Differentiation of CD11c⁺CX₃CR1⁺ cells in the small intestine requires Notch signaling

Significance

The gastrointestinal tract directly faces an array of environmental agents, including bacteria and food. Intestine-specific subsets of immune cells maintain gut homeostasis by continuously sampling luminal antigens and maintaining immune tolerance. We here demonstrated that Rbpj, an essential molecule for Notch signaling, is essential for the development of CD11c⁺CX₃CR1⁺ cells that are crucial for sampling luminal antigens with dendrites projecting through the epithelial cell layer. These findings indicate that Notch signaling is required for maintaining the repertoires of intestinal antigen-presenting cells and suggest new ways to keep immune tolerance by modulating Notch-mediated CD11c⁺CX₃CR1⁺ cells.

Abstract

The gastrointestinal tract comes into direct contact with environmental agents, including bacteria, viruses, and foods. Intestine-specific subsets of immune cells maintain gut homeostasis by continuously sampling luminal antigens and maintaining immune tolerance. CD11c⁺CX₃CR1⁺ cells sample luminal antigens in the small intestine and contribute to the trafficking of bacteria to lymph nodes under dysbiotic conditions. The molecular mechanisms crucial for the differentiation of CD11c⁺CX₃CR1⁺ cells remain unclear. Here we demonstrate that the Notch1– or Notch2–Rbpj axis is essential for the development of CD11c⁺CX₃CR1⁺ cells. In mice in which Rbpj or Notch1 and Notch2 were deleted from CD11c⁺ cells, there was a deficit of CD11c⁺CX₃CR1⁺ cells and an accumulation of CD11c^lowCX₃CR1⁺ cells. The CD11c^lowCX₃CR1⁺ cells could not differentiate to CD11c⁺CX₃CR1⁺ cells, suggesting that CD11c^lowCX₃CR1⁺ cells represent a lineage distinct from CD11c⁺CX₃CR1⁺ cells. These data indicate that Notch signaling is essential for lineage fixation of intestinal CD11c⁺CX₃CR1⁺ cells.

The intestinal mucosa comes into direct contact with a complex environment that includes both foods and microorganisms. Consequently, it has unique subsets of immune cells to protect the host (1–4), and complex immunoregulatory mechanisms determine the immune tolerance to commensal microbes and food antigens (2, 4).

Dendritic cells (DCs) and macrophages in the intestine are required for sensing the presence of invading pathogens. The lamina propria (LP) contains a variety of DCs and macrophage subsets. One such subset expresses integrin αE (CD103) and has the ability to drive IL-17–producing T cells (5, 6). The CX3C chemokine receptor 1 (CX₃CR1)⁺ cells that express integrin αX (CD11c) and a macrophage marker, F4/80 but not CD103 are crucial for the sampling of luminal antigens through their use of projecting dendrites that penetrate the epithelial cell layer (7–9). Furthermore, CD11c⁺CX₃CR1⁺ cells are involved in the trafficking of commensal bacteria into mesenteric lymph nodes in the absence of Myd88 or under dysbiotic conditions (10) and transfer the luminal soluble antigen to integrin αM (CD11b)⁺CD103⁺ DCs via the gap junction to induce oral tolerance (11). The exact molecular mechanism that controls the development or differentiation of CD11c⁺CX₃CR1⁺ cells remains unclear.

Rbpj is a transcriptional regulator required for Notch signaling (12). Accumulating evidence indicates that Notch signaling is crucial for the development of immune cells and for the functional differentiation of T cells (12–19). Recent studies have revealed that Notch signaling also regulates the development of DCs in the spleen and intestine (20) (21, 22). These data indicate that Rbpj regulates the survival of CD8⁻ DCs in the spleen and that Notch2 signaling controls CD11b⁺CD103⁺ DCs in the intestinal LP. A recent study has revealed that Notch2-regulated intestinal DCs produce IL-23 required for controlling Citrobacter rodentium (22). In addition to the development or differentiation of DCs, Notch plays an important role in the activation of DCs and macrophages (23, 24).

Here we demonstrate that Rbpj is required for the development of CD11c⁺CX₃CR1⁺ cells and that both Notch1 and Notch2 are involved in the development of CD11c⁺CX₃CR1⁺ cells. The lack of CD11c⁺CX₃CR1⁺ cells in mice in which Rbpj or Notch1 and Notch2 was deleted from CD11c⁺ cells was accompanied by an accumulation of CD11c^lowCX₃CR1⁺ cells that could not differentiate toward CD11c⁺CX₃CR1⁺ cells. These findings indicate that Notch–Rbpj axis is required for the lineage fixation of CD11c⁺CX₃CR1⁺ cells and suggest that Notch signaling has a complex role in maintaining repertoires of intestinal antigen-presenting cells.

Results

Deficiency of Rbpj in CD11c⁺ Cells Reduces CD11c⁺CX₃CR1⁺ Cells.

