Initially, DCs were noted to be bone marrow derived and to share a progenitor with macrophages and granulocytes in colony forming assays. Granulocyte-macrophage colony stimulating factor (GM-CSF) was recognized as a key stimulating cytokine. The DCs were regarded to be a separate cell lineage, because their properties were so distinct from phagocytes, in particular the potency with which DCs stimulated T cells and their capacity to initiate immune responses in culture and in mice. This initial view of DC development, i.e., as a parallel track to the development of macrophages and granulocytes, has proven to be oversimplified. The life history of DCs is replete with distinctive and sometimes perplexing features, so much so that this section of the introduction will be the longest.
Subsets. Another unusual feature of DCs is that there are several subsets, as assessed by surface markers and functions (reviewed in (2)). The existence of subsets was first emphasized in mice, where two major groups of CD8+ CD4− and CD8− CD4+ DCs were distinguished (3). Subsets of DCs have also been identified in the rat, marked as CD4− OX41− and CD4+ OX41+ (4, 5). Perhaps these rat subsets correspond to the CD8+ and CD8− mouse DC subsets (CD8 is not found on rat or human DCs). A major functional difference between these subsets is that the mouse CD8+ and rat CD4− DCs selectively take up certain types of dying cells (6–9).
In both mice and humans, there are additional DCs called “plasmacytoid” which are distinguished from other DCs, often termed “myeloid”. This term leaves something to be desired, because it implies that plasmacytoid DCs are nonmyeloid. This is not established, and a myeloid origin is actually suggested by the expression in plasmacytoid DCs of c-fms, the receptor for macrophage colony stimulating factor (10). I suspect the term “plasmacytoid DC” will remain useful, since it portrays their resemblance to plasma cells, which led to their original definition; other DC subsets may need to be denoted with more specific terms such as “CD8+ mouse DCs”.
Importantly, DC subsets are now proving to have major “innate” differences involving cytokine production, receptors for antigen uptake, and receptors for microbial ligands (Toll-like receptors), cytokines and chemokines. The December 2002 volume of Human Immunology has an entire issue devoted to plasmacytoid DCs. These cells, for example, are able to make unusually high amounts of interferon-α in response to viral infection (11, 12), to express a distinct marker called BDCA-2 that is capable of endocytosis and antigen presentation (13) (although note should be taken that plasmacytoid DCs are very weak at capturing soluble and particulate antigens), to respond to IL-3 rather than GM-CSF (14), and to express Toll-like receptor 9 that responds to bacterial DNA and demethylated CpG deoxyoligonucleotides (15, 16). Other DCs can make large amounts of IL-12 in response to bacterial cell wall components (17–19), express a number of endocytic receptors that are not found on plasmacytoid DCs such as Langerin and an asialoglycoprotein receptor (20–22), respond to GM-CSF rather than IL-3 (23), and express Toll-like receptors 2, 3 and 4 (24).
There are additional DC subsets that have recently been identified. One is found in human blood and is marked by a 6-sulfo LacNAc carbohydrate modification of the selectin ligand PSGL-1; these DCs are capable of producing large amounts of TNF (25). In the cornea, there are abundant immature DCs that lack detectable MHC class II molecules; however, following transplantation, these DCs undergo typical differentiation to express characteristically high levels of MHC class II (26). There also are CD11c+ DCs that develop during infection and move from blood into infected lymphoid and nonlymphoid tissues (27–29).
It remains unclear whether DC subsets have distinct patterns of migration in vivo. Some of the current findings are as follows. Plasmacytoid DCs have a long life span relative to other subsets of DCs (30, 31). Perhaps this long life span is associated with a continuous recirculation from blood through lymphoid organs via high endothelial venules, for which plasmacytoid DCs express the CD62L selectin homing molecule. Many DCs can express CLA-4 (25). This is an epitope on PSGL-1, a ligand for CD62E and CD62P, i.e., the E and P selectins expressed by vessels in peripheral tissues. Some DCs rapidly extravasate into tissues in the steady state, but this occurs by a PSGL-1-independent mechanism (32, 33). Plasmacytoid DCs also can be found in an extravascular location in conditions such as nasal allergy and cutaneous lesions of lupus erythematosus (34, 35).
Another subset distinction originates from work on DCs within the skin, but it may apply to DCs from other organs. Epidermal DCs (Langerhans cells) express high levels of markers, such as CD1a, Langerin and E-cadherin; these are not found on most DCs in the dermis and interstitial spaces of other organs (dermal or interstitial DCs). Reciprocally, dermal DCs can express DC-SIGN/CD209, the macrophage mannose receptor/CD206, and CD13. One interesting functional distinction between epidermal and dermal DCs is that the latter stimulate B cells in certain assay systems (36).
