Dendritic Cell and Macrophage Heterogeneity In Vivo

Abstract

Macrophage and dendritic cell (DC) are hematopoietic cells found in all tissues in the steady state that share the ability to sample the environment but have distinct function in tissue immunity. Controversies remain on the best way to distinguish macrophages from DCs in vivo. In this Perspective, we discuss how recent discoveries in the origin of the DC and macrophage lineage help establish key functional differences between tissue DC and macrophage subsets. We also emphasize the need to further understand the functional heterogeneity of the tissue DC and macrophage lineages to better comprehend the complex role of these cells in tissue homeostasis and immunity.

History of the Discovery of the Mononuclear Phagocyte System

In 1908, Elie Metchnikoff (1845–1916), together with Paul Ehrlich, was awarded the Nobel Prize for his work on phagocytosis (Kaufmann, 2008). Metchnikoff classified phagocytes (named after the Greek “phago” for devour and “cytes” for cells) into “macrophages” (large eaters) and “microphages” (a smaller type of phagocytic cell, the polymorphonuclear leukocyte now known as granulocytes) and argued that both types of phagocytes played an important role in host resistance against infections. Metchnikoff also recognized the close relationship between mononuclear phagocytic cells in the spleen, lymph nodes, bone marrow, and connective tissues, leading him to introduce for the first time the term “macrophage system” (Gordon, 2008). Karl Albert Ludwig Aschoff, a German physician and pathologist, developed this concept further and grouped several cell types into what he called the reticulo-endothelial system and subsequently the reticulo-histiocyte system. The reticuloendothelial system included reticular cells (or fixed macrophages) of the spleen and lymph nodes, endothelial cells of the lymph and blood sinuses, monocytes, and histiocytes, a term referred to as “tissue wandering as opposed to fixed” macrophages. These cells were grouped on the basis of their capacity to uptake vital dye in vivo, an assay thought to measure phagocytic activity. However, poorly phagocytic cells, such as endothelial cells can also become labeled as a result of pinocytosis, especially when large amounts of dye are applied. Such labeling is therefore unreliable as a criterion for the identification of mononuclear phagocytes and a new classification was devised in 1969 to group only highly phagocytic cells and their precursors in one system called the mononuclear phagocyte system (MPS), a term that is still in use today (van Furth et al., 1972).

In the early 1970s, Ralph Steinman and Zanvil Cohn identified a population of hematopoietic cells in the mouse spleen that excel at antigen presentation and T cell stimulation, named dendritic cells (DCs) (Steinman and Cohn, 1973, 1974). Similar to macrophages, DCs are derived from hematopoietic progenitors and populate most lymphoid and nonlymphoid tissues. Initially thought to be poorly phagocytic, there is now evidence that DCs have a potent phagocytic activity that is mostly dedicated to inform the adaptive immune system about peripheral tissue cues through their unique ability to sample tissue antigens, migrate to the draining lymph nodes, present extracellular antigens, and initiate tissue-specific T cell immunity (Mellman and Steinman, 2001).

Accumulating evidence now suggests that macrophages and DCs are tissue phagocytic cells that specialize in the sensing and sampling of the tissue environment. Although the term MPS has created a framework that helped our understanding of tissue phagocyte biology, it also led to the assumption that all tissue phagocytes are identical in origin and function. Here, we describe the limitation of the MPS definition and its failure to account for the heterogeneous origin of tissue phagocytes. We also argue that defining DC and macrophage population on the basis of their origin should help understand the discrete role of these cells in tissue immunity and homeostasis.

Heterogeneity of Tissue-Resident DCs

DCs are classified into two main subsets that include classical DCs and plasmacytoid DCs (Heath and Carbone, 2009). Because the main theme of this Perspective is to discuss tissue phagocyte heterogeneity, plasmacytoid DCs will not be discussed here, considering that the ability of these cells to populate nonlymphoid tissues in the steady state as well as to phagocytose, process, and present extracellular tissue antigens is limited (Villadangos and Young, 2008).

Classical DCs (referred to as DCs here) form a heterogeneous population of hematopoietic cells that reside in all lymphoid, interface, and connective tissues (Figure 1). One of the biggest challenges to understand the molecular mechanisms that control DC function has been the lack of specific markers to define the DC lineage. DCs continue to be defined to date as hematopoietic cells lacking hematopoietic lineage markers while expressing high amounts of MHC class II (shared by B cells and activated macrophages) and the integrin CD11c (also expressed by some macrophage population, activated T cells, B cells and NK cells).

Figure 1. Origin of Batf3-IRF8-Id2-Dependent and -Independent Tissue-Resident DCs in Mice.

Lymphoid and peripheral tissues have at least four distinct resident DC subsets: CD4⁻ CD8α⁺, CD4⁺ CD8α⁻, CD103⁺ CD11b⁻, and CD103⁻ CD11b⁺ DCs. A CD103⁺ CD11b⁺ DC subset has been best characterized within the small intestine. This figure illustrates the precursors, transcription factors (black), and cytokine receptors (red) required for the development of each population. Commitment to the mononuclear phagocyte lineage is determined at the stage of the macrophage-dendritic cell progenitor (MDP). Granulocyte/macrophage progenitor (GMP) give rise to MDP, which has lost granulocyte potential and gives rise only to monocytes and DCs restricted precursors (common DC precursors [CDPs]). CDPs have lost monocyte-macrophage differentiation potential and give rise exclusively to plasmacytoid DCs (pDCs) and pre-DCs circulating precursors that migrate to lymphoid tissues where they differentiate into lymphoid tissue CD4⁻CD8α⁺ and CD4⁺CD8α DCs and to nonlymphoid tissue to give rise to CD103⁺CD11b⁻ DCs and CD103⁺CD11b⁺ DCs. Flt3 controls myeloid precursors commitment to the DC lineage as well as the differentiation of mature DCs in tissue. The transcription factors Batf3 Id2 and IRF8 control the differentiation of lymphoid tissue CD8⁺ DC and nonlymphoid tissues CD103⁺CD11b⁻ DC but do not control CD11b⁺ DC differentiation in lymphoid and nonlymphoid tissues.

Progress has been made in recent years in understanding the origin and the transcriptional program that control DC differentiation in vivo. These results allowed the study of homogenous cell populations that derive from committed precursors along a distinct differentiation program shifting the focus from studies on phenotypically defined subsets to developmentally regulated subsets. Below we discuss the importance of revisiting the role of DCs by separating the contribution of these developmentally distinct DC subsets in tissue immunity.

