Report

A Clonogenic Bone Marrow Progenitor Specific for Macrophages and Dendritic Cells

See allHide authors and affiliations

Science  06 Jan 2006:
Vol. 311, Issue 5757, pp. 83-87
DOI: 10.1126/science.1117729

This article has a correction. Please see:

Abstract

Macrophages and dendritic cells (DCs) are crucial for immune and inflammatory responses and belong to a network of cells that has been termed the mononuclear phagocyte system (MPS). However, the origin and lineage of these cells remain poorly understood. Here, we describe the isolation and clonal analysis of a mouse bone marrow progenitor that is specific for monocytes, several macrophage subsets, and resident spleen DCs in vivo. It was also possible to recapitulate this differentiation in vitro by using treatment with the cytokines macrophage colony-stimulating factor and granulocyte-macrophage colony-stimulating factor. Thus, macrophages and DCs appear to renew from a common progenitor, providing a cellular and molecular basis for the concept of the MPS.

Macrophages (MΦs) and dendritic cells (DCs) are involved in the scavenging of dying cells, pathogens, and molecules through phagocytosis and endocytosis and the use of pattern recognition receptors (1). As a result, both cell types make a vital contribution to immunity and inflammatory responses to pathogenic microorganisms (2). At the same time, these cells are distinct: MΦs represent scavenging tissue-resident cells and take part in the innate immune response (3), whereas DCs represent professional antigen-presenting cells that trigger and regulate the adaptive immune response (4). Both lineages also each display a remarkable heterogeneity in phenotype, tissue localization, and function and have been divided into several subsets (5).

Although the origin and renewal of MΦs and DCs have been intensively investigated, the lineage relation between these cells types is unclear (fig. S1). Most MΦs are dependent on macrophage colony-stimulating factor (M-CSF) for their development in vitro and in vivo, whereas most DC subsets are not (6). Conversely, the ligand for the receptor tyrosine kinase Flk-2 (FLT3-L) is important for the development of many DC subsets but appears not to be required for MΦ development (7). MΦs are believed to derive solely from myeloid progenitors, but recent in vivo experiments have shown that most steady-state DC subsets, including CD11c+ CD8α, CD11c+ CD8α+ DCs, and plasmacytoid DCs (PDCs), can differentiate from early bone marrow (BM) progenitors such as the common lymphoid progenitors (CLPs) as well as the common myeloid progenitors (CMPs) (812). From these and other reports, it has been hypothesized that MΦs and DCs have distinct progenitors (11, 13).

However, an alternative hypothesis is that many DC and MΦ subsets derive from a common precursor able to develop into either cell type depending on cytokine signals or other cues encountered at tissue sites (14). This hypothesis of a common progenitor shared by cells of the mononuclear phagocyte system (MPS) (15, 16) is supported by the capacity of human and mouse monocytes and recently described mouse pre-immunocytes to give rise to MΦs and DCs in vitro (1720). Furthermore, mouse blood monocytes can differentiate in vivo into a particular subset of antigen-presenting DCs under inflammatory conditions (21, 22), although it is unknown whether blood monocytes also contribute to classical steady-state DC subsets, such as the spleen CD8α+ and CD8α DCs, and to tissue MΦs (22, 23).

In a search for progenitors specific for MΦs and DCs, we identified a mouse BM population that expresses both CD117 (c-kit, the receptor for stem cell factor) and the chemokine receptor CX3CR1 but not markers of lineage-committed precursors (Lin)(Fig. 1A) (24). CX3CR1 is widely expressed among monocytes, DCs (22, 25), and MΦs (fig. S2), and its expression can be easily followed with the use of a reporter system in which green fluorescent protein (GFP) is driven at the locus of the Cx3cr1 gene in mice (22, 25). CX3CR1+ CD117+ Lin cells represented ∼0.5% of total BM cells and could be highly purified (>99%) by using flow cytometry fluorescence-activated cell sorting (FACS) on the basis of the expression of GFP and CD117 and the absence of expression of CD11b (Fig. 1A). Morphology of these cells appeared to parallel immature myeloid progenitors (Fig. 1A).

Fig. 1.

