CX₃CR1⁺ CD8α⁺ dendritic cells are a steady-state population related to plasmacytoid dendritic cells

Abstract

Lymphoid organs are characterized by a complex network of phenotypically distinct dendritic cells (DC) with potentially unique roles in pathogen recognition and immunostimulation. Classical DC (cDC) include two major subsets distinguished in the mouse by the expression of CD8α. Here we describe a subset of CD8α⁺ DC in lymphoid organs of naïve mice characterized by expression of the CX₃CR1 chemokine receptor. CX₃CR1⁺ CD8α⁺ DC lack hallmarks of classical CD8α⁺ DC, including IL-12 secretion, the capacity to cross-present antigen, and their developmental dependence on the transcriptional factor BatF3. Gene-expression profiling showed that CX₃CR1⁺ CD8α⁺ DC resemble CD8α⁻ cDC. The microarray analysis further revealed a unique plasmacytoid DC (PDC) gene signature of CX₃CR1⁺ CD8α⁺ DC. A PDC relationship of the cells is supported further by the fact that they harbor characteristic D–J Ig gene rearrangements and that development of CX₃CR1⁺ CD8α⁺ DC requires E2-2, the critical transcriptional regulator of PDC. Thus, CX₃CR1⁺ CD8α⁺ DC represent a unique DC subset, related to but distinct from PDC. Collectively, the expression-profiling data of this study refine the resolution of previous DC definitions, sharpen the border of classical CD8α⁺ and CD8α⁻ DC, and should assist the identification of human counterparts of murine DC subsets.

Classic splenic dendritic cells (cDC), also termed “conventional DC,” are a subpopulation of mononuclear phagocytes defined in the mouse by high expression of the β integrin CD11c, migratory capacity, and an unrivalled ability to stimulate naïve T cells (1, 2). Beyond phenotypic and functional definitions, recent studies indicate that the short-lived CD11c^hi cDC are derived from dedicated nonmonocytic bone marrow-derived precursors termed “precDC” (3). Splenic cDC display considerable phenotypic heterogeneity, and their subsets are believed to have distinct functions in pathogen recognition and immunostimulation (4). Splenic CD11b⁺ cDC, which can be subdivided further into CD4⁺ and CD8/CD4 double-negative (DN) cDC, efficiently form peptide–MHC class II complexes (5). CD11b⁺ cDC secrete IL-10 and have been shown to induce T-helper cell type 2 CD4 responses preferentially (6). Development and/or maintenance of CD11b⁺ cDC require the transcription factors RelB (7), interferon regulatory factor 4 (IRF4) (8, 9), and RBP-J (10).

The second main murine cDC subset is characterized by the expression of CD8α homodimers and the C-type lectin CD205 (4). In vivo, CD8α⁺ cDC preferentially endocytose dying cells (11) and are considered specialized to cross-present engulfed cellular antigens in the context of MHC class I to CD8⁺ T cells (12). CD8α⁺ cDC have a predetermined capacity to secrete IL-12 (p70) and consequently stimulate T-helper type 1 CD4⁺ T-cell responses (13, 14). CD8α⁺ cDC also were reported to promote the development of T regulatory cells via production of TGF-β (15). Generation of CD8α⁺ cDC requires the transcription factors IRF8/ICSBP (16, 17) and Id2 (18, 19) and is specifically controlled by the transcription factor BatF3 (20).

In addition to cDC, lymphoid organs also harbor plasmacytoid DC (PDC) which are specialized in type I IFN secretion in response to viral challenge (21). PDC share a common developmental origin with cDC, although they branch off before the precDC (3, 22, 23), develop locally in the bone marrow, and are relatively long lived in the periphery (24, 25). PDC display a number of lymphocytic features, including the presence of Ig D–J rearrangements as the result of RAG protein expression during their development (26). The generation of PDC is controlled specifically by the transcriptional regulator E2-2 (27).