The Batf3-IRF8-Id2 DC Lineage

The Batf3-IRF8-Id2 DC lineage refers to a group of DCs located in lymphoid and non-lymphoid tissues that share a common origin, phenotype and function.

Lymphoid Tissue CD8⁺ DCs and Nonlymphoid Tissue CD103⁺CD11b⁻ DCs Have a Common Origin

As shown in Figure 1, the presence in lymphoid tissues of a discrete population of DCs that expressed the CD8αα homodimer but lacked the CD8β chain and CD11b marker, most commonly referred to as CD8⁺ DCs, has been known for more than a decade (Shortman and Heath, 2010). Several studies have now established that CD8⁺ DCs arise from a distinct differentiation program that is dependent on key transcription factors that include Batf3 (Hildner et al., 2008) and IRF8 (Aliberti et al., 2003; Schiavoni et al., 2002; Tailor et al., 2008) and the inhibitor of DNA protein Id2 (Hacker et al., 2003). Absence of any of these factors prevents the differentiation of CD8⁺ DCs in lymphoid organs, but does not affect the differentiation of CD8⁻ DCs (Hacker et al., 2003; Hildner et al., 2008; Tailor et al., 2008). Data have also shown that the mammalian target of rapamycin (mTOR), promotes the differentiation of CD8⁺ DCs and that the loss of the tumor suppressor gene PTEN, a negative regulator of mTOR, expands the number of CD8⁺ DCs in murine lymphoid tissues (Sathaliyawala et al., 2010).

A DC subset equivalent to lymphoid tissue CD8⁺ DCs has been identified in nonlymphoid tissues. These cells lack the CD8 marker and express the integrin CD103 (Bedoui et al., 2009; del Rio et al., 2007; Sung et al., 2006). Similar to CD8⁺ DCs, they express low amounts CD11b (del Rio et al., 2007; GeurtsvanKessel et al., 2008; Sung et al., 2006) and lack macrophage markers such as CD172a, F4/80, and CX₃CR1 (Ginhoux et al., 2009). Nonlymphoid tissue CD103⁺CD11b⁻ DCs share the same differentiation requirements as CD8⁺ DCs. They are absent in Batf3-deficient (Edelson et al., 2010), Id2-deficient (Ginhoux et al., 2009; Edelson et al., 2010) and IRF8-deficient mice (Ginhoux et al., 2009) and expand in PTEN-deficient mice (Sathaliyawala et al., 2010). Of note, CD103 is not a specific marker of Batf3-IRF8-Id2-dependent DCs as CD103 is also expressed on a subset of CD11b⁺ DCs in the lamina propria (Coombes et al., 2007; Jaensson et al., 2008; Johansson-Lindbom et al., 2005; Sun et al., 2007). Lamina propria CD103⁺CD11b⁺ DCs develop normally in Batf3-deficient and IRF8 deficient mice (Edelson et al., 2010) and in Id2-deficient mice (M. Bogunovic, F. Ginhoux and M.M., unpublished data) as discussed below.

Importantly, lymphoid tissue CD8⁺ DCs and nonlymphoid tissue CD103⁺CD11b⁻ DCs, which will be referred to as Batf3-IRF8-Id2 DCs in this Perspective, are critically dependent on the cytokine fms-like thyrosine kinase 3 ligand (Flt3L) and its receptor (Flt3) for their development and homeostasis (Figure 1, Table 1). Flt3 and Flt3L are key regulators of DC commitment in hematopoiesis. Flt3 is expressed on short-term repopulating hematopoietic stem cells and is progressively extinguished on most hematopoietic cell lineages with the exception of the DC precursors (Merad and Manz, 2009). Loss of Flt3 expression in hematopoietic progenitors correlates with the loss of DC differentiation potential, whereas enforcement of Flt3 expression on Flt3-negative progenitors rescues their ability to differentiate into DC (Onai et al., 2006). Flt3 expression is maintained in lymphoid tissue DCs and Flt3L has been shown to control the proliferation and homeostasis of DCs in lymphoid organs (Liu et al., 2007; Waskow et al., 2008). Importantly, nonlymphoid tissue CD103⁺CD11b⁻ DCs also express high amounts of Flt3 receptor on the cell surface and are more severely reduced in mice reconstituted with Flt3-deficient hematopoietic progenitors compared to CD103⁻ DCs or tissue macrophages (Ginhoux et al., 2009). It is interesting to note that in bone marrow chimeric mice reconstituted with Flt3-deficient hematopoietic progenitors, CD103⁺CD11b⁻ DCs are completely absent in nonlymphoid tissue whereas CD8⁺ DCs are reduced by one-third in lymphoid tissues (Ginhoux et al., 2009). These results are likely to reflect the more stringent role of Flt3 in maintaining DC homeostasis in nonlymphoid tissues compared to lymphoid organs.

Table 1.

Dendritic Cell Heterogeneity

Pattern recognition receptors
	TLR3	TLR7	TLR11–12	RIG-I	MDA5	NOD1	NLRC4
Batf3-dependent DC	+	−	+	−	−	−	−
Batf3-independent DC	−	−/+	−/+	+	+	−/+	−/+
References	Edwards 2003	Edwards 2003	Yarovinsky 2005	Luber 2010	Luber 2010	Luber 2010	Luber 2010

Lectin-like receptors
	CD205	CD207	CD209	33D1	Clec7a	Clec9a	Clec12a	CD24
Batf3-dependent DC	++	int/−	−	−	−	+	++	++
Batf3-independent DC	+	−	+	−/+	+	−	+	+
References	Shortman 2010	Shortman 2010	Our data	Dudziak 2007	Shortman 2010	Shortman 2010	Shortman 2010	Shortman 2010

Other differential cell markers
	CD36	Sirpa	CX3CR1	F4/80	XCR1	Cystatin C	Necl2
Batf3-dependent DC	+	−	−	−	+	+	−
Batf3-independent DC	−	+	−/+	+	−	−	+
References	Shortman 2010	Lahoud 2006	Ginhoux 2009	Ginhoux 2009	Domer 2009	Shortman 2010	Galibert 2005

This table summarizes published data on differential expression of Toll-like receptors (Edwards et al., 2003; Luber et al., 2010; Yarovinsky et al., 2005), lectin-like receptors (Dudziak et al., 2007; Shortman and Heath, 2010), and other cell markers (Dorner et al., 2009; Galibert et al., 2005; Ginhoux et al., 2009; Lahoud et al., 2006; Shortman and Heath, 2010) between Batf3-IRF8-Id2 dependent DCs (including CD8⁺and/or CD103⁺DCs) and Batf3-IRF8-Id2- independent DCs.