To reveal the involvement of Notch signaling in the development of CD11c⁺CX₃CR1⁺ cells, Rbpj^flox/flox mice were crossed with CD11c promoter-driven Cre transgenic (Rbpj^−/−) mice, and this strain was crossed further with Cx3cr1 knockin mice in which Cx3cr1 was replaced by GFP (7). (Hereafter, this strain is called “Rbpj^−/−:CX₃CR1^gfp/+ mice.”) We first investigated CX₃CR1⁺ cells in the small intestine of Rbpj^−/−:CX₃CR1^gfp/+ mice. The expression of CD11c and CD11b was evaluated by gating on F4/80, CX₃CR1, and MHC class II⁺ LP cells from Rbpj^−/−:CX₃CR1^gfp/+ mice and control Rbpj^+/+:CX₃CR1^gfp/+ mice (Fig. 1A). We found a marked reduction of CD11c⁺CX₃CR1⁺ cells in Rbpj^−/−:CX₃CR1^gfp/+ mice (Fig. 1A). In contrast, in Rbpj^−/−:CX₃CR1^gfp/+ mice, there was an accumulation of CD11c^lowCX₃CR1⁺ cells that were rare in control Rbpj^+/+:CX₃CR1^gfp/+ mice (Fig. 1A). Both CD11c⁺CX₃CR1⁺ cells in Rbpj^+/+:CX₃CR1^gfp/+ mice and CD11c^lowCX₃CR1⁺ cells in Rbpj^−/−:CX₃CR1^gfp/+ mice were negative for CD103 and positive for the macrophage marker CD68 (Fig. 1A). These data indicate that CD11c⁺CX₃CR1⁺ cells and CD11c^lowCX₃CR1⁺ cells are not DCs. The sideways scatter (SSC) and forward scatter (FSC) were relatively higher in CD11c^lowCX₃CR1⁺ cells than in CD11c⁺CX₃CR1⁺ cells (Fig. 1B). Giemsa staining of cells demonstrated that CD11c^lowCX₃CR1⁺ cells had larger cytosols with irregular shapes and contained more granules than did CD11c⁺CX₃CR1⁺ cells (Fig. 1B).

Fig. 1.

Rbpj deficiency in CD11c⁺ cells disturbs the differentiation of CD11c⁺CX₃CR1⁺ cells. (A) Small intestinal LP cells from Rbpj^−/−:CX₃CR1^gfp/+ and Rbpj^+/+:CX₃CR1^gfp/+ mice were stained with anti-F4/80, MHC class II, CD11b, CD11c, CD103, and CD68 antibodies. (Upper) Seven-amino actinomycin D–negative (7AAD⁻) cells were analyzed by flow cytometry. The gray histogram shows the negative control. The numbers shown in each figure indicate the percentage of cells in parental gates. (Lower) The size of each population in Rbpj^−/−:CX₃CR1^gfp/+ mice (open bars) and Rbpj^+/+:CX₃CR1^gfp/+ mice (filled bars) was calculated by counting the total number of small intestinal cells and the percentage of each population. The data are shown as means ± SE. *P < 0.05; **P < 0.01. N.S., not significant. (B) (Upper) Giemsa staining of sorted F4/80⁺MHC class II⁺CD11b⁺CD11c^lowCX₃CR1⁺ (CD11c^lowCX₃CR1⁺) cells from Rbpj^−/−:CX₃CR1^gfp/+ mice and F4/80⁺MHC class II⁺CD11b⁺CD11c⁺CX₃CR1⁺ (CD11c⁺CX₃CR1⁺) cells from Rbpj^+/+:CX₃CR1^gfp/+ mice. (Original magnification: 1,000×.) (Scale bar: 10 μm.) (Lower) The sizes of CD11c^lowCX₃CR1⁺ cells from Rbpj^−/−:CX₃CR1^gfp/+ mice (red dots) and CD11c⁺CX₃CR1⁺ cells from Rbpj^+/+:CX₃CR1^gfp/+ mice (black dots) were assessed by FSC and SSC. (C) CD11c^lowCX₃CR1⁺ cells from Rbpj^−/−:CX₃CR1^gfp/+ mice (red line) or CD11c⁺CX₃CR1⁺ cells from Rbpj^+/+:CX₃CR1^gfp/+ mice (black line) were stained with anti-CD40, CD80, or CD86 for evaluation by flow cytometry. Fluorescence-conjugated streptavidin or isotype-matched control antibody (gray) was used as the negative control. (D) Expression of Il10 mRNA from sorted R1–R4 populations from Rbpj^+/+ mice or from the R5 population from Rbpj^−/− mice was evaluated by real-time PCR and is shown as the relative mRNA expression levels of Il10 versus Actb. The data are shown as means ± SE. **P < 0.01. The gating strategy is the same as in Fig. S3A. (E) Alexa Fluor 647-conjugated OVA (black line) or unconjugated OVA (gray) was injected through the duodenum in Rbpj^−/−:CX₃CR1^gfp/+ mice or Rbpj^+/+:CX₃CR1^gfp/+ mice. Five hours after injection, expression of Alexa Fluor 647 in CD11c^lowCX₃CR1⁺ cells and CD11c⁺CX₃CR1⁺ cells was evaluated by flow cytometry. The data shown in these figures are representative of four independent experiments.

