Batf3-Dependent CD11blow/− Peripheral Dendritic Cells Are GM-CSF-Independent and Are Not Required for Th Cell Priming after Subcutaneous Immunization

Dendritic cells (DCs) subsets differ in precursor cell of origin, functional properties, requirements for growth factors, and dependence on transcription factors. Lymphoid-tissue resident CD8α+ conventional DCs (cDCs) and CD11blow/−CD103+ non-lymphoid DCs are developmentally related, each being dependent on FMS-like tyrosine kinase 3 ligand (Flt3L), and requiring the transcription factors Batf3, Irf8, and Id2 for development. It was recently suggested that granulocyte/macrophage colony stimulating factor (GM-CSF) was required for the development of dermal CD11blow/−Langerin+CD103+ DCs, and that this dermal DC subset was required for priming autoreactive T cells in experimental autoimmune encephalitis (EAE). Here, we compared development of peripheral tissue DCs and susceptibility to EAE in GM-CSF receptor deficient (Csf2rb −/−) and Batf3 −/− mice. We find that Batf3-dependent dermal CD11blow/−Langerin+ DCs do develop in Csf2rb −/− mice, but that they express reduced, but not absent, levels of CD103. Further, Batf3 −/− mice lacking all peripheral CD11blow/− DCs show robust Th cell priming after subcutaneous immunization and are susceptible to EAE. Our results suggest that defective T effector priming and resistance to EAE exhibited by Csf2rb −/− mice does not result from the absence of dermal CD11blow/−Langerin+CD103+ DCs.


Introduction
GM-CSF regulates the development and activity of several myeloid cell types and influences both the initiation and maintenance of adaptive immune responses [1]. Mice lacking GM-CSF or its receptor show decreased antigen specific T cell priming against the encephalitogenic peptide of myelin oligodendrocyte glycoprotein (MOG  ) and are resistant to EAE [2]. Beyond this role in priming adaptive immune responses, GM-CSF acts to sustain ongoing effector responses, both in EAE initiated by MOG  peptide [2] and in collagen-induced arthritis [3]. Similarly, in a murine model of autoimmune myocarditis, GM-CSF acts during priming of Th17 responses by promoting IL-6 and IL-23 production from DCs and during ongoing autoimmune responses by promoting survival of autoreactive CD4 + T cells [4].
Several recent studies indicate that it is GM-CSF rather than IL-17 that is the primary Th17-derived cytokine responsible for the development of EAE [5][6][7]. Initially, it was thought that IL-17 production would explain why Th17 cells and RORct were required for EAE. However, in vivo neutralization of IL-17 only reduced EAE severity but did not eliminate disease, and non-IL-17 and non-IFN-c pathways have been implicated [8][9][10]. IL-23, but not IL-6 or TGF-b, was found to be critical for regulating the pathogenicitiy of Th17 cells [11,12]. GM-CSF derived from T cells, but not other CNS-resident or peripheral immune cells, was required for development of EAE [5], and acted during the effector phase to augment pathogenicity of Th17 cells [6,7]. IL-23 induced Th17 cells to increase production of GM-CSF, which further enhanced IL-23 production by antigen-presenting cells [6]. RORct was shown to induce GM-CSF, which acted on infiltrating myeloid cells entering the CNS, rather than resident microglia, during the development of EAE [7]. These studies suggest a model of EAE in which pathogenic T cells secrete GM-CSF that activates myeloid cells infiltrating the CNS to produce pathogenic lesions and further amplifies IL-23 production by DCs [13].