In contrast, lymphoid tissue CD8⁺ DCs and nonlymphoid tissue CD103⁺CD11b⁻ DCs lack the receptor for colony stimulating factor 1 (Csf-1), a cytokine that controls the differentiation of many tissue macrophages in vivo (Chitu and Stanley, 2006) and remain unaffected in mice that lack Csf-1 receptor (Csf-1R) and in mice that are reconstituted with Csf-1 receptor (Csf-1R)-deficient hematopoietic progenitors (Ginhoux et al., 2009).

Lymphoid tissue CD8⁺ DCs and nonlymphoid tissue CD103⁺CD11b⁻ DCs are derived from DC-restricted precursors that have been identified in mice blood and bone marrow. Commitment to the mononuclear phagocyte lineage is determined at the stage of the macrophage-dendritic cell progenitor (MDP) (Auffray et al., 2009; Fogg et al., 2006), at which point, erythroid, megakaryocyte, lymphoid, and granulocyte fates have been precluded. MDPs give rise to monocytes and DCs restricted precursors (also called common DC precursors [CDPs]), which have lost monocyte-macrophage differentiation potential. CDPs give rise exclusively to plasmacytoid DCs and pre-DCs, circulating precursors that migrate to lymphoid tissues where they differentiate into lymphoid tissue CD8⁺ and CD8⁻ DCs (Naik et al., 2007; Onai et al., 2007). Subsequent studies revealed that pre-DCs also migrate to nonlymphoid tissues and give rise to nonlymphoid tissue CD103⁺CD11b⁻ DCs but also to CD103⁺CD11b⁺ DCs (Bogunovic et al., 2009; Ginhoux et al., 2009; Varol et al., 2009) (Figure 1).

Interestingly, Batf3 Id2 and IRF8 are differentially expressed along the DC lineage (Edelson et al., 2010; Jackson et al., 2011). IRF8 gene is expressed in DC-restricted precursors in the bone marrow and remained expressed in circulating precursors and in terminally differentiated CD8⁺ and CD103⁺ DC subsets, whereas its expression is extinguished in lymphoid tissue-resident CD8⁻ DCs, which also derive from DC-restricted precursors (Edelson et al., 2010; Jackson et al., 2011). Reporter mice and expression array data revealed that Id2 is absent in DC-restricted precursors in the bone marrow, starts to be expressed in pre-DCs that reach the spleen, increases in all DC subsets but reach higher amounts in CD8⁺ and CD103⁺ DCs (Edelson et al., 2010; Jackson et al., 2011). In contrast, Batf3 gene becomes expressed only in the terminal stages of DC development and is highly expressed by all DCs (Edelson et al., 2010; Jackson et al., 2011). The overlap of high IRF8, Id2, and Batf3 expression was only found within the CD8⁺ and CD103⁺ DC subsets, strongly suggesting that it is the combination of these three factors that controls the differentiation and/or function of lymphoid tissue CD8⁺ DCs and nonlymphoid tissue CD103⁺CD11b⁻ DCs.

CD8⁺ and CD103⁺CD11b⁻ DCs Share a Similar Gene Expression Pattern and a Similar Sensing Repertoire

Transcriptional analysis of purified lymphoid tissue CD8⁺ and nonlymphoid tissue CD103⁺CD11b⁻ DCs isolated from different tissue sites share a remarkably similar gene expression pattern (Edwards et al., 2003; Edelson et al., 2010) (see also www.immgen.org). Lymphoid tissue-resident CD8⁺ DCs share several sensing receptors with nonlymphoid tissue CD103⁺CD11b⁻ DCs. For example, lymphoid tissue CD8⁺ DCs are the only lymphoid tissue DCs expressing toll-like receptor 3 (TLR3) and TLR11– TLR12 but they lack TLR7 (Edwards et al., 2003; Shortman and Heath, 2010; Yarovinsky et al., 2005). Similarly, skin migratory CD103⁺CD11b⁻ DCs are the only migratory DCs expressing TLR3 (Jelinek et al., 2011); they also express high amounts of TLR11 transcripts and lack TLR7. Lymphoid tissue CD8⁺ DCs also lack the retinoic acid inducible gene (RIG-I), a sensor required to sense certain viruses in the cytoplasm (Luber et al., 2010), and similarly, RIG-I transcripts are expressed at much lower levels in nonlymphoid tissue CD103⁺CD11b⁻ DCs compared to other nonlymphoid tissue DC (www.immgen.org) (Table 1).

Lymphoid tissue CD8⁺ DCs express the lectin-like receptors CD207 and CD205, and the recently described Clec9A specialized in the sensing of necrotic bodies, whereas these lectins are absent or expressed at lower amounts in CD8⁻ DCs (Dudziak et al., 2007; Sancho et al., 2009; Shortman and Heath, 2010). Similarly, nonlymphoid tissue CD103⁺CD11b⁻ DCs express CD205 and can express CD207 at least in some tissues including skin and lung, whereas CD207 is always absent from lung and skin CD103⁻CD11b⁺ DCs (Helft et al., 2010). Our unpublished data also suggest that Clec9A is expressed specifically by nonlymphoid tissue CD103⁺CD11b⁻ DC and is absent from CD103⁻CD11b⁺ DC isolated from the same tissues.

Lymphoid tissue CD8⁺ DCs (Dorner et al., 2009) and nonlymphoid tissue CD103⁺CD11b⁻ DCs (J. Helft and M.M., unpublished data) also uniquely express the chemokine receptor XCR1.

