I. INTRODUCTION

Top Previous Next References

Dendritic cells (DCs) were first described in the mid 1970s by Ralph Steinman, who observed in the spleen a subpopulation of cells with a striking dendritic shape. These cells were nonphagocytic, loosely adherent, and of low buoyant density (325-327). It was soon appreciated that these bone marrow-derived cells existed in all lymphoid and most nonlymphoid tissues. DCs were described as cells that constitutively expressed both major histocompatibility complex (MHC) class I and class II antigens, spontaneously clustered T cells via antigen-independent mechanisms (later understood to represent the interplay of surface molecules on DCs that were mutually complementary to surface molecules on T cells), and, most importantly, stimulated naive CD4 and CD8 T cells to respond to nominal and alloantigens more effectively than any other previously described antigen presenting cell (APC).

In recent years, DCs have been increasingly studied for their role as critical adjuvants in vaccines for prevention of microbial infection and allograft rejection and treatment of cancer and autoimmune diseases. Several reviews on DCs and their role in immune regulation have appeared recently, because of the increased realization of their importance in immunoregulation and possibilities for exploiting them for biomedical purposes (Refs. 21, 22, 134, 188 are representative). This review overviews DC biology, highlighting more recent literature. DC origins and differentiation pathways are discussed, including factors that regulate their migration to sites where they play their surveillance role. How DCs link innate and adaptive immunity will be reviewed, with a separate section on how certain pathogens, to survive in the infected host, subvert the immunostimulating activity of DCs. The role DCs play in autoimmune and allergic diseases, transplantation, and cancer is described. In relevant sections, representative studies that have manipulated DCs for therapeutic purposes are summarized.

	II. DENDRITIC CELL ORIGINS

Top Previous Next References

DCs are a heterogeneous group of cells that display differences in anatomic localization, cell surface phenotype, and function. However, DCs have several features in common (22, 134). First, originating from CD34 bone marrow stem cells, precursor DCs are seeded via the bloodstream to the tissues where they give rise to immature DCs that include Langerhans cells (LCs) and interstitial DCs (also called dermal DCs). Second, immature DCs have the ability to take up antigen, via both receptor- and non-receptor-mediated mechanisms, and readily degrade antigens in endocytic vesicles to produce antigenic peptides capable of binding to MHC class II. Third, in response to danger signals, i.e., tissue damage, pathogen-derived products, or inflammatory cytokines, DCs mature and migrate to lymphoid organs where they interact with antigen-specific CD4 T cells to initiate immune responses (83, 169, 183, 205, 257, 362, 366). Fourth, distinct chemokine receptors occur on immature DCs, compared with mature DCs, which regulate their traffic into tissue sites in response to inflammatory chemokines (71, 143, 283, 396). Fifth, as DCs mature, they express a high density of MHC class II molecules complexed with antigen for recognition by the T cell receptor (TCR) expressed on CD4 T cells and costimulatory molecules to stimulate CD4 T cell proliferation. Finally, other factors in the microenvironment at the time of DC maturation have been shown to dictate whether DCs will produce IL-12 and initiate Th1 responses or have their IL-12-producing capacity suppressed and initiate Th2 responses (366).

DCs generally have a low buoyant density and are initially adherent to plastic but then readily detach (227, 329). Early methods of DC isolation used enzyme-digested tissue and exploited these attributes to obtain enriched populations of DCs from murine tissue. Selection of low-buoyant-density cells enriches for mononuclear cells, and the adherence step helps eliminate T cells and B cells from the preparation. Further enrichment of DCs utilizes various combinations of additional negative and positive selection steps. One negative selection step is based on phagocytosis of silica particles, latex beads, or carbonyl iron particles to remove avidly phagocytic macrophages from preparations. Other negative selection steps eliminate contaminating B cells, T cells, and NK cells from the preparation by a combination of immunophenotyping and cell sorting. Positive selection steps use various combinations of monoclonal antibodies to isolate cells expressing important DC cell surface markers, typically MHC class II and CD11c. With the use of multiple enrichment steps, pure populations of tissue-derived DCs have been obtained.

DCs can also be propagated from bone marrow and blood using various combinations of growth factors, such as granulocyte macrophage-colony stimulating factor (GM-CSF), tumor necrosis factor- (TNF-), interleukin (IL)-4, stem cell factor (SCF), transforming growth factor- (TGF-), IL-3, and Flt3 ligand (Flt3L) (47, 151, 221, 264, 279, 283, 332, 392). GM-CSF in combination with IL-4 or TNF- and other cytokines provides important growth factors for interstitial DCs and LCs. In addition, LCs also require TGF- for their differentiation (35, 332). IL-3 is a cytokine required by plasmacytoid DCs, a noninterstitial, non-LC DC subtype, that express CD123 (IL-3 R) and are found in lymphoid tissue (127). Flt3L has been used to stimulate the proliferation of stem cells and progenitor cells in vitro and expand and mobilize all DCs and their progenitors in vivo (221, 222, 262, 264).

