Lactation-associated macrophages exist in murine mammary tissue and human milk

Abstract

Macrophages are involved in immune defense, organogenesis and tissue homeostasis. Macrophages contribute to the different phases of mammary gland remodeling during development, pregnancy and involution postlactation. Less is known about the dynamics of mammary gland macrophages in the lactation stage. Here, we describe a macrophage population present during lactation in mice. By multiparameter flow cytometry and single-cell RNA sequencing, we identified a lactation-induced CD11c⁺CX3CR1⁺Dectin-1⁺ macrophage population (liMac) that was distinct from the two resident F4/80^hi and F4/80^lo macrophage subsets present pregestationally. LiMacs were predominantly monocyte-derived and expanded by proliferation in situ concomitant with nursing. LiMacs developed independently of IL-34, but required CSF-1 signaling and were partly microbiota-dependent. Locally, they resided adjacent to the basal cells of the alveoli and extravasated into the milk. We found several macrophage subsets in human milk that resembled liMacs. Collectively, these findings reveal the emergence of unique macrophages in the mammary gland and milk during lactation.

Fig. 1. Lactating murine mammary gland contains CD11c^hi macrophages.

a,b, UMAP plots and corresponding heatmap (a) and representative flow cytometry plots, frequency and total cell counts (b) of the myeloid compartment (pregated on CD45⁺CD11b⁺ and/or CD11c⁺ cells) in the mammary glands of virgin or lactating (day 10 pp) (a) or virgin (day −20 pp), lactating (days 1–15 pp) or postlactating (day 21 pp) (b) wild-type mice, analyzed by flow cytometry. Heatmap shows the mean marker expression level. Data were transformed and percentile normalized, n = 3 per group (a). Data (n = 5–6 per timepoint) were pooled from six independent experiments (b). Kruskal–Wallis test was corrected with Dunn’s multiple comparisons test, *P < 0.05; **P < 0.01; ****P < 0.0001. Mac, macrophages, Mo, monocytes. c, Heatmap of the fold change in total cell counts of the different myeloid cell populations (as in a) in lactating mammary glands at early lactation (days 1–6 pp), late lactation (day 8–15 pp), postlactation (day 21 pp) compared with nonlactating mammary glands of wild-type mice. d, Immunohistochemistry of the virgin and lactating (day 8 pp) mammary glands from Cx3cr1^GFP/+ mice. SMA (blue), Cx3cr1-GFP (green), CD11c (magenta). Insets, magnifications of outlined regions showing SMA, and CD11c or Cx3cr1-GFP. Images are representative of n = 2 mice. Scale bar, 75 µm.

Fig. 2. Murine liMacs exhibit a unique transcriptional signature.

a,b, Seurat guided clustering and dimensionality reduction by UMAP showing eight and nine distinct myeloid cell populations (a) and Venn diagram of genes expressed in F4/80^lo macrophages, F4/80^hi macrophages and liMacs (b) in scRNA-seq performed on CD11b⁺ and/or CD11c⁺ myeloid cells sorted from virgin (n = 1) and lactating (day 7 pp, n = 2) mammary glands from wild-type mice. c, Dot plot of top 100 liMac signature genes with significantly higher expression in liMacs compared with the other defined immune populations, plotted versus F4/80^hi and F4/80^lo macrophages (as in a). Dot size represents percentage of cells in a cluster expressing each gene; dot color reflects expression level (as indicated on legend). d,e, Seurat guided clustering and dimensionality reduction by UMAP showing liMac and proliferating liMacs (liMac^prolif) in scRNA-seq performed on sorted (CD45⁺Ly6G^–Siglec-F^–NK1.1^–) CD11b⁺ and/or CD11c⁺ cells (d) and heatmap showing genes associated to liMac and liMac^prolif (e) in lactating mammary glands at day 8 (n = 3), day 11 (n = 3), and day 14 pp (n = 3) of wild-type mice.