A recent report has suggested that the role for GM-CSF in EAE was based on its requirement for the development of dermal CD11b low/2 Langerin + CD103 + DCs [14]. Dermal CD11b low/2 Langerin + CD103 + DCs represent one anatomic subtype of peripheral tissue CD11b low/2 CD103 + DCs, all of which share with lymphoid tissue CD8a + cDCs developmental dependence on Batf3, Irf8, and Id2 [15,16]. CD8a + cDCs are cross-presenting DCs, characterized by consistent expression of DEC205, and variable expression of CD103 which can be regulated by GM-CSF, IL-3, and TGF-b1 [17][18][19]. Mice lacking GM-CSF or the GM-CSF receptor were reported to lack dermal CD11b low/2 Langerin + CD103 + DCs in the skin and peripheral lymph nodes under steady state and inflammatory conditions, and were resistant to EAE [14]. In addition, radiation chimeras expressing the human diphtheria toxin receptor (DTR) controlled by the Langerin promoter [20] and depleted of peripheral CD11b low/2 Langerin + CD103 + DCs were resistant to EAE and had reduced priming of MOG 35-55 -specific T cells [14]. This study concluded that GM-CSF was required for the initiation of EAE because it was required for the development of dermal CD11b low/2 Langerin + CD103 + DCs, which were required to prime pathogenic T cells [14].
To test whether loss of GM-CSF receptor protects from EAE by eliminating dermal CD11b low/2 Langerin + CD103+ DCs, we asked if Batf3 2/2 mice, which also lack this DC subset, are also protected from EAE. We find that Batf3 2/2 mice develop EAE after immunization and have increased numbers of MOG  peptidespecific T cells, excluding a requirement for either peripheral CD11b low/2 CD103 + DCs or lymphoid tissue-resident CD8a + DCs in EAE development. Direct comparison of Batf3 2/2 and GM-CSF receptor deficient mice revealed that GM-CSF signaling was not required for the development of dermal CD11b low/2 Langerin + CD103 + DCs, but rather regulated the expression level of CD103 on all peripheral DCs.
In draining lymph nodes, examination by ELISPOT assay showed that immunized Batf3 2/2 mice had an increased frequency of MOG 35-55 -specific T cells producing IL-17 and IFN-c (Fig. 1B), while Csf2rb 2/2 mice had a significantly reduced frequency of such cells, as previously reported [14]. Thus, Batf3 2/2 mice, which lack all peripheral CD11b low/2 CD103 + DCs [16], are able to prime pathogenic CD4 + T cells, and remain susceptible to EAE. These findings would disagree with the interpretation that resistance to EAE in Csf2rb 2/2 mice results from the absence of dermal CD11b low/2 Langerin + CD103 + DCs as was recently suggested [14].
While loss of Batf3 does not eliminate the population of DEC205 + CD8a + cDCs in SDLNs, these cells did not express CD103 normally (Fig. S1E). 9% of the DEC205 + CD8a + cDCs in SDLNs of control mice expressed CD103, while only 1.8% expressed CD103 from Batf3 2/2 mice. Notably, CD103 expres sion was also reduced on this population from SDLNs of Csf2rb 2/2 mice. Conceivably, the persistence of DEC205 + CD8a + cDCs in SDLNs on the C57BL/6 background could result from genetic polymorphisms that impact the redundancy of Batf3 in the development of these cells. Alternately, variable inflammation between strains could be involved, since infection with Toxoplasma gondii can regulate the size of splenic CD11b low/2 CD103 + DC populations [28].
Basal CD103 expression by CD8a + -equivalent cDCs requires Batf3 but not GM-CSF signaling While CD103 has been used as a marker for the dermal CD11b low/2 Langerin + DC lineage [14], recent work shows that CD103 expression can be dynamically regulated by cytokines [17,18]. Therefore, we wished to re-examine the relationships between CD103 expression, Batf3, and GM-CSF signaling. We selected the Flt3L-induced DC differentiation from bone marrow cultures for this purpose since it has allowed the analysis of CD103 expression as dependent on both Batf3 and GM-CSF [17,18] (Fig. 4).
