A Lineage of Myeloid Cells Independent of Myb and Hematopoietic Stem Cells

See allHide authors and affiliations

Science  06 Apr 2012:
Vol. 336, Issue 6077, pp. 86-90
DOI: 10.1126/science.1219179


Macrophages and dendritic cells (DCs) are key components of cellular immunity and are thought to originate and renew from hematopoietic stem cells (HSCs). However, some macrophages develop in the embryo before the appearance of definitive HSCs. We thus reinvestigated macrophage development. We found that the transcription factor Myb was required for development of HSCs and all CD11bhigh monocytes and macrophages, but was dispensable for yolk sac (YS) macrophages and for the development of YS-derived F4/80bright macrophages in several tissues, such as liver Kupffer cells, epidermal Langerhans cells, and microglia—cell populations that all can persist in adult mice independently of HSCs. These results define a lineage of tissue macrophages that derive from the YS and are genetically distinct from HSC progeny.

Two different types of hematopoietic cells can give rise to macrophages in vertebrates (1, 2). In mice, macrophages develop in the yolk sac (YS) from embryonic day 8 (E8) (3). In contrast, definitive hematopoietic stem cells (HSCs) appear within the hematogenic endothelium of the aorto-gonado-mesonephros region (46) at E10.5 (6) and migrate to the fetal liver where they expand and differentiate starting from E12.5 (2). Macrophages and dendritic cells (DCs) are present in all tissues and are critical effectors and regulators of immune responses. A large number of these cells, including classical DCs, plasmacytoid DCs and monocytes, originate from HSCs and are replaced continually from a macrophage and DC precursor (7, 8). However, bone marrow (BM) transplantation leads to relatively inefficient replacement of tissue macrophages. Classical studies have proposed a dual origin for tissue macrophages, with half of the population being renewed from circulating precursors, and the other half from local production (9). More recently, mutations in GATA2 (10) and IRF8 (11) have been associated with profound defects in BM-derived monocytes and DCs, whereas many tissue macrophages were unaffected (11, 12). Liver Kupffer cells (13), epidermal Langerhans cells (14, 15), microglia (16, 17), and pleural macrophages (18) were shown to be able to proliferate and renew independently from the BM.

Self-renewal or independence from the BM does not preclude the initial development of macrophages from HSCs. However, the hypothesis that the myeloid lineage may be split into cells originating from the YS and from HSCs has been raised (19). Recent fate-mapping studies in Runx1MER-cre-MER embryos resulted in labeling of 30% of the microglia (17) but also of 10% of HSCs (20). Thus, the respective contribution of YS and HSC to the macrophage pools remains unclear.

We first reexamined the kinetics of myeloid cell development using Cx3cr1gfp/+ (green fluorescent protein, GFP) reporter mice (3, 21). CD45+ CX3CR1bright F4/80bright YS-derived macrophages circulated in the blood and colonized the developing mouse embryo between E9.5 and E10.5, starting with the cephalic area (Fig. 1A and figs. S1 and S2). By E10.5, YS-derived macrophages were proliferating and detected in most tissues (Fig. 1B and figs. S1 and S2). Throughout development, CD45+ CX3CR1bright F4/80bright cells remained detectable in tissues (fig. S1). From E12.5 onwards, at the time when fetal liver hematopoiesis is active, a novel population of CD45+ CX3CR1+ cells appeared expressing low levels of F4/80 (F4/80low) but high CD11b (CD11bhigh) (fig. S1).

