The role of CSF1R-dependent macrophages in control of the intestinal stem-cell niche

Abstract

Colony-stimulating factor 1 (CSF1) controls the growth and differentiation of macrophages.CSF1R signaling has been implicated in the maintenance of the intestinal stem cell niche and differentiation of Paneth cells, but evidence of expression of CSF1R within the crypt is equivocal. Here we show that CSF1R-dependent macrophages influence intestinal epithelial differentiation and homeostasis. In the intestinal lamina propria CSF1R mRNA expression is restricted to macrophages which are intimately associated with the crypt epithelium, and is undetectable in Paneth cells. Macrophage ablation following CSF1R blockade affects Paneth cell differentiation and leads to a reduction of Lgr5⁺ intestinal stem cells. The disturbances to the crypt caused by macrophage depletion adversely affect the subsequent differentiation of intestinal epithelial cell lineages. Goblet cell density is enhanced, whereas the development of M cells in Peyer's patches is impeded. We suggest that modification of the phenotype or abundance of macrophages in the gut wall alters the development of the intestinal epithelium and the ability to sample gut antigens.

Fig. 1

Prolonged anti-CSF1R blockade depletes macrophages in the gut. Csf1r-EGFP mice were treated with anti-CSF1R mAb or control-IgG (control) for 6 weeks before analysis. a Depletion of Csf1r-EGFP⁺ (green) macrophages in the gut wall of anti-CSF1R mAb-treated mice. Scale bar, 100 µm. Images are representative of four mice/group from three independent experiments. b Detection of Csf1r-EGFP⁺ (green) and CD68⁺ (red) macrophages in Peyer’s patches by IHC. FAE, follicle-associated epithelium; SED, subepithelial dome; V, villus. Broken lines indicate the FAE boundary. Scale bar, 100 µm. c Morphometric analysis of the density of Csf1r-EGFP⁺CD68⁺ macrophages in Peyer’s patches. Data represent mean ± SD from 2 to 10 sections from 3 to 4 mice/group. ***P < 0.001, two-tailed unpaired Student’s t-test. d RT–qPCR analysis of Cd68, Emr1, and Csf1r expression in Peyer’s patches. Bars represent mean ± SEM. Data are derived from 3 to 4 mice/group. *P < 0.05, two-tailed unpaired Student’s t-test. NS not significant

Fig. 2

Prolonged anti-CSF1R blockade leads to a loss of lysozyme-expressing Paneth cells in the crypts. Mice were treated with anti-CSF1R mAb or control-IgG (control) for 6 weeks before subsequent analysis. a IHC analysis shows loss lysozyme-expressing Paneth cells (red) in intestinal crypts after anti-CSF1R-treatment. Broken line indicates crypt and Paneth cell boundaries. Images are representative of 4 mice/group from 3 independent experiments. b Morphometric analysis of the density of lysozyme-expressing Paneth cells in the crypts after anti-CSF1R mAb treatment (open bars) when compared to controls (closed bars). Data represent mean ± SD from 6 to 20 sections from four mice/group, and are representative of data derived from three independent experiments. ***P < 0.001, Mann–Whitney U-test. c RT–qPCR analysis shows negligible expression of Lyz1, Lyz2, Wnt3, and Wnt3a in mRNA from isolated intestinal crypts following treatment with anti-CSF1R mAb. Data are normalized to the expression of Gapdh. Bars represent mean ± SEM. Data are derived from four mice/group. *P < 0.05; ***P < 0.001, two-tailed unpaired Student’s t-test. d Mice were treated with anti-CSF1R mAb or control-IgG (control) for 6 weeks. A parallel group of mice were allowed to recover for 8 week following anti-CSF1R mAb. IHC analysis shows loss lysozyme-expressing Paneth cells (red) in intestinal crypts and CD68⁺ macrophages (green) in the gut wall after anti-CSF1R-treatment. The presence of these cells was restored within an 8 week recovery period following anti-CSF1R-treatment. Scale bar 100 µm. e Morphometric analysis of the % crypts with lysozyme-expressing Paneth cells mice from each treatment group. Data represent mean ± SD from sections from 3 to 4 mice/group, and 30 to 120 crypts/section, and are representative of data from three independent experiments. ***P < 0.001, one-way ANOVA with Tukey’s post hoc test. NS not significant