Here we report the characterization of a murine CD8α⁺ DC subset that is marked by high-level expression of the chemokine receptor CX₃CR1 and low expression of the costimulator CD86. CX₃CR1⁺ CD8α⁺ DC lacked hallmark features of classical CD8α⁺ DC, including the ability to produce IL-12, to cross-present antigen, and BatF3 dependence. Instead, their gene-expression profile, the presence of IgH gene rearrangements, and dependence on E2-2 define CX₃CR1⁺ CD8α⁺ DC as a steady-state DC population related to but distinct from plasmacytoid DC.

Results

Identification of the CX₃CR1⁺ CD8α⁺ DC Subset.

Flow cytometric analysis of spleen cells of Cx₃cr ^gfp mice—a transgenic mouse strain in which the gene encoding the CX₃CR1 chemokine receptor was replaced by an EGFP reporter gene (28)—allows the subdivision of splenic CD8α⁺ DC into CX₃CR1/GFP⁻ and CX₃CR1/GFP⁺ cells, respectively (Fig. 1A). Staining with a CX₃CL1-Fc fusion protein confirmed the expression of CX₃CR1 on the GFP⁺ DC but not on the GFP⁻ CD8α⁺ DC in Cx₃cr^gfp/+ mice (Fig. 1B). CX₃CR1⁺ CD8α⁺ DC generally comprised 15–30% of splenic CD8α⁺ DC in naïve adult C57BL/6 mice but reached 50% depending on the genetic background and housing facility. CX₃CR1⁻ and CX₃CR1⁺ CD8α⁺ DC populations also could be detected in lymph nodes of Cx₃cr ^gfp C57BL/6 mice (Fig. 1C) and Cx₃cr1^gfp BALB/c mice (Fig. 1D). Moreover CX₃CR1⁺ CD8α⁺ DC were present in CX₃CR1-deficient Cx₃cr1 ^gfp/gfp mice, indicating that the CX₃CR1 chemokine receptor is dispensable for their generation (Fig. 1E). In WT mice that lack the CX₃CR1/GFP label, the CX₃CR1⁺ subset of CD8α⁺ DC can be identified by low side scatter, low levels of the costimulatory molecule CD86, and the lack of the integrin CD103 (Fig. 1F).

Fig. 1.

Identification of the CX₃CR1⁺ CD8α⁺ DC subset. (A) Flow cytometric analysis of splenocytes from Cx₃cr1^gfp/+ mice. cDC were identified as CD11c^hi cells. Histogram shows GFP expression in CD11c^hi CD8α⁺ DC. (B) Staining of splenic cDC subpopulations for surface expression of CX₃CR1 using a CX₃CL1-Fc fusion protein. (C) Flow cytometric analysis of lymph node cells from Cx₃cr1^gfp/+ mouse. (D) Flow cytometric analysis of splenocytes from Cx₃cr1^gfp/+ BALB/c mouse. (E) Flow cytometric analysis of splenocytes from CX₃CR1-deficient Cx₃cr1^gfp/gfp mice. (F) Flow cytometric analysis of splenocytes from Cx₃cr1^gfp/+ mouse. Note characterization of CX₃CR1/GFP⁺ CD8α⁺ DC as a CD86^lo CD103^neg population.

CX₃CR1⁺ CD8α⁺ DC Lack Hallmark Features of Classical CD8α⁺ DC.

We first investigated whether CX₃CR1⁺ and CX₃CR1⁻ CD8α⁺ DC are distinct entities. CX₃CR1⁺ CD8α⁺ expressed high levels of MHC class II and could stimulate naïve alloreactive CD4⁺ T cells efficiently, establishing that they are bona fide DC (2) (Fig. 2A). Classical CD8α⁺ DC have been reported to be overrepresented in spleens of young mice (29). Interestingly, this age-related skewing of the cDC subset balance was restricted to the CX₃CR1⁻ CD8α⁺ cDC; CX₃CR1⁺ CD8α⁺ DC were found at equal frequencies in young and old mice (Fig. 2B).

Fig. 2.

