RXRs control serous macrophage neonatal expansion and identity and contribute to ovarian cancer progression

Abstract

Tissue-resident macrophages (TRMs) populate all tissues and play key roles in homeostasis, immunity and repair. TRMs express a molecular program that is mostly shaped by tissue cues. However, TRM identity and the mechanisms that maintain TRMs in tissues remain poorly understood. We recently found that serous-cavity TRMs (LPMs) are highly enriched in RXR transcripts and RXR-response elements. Here, we show that RXRs control mouse serous-macrophage identity by regulating chromatin accessibility and the transcriptional regulation of canonical macrophage genes. RXR deficiency impairs neonatal expansion of the LPM pool and reduces the survival of adult LPMs through excess lipid accumulation. We also find that peritoneal LPMs infiltrate early ovarian tumours and that RXR deletion diminishes LPM accumulation in tumours and strongly reduces ovarian tumour progression in mice. Our study reveals that RXR signalling controls the maintenance of the serous macrophage pool and that targeting peritoneal LPMs may improve ovarian cancer outcomes.

Fig. 1. RXR deficiency alters serous cavity macrophage populations.

a–b Flow cytometry of serous cavities from 7 to 9-week-old LysM^Cre+Rxra^fl/fl mice and Rxra^fl/fl littermates. a Representative flow cytometry plots show the frequencies of LPMs (F4/80^HIMHCII^LO) and SPMs (F4/80^LOMHCII^HI) pregated on CD45⁺B220^-CD11b⁺CD115⁺ cells (see also Supplementary Fig. 1a). b Graphs show frequencies among CD45⁺ leukocytes and absolute numbers. n = 3–5 from at least two independent experiments. c Immunofluorescence images showing F4/80 and DAPI staining of peritoneal lavage cytospins from LysM^Cre+Rxra^fl/fl and Rxra^fl/fl mice. Scale bar: 20 μm. d–e Flow cytometry of serous cavities from 9-week-old LysM^Cre+Rxrab^fl/fl mice and Rxrab^fl/fl littermates (see also Supplementary Fig. 1h). d Representative flow cytometry plots show the frequencies of LPMs and SPMs pregated on CD45⁺B220^-CD11b⁺CD115⁺ cells (left) and the percentages of TIM4⁺ and TIM4⁻ cells pregated on LPMs (right). e Graphs show frequencies among CD45⁺ leukocytes and absolute numbers. Data (n = 5–7 per genotype) are representative of two independent experiments; ^#p ≤ 0.05 (unpaired Student’s t test) vs total macrophage percentage or absolute numbers in Rxrab^fl/fl mice; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001 (unpaired Student’s t test) vs the same population in Rxrab^fl/fl mice. f Annotated t-SNE plots in the identified populations among CD45⁺ peritoneal cells from LysM^Cre+Rxrab^fl/fl and Rxrab^fl/fl mice as in (d) (left) and overlaid with biexponential transformed marker expression levels (right). All data are presented as mean ± SEM; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001 (unpaired Student’s t test). Source data are provided as a Source Data file.

Fig. 2. RXRs control the identity of LPMs.