CD8⁺ and CD103⁺ CD11b⁻ DCs Share a Common Function

Lymphoid tissue CD8⁺ DCs and nonlymphoid tissue CD103⁺ CD11b⁻ DCs are very potent at producing interleukin-12 (IL-12), cross-presenting antigens to CD8⁺ T cells, and inducing the differentiation of effector CD8⁺ T cells (Sung et al., 2006; del Rio et al., 2007; GeurtsvanKessel et al., 2008; Henri et al., 2010; Kim and Braciale, 2009; Bedoui et al., 2009; Heath and Carbone, 2009; Shortman and Heath, 2010). The GTPase Rac2 specifically expressed by CD8⁺ DCs and absent from CD8⁻ DCs is thought to control antigen processing and cross-presentation through the regulation of phagosomal pH and oxidation (Savina et al., 2006; Savina et al., 2009). Although likely, it remains to be determined whether a similar machinery is also in place in nonlymphoid tissue CD103⁺CD11b⁻ DCs. The increased ability to interact with CD8⁺ T cells was also recently shown to be dependent on the expression of XCR1, a chemokine receptor which ligand XCL1 is produced by naive CD8⁺ T cells (Dorner et al., 2009). XCR1 is restricted to lymphoid tissue CD8⁺ DCs (Dorner et al., 2009) and nonlymphoid tissue CD103⁺CD11b⁻ DCs (unpublished data) and is absent from Batf3-IRF8-Id2 independent DCs.

In agreement with their unique TLR3 expression, CD8⁺ DCs and CD103⁺CD11b⁻ DCs share a superior ability to respond to TLR3 ligand adjuvant (Hochrein et al., 2001; Jelinek et al., 2011; Longhi et al., 2009; Sung et al., 2006). In addition, recent data revealed that CD8⁺ DCs are also the only lymphoid tissue DC subset that produces interferon lambda in response to the TLR3 ligand Poly: I:C (Lauterbach et al., 2010), and similar results were found for lung CD103⁺CD11b⁻ DC (unpublished data).

Overall, these results show that Batf3-IRF8-Id2-dependent DCs regardless of their tissue localization share a similar origin, differentiation program, sensing repertoire, and T cell stimulatory function, suggesting that at least some diversity of the DC lineage is determined at the precursor stage.

Batf3-IRF8-Id2 DC Equivalent in Humans and Other Species

In comparison to mice DCs, considerably less is known about the origin of human DCs, their differentiation program and their functional differentiation in situ due to their rarity in blood and poor accessibility of human tissues, with the exception of the skin. Circulating DC subsets have been distinguished based on three cell surface markers including CD1c (or BDCA1) expressed on the majority of circulating DCs in humans, CD141 (also called BDCA3) expressed on a minute population, and Fcγ receptor III (CD16) (Dzionek et al., 2000; Huysamen et al., 2008; Schäkel et al., 2002; Ueno et al., 2010).

Similar to Batf3-IRF8-Id2-dependent DCs in mice, CD141⁺ CD1c⁻ DCs were found to uniquely express the lectin Clec9A and the chemokine receptor XCR1 and excel in the production of IL-12 and the cross-priming of CD8⁺ T effector cells (Bachem et al., 2010; Crozat et al., 2010; Jongbloed et al., 2010; Lauterbach et al., 2010; Poulin et al., 2010). CD141⁺CD1c⁻ DCs were also found to uniquely express TLR3 and to be the main producers of interferon lambda in response to poly:I:C activation (Lauterbach et al., 2010). However, the human CD1c⁺ DC subset also shared some functional properties with mice Batf3-IRF8-Id2 DCs given that they were found to produce high amounts of IL-12 and to cross-prime antigens to CD8⁺ T cells (Mittag et al., 2011). Interestingly, the genetic analysis of patients with disseminated mycobacterial infection identified two distinct mutations affecting IRF8 transcriptional activity. One mutation led to the complete depletion of DCs and circulating monocytes (Hambleton et al., 2011). A distinct mutation led to the specific depletion of circulating CD1c⁺ DCs, whereas CD141⁺ DCs remained intact in these patients (Hambleton et al., 2011). These results suggest that IRF8 transcription factors also controls DC differentiation in humans but, in contrast to mice IRF8, is also involved in the development of monocytes. Further studies are required to fully understand the transcriptional program that controls DC development in humans, which could potentially differ from mice. More studies are also required to further understand the functional heterogeneity of human DC subsets and to assess whether the phenotypic markers currently used to define these subsets in the blood help define functionally distinct populations.

Interestingly, analysis of skin migratory DC isolated from the sheep skin draining lymphatic vessels revealed that the presence of a DC subset, which, similar to mouse Batf3-IRF8-Id2 DCs, lacks the macrophage marker CD172a, expresses high amounts of CD205, Clec9A, and XCR1, and excels in the cross-presentation of soluble antigens and in the differentiation of CD8⁺ effector T cells (Contreras et al., 2010). The molecular program that controls the development of this functional and phenotypically distinct DC subset remains to be analyzed but if identical to mice, these results will support the presence of an evolutionary conserved DC subset further emphasizing the functional relevance of this DC population to tissue immunity.

The Batf3-IRF8-Id2-Independent DC Lineage

The Batf3-IRF8-Id2 independent DC lineage refers to DCs that develop independently of Batf3, Id2, and IRF8. Phenotypically, Batf3-IRF8-Id2 independent DCs are more heterogeneous than Batf3-IRF8-Id2 -dependent DCs. Although they always express CD11b and lack CD8, they are heterogenous for CD103, CD172a, F4/80, and CX₃CR1 expression (Ginhoux et al., 2009; Helft et al., 2010). The transcription factors IRF4, IRF2 (Ichikawa et al., 2004; Suzuki et al., 2004), and RBPJ (Caton et al., 2007) partly control the differentiation of CD8⁻ DCs in lymphoid organs, but the molecular program that controls the differentiation of Batf3-IRF8-Id2 -independent DCs in nonlymphoid tissues needs to be clarified and the exact relationship between Batf3-IRF8-Id2 -independent DCs in lymphoid versus nonlymphoid tissues remains unclear.

The heterogeneity of Batf3-IRF8-Id2 independent DCs has been most clearly identified in the lamina propria, where they consist of two distinct DC populations best distinguished on the basis of CD103 expression. Lamina propria Batf3-IRF8-Id2 independent DCs include CD103⁺CD11b⁺ DCs and CD103⁻CD11b⁺ DC subsets. CD103⁺CD11b⁺ lamina propria DCs express low amounts of the macrophage markers CD172a, F4/80 and CX₃CR1 and low Csf-1R. They express high amounts of Flt3 receptor and are dependent on Flt3L and GM-CSF (also called Csf-2) for their development, whereas they are unaffected in Csf-1R deficient mice (Bogunovic et al., 2009; Schulz et al., 2009; Varol et al., 2009). Adoptive transfer studies of purified precursors revealed that lamina propria CD103⁺CD11b⁺ DCs derive from DC-restricted precursors (Bogunovic et al., 2009; Varol et al., 2009) and potently migrate to the draining lymph nodes upon oral microbial stimuli (Bogunovic et al., 2009; Schulz et al., 2009).