CX₃CR1⁺ CD8α⁺ DC lack hallmarks of classical CD8α⁺ DC. (A) Mixed leukocyte reaction with indicated numbers of sorted splenic DC subsets isolated from Cx3cr1^gfp/+ C57BL/6 mice and BALB/c CD4⁺ T cells (10⁵). Data are representative of two experiments. (B) Flow cytometric analysis of splenic DC subsets in 3- and 8-wk-old Cx3cr1^gfp/+ C57BL/6 mice. Bars represent the percentages of CX₃CR1⁻ CD8α⁺ and CX₃CR1⁺ CD8α⁺ subsets out of total cDC (CD11c^hi cells). Data are representative of two experiments. (C) Analysis of IL-12p70 (Left) and IL-12p40 (Right) secretion by sorted splenic DC subsets in response to in vitro exposure to CpG. Data are representative of two experiments.(D) Selective deletion of CX₃CR1⁻ CD8α⁺ splenic DC by CytC injection. Bars represent the percentages of CX₃CR1⁻ CD8α⁺ and CX₃CR1⁺ CD8α⁺ subsets out of total DC (CD11c^hi cells). Data are representative of two experiments. (E) Flow cytometric analysis of splenic DC subsets of Cx3cr1^gfp/+ C57BL/6 mice bearing a WT tumor or a tumor secreting FLt3L. Bars represent the percentages of CX₃CR1⁻ CD8α⁺ and CX₃CR1⁺ CD8α⁺ subsets out of total cDC (CD11c^hi cells) (n = 3). Data are representative of two experiments. (F) Flow cytometric analysis of splenic DC from Batf3^+/+Cx3cr1^gfp/gfp, Batf3^+/−Cx3cr1^gfp/gfp, and Batf3^−/−Cx3cr1^gfp/gfp mice. cDC were gated as CD11c^hi B220⁻ cells.

A prominent functional feature of classical CD8α⁺ DC is IL-12 production in response to microbial challenge (13, 14). We previously reported that after mice were injected with Toxoplasma gondii tachyzoite extract only CX₃CR1⁻ CD8α⁺ DC, but not CX₃CR1⁺ CD8α⁺ DC, produce IL-12 (28). This difference could result from a general inability of the latter cells to produce IL-12 or their lack of the specific Toxoplasma sensor TLR11 (30). To address this issue, we sorted CX₃CR1⁺ CD8α⁺, CX₃CR1⁻ CD8α⁺, and CD8α⁻ cDC and exposed them in vitro to a Toll-like receptor 9 (TLR9) agonist. As seen in Fig. 2C, C–phosphate–G (CpG) stimulation in the presence of accessory cytokines induced IL-12 production as measured by secretion of the p70 subunit specifically by classical CD8α⁺ but not CX₃CR1⁺ CD8α⁺ or CD8α⁻ cDC. Interestingly, however, the TLR9 stimulus boosted production of the p40 subunit (indicative of IL-23) by all the populations tested, confirming the TLR9 responsiveness of the cells (Fig. 2C). This result establishes that, like cDC, CX₃CR1⁺ CD8α⁺ DC respond to TLR9 engagement but, unlike classical CD8α⁺ DC, do not produce IL-12.

The main functional hallmark of classical CD8α⁺ DC is their unique capability to channel exogenous antigens into the MHC class I presentation pathway for cross-presentation (12). Lew and colleagues recently established an elegant method that allows the identification of cross-presenting cells in the in vivo context based on their unique sensitivity to extracellular cytochrome c (CytC) (31). However, in this study only a fraction of splenic CD8α⁺ DC was depleted; the remainder was unable to cross-present and therefore was deemed functionally impaired, as indicated by impaired IL-12 production (31). To explore the possibility that the CytC-resistant CD8α⁺ DC were, in fact, CX₃CR1⁺ CD8α⁺ DC, we injected Cx₃cr ^gfp/+ mice with CytC. Flow cytometric analysis of the CytC-challenged mice revealed that only CX₃CR1⁻ CD8α⁺ cDC were ablated by this procedure, whereas the percentage of CX₃CR1⁺ CD8α⁺ DC remained unchanged (Fig. 2D). This result suggests that CX₃CR1⁺ CD8α⁺ DC lack characteristic cytosolic export mechanisms assigned to classical CD8α⁺ DC that allow them to cross-present antigens.