Fig. 3. Murine liMacs are derived predominantly from monocytes and expand during lactation.

a, Representative flow cytometry histograms and graph (±s.d.) showing the percentage of tdTomato labeling in F4/80^lo macrophages (F4/80^lo Mac), F4/80^hi macrophages (F4/80^hi Mac), liMacs (pregated on CD45⁺SiglecF^–Ly6C^–Ly6G^–) and Ly6C^hi monocytes (Mo) (pregated on CD45⁺SiglecF^–Ly6G^– cells) in mammary glands from lactating (day 10 pp) Ms4a3^CreR26^Ai14 mice. Pooled data from three independent experiments, n = 9. b, Immunohistochemistry of lactating mammary glands from Ms4a3^CreR26^Ai14 mice (day 10 pp). DAPI (blue), SMA (green), Dectin-1 (magenta), tdTomato (red). Right panels show DAPI and single stains for Dectin-1 and tdTomato, respectively, of the outlined region in the overview image on the left. Scale bar, 70 µm; n = 4. c, Representative flow cytometry histograms and graph (±s.d.) showing the percentage of tdTomato labeling in F4/80^lo Macs, F4/80^hi Macs, liMacs and Ly6C^hi Mo as in a in the mammary glands of lactating (day 8–15 pp) Ccr2^CreERR26^Ai14 mice treated with tamoxifen on day 1 and day 3 pp. Pooled data of three experiments, n = 6 (n = 4 from 8 days pp and n = 2 from 15 days pp). d, Representative immunofluorescence image of lactating mammary glands (day 15 pp) from Ccr2^CreERR26^Ai14 dams treated with tamoxifen on day 1 and day 3 pp. DAPI (blue), SMA (green), Iba1 (yellow) and tdTomato (red). Middle panels (1) and bottom panels right (2) show DAPI and single stains for Iba1 and tdTomato of the outlined region in the overview image on the left (1) or of the image on top (2), respectively. Scale bars, 70 µm for (1), 35 µm for (2); n = 6. e, Representative flow cytometry plots and violin plots show the frequency and total cell counts of monocytes, F480^hi Macs, F4/80^lo Macs and liMacs as in a in the mammary glands from lactating (days 13−15 pp) Ccr2^–/– and wild-type dams. Pooled data from five independent experiments; n = 8 (WT) and 7 (Ccr2^−/−). Two-tailed Mann–Whitney test was performed, **P < 0.01; ***P < 0.001; NS, not significant. f, Representative flow cytometry plots and a graph (±s.e.m.) showing the percentage of EdU⁺ F4/80^lo macrophages, F4/80^hi macrophages and liMacs in mammary glands from lactating (day 7 pp) wild-type mice 20 h post EdU injection. Pooled data from three independent experiments, n = 4. Kruskal–Wallis test followed by Dunnʼs multiple comparison test. *P < 0.05.

Fig. 4. CSF-1 and the microbiota regulate the development of murine liMacs in the lactating mammary gland.

a, Representative flow cytometry plots and violin plots showing the frequency and total cell counts of F4/80^hi macrophages (Mac), F4/80^lo Macs and liMacs in lactating mammary glands (days 8–14 pp) from control (WT or Csf1r^fl/fl) and Cd11c^CreCsf1r^fl/fl dams. Pooled data from five to six mice analyzed in four independent experiments. Two-tailed Mann–Whitney test was performed. **P < 0.01. b, Representative flow cytometry plots and violin plots showing the frequency and total cell counts of F4/80^hi Macs, F4/80^lo Macs and liMacs (per 10⁶ total cells) in lactating mammary glands (day 9 pp) from Il34^LacZ/LacZ and control (Il34^+/+ or Il34^LacZ/+) dams. Pooled data of five mice analyzed in four independent experiments. Two-tailed Mann–Whitney test was performed. c, Representative flow cytometry plots and violin plots showing the frequency and total cell counts of F4/80^hi Macs, F4/80^lo Macs and liMacs in lactating mammary glands (days 7–14 pp) from dams treated with CSF-1 antibodies (Ab) (two to three times) or control dams (treated with isotype antibody or left untreated). Pooled data from 11 mice analyzed in five independent experiments. Two-tailed Mann–Whitney test was performed. *P < 0.05, ***P < 0.001, ****P < 0.0001. d, Representative flow cytometry plots and violin plots showing the frequency and total cell counts of F4/80^hi Macs, F4/80^lo Macs and liMacs in lactating mammary glands (day 7 pp) of conventional (CV), Cuatro (Cu) and germ-free (GF) dams. Pooled data from three independent experiments are shown; n = 4–7. Kruskal–Wallis test followed by Dunn’s multiple comparison test. **P < 0.01.

