Bone marrow–resident DC progenitors

Bone marrow–derived DC precursors or MDPs do not express lineage markers (Lin^- : CD3^- CD19^- NK1.1^- I-A^b- CD11c^- B220^- Ter119^- CD11b^- Gr-1^- ) and were originally identified by their expression of the stem cell factor receptor c-Kit (CD117) and a transgene encoding green fluorescent protein (GFP) driven by the promoter of the gene encoding the chemokine receptor CX3CR1, which has made analysis of mutant mouse strains cumbersome⁵. MDPs can also give rise to monocytes and some tissue macrophages^{5, 6}; CDPs (Lin^- c-Kit^intFlt3⁺CSF1R⁺) give rise to pDCs and cDCs in vivo^{7, 8}. To further define DC progenitors, we purified Lin^- cells expressing the receptor for colony-stimulating factor 1 (CSF1R) and analyzed their developmental potential in vitro and in vivo (Fig. 1).

Figure 1: Identification of cDC progenitors.

(a) Flow cytometry of the expression of c-Kit and Sca-1 (left) by 2.5 10⁵ Lin^- (CD3^- CD19^- B220^- NK1.1^- CD11c^- CD11b^- Ter119^- Gr-1^- ) bone marrow cells (BM; top) and 3 10⁵ Lin^- spleen cells (bottom), and CSF1R expression by Sca-1^- cells (middle) and Sca-1⁺ cells (right). Numbers in plots at left indicate percent Sca-1^- cells (middle) or Sca-1⁺ cells (right); numbers in outlined areas (middle and right) indicate percent CSF1R⁺ cells. This analysis was done twice (spleen) or over 50 times (bone marrow). (b) Colony growth of total bone marrow cells, Lin^- CSF1R⁺ cells and Lin^- CSF1R^- cells cultured in semisolid medium in the presence of erythropoietin (Epo) and IL-3 (BFU-E (burst-forming unit–erythroid)) or erythropoietin alone (CFU-E (colony-forming unit–erythroid)); growth in GM-CSF, CSF1 or Flt3L was analyzed after 7 d. These assays were done four times in duplicate. (c) Flow cytometry of the expression of Gr-1 (PMNs), F4/80 (red pulp macrophages (RP-Mp)), PDCA1 (pDCs), CD11b, CD11c and CD8 (monocytes (Mono) and cDCs) by spleen cells 14 d after adoptive transfer of purified Lin^- CSF1R⁺c-Kit^- cells (middle) or Lin^- CSF1R⁺c-Kit⁺ cells (bottom), with gating on live (DAPI^- ) CD3^- CD19^- NK1.1^- Ter119^- donor cells (CD45.2⁺). Dark green dots indicate CD8⁺ cDCs; light green dots indicate CD8^- cDCs. Numbers along right margin indicate percent cells of each type (identified by color in plots). Data are representative of ten transfers. (d) Flow cytometry³³ of the expression of CX3CR1, c-Kit and CSF1R (top) or CSF1R, c-Kit and Flt3 (bottom) by Lin^- bone marrow cells, indicating the phenotype correlation among Lin^- c-Kit⁺CX3CR1⁺ cells (MDP^*; original MDP definition⁵), Lin^- CSF1R⁺ cells (MDP; as reported here) and CDPs⁷. Numbers in plots indicate percent cells in outlined areas (left) or quadrants (right). This analysis was repeated twice. (e) Flow cytometry of the expression of CD11b and CD1c by donor-derived spleen cells at 14 d after transfer of purified Lin^- CX3CR1⁺c-Kit⁺ cells (MDPs⁵; top), Lin^- CX3CR1⁺c-Kit^- cells (middle) or Lin^- CSFR1⁺c-Kit^- cells (bottom) into sublethally irradiated congenic wild-type recipient mice, with gating on donor-derived cells (CD45.2⁺). Outlined areas indicate monocytes and cDCs. These transfers were done twice.