We next compared the expression of cell-surface molecules on CD11c^lowCX₃CR1⁺ and CD11c⁺CX₃CR1⁺ cells (Fig. 1C). CD11c^lowCX₃CR1⁺ cells expressed CD40 and CD80 at levels equivalent to CD11c⁺CX₃CR1⁺ cells (Fig. 1C). In contrast, the expression of CD86 (Fig. 1C) and CD68 (Fig. 1A) was higher in CD11c^lowCX₃CR1⁺ cells than in CD11c⁺CX₃CR1⁺ cells, and the expression of CD4 was lower in CD11c^lowCX₃CR1⁺ cells than in CD11c⁺CX₃CR1⁺ cells (Fig. S1A). The expression pattern of cell-surface molecules on CD11c^lowCX₃CR1⁺ cells in Rbpj^−/−:CX₃CR1^gfp/+ mice was similar to that in Rbpj^+/+:CX₃CR1^gfp/+ mice (Fig. S1B). The absence of the Cx3cr1 does not affect the development of CD11c⁺CX₃CR1⁺ cells (Fig. S2), indicating that CX₃CR1 is not involved in the development of Rbpj-mediated differentiation of CD11c⁺CX₃CR1⁺ cells.

Previous studies categorized intestinal antigen-presenting cells into four groups (R1–R4) based on their expression of CD11c and CD11b (Fig. S3A) (5). The R1, R2, R3, and R4 populations consist of CD11b⁻CD103⁺ DCs, CD11b⁺CD103⁺ DCs, macrophages, and eosinophils, respectively (5). The R3 population expresses MHC class II (Fig. S3B), and about 90% of the R3 population is positive for CX₃CR1 and F4/80 (Fig. S3C), indicating that the R3 population is almost identical to CD11c⁺CX₃CR1⁺ cells. Therefore, when analyzing mice that do not have CX₃CR1-gfp allele, we used R3 as a population that corresponds to CD11c⁺CX₃CR1⁺ cells. By analyzing CD11c and CD11b in Rbpj^f/f-CD11c mice, we found an increased proportion of CD11c^lowCD11b⁺ cells that we named “R5” (Fig. S3A). CD11c^lowCD11b⁺ cells were positive for MHC class II (Fig. S3B), CX₃CR1, and F4/80 (Fig. S3C), indicating that the R5 population is almost identical to CD11c^lowCX₃CR1⁺ cells. A previous study revealed that the R3 population secretes IL-10, which is involved in the low T-cell–stimulatory activity of this population (25, 26). We tested the mRNA expression of Il10 in R1–R4 populations from control mice and in the R5 population from Rbpj^−/− mice. As previously reported, the R3 (CD11c⁺CX₃CR1⁺) population expressed Il10 at a high level, and the R5 (CD11c^lowCX₃CR1⁺) population also expressed substantial levels of Il10 mRNA (Fig. 1D).

To compare the antigen-capturing activity of CD11c⁺CX₃CR1⁺ and CD11c^lowCX₃CR1⁺ cells, Alexa Fluor 647-conjugated ovalbumin (OVA) was injected through the duodenum, and OVA uptake was evaluated 5 h later. The CD11c^lowCX₃CR1⁺ population in Rbpj^−/−:CX₃CR1^gfp/+ mice and the CD11c^lowCX₃CR1⁺-like population in Rbpj^+/+:CX₃CR1^gfp/+ mice took up more OVA than did the CD11c⁺CX₃CR1⁺ population (Fig. 1E), indicating that CD11c^lowCX₃CR1⁺ cells have a greater ability than CD11c⁺CX₃CR1⁺ cells to take up exogenous antigens.

We assessed functional difference in intestinal immune responses between Rbpj^+/+ and Rbpj^−/− mice by using a dextran sodium sulfate (DSS)-induced colitis model. However, severity as evaluated by the body weight loss was comparable between two groups (Fig. S4).

Rbpj^−/− Mice Have Few CD11c⁺Cells in the Small Intestine.

