Peripheral CD103+ dendritic cells form a unified subset developmentally related to CD8alpha+ conventional dendritic cells

Abstract

Although CD103-expressing dendritic cells (DCs) are widely present in nonlymphoid tissues, the transcription factors controlling their development and their relationship to other DC subsets remain unclear. Mice lacking the transcription factor Batf3 have a defect in the development of CD8alpha+ conventional DCs (cDCs) within lymphoid tissues. We demonstrate that Batf3(-/-) mice also lack CD103+CD11b- DCs in the lung, intestine, mesenteric lymph nodes (MLNs), dermis, and skin-draining lymph nodes. Notably, Batf3(-/-) mice displayed reduced priming of CD8 T cells after pulmonary Sendai virus infection, with increased pulmonary inflammation. In the MLNs and intestine, Batf3 deficiency resulted in the specific lack of CD103+CD11b- DCs, with the population of CD103+CD11b+ DCs remaining intact. Batf3(-/-) mice showed no evidence of spontaneous gastrointestinal inflammation and had a normal contact hypersensitivity (CHS) response, despite previous suggestions that CD103+ DCs were required for immune homeostasis in the gut and CHS. The relationship between CD8alpha+ cDCs and nonlymphoid CD103+ DCs implied by their shared dependence on Batf3 was further supported by similar patterns of gene expression and their shared developmental dependence on the transcription factor Irf8. These data provide evidence for a developmental relationship between lymphoid organ-resident CD8alpha+ cDCs and nonlymphoid CD103+ DCs.

Figure 1.

Batf3^−/− mice lack lung CD103⁺ DCs. (A) Cell suspensions of lung digests from Batf3^+/+ and Batf3^−/− mice were stained for CD45, CD11c, autofluorescence (FITC channel; auto), CD11b, and CD103. Live CD45⁺CD11c⁺ cells, representing a mixture of lung macrophages and DCs, were initially gated (left), with the low autofluorescent population (DCs) further analyzed for the discrimination of three pulmonary DC subsets (right). Numbers represent the percentage of cells within the indicated gates. (B) Numbers of pulmonary DCs per mouse for each of the three pulmonary DC subsets as gated in A. Data represent the combination of two separate experiments, with each point representing the pooled lung cells from an individual mouse. Horizontal lines represent the mean cell number. (C) Cell suspensions of lung digests from Batf3^+/+ and Batf3^−/− mice were stained for CD45, CD11c, CD11b, CD103, and intracellular Langerin. Live CD45⁺CD11c⁺ cells, representing a mixture of lung macrophages and DCs, were initially gated (left). A further gate was drawn to encompass all pulmonary DCs and exclude CD11b⁻CD103⁻ lung macrophages. These DCs were analyzed for the expression of CD103 and Langerin (right). Numbers represent the percentage of cells within the indicated gates. Staining was performed on two individual mice per group in one independent trial. (D) Batf3^+/+ and Batf3^−/− mice were infected with 100 PFU SeV intranasally or left uninfected (three or four uninfected mice per genotype; four infected mice per genotype). At day 8 after infection, SeV-specific CD8⁺ T cells were identified in cell suspensions of lung digests stained for CD19, Mac3, CD8α, and Kb-SeV NP 324–332 MHC pentamers. CD8α⁺CD19⁻Mac3⁻ cells were gated. Numbers represent the percentage of SeV-specific CD8⁺ T cells. Data are representative of two separate experiments (one in which staining was performed at day 9 after infection). (E) Percentage of SeV-specific CD8⁺ T cells in the lungs of uninfected or infected mice, as stained in D. Horizontal lines represent the mean percentage.

Figure 2.

Comparative microarray expression analysis of lung and splenic DC subsets. Microarray data for two lung DC and two splenic cDC subsets were normalized and modeled (three replicate arrays for each DC subset). Model-based expression indices were calculated using dChip software for selected (A) transcription factors, (B) surface receptors, and (C) growth factor receptors. Values represent the mean index, with error bars representing the standard deviation.

Figure 3.

Batf3^−/− mice lack intestinal CD103⁺CD11b⁻ DCs. (A) Cell suspensions of LP from Batf3^+/+ and Batf3^−/− mice were stained for CD11c, I-A^b, CD11b, and CD103 (representative data from a single experiment involving two to three mice per group; two independent experiments were performed). Live CD11c^highI-A^b+ cells were gated. Numbers represent the percentage of cells within the indicated gates. (B) Percentage of CD103⁺CD11b⁻ LP DCs as gated in A. Horizontal lines represent the mean percentage. (C) Cell suspensions of LP from Batf3^+/+ and Batf3^−/− mice were stained for CD3e, CD4, CD8, and intracellular Foxp3 (representative data from a single experiment with three mice per group). Live CD3e⁺CD4⁺CD8⁻ T cells were gated. Numbers represent the percentage of cells within the indicated gates. (D) Percentage of Foxp3⁺ LP T cells as gated in C. Horizontal lines represent the mean percentage. (D) Light microscopy of H&E-stained sections of the small intestine and colon from Batf3^+/+ and Batf3^−/− mice (data are representative of a single experiment involving two mice per group at 8–10 mo of age; a separate trial examined five mice per group at 3 mo of age with similar findings). Note the normal villous architecture and LP leukocytes, without evidence of inflammation. Bars: (left and right) 100 µm; (middle) 500 µm.