Fig. 5. LiMacs are present in murine milk and contribute to neutrophil recruitment in mastitis.

a, Representative immunofluorescence image of a lactating mammary gland (day 15 pp). DAPI (blue), SMA (green) and Iba1 (yellow). Arrowhead shows an Iba1⁺ cell inside an alveoli. Scale bar, 70 µm. b,c, Representative flow cytometry plots and gating strategy (b) and violin plots showing frequencies of eosinophils, neutrophils, Ly6C^hi monocytes and liMacs within CD45⁺ cells (c) in lactating mammary glands (days 7–11 and 12–15 pp) from wild-type mice; n = 3 for days 7–11 pp, and n = 6 for days 12–15 pp, data pooled from two independent experiments. Two-tailed Mann–Whitney test was performed. d, Survival curve and weight of WT pups nursed by WT dams treated with CSF-1 antibodies or isotype control at E18.5 and every other day pp; n = 15, pooled data from three independent experiments. e, Survival curve and weight of pups (Cd11c^CreCsf1r^fl/+ or Csf1r^fl/+) nursed by CD11c^CreCsf1r^fl/fl or Csf1r^fl/fl mice; n = 30, pooled data from four experiments. f, H&E staining of mammary gland sections from unchallenged (control) WT mice or WT mice injected with LPS into the mammary gland (18 h post LPS challenge). Scale bar, 100 µm; n = 2–3 mice per group. g, UMAP plots and bar graph showing the frequencies of eosinophils, F4/80^hi macrophages (Mac), F4/80^lo macrophages, liMacs, Ly6C^hi monocytes (Mo), Ly6C^lo monocytes, neutrophils and CD11b^intMHCII⁺ cells (pregated on CD45⁺CD3^–CD19^–NK1.1^– and CD11b⁺ and/or CD11c⁺ and/or F4/80⁺ cells) in lactating mammary glands (days 11–14 pp) of WT mice, injected with LPS into the fourth mammary glands 18 h earlier or left untreated (control). The corresponding heatmap shows the median marker expression level. Data were transformed and quantile normalized, n = 2–3 per group. h, Representative flow cytometry plots and violin plots showing total cell counts of liMacs (top) (pregated on CD45⁺CD19^–NK1.1^–CD3^–SiglecF^–Ly6G^–) and neutrophils (bottom) (pregated on CD45⁺LiMacs^–) in lactating mammary glands (days 10–14 pp) in WT mice left untreated or pretreated with CSF-1 antibodies on days 7–11 pp, and injected with LPS into the fourth mammary glands 18 h earlier or left uninjected. Data (n = 5–6 per group) were pooled from two independent experiments. One-way ANOVA test was applied, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 6. Human milk contains several subsets of macrophages.