We analyzed the expression of F4/80 and CD11c in the small intestine of Rbpj^−/− and Rbpj^+/+ mice. The numbers and distributions of F4/80⁺ cells were comparable in Rbpj^+/+ mice and Rbpj^−/− mice (Fig. 2A). CD11c⁺ cells were restricted to the LP region in Rbpj^+/+ mice but were scarcely detected in Rbpj^−/− mice (Fig. 2A). Those results are consistent with findings that Rbpj^−/− mice have few CD11c⁺CX₃CR1⁺ cells. We next compared the intestinal sites in which CD11c⁺CX₃CR1⁺ cells and CD11c^lowCX₃CR1⁺ cells were found. The number and distribution of CX₃CR1⁺ cells were comparable in Rbpj^+/+:CX₃CR1^gfp/+ and Rbpj^−/−:CX₃CR1^gfp/+ mice (Fig. 2B). Those data suggest that CD11c-dependent Rbpj deficiency skews the development of CD11c⁺CX₃CR1⁺ cells to CD11c^lowCX₃CR1⁺ cells, but the distributions of those two populations in the tissue were similar.

Fig. 2.

Rbpj deficiency in CD11c⁺ cells disturbs the development of CD11c⁺CX₃CR1⁺ cells in the small intestine. (A) Sections of ileum of Rbpj^+/+ or Rbpj^−/− mice were stained with anti-CD11c (red) and F4/80 (green) antibodies. The nuclei were stained with DAPI (blue). Sections were evaluated by confocal microscopy. (Original magnification: 200×.) (Scale bar: 100 μm.) (B) Sections of small intestine of Rbpj^+/+:CX₃CR1^gfp/+ mice or Rbpj^−/−:CX₃CR1^gfp/+ mice were stained with anti-CD11c (red) and GFP (green) antibodies. The nuclei were stained with DAPI (blue). Sections were evaluated by confocal microscopy. (Original magnification 400×.) (Scale bar: 50 μm.) The data shown in these figures are representative of four independent experiments.

Notch1 and Notch2 Are Required for the Differentiation of CD11c⁺CX₃CR1⁺ Cells.

We analyzed the expression of Notch receptors on CD11c⁺CD11b⁺ (R3: CD11c⁺CX₃CR1⁺), CD11c⁺CD11b⁻ (R1: CD11b⁻CD103⁺ DCs), and CD11c^highCD11b^high (R2: CD11b⁺CD103⁺ DCs) populations. R2 (CD11b⁺CD103⁺ DCs) populations expressed Notch2 and Notch3 at low levels. R3 (CD11c⁺CX₃CR1⁺) populations expressed Notch1, 2, and 3, and the expression of Notch 3 was relatively higher than in the R1 (CD11b⁻CD103⁺ DCs) and R2 (CD11b⁺CD103⁺ DCs) populations (Fig. S5A). To determine if the R3 (CD11c⁺CX₃CR1⁺) population received Notch signals, the expression of Hes1 (a Notch target gene) was evaluated by real-time PCR (Fig. S5B). The R3 (CD11c⁺CX₃CR1⁺) population expressed higher amounts of Hes1 than did the R1 (CD11b⁻CD103⁺ DCs) and R2 (CD11b⁺CD103⁺ DCs) populations (Fig. S5B), suggesting that the R3 (CD11c⁺CX₃CR1⁺) population received stronger Notch signaling than other types of intestinal antigen-presenting cells.

To examine which Notch receptors are required for the differentiation of CD11c⁺CX₃CR1⁺ cells, we crossed Notch1^flox/flox mice and Notch2^flox/flox mice with CD11c promoter-driven Cre transgenic mice and with Cx3cr1 knockin mice (N1-KO or N2-KO, respectively). Mice deficient in both N1 and N2 were termed “N1/N2-DKO mice.” N1-KO mice and N2-KO mice had reduced numbers of CD11c⁺CX₃CR1⁺ cells. N1/N2-DKO mice had markedly reduced numbers of CD11c⁺CX₃CR1⁺ cells (Fig. 3A). The N1-KO mice, the N2-KO mice, and N1/N2-DKO mice had increased numbers of CD11c^lowCX₃CR1⁺ cells with higher FSC and SSC (Fig. 3A) that expressed higher levels of CD86 than control cells (Fig. 3B). In contrast, the R3 (CD11c⁺CX₃CR1⁺) population in Notch3-deficient mice was comparable to that in wild-type mice (Fig. 3C and Fig. S6A). A previous report (21) showed that Notch2 controls the development of CD11b⁺CD103⁺ DCs. We also found that numbers of CD11b⁺CD103⁺ DCs in small intestine were reduced in N2-KO and N1/N2-DKO mice (Fig. S6B). In contrast, the number of CD11b⁺CD103⁺ DCs is partially decreased in Rbpj^−/−:CX₃CR1 mice (Fig. S3A and Fig. S6B), and there was no significant reduction of CD11b⁺CD103⁺ DCs in N1-KO mice (Fig. S6B). Those data indicate that both Notch1 and Notch2 contribute to the differentiation of CD11c⁺CX₃CR1⁺ cells and that Notch2, but not Notch1, controls the development of CD11b⁺CD103⁺ DCs. Although Notch3-deficient mice did not show any impaired differentiation of R3 (CD11c⁺CX₃CR1⁺) cells, it would be necessary to assess the contribution of Notch3 to the development of R3 (CD11c⁺CX₃CR1⁺) on a Notch1- and Notch2-deficient background.

