death rates detected by flow cytometry. Data are presented as the mean ± SD; n = 3. *P < 0.05, **P < 0.01, ***P < 0.001. PE-A, phycoerythrin-area.

To verify that CD36 silencing inhibited IL-β-stimulated LCFA uptake, CD36-deficient TECs were exposed to the CM of HGPA-challenged Trem2^−/− BMDMs. Uptake of LCFA in CD36-deficient TECs was attenuated compared with that in normal TECs (Supplementary Fig. 10G and H), indicating that CD36 played a critical role in IL-1β–mediated LCFA uptake and lipid disorder in TECs.

NF-κB Activity Is Required for IL-1β–Induced CD36 Expression

To investigate the potential regulatory mechanism underlying IL-1β–induced CD36 expression in TECs, gene set enrichment analysis of the RNA-seq data was performed. TECs stimulated by CM from HGPA-challenged Trem2^−/− BMDMs upregulated the expression of genes related with NF-κB pathways (Fig. 8A). Moreover, NF-κB–specific DNA-binding motifs were identified in the promoter region of CD36 based on JASPAR (28) prediction (Fig. 8B), suggesting the NF-κB pathway may participate in the transcriptional regulation of CD36. Notably, knockdown of p65 using siRNA suppressed the expression of CD36 in TECs with the treatment of either IL-1β or CM from HGPA-challenged Trem2^−/− BMDMs (Fig. 8C–E and G). Moreover, treatment with the NF-κB inhibitor BAY 11-7082 also significantly inhibited IL-1β–induced CD36 protein expression in TECs (Fig. 8F). These results indicate the NF-κB pathway is critical for IL-1β–induced CD36 expression in TECs.

Figure 8.

NF-κB activity is required for IL-1β–induced CD36 expression. A: Gene set enrichment analysis showed the NF-κB translation pathways of renal tissues in TCMK-1 cells after stimulation by CM of WT or Trem2^−/− BMDMs. B: NF-κB binding motif and its binding sites in the CD36 gene promoter. C: Expression of p65 was measured in siRNA-transfected TECs. D: Effects of p65 siRNA knockdown on CD36 expression in TECs. E: IL-1β inducing of these transfected cells significantly inhibited CD36 protein levels. F: CD36 protein expression of pretreated the TCMK-1 cells with NF-κB inhibitors (BAY 11-7082) before inducing IL-1β. G: The p65-deficient TEC CD36 expression when exposed to CM from HGPA-challenged Trem2^−/− BMDMs. Data are presented as the mean ± SD; n = 3. **P < 0.01, ***P < 0.001. BAY, BAY 11-7082; Ctr, control; ES, enrichment score.

Discussion

The pathogenesis of DN is associated with glucose and lipid metabolism disorders and oxidative stress (29,30). Lipid deposition induced kidney injuries due to lipotoxicity, ROS, and ferroptosis in the renal cells of DN (7,12,31,32). With the metabolic disorders and the onset of injury and cell death, a significant infiltration of immune cells occurred, leading to the exacerbation of renal tubular damage. Previous studies have shown the presence of macrophages in both the interstitial areas and glomeruli of patients at different stages of DN (13). Trem2⁺ macrophages have recently been identified in the kidneys of both mice and humans with diabetes, indicating the possible correlation of Trem2⁺ macrophages with disease progress (4,33). Trem2 is recognized as a significant immune signaling hub that detects tissue damage and triggers a vigorous immune remodeling process as an adaptive reaction to injury (34,35). Studies of liver diseases revealed that Trem2 KO exacerbates lipid accumulation in nonalcoholic steatosis hepatitis (36); in contrast, high expression of Trem2 in macrophages may protect the liver from oxidative stress (37). Notably, our research uncovered the conserved function of Trem2⁺ macrophages on metabolic dysfunction and related cell injury in the kidney of DN. We found that Trem2 deficiency exacerbated lipid accumulation both in serum and in the kidney of diabetic mice, and the underlying mechanisms were further deciphered.