a,b, UMAP and corresponding heatmap of mean marker expression (a) showing immune cells in human milk (collected on days 4–63 pp and pregated on CD45⁺ cells) and graphs showing percentage of immune cell populations among CD45⁺ cells in human milk (b) (n = 13), analyzed by flow cytometry. Representative UMAP is shown from one of four independent experiments, n = 4. Two-tailed unpaired Student’s t test was performed. ****P < 0.0001. c, Seurat guided clustering and dimensionality reduction by UMAP showing seven immune cell populations including B cells, T cells, three populations of macrophages (Mac 1–3), monocytes (Mo), DC2/DC3 and unidentified clusters termed ‘Rest’ in scRNA-seq analysis of immune cells (CD45⁺) sorted from human milk samples (n = 6). d, Dot plot showing expression of selected genes in the three human macrophage subsets, monocytes and DC2/DC3 as in c. Dot size represents percentage of cells in a cluster expressing each gene; dot color reflects expression level. e, Heatmap showing the expression of liMac signature genes (as in Fig. 2c) by human milk macrophages (Mac 1, Mac 2 and Mac 3) and T cells as in c. f, Prediction similarity scores of mouse mammary gland macrophages applied to human milk macrophage subsets as in c. Each dot represents a cell. g, UMAP of the MoMac-VERSE compendium with annotated clusters (left) and Mac 1, Mac 2, Mac 3, monocytes (Mo) and DC2/DC3 (as in c) projected on the MoMac-VERSE (right). h, UMAP and frequencies of the MoMac-VERSE populations projected on human milk Mac 1, Mac 2, Mac 3, monocytes (Mo) and DC2/DC3 (as in c). i, Dot plot showing expression of selected ‘TREM2 signature’ genes^,, in the human macrophage clusters Mac 1, Mac 2, Mac 3 as in c. Dot size represents percentage of cells in a cluster expressing each gene; dot color reflects expression level (as indicated on legend).

Extended Data Fig. 1. Gating strategy and population dynamics of myeloid cells in the murine mammary gland during lactation.

a-c, Representative flow cytometry plots for the gating strategy of eosinophils, neutrophils, Ly6C^hi monocytes, F4/80^lo macrophages (Mac), F4/80^hi macrophages, liMacs and dendritic cells (DC) (pre-gated on live, singlets and CD45⁺ cells) (a), violin plots of total cell counts (b) and bar graph with the percentages of total myeloid cells (c) in mammary glands from virgin, lactating (day 10 pp) and post-lactating (day 21 pp) wild-type mice. Data (n = 5-6 per timepoint) are pooled from 6 independent experiments. Kruskal-Wallis test was corrected with Dunn’s multiple comparisons test, *p < 0.05; **p < 0.01; ***p < 0.001. d, UMAP plot (pre-gated on CD45⁺CD11b⁺ and/or CD11c⁺) showing the myeloid compartment of the mammary glands in late gestation (at E18.5), analyzed by flow cytometry. Heatmap shows the mean marker expression. Data were transformed and percentile normalized. Related to Fig. 1.

Extended Data Fig. 2. LiMacs are localized at the site of milk production in mice.

a-b, Immunofluorescence images of lactating mammary glands from Cx3cr1^GFP/+ dams (day 8 pp) showing SMA (blue), CD11c (magenta), Cx3cr1-GFP (green) and MHCII (yellow) (a) or Lyve1 (b) (yellow). Insets are magnifications of outlined regions showing single stainings for Cx3cr1-GFP, CD11c, MHCII (a) or Cx3cr1-GFP, CD11c, Lyve1 (b). Images are representative of n = 2 mice. Scale bar: 75 µm for low magnification, 10 µm (a) or 38 µm (b) for insets. c-d, Immunofluorescence images of lactating mammary glands from Cx3cr1^GFP/+ dams (day 9–12 pp) showing DAPI (blue), SMA (red), Cx3cr1-GFP (green), and CD64 (cyan) (c) or Iba1 (cyan) (d). Insets are magnifications of outlined regions showing single stainings for Cx3cr1-GFP and CD64 (c) and Iba1 (d). Scale bar: 100 µm for low magnification, 50 µm for insets. Images are representative of n = 4 mice. Related to Fig. 1.