In contrast, lamina propria CD103⁻ CD11b⁺ cells express the macrophage markers CD172a, F4/80, CX₃CR1, and the Csf-1R. They are derived from circulating monocytes and not from DC restricted precursors and require Csf-1R but not Flt3, Flt3L, Csf-2, or Csf-2R for their development (Bogunovic et al., 2009; Schulz et al., 2009; Varol et al., 2009). They are much less potent than CD103⁺CD11b⁺ DCs at stimulating T cell proliferation (Schulz et al., 2009) and migrate poorly to the draining lymph nodes during oral infection (Bogunovic et al., 2009; Ginhoux et al., 2009; Varol et al., 2009), suggesting that this population despite expression of high MHC II and CD11c, probably corresponds to tissue macrophages. Altogether, CD103 expression helps decipher the heterogeneity of Batf3-IRF8-Id2-independent CD11b⁺ DCs in the lamina propria, which consists of CD103⁺CD11b⁺ DC and CD103⁻ CD11b⁺ macrophages.

However, such markers are currently lacking in other tissues. For example in the lung, CD11b⁺ DCs lack CD103 but are heterogeneous for macrophage markers and for the DC marker CD24. They are partly reduced in mice reconstituted with Flt3^−/− or Csf1r^−/− hematopoietic progenitors (Ginhoux et al., 2009), which probably reflect the heterogeneity of this DC population. The use of Flt3 or Csf-1R as markers of heterogeneity as well as other profiling tools together with lineage tracing and functional genomic studies as discussed below should help clarify the regulatory and functional attributes of this DC subset.

The Langerhans Cell Exception

Langerhans cells (LCs) refer to the DCs that reside in the epidermal layer of the skin (Merad et al., 2008). These cells stand apart from other tissue DCs because they have unique ontogeny and homeostatic properties (Ginhoux and Merad, 2010). Phenotypically, LCs express lower MHC II, intermediate CD11c, and very high amounts of the lectin receptor CD207, which is also expressed, although at lower amounts, by Batf3-IRF8-Id2-dependent DCs (Merad et al., 2008). They express CD11b, F4/80 and lack CX₃CR1 (Merad et al., 2008). In contrast to most tissue DCs, they arise from embryonic precursors that are recruited to the skin prior to birth and are maintained in the skin throughout life in steady-state conditions (Chang-Rodriguez et al., 2005; Chorro et al., 2009; Merad et al., 2002). LCs are absent in Id2-deficient mice (Hacker et al., 2003) and have been reported to be reduced in IRF8-deficient mice (Schiavoni et al., 2004). However, epidermal LCs remained unaffected in patients with mutations impairing IRF8 gene function (Hambleton et al., 2011), which is consistent with our unpublished data showing that epidermal LC are not reduced in numbers in IRF8-deficient mice but express reduced MHC II levels.

In agreement with their distinct origin and homeostasis, LCs express a very distinct gene expression pattern compared to other DCs (www.immgen.org). In contrast to most DCs, LCs lack Flt3 and develop independently of the DC cytokine Flt3 ligand but require Csf-1R but not Csf-1 for their development (Ginhoux et al., 2006). These results came as a surprise given that Csf-1R was thought to control mainly macrophage development (Chitu and Stanley, 2006). In the epidermis, LCs are highly phagocytic and have low MHC class II expression on the cell surface, but migratory LC that reach the lymph nodes become indistinguishable phenotypically from other migratory DC (Merad et al., 2008). LC are also thought to modulate skin contact hypersensitivity response (Kaplan et al., 2005) and are required to induce antigen specific T helper 17 cell differentiation in response to skin candida albicans infection (Igyártó et al., 2011). In vitro studies using primary LCs isolated from mice and humans also suggest that LCs are very potent at initiating T cell response and at cross-presenting cell-associated antigens (Klechevsky et al., 2008; Stoitzner et al., 2008).

Monocyte-Derived DC and Tip DC

Monocytes can easily be differentiated into DCs in vitro (Sallusto and Lanzavecchia, 1994,) but the exact contribution of monocytes to the DC pool in the steady state remains unclear. In contrast to the steady state, monocyte differentiation into DC dramatically increases in inflamed tissues. The term TNF-iNOS DC (Tip DC) was first used to describe monocytes that differentiate into DCs in the spleens of listeria-infected animals (Serbina et al., 2003). Tip DCs were described as cells expressing Gr-1 (Ly6C), CD163, intermediate MHC II, and the integrin CD11c and producing TNF and iNOS (Serbina et al., 2003). This term was subsequently extended to all inflammatory monocyte-derived DCs without always providing supportive data showing TNF and iNOS secretion. Recent studies identified new means to distinguish monocyte-derived DCs in inflamed lymphoid organs, which should help the purification and the study of these cells (Cheong et al., 2010). The cytokines and transcription factors that control monocyte-derived DC differentiation remain unclear. Csf-2, a cytokine essential for the generation of in vitro-derived DCs, is thought to be critical for the differentiation of inflammatory DCs, although definite proof of the role of this cytokine in DC differentiation in vivo remains to be established (Shortman and Naik, 2007).

In Vitro-Derived DC

The identification of in vitro culture conditions that promote the differentiation of DCs contributed substantially to the expansion of the DC field (Caux et al., 1992; Inaba et al., 1992; Sallusto and Lanzavecchia, 1994). In mice, bone marrow progenitors represent the main source of in vitro-derived DCs, whereas in humans, DCs are generated from circulating monocytes or CD34⁺ hematopoietic progenitors. In DC cultures, Csf-2 is a key cytokine to drive DC differentiation (Caux et al., 1992; Inaba et al., 1992; Sallusto and Lanzavecchia, 1994). However, DC populations obtained in Csf-2-driven cultures do not faithfully resemble DC subsets identified in vivo. These results together with the realization that Flt3L rather than Csf-2 drives DC differentiation in vivo (Vremec et al., 1997) have led to substitution of Flt3L for Csf-2 in DC differentiation cultures (Naik et al., 2007). Bone marrow-derived DCs cultured in the presence of Flt3L leads to the differentiation of DC subsets that phenotypically and functionally resemble CD8⁺ and CD8⁻ DCs (del Rio et al., 2008; Naik et al., 2007). Further analysis of these DC populations both at the transcriptional and functional levels should help establish the physiological relevance of this in vitro differentiation DC system.