Fig. 3

Paneth cells do not express CSF1R mRNA and protein. Mice were treated with anti-CSF1R mAb or control-IgG (control) for 6 weeks before analysis. A parallel group of mice were allowed to recover for 8 wk following anti-CSF1R mAb treatment. a Analysis of H&E stained intestines revealed that Paneth cells containing secretory granules (arrows) were present following prolonged CSF1R blockade. Scale bar, 20 µm. Representative images from 3 mice/group are shown. b RNA in situ hybridization analyses showed that Defa1 mRNA (red) was abundantly expressed in the intestinal crypts of mice treated with anti-CSF1R mAb. Scale bar, 100 µm. Representative images from 3 mice/group are shown. c Morphometric analysis of the % crypts with Defa1-expressing Paneth cells in the crypts of mice from each treatment group. Data represent mean ± SD derived from 3 mice/group, and 96–100 crypts/section. ns, not significant, one-way ANOVA with Tukey’s post hoc test. d IHC analysis of the crypts of untreated Csf1r-EGFP mice shows Paneth cells do not express CSF1R. In the crypt EGFP expression was only detected in cells with a macrophage morphology. e RT–qPCR analysis shows negligible expression of Cd68, Emr1, and Csf1r in mRNA from isolated intestinal crypts. Data are normalized to the expression of Gapdh. f RNA in situ hybridization analyses showed that Csf1r mRNA (blue) was not expressed in the epithelium of the small intestine. Scale bar, 50 µm. Inset shows absence of Csf1r mRNA in the epithelium of the intestinal crypt. g IHC analysis of lysozyme-expressing Paneth cells (red) in intestinal crypts and CD68⁺ macrophages (green) in the gut wall of Csf1r^ΔIEC, Vil1-Cre and Csf1r^FL/FL mice. Representative images from three mice/group are shown. Scale bar 100 µm. h Morphometric analysis revealed that the % crypts with lysoszyme-expressing Paneth cells was similar in the intestines of Csf1r^ΔIEC, Vil1-Cre and Csf1r^FL/FL mice. Data represent mean ± SD derived from three mice/group, and 34–118 crypts/section. NS not significant, one-way ANOVA with Tukey’s post hoc test

Fig. 4

Macrophages are the only cells which express CSF1R within intestinal crypts. a IHC analysis shows Csf1r-EGFP⁺CD68⁺ macrophages are in direct intimate contact with the crypt epithelial cell layer. Scale bar, 20 µm. b Left-hand panel shows the distribution of Csf1r (blue) and eGfp (red) mRNA expression in Peyer’s patches. Scale bar, 200 µm. In the gut wall Csf1r (red, middle panel; blue, right-hand panel) and eGfp (blue) mRNA expression was only expressed by macrophages (Emr1, red; right-hand panel). Scale bar, 50 µm. c IHC analysis shows that in the crypts and gut wall of Csf1r-EGFP mice, Csf1r-EGFP⁺ macrophages (green) were depleted after CSF1R blockade. Sections in a, c were counterstained with DAPI to detect cell nuclei (blue). Broken lines indicate the crypt epithelial cell layer boundary. Scale bar, 50 µm. Representative images are shown from 3 to 4 intestinal segments from 4 to 8 mice/group, from three independent experiments

Fig. 5

Prolonged anti-CSF1R treatment impairs crypt homeostasis. Mice were treated with anti-CSF1R mAb or control-IgG (control) for 6 weeks before analysis. A parallel group of mice were allowed to recover for 8 wk following anti-CSF1R mAb treatment. a Comparison of Lgr5 mRNA expression in isolated intestinal crypts from anti-CSF1R (open bars) or control-treated (closed bars) mice by RT–qPCR analysis. Bars represent mean ± SEM derived from four mice/group. *P < 0.05, two-tailed unpaired Student’s t-test. b Two-color mRNA in situ hybridization analysis confirmed a reduction in Lgr5 mRNA expression (red) at the base of intestinal crypts coincident with the ablation of Csf1r mRNA (blue) expressing macrophages. Representative images from three mice/group are shown. Scale bar, 25 µm. c Detection of proliferating Ki-67⁺ cells (red) in intestinal crypts by IHC. Sections counterstained with DAPI to detect nuclei (blue). Broken lines indicate the crypt epithelial cell layer boundary. Scale bar, 20 µm. d Morphometric analysis of (left) the magnitude of Ki-67-specific immunostaining in intestinal crypts, and (right) the distribution of Ki-67-expression within the crypt. Data represent mean ± SD derived from 5 to 9 sections from four mice/group. ***P < 0.001, two-tailed unpaired Student’s t-test. e IHC detection of SOX9 expression (green) in intestinal crypts. Sections were counterstained with DAPI to detect cell nuclei (blue). Scale bar, 50 µm. f Morphometric analysis confirmed that following prolonged CSF1R blockade the % area of the crypts with SOX9 + cell nuclei was significantly increased. Bars represent mean ± SD derived from four mice/group with 50–100 crypts/mouse, and are representative of data from two independent experiments. *P < 0.05, one-way ANOVA with Tukey’s post hoc test. g RT–qPCR analysis of Bmi1 mRNA expression in isolated intestinal crypts from anti-CSF1R or control-treated mice. Bars represent mean ± SEM derived from four mice/group. *P < 0.05, two-tailed unpaired Student’s t-test. h mRNA in situ hybridization analysis revealed an increase in the abundance of Bmi1-expressing cells (red, arrows) in the intestinal crypts of anti-CSF1R-treated mice. Scale bar, 20 µm. i Mean no. Bmi1-expressing cells/crypt. Bars represent mean ± SD derived from four mice/group with 40–50 crypts/mouse. *P < 0.05, one-way ANOVA with Tukey’s post hoc test