Fig. 1

Myb-independent F4/80+ YS macrophages. (A) Flow cytometric analysis of E10.5 YS, skin, and blood macrophages in Cx3cr1gfp/+ embryos, gated on CD45+ cells. YS-derived macrophages (3) are color-gated in blue. Representative data from individual embryos, n = 6. (B) Expression of Ki67 in E10.5 limb buds CX3CR1-GFP macrophages. Scale bar is 20 μm, 20× objective. n = 3, data from one representative experiment are shown. (C) Flow cytometric analysis of YS from Pu.1−/− (n = 5), Pu.1+/− (n = 8), Myb−/− (n = 5), Myb+/− (n = 7), and their WT littermates (n = 3 to 4), at E10.5. Histograms indicate the percentage of CD45+ cells among live cells (top), and the percentage of F4/80+ CD11b+ cells among CD45+ cells (bottom). Means ± SEM, ns, not significant; *P < 0.05 versus WT. (D) Flow cytometric analysis of E14.5 and E16.5 fetal liver, gated on CD45+ cells. Histograms represent the percentage of CD45+ subsets. Means ± SEM from WT (n = 3 and 8), Myb+/− (n = 12 and 4) and Myb−/− (n = 3 and 4) embryos at E14.5 and E16.5, respectively. (E) Frozen tissue sections from E16.5 livers from Myb−/− and WT littermates stained for F4/80 (red), collagen IV (green), and 4′,6′-diamidino-2-phenylindole (DAPI) (blue). Scale bar is 25 μm, 20× objective (a representative experiment is shown, n = 3).

The transcription factor PU.1 is required for the development of macrophages (22, 23) but is dispensable for the development of HSCs (24). In contrast, Myb is dispensable for YS myelopoiesis although it is required for the development of HSCs (2527). We therefore used deletion of Myb and Pu.1 in mice to determine the contribution of YS to macrophage lineages.

F4/80+ CD11b+ macrophages were absent from the YS of Pu.1−/− embryos, whereas they developed in normal numbers in E10.5 Myb−/− YS (Fig. 1C). In addition, Myb−/− fetal liver lacked all Kit+ cells, including Kit+ CD45+ hematopoietic precursors at E14.5 and E16.5, whereas CD11b+ and F4/80+ cells were still present (Fig. 1, D and E, and fig. S3). On the basis of these findings, we investigated the development of myeloid subsets in the absence of Myb. In Myb−/− mice, CD45+ F4/80low CD11bhigh macrophages were largely absent in all organs examined (Fig. 2, A and B); however, CD45+ F4/80bright CD11blow macrophages developed normally (Fig. 2, A and B). They accounted for nearly all of the macrophages found in the skin, spleen, pancreas, kidney, and lung of Myb−/− mice at E16.5 (Fig. 2B). In both Myb−/− and Myb+/+ littermate fetuses, CD45+ F4/80bright CD11blow macrophages were located in the upper dermis and invaded the developing keratin-14+ epithelium (fig. S4). Likewise, in lung and kidney of wild-type (WT) and Myb−/− embryos, F4/80bright dendritic-shaped macrophages were located adjacent to collagen IV+ tubular structures, which correspond to developing bronchiolar and tubular epithelia, respectively (fig. S4). Pancreatic F4/80bright macrophages were found in close proximity to insulin+ β cells (fig. S4). F4/80bright macrophages were also found in the stroma of liver and spleen (Fig. 1E and figs. S4 and S5) (28). In contrast Pu.1−/− mice lacked all macrophages in all tissues examined (Fig. 2A and fig. S5) (22, 23). In the brain, Iba1+ F4/80+ microglia were present at normal density in the developing brain of Myb−/− embryos (fig. S5). Our results suggest that Myb was required for differentiation of CD45+ F4/80low CD11bhigh monocyte/macrophages, as expected if they derived from HSCs. In contrast, the development of CD45+ F4/80bright CD11blow macrophages and microglia did not require Myb or functional HSCs but was Pu.1-dependent.

Fig. 2

Myb-dependent and independent tissue macrophages in fetuses. (A and B) Flow cytometric analysis of the indicated organs at E14.5 and E16.5 of WT (n = 11 and 6, respectively), Myb+/− (n = 8 and 15), Myb−/− (n = 4 and 4), and Pu.1−/− (n = 3 to 5) mice, gated on CD45+ cells. F4/80bright CD11blow macrophages are color-gated in blue in dot plots and represented by white bars in histograms. F4/80low CD11bhigh macrophages are color-gated in red and represented by black bars in histograms. KIT+ cells (color-gated green) are shown in skin and spleen. Histograms represent means ± SEM, *P < 0.05 of Myb−/− versus WT.