Murine DCs have been classified into two main lineages: myeloid DCs as originally described by Steinman and Cohn (327) and lymphoid DCs described by Suss and Shortman (339). However, researchers now recognize that enrichment steps led to selective loss of DC subpopulations and that murine DCs are not readily separated into these two distinct lineages. A current method to obtain DCs that avoids depletion of DC subpopulations involves mild collagenase digestion; breaking DC-T cell complexes with EDTA; selecting low-density cells; depleting T cells (CD3+ or Thy1+), B cells (B220+), granulocytes (Gr1+), and erythroid lineage cells (TER-119+) by immunomagnetic bead depletion; and finally positive sorting for cells expressing CD11c and MHC class II (226, 368). DCs can be further segregated into subtypes based on expression of CD4 and the CD8 chain homodimer (CD8), markers originally thought to be confined mainly to T cells (226, 368). Two important problems in subtyping DCs based on expression of CD4 and CD8 are autofluorescence and adsorption of CD4, CD8, and Thy-1 surface antigens from other cells (368). Other difficulties that further confuse the subtyping of DCs are the different stages of maturation that DCs exhibit in situ, i.e., bone marrow progenitors, precursor DCs in blood and lymphatics, immature DCs in tissue, and mature DCs within secondary lymphoid organs. Thus, in evaluating the literature on DC subtypes, the procedure used for isolation, the controls used to minimize immunofluorescent staining artifacts, the authenticity of surface CD4 and CD8 markers, and the developmental state of the DC must all be considered.

Over the past few years, numerous reports detailing different isolation procedures and dealing with the phenotype, localization, and function of murine CD8+ and CD8 DCs have been published, contributing to our understanding of DC biology (125, 170, 178, 179, 195, 219, 251, 263, 264, 269, 280, 305, 339, 368, 378, 381, 382). At least five major populations of DCs have been described in the central and peripheral lymphoid organs of mice (see Table 1). In murine spleen, three DCs subtypes are delineated, namely, CD4-8+DEC205+CD11b, CD4+8DEC205CD11b+, and CD4-8DEC205CD11b+ (170, 305, 368, 381). In lymph nodes, these three subtypes are present together with a fourth population, CD4-8(lo)DEC205+ with various levels of CD11b (11, 305). The mouse thymus appears to contain two DC types, one that overlaps with a lymph node subtype and one that may be unique, CD4CD8/loDEC205+CD11b and CD4CD8DEC205+CD11b, respectively. Based on phenotype and maturation kinetics, CD4-bearing DC depletion studies, bromodeoxyuridine (BrdU) labeling kinetics, and bone marrow reconstitution studies, the three spleen DC subtypes appear to be products of three independent developmental streams, not different states of maturation. All three subtypes were classed as mature, because they expressed CD80, CD86, and CD40 and efficiently activated allogeneic T cells (368). However, further maturation was induced in all these subtypes by bacterial stimuli (170). Contradictory to conventional views about DC maturation was that these three DC subtypes phagocytosed particulate material in vivo and upon maturation retained phagocytic capacity. Upon maturation, no DC subtype converted to the other, and continuous elimination of CD4-bearing DCs by antibody depletion had no effect on numbers of the other two DC subtypes. BrdU labeling experiments indicated all three DCs subtypes had a rapid turnover in the spleen, with the CD4CD8+ DCs showing the fastest turnover and with none being the precursor of the other. Immunofluorescent staining of spleen sections showed that the two CD8 populations, i.e., CD4+8 and CD4-8 DCs, were in the marginal zones of the spleen, with only CD4-8+ DCs concentrated in T cell areas. However, in response to microbial stimuli such as lipoplysaccharide (LPS), CD8 DCs rapidly migrated to T cell areas (82). Not delineated in this microbial stimuli study were the migratory responses of CD4+ and CD4 subsets of the CD8 DC population. Functionally CD8+ and CD8 appeared distinct with CD8+ DCs producing much higher levels of IL-12 than the CD8 DCs in vitro (179, 218, 265). Whether this distinction holds true in vivo when the two CD8+ and the one CD8 subsets are evaluated as three subsets, i.e., CD4+8, CD4-8+, and CD4-8, must be determined.