Fig. 3.

Notch1 and Notch2 are required for the differentiation of CD11c⁺CX₃CR1⁺ cells. LP cells from the small intestine were isolated from four mouse strains that have the CD11c-Cre transgene and are heterozygous for the CX₃CR1-gfp allele: N1-KO, N2-KO, N1/N2-DKO, and control mice. (A) Cells were stained for evaluation with antibodies against F4/80, MHC class II, CD11c, and CD11b. CD11b and CD11c expression (Upper; numbers indicate the percentage of cells) or FSC and SSC (Lower; numbers show the value of FSC and SSC) were analyzed by gating on the 7AAD⁻F4/80⁺CX₃CR1⁺MHC class II⁺ population. The gating strategy is the same as in Fig. 1A. (B) Expression of CD40, CD80, and CD86 in CD11c⁺CX₃CR1⁺ cells or CD11c^lowCX₃CR1⁺ cells was evaluated by flow cytometry. Fluorescence-conjugated streptavidin or isotype-matched control antibody (gray area) was used as the negative control. (C) LP cells from the small intestine were isolated from Notch3-deficient mice. Cells were stained with antibodies against CD11c and CD11b, and their expression was evaluated. The gating strategy was the same as in Fig. S3A. (D) Sections of ileum from Dll1^f/+-Vil or Dll1^f/f-Vil mice were stained with anti-Dll1 (red) antibody. The nuclei were stained with DAPI (blue). Sections were evaluated by confocal microscopy. (Original magnification: 200×.) (Scale bar: 50 μm.) (E) GFP expression in the sections of ileum in Vil-CAG mice was evaluated by confocal microscopy. (Original magnification: 200×.) (Scale bar: 50 μm.) (F) LP cells from the small intestine were isolated from Dll1^f/+-Vil and Dll1^f/f-Vil mice. Cells were stained with antibodies against CD11c and CD11b, and their expression was evaluated. The gating strategy is the same as in Fig. S3A. The data shown in these figures are representative of four independent experiments.

We sought to assess the Notch ligands that regulated the development of CD11c⁺CX₃CR1⁺ cells. We first assessed the expression of Delta-like 1 (Dll1) in the small intestine. To assess the expression of Dll1, we used mice in which Dll1 was deleted by Vil1-Cre (Dll1^f/f-Vil mice) and control mice. We found that Dll1 is expressed in the crypt base where precursors of intestinal epithelial cells are present in control Dll1^f/+-Vil mice (Fig. 3D). This result was confirmed by the absence of Dll1 staining in the small intestine from Dll1^f/f-Vil mice (Fig. 3D). To evaluate the expression pattern of Vil1-Cre, Vil1-Cre mice were crossed with CAG-CAT-GFP (Vil-CAG) mice, in which GFP expression was directed upon Cre-mediated excision of the loxP-flanked chloramphenicol acetyltransferase gene located between the CAG promoter and Gfp. The expression of Vil1-Cre was restricted to the crypt base and intestinal epithelial cells (Fig. 3E). However, the numbers of the R3 (CD11c⁺CX₃CR1⁺) population in Dll1^f/f-Vil mice were comparable to those in control mice (Fig. 3F). These results demonstrate that Dll1 expression on intestinal epithelial cells is not required for the development or differentiation of CD11c⁺CX₃CR1⁺ cells. The numbers of the R3 (CD11c⁺CX₃CR1⁺) population in Jagged1^f/f-Vil mice, in which Jagged1 was deleted by Vil1-Cre, also were comparable to those in control mice (Fig. S6C).

Intrinsic Notch Signaling Is Required for CD11b⁺CD11c⁺ Development.

We asked whether the deletion of Rbpj from CD11c⁺ cells is intrinsically involved in the defective development of the R3 (CD11c⁺CX₃CR1⁺) and R5 (CD11c^lowCX₃CR1⁺) populations. Therefore, we crossed Rbpj^+/−or Rbpj^−/− mice with CAG-CAT-GFP mice (Rbpj^+/−-CAG or Rbpj^−/−-CAG, respectively). About 90% of the R3 (CD11c⁺CX₃CR1⁺) population from Rbpj^+/−-CAG mice and the R5 (CD11c^lowCX₃CR1⁺) population from Rbpj^−/−-CAG mice expressed GFP (Fig. 4A), indicating that Rbpj is deleted efficiently in both the R3 (CD11c⁺CX₃CR1⁺) and R5 (CD11c^lowCX₃CR1⁺) populations. To exclude the possibility that Rbpj deficiency in CD11c⁺CX₃CR1⁺ cells affect neighboring cells and so contribute to the differentiation of CD11c^lowCX₃CR1⁺ cells, we transplanted bone marrow cells from Rbpj^+/+:CX₃CR1^gfp/+ mice (CD45.1/CD45.2) or from Rbpj^−/−:CX₃CR1^gfp/+ (CD45.2) mice into lethally irradiated C57BL/6 (CD45.1) mice. Four weeks after transplantation, we analyzed the development of CD11c⁺CX₃CR1⁺ cells in the small intestine (Fig. 4B). CD11c⁺CX₃CR1⁺ cells were more numerous in the CD45.1⁺CD45.2⁺ fraction, whereas large numbers of CD11c^lowCX₃CR1⁺ cells differentiated in CD45.2⁺ cells (Fig. 4B). These data demonstrate that the lack of intrinsic Rbpj-mediated signaling was responsible for the differentiation of CD11c^lowCX₃CR1⁺ cells.