Full size image (133 KB)

We found that about 8% of all Lin^- Sca-1^- cells expressed CSF1R in the bone marrow (absolute numbers, 7.2 10⁴ 0.3 10⁴ per two femurs; n = 7 mice; Fig. 1a). This fraction was rapidly proliferating (35% 4% in S or G2 phase; n = 4 mice; Supplementary Fig. 1 online) and did not include early self-renewing hematopoietic precursors (Sca-1⁺ cells; Fig. 1a). Lin^- CSF1R⁺ cells responded to granulocyte-macrophage colony-stimulating factor (GM-CSF) in vitro by giving rise to DCs, which were Gr-1^- MHCII⁺CD11c⁺ or Gr-1^- MHCII⁺CD11c^- (Supplementary Fig. 2 online) but failed to respond to CSF1 or to erythropoietin with or without interleukin 3 (IL-3; Fig. 1b). Although most cells in the cultures treated with GM-CSF were DCs, there was little expansion (threefold), which indicated that this progenitor population undergoes only a limited number of cell divisions before differentiation in response to GM-CSF (Lin^- c-Kit⁺CSF1R^- cells expanded 18-fold). Consistent with published work⁷, in vitro colony formation from single cells in response to Flt3L was undetectable (Fig. 1b). We conclude that Lin^- CSF1R⁺ cell populations are devoid of early and late erythroid progenitor cells and they produce DCs in response to GM-CSF in vitro. In contrast, Lin^- CSF1R^- cells responded to GM-CSF in vitro by giving rise to cells with diverse phenotypes (Fig. 1b and Supplementary Fig. 2).

In adoptive transfer experiments, purified Lin^- CSF1R⁺ cells (Supplementary Fig. 3 online) gave rise exclusively to spleen CD8⁺ CD8^- cDCs and SIGNR1⁺ marginal zone macrophages but not to pDCs, PMNs, red pulp macrophages (F4/80⁺) or metallophilic macrophages (Ser-4⁺), regardless of c-Kit expression (Fig. 1c and Supplementary Fig. 4 online). In contrast, Lin^- c-Kit⁺CSF1R^- cells gave rise to all other myeloid cell types (data not shown). We detected no donor-derived cells in the bone marrow of recipients (data not shown), which suggested that Lin^- CSF1R⁺ cells cannot migrate back to their site of origin.

MDPs were originally defined as Lin^- cells expressing a CX3CR1 promoter–driven GFP transgene, c-Kit⁵ and CSF1R⁶ (Fig. 1d). Lin^- CX3CR1⁺ cells included c-Kit⁺ and c-Kit^- cells and both populations had identical DC potential in adoptive transfer experiments in vivo (Fig. 1e); thus, Lin^- CX3CR1⁺c-Kit⁺ and Lin^- CX3CR1⁺c-Kit^- cells were MDPs (Fig. 1e). Only 44% of Lin^- CX3CR1⁺ cells expressed CSF1R (21% c-Kit⁺ and 23% c-Kit^- ; Fig. 1d), but all Lin^- CSF1R⁺ cells were CX3CR1⁺ (data not shown). We found that 75% of Lin^- CSF1R⁺ cells (36% c-Kit⁺ and 39% c-Kit^- ) expressed Flt3 (Fig. 1d) and, when separated on the basis of Flt3 expression, cDC progenitor potential was 'concentrated' in Flt3⁺ cells in vivo (data not shown). Lin^- CSF1R⁺Flt3^- cells failed to give rise to any other cell type (data not shown). CDPs⁷ (Lin^- c-Kit^intFlt3⁺CSF1R⁺) represented less than 39% of Lin^- CSF1R⁺ cells and they are therefore included in MDPs (Fig. 1d). We conclude that bone marrow–derived Lin^- CSF1R⁺ cells give rise to cDCs in vivo and that for analysis of Flt3-mutant mice (discussed below), the Lin^- CSF1R⁺ phenotype is sufficient to define MDPs.

Granulocyte-monocyte progenitors (GMPs), common myeloid progenitors (CMPs) and common lymphoid progenitors all have cDC progenitor activity^{17, 18, 19}. To determine how MDPs relate to those other progenitor fractions, we analyzed the expression of CD34 and the transmembrane receptor FcRII/III (CD16/32) by Lin^- c-Kit⁺CX3CR1⁺ (ref. 5) and Lin^- CSF1R⁺ MDPs (Supplementary Fig. 5 online). We found that the expression profiles of both sets of MDPs overlapped those of GMPs and CMPs and, because of the restricted in vivo developmental potential of MDPs, we conclude that GMPs and CMPs are mixed progenitor populations that contain MDPs¹⁹.