Isolation of a CX3CR1+ clonogenic proliferating BM progenitor. (A) BM cells expressing GFP and CD117 but not CD11b (top right) CD3, CD19, NK1.1, Iab, CD11c, B220, TER-119, or Gr1 (Lin; bottom left) were purified and stained with MGG (bottom right). (B) Cloning efficiency was assessed by limiting dilution series on S17 (open circles) or OP9 (solid circles) stromal cells. (C) Real-time polymerase chain reaction analysis of gene expression was performed for CX3CR1+ CD117+ (black) and CX3CR1 CD117+ (gray). (D) Lin IL7Rα CD117+ Sca-1 cells (MPs) were purified from BM and analyzed for expression of GFP, CD16/32, and CD34. (E) CX3CR1+ CD117+ Lin cells (solid squares) or OP9 cells in the absence (solid circles) or presence of exogenous M-CSF (open triangles) or FLT3-L (open squares), or in the presence of M-CSF alone (open circles) were plated at 1, 10, or 100 cells per well onto S17. (F and G) Cells were plated in the absence of stroma at varying cell numbers per well (E) and at one cell per well (F) in the presence of M-CSF (solid circles), GM-CSF (solid squares), M-CSF and GM-CSF (open squares), M-CSF and GM-CSF and FLT3-L (open circles), or FLT3-L (solid triangles).

A cloning efficiency of CX3CR1+ CD117+ Lin cells was calculated as 50%, as determined by limiting dilution analysis (Fig. 1B) and by single-cell sorting followed by culturing on S17 stromal cells (26) (fig. S3A). Clones expanded exponentially in culture to colonies of 3 × 103 to 1 × 104 cells after 7 days, indicating that each proliferating progenitor had undergone an average of 12 mitoses (fig. S3B).

mRNA for the M-CSF receptor c-fms, the granulocyte-macrophage colony-stimulating factor (GM-CSF) receptor α chain, and the transcription factor CEBPα were enriched by a factor of 100 to 1000 in the CX3CR1+ CD117+ Lin cells as compared with CD117+ Lin progenitor cells depleted of CX3CR1-expressing cells (Fig. 1C). Genes usually associated with lymphoid progenitors such as Gata3, Flt3, and Pu.1 were also expressed, in contrast to no or weak expression of genes associated with the erythrocyte-megakaryocyte pathway Gata1, Epo-R, and c-mpl (Fig. 1C).

CX3CR1+ CD117+ Lin cells existed within the CD117+ Sca1 IL7Rα myeloid progenitors (MPs) (8) (Fig. 1D). CX3CR1+ CD117+ Lin cells homogenously expressed CD34 and CD16/32, a phenotype shared by granulocyte-macrophage progenitors (GMPs) found among MPs (8) (Fig. 1D). However, they could be distinguished from GMPs by lower expression of CD117 and elevated expression of CX3CR1 (Fig. 1D).

The efficiency of cloning CX3CR1+ CD117+ Lin cells by culturing on M-CSF-deficient stromal cells [OP9 cells (27)] was reduced by two orders of magnitude compared with the efficiency of cells cultured on S17 cells (Fig. 1B). Addition of recombinant M-CSF completely restored cloning efficiency (Fig. 1E), and recombinant GM-CSF could substitute for this effect. Furthermore, stroma was dispensable for the growth of both M-CSF and GM-CSF cultures but not FLT3-L cultures of progenitors (Fig. 1, F and G). Lastly, an additive effect of M-CSF and GM-CSF on cloning efficiency was not observed (Fig. 1, F and G), suggesting that the cytokines were acting on the same cell.

In contrast to sorted CX3CR1 CD117+ Lin cells, which gave rise to colonies of polymorphonuclear cells (PMN), MΦs, and DCs; nearly all colonies (96%) arising from single CX3CR1+ CD117+ Lin cells grown on S17 stroma possessed morphology typical of MΦs and DCs but not PMN (Fig. 2A). Limiting dilution analysis showed that CX3CR1+ CD117+ Lin cells were also devoid of B or natural killer (NK) lymphoid potential (fig. S4A). When grown in the absence of stroma, clonal colonies cultured with M-CSF expressed CX3CR1, appeared macrophage-like by morphology and phenotype (CD11b+ CD11c) (Fig. 2B), and were efficient at phagocytosing heat-killed Escherichia coli (fig. S4B). Colonies cultured with GM-CSF down-regulated CX3CR1, had the morphology and phenotype of DCs (CD11bint CD11c+) (Fig. 2B), and were capable of processing and presenting a model antigen (ovalbumin) to naïve T cells (fig. S4C). Notably, single cells cloned in the presence of both cytokines gave rise to both macrophage- and DC-like cells (Fig. 2B). Addition of FLT3-L to GM-CSF did not significantly modify CX3CR1, CD11b, or CD11c expression (fig. S5).

Fig. 2.