Previously, CD24 high SIRPa low DCs were considered an ''equivalent'' of CD8a + cDCs in Flt3L-treated bone marrow cultures [30], despite the fact that only a subpopulation of these cells express CD103 (Fig. 4). The CD103 + fraction of CD24 high SIRPa low DCs was more potent than the CD103 2 fraction for cross-presentation [17], suggesting a correlation between CD103 expression and CD8a + cDC function. We find that addition of GM-CSF during day 7 through 9 of Flt3L culture induces CD103 expression on all CD24 high SIRPa low DCs (Fig. 4), in agreement with recent reports which showed that such GM-CSF-induced cells exhibit robust cross-presentation [17,18]. However, while addition of IL-3 and TGF-b1 to Flt3L cultures can both induce expression of CD103, only IL-3 induced the capacity for crosspresentation [17], dissociating CD103 expression from crosspresentation. Likewise, we found that GM-CSF treatment induced CD103 expression on all CD24 high SIRPa low DCs, even in bone marrow cultures from Batf3 2/2 mice (Fig. 4), which completely lack CD103 + DCs in vivo [16], further dissociating CD103 expression from the properties of true CD8a + cDC equivalents. In agreement, a recent report found that GM-CSF-induced CD103 + CD24 high SIRPa low Flt3L-derived DCs from Batf3 2/2 mice lacked the capacity for cross-presentation [18]. Finally, Csf2rb 2/2 bone marrow generated a subpopulation of CD103 + CD24 high SIRPa low DCs, indicating development that is independent of GM-CSF signaling, although whether these arise from the actions of cytokines such as IL-3, or arise independently of soluble factors present in the culture remains unknown. In summary, these results indicate that CD103 expression can be regulated by GM-CSF, independently of Batf3, and may not represent a static marker of Batf3-dependent DC lineages.

Discussion
Our study makes two new observations. First, we show that CD11b low/2 non-lymphoid DCs develop in the absence of GM-CSF signaling. Second, we show that Batf3-dependent CD11b low/2 non-lymphoid DCs are not required for T effector cell priming or clinical EAE after subcutaneous immunization. King et al. recently concluded that mice deficient for GM-CSF or GM-CSF receptor were resistant to EAE because they lacked dermal CD11b low/2 Langerin + CD103 + DCs, which they proposed were required for priming effector CD4 T cells with MOG 35-55 [14]. Dermal CD11b low/2 Langerin + CD103 + DCs are a conventional DC subset that requires the transcription factor Batf3 for their development [16]. Here, we directly compared Csf2rb 2/2 with Batf3 2/2 mice for their susceptibility to EAE and for their development of various DC subsets using a broad panel of DC markers, including Langerin, CD11b, EpCAM, CD24, and CD103. We confirm that the dermal CD11b low/2 Langerin + CD103 + DC subset is absent in Batf3 2/2 mice, but surprisingly find that Csf2rb 2/2 mice harbor normal numbers of this DC subset in SDLNs as a CD11b low/2 Langerin + EpCAM mid CD24 high cell that expresses reduced, but not absent, levels of CD103. Moreover, in other peripheral tissues, Csf2rb 2/2 mice have similar CD11b low/2 CD103 + DCs that express reduced, but not absent, levels of CD103.
Our data agree with recent reports that GM-CSF directly regulates CD103 expression on Flt3L-derived CD8a + equivalent cDCs and splenic DEC205 + CD8a + cDCs [17][18][19]. Beyond this, we show that GM-CSF also regulates levels of CD103 expression on peripheral tissue-resident CD11b low/2 DCs, which were not examined in the previous studies. Notably, a separate report has suggested that GM-CSF signaling is required for the development of small intestinal lamina propria CD11b + CD103 + DCs [31], although their data could be reinterpreted as simply representing reduced CD103 expression by these DCs. In that report, Csf2rb 2/2 mice had half as many lamina propria CD11b + CD103 + DCs as wild type mice, but 1.7-fold more CD11b + CD103 2 DCs than wild type mice, consistent with a reduction of CD103 expression as an explanation for the ''missing'' subset.
