Introduction
Dendritic cells (DCs) and macrophages form networks of phagocytic cells throughout most tissues and have important roles in development, scavenging, inflammation and antipathogen defenses, and in the initiation, maintenance and regulation of immune memory1, 2. Considerable attention is currently focused on the characterization of their progenitors and precursors, the signals driving their development in the bone marrow, their migration to tissues and their homeostasis in peripheral tissues. Development of DCs and macrophages begins in the bone marrow, where precursors are dependent on outside signals for their proliferation and emigration. In the current issue of Nature Immunology, Waskow et al.3 identify yet another level of control whereby bone marrow DC precursors that have entered the blood stream and reached their sites of residence in lymphoid organs undergo a local homeostatic expansion, a process that is dependent on the growth factor Fms-related tyrosine kinase-3 ligand (Flt3L) and its receptor and that is critical for regulating the size of the DC pool in the periphery.
Until recently, homeostatic control of DCs and macrophages in the periphery had been attributed mainly to the proliferation of precursors in the bone marrow, a process believed to be controlled by cytokines such as M-CSF4, GM-CSF5 and Flt3L (also known as Flk2)6. The M-CSF receptor (c-fms, CD115, M-CSF-R) is expressed on myeloid precursors and is critical for the development of many cells of this lineage4. Mice homozygous for the osteopetrosis spontaneous mutation (op/op mice) and M-CSF-deficient mice both have milder phenotypes than M-CSF-R-deficient mice7, suggesting that M-CSF-R may have additional ligand(s). Flt3 (Flk2, CD135) is broadly expressed on early hematopoietic precursors in the bone marrow, and injection of Flt3L was shown to increase the numbers of myeloid and lymphoid progenitors and peripheral DCs, monocytes and granulocytes. However, addition of Flt3L to cultured DC precursors has only modest effects on their proliferative expansion3, 8, 9, 10.
Local homeostatic control of their renewal within target tissues has recently been shown to be an important feature of the biology of DC and macrophage networks. Local proliferation is indeed sufficient for the renewal of microglia11 and Langerhans cells (LCs)12 throughout life in the steady state and also in a variety of diseases. It is only under defined conditions that microglia and Langerhans cells are replaced with bone marrow–derived Gr1hi CCR2+ monocytes. However, the molecular mechanisms that control LC and microglia homeostasis are not yet understood.
Although conventional DCs (cDCs) of lymphoid organs, monocytes and some tissue macrophages must be continually replenished in the periphery from a bone marrow common precursor designated as the macrophage DC progenitor (MDP)8, local homeostatic expansion is also crucial to regulating the size of the cDC pool. A previous study showed that the committed cDC precursors that enter the circulation rapidly migrate into lymphoid tissues, where they undergo cell division locally and differentiate into conventional CD11c+ CD8
+ and CD11c+ CD8-
DCs13. Thus, DC homeostasis is dependent on the rate of DC progenitor input from blood, cell division and cell death13.
Waskow et al.3 now uncover the molecular mechanisms of cDC local homeostatic control. They report on the effect of Flt3 on DC development in the bone marrow and in peripheral lymphoid organs and show that Flt3 is an important regulator of homeostatic cDC division in the periphery in vivo (Fig. 1). They show that DC precursors in the bone marrow and spleen do not require Flt3-mediated signals for their generation; by contrast, Flt3 was needed to obtain an adequate number of DCs in the spleen. Supraphysiological levels of Flt3L (produced by intravenous treatment with Flt3L) did promote DC expansion and emigration from the bone marrow. In Flt3-deficient mice, however, the numbers of bone marrow MDPs and the proportion of spleen pre-DCs were normal, but the number of spleen cDCs was markedly reduced.
Figure 1: Flt3 controls homeostasis of the dendritic cell pool in lymphoid organs.
Macrophage dendritic cell progenitors (MDPs) are a population of macrophages and DC precursor that are identified by the expression of the M-CSF receptor (CSF1R, cfms, CD115) and the absence of lineage markers and that overlap with the more recently described common DC precursors (CDPs). MDPs give rise to conventional CD11c+ CD8+ and CD11c+ CD8-
DCs, monocytes and macrophages, and they may also give rise to plasmacytoid DCs (pDCs)a, although pDC potential was not observed in the present study. DC progenitors in the bone marrow do not depend on Flt3 for their homeostatic maintenance in vivo; however, supraphysiologic levels of Flt3L increase their expansion and their emigration from the bone marrow. The present study shows that Flt3 is essential to maintain normal numbers of conventional DCs (cDCs) in lymphoid organs, by controlling locally the proliferation of DC precursors. In homeostatic conditions, a small number of DC precursors rapidly transit from bone marrow to spleen and lymph nodes via the blood as relatively immature cells that do not express surface MHC II or CD11c. These cells undergo cell divisions within lymphoid organs, under the control of Flt3L, and differentiate into conventional CD11c+ CD8+ and CD11c+ CD8-
DCs. Lymphotoxin- (LT1
2) has a similar role, restricted to the homeostasis of the CD8-
subset of cDCs (14). CMP, common myeloid progenitors; GMP, granulocyte/macrophage progenitors; HSC, hematopoietic stem cells.