Classical CD8α⁺ DC are expanded preferentially upon systemic exposure to the growth factor Fms-related tyrosine kinase 3 ligand (Flt3L) (32). Interestingly, however, CX₃CR1⁺ CD8α⁺ DC were significantly underrepresented among CD11c^hi cells of mice bearing an Flt3L-secreting tumor (Fig. 2E). This finding suggests that CX₃CR1⁺ CD8α⁺ DC arise from developmental pathways distinct from those of classical CD8α⁺ DC. Development of the latter has been shown to require the basic leucine zipper transcription factor BatF3 (20). However, closer examination revealed that Batf3^−/− mice retain a sizable population of CD8α⁺ DC. Moreover, subsequent flow cytometric analysis of Cx₃cr1^gfp Batf3^−/− mice showed that these residual CD8α⁺ DC almost uniformly expressed CX₃CR1 (Fig. 2F). Together with the results of the Flt3L exposure, this result indicates that developmental requirements of CX₃CR1⁺ CD8α⁺ DC are distinct from those of classical CD8α⁺ DC. Collectively these findings establish that CD11c^hi CX₃CR1⁺ CD8α⁺ cells are DC but lack hallmark features assigned to classical CD8α⁺ DC, such as IL-12 production, the ability to cross-present, and developmental BatF3 dependence.

Gene Expression Profiling of Splenic DC Subsets.

To compare the expression profiles of CX₃CR1⁺ and CX₃CR1 CD8α⁺ DC and to study their relationship to the CD8α⁻ cDC subsets, we next isolated CD11c^hi DC from spleens of heterozygous mutant Cx₃cr1^gfp/+ C57BL/6 mice and sorted the two CD8α⁺ DC populations, as well as the CD4⁺ and DN cDC, to purity (Fig. 3 A and B). RNA was extracted from the four samples and subjected to gene-expression profiling using Mouse Genome 430.2 Affymetrix GeneChip arrays. Notably, all samples expressed equal levels of mRNAs encoding CD11c, MHC class II (I-A^b), Flt3, and CD40 (Table S1) but lacked expression of B- and T-cell markers such as CD19 and CD90. Moreover CX₃CR1, CD4, and CD8α mRNAs were all expressed by the appropriate populations; the expression of CD8α mRNA establishes that the CD8α molecules on the CX₃CR1⁺ population result from cell-intrinsic expression and not from passive acquisition from CD8⁺ T cells (33). Microarray results of selected genes were validated by RT-PCR analysis (Fig. S1).

Fig. 3.

Sorting strategy and gene-expression matrix of splenic cDC subsets. (A) Flow cytometric analysis of magnetic bead-enriched CD11c^hi cells isolated from Cx₃cr1^gfp/+ mice indicating sorting gates. (B) Analysis of sorted DC subsets. Percentages indicate purity of the respective cDC populations. (C) Expression matrix of the 500 modulated genes. Rows represent individual genes, and columns represent splenic cDC subsets. Colors indicate the relative expression levels of the genes in the different subsets, according to the code shown on the right.