King et al. also found decreased T cell priming to subcutaneous immunization with MOG 35-55 peptide and partial EAE resistance after in vivo depletion of peripheral Langerin + CD103 + DCs using radiation chimeras reconstituted with bone marrow from transgenic mice expressing DTR under the control of the Langerin promoter [20]. The use of radiation chimeras allowed the selective depletion of dermal CD11b low/2 Langerin + DCs, avoiding depletion of epidermal CD11b + Langerin + Langerhans cells, because the latter are radiation-resistant, and were therefore of host origin and did not express the DTR. Our results indicate that Csf2rb 2/2 mice are not deficient in dermal CD11b low/2 Langerin + DCs, but rather that these cells have reduced CD103 expression. However, Batf3 2/2 mice lack this dermal DC subset, and yet are susceptible to EAE induction and display increased, not decreased, T cell priming to MOG  peptide immunization. It is unclear why DTR-mediated depletion of donor-derived, radiation-sensitive Langerin + DCs reduces priming and delays EAE, when genetic ablation of these cells in Batf3 2/2 mice leaves priming and EAE intact. Their use of DTR-mediated depletion involved irradiation of mice prior to immunization, which may have had unintended consequences. Furthermore, a study by Henri et al. has identified additional dermal DC subsets [26], including one that is Langerin + but CD103 2 , although it's Batf3-dependence was not characterized. Conceivably, the depletion of this subset by diphtheria toxin treatment could contribute to differences between our results and those of King et al. Notably, DTR-mediated depletion reduced the severity of EAE, but with a profound decrease only in MOG 35-55specific IFN-c-secreting T cells, with minimal change in the number of IL-17-secreting T cells. In contrast, MOG 35-55 -immunized Csf2rb 2/2 mice had profound reductions in antigen-specific IFNc-secreting and in IL-17-secreting T cells, suggesting that Csf2rb 2/2 mice differ from mice in which Langerin + DCs are depleted using DTR.
Our results suggest that the absence of dermal CD11b low/2 Langerin + CD103 + DCs leads to augmented T cell priming of both Th17 and Th1 responses. Igyártó et al. have recently shown that skin infection of Batf3 2/2 mice with recombinant Candida albicans led to reduced Th1 but increased Th17 responses compared to control mice [32]. These authors demonstrated that dermal Langerin + CD103 + DCs expressed IL-12, explaining their ability to promote Th1 cell responses, and IL-27, a cytokine known to inhibit Th17 cell responses [33,34]. While we found increased Th17 responses in Batf3 2/2 mice immunized with MOG  and CFA, we also observed increased Th1 responses compared to control Figure 3. GM-CSF controls the level of CD103 expression on Batf3-dependent CD11b low/2 peripheral DCs. (A) FACS analysis of lung DCs from C57BL/6, Batf3 2/2 (C57BL/6 background), and Csf2rb 2/2 mice. Left plots are gated on live, single cells. The second column of plots is gated on CD11c + CD45.2 + cells, consisting of a mixture of lung macrophages and DCs. Note that in Csf2rb 2/2 mice, lung macrophages lack CD11c expression [38], and therefore this gate includes only DCs. The third and fourth columns of plots are gated on I-A bhigh autofluorescence low DCs. Numbers represent the percentage of cells within the indicated gates. Data are representative of six individual lungs from three mice per genotype over two experiments. (B) FACS analysis of kidney DCs from C57BL/6, Batf3 2/2 (C57BL/6 background), and Csf2rb 2/2 mice. Left plots are gated on live, single cells. The second column of plots is gated on CD11c high CD45.2 + cells, consisting of DCs. The third column of plots is gated on CD11c high I-A b+ DCs. Numbers represent the percentage of cells within the indicated gates. Data are representative of six individual kidneys from three mice per genotype over two experiments. doi:10.1371/journal.pone.0025660.g003 Figure 4. GM-CSF induces CD103 expression on Flt3L-derived CD8a-equivalent DCs. FACS analysis of developing DCs in bone marrow cells from C57BL/6, Batf3 2/2 (C57BL/6 background), and Csf2rb 2/2 mice cultured with Flt3L for 9 days, with or without the addition of GM-CSF to the culture on day 7. The left column is gated on live single cells, the