Effects of Flt3L on MDPs

To determine how Flt3L affects DC development, we examined the effects of its administration on bone marrow and peripheral DCs. After administration of Flt3L, the expression of its receptor, Flt3, was downregulated on MDPs and on spleen cDCs, which indicated a rapid response to Flt3L in all DC compartments (Fig. 2a and data not shown). MDPs (Lin^- CSF1R⁺; Fig. 2b) and Lin^- c-Kit⁺CX3CR1⁺ cells in mice transgenic for the CX3CR1 promoter–driven GFP transgene (Supplementary Fig. 6 online) were higher by a factor of about ten, whereas GMPs and CMPs were higher by a factor of only about five, and megakaryocytic and erythroid-prone progenitor cells were unaffected¹² (Supplementary Figs. 6 and 7 online). However, treatment with Flt3L did not alter the developmental potential of MDPs on a 'per-cell' basis, as shown by adoptive transfer experiments in which MDPs from untreated and Flt3L-treated mice had the same ability to give rise to cDCs (Fig. 2c). We conclude that Flt3L increases the number of MDPs but does not alter their ability to give rise to DCs.

Figure 2: Effects of stimulation with Flt3L.

(a) Flow cytometry of spleen and bone marrow cells from mice injected daily for 8 d with PBS (- ; left) or Flt3L (right), gated on CD3^- CD19^- NK1.1^- Ter119^- cells (spleen) or Lin^- (CD3^- CD19^- B220^- NK1.1^- CD11b^- CD11c^- Gr1^- Ter119^- ) Sca-1^- cells (bone marrow). Numbers above plots indicate percent of each myeloid cell type (identified by color in plots). This analysis was done four times. (b) Absolute numbers of cDCs, pDCs, PMNs, red pulp macrophages, monocytes, T cells (Tc) and B cells (Bc) (top) and MDPs in bone marrow (bottom) in response to sustained Flt3L stimulation (8 d). Data are representative of two independent experiments. (c) Developmental potential of Lin^- CSF1R⁺ cells from mice left unstimulated (–) or stimulated for 6 d with Flt3L (+); cDC yield was calculated on the basis of 1 10⁵ injected cells. Data are representative of four independent experiments. (d) Contribution of cells from mice stimulated for 6 d with Flt3L (open bars; CD45.2⁺) and from congenic untreated mice (filled bars; CD45.1⁺) to CD3^- CD19^- B220^- NK1.1^- Ter119^- CD11b^- CD11c^- Gr1^- immature spleen cells (Input) and to spleen cDCs, PMNs and monocytes at 6 d after adoptive transfer of mixed cDC-depleted spleen cell populations to irradiated recipients (CD45.1⁺CD45.2⁺). Data are representative of two independent experiments.

Full size image (98 KB)

In the spleen, absolute numbers of cDCs, pDCs, PMNs, monocytes and red pulp macrophages were increased in response to Flt3L (Fig. 2a,b). To determine whether Flt3L injection also altered the number of DC precursors in the spleen, we did adoptive transfer experiments with mixtures of untreated and Flt3L-treated splenocyte populations depleted of DCs (Fig. 2d). Because the donor populations were depleted of DCs, newly generated cDCs developed from intrasplenic precursors. The relative increase in DCs from the Flt3L-treated spleens suggested that DC progenitors were mobilized to the spleen in Flt3L-treated mice (Fig. 2d).