Fig. 4.

Intrinsic Rbpj is required for the differentiation of CD11c⁺CX₃CR1⁺ cells. (A) LP cells from the small intestine were isolated from Rbpj^+/−-CAG or Rbpj^−/−-CAG mice. Cells were stained with antibodies against CD11c and CD11b, and R3 (CD11c⁺CX₃CR1⁺) and R5 (CD11c^lowCX₃CR1⁺) populations were evaluated as in Fig. S3A. (B) Lethally irradiated C57BL/6 (CD45.1) mice were reconstituted with bone marrow cells from Rbpj^+/+:CX₃CR1^gfp/+ (CD45.1/CD45.2) or from Rbpj^−/−:CX₃CR1^gfp/+ (CD45.2) mice (1:1). Four weeks after transplantation, the expression of CD11b and CD11c by CX₃CR1⁺F4/80⁺MHC class II⁺CD45.1⁺CD45.2⁺ cells or CX₃CR1⁺F4/80⁺MHC class II⁺CD45.1⁻CD45.2⁺ cells was evaluated by flow cytometry. The data shown in these figures are representative of four independent experiments.

Notch Alters the Differentiation of CD11c⁺CX₃CR1⁺ Cells.

CD11c⁺CX₃CR1⁺ cells are derived from monocyte precursors (27), and CX₃CR1⁺CD115⁺CD11b^highCD117⁻Gr1^high cells (Gr1^high cells) differentiate toward CX₃CR1⁺CD115⁺CD11b^highCD117⁻Gr1^low cells (Gr1^low cells). We examined the expression of Notch receptors on Gr1^high cells and Gr1^low cells. The Gr1^high cells expressed Notch2, and Gr1^low cells expressed Notch2 and 3 (Fig. 5A). Notch1 was not expressed by either cell type. We sought to determine whether Gr1^low cells could differentiate toward CD11c⁺CX₃CR1⁺ cells on OP9 or OP9-Dll1 cells in which Dll1 was overexpressed. Gr1^low cells could differentiate to CD11c⁺CX₃CR1⁺ cells on OP9 cells but could not differentiate toward CD11c⁺CX₃CR1⁺MHC class II⁺ cells (Fig. 5B). In contrast, Gr1^low cells could differentiate toward CD11c⁺CX₃CR1⁺MHC class II⁺ cells in vitro on OP9-Dll1 (Fig. 5B), supporting the notion that Gr1^low cells are progenitors for CD11c⁺CX₃CR1⁺ (MHC class II⁺) cells and that Notch signaling is crucial for the differentiation of CD11c⁺CX₃CR1⁺ (MHC class II⁺) cells.

Fig. 5.

CD11c^lowCX₃CR1⁺ cells represent a lineage distinct from that of CD11c⁺CX₃CR1⁺ cells. (A) Expression of Notch1–Notch4 by the Gr1^low or Gr1^high fraction from CX₃CR1^gfp/+ mouse bone marrow cells present in the CD115⁺CX₃CR1⁺CD11b^highCD117⁻ gate. Fluorescence-conjugated streptavidin or isotype-matched control antibody (gray). (B) CD115⁺CX₃CR1⁺CD11b^highCD117⁻Gr1^low cells from CX₃CR1^gfp/+ mice were cultured on OP9 or OP9-Dll1 cells for 5 d in the presence of macrophage colony-stimulating factor (M-CSF). Expression of CX₃CR1 and MHC class II was evaluated by flow cytometry by gating on both CD11c⁺ and CD11b⁺ cells. (C) CD11c⁺CX₃CR1⁺ cells or CD11c^lowCX₃CR1⁺ cells were sorted from Rbpj^−/−:CX₃CR1^gfp/+ mice and subsequently were cultured on OP9 or OP9-Dll1 cells with M-CSF, granulocyte macrophage colony-stimulating factor (GM-CSF), and Flt3L for 5 d. Expression of CD11b and CD11c was evaluated by flow cytometry by gating on MHC class II⁺ cells. (D) CD11c⁺CX₃CR1⁺ cells and CD11c^lowCX₃CR1⁺ cells were sorted from Rbpj^+/+:CX₃CR1^gfp/+ mice and subsequently were cultured on OP9 or OP9-Dll1 cells with M-CSF, GM-CSF, and Flt3L for 5 d. Expression of CD11b and CD11c was evaluated by flow cytometry by gating on MHC class II⁺ cells. The data shown in these figures are representative of four independent experiments.