a Volcano plot showing the global transcriptional changes in LysM^Cre+Rxra^fl/fl vs Rxra^fl/fl LPMs determined by RNA-seq. Each circle represents one DEG and coloured circles represent DEGs significantly upregulated (Benjamini–Hochberg adjusted p value ≤0.05 and Log fold change (FC) ≥1.5 (in red)) or significantly downregulated (Benjamini–Hochberg adjusted p value ≤0.05 and Log FC ≤ 1.5 (in blue)). Normalized expression values from RNA-seq data are provided in Supplementary Tables 1 and 2. b Gene set enrichment analysis (GSEA) of RNA-seq data showing downregulated and upregulated functions in LysM^Cre+Rxra^fl/fl vs Rxra^fl/fl LPMs. NES Normalized enrichment score; FDR false-discovery rate. c Genomic distribution of enriched regions in LysM^Cre+Rxrab^fl/fl and Rxrab^fl/fl LPMs, identified in the ATAC-seq data set. d Scatter plot comparing accessibility to Tn5 transposase for differentially accessible peaks in LysM^Cre+Rxrab^fl/fl and Rxrab^fl/fl LPMs (y axis, logFC in normalized read counts) and mRNA expression changes in LysM^Cre+Rxra^fl/fl and Rxra^fl/fl LPMs for the nearest gene (x axis; logFC values). Grey dots represent the association between differentially accessible regions and the nearest differentially expressed genes, as detected by HOMER. Peaks associated with DEGs related to SPMs are highlighted in red (upregulated genes), and those related to the LPM-specific signature are highlighted in blue (downregulated genes). Chi-squared and Pearson correlation tests, R² = 0.325, Chi-square = 202.99, p value <10⁻⁶. e Genome browser views of proliferation-related (Cdca2 and Cenpe), apoptosis-related (Naip1 and Capn2), LPM-specific (Gata6 and Thbs1) and SPM-specific (Retnla and CD209d) gene bodies in LysM^Cre+Rxrab^fl/fl and Rxrab^fl/fl LPM ATAC-seq data set. H3K4me2 and H3K27ac-marked regions previously defined by Gosselin et al. in LPMs from wild-type C57BL6/J mice are included as grey-to-black bars (tone intensity indicates read length). Vertical highlights correspond to regions of interest for the specified loci. f HOMER known motif analysis of Rxrab^fl/fl- and LysM^Cre+Rxrab^fl/fl-specific ATAC-seq peak sequences in LPMs. Top table shows transcription factor motifs enriched in Rxrab^fl/fl LPMs using a background corresponding to LysM^Cre+Rxrab^fl/fl LPM peaks. Bottom table shows transcription factor motifs enriched in LysM^Cre+Rxrab^fl/fl LPMs using a background corresponding to Rxrab^fl/fl LPM peaks. Percentages of Rxrab^fl/fl and LysM^Cre+Rxrab^fl/fl peaks relative to levels in their respective backgrounds are shown.

Fig. 3. RXRs are required for the neonatal expansion of peritoneal LPMs.

a–b Frequency among CD45⁺ cells and absolute numbers of TIM4⁺ LPMs (a) and SPMs (b) from peritoneal exudates of LysM^Cre+Rxrab^fl/fl mice (purple) and Rxrab^fl/fl mice (black) from the day of birth (0) through postnatal day 70. n = 3–17 per genotype and age, pooled from up to three independent experiments per age. c Representative dot plots showing Ki-67 and DAPI staining gated on peritoneal TIM4⁺ LPMs, with quantification showing frequencies of proliferating (G₂/M/S) TIM4⁺ LPMs over time. n = 4–9 per genotype and age, pooled from one to two independent experiments per age. d Flow cytometry density plots showing BrdU incorporation by TIM4⁺ LPMs after treatment and quantification showing frequencies of TIM4⁺ LPMs with BrdU incorporation. n = 3–12 per genotype and age, pooled from up to four independent experiments per age. e mRNA expression of cell-cycle-related genes in peritoneal LPMs from DAB1 LysM^Cre+Rxrab^fl/fl mice. Gene expression is normalized to DAB1 Rxrab^fl/fl LPMs (dashed line). n = 5 per genotype and gene. All data are presented as mean ± SEM; *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 and ****p ≤ 0.001 compared with age-paired Rxrab^fl/fl mice; ^#p ≤ 0.05, ^##p ≤ 0.01, ^###p ≤ 0.001 and ^####p ≤ 0.001; (a–d) two-way ANOVA followed by Tukey's multiple comparisons test; and (e) unpaired Student’s t test. DAB: day after birth. Source data are provided as a Source Data file.

Fig. 4. RXR deficiency leads to lipid accumulation and apoptosis of peritoneal TIM4⁺ LPMs in adult mice.