Four-way comparison of all splenic DC subsets revealed only a few genes preferentially expressed in both CX₃CR1⁺ and CX₃CR1⁻ CD8α⁺ DC as compared with CD8α⁻ DC, including CD8α, CD24, and Sca1 (Table S2). Interestingly, however, expression of many hallmark proteins of classical CD8α⁺ DC, such as CD103, CD205, TLR3, XCR1, IL-15, and IL-12 (34), was absent from CX₃CR1⁺ CD8α⁺ DC (Table S3), supporting the notion that CX₃CR1⁺ CD8α⁺ DC are distinct entities. Rather, the microarray expression profile of CX₃CR1⁺ CD8α⁺ DC revealed a significant overlap with the profile of both CD4⁺ and DN cDC (Fig. 3C). Thus, molecules that have been associated with the DN cDC, including Sirpβ, Notch3, TLR5, TLR7, DCIR2, and CD209 (34), were all expressed at >3-fold higher levels in CX₃CR1⁺ than in CX₃CR1⁻ CD8α⁺ DC (Table S4). Notably, the expression of the transcription factors IRF4 and RelB, which are critical for the generation of CD4⁺ cDC but reportedly are dispensable for CD8α⁺ cDC production (8, 9), was also elevated (Table S4). Conversely, IRF8/ICSBP and Id2, reported to be essential for CD8α⁺ cDC development (16, 18, 19, 17), were prominently expressed in CX₃CR1⁻ CD8α⁺ cDC but were absent from CX₃CR1⁺ CD8α⁺ DC (Table S3). Collectively, these results establish that, according to their gene-expression profile, CX₃CR1⁺ CD8α⁺ DC are more closely related to CD4⁺ and DN cDC than to CD8α⁺ cDC.

CX₃CR1⁺ CD8α⁺ DC Share Gene Expression and Somatic Rearrangements with Plasmacytoid Dendritic Cells.

We next used the microarray data to define gene sets overexpressed in each population. These high-confidence sets then were compared with published expression profiles of DC populations (35). Principal component analysis (PCA) of the genes overexpressed specifically in the CD8⁺ CX₃CR1⁺ population revealed a striking enrichment for PDC-specific genes (Fig. 4A and Table S5), including the surface markers Klra17/Ly-49Q, SiglecH, Bst2/mPDCA-1, and Ly-6C, as well as the transcription factors Tcf4/E2-2 and SpiB. Thus, PDC-specific genes were highly enriched in the CX₃CR1⁺ subset, whereas classical CD8⁺ DC-specific genes were enriched only in the CX₃CR1⁻ subset (Fig. 4B). In further support of a potential link between PDC and CX₃CR1⁺ CD8α⁺ DC, we identified a significant list of genes that were silent in these two populations but were expressed in both CD8α⁺ and CD8α⁻ cDC (Table S6).

Fig. 4.

CX₃CR1⁺ CD8α⁺ DC share expression profile, somatic Ig gene rearrangements, and E2-2 dependence with PDC. (A) PCA of the genes overexpressed in CX₃CR1⁺ CD8α⁺ cells compared with CD8α⁺ cDC. Shown are the three principal component sets of genes analyzed against the expression data of Robbins et al. (35). Note the preferential expression in PDC (Top), in CD8α⁻ DC (Middle), and in both PDC and CD8α⁻ DC (Bottom). (B) Distribution of genes specific for CD8α⁺ DC or PDC in the two CD8α⁺ DC subsets. Shown is pairwise comparison of average probe intensities in CX₃CR1⁺ and CX₃CR1⁻ CD8α⁺ DC; genes specific for CD8α⁺ DC or PDC were selected based on reference 35. (C) IFN-α production by sorted splenic DC subsets after incubation for 20 h in the presence or absence of 400 HAU/mL influenza virus. (D) Expression of the TLR/IFN signaling genes in sorted CD8α⁻ cDC, PDC, and CX₃CR1⁺ CD8α⁺ DC as determined by quantitative RT-PCR (mean ± SD of triplicate reactions). Expression levels are normalized relative to the CX₃CR1⁺CD8α⁺ subset. (E) D–J rearrangement of IgH gene in sorted DC subsets. IgH D–J rearrangements were detected by genomic PCR using primer sets for the D_HQ52 element. Data are representative of three experiments. (F) Absence of CX₃CR1⁺ CD8α⁺ DC and PDC in absence of the transcription factor E2-2. (Left) Gated donor-derived (CD45.2⁺) splenocytes from E2-2^−/− chimeras or control E2-2^+/+ chimeras mice were analyzed for the presence of CD11c^int Bst2⁺ PDC. (Right) Gated CD11c^hi CD8α⁺ DC comprise the CD86^lo SSC^lo and the CD86^hi SSC^hi subsets (corresponding to CX₃CR1⁺ and CX₃CR1⁻ populations, respectively). Data are representative of three independent experiments.