Extended Data Fig. 3. Genes highly expressed in liMacs.

a-e, scRNA-seq analysis of myeloid cells (CD11b⁺ and/or CD11c⁺) of lactating mammary glands derived from WT mice at 7 dpp. a, Heatmap of cluster-defining genes. Related to Fig. 2a. b, Dot plot showing expression of selected ‘Folr2 signature’ genes in liMacs, F4/80^hi macrophages (Mac) and F4/80^lo macrophages. Dot size represents percentage of cells in a cluster expressing each gene; dot color reflects expression level (as indicated in the legend). c, Heatmap showing expression of macrophage and DC signature genes^, in liMacs, F4/80^lo macrophages, F4/80^hi macrophages, cDC2s and cDC1s. d, Dot plot showing expression of selected ‘Cadm1/Trem2 signature’ genes in liMacs, F4/80^hi macrophages and F4/80^lo macrophages. Dot size represents percentage of cells in a cluster expressing each gene; dot color reflects expression level (as indicated in the legend). e, Enrichment of GO terms according to the gene expression pattern in liMacs, F4/80^hi macrophages and F4/80^lo macrophages of lactating mammary glands. f, Immunofluorescence image of lactating mammary glands from Cx3cr1^GFP/+ dams (day 8 pp) showing DAPI (blue), SMA (white), Cx3cr1-GFP (green) and Dectin-1 (red) (left) and single stainings for Cx3cr1-GFP and Dectin-1 (right). Images are representative of n = 2 mice. Scale bar: 100 µm.

Extended Data Fig. 4. LiMacs proliferate during the lactation period in mice.

a, UMAP showing proliferating cells (in red) of scRNA-seq data of myeloid cells (CD11b⁺ and/or CD11c⁺) and dot plot showing expression of proliferation-associated genes in lactating mammary glands (at day 7 pp) of wild-type mice. Dot size indicates percentage of cells expressing the gene; color indicates the average gene expression level per cell. Related to Fig. 2. b, Polar volcano plot showing differentially expressed genes between liMacs at days 8, 11 and 14 pp, analyzed by the limma voom method. Coordinates correspond to relative mean expression values per group. 86 genes out of 8052 resulted differentially expressed, of which 46 were specifically upregulated in day 8 pp (adjusted p values < 0.01). 30 genes with top p values are labeled. Related to Fig. 2. c, Representative immunofluorescence image and a bar graph (±SEM) showing percentage of Ki67⁺ liMacs in lactating mammary glands from Cx3cr1^GFP/+ dams (day 8–10 pp). Ki67 (magenta), SMA (yellow), Cx3cr1-GFP (green) and DAPI (blue). Insets are magnifications showing single stainings of the outlined regions. N = 4 mice. Scale bar: 75 µm, inset 18.75 µm. d, Violin plots showing total cell counts of Ly6C^hi monocytes, neutrophils and eosinophils in lactating mammary glands (day 8–14 pp) of Csf1r^fl/fl and Cd11c^CreCsf1r^fl/fl dams. Pooled data from 5-6 mice analyzed in 4 independent experiments. Two-tailed Mann-Whitney test was performed, ns = not significant. Related to Fig. 4. e, Violin plots showing total cell counts of Ly6C^hi monocytes, neutrophils, eosinophils, B cells and DCs in lactating mammary glands (day 7–14 pp) of control (isotype control antibody-treated or untreated) or CSF-1 antibody (Ab)-treated (2-3 times) from wild-type dams. Pooled data from 11 mice analyzed in 5 independent experiments. Two-tailed Mann-Whitney test was performed, ns = not significant. Related to Fig. 4.

Extended Data Fig. 5. Absence of liMacs does not alter protein composition of murine milk.