It was originally thought that Langerhans cells are immediately derived from monocytes. However, their numbers in the steady state are maintained by local proliferation in the skin; only during inflammation is there recruitment of Langerhans cells from CCR2-bearing precursors from the blood and bone marrow (37), possibly from DCs expressing CD1a (38). TGFβ has a major effect on LC development (39–43), and this cytokine may additionally induce the CCR6 chemokine (for CCL 19 or MIP-3α) receptor (44) for homing to a variety of epithelia.
Much of the information on DC subsets at this time is derived from cells isolated ex vivo. It will be valuable to learn to manipulate these subsets in vivo, to determine their consequences for tolerance and immunity. One way that may help in this regard is to learn to selectively deliver an antigen to a DC subset, as can now be done with dying cells that selectively target the CD8+ subset of mouse DCs, for example (8, 9). The plethora of DC subsets identified with current markers can seem confusing, but the data are suggesting that the different subsets at a minimum are specialized for recognizing distinct pathogens and carrying out distinct innate functions. As learned from the history of lymphocyte subsets, the initial markers that define a subset eventually allow for experiments that yield clearer criteria applicable to intact animals and patients.
Migration. The migration and positioning of DCs are among the hallmarks of this lineage. Different chemokine receptors are valuable at different stages of the life history of DCs. For example, CCR2 seems important for DCs to translocate into the T cell rich regions of lymphoid tissues during contact allergy and infection with L. major (54); CCR5 may help recruit DCs to inflammatory sites (55, 56); CCR6 appears to be important for positioning DCs at epithelial surfaces (57–59); CCR7 may increase entry to lymphatics and migration to the T cell areas of lymph nodes (60). Other mediators influence DC migration perhaps by controlling the function of chemokine receptors, such as lipids like lipoxins, leukotrienes, prostaglandins (61, 62).
There may be some misunderstanding that DCs only migrate into lymphatics during inflammation and infection. However, DCs seem continuously on patrol through peripheral organs, lymph and lymphoid tissues in the steady state. This steady state migration provides DCs an opportunity to sample self antigens and environmental proteins continuously for purposes of immune tolerance (6, 63). There is new evidence that the uptake of apoptotic cells can increase expression of CCR7 (a chemokine receptor that mediates homing to the T cell area), without driving other features of DC maturation (below) (64). Also, the accumulation of CCR7+ but otherwise immature DCs occurs in certain clinical states associated with increased numbers of Langerhans cells in lymph nodes (65). One possibility is that DCs (or a subset of DCs) entering the lymph in the steady state are cells that have taken up dying cells and started to express CCR7.
Transcriptional controls. The analysis of DC development requires more data on transcriptional controls. Several are under study, with the NF-κB/rel family providing intriguing guidelines already. There are several NF-κB proteins, which in turn can form a number of heterodimers. These different rel family members contribute to different stages of DC development. Mice lacking rel B lack the CD8− subset of spleen DCs (66), while mice lacking both rel A (p65) and p50 lack most DCs, probably because of diminished survival (67). Mice lacking cRel and p50 fail to respond to TRANCE and CD40L to control survival and cytokine (IL-12 p40) production by maturing DCs (67).
Interestingly, DCs express high levels of all NF-κB proteins, not just activities (68). This may help to explain how DCs react so quickly and vigorously to many stimuli that signal through NF-κB, for example Toll-like and TNF-family receptors.
DCs are nonproliferating cells that are quickly responsive to different environmental stimuli. Because some types of DCs are available in large numbers e.g., monocyte-derived DCs, they may be ideal to study the regulation and recruitment of stimulus-dependent transcriptional factors to chromatin, e.g., the p38 MAP kinase-dependent recruitment of NF-κB (69). The new methodologies for studying access and binding of transcriptional factors to their promoters promise to accelerate this field of research and to identify distinct signal transduction pathways that control sets of genes during DC differentiation.
Maturation. A major feature of the life history of DCs is termed maturation. The events that take place before and during DC maturation are important in understanding the control of immunity and tolerance. When ex vivo-derived, antigen-pulsed DCs are used to immunize mice (72, 73) or humans (74, 75), the mature DCs are immunogenic, whereas immature DCs can induce regulatory cells (76). Formal proof that DC maturation in vivo (rather than ex vivo) controls the induction of CD4+ and CD8+ T cell immunity was only recently obtained. The experiments used α-galactosylceramide, which acts via NKT cells to mature DCs (77). These DCs, following antigen capture and maturation in vivo, were able to induce immunity upon adoptive transfer to naive animals, without an additional antigen or maturation stimulus.