A prime function of PDC is their production of type I interferons in antiviral responses (36). To probe for a functional overlap of CX₃CR1⁺ CD8α⁺ DC and PDC, we tested their response to viral challenge. As seen in Fig. 4C, only PDC, but not CX₃CR1⁺ CD8α⁺ DC, produced type I IFN upon in vitro exposure to influenza virus. In accordance with this finding, quantitative PCR analysis revealed that CX₃CR1⁺ CD8α⁺ DC express lower levels of TLR9, TLR7, and IRF-7 than do PDC (Fig. 4D). Moreover, CX₃CR1⁺ CD8α⁺ DC lacked surface expression of classical PDC markers such as B220 and mPDCA (Fig. S2). These findings establish that CX₃CR1⁺ CD8α⁺ DC are phenotypically and functionally distinct from PDC.

As a result of the activation of a unique “lymphoid” gene-expression program, PDC, unlike cDC, harbor Ig heavy-chain (IgH) D–J rearrangements that can serve as distinctive and stable genetic lineage markers (37, 26). To probe for a developmental connection between CX₃CR1⁺ CD8α⁺ DC and PDC, we investigated whether these two populations share this genetic feature by probing the rearrangement status of their IgH loci using a genomic PCR assay. As shown in Fig. 4E, D–J rearrangements were detected readily in both PDC and CX₃CR1⁺ CD8α⁺ DC, whereas classical CD8α⁺ DC and CD8α⁻ DC harbored only germline IgH alleles. This finding suggests that CX₃CR1⁺ CD8α⁺ DC are developmentally related to PDC.

We next decided to probe for potential developmental connections between the two cell populations. Because the transcription factor E2-2 is required for PDC generation (27), we tested whether CX₃CR1⁺ CD8α⁺ DC develop in the absence of E2-2. We found both PDC and CX₃CR1⁺ CD8α⁺ DC were absent in chimeras established from E2-2–deficient fetal livers (38) (Fig. 4F). We also examined a conditional strain in which E2-2 was deleted using a DC/PDC-specific CD11c-Cre transgene (10, 27). In these animals, some PDC develop in the bone marrow but disappear from the spleen because of spontaneous differentiation. In these animals, splenic PDC were absent, but SSC^low CD86^lo CD8α⁺ DC were present at the same frequencies as in the control animals (Fig. S3). These data suggest that CX₃CR1⁺ CD8α⁺ DC may be generated independently of mature PDC, possibly from a common bone marrow progenitor.

Collectively, these results show that CX₃CR1⁺ CD8α⁺ DC share with PDC a significant expression signature, the presence of unique Ig rearrangements, and the dependence on the transcription factor E2-2. However, they are functionally distinct from PDC in that they are not specialized in IFN-α production.