Fewer cDCs and pDCs in Flt3^{- /-} mice

To further investigate the function of Flt3 in DC development, we examined Flt3-deficient mice (Flt3^{- /-} mice¹⁷; Fig. 3). The absence of Flt3 led to fewer pDCs (Fig. 3a,b) and cDCs (Fig. 3c), which was more notable in younger mice^{20, 21}, but MDP numbers (Fig. 3j) and spleen pre-cDC numbers²² (Fig. 3h and Supplementary Fig. 7) remained unaltered. The finding of fewer cDCs and pDCs was not due to a general defect in myeloid cells, as monocyte, PMN, Ser-4⁺ metallophilic macrophage and SIGNR1⁺ marginal zone macrophage numbers were not significantly altered (Fig. 3d–f and Supplementary Fig. 8 online). Unexpectedly, we found that double-mutant mice lacking Flt3 and the GM-CSF receptor did not have a further diminution in pDCs or cDCs (Supplementary Fig. 9 online). Thus, the GM-CSF receptor is not required for DC development in vivo, even in the absence of Flt3. We conclude that there is a specific DC defect in the spleens of Flt3^{- /-} mice.

Figure 3: Myeloid compartments of Flt3^{- /-} mice.

(a) Flow cytometry of the relative contributions of myeloid cells to spleen cells in 4-week-old mice wild-type mice (WT) and Flt3^{- /-} mice (KO), for cells gated on CD3^- CD19^- NK1.1^- Ter119^- cells. Numbers above plots indicate percent of each myeloid cell type (identified by color in plots). (b–f) Absolute numbers of pDCs (CD3^- CD19^- NK1.1^- Ter119^- Gr-1^- PDCA1⁺; b), cDCs (CD3^- CD19^- NK1.1^- Ter119^- Gr-1^- PDCA1^- F4/80^- CD11c⁺; c), monocytes (CD3^- CD19^- NK1.1^- Ter119^- PDCA1^- F4/80^- CD11c^- Gr-1^neg–loCD11b⁺; d), PMNs (CD3^- CD19^- NK1.1^- Ter119^- Gr-1⁺; e) and spleen cells (f). (g,h) Absolute numbers of intrasplenic cDC progenitors²², assessed by measurement of medium-density spleen cell numbers (MDC; g) and calculation of pre-cDC numbers (CD3^- CD19^- NK1.1^- Ter119^- CD45RA^neg–loCD11c⁺CD43⁺SIRP-^lo; h). (i,j) Absolute bone marrow cell numbers (i) and calculation of MDP (Lin^- CSF1R⁺) numbers (j). In b–j, arrow with adjacent numbers indicate change, wild-type versus Flt3^{- /-} ; each symbol represents one mouse assayed at 2, 4 or 9 weeks of age. Data are representative of two experiments.

Full size image (72 KB)

Flt3 in DC homeostasis

To determine whether the DC defect in Flt3^{- /-} mice was cell intrinsic, we created bone marrow chimeras by engrafting wild-type mice with Flt3^{- /-} or wild-type cells and analyzed the recipient mice 4 months later (Fig. 4a). Recipients had similar engraftment of Flt3^{- /-} and wild-type hematopoietic stem cells (HSCs), as determined by flow cytometry (Lin^- c-Kit⁺Sca-1⁺ cells; data not shown), and had similar numbers of PMNs, red pulp macrophages and monocytes but fewer Flt3^{- /-} pDCs and cDCs (Fig. 4a). We conclude that the DC deficiency in Flt3^{- /-} mice was cell intrinsic.

Figure 4: DC development in Flt3^{- /-} bone marrow chimeras.

(a) Absolute numbers of wild-type (filled) and Flt3^{- /-} (open) spleen PMNs, red pulp macrophages, monocytes, pDCs and cDCs in bone marrow chimeras 4 months after injection of Lin^- c-Kit⁺ cells into lethally irradiated recipients. Each symbol represents an individual mouse. Data are representative of two experiments. (b) Relative contribution of Flt3^{- /-} (CD45.2⁺) cells to bone marrow HSCs (KSL; c-Kit⁺Sca-1⁺Lin^- ), MDPs (Lin^- CSF1R⁺) and splenic T cells (T; CD3⁺), B cells (B; CD19⁺), PMNs, monocytes, red pulp macrophages, pDCs and cDCs in mixed bone marrow chimeras 1.5 months after transfer. Data are pooled from three recipient mice and are representative of two independent experiments. (c) Flow cytometry of the contribution of either partner of 5-week parabiosis between wild-type control mice and between wild-type and Flt3^{- /-} mice, for cells gated on spleen T cells (CD3⁺; top row) and cDCs (CD3^- CD19^- Ter119^- NK1.1^- Gr1^- PDCA1^- F4/80^- CD11c^hi; bottom row), with the origin of cells determined by distinct CD45 isoforms: wild-type–wild-type, CD45.1⁺ and CD45.2⁺; wild-type–Flt3^{- /-} , CD45.1⁺ and CD45.2⁺. Data are representative of four experiments.