second on CD11c + CD45RA 2 cDCs, the third on I-A b+ cDCs, and the fourth on CD24 high Sirpa low cells. Numbers represent the percentage of cells within the indicated gates. Data are representative of bone marrow cells from two mice per genotype over two experiments. doi:10.1371/journal.pone.0025660.g004 mice. Conceivably, the different Th1 responses between our study and Igyártó et al. could arise from differences in the form of immunization, relying on CFA versus fungal infection. Notably, blood-derived inflammatory monocytes have been reported to serve as a critical source of IL-12 after CFA-based subcutaneous immunization [35]. While we observed between a 2 to 3-fold higher frequency of IL-17-producing T cells in MOG 35-55immunized Batf3 2/2 mice relative to controls, the clinical EAE scores were similar between these groups. Conceivably, the frequency of IL-17-producing T cells is unrelated to the severity of EAE when Th17 cells are above some threshold frequency. Alternatively, IL-17 itself may not be the limiting factor in regulating the intensity of EAE, since GM-CSF also recently was implicated as the major Th17-derived cytokine responsible for driving pathogenic lesions in this model [5][6][7]. In fact, however, we do observe a slightly more rapid onset of EAE in Batf3 2/2 mice, which was more pronounced in the 129S6/SvEv genetic background (Fig. 2D), which could relate to increased Th17 development.
If a lack of peripheral CD11b low/2 DCs does not abrogate priming of T effector cells after subcutaneous immunization, why then does this priming defect occur in Csf2rb 2/2 mice? One possibility is that other DC subsets besides dermal CD11b low/2 DCs can prime CD4 T cells [35], but that the ability of these other DCs to prime requires that they receive a GM-CSF signal. Indeed, GM-CSF is essential for IL-6 and IL-23 production by DCs during the priming phase of autoimmune myocarditis [4] and is required for inflammatory DC development from monocytes in a model of repeated immunization with methylated BSA and CFA [36]. T cells do not express GM-CSF receptor, so an intrinsic T cell defect seems unlikely. While the source of GM-CSF after subcutaneaous immunization is still unclear, antigen-specific T cells can produce GM-CSF [5][6][7] and may augment responses by DCs during initial priming of T effector cells.

Ethics statement
This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Animal Studies Committee of Washington University (#20090320).

Mice
Wild-type 129S6/SvEv and C57BL/6 mice were purchased from Taconic. Batf3 2/2 mice on a 129S6/SvEv background were previously generated in our laboratory [27], and were backcrossed for 10 generations to the C57BL/6 background. Csf2rb 2/2 mice on the C57BL/6 background were purchased from Jackson Laboratory. Experiments were performed with sex-matched mice at 8-20 weeks of age. Mice were bred and maintained in our specific pathogen-free animal facility according to institutional guidelines.

Induction of EAE
EAE was induced as previously described [14]. Briefly, mice were immunized subcutaneously with 100 micrograms MOG  peptide (Sigma Genosys) emulsified in complete Freund's adjuvant (CFA) (made with 5 mg/ml heat-killed Mycobacterium tuberculosis H37Ra (BD Difco) in incomplete Freund's adjuvant (BD Difco)). Pertussis toxin (List Biological Laboratories) was injected intraperitoneally (300 ng) on days zero and two. Mice were observed for signs of EAE and graded on a standard 0-5 scale as described [37].

T cell priming and ELISPOT assays
Mice were immunized subcutaneously in the hind footpads with 10 nanomoles MOG  peptide (Sigma Genosys) emulsified in complete Freund's adjuvant (CFA) (BD Difco). Popliteal lymph nodes were collected at day 7, and single cell suspensions were used in ELISPOT assays with the IFNc ELISPOT antibody pair from BD Biosciences on Multiscreen Filter Plates from Millipore. An IL-17 ELISPOT was designed using anti-IL17A capture (clone TC11-18H10) and detection (clone TC11-8H4.1) antibodies (BD Biosciences). Cells were plated at 1610 6 cells/well, in duplicate, and stimulated with either no antigen, or 10 micromolar MOG  peptide. Plates were developed after 16 hours of stimulation at 37uC, and spots were counted on an Immunospot counter (Cellular Technology Ltd.).