F4/80bright macrophages were originally described 30 years ago by Gordon and Hume in adult mice and developing embryos; these were closely associated with epithelia (29), and included epidermal Langerhans cells, liver Kupffer cells, and macrophages of the pancreatic islet of Langerhans, kidney interstitial cells, and splenic red pulp macrophages (28, 3032). We thus investigated whether the F4/80bright cells are related to YS macrophages.

We first analyzed their pattern of gene expression (21). Unsupervised clustering analysis of genome-wide expression arrays from E10.5 YS macrophages and E16.5 F4/80bright and F4/80low macrophages indicated coclustering of E10.5 YS and F4/80bright E16.5 macrophages, whereas F4/80low macrophages clustered separately (Fig. 3A). In a supervised analysis, lung, skin, and kidney E16.5 F4/80bright macrophages and YS macrophages shared a common signature of differentially regulated genes, distinct from that of E16.5 F4/80low macrophages (P < 0.05, twofold changes) (Fig. 3B and fig. S6). The signatures from Myb−/− and Myb+/+ F4/80bright macrophages clustered together, which indicated that the presence or absence of Myb did not affect their gene expression profile at the genome level (Fig. 3, A and B).

Fig. 3

Gene expression profiling of embryonic and fetal macrophages. (A) Unsupervised hierarchical clustering of whole-genome expression arrays from E10.5 YS macrophages and E16.5 F4/80bright and F4/80low macrophages (21). Unsupervised clustering analysis (of all probes) was performed using Spearman rank correlation to measure the similarity between samples. (B) Supervised analysis. The heat map represents the gene signature of YS and F4/80bright macrophages compared with that of F4/80low macrophages (n = 1132, twofold change-filtered, P < 0.05, see fig. S6). (A) and (B) Results from individual mice. (C) Among the genes identified within the signatures, 55 genes, selected on the basis of their putative functions (GO analysis) are shown. The heat map represents mean log2 normalized intensity (NI, n = 3). Bar graphs indicate NI (means ± SD). (D) Fate mapping of FLT3 expression in blood and tissue of 4-week-old Flt3Cre x R26LSL-YFP mice. Bars represent percentage of YFP-positive cells (means ± SEM, n = 3). F4/80bright and F4/80low macrophages were gated on CD45+ CD11b+. No YFP was detected in Cre littermates (n = 3, see fig. S6).

A detailed analysis of the YS and F4/80bright macrophages signature is presented in Fig. 3C. The transcription factor Maf that controls macrophage proliferation (33), the chemokine receptor Cx3cr1, and the macrophage colony-stimulating factor (M-CSF) receptor (Csf1r) were enriched in F4/80bright and YS macrophages, whereas Gata2, which controls monocyte differentiation (10), the chemokine receptor Ccr2, the growth factor receptor Flt3 expressed in pluripotent hematopoietic progenitors (3436), and genes responsible for peptide editing and presentation were selectively expressed by F4/80low macrophages (Fig. 3C). Analysis of 4-week-old Flt3-Cre × RosaLSL-YFP (yellow fluorescent protein, YFP) mice (Fig. 3D and fig. S6) indicated that recombination was high in blood cells and in F4/80low macrophages, but low in F4/80bright macrophages. This indicated that expression of FLT3 remains low in adult F4/80bright tissue macrophages and suggested that their development occurs largely independently of FLT3+ multipotent precursors.

Altogether, these data suggested a close developmental relation between E10.5 YS macrophages and E16.5 F4/80bright macrophages, whereas the F4/80low macrophages appeared related to adult hematopoiesis and antigen presentation. These data also supported the observation that the differentiation of F4/80bright macrophages was not affected by Myb deficiency. Thus, our observations suggested a possible YS origin of these F4/80bright macrophages.