Full size image (98 KB)

To directly compare the developmental potential of mutant and wild-type progenitors, we constructed mixed bone marrow chimeras and analyzed them after 1.5 months (Fig. 4b and Supplementary Fig. 10 online). In the bone marrow, there was similar reconstitution of HSCs and MDPs, and in the periphery, PMN and monocyte reconstitution was 47% and 37%, respectively (Fig. 4b). In contrast, the contribution of Flt3^{- /-} cells to both subsets of DCs, red pulp macrophages and, as reported before, T lymphocytes and B lymphocytes²³ was very low, which indicated specific defects in the generation of these lineages in the absence of Flt3 (Fig. 4b). The defects in DCs were more notable than those of MDPs, which suggested the existence of a Flt3-dependent checkpoint for DC development beyond the MDP stage (Fig. 4b).

To further investigate the possible involvement of Flt3 in peripheral DC homeostasis, we connected Flt3^{- /-} mice with wild-type mice by parabiosis and compared them with congenic wild-type mice connected by parabiosis (control mice; Fig. 4c and Supplementary Fig. 11 online). In contrast to control mice, only 2% of the cDCs in the spleen of each wild-type parabiotic partner were of Flt3^{- /-} origin and a disproportionately large number of wild-type cDCs were present in the Flt3^{- /-} spleen relative to that in wild-type versus wild-type controls (wild-type–wild-type, 15%; wild-type–Flt3^{- /-} , 64%; Fig. 4c and Supplementary Fig. 11). To determine whether the imbalance in Flt3^{- /-} versus wild-type parabiotic mice was due to lower DC progenitor potential or fewer progenitors, we transplanted a mixture of purified MDPs from wild-type and Flt3^{- /-} mice into irradiated recipients. At 11 d after transfer, most cDCs were of wild-type origin, which suggested that Flt3^{- /-} MDPs had less potential to develop into cDCs (Fig. 5a). The contribution of Flt3^{- /-} cells was significantly lower after transfer into wild-type mice. To determine whether the difference in DC differentiation potential of wild-type and Flt3^{- /-} MDPs was Flt3L dependent, we transferred the mixture of the two types of MDPs into Flt3l^{- /-} mice. In contrast to results obtained with wild-type recipients, the ratio of wild-type and Flt3^{- /-} cDCs in the spleens of Flt3l^{- /-} recipients reflected the input ratio of MDPs (Fig. 5a). We obtained the same result by transplanting peripheral blood cell populations containing circulating cDC precursors (Fig. 5a). Thus, in the absence of Flt3L, wild-type and mutant MDPs and peripheral blood–resident cDC precursors had the same potential to develop into cDCs. We conclude that Flt3 is an important mediator of DC development after the MDP and circulating cDC precursor stage.

Figure 5: Flt3 in peripheral expansion.