Discussion

Here we report the identification and characterization of a CX₃CR1⁺ CD8α⁺ DC subpopulation that coexists with cDC in lymphoid tissues of naïve mice and can comprise a major fraction of splenic and lymph node CD8α⁺ DC. CX₃CR1⁺ CD8α⁺ DC lacked hallmark features assigned to classical CD8α⁺ DC, such as the ability to produce IL-12 and cross-present, as well as the developmental dependence on the transcription factor BatF3. Moreover, their gene-expression profile also confirmed that CX₃CR1⁺ CD8α⁺ DC are distinct from CD8α⁺ cDC but rather resemble CD8α⁻ cDC. Comparative microarray analysis and the presence of characteristic Ig gene rearrangements established that CX₃CR1⁺ CD8α⁺ DC are ontologically related to PDC. This notion is supported further by the fact that development of both PDC and CX₃CR1⁺ CD8α⁺ DC depends on the transcription factor E2-2. However, conditional ablation of E2-2 using CD11c-Cre revealed that, in contrast to PDC, the maintenance of CX₃CR1⁺ CD8α⁺ DC does not require continuous E2-2 expression. Analysis of these conditional E2-2–deficient animals furthermore established that CX₃CR1⁺ CD8α⁺ DC can emerge without an obligatory PDC intermediate and that their numbers remain unaffected by the lack of splenic PDC. Notably, PDC have been reported to give rise to cells with phenotypic and functional cDC features upon microbial challenge in vitro and in vivo (25, 39, 40). Specifically, these PDC-derived DC were shown to express CD8α, but, unlike classical CD8α⁺ DC, to lack CD205 expression (25). However, our data argue that steady-state CX₃CR1⁺ CD8α⁺ DC arise from Rag-expressing immature bone marrow precursors shared with PDC, possibly the CD11c⁻ Ly-6C⁺ B220⁻ subset (27) (Fig. S4). Supporting this notion, Irf8^−/−-deficient mice that lack both classical CD8α⁺ cDC and PDC (16, 8, 41) were reported to retain a residual population of CD8α⁺ DC. Moreover, these cells share striking similarity with the CX₃CR1⁺ CD8α⁺ DC reported in this study, including low-level expression of CD86, absence of TLR3, and the inability to produce IL-12 (8). Together with our results of the constitutive and conditional E2-2 animals, this finding strongly suggests that CX₃CR1⁺ CD8α⁺ DC share developmental pathways with PDC but develop in absence of the latter.

Importantly, our analysis revealed that the expression of a number of genes previously assigned to classical CD8α⁺ DC (34) are, in fact, restricted to the CX₃CR1⁺ CD8α⁺ subset. By removing contaminating PDC-related CX₃CR1⁺ CD8α⁺ DC from the CX₃CR1⁻ CD8α⁺ cDC population, our study thus significantly refines the resolution of previous expression profile-based DC definitions (5, 34) and sharpens the border between classical CD8α⁺ and CD8α⁻ DC. Our findings should assist the identification of human correlates of murine DC subsets as exemplified by the recent reports on the human equivalent of classical mouse CD8α⁺ DC (42). Taken together, we identified and molecularly defined an additional steady-state DC subset in the mouse that is developmentally related to PDC but is functionally distinct. Future studies should reveal potential unique roles of CX₃CR1⁺ CD8α⁺ DC in pathogen recognition and immunostimulation.

Materials and Methods

Mice.

The following mice were used in this study: WT C57BL/6 mice; heterozygous and homozygous mutant Cx₃cr1^gfp C57BL/6 mice, Cx₃cr1^gfp BALB/c mice (28), and E2-2^+/− (38) and Batf3^−/− mice (20) crossed with Cx₃cr1^gfp mice. Fetal liver chimeras were established as described (10, 27) by intercrossing E2-2^+/− parents, one of which was Cx₃cr1^gfp/WT. All animals were maintained under specific pathogen-free conditions and handled according to protocols approved by each of the investigator's institutions in accordance with international guidelines.

Microarrays.

After collagenase D digestion, spleens from Cx₃cr1^gfp/+ C57BL/6 mice were enriched for CD11c⁺ cells by magnetic separation according to the manufacturer's protocol (Miltenyi Biotec GmbH). Splenic CD11c^hi cells were isolated using the FACS ARIA high-speed sorter (Becton-Dickinson). Total RNA was extracted and subjected to gene-expression profiling using the Mouse Genome 430.2 Affymetrix GeneChip. We then applied the Sorting Points into Neighborhoods (SPIN) algorithm, an unsupervised analysis tool for organization and visualization of the data. For pairwise comparison of the two CD8α⁺ DC subsets, microarray data from two independently sorted samples were combined and analyzed using NIA Array software (43).

Flow Cytometric Analysis.