Heterogeneity of Tissue Macrophages

Macrophages are hematopoietic cells that populate every tissue including the brain. Whereas DCs are focused on initiating tissue immune responses, macrophages main role is to ensure tissue integrity. Functions shared by most tissue macrophages include a high phagocytic function and degradative potential, allowing them to clear foreign and damaged cells (Gordon and Taylor, 2005). Macrophages also participate in the induction of innate immunity in response to tissue infection, which plays a critical role in the killing of micro-organisms and their pathogenic factors (Gordon and Taylor, 2005). They can also load extracellular antigen in MHC class II compartments but they are less efficient than DCs at priming naive T cells and mostly interact with effector CD4⁺ T cells. Different names have been used to identify macrophage populations in vivo, which are mostly defined on the basis of a set of defined cell surface markers (Table 2). These names can reflect those of the scientists who discovered these cells (i.e., Kupffer cells in the liver) or the location in which they reside (i.e., marginal zone macrophages and the red pulp macrophages of the spleen, subcapsular sinus, and medullary macrophages in lymph nodes), but are rarely informative of the cells’ origin and function. Similar to DCs, the lack of genetic tools targeting specifically the macrophage lineage compromises our understanding of the regulatory mechanisms that control macrophage development and function in vivo. The recent development of engineered mouse models to track, deplete, modify, and alter the macrophage compartment will undoubtedly transform our view of the macrophage lineage.

Table 2.

Heterogeneity of Tissue Macrophages in Mice

Lymphoid tissue macrophages
Spleen		Lymph node	Bone marrow	Thymus
Phenotype and Localization	Red pulp MØ -Entire red pulp -F4/80hiCD11bloCD169loMHCIIlo CD163+CD68+CD115+CD172a+	Subcapsular sinus MØ -Lining subcapsular sinus -F4/80loCD11b+CD169+CD11clo MARCO+	Bone marrow CD169+ MØ -Distributed throughout the marrow but enriched around the hematopoietic stem cell niche F4/80+CD11bloCD169+CD11clo CD68+CX3CR1−CD115+	Subcapsular MØ -MHCII+F4/80+Mac-2−FcγRII+FcγRIII+
	Marginal zone MØ -Outer layer of marginal zone sinus -F4/80−SIGN-R1+MARCO+	Medullar MØ -Lining medullar sinus -F4/80hiCD11b+CD169+CD11clo		Cortex MØ -MHCII−F4/80+Mac-2+FcγRII+FcγRIII+
	Marginal zone metallophilic MØ -Inner layer of marginal zone sinus -F4/80−CD169+	CD11chiCD169+ MØ -Boundary between the sinus and the T cell zone or B cell follicle -MHCII+F4/80+CD169+CD11chiCD8+		Cortico-medullary MØ -MHCII+F4/80+Mac-2+ FcγRIIhiFcγRIIIhi
	Tingible body MØ -Germinal center -F4/80−CD11b−CD68+MFG-E8+	Tingible body MØ -Germinal center -F4/80−CD11b−CD68+MFG-E8+		Medulla MØ -Germinal center -MHCIIhiF4/80+Mac-2+ FcγRII+FcγRIII+
References	Lloyd et al., 2008 Kohyama et al., 2009 Miyake et al., 2007 You et al., 2011 Hanayama et al., 2004 Rabinowitz et al., 1991	Phan et al., 2009 Asano et al., 2011	Chow et al., 2011 Crocker and Gordon, 1986	Soga et al., 1997

Non-lymphoid tissue macrophages
Liver		Lung	Gut	Skin	Kidney
Phenotype and Localization	Kupffer cell -In the liver sinusoid -F4/80hiCD11bloCD169+ CD68+Mac-2+	Alveolar MØ -Alveolar airspace -F4/80+CD11bloCD169+ CD11chiCD68+SiglecF+ MARCO+Mac-2+	Lamina propria MØ -Intestinal lamina propria -MHCII+F4/80+CD11b+ CD11c+CD103−CD115+ CX3CR1+CD172a+	Dermal MØ -F4/80+CD11b+CD11clo mMGL+CD206+MHCIIlo In deep dermis, MØ can express CD169+	Kidney MØ -F4/80+CD11bloCD103− CX3CR1+CD172a+
		Interstitial MØ -Alveolar interstitium -CD11c−F4/80+CD68+ MHCII+	Serosal DC or MØ -Intestinal muscular layer and serosa -MHCIIhiF4/80+CD11b+ CD169+CD11cloCX3CR1+ CD103−CD115+
References	Flotte et al., 1983 Kinoshita et al., 2010 Crocker and Gordon, 1989	Flotte et al., 1983 Tateno et al., 2007 Ducreux et al., 2009 Palecanda et al., 1999 Bedoret et al., 2009 Lagranderie et al., 2003	Bogunovic et al., 2009 Flores-Langarica et al., 2005	Dupasquier et al., 2004 Dupasquier et al., 2006	Our unpublished data

This table describes the name, localization, and phenotype of macrophage populations that reside in the spleen (Hanayama et al., 2004; Kohyama et al., 2009; Lloyd et al., 2008; Miyake et al., 2007; Rabinowitz and Gordon, 1991; You et al., 2011), lymph nodes (Phan et al., 2009); (Asano et al., 2011), bone marrow (Chow et al., 2011; Crocker and Gordon, 1986), thymus (Soga et al., 1997), liver (Crocker and Gordon, 1989; Flotte et al., 1983; Kohyama et al., 2009), lungs (Bedoret et al., 2009; Ducreux et al., 2009; Flotte et al., 1983; Lagranderie et al., 2003; Palecanda et al., 1999; Tateno et al., 2007), gut (Bogunovic et al., 2009; Flores-Langarica et al., 2005), and skin (Dupasquier et al., 2004; Dupasquier et al., 2006).