(a) Contribution of donor wild-type cells (WT) and Flt3^{- /-} cells (Flt3-KO) to the input and to reconstituted cDCs 11 d after adoptive transfer of mixed bone marrow Lin^- CSF1R⁺ cells (MDP; left) or white blood cells (WBC; right) into sublethally irradiated wild-type (WT) or Flt3l^{- /-} (Flt3L-KO) recipient mice. Contribution of Flt3^{- /-} cells to cDCs after transfer: P = 0.001, MDP in wild-type recipients; P = 0.04, WBC in wild-type recipients; P = 0.2, MDP in Flt3l^{- /-} recipients; P = 0.2, WBC in Flt3l^{- /-} recipients (paired t-test). Data from three independent experiments are pooled. (b) Flow cytometry of BrdU incorporation into cDCs (CD3^- CD19^- Ter119^- NK1.1^- F4/80^- PDCA1^- CD11c⁺) from wild-type–wild-type (top) and Flt3^{- /-} –wild-type (bottom) mixed bone marrow chimeras, after a pulse of 2 h. Numbers in plots indicate percent BrdU incorporation into CD45.2⁺ cells (top right) or CD45.2^- cells (bottom right). (c) BrdU incorporation into MDPs (top) and cDCs (bottom) after a pulse of 2 h (left) or into pDCs (top) and cDCs (bottom) after a pulse of 4 d (right). BrdU incorporation in either wild-type–wild-type or Flt3^{- /-} –wild-type cells in the same mouse (paired t-test): Flt3^{- /-} cDCs, P = 0.018 (2 h) and P = 0.011 (4 d); Flt3^{- /-} MDPs, P = 0.26 (2 h); Flt3^{- /-} pDCs, P = 0.56 (4 d). (d) CFSE dilution (right) at 5 d after transfer of mixed Lin^- CSF1R⁺ wild-type (CD45.1⁺) cells and Flt3^{- /-} (CD45.2⁺) cells into irradiated recipient mice, showing retention of CFSE in wild-type cells (green; top) or Flt3^{- /-} cells (orange; bottom) with gating on CD3^- CD19^- NK1.1^- Ter119^- Gr1^- PDCA1^- CD11c^hiMHCII⁺ cells (left). Numbers adjacent to outlined areas (middle) indicate percent contribution of each donor cell type after exclusion of recipient cDCs; numbers in top right corners (right) indicate percent CFSE⁺ cells. Data are representative of three independent experiments. (e) CFSE dilution at 6 d after transfer of wild-type spleen and lymph node cells (7 10⁶) into wild-type mice (WT) or Flt3l^{- /-} mice (KO), for cells gated on donor CD11c^hi cells. Numbers above bracketed lines indicate percent CSFE^- cells. Data are representative of three experiments.

Full size image (97 KB)

To analyze the proliferation of Flt3-deficient and Flt3-sufficient cells in the same microenvironment, we generated bone marrow chimeras and measured incorporation of the thymidine analog BrdU into MDPs and cDCs after a pulse of 2 h (Fig. 5b,c). There was no difference in BrdU uptake in MDPs derived from wild-type or Flt3^{- /-} HSCs, which confirmed the Flt3-independent generation. In contrast, Flt3^{- /-} cDCs incorporated less BrdU than did Flt3^+/+ cDCs (Fig. 5b,c), which suggested that in situ cDC proliferation depended on Flt3. After a BrdU pulse of 4 d, Flt3^{- /-} cDCs incorporated about 10% less BrdU than wild-type cells did, but pDCs show no difference in BrdU incorporation relative to that of wild-type pDCs (Fig. 5c), which suggests that a distinct mechanism underlies the paucity of pDCs in Flt3^{- /-} mice.

To determine whether a defect in cell division in the periphery contributed to the DC defect in the absence of Flt3, we labeled Flt3^{- /-} and wild-type MDPs with the cytosolic dye CFSE, transferred mixtures of the labeled cells into irradiated recipients and measured cell division by CFSE dilution. We found that Flt3^{- /-} cells divided less than wild-type cells did in the spleen (Fig. 5d), which suggested that Flt3 regulates DC cell division in the periphery. To examine the effects of Flt3 on peripheral DC division directly, we injected CFSE-labeled spleen and lymph node cells into unirradiated wild-type and Flt3l^{- /-} mice and measured cDC proliferation by CFSE dilution. We readily detected DC proliferation in wild-type mice, but it was difficult to document in Flt3l^{- /-} recipients (Fig. 5e). We conclude that Flt3 regulates DC division in the periphery in the steady state.

Mice, parabiosis and Flt3L injection.