Using this culture system, we evaluated the differentiation of CD11c^lowCX₃CR1⁺ cells and CD11c⁺CX₃CR1⁺ cells in vitro. The increase of CD11c^lowCX₃CR1⁺ cells in Rbpj^−/−:CX₃CR1^gfp/+ mice allowed us to evaluate whether CD11c^lowCX₃CR1⁺ cells are precursors of CD11c⁺CX₃CR1⁺ cells or can change to CD11c⁺CX₃CR1⁺ cells. CD11c⁺CX₃CR1⁺ cells and CD11c^lowCX₃CR1⁺ cells were sorted from Rbpj^−/−:CX₃CR1^gfp/+ mice and subsequently were cultured on OP9 or OP9-Dll1 cells (Fig. 5C). CD11c⁺CX₃CR1⁺ cells did not differentiate to CD11c^lowCX₃CR1⁺ cells; moreover, CD11c^lowCX₃CR1⁺ cells could not differentiate toward CD11c⁺CX₃CR1⁺ cells. Furthermore, CD11c⁺CX₃CR1⁺ cells from Rbpj^+/+:CX₃CR1^gfp/+ mice were unable to give rise to CD11c^lowCX₃CR1⁺ cells; similarly, CD11c^lowCX₃CR1⁺ cells from Rbpj^+/+:CX₃CR1^gfp/+ mice did not produce CD11c⁺CX₃CR1⁺ cells (Fig. 5D). Those data suggest that CD11c^lowCX₃CR1⁺ cells are not precursors of CD11c⁺CX₃CR1⁺ cells, nor can they change to CD11c⁺CX₃CR1⁺ cells.

Discussion

The intestine is in continuous contact with both pathological and beneficial commensal bacteria (2, 4, 28). Epithelial cell recognition of such microorganisms is required for the maintenance of immune homeostasis in the gut (4, 28). CD11c⁺CX₃CR1⁺ cells contribute to the sampling of intestinal antigens (7–9), and they are crucial for the trafficking of bacteria to lymph nodes under dysbiotic condition (10). However, the molecular mechanisms that regulate the differentiation or development of CD11c⁺CX₃CR1⁺ cells remain unclear. Our present studies show that Notch signaling in CD11c⁺ cells is essential for the differentiation of intestinal CD11c⁺CX₃CR1⁺ cells. Moreover, the lack of Notch1 and Notch2 signaling skews the cell lineage toward CD11c^lowCX₃CR1⁺ cells that have stronger antigen uptake activity, differential expression of cell-surface markers, and distinct morphology as compared with CD11c⁺CX₃CR1⁺ cells. Those data suggest that Notch is crucial for fixing the CD11c⁺CX₃CR1⁺ cell lineage among intestinal antigen-presenting cells.

CX₃CR1⁺CD115⁺CD11b^highCD117⁻Gr1^high cells in the bone marrow differentiate to CX₃CR1⁺CD115⁺CD11b^highCD117⁻Gr1^low cells that are precursors for CD11c⁺CX₃CR1⁺ cells (27). Here, we found that CX₃CR1⁺CD115⁺CD11b^highCD117⁻Gr1^low cells differentiate to CD11c⁺CX₃CR1⁺MHC class II⁺ cells on OP9-Dll1 cells. In contrast, in the OP9 culture system, CD11c^lowCX₃CR1⁺ cells could not differentiate toward CD11c⁺CX₃CR1⁺ cells. Moreover, CD11c⁺CX₃CR1⁺ cells could not develop into CD11c^lowCX₃CR1⁺ cells. Those results demonstrate that Notch signaling is required for the differentiation of CD11c⁺CX₃CR1⁺ cells and that CD11c^lowCX₃CR1⁺ cells are not precursors for CD11c⁺CX₃CR1⁺ cells or vice versa. Rather, the data suggest that Notch is required for lineage fixation of CD11c⁺CX₃CR1⁺ cells and that the lack of Notch signaling skews the cell lineage from CD11c⁺CX₃CR1⁺ to CD11c^lowCX₃CR1⁺ cells. However, we have not eliminated the possibility that CD11c^lowCX₃CR1⁺ cells and CD11c⁺CX₃CR1⁺ cells are derived from distinct precursors or that the role of Notch in each precursor is to promote or inhibit their differentiation. This issue could be clarified by identifying the precursor cells for each population.