Flow cytometry
Inguinal lymph nodes and spleens were minced and digested in 5 ml Iscove's modified Dulbecco's media +10% FCS (cIMDM) with 250 mg/ml collagenase B (Roche) and 30 U/ml DNase I (Sigma-Aldrich) for 1 h at 37uC with stirring. Lungs and kidneys were perfused with 10 ml DPBS via injection into the right ventricle after transection of the lower aorta. Dissected lungs and kidneys were minced and digested in 5 ml of cIMDM with 4 mg/ ml collagenase D (Roche) for 1 h at 37uC with stirring. EDTA (5 mM final concentration) was added to cell suspensions, and cells were incubated on ice for 5 min. Cells were passed through an 80 mm strainer before red blood cells lysis with ACK lysis buffer. Cells were counted on a Vi-CELL analyzer, and 2-5610 6 cells were used per antibody staining reaction.
Staining was performed at 4uC in the presence of Fc Block (clone 2.4G2, BD Biosciences or BioXcell) in FACS buffer (DPBS+0.5% BSA+2 mm EDTA+0.02% sodium azide). For experiments involving intracellular anti-Langerin staining, cells were surface antibody stained prior to fixation in 4% formaldehyde (Thermo Scientific) for 15 minutes at room temperature. Cells were subsequently permeabilized (DPBS+0.1% BSA+0.5% saponin) for 5 min at 4uC and stained in the same permeabilization buffer with anti-Langerin antibody. Cells were washed twice in permeabilization buffer before returning them to FACS buffer. Flow cytometry was performed on a FACSCantoII (BD Biosciences), and data was analyzed with FlowJo software (Tree Star, Inc.).

Bone marrow cultures
Bone marrow cells from femurs and tibias were collected and red blood cells were lysed in ACK lysis buffer. Cells were cultured in four ml cIMDM in six-well plates at 2610 6 cells/ml containing 160 ng/ml murine Flt3L (Peprotech). At day seven of culture, murine GM-CSF (Peprotech) was added to some wells at 20 ng/ ml. Non-adherent cells were collected at day nine for flow cytometry. Figure S1 Flow cytometric analyses of DCs in Batf3 2/2 mice. (A) FACS analysis of SDLN (inguinal) DCs from Batf3 +/+ and Batf3 2/2 mice on both the C57BL/6 and 129S6/SvEv backgrounds. Left plots from mice of each background are gated on CD11c + BST2 2 cDCs. Right plots from mice of each background are gated on migratory (DEC205 + CD8a 2 ) DCs. Numbers represent the percentage of cells within the indicated gates. Data are representative of at least six mice per genotype and background obtained in several independent experiments. (B) Absolute cell numbers of DEC205 + CD8a + DCs per individual inguinal lymph nodes from C57BL/6 and Batf3 2/2 mice (C57BL/6 background) calculated from the FACS analysis performed in Figure 2A. Horizontal bars represent the geometric mean, p value by unpaired student's t test. (C) and (D) FACS analysis of splenic DCs from Batf3 +/+ and Batf3 2/2 mice on both the (C) C57BL/6 and (D) 129S6/SvEv backgrounds. Plots are gated on CD11c + I-A b+ cDCs. Numbers represent the percentage of cells within the indicated gates. Data are representative of at least six mice per genotype and background obtained in several experiments. (E) FACS analysis of SDLN (inguinal) DCs from C57BL/6, Batf3 2/2 (C57BL/6 background), and Csf2rb 2/2 mice. Left plots are gated on CD11c high cells. Right plots are gated on DEC205 + CD8a + cDCs. Numbers represent the percentage of cells within the indicated gates. Data are representative of eight to ten individual inguinal lymph nodes from four to five mice per genotype obtained in two experiments. Data is depicted in these plots are derived from analysis of the same mice used in Figure 2A, but gated to analyze CD103 expression on DEC205 + CD8a + cDCs. (TIF)

Supporting Information
Author Contributions