Flt3^{- /-} mice were generated and provided by I. Lemischka²³; mice deficient in GM-CSF receptor- were generated and provided by L. Robb and G. Begley³¹. C57BL/6, C57BL/6 C57BL/6 Pep^3b CD45.1⁺ (SJL) and CD45.1⁺CD45.2⁺ F₁ mice were from Jackson Laboratories or were bred at Rockefeller University. Parabiotic mice were produced as described⁹. Mice were anesthetized (2.5% (vol/vol) Avertin; Fluka) and shaved. Skin incisions were made in the sides of two adjacent mice from hip to elbow and ligaments were sutured together with chromic gut (Ethicon), then the skin incisions were closed with 9-mm stainless-steel wound clips. Mice were kept on antibiotics and ibuprofen (0.16 mg/l) for the ensuing 3 weeks. For Flt3L injection, 10 g recombinant human Flt3L (Amgen) in 100 l PBS was injected subcutaneously daily. All mice were maintained in specific pathogen–free conditions, and mouse protocols were approved by the Rockefeller University Animal Care and Use Committee.

Cell preparation, flow cytometry and immunohistology.

Spleens and thymi were gently dispersed between frosted slides and were digested for 30 min at 37 °C in PBS with 5% (vol/vol) FCS containing collagenase D (type II; Roche) at a final concentration of 0.4 U/ml. Cell numbers were determined with a FACSCalibur (BD); beads were counted (Caltag) and absolute leukocyte numbers were measured with anti-Ter119 (Ter119; eBioscience) according to the manufacturer's instructions. Medium-density spleen cells (MDPs; 1.076–1.084 g/cm³; Optiprep; Sigma) were prepared as described¹⁹. For the preparation of cells for flow cytometry, before cells were stained with specific antibodies, nonspecific binding sites were blocked with total rat immunoglobulin (012-000-002; Jackson Laboratories) and purified anti-FcRII/III (93; eBioscience). All cells were stained at 4 °C in PBS with 5% (vol/vol) FCS. For immunohistology, cryosections were fixed in 100% ethanol at - 20 °C, nonspecific binding sites were blocked with purified anti-FcRII/III and a streptavidin-biotin–blocking kit (Vector Laboratories), and sections were stained in PBS with 5% (vol/vol) FCS. The following reagents were used for staining (flow cytometry and immunohistology): anti-CD3 (2C11), anti-CD11c (HL3), anti-CD45RA (14.8), anti-CD43 (S7) and anti-SIRP (P84; all from Pharmingen); anti-CD8 (53–6.7), anti-CD19 (1D3), anti-CD45R (RA3-6B2), anti-CD11b (M1/70), anti-Gr-1 (RB6-8C5), anti-Ter119 (Ter119), anti-c-Kit (2B8), anti-Sca-1 (D7), anti-CSF1R (AFS98), anti-FcRII/III (93), anti-CD34 (RAM34), anti-Flt3 (A2F10), anti-I-A^b (MS/114.15.2), anti-F4/80 (BM8), anti-CD45.1 (A20) and anti-CD45.2 (104; all from eBioscience); and anti-PDCA1 (JFOS-1C2.4.1; Myltenyi); anti-CD169 (Ser-4), anti-SIGNR1 (22D1) were generated and provided by C.G. Park. The following reagents were used for secondary steps: streptavidin-allophycocyanin (Pharmingen); streptavidin–Pacific orange, DAPI (4,6-diamidino-2-phenylindole) and anti-fluorescein–Oregon green, goat immunoglobulin G fraction, Alexa Flour 488 conjugate (A11096; all from (Molecular Probes); streptavidin–phycoerythrin–carbocyanine 5.5 (Caltag); and indocarbocyanine–anti-hamster (Jackson Laboratories).

Methyl cellulose colony assay and liquid culture.

Total bone marrow cells (1 10⁵) or isolated progenitors (1 10³) were plated in medium containing methylcellulose (M3334; StemCell Technologies) supplemented with erythropoietin (10 ng/ml) or erythropoietin and IL-3 (50 ng/ml). Erythroid colonies were counted after 48 h (erythropoietin) or 7 d (erythropoietin and IL-3) as described³². Cultures were overlaid with an equal volume of 0.4% (wt/vol) benzidine in 12% (vol/vol) acetic acid and were assigned scores according to blue staining of colonies. Burst-forming unit–erythroid colonies contained at least one benzidine-positive erythroid cell, and colony-forming unit–erythroid colonies were uniformly benzidine positive. Sorted total bone marrow cells (2 10⁵) or isolated progenitors (1 10³) were plated in medium containing methyl cellulose (M3231; StemCell Technologies) supplemented with GM-CSF (10 ng/ml), CSF1 (10 ng/ml) or Flt3L (100 ng/ml). Colony formation was assessed after 7 d. For liquid cultures, 2 10⁵ total bone marrow cells or 5 10⁴ sorted cells were cultured in 1 ml medium supplemented with GM-CSF (10 ng/ml). At day 3, medium was 'exchanged' and at day 6 cultures were analyzed.