a, Hematoxylin and eosin staining of mammary gland sections of lactating mammary glands (day 8 pp) from Csf1r^fl/fl and Cd11c^CreCsf1r^fl/fl dams. The number of alveoli and the alveolar area were quantified using ilastik software. Images are representative of n = 4 and 5 mice, respectively. Scale bar: 200 µm. Two-tailed unpaired Student’s t test was performed, ns = not significant. Related to Fig. 5. b, Heatmap showing the 50 most abundant proteins in milk derived from Csf1r^fl/fl and Cd11c^CreCsf1r^fl/fl dams at day 7 pp. N = 13-14, pooled data from 3 individual experiments, indicated with a gray line. c, Violin plots showing concentrations of IgA, IgM, IgG1, IgG2a, IgG2b, IgG3 in mouse milk collected at day 7 pp from lactating Csf1r^fl/fl and Cd11c^CreCsf1r^fl/fl dams. N = 9 and 7 mice, respectively. Two-tailed unpaired Student’s t test was performed, ns = not significant. d, Representative immunofluorescence images of lactating mammary glands (day 11–14 pp) of control (untreated) or CSF-1 antibody-treated (on day 8–11 pp) dams showing DAPI (blue), IgA (green) and CD138 (red). Insets are magnifications showing single stainings of the outlined regions. Scale bar: 50 µm in the overview image, 37.5 µm in the inset. N = 2-3 mice. e, Representative flow cytometry plots and violin plots showing total cell counts of IgA⁺ cells in lactating mammary glands (day 7 pp) of isotype-treated or CSF-1 antibody-treated (on day 1, 3, 5 pp) dams. Data were pooled from 3 independent experiments, n = 5-6. Two-tailed Mann-Whitney test was performed, ns = not significant.

Extended Data Fig. 6. The three mammary gland macrophages respond differently to microbial stimuli.

a-b. Representative flow cytometry plots (a) and violin plots (b) showing the percentage of proIL-1β⁺ or TNF⁺ liMacs (a,b) and F4/80^lo macrophages (Mac) and F4/80^hi macrophages (b) from lactating mammary glands (day 11–14 pp) of wild-type mice after exposure to LPS or zymosan in vitro for 6 h. Data was pooled from two independent experiments, n = 7. Each value is a mean of 3 replicates. Kruskal-Wallis test with Dunn’s multiple comparisons test was performed, *p < 0.05; **p < 0.01; ***p < 0.001. c, Representative flow cytometry plots and violin plots showing the percentage of pHrodo Red⁺ F4/80^lo macrophages, F4/80^hi macrophages and liMacs from the lactating mammary glands (day 10–15 pp) of wild-type mice after exposure in vitro to E. coli, zymosan and S. aureus pHrodo Red bioparticles for 90 minutes. N = 10 for E. coli and zymosan pHrodo Red bioparticles, pooled from 3 independent experiments (day 10–15 pp); n = 7 for S. aureus pHrodo Red bioparticles, pooled from 2 independent experiments (day 12–15 pp). All samples were run in triplicates and the means were calculated. Kruskal-Wallis test with Dunn’s multiple comparisons test was performed, *p < 0.05; **p < 0.01; ***p < 0.001, ns = not significant. d, Violin plots showing total cell counts of other mononuclear phagocytes (MNPs) (CD45⁺CD19^-NK1.1^-CD3^-SiglecF^-Ly6G^- ‘LiMac^-‘), eosinophils, NK cells, B cells and T cells in the lactating mammary glands (day 10–14 pp) of control (untreated) and CSF-1 antibody-treated (on day 7–11 pp) mice, challenged with LPS 18 hours prior to analysis or left untreated. Data (n = 5-6 per group) were pooled from 2 independent experiments. One-way ANOVA test was applied, *p<0.05, **p<0.01, ****p<0.0001, ns = not significant. Related to Fig. 5.

Extended Data Fig. 7. Three subsets of macrophages are present in human milk.

a, Representative flow cytometry plots of human breast milk T cells, eosinophils, neutrophils and macrophages (pre-gated on CD45⁺ cells) and representative histograms for the different markers for each population. b, Heatmap showing cluster-defining genes of CD45⁺ cells sorted from human milk samples (n = 6). Related to Fig. 6. c, Dot plot showing FOLR2⁺ macrophage signature gene expression in Mac 1, Mac 2 and Mac 3. d, Principal component analysis of F4/80^lo macrophages, F4/80^hi macrophages and liMacs of murine mammary gland (day 7 pp) and Mac 1, Mac 2 and Mac 3 populations of human milk (day 8–63 pp). e, Enrichment of GO terms according to the gene expression pattern in human milk Mac1, Mac2 and Mac3 as compared to T cells. Related to Fig. 6.