Differentiation of CX3CR1+ CD117+ Lin cells in vitro. (A) CX3CR1+ CD117+ Lin progenitors gave rise to colonies with MΦ- and DC-like morphology (black bars and top left photo). CX3CR1 CD117+ Lin progenitors gave rise to colonies of PMN (open bars and bottom left photo), colonies of MΦ and DC (black bars and top right photo), and colonies of MΦ and DC plus PMN (hatched bars and bottom right photo). (Right) CD11b and CD11c expression on colonies from single cells cultured on S17 cells with exogenous GM-CSF and FLT-3L compared with those cultured on S17 cells alone. S17 cells appear as CD11b CD11c. (B) CX3CR1+ CD117+ Lin cells cultured without stroma but in the presence of M-CSF (left profiles and top photos) or GM-CSF (middle profiles and bottom left photo) or both (right profiles and bottom right photo). M-CSF cultures display macrophage-like phenotype and morphology, including phagocytosis of dead cells (open arrow), whereas GM-CSF cultures display DC-like phenotype and morphology (bottom left photos). In the presence of both cytokines (bottom right photo), macrophage- and DC-like cells (arrowhead and arrows, respectively) are observed.

In order to test the in vivo potential of CX3CR1+ CD117+ Lin cells, we intravenously transferred 2 × 104 cells purified from Cd45.2 donor mice with the use of FACS to irradiated congenic Cd45.1 C57BL/6 recipient mice (Fig. 3A). Donor-derived cells first became detectable in the spleen at day 2, expressed intermediate amounts of GFP, and were CD45.2+ CD11bint CD11c (Fig. 3B). By day 7, three new populations of donor-derived spleen cells were easily detectable: CD11b CD11c+, CD11b+ CD11c+, and CD11b+ CD11c (Fig. 3B). In contrast, we did not detect donor-derived Gr1hi PMNs, B220+ B cells, CD3+ T cells, NK1.1+ NK cells, or Gr1+ B220+ PDCs in any tissues between 2 and 102 days after transfer (Fig. 3C, left, and fig. S6). The appearance of CD11b+ and CD11c+ cells was delayed by a week when CX3CR1+-depleted CD117+ Lin progenitors were transferred in place of CX3CR1+ CD117+ Lin cells (Fig. 3C, right). At day 7 after transfer, CX3CR1+ CD117+ Lin cells gave rise to more donor-derived CD11c+ cells than did CX3CR1 CD117+ Lin cells, despite a lower proliferation potential (Fig. 3D and fig. S3B). These data indicate that the CX3CR1+ CD117+ subset was the main progenitor intermediate for donor-derived CD11b+ and CD11c+ cells in vivo.

Fig. 3.

Differentiation of CX3CR1+ CD117+ Lin cells in vivo. (A) Sorted CX3CR1+ (GFP+) CD117+ Lin cells (2 × 104) from mice expressing CD45.2 were transferred intravenously into irradiated (6 gray) wild-type congenic mice expressing CD45.1. (B) Donor-derived spleen cells (CD45.2+ GFP+/–, left) were analyzed for expression of CD11b, F4/80, and CD11c (right) 2 days (top) or 7 days (bottom) after transfer. (C) Sorted progenitors (CD117+ Lin cells), expressing CX3CR1 (top) or not (bottom), were transferred into recipient mice, and spleen cells were harvested at various times after transfer and analyzed by FACS. Donor-derived (CD45.1 CD45.2+) Gr1+ GFP cells (PMN) were detectable 7 days after transfer of CX3CR1 progenitors but not CX3CR1+ progenitors (left). CD11b+ and/or CD11c+ cells were analyzed for donor origin (CD45.2 and GFP expression) between 2 and 15 days after transfer (right). (D) Seven days after transfer, donor-derived cells (CD45.1 CD45.2+, left) from CX3CR1 progenitors were mostly CD11c, whereas those from CX3CR1+ progenitors were mostly CD11c+ (middle) and were the major source of spleen CD11c+ cells (right).

Donor-derived CD11c+ cells were detectable in the spleen of recipients beginning at day 4 after transfer of 2 × 104 CX3CR1+ CD117+ Lin cells (Fig. 4). These cells represented between 0.18 and 52.6% of total spleen CD11c+ (average of 7.15%, n = 17). Among CD11c+ cells, CD11b cells expressed CD8α but not CX3CR1, whereas CD11b+ cells expressed CX3CR1 but not CD8α, which is consistent with observations in mouse spleen under steady-state conditions (25) (Fig. 4A and fig. S7). Immunohistochemical analysis performed at day 7 after transfer confirmed the presence of donor-derived DCs in the white pulp of the spleen, surrounded by marginal sinus MΦs (Fig. 4B). The number of donor-derived DCs increased until day 6 to 7, decreasing thereafter and reaching background numbers after day 15 (Fig. 4C). Thus, both CD8α+ and CD8α DCs in these studies were short-lived, as suggested by previous studies (5, 28), and arose from a common proliferating CX3CR1+ CD117+ Lin BM progenitor with apparent limited self-renewal. BM monocytes were at least two orders of magnitude less efficient than progenitors at generating CD8α DCs after transfer, and we did not observe convincing evidence that monocytes could give rise to CD8α+ DCs (fig. S8). Whether inflammation was involved in the differentiation of progenitors into steady-state spleen DCs was difficult to assess. However, irradiation of the host, which causes inflammation, was not required for the differentiation of donor-derived DCs (fig. S9), further suggesting that progenitors gave rise to steady-state splenic DCs.