Staining reagents used in this study included the phycoerythrin-coupled antibodies anti-MHC II, CD8, CD24, CD11c, CD103, and mPDCA1, the biotinylated antibodies anti-Ly6c, CD4, and CD8, the Allophycocyanin (APC)-coupled antibodies anti-CD11c, CD4, and CD8, the APC-Alexa750–coupled antibody anti-B220, and peridinin chlorophyll protein complex (PerCP)-coupled streptavidin. Unless indicated otherwise, the reagents were obtained from eBioscience, Biolegend, BD Biosciences, and Caltag. For CX₃CR1 surface staining, cells were incubated with an CX₃CL1-Fc fusion peptide (NTN-Fc; kindly provided by Millennium Biotherapeutics) and subsequently stained with Cy5-conjugated F(ab)₂ goat anti-human Ig G1 (IgG1; Jackson ImmunoResearch Laboratories). The cells were analyzed on a FACSCalibur cytometer (Becton-Dickinson) using CellQuest software (Becton-Dickinson).

Mixed Leukocyte Reactions.

We cultured 10⁴ and 5 × 10⁴ splenic DC (C57BL/6) with 10⁵ responder CD4⁺ T cells (BALB/c). Cultures were pulsed after 72 h with 1 μCi of [H³] thymidine, and incorporation was measured 16 h later.

In Vivo Exposure to FLT3L.

To test the impact of Flt3L exposure on the distribution of CD8α⁺ DC subsets, Cx₃cr1^gfp/+ C57BL/6 mice were inoculated with B16 tumor cells (3 × 10⁶) that had been manipulated to overexpress Flt3L (44).

In Vitro Cytokine Secretion Assay.

We cultured 5 × 10⁴ sorted DC subsets in 200 uL RPMI (Gibco/BRL) with CpG (5 ug/mL) for 24 h. Supernatant IL-12 levels were measured using IL-12 p40/p70 ELISA Abs (554476; BD Pharmingen) or the IL-12 p70 ELISA kit (88–7121-22; eBioscience). For IL-12 p70 production, the stimulation consisted of the cytokines IL-4 (200 ng/mL), GM-CSF (40 ng/mL), and IFN-γ (20 ng/mL).

Ablation of Cross-Presenting Cells.

To ablate cross-presenting cells, mice were injected i.v. with 7.5 mg equine CytC dissolved in PBS (C2506; Sigma) (31). Animals were analyzed 24 h after the CytC injection.

Diagnostic Genomic PCR for Ig Loci Rearrangements.

Genomic DNA was extracted from 2 × 10⁵ sorted cells by incubation with proteinase K at 55 °C for 2 h followed by heat inactivation for15 min at 95 °C. PCR primers specific for the D_HQ52 element were used to amplify rearranged IgH D–J as previously described (45).

Type I IFN Assay.

Sorted cells were cultured at 2 × 10⁶ cells/mL in complete RPMI 1640 for 20 h, with 400 hemagglutinin (HA) units/mL influenza virus A/Texas/1/77 (kindly provided by R. Arnon, The Weizmann Institute, Rehovot, Israel). Supernatants were assayed using an IFN-α ELISA kit (Performance Biomedical Laboratories).

Real-Time PCR.

Splenocytes from Cx₃cr1^gfp mice were stained with antibody conjugates, including Bst2-APC (Miltenyi Biotec) and sorted by a FACSAria flow sorter (BD Immunocytometry Systems) into CD11c^int Bst2⁺ (PDC), CD11c^hi CD8α⁻ CD11b⁺ GFP⁺ (CD8α⁻ DC), and CD11c^hi CD8α⁺ CD11b⁻ GFP⁺ (CX₃CR1⁺ CD8α⁺ DC) fractions. Cells were sorted directly into TRIzol LS reagent (Invitrogen). Total RNA from the sorted cells was extracted and reverse transcribed. Gene-expression levels were assayed by SYBR Green-based real-time PCR on an MX3000P instrument (Stratagene). All expression was normalized to β-actin and is presented relative to the CX₃CR1⁺ CD8α⁺ DC sample using the ΔΔC_T method. All primers were validated for linear amplification (sequences are available on request).