Regulation of Macrophage Development

Although the current dogma suggests that most tissue macrophages derive from circulating monocytes that originate in the bone marrow, direct evidence that circulating monocytes contribute to all tissue macrophages is lacking. Adoptive transfer studies revealed that monocytes contribute to the red pulp macrophages (Fogg et al., 2006) and the mucosal macrophage (Varol et al., 2007; Bogunovic et al., 2009; Varol et al., 2009) pool, but further studies are needed to confirm the contribution of circulating monocytes to other tissue macrophage subsets. There is now clear evidence that macrophages can proliferate locally in the steady state and in response to specific tissue injury, and the contribution of circulating precursors versus self-renewal to tissue macrophage homeostasis remains to be established (Gordon and Taylor, 2005; Jenkins et al., 2011; Randolph, 2011). The capacity of macrophage to self-renew and contribute to their homeostasis highlights one of the limitations of the MPS definition that suggests that all tissue phagocytes are derived from a mobile pool of circulating monocytes.

The dramatic reduction of tissue macrophages observed in Csf1r^−/− mice, that lack the Csf-1R (Dai et al., 2002) and Csf1^op/op mice (Yoshida et al., 1990), that have a natural null mutation in the Csf-1 cytokine gene, established the key role of Csf-1 and its receptor in macrophage homeostasis in vivo (Pixley and Stanley, 2004). However, the precise role of Csf-1 and its receptor in monocyte commitment remains controversial. One hypothesis is that Csf-1 drives the differentiation of monocytes into tissue macrophages (Metcalf, 1985), whereas a different hypothesis suggests that Csf-1 provides a survival signal to the differentiating monocytes and that surviving cells utilize an intrinsic developmental program to become mature macrophages (Korn et al., 1973; Lagasse and Weissman, 1997; Nakahata et al., 1982). Importantly, some tissue macrophages are more profoundly affected in the absence of Csf-1R than in the absence of its ligand Csf-1. For example, the brain resident macrophages, also called microglia and the epidermal LCs, develop normally in Csf1^op/op mice but are absent in Csf1r^−/− mice (Ginhoux et al., 2010; Ginhoux et al., 2006). Interestingly, a second CSF-1R ligand called interleukin 34 (IL-34) has been identified (Lin et al., 2008). Evolutionarily conserved IL-34 has been identified in humans, mice, and birds (Garceau et al., 2010). IL-34 is highly expressed in the brain of postnatal mice (Wei et al., 2010) and the exact role of IL-34 in the development and homeostasis of microglia and other tissue macrophages remains to be determined.

The transcriptional program that control monocyte-macrophage differentiation has been reviewed elsewhere (Auffray et al., 2009; Friedman, 2007; Geissmann et al., 2010). However, most of the studies on the transcriptional regulation of macrophage differentiation were done in vitro with progenitor-enriched populations, and the exact role of these factors in driving the differentiation of monocytes into distinct macrophage subsets in vivo remains to be examined. This is particularly important given that functional macrophage specialization is likely to be regulated at the tissue levels and not at the progenitor level.

Functional Specialization of Tissue Macrophages

Although the developmental pathways that give rise to distinct macrophage subsets are still unclear, there is evidence that different macrophage populations play distinct roles in vivo. In most tissues, macrophages promote tumor surveillance through their ability to capture and kill malignant hematopoietic cells (Jaiswal et al., 2010), but can also promote tumor growth through their ability to increase angiogenesis and immunosupression (Qian and Pollard, 2010). Osteoclasts represent a population of bone macrophages that specialize in bone resorption and are critical to maintain bone mass homeostasis (Edwards and Mundy, 2011). Bone marrow macrophages promote the retention of hematopoietic stem cells and the depletion of bone marrow macrophages mobilizes hematopoietic progenitors to the blood, enhancing yields of clinical mobilization regimen for hematopoietic cell transplants (Chow et al., 2011). In allogeneic hematopoietic cell transplant recipient, host lymphoid tissue macrophages reduce the allogeneic T cell pool and limit graft versus host reactions through their ability to capture live allogeneic T cells and inhibit T cell proliferation (Hashimoto et al., 2011). In lymph nodes, subcapsular macrophages provide a key barrier that protects from the systemic dissemination of viruses (Iannacone et al., 2010) and promote the presentation of viral antigens to B cells (Phan et al., 2009; Phan et al., 2007). These results emphasize the functional diversity of the macrophage network, which may be partly dictated by the environment in which the macrophage resides. In support of this hypothesis are the findings showing that whereas Csf-1R and the transcription factor PU.1 (Friedman, 2007) control the commitment to a common macrophage program, the transcription factors c-fos and MITF control specifically osteoclast differentiation (Edwards and Mundy, 2011), and the transcription factor Spi-C selectively controls the development of red pulp macrophages but no other macrophage population (Kohyama et al., 2009). Therefore, one of the biggest challenges in the macrophage field is to continue to explore the origin of the precursor cell population and the molecular program that control global macrophage commitment in bone marrow progenitors, but in addition, to decipher the regulatory program that controls macrophage specialization and function in the periphery.

The Microglia Exception

Microglia cells (from the Greek “glia” for glue) refer to cells in the brain that are nonneuronal and nonastrocytic playing a supportive role in brain homeostasis. Although microglia share many key features with other tissue macrophages, they have a distinct origin in that they derive from embryonic precursors that have been recruited to the brain prior to birth (Ransohoff and Perry, 2009). Using in vivo cell fate mapping studies, recent data suggest that primitive macrophages are the main precursors of adult microglia that populate the adult brain (Ginhoux and Merad, 2010). In addition and in contrast to most adult tissue macrophages, some studies have reported that microglia are maintained throughout life independently of any blood input and can resist high dose γ-ray irradiation (Ajami et al., 2007; Mildner et al., 2007), although arguments against this hypothesis have also been presented (Soulet and Rivest, 2008). Similar results were found for epidermal LCs (Merad et al., 2002). Importantly, microglia and LCs can be repopulated by circulating monocytes upon brain or skin injuries (Ginhoux et al., 2006; Mildner et al., 2007). Whether embryonic-derived cells that populate the normal skin and brain have a similar function to that of monocyte-derived cells that arise in injured tissues remains to be examined.

Altogether, these results emphasize yet another limitation of the MPS definition that suggests that two prominent populations of phagocytes that reside in the brain and skin are in fact derived from embryonic myeloid precursors and not from bone marrow-derived monocytes.