Fig. 4.

Characterization of DCs and MΦs derived in vivo from CX3CR1+ CD117+ Lin progenitors. (A) CD45.2+ CD11c+ cells (left) consist of CD11b+ GFPhigh (CX3CR1high) CD8α and CD11b GFPlow (CX3CR1low) CD8α+ populations (right and bottom). (B) Donor-derived (CD45.2+, green) cells in sections of recipient spleens stained with MOMA-1 (red). (C) After transfer of 2 × 104 CX3CR1+ CD117+ Lin progenitors, the number of donor-derived (CD45.2+ and/or GFP+) CD11c+ CD11b cells (black) or CD11c+ CD11b+ cells (white) in the entire spleen (y axis) was estimated as described in (24). (D) Donor-derived (GFP+, green), F4/80+ (red) cells within spleens of transfer recipients. Details of the top row of photos are shown in the bottom row. At day 7, F4/80+ GFP+ (arrowhead) and F4/80 GFP+ cells (arrow) could also be detected. (E) GFP+ MΦ detected with MOMA-1 (left), SIGNR1 (middle), or MARCO (right). Scale bar is 20 μm for all images. (F) Thioglycolate-elicited CD45.1 CD45.2+ F4/80+ CD11b+ MΦ were detectable in peritoneal exudates of Cd45.2 Cx3cr1gfp/+ mice (middle) and Cd45.1 mice that received 2 × 104 CX3CR1+ CD117 + Lin progenitors (M Φ and DC progenitors or MDP) (right) but not controls (Cd45.1, left). (G) Summary of the proposed differentiation potential of the MDP, and its origin and relationship with other progenitors, on the basis of our current findings.

F4/80+ GFP+ cells were detected at day 2 by immunohistochemical analysis as scattered rounded cells in the red pulp (Fig. 4D) and likely correspond to the F4/80+, CD11b+ GFP+ CD45.2+ monocytes detected by flow cytometry (Fig. 3B). At time points later than 1 week after transfer, F4/80+ GFP+ MΦs were difficult to detect by FACS (most likely because of the technical difficulty of extracting MΦ tissue) but were easily identified by histology in the red pulp up to several weeks posttransfer (Fig. 4D). GFP+ marginal sinus and marginal zone MΦs were identified on the basis of the expression of MOMA1 and of SIGN-R1 and MARCO, respectively (Fig. 4E). Lastly, donor-derived F4/80+ peritoneal MΦs were detectable 3 days after thioglycollate injection (Fig. 4F), and donor-derived cells with a microglial phenotype were detectable in the brain of irradiated recipients (fig. S10). Together, these data indicate that CX3CR1+ CD117+ Lin cells may be progenitors for several populations of MΦ in this experimental setting.

Previous reports described precursor populations that can generate DCs or MΦs (14), but their differentiation potential was not restricted to these lineages. Monocytes differentiate into CD11cint DCs under inflammatory conditions (21, 22) but have a very limited potential to differentiate into steady-state spleen DCs in vivo (fig. S8). The novel progenitor described here and that we name MDP, for MΦ and DC progenitor, gives rise to monocytes, to several subsets of MΦs, and to steady-state CD11c+ CD8α+ and CD11c+ CD8α DCs in vivo (Fig. 4G). In contrast, the MDP is devoid of lymphoid, erythroid, and megakaryocytic potential and, reminiscent of the mono-blast described by van Furth in 1975 (29), also lacks PMN differentiation potential. In this experimental system, we could not detect PDCs in the progeny of the MDP, suggesting that this DC subset may originate from a distinct lineage. Because the MDP can be selected or instructed to differentiate by external cues, we suggest that it might be possible to modulate differentiation toward particular MΦ or DC subsets in a therapeutic setting and thereby influence physiological and pathological processes.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1117729/DC1

Materials and Methods

Figs. S1 to S10

Table S1

References and Notes

References and Notes

View Abstract

Navigate This Article