Fig. 6

Anti-CSF1R-treatment or CSF1-stimulation do not directly affect crypt homeostasis. a RT–qPCR analysis shows enteroids prepared from the small intestines of untreated C57BL/6J mice express negligible levels of Cd68, Emr1, and Csf1r. Expression of these genes is undetectable after passage. Data represent mean ± SEM from triplicate cultures. b, c Enteroids were prepared from the small intestines of C57BL/6J mice and at passage 3 were cultivated in the presence or absence of anti-CSF1R mAb (10 µg/ml), control rat IgG (10 µg/ml) or CSF1-Fc (500 ng/ml). b Mean enteroid yield and size at day 5 after treatment. Data represent mean ± SD from eight wells/treatment. Enteroid yield; NS Kruskal–Wallis one-way ANOVA. Enteroid size; NS one-way ANOVA with Tukey’s post hoc test. c Representative morphology of the enteroids at intervals after treatment. Scale bar, 50 µm. d RT–qPCR analysis shows the expression of Lyz1, Lyz2, Lgr5, and Olfm4 in enteroids was unaffected by treatment with anti-CSF1R mAb. Data represent mean ± SEM from triplicate cultures. e Representative images of day 7 enteroids prepared from the crypts of Csf1r^ΔIEC, Vil1-Cre and Csf1r^FL/FL mice. Scale bar, 150 µm. f The mean yield of enteroids prepared from the crypts of Csf1r^ΔIEC, Vil1-Cre and Csf1r^FL/FL mice on day 7 of cultivation was similar. Data represent mean ± SD from enteroids prepared from 3 mice/group, 9 wells/mouse. ns, one-way ANOVA with Tukey’s post hoc test. g Representative images of day 5 enteroids prepared from the crypts of mice treated with anti-CSF1R mAb or control-IgG (control) for 6 weeks. Scale bar, 150 µm. h The mean yield of enteroids prepared from the crypts of mice from each group was similar. Data represent mean ± SEM from 14 to 16 wells of organoids from four mice/group. NS two-tailed unpaired Student’s t-test

Fig. 7

M cells are dramatically reduced in Peyer’s patches after prolonged CSF1R blockade. a Histological analysis of Csf1r-EGFP mice shows CSF1R (green) is not expressed by epithelial cells in Peyer’s patches. FAE, follicle-associated epithelium; SED, subepithelial dome; dotted lines indicate the FAE boundary. Scale bar, 100 µm. b Expression of Csf1r (middle panel) and Gp2 in individual deep CAGE sequence datasets of mouse bone marrow-derived macrophages (BMDM) and M cells. Csf1r mRNA is undetectable in M cells. c–g Mice were treated with anti-CSF1R mAb or control-IgG (control) for 6 weeks before subsequent analysis. c Peyer’s patches were whole-mount immunostained to detect GP2 (green), UEA-1 (red) and f-actin (blue). Closed arrows, GP2⁺ M cells with characteristic basolateral pockets; chevrons, GP2-UEA-1⁺ goblet cells. The positions of the X–Z and Y–Z projections of the FAE are indicated by the broken lines in the X–Y images. Scale bar, 60 µm. Boxed areas are shown below in higher magnification. Scale bar, 20 µm. d Morphometric analysis showed that the density of GP2⁺ M cells was significantly reduced after prolonged CSF1R blockade, whereas the density of GP2⁻UEA-1⁺ goblet cells was significantly increased. Data are mean ± SD derived from 10 to 13 mice/group, 1 to 9 FAE/mouse from two independent experiments. ***P < 0.001, Mann–Whitney U-test; ****P < 0.001, two-tailed unpaired Student’s t-test. e Morphometric analysis showed that FAE size was unchanged after prolonged CSF1R blockade. Data are mean ± SEM derived from 10 to 13 mice/group, 1 to 9 FAE/mouse from two independent experiments. NS Mann–Whitney U-test. f, g The uptake of particulate antigen (fluorescent beads, green) into the Peyer’s patches was significantly impaired in anti-CSF1R-treated mice. Circles highlight individual beads in the SED. Sections counterstained to detect F-actin (blue). Scale bar, 5 µm. g The number of beads transcytosed across the FAE was significantly reduced in anti-CSF1R-treated mice. Data represent mean ± SD from 6 to 12 sections from 4 mice/group. ***P < 0.001, Mann–Whitney U-test