Figure 4.

Batf3^−/− mice lack dermal CD103⁺ DCs. (A) Cell suspensions of epidermal and dermal digests from Batf3^+/+ and Batf3^−/− mice were stained for CD45, CD11c, intracellular Langerin, CD103, and CD11b (representative data from a single experiment involving two mice per group). Live CD45⁺CD11c⁺ cells were initially gated (left), allowing for the discrimination of CD103⁻Langerin⁺ LCs and CD103⁺Langerin⁺ dermal DCs. Langerin-expressing cells were gated for further analysis (right), allowing good discrimination of these two populations based on their differential expression of CD11b. Numbers represent the percentage of cells within the indicated gates. (B) Cell suspensions of pooled SDLNs from Batf3^+/+ and Batf3^−/− mice were stained for CD11c, intracellular Langerin, CD103, DEC205, CD11b, and CD8α (representative data from a single experiment involving two mice per group; experiments with similar combinations of stains have been performed in two other independent trials involving at least one mouse per genotype in each trial). Live CD11c^high cells were gated. Note the complete absence of CD103⁺Langerin⁺ dermal DCs (far left) and CD8α⁺Langerin^low lymph node–resident DCs in Batf3^−/− mice. Langerin-expressing cells present in Batf3^−/− mice represent epidermal LCs. (C) Percentage and number of DCs per inguinal lymph node for three DC subsets. In this case, individual cell suspensions of inguinal lymph nodes (three mice per group, six lymph nodes total) were stained for CD11c, DEC205, CD8α, CD103, and intracellular Langerin. Live CD11c^high cells were initially gated. Lymph node–resident DEC205⁺CD8α⁺ cDCs were identified as a percentage of this gate. Migratory DCs were further gated as DEC205⁺CD8α⁻ cells, and Langerin⁺CD103⁺ dermal DCs and Langerin⁺CD103⁻ LCs were identified as percentages of this gate. Horizontal lines represent the mean percentage and cell number.

Figure 5.

Irf8 mutant mice lack nonlymphoid CD103⁺ DCs. (A) Cell suspensions of spleens from WT (B6C3HF1/J) and Irf8 mutant (BXH2) mice were stained for CD11c, CD11b, CD8α, and DEC205 (representative data from a single experiment involving two mice per group; a second independent experiment was performed with one mouse per group). Live CD11c^high cells were gated. Numbers represent the percentage of cells within the indicated gates. (B) Cell suspensions of lung digests from WT (B6C3HF1/J) and Irf8 mutant (BXH2) mice were stained for CD45, CD11c, CD11b, and CD103 (representative data from a single experiment involving two mice per group; two independent experiments were performed; a second independent experiment was performed with one mouse per group). Live CD45⁺CD11c⁺ cells, representing a mixture of lung macrophages and DCs, were gated. Numbers represent the percentage of cells within the indicated gates. (C) Cell suspensions of LP from WT (C57BL/6) and Irf8 mutant (BXH2) mice were stained for CD11c, CD11b, and CD103 (representative data from a single experiment involving two mice per group; a second independent experiment was performed with one mouse per group). Live CD11c^high cells were gated. Numbers represent the percentage of cells within the indicated gates. (D) Cell suspensions of pooled SDLNs from WT (B6C3HF1/J) and Irf8 mutant (BXH2) mice were stained for CD11c, CD8α, DEC205, CD11b, and CD103 (representative data from a single experiment involving two mice per group; a second independent experiment was performed with one mouse per group). Live CD11c^high cells were initially gated (left) to allow the identification of lymph node–resident DEC205⁺CD8α⁺ cDCs. Migratory DCs were further gated as DEC205⁺CD8α⁻ cells and analyzed for their expression of CD11b and CD103. Numbers represent the percentage of cells within the indicated gates.

Figure 6.

Batf3^−/− mice display a normal CHS response. (A) Sensitized and nonsensitized Batf3^+/+ and Batf3^−/− mice were treated on their ears with DNFB to elicit a hapten-specific CHS response. Ear swelling was measured at days 1 and 2 after elicitation. Each point represents an individual ear (three mice per group, six ears per group; Fig. S6 A shows an independent trial of this same protocol). Horizontal lines represent the mean. (B) Light microscopy of H&E-stained sections of treated ears from nonsensitized and sensitized Batf3^+/+ and Batf3^−/− mice. Sensitized ears demonstrate edema and focal areas of leukocyte infiltration (arrowheads) composed largely of PMNs. Bars: (left) 1 mm; (right) 100 µm.