CD11c-Cre–mediated deletion of Notch1 or Notch2 disturbs the differentiation of CD11c⁺CX₃CR1⁺ cells. The Notch3-deficient population was normal in size, although Notch3 was highly expressed on the R3 population. The deficiency of both Notch1 and Notch2 leads to defects of CD11c⁺CX₃CR1⁺ cell differentiation similar to those seen in Rbpj-deficient cells, suggesting that Notch1 and Notch 2 are major receptors controlling CD11c⁺CX₃CR1⁺ cell differentiation. Previous studies reported that Notch2 signaling regulates CD11c⁺ESAM⁺ DCs in the spleen and CD11b⁺CD103⁺ DCs in the intestine (20–22), as confirmed in the present study, and also suggested that CD11b⁺CD103⁺ DCs are a major producer of IL-23, controlling Citrobacter rodentium infection (22). However, those studies did not find a defect in intestinal CD11c⁺CX₃CR1⁺ cell differentiation caused by Notch2 deficiency. Those studies did not use CX₃CR1 as a marker or analyze CD11c and CD11b by gating in MHC class II⁺ cells. Those approaches could explain why those reports could not detect the defect in CD11c⁺CX₃CR1⁺ cell differentiation caused by Notch2 deficiency. Therefore, the alteration of the intestinal immune system of mice in which Notch2 was deleted by CD11c-Cre is attributable to defective differentiation of both CD11b⁺CD103⁺ DCs and CD11c⁺CX₃CR1⁺ cells, together with an increased number of CD11c^lowCX₃CR1⁺ cells. In any case, Notch signaling plays key roles in establishing the repertoire of antigen-presenting cells in the intestine.

Five canonical Notch ligands that regulate CD11c⁺CX₃CR1⁺ cells are present in mice (12), and Vil1-Cre–mediated deletion of Jagged1 or Dll1 did not affect the development of CD11c⁺CX₃CR1⁺ cells. BecauseVil1-Cre is expressed only in intestinal epithelial cells, it is unlikely that the expression of Jagged1 or Dll1 in intestinal epithelial cells is involved in the differentiation of CD11c⁺CX₃CR1⁺ cells. The expression of Dll1 in the intestine is restricted in the intestinal epithelial cells; however, the expression pattern of other Notch ligands has not been studied. Therefore, it would be necessary to determine the expression pattern of Notch ligands in the intestine and bone marrow to identify the crucial Notch ligands that support CD11c⁺CX₃CR1⁺ cell differentiation.

The current results demonstrate that Notch signaling is indispensable for lineage fixation of CD11c⁺CX₃CR1⁺ cells. Moreover, the lack of Notch signaling in CD11c⁺ cells increases the number of CD11c^lowCX₃CR1⁺ cells. Although we did not observe any difference between control and Rbpj^−/− mice in the severity of DSS-induced colitis, the deletion of Rbpj by CD11c-Cre also affects the development of CD11b⁺CD103⁺ dendritic cells, as previously reported and shown here, and thereby might affect the immunological responses in the small intestine (21). Therefore, to analyze the functional role of Rbpj in CD11c⁺CX₃CR1⁺ cells precisely in future studies, it will be necessary to find more a specific marker that deletes only Rbpj in CD11c⁺CX₃CR1⁺ cells. Nevertheless, these results provide molecular insights into the differentiation of CD11c⁺CX₃CR1⁺ cells. CD11c⁺CX₃CR1⁺ cells are involved not only in sampling luminal antigens but also in conveying bacteria to lymph nodes under dysbiotic conditions. Therefore, we suggest that the identification of Notch signaling as the essential pathway for their differentiation may open new ways to modulate CD11c⁺CX₃CR1⁺ cells in dysbiosis-mediated pathological bacterial infections and oral tolerance by targeted inhibition of Notch signaling.

Methods

Mice.

C57BL/6 mice (6- to 8-wk-old) were purchased from Japan SLC. Notch1^flox/flox, Cx3cr1^gfp/gfp, Notch1^flox/flox, Vil1-Cre, and CD11c-Cre transgenic mice were purchased from Jackson Laboratory. Notch2^flox/flox (29), Rbpj^flox/flox (30), Notch3-deficient (31, 32), and CAG-CAT-GFP (33) mice were provided by S. Chiba (University of Tsukuba, Tsukuba, Japan), T. Honjo (Kyoto University, Kyoto, Japan), R. Kopan (University of Cincinnati College of Medicine, Cincinnati), and J. Miyazaki (Osaka University, Osaka, Japan), respectively. Dll1^flox/flox (34) and Jagged1^flox/flox (35) mice were previously described. All mice were maintained under specific pathogen-free conditions in the animal facilities at the University of Tokushima, Japan. All experiments were performed in accordance with institutional guidelines for animal care at the University of Tokushima.

Statistical Analysis.

For all experiments, the significant differences between groups were calculated using the Mann–Whitney u test for unpaired data. Differences were considered significant when P < 0.05.

Other methods are described in SI Methods.