Fig. 8

Loss of immature and mature M-cell marker expression in Peyer’s patches after prolonged CSF1R blockade. Mice were treated with anti-CSF1R mAb or control-IgG (control) for 6 weeks before subsequent analysis. a Detection of Anxa5 (red) and GP2 (green) in the follicle-associated epithelium (FAE) by IHC. Scale bar, upper panels, 50 µm. Scale bar, lower panel, 20 µm. b Morphometric analysis of the number of GP2⁺ and Anxa5⁺ M cells in the FAE. SED, subepithelial dome. Data represent mean ± SD from 1 to 7 FAE from 8 mice/group. ***, P < 0.001, two-tailed unpaired Student’s t-test; ****P < 0.001, Mann–Whitney U-test. c IHC analysis of SpiB expression (green) in the FAE. Dotted line in A & C shows the FAE boundary. Scale bar, upper panels, 100 µm. Scale bar, lower panel, 20 µm. d Morphometric analysis of the number of SpiB⁺ cells in the FAE. Data represent mean ± SD from 6 to 9 FAE from four mice/group. ***, P < 0.001, Mann–Whitney U-test. e RT–qPCR analysis shows the expression of immature (Anxa5, Marcksl1, and Spib) and mature (Ccl9, Sgne1, and Gp2) M cell-related genes was reduced in Peyer’s patches after prolonged CSF1R blockade. Bars represent mean ± SEM derived from four mice/group. *P < 0.05; **P < 0.01, two-tailed unpaired Student’s t-test. NS not significant

Fig. 9

Prolonged CSF1R blockade does not affect RANKL or CCL20 expression in Peyer’s patches. a IHC analysis shows no observable difference in the expression or distribution of RANKL on subepithelial dome (SED) stromal cells after anti-CSF1R mAb treatment. Scale bar 100 µm. Morphometric analysis (histogram, right) confirmed the magnitude of the RANKL-specific immunostaining in the SED of Peyer’s patches from each group of mice was similar. Data represent mean ± SD from 3 to 4 mice/group and 1 to 3 SED regions/mouse. NS Mann–Whitney U-test. b Expression of Csf1r (black), Madcam1 (red) and Pdpn (blue) in individual mRNA sequencing datasets of mouse “M cell-inducing-” (MCi)-subepithelial mesenchymal cells and mesenchymal marginal reticular cells. c RT–qPCR analysis shows no significant difference in the expression of Tnfsf11 (RANKL), Tnfrsf11a (RANK), or Tnfrsf11b (OPG) mRNA in Peyer’s patches following anti-CSF1R treatment. Bars represent mean ± SEM derived from four mice/group. NS not significant, two-tailed unpaired Student’s t-test. d IHC analysis shows no observable difference in the expression of CCL20 in the follicle-associated epithelium (FAE) after anti-CSF1R mAb treatment. Scale bar, 100 µm. Morphometric analysis (histogram, right) confirmed the magnitude of the CCL20-specific immunostaining in Peyer’s patches from each group of mice was similar. Data represent mean ± SD from 3 to 9 FAE from four mice/group. NS two-tailed unpaired Student’s t-test. e IHC analysis of the distribution of CD11c⁺ (red) and B220⁺ (green) cells in the Peyer’s patches after anti-CSF1R mAb treatment (scale bar, 100 µm). Boxed areas are shown adjacently in higher magnification (scale bar, 25 µm). Arrows indicate CD11c⁺B220⁺ “M cell-inducing” B cells. Arrow-heads indicate CD11c⁻B220⁺ B cells. Sections counterstained with DAPI to detect nuclei (blue). f After anti-CSF1R treatment morphometric analysis showed that the number of CD11c⁺B220⁻ mononuclear phagocytes in the FAE was significantly decreased (left-hand panel), the number of CD11c⁻B220⁺ B cells was increased (right-hand panel), but the number of CD11c⁺B220⁺ “M cell-inducing” B cells was unchanged (lower panel). Dotted line indicates the FAE boundary. Data represent mean ± SD from 1–11 FAE from four mice/group. ***P < 0.001, two-tailed unpaired Student’s t-test. NS not significant