We thus devised a fate-mapping strategy to address directly the relation between YS precursors and F4/80bright macrophages in late embryos and adult mice. We first transplanted limb buds from E10.5 Cx3cr1gfp/+ mouse embryos onto the chorioallantoic membrane (CAM) of E7 chicken embryos (21) (Fig. 4A) (37). CAM grafting should permit the study of murine limb development in the absence of a source of HSCs. After 12 days of culture on the CAM, limb buds gave rise to limb-like structures consisting of a stratified epithelium of mouse origin enveloped by the chicken chorion. In all cases, CX3CR1+ YS murine macrophages migrated toward the epithelium (movie S1) and colonized the epidermis on top of the keratin-14+ basal keratinocyte layer, a feature of Langerhans cells (Fig. 4A and fig. S7).

Fig. 4

Fate-mapping analyses of F4/80+ YS macrophages. (A) Limbs from E10.5 Cx3cr1gfp/+ embryos were grafted onto the chorioallantoic membrane (CAM) of an E7 chicken embryo (left, 0.8 objective) and cultured for 12 days. Murine tissue was detected with digoxigenin (DIG)–labeled in situ hybridization to the mouse B2 SINE (dark purple, middle left, and fig. S7). Optical sections from whole-mount limb graft (middle right and right, and fig. S7, movie S1) were analyzed by confocal microscopy. Red staining represents mouse keratin-14; green represents GFP. Arrows point to epidermal basal membrane. Scale bars are 20 μm, 20× objective. Data are representative of four independent experiments. (B) Pulse-labeling of CSF1R+ cells in E8.5 embryos. FVB strain Csf1rMer-iCre-Mer/+ females mated with B6 RosaLSL-YFP/LSL-YFP males received an intraperitoneal injection of 75 μg/g OH-TAM and 37.5 μg/g progesterone at E8.5 of timed pregnancies. (Top) All events in blood and liver; (middle and right) expression of YFP in CD45+ cells in 4-week-old Csf1rMer-iCre-Mer/+;RosaLSL-YFP/+ F1 mice (Cre+) and Csf1rwt;RosaLSL-YFP/+ F1 control littermates (Cre).(Bottom) F4/80 and CD11b expression on CD45+cells and expression of F4/80 and CD11b in total CD45+YFP+ cells in Cre+ and Cre littermates in the indicated organs. A representative experiment is shown out of two, with n = 4 Cre+ and 8 Cre. (C) Myb-deficient BM chimera. Deletion of floxed Myb alleles was induced by poly(I:C) in 6- to 8-week-old Cd45.2;Mx1-Cre;Mybf/f mice (fig. S10). Mice were transplanted with 107 whole BM cells from Cd45.1 mice (fig. S10) and examined 3 months after transplantation for donor (CD45.1) or recipient (CD45.2) origin of the indicated cell populations. F4/80bright cells are color-gated blue. F4/80low CD11bhigh cells are color-gated red. Representative results from two experiments.

In a second, genetic approach, we took advantage of the early expression of the receptor for M-CSF (CSF1R) by YS macrophages (38) to label CSF1R-expressing cells in the YS and to follow their progeny in adult mice using Csf1rMer-iCre-Mer reporter mice (39). Pregnant Csf1rMer-iCre-Mer mice crossed to RosaLSL-YFP received a single intraperitoneal low dose of OH-tamoxifen (TAM) (40) together with progesterone (41) at E8.5, in order to pulse-label Csf1r-expressing YS macrophages in Csf1rMer-iCre-Mer;RosaLSL-YFP F1 embryos between E8.5 and E9.5 without terminating the pregnancy (Fig. 4B and fig. S8). Analysis of TAM-pulsed Csf1rMer-iCre-Mer;RosaLSL-YFP F1 mice 4 weeks after birth showed labeling of F4/80bright macrophages in all tissues examined: in Kupffer cells and Langerhans cells and also in pancreas, lung, spleen, and kidney F4/80bright macrophages (Fig. 4B and fig. S8). YFP labeling was also present in microglia (fig. S9). No YFP expression was detected elsewhere, including the typical progeny of adult HSCs, such as blood leukocytes, or in tissue CD11bhigh F4/80low macrophages, and no YFP was detected in TAM-treated Csf1rwt;RosaLSL-YFP littermates (Fig. 4B).