Distinguishing DCs from Macrophages In Vivo

Most cells types in vivo are identified on the basis of their ontogeny, phenotype, and function. In contrast, the definition of tissue DCs and macrophages remains mostly phenotypic and the use of the cell surface markers CD11c and F4/80 as surrogate for origin and function have lead to confusion on the exact contribution of DCs versus macrophages to tissue immunity. Indeed, the CD11c integrin can be expressed on macrophages and monocytes, and therefore the role of CD11c⁺ cells does not always reflect the role of DCs in vivo. Similarly, F4/80 is also expressed by eosinophils in the gut and neutrophils in the bone marrow and is not a specific marker of macrophages. Whereas differentiating macrophages from Batf3-IRF8-Id2-independent MHCII⁺ CD11c⁺ cells continues to be difficult probably because these cells as currently defined are heterogeneous and include tissue macrophages as discussed above, Batf3-IRF8-Id2 DCs are easily distinguished from macrophages as described above and summarized below.

1. Origin: Batf3-IRF8-Id2 DCs arise from CDP in an Flt3L-dependent and Csf-1R ligand-independent manner, whereas macrophages arise independently of CDP and require Csf-1R ligand but not Flt3L for their development.

2. Phenotype: Batf3-IRF8-Id2 DCs lack Csf-1R, CD11b, CX3CR1, F4/80, and CD172a, whereas macrophages always express some or all of these markers.

3. Function: Batf3-IRF8-Id2 DCs are much more potent at initiating T cell differentiation compared to macrophages. Batf3-IRF8-Id2 DCs have a much lower degradative potential compared to macrophages and excel in antigen cross-presentation to CD8⁺ T cells, whereas macrophages can only cross-present very large amount of antigens.

4. Migratory Potential : Batf3-IRF8-Id2 DCs are much more potent in migrating to the draining LN compared to macrophages.

In sum, prior confusion on DCs and macrophages contribution to tissue immunity is partly due to the fact that lineage marker ⁻ MHCII⁺ CD11c⁺ cells are not synonymous of DCs and an effort to better define the heterogeneity of Batf3-IRF8-Id2-independent MHCII⁺ CD11c⁺ cells is critically needed to help clarify the role of these cells in the modulation of tissue immune responses.

Focusing on DCs and Macrophage Heterogeneity Rather than Similarities

The term mononuclear phagocyte system was critical to alert the scientific community of the importance of phagocytes in innate immune defense. Stressing the commonality of a cellular system widely distributed throughout the body to promote host defense against foreign invasion facilitated the acceptance of a new cellular lineage by the scientific community. More than a century after their discovery, the identification of human diseases due to compromised phagocyte function together with the development of mouse models with engineered phagocyte-specific genetic defects established the key role of phagocytes in tissue immunity and homeostasis. However, these studies have also revealed the complexity and heterogeneity of this cellular system. In fact, while Aschoff stressed the importance of a certain fundamental resemblance between the cells with respect to phagocytosis and storage, he insisted that similarities do not amount to identity and expected that future research would make it possible to distinguish more accurately between different populations of tissue phagocytes.

There have been controversies on whether cross-presentation function is unique to DCs, given that macrophages can also cross-present extracellular antigens to CD8⁺ T lymphocytes if incubated with large amount of soluble antigens. Subsequent studies using genetically engineered mouse models revealed that absence of Batf3-IRF8-Id2-dependent DCs abolishes the cross-presentation of cell-associated antigens despite the presence of normal tissue macrophages (Hildner et al., 2008). These results are consistent with in vitro studies showing that macrophages are unable to cross-present limiting amounts of soluble antigens. These data also suggest that in considering differences among tissue phagocytes, it is critical not only to assess the capacity of the cells to perform a function but also the efficiency with which this function is performed. Focusing on differences in cross-presentation among DCs and macrophages has helped investigators identify pathways that promote DC superior ability to cross-present tissue antigens (Savina et al., 2006; Savina et al., 2009). These studies have also helped the development of targeting strategies to promote T cell immunity in vivo (Bonifaz et al., 2002; Bonifaz et al., 2004; Ueno et al., 2011). There are also controversies on whether tissue cell migration to the draining lymph node is unique to DCs because macrophages also can migrate to tissue draining lymph nodes in response to tissue injury, although this occurs at low frequency, and thus its functional relevance is not known. DCs, in contrast to macrophages, can migrate to the draining lymph nodes in the absence of tissue injury, and steady state migration to the lymph nodes and presentation of self-antigens provides a mechanism to promote peripheral tolerance (Hemmi et al., 2001; Huang et al., 2000; Mellman and Clausen, 2010; Ohl et al., 2004; Pugh et al., 1983) that could be exploited for the treatment of autoimmune diseases (Steinman et al., 2003). DCs’ ability to migrate to the draining LN in response to tissue injury augments many folds, is controlled in part by CCR7 ligands, and is required to induce adaptive immune response to peripheral tissue antigens (Ohl et al., 2004; Randolph et al., 2008).

What Next in MPS Research?

Although we have highlighted the importance of acknowledging phagocyte heterogeneity in vivo, it is critical to distinguish between phenotypic plasticity and discrete mononuclear phagocyte subsets. The definition of a mononuclear phagocyte subset should only be based on developmental and functional evidence and not on a set of phenotypical differences. System-wide investigation of gene expression at the mRNA transcript level should help revisit phagocyte subset classification and identify gene regulatory networks that control MPS development, differentiation, and function. Quantitative proteomics also holds great promise to enhance or complement the picture of gene expression in cells, and thus to contribute to the understanding of tissue phagocytes’ molecular program. The paucity of available tissue phagocytes especially in nonlymphoid tissues remains, however, a great limitation.

System biology studies should also help the development of genetic tools allowing inducible gene regulation in vivo and lineage tracing of genetically marked, defined myeloid precursor population to further comprehend the developmental complexity of the phagocyte system. RNA interference (RNAi) is the most rapid means to identify gene function in human and mouse cells. At present, high-throughput screening is only feasible in cell culture, which limits the approach to in vitro derived macrophage and DC with the limitation discussed above as primary tissue phagocytes cannot be easily manipulated in culture. However, as we learn more about the developmental requirements of macrophages and DCs, its hould be possible to generate better in vitro and in vivo models for performing gain- and loss-of-function studies. These studies will then need to be validated in the human setting using similar genetics approaches. Unraveling the transcriptional program that controls macrophage and DC functional heterogeneity will undoubtedly opens new strategies for taking MPS biology into medicine (Steinman and Banchereau, 2007).