Immature DCs are adept at capturing antigens, especially through receptor-mediated uptake (below). Also, immature DCs respond quickly and vigorously to many microbial and inflammatory (type I and II interferons, GM-CSF, TNFα) stimuli via Toll-like and cytokine receptors. Nevertheless, the single term “immature” is used to describe DCs in different circumstances, e.g., the Langerhans cells of the epidermis, the cells generated from monocytes with GM-CSF and IL-4, and many of the DCs within lymphoid organs in the steady state. These DC populations likely have differences, e.g., in their capacities to form MHC-peptide complexes (see below). However, each of these immature DCs captures antigens and responds rapidly to maturation stimuli to become potent stimulators of immunity.
During maturation, DCs dampen endocytic receptor expression and endocytosis itself (78), and they undergo wholesale changes in the expression of many molecules used to interact with T cells, such as several B7-family members (CD80, CD86, PD-L2/B7-DC, ICOS-L), TNF family members (CD137/4–1BBL, CD134/OX40L, CD70), as well as chemokine receptors (CCR5, CCR7). The term maturation initially was meant to highlight the intricate and multifaceted nature of DC differentiation (79–83). Now it is clear that dozens of genes are altered in their expression in maturing DCs at least ex vivo (84–86). This exciting area now needs to be addressed more fully in vivo.
There are many stimuli for DC maturation. Many microbial ligands and synthetic compounds act on distinct Toll-like receptors to control DC maturation, e.g., viral RNA and poly IC on TLR3 (87), mycobacterial extracts on TLR2 and TLR4 (88), imidazoquinolines on TLR7 (18, 89, 90), and bacterial DNA and CpG deoxyoligonucleotides on TLR 9 (91, 92). Physiologic molecules like β-defensins may also exploit Toll-like receptors in order to mature DCs (93). Other maturation stimuli signal DCs through pathways distinct from the Toll-like receptors. These include Fcγ receptors for immune complexes (94, 95), and PIR-B (96) and TREM-2 (97) whose ligands are not yet known; CD100 (98); and several TNF family members like TNFα itself, fasL (99) and CD40L. CD40 signaling in particular is a powerful inducer of most of the immunogenic functions of DCs (100–102) and also stimulates DC maturation and immunogenicity in vivo (8, 103, 104). Innate cell types like NK cells (105, 106), γδ T cells (107) and NKT cells (77, 108) can mature DCs by pathways that remain to be worked out. An important finding is that certain types of necrotic cells can induce maturation in vitro (109). This may require the function of inflammatory cytokines like TNF (110) or heat shock proteins (111), the latter acting at the level of CD40 (112) and possibly through Toll-like receptors (113, 114). Type I interferons, which can be produced in abundance by plasmacytoid DCs (11, 12), and to some extent by other DCs (25), are emerging as significant inducers of DC maturation (115–118). IFNα induces DC maturation in two longstanding settings of immune enhancement, i.e., the spontaneous maturation of certain types of DCs in culture (119), and the adjuvant action of complete Freund's adjuvant (120). Actually, the above lengthy summary of inducers of DC maturation does not represent a complete summary of the literature.
It seems unlikely that functionally identical DCs will develop in response to the breadth of distinct maturation stimuli just outlined above. Nonetheless, each stimulus can provide control of the immune response through the aegis of DC maturation. It may be important to stress that the critical cell biology linking innate and adaptive immunity is the process of DC maturation, rather than signaling via Toll-like or other specific receptors per se. Signaling other cell types via these receptors does not generate cells with the potent and specialized properties of DCs, which additionally are able to control the quality of the adaptive response, as we shall discuss below.
In a reciprocal sense, many agents are being defined that block DC maturation. Most notable are a number of pathogens, which may evade the immune system by blocking maturation (reviewed in (121)), as well as some natural compounds such as 1α, 25-dihydroxyvitamin D3 (122, 123). There also are mechanisms to dampen select components of DC maturation, particularly IL-12 production (124–127). Papers have begun to appear on the pharmacologic blockade of DC maturation (122, 128, 129), a strategy that may reduce chronic inflammatory disease and the initiation of transplant rejection.
There has been a major change in emphasis with the new information that immature DCs are not simply ignored but instead are able to mediate tolerance in the steady state, e.g., by a deletional mechanism (8, 103, 104). The term “mature” in a functional sense has always connoted immunogenicity, e.g., the differentiation and expansion of effector functions and memory. This concept of maturation was one of the first to distinguish antigen handling by immature DCs, which is required for antigen-specific tolerance, from the many other accessory functions used by DCs and other cells to control immune responsiveness (82, 83). Admittedly, the two terms – immature and mature – are insufficient to precisely describe the different kinds of DCs that have been generated in vitro or have been isolated from different tissues in vivo. There is interest in identifying new terms to describe different states of DC function, but I would like to suggest that rather than more descriptive nomenclature, it might be more fruitful to concentrate on defining specific stimulatory pathways and molecular markers for different functional states of DCs.