BrdU labeling and detection, CFSE labeling and cell cycle analysis.

Mice were injected intraperitoneally with 1 mg BrdU (5-bromodeoxyuridine); 2 h or 4 d later, mice were killed and BrdU incorporation was analyzed according to the manufacturer's instructions (BD). Cell surface antigens were stained with antibodies, then cells were fixed and made permeable. BrdU epitopes were exposed by digestion for 1 h at 37 °C with DNAse I and were detected with monoclonal antibody to BrdU (3D4; BD). For CFSE labeling, cells at a density of 1 10⁷ cells per ml were stained for 10 min at 37 °C with 5 M CFSE (carboxyfluorescein diacetate succinimidyl; Molecular Probes) in PBS with 0.1% (vol/vol) FCS. Reactions were 'quenched' by the addition of an equal volume of 100% FCS followed by three washes in PBS with 5% (vol/vol) FCS to ensure removal of CFSE. For cell cycle analysis, cells were sorted and then were incubated overnight in the dark at 25 °C in 80% (vol/vol) PBS, 20% (vol/vol) water, 0.1% (vol/vol) TritonX-100 (Sigma), propidium iodide (50 mg/ml; Sigma) and RNAse A (10 g/ml; Roche). A FACSCalibur with doublet discrimination mode was used for analysis.

Cell-transfer experiments.

Recipient mice were conditioned (lethal irradiation, 550 cGy plus 500 cGy with 3 h between irradiations; sublethal irradiation, 600 cGy) and cell suspensions were injected intravenously approximately 3 h after irradiation. Mice were fed antibiotic-supplemented food (TestDiet) and were given free access to water. Cell sorter–purified progenitors were injected into sublethally irradiated recipient mice and donor-derived cDCs recovered from recipient spleens were counted as follows. For Figure 1c, 3.9 10⁵ purified Lin^- CSF1R⁺Kit^- cells were injected and 14 d later, 7.5 10² donor-derived cDCs were recovered; and 2.7 10⁵ purified Lin^- CSF1R⁺Kit⁺ cells were injected and 14 d later, 2.2 10³ donor-derived cDCs were recovered. For Figure 1e, 2.1 10⁵ purified GFP-CX3CR1⁺Kit⁺ cells were injected and 14 d later, 7.3 10³ donor-derived cDCs were recovered; 7.2 10⁵ purified GFP-CX3CR1⁺Kit^- cells were injected and 14 d later, 2.5 10³ donor-derived cDCs were recovered; and 4.3 10⁵ purified CSF1R⁺Kit^- cells were injected and 14 d later, 2.1 10³ donor-derived cDCs were recovered. For Figure 2c, 1 10⁵ purified Lin^- CSF1R⁺ MDPs from untreated wild-type mice were injected and 9 d later, 8.91 10³ and 5.73 10³ donor-derived cDCs were recovered; and 3 10⁵ purified Lin^- CSF1R⁺ MDPs from mice treated for 6 d with Flt3L were injected and 9 d later, 1.88 10⁴ or 2.26 10⁴ donor-derived cDCs were recovered. Single and mixed bone marrow chimeras were produced by injection of Lin^- c-Kit⁺ cells from Flt3^{- /-} mice or wild-type mice (single), or Flt3^{- /-} and wild-type mice (CD45.2⁺ and CD45.1⁺; mixed), into lethally irradiated wild-type recipient mice (SJL (CD45.1⁺), single; SJL C57BL/6 F₁ (CD45.1⁺CD45.2⁺), mixed).

Accession code.

UCSD-Nature Signaling Gateway (http://www.signaling-gateway.org): A000949.

Note: Supplementary information is available on the Nature Immunology website.