The activation of macrophages in response to elevated glucose and lipid levels plays a critical role in the interplay between inflammation and renal epithelial cells, mediated by a diverse array of cytokines and chemokines (38). IL-1β is an important inflammatory cytokine involved in kidney injury and repair (39). Previous studies showed that IL-1β–induced lipid accumulation increased the accumulation of intracellular cholesterol in podocytes and mesangial cells (7,40). Both genetic and pharmacological inhibition of IL-1β protected against podocyte damage in DN by hindering lipid buildup and ROS production (32). Here, we demonstrated that macrophages from diabetic Trem2^−/− mice increased production of proinflammatory mediators IL-1β by activating ERK signaling pathways. We showed that in addition to the direct effect on activating inflammation, IL-1β released by Trem2 macrophages regulated renal tubular lipid uptake by enhancing the expression of CD36 in TECs. The upregulated CD36 subsequently led to increased lipid accumulation and ferroptosis in renal tubular cells, which is consistent with the notion that lipids accumulate in the kidney and are associated with tubulointerstitial injury (39).

Lipotoxicity caused damage to renal cells by triggering inflammation, producing ROS, and leading to ferroptosis (12,31,32). Ferroptosis has been highlighted as playing an essential role in the onset and advancement of DN, which is primarily driven by lipid peroxidation (41). CD36, as scavenger receptor, is a major receptor mediating FFA uptake and is involved in lipid metabolism (42). Previous studies have reported the correlation of CD36 expression with CKD, especially DN kidney (7,43), and the contribution of CD36 to renal tubular damage (44,45). Studies also showed that inflammatory cytokines enhanced the translational efficiency of hepatic CD36 through the activation of the mTOR signaling pathway in nonalcoholic fatty liver disease (46). In this study, we revealed that IL-1β enhanced the expression of CD36 via the transcription factor NF-κB in TECs, promoting FFA transport and inducing lipid accumulation. Inhibition of NF-κB blocked IL-1β–induced CD36 production.

Cell clearance supports resolving inflammation and triggering tissue remodeling needed to restore homeostasis (47,48). The phagocytic process occurring upon engagement of Trem2, similar to what is described for the uptake of apoptotic neurons (49) and bacterial products (50), leads to the elimination of the pathogen and of potentially harmful dying cells. Furthermore, Trem2 also enhanced microglial phagocytosis and decreased the transcription of cytokines such as IL-1β, IL-6, and TNF-α in vivo (51). Consistent with this, our transcriptome data identified phagocytosis-regulating genes. Trem2-deficient macrophages exhibit reduced ferroptotic cell phagocytosis, which may drive DN progression.

In summary, our study shows elevated Trem2 in human and mouse DN kidneys correlates with DN progression. Trem2 KO macrophages activate ERK phosphorylation, increasing IL-1β secretion. This excess IL-1β activates tubular NF-κB, promoting CD36-dependent lipid uptake and ferroptosis. Trem2-deficient macrophages also exhibit reduced phagocytosis of ferroptotic cells in diseased kidneys. Targeting Trem2 could both dampen IL-1β–driven inflammation and enhance macrophage phagocytosis. Strategies like Trem2 agonists or macrophage-specific overexpression via gene therapy may mitigate renal lipid accumulation and ferroptosis in DN. Our findings identify Trem2⁺ macrophages as key DN regulators and propose Trem2 as a promising therapeutic target.

This study has limitations. First, our study establishes a correlative rather than direct mechanistic link between Trem2 and ERK inhibition; precise mechanisms remain unclear. Second, tubular cell–secreted chemoattractants for Trem2 macrophages in DN and Trem2’s specific ligand need identification. Third, Trem2 pathway activation via agonists or antibodies is required to explore its therapeutic potential in halting DN progression.

This article contains supplementary material online at https://doi.org/10.2337/figshare.29954540.

Funding Statement

This work was supported by the National Natural Science Foundation of China (grants 82030025 and 32141005) and the Beijing Natural Science Foundation (grant L232122).