a Representative May–Grünwald–Giemsa cytospins of sorted TIM4⁺ LPMs (left), and Oil Red O stained TIM4⁺ LPMs (right) from 9-week-old LysM^Cre+Rxrab^fl/fl and Rxrab^fl/fl mice. Scale bars: 20 μm (May–Grünwald–Giemsa) and 10 μm (Oil Red O). b Heatmap showing normalized log2 FC in the expression of protein transport-related genes in adult LysM^Cre+Rxra^fl/fl and Rxra^fl/fl LPMs (left), and qPCR analysis of LysM^Cre+Rxrab^fl/fl and Rxrab^fl/fl LPMs (right). n = 4–6 per genotype and gene. c Confocal images showing overlaid channels for LysoTracker (red) and BODIPY493/503 (green) from sorted and cultured TIM4⁺ LPMs from adult LysM^Cre+Rxrab^fl/fl and Rxrab^fl/fl mice (see also Supplementary Fig. 4f). Panels show a representative image of Rxrab^fl/fl LPMs and three representative fields of view of LysM^Cre+Rxrab^fl/fl TIM4⁺ LPMs: (1) cells with non-acidic lipid vesicles; (2) cells with lipid and lysosome markers overlapped (yellow arrowhead); and (3) cells with a central lipid core surrounded by an acidic ring (yellow arrowheads). Scale bar = 10 μm. d Flow cytometry dot plots showing BODIPY493/503 and LysoTracker (LT) staining gated on TIM4⁺ LPMs (left), and the percentage of double-positive LT^HIBODIPY⁺ TIM4⁺ LPMs from LysM^Cre+Rxrab^fl/fl and Rxrab^fl/fl mice (right). n = 4–8 per genotype and age, pooled from one to two independent experiments per age. e Flow cytometry dot plots showing Annexin-V staining gated on TIM4⁺ LPMs (left), and frequency quantification of apoptotic TIM4⁺ LPMs (right) in LysM^Cre+Rxrab^fl/fl and Rxrab^fl/fl mice. n = 3–9 per genotype and age, pooled from up to three independent experiments per age. f Flow cytometry dot plots showing Annexin-V and BODIPY493/503 staining gated on TIM4⁺ LPMs (left), and frequency quantification of lipid-loaded and lipid-free apoptotic TIM4⁺ LPMs (right) in adult LysM^Cre+Rxrab^fl/fl and Rxrab^fl/fl mice. n = 3–4 per genotype, representative of four independent experiments. g Heatmap showing normalized log2 FC in the expression of apoptosis-associated genes in adult LysM^Cre+Rxra^fl/fl and Rxra^fl/fl LPMs (left), and qPCR analysis of LysM^Cre+Rxrab^fl/fl and Rxrab^fl/fl LPMs (right). n = 3–4 per genotype and gene. qPCR data are presented as gene expression in LysM^Cre+Rxrab^fl/fl normalized to Rxrab^fl/fl LPMs (dashed line). All data are presented as mean ± SEM; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001; (d and e) two-way ANOVA and (b, f and g) unpaired Student’s t test. Source data are provided as a Source Data file.

Fig. 5. Peritoneal macrophages invade early ovarian tumour lesions and promote ovarian cancer progression.

a Experimental design of the orthotopic model for Upk10 ovarian tumours. i.p. intraperitoneal; i.b. intrabursal; PMS pregnant mare serum gonadotropin; hCG human chorionic gonadotropin; Upk10 mouse ovarian tumour cell line. b Representative image of ovarian tumours (left) and quantification of primary ovarian tumour size and weight (right) in Rxrab^fl/fl and LysM^Cre+Rxrab^fl/fl mice. Data are presented as median (lower-upper quartiles and minimum and maximum values). n = 37–39 per genotype, pooled from seven independent experiments. c Hematoxylin and eosin staining of naïve and tumour-bearing ovaries in the same mice as in (b). Scale bar = 1 mm. Dotted lines delimit the tumour border. d Flow cytometry analysis of GATA-6⁺F4/80⁺ macrophages (red box gate) in primary ovarian tumours (top). Frequency of GATA-6⁺F4/80⁺ macrophages gated as CD45⁺Upk10^-B220^-Ly6G^-CD3^-CD11b⁺ cells (bottom, see also Supplementary Fig. 6d). Negative control shows non-specific staining with the secondary antibody for GATA-6. Data are presented as median (lower-upper quartiles and min-max values). n = 18–27 per genotype, pooled from four independent experiments. e Immunofluorescence staining for GATA-6 (red), F4/80 (green) and DAPI (blue) on frozen tumour sections from the same mice as in (b). Scale bars: 100 and 10 μm for insets (white squares). Data are representative of four animals per genotype. f Quantification of GATA-6⁺ F4/80⁺ cells in the images in (e). Data are presented as mean ± SEM. n = 4 per genotype. *p ≤ 0.05; ****p ≤ 0.0001; (b) Mann–Whitney U test, (d) generalised lineal model with gamma distribution and (f) unpaired Student’s t test. Source data are provided as a Source Data file.