To further investigate the role of Myb and of HSCs in the maintenance of adult F4/80bright macrophages, we performed conditional deletion of the Myb gene in adult Cd45.2;Mx1 Cre;Mybflox/flox mice (Fig. 4C and fig. S10) (21). This resulted in a rapid loss of blood monocytes and granulocytes and successful engraftment of CD45.1 BM cells by injection of 107 Cd45.1;Myb+/+ BM cells in the absence of irradiation or any further conditioning regimen (fig. S10). BM chimeras were analyzed 1 and 3 months after BM rescue. At both time points, all monocytes and granulocytes were of donor CD45.1 origin, which indicated complete chimerism (Fig. 4C and fig. S10). In peripheral tissues, CD11bhigh F4/80low macrophages and DCs had also been replaced by CD45.1 BM-derived cells after 3 months (Fig. 4C). In contrast, F4/80bright macrophage subsets remained of the host CD45.2 genotype in the liver, epidermis, and brain and were not replaced. Only 10% of F4/80bright macrophages were of donor origin in the pancreas and spleen. In the lung and kidney, macrophages were of mixed recipient and donor origin, which suggested more heterogeneity of these cell populations (Fig. 4C). Recombination and Myb deletion in CD45.2 F4/80bright macrophages was confirmed by genomic polymerase chain reaction (fig. S10).

Taken together, data from fate mapping studies indicated that YS-derived precursors give rise to populations of Myb-independent F4/80bright macrophages and Langerhans cells in mouse tissues in the presence of WT HSCs. They persist in adult mice independently of HSCs and of Myb under homeostatic conditions. In contrast, Myb-dependent BM precursors continuously replace classical DCs (8) and CD11bhigh F4/80low macrophages and some F4/80bright macrophages, in particular in the kidney and lung, which indicates a mixed origin of these populations. These results also suggest that deletion of Myb is sufficient to ablate adult HSCs in the absence of any other conditioning regimen.

Collectively, our experiments define two lineages of macrophages and DCs in the mouse. These data modify the concept of hematopoietic “stem cells,” because not one, but two, myeloid systems overlap and renew independently in mice. As F4/80bright macrophages develop and stay in close association with epithelial structures (29), these represent a candidate regulator of macrophage homeostasis (14), together with transcription factors such as Maf (33). Of note, the Cd45.2;Mx1-Cre;Mybf/f/Cd45.1 BM transplantation model we describe here should be useful to dissect the distinct functions and molecular wiring of the two myeloid systems in the future.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S10

References (4248)

Movie S1

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: The authors are indebted to S. McKercher, Sanford-Burnham Medical Research Institute La Jolla, California, USA, for providing Pu.1+/− mice and to T. Boehm and C. Bleul, Max Planck Institute of Immunobiology and Epigenetics, Freiburg, Germany, for providing Flt3-Cre mice. We also thank P. Saha for help with surgical procedures; C. Trouillet for help with flow cytometry and general organization of the lab; A. McGuigan and the staff of the Biological Service Unit at King’s College London (KCL), Guy’s Campus, for support with mouse husbandry; and S. Heck and the Flow facility of the Biomedical Research Centre (BRC) at King’s Health Partners. The data presented in this manuscript are tabulated in the main paper and the supplementary materials. MIAME data are available at GomezElisa_AgilentMouse, ArrayExpress accession: E-MEXP-3510. C.S. is supported by a fellowship program of the German National Academy of Sciences Leopoldina (LPDS 2009-31). K.J.L. is supported by the Wellcome Trust (WT081880AIA) and U.K. Biotechnology and Biological Sciences Research Council (BB/E013872). F.G. is the Arthritis Research U.K. Chair of Inflammation Biology, at KCL. This work was funded by grants MRC G0900867, from the U.K. Medical Research Council, and ERC-2010-StG-261299 MPS2010 from the European Research Council to F.G.
View Abstract

Navigate This Article