The authors therefore hypothesized, and went on to demonstrate, that in homeostatic conditions, DC progenitors rapidly transit from bone marrow to spleen and lymph nodes via the blood—as relatively immature progenitors that do not express either surface major histocompatibility complex class II (MHC II) or CD11c- and that undergo a limited number of cell division in the lymphoid organs under the control of Flt3L. Using a bone marrow chimera, the authors showed that the DC deficiency occurring in Flt3-deficient mice was the result of a cell-intrinsic mechanism. Using parabiotic mice, they further demonstrated that Flt3 is a mediator of DC development after the MDP and circulating cDC precursor stage.
Finally, BrdU incorporation and CFSE dilution experiments indicated that Flt3 regulates DC cell division in the periphery. Notably, the numbers of plasmacytoid DC (pDC) were also reduced in Flt3-deficient mice, but Flt3-deficient pDCs showed no difference in BrdU incorporation compared to wild-type pDCs, suggesting that another mechanism, distinct from homeostatic control of proliferation, underlies the paucity of pDCs in Flt3-deficient mice. It is not known whether Gr1hi CCR2+ spleen monocytes or inflammatory 'TipDCs' are reduced in steady state or during infection in Flt3-deficient mice.
In a very elegant series of experiments, a recent study previously identified local homeostatic expansion of CD8-
cDC, controlled by lymphotoxin-, as being critical in maintaining the size of the CD8-
cDC pool in the spleen14. At that time, the results were interpreted as suggesting that the renewal of this cDC subset is independent of bone marrow input. Together with the work of Liu et al.13, the present study provides the conceptual framework to unify these results. The peripheral cDC compartment is continually replenished from blood borne precursors that undergo cell division in peripheral lymphoid organ while differentiating into cDCs. Both Flt3 and lymphotoxin- are required, at different levels, for the homeostatic proliferation of spleen DC precursors.
The data from Waskow et al.3 also refine the current understanding of cDC differentiation by showing that MDP, originally defined as lineage-negative (Lin-
) cells expressing a CX3CR1 promoter–driven GFP transgene and low CD117 and responding to M-CSF via the M-CSF receptor (CSF1R, CD115)8, can be more simply defined as Lin-
CSF1R+ bone marrow cells. Moreover, the authors show that the recently described common DC precursor (CDP)9, 10 that has been proposed to be another stem or branch for DCs development was in fact included in the MDP population. This was expected, as CDPs have been described as a Lin-
IL7R-
CD135+ CD115+ CD117lo, whereas MDPs were originally described as Lin-
IL7R-
CD117lo cells with a functional M-CSF receptor (CD115) and high levels of Flt3 (CD135) and M-CSF receptor (CD115) mRNA8.
However, MDPs and CDPs were reported to differ in their differentiation potential. CDPs gave rise to cDCs and pDCs in vivo9, 10, whereas MDPs did not8. Waskow et al.3 speculate that either Flt3L treatment of the recipients may be important to detect MDP-derived PDCs or the CDPs may have been contaminated with small numbers of multipotent progenitors. We recently observed that PDCs express CX3CR1, and, in contrast to our initial description, we have also been able to detect PDCs of donor origin after adoptive transfer of MDPs (unpublished data). More work may be needed to fully understand the origin and differentiation pathway of PDCs.
A second key difference between MDPs and CDPs lies in CDPs' lack of response to M-CSF9, 10. Although both cell types are defined by the expression of the M-CSF receptor, MDPs proliferate in vitro in response to M-CSF and gave rise to macrophage-like cells8, whereas CDPs do not9, 10. Waskow et al.3 found a poor proliferative response of MDPs to M-CSF, and they speculate that the lack of response to M-CSF by MDPs in their study—and of CDPs in previous studies—may be due to the isolation of MDPs and CDPs using anti-CD115. Indeed, we have observed that MDPs isolated using anti-CD115 presented with an impaired cloning efficiency and proliferation in response to M-CSF (unpublished data). If confirmed, this observation would imply that using anti-CD115 may not be the ideal strategy to isolate MDPs or CDPs for functional studies.
In conclusion, the present study by Waskow et al.3, together with previous work from the same group13, provides a conceptual framework to further explore the mechanisms of DC homeostasis in lymphoid organs. They describe cDC homeostasis as being dependent on the production of MDPs in the bone marrow, the rate of DC progenitor input from blood, local Flt3-dependent cell division and cell death. Importantly, this work also shows that, as could have been expected from the original publication8, the DC progenitors MDPs and CDPs constitute overlapping populations in the bone marrow.