The question remains whether the subsets of DCs have a common progenitor. Because CD8+ DCs lack the myeloid marker CD11b, they were originally thought to arise from a lymphoid-committed progenitor and were generated at low frequencies from thymic T cell progenitors (13). In contrast, CD8 DCs, which may be either CD4+ and CD4, generally express CD11b, were considered myeloid related, and could be derived from myeloid progenitors (150). Additional arguments were made that the CD8 marker on DCs reflected an origin from a precursor different from the CD8 populations, because CD8+ DCs have distinct cytokine requirements for their in vitro generation and utilize different transcription factors (293, 380). For example, thymic CD8 DCs precursors require GM-CSF for differentiation in culture. In contrast, CD8+ DCs precursors require IL-3, but not GM-CSF, to differentiate (293). The absence of CD8 and the presence of CD8+ DCs in RelB and PU.1 knockout mice suggest relB and PU.1 play a role in the development of CD8 DCs (129, 380). A recent study assessed the ontogeny of CD8+ and CD8 DCs. Traver et al. (357) showed by transfer of marked, lineage-restricted progenitors that both CD8+ and CD8 DCs arise from common myeloid and lymphoid progenitors in both murine thymus and spleen. In addition, RelB and PU.1 were expressed in both CD8+ and CD8 DCs. Clearly, DCs can be derived from either myeloid or lymphoid precursors. However, the study by Traver et al. (357) indicates that CD8 on DCs does not indicate a lymphoid origin but rather may reflect the maturation or differentiation status and once defined may predict the function of the DC.

In humans, DCs are also found as precursor populations in bone marrow and blood and as more mature forms in lymphoid and nonlymphoid tissues. Three distinct subtypes of human DCs have been delineated based on studies of skin DCs (57), DCs generated in vitro from CD34+ hematopoietic progenitors (51), and blood DC precursors (see Fig. 1) (279). Human skin contains two of the three DC subtypes in immature form: LCs and interstitial DCs. Both subtypes emerge in cultures from CD34+ bone marrow and CD11c+ blood precursors in the presence of GM-CSF and either IL-4 or TNF- (48, 279, 283). The CD11c+ DC precursor expresses myeloid markers, including CD13 and CD33. Upon activation by CD40L, immature myeloid DCs undergo maturation and produce IL-12 (56). A distinction from interstitial DCs is that LCs also require TGF- (154) and arise from either a CD11c+CD14+ monocyte or a CD11c+CD14 precursor, whereas interstitial DCs arise from a CD11c+CD14+ precursor that can also differentiate into macrophages in the presence of only M-CSF (48, 279, 283, 393). The demonstration that LCs can arise from CD11c+CD14+ monocyte is controversial and may ultimately reflect the plasticity of DCs. LCs and interstitial DCs subtypes share several markers, but LCs uniquely express CD1a, Birbeck granules, langerin, and the adhesion molecule E-cadherin. In contrast, interstitial DCs uniquely express the coagulation factor XIIIa. LCs and interstitial DCs also share the capacity to activate both CD4 and CD8 naive T cells and secrete IL-12. One striking difference between LCs and interstitial DCs is the ability of interstitial DCs, but not LCs, to induce the differentiation of naive B cells into immunoglobulin-secreting plasma cells (49, 92).

View larger version (36K): [in this window] [in a new window]   Fig. 1. Schema for derivation of human dendritic cell (DC) subsets from CD34+ myeloid and lymphoid progenitors. In humans, DCs are found as precursor populations in bone marrow and blood and as more mature forms in lymphoid and nonlymphoid tissues (top). Both myeloid and lymphoid lineage DCs can be propagated from bone marrow progenitors and blood precursors using various combinations of growth factors, such as granulocyte macrophage-colony stimulating factor (GM-CSF), tumor necrosis factor- (TNF-), interleukin (IL)-4, transforming growth factor- (TGF-), and IL-3. Interstitial DCs and Langerhans DCs subsets (middle) are found at sites that interface with the external environment. Plasmacytoid DCs are found in the T cell zones of lymphoid organs and in the thymus and blood. Shown below each DC subset are phenotype and functional characteristics defining that DC subset. See text for additional details. MHC, major histocompatibility complex; IFN, interferon.

Plasmacytoid DCs are a third type of DC and are so named because at the ultrastructural level they resemble Ig-secreting plasma cells. These DCs are found in the T cell zones of lymphoid organs and in the thymus and blood and were previously described as plasmacytoid T cells or plasmacytoid monocytes (98, 127, 249, 272, 314, 333). Plasmacytoid DCs are characterized by a unique phenotype, CD11c-CD4+CD123+CD45RA+HLA-DR+, and possess the unique ability to secrete large amounts of interferon (IFN)-/B upon viral stimulation (54, 127, 174, 307). In this context, precursor plasmacytoid DCs in blood correspond to natural IFN--producing cells, suggesting an important role during viral infections (33, 292). Precursor plasmacytoid DCs in blood express CD62L and the chemokine receptor CXCR3, which mediate homing and migration of these cells into the lymph node via high endothelial venules (HEVs) in response to inflammatory chemokines (54). Unlike LCs and interstitial DCs, plasmacytoid DCs require IL-3 for their differentiation and are derived from a CD11c- blood precursor that has low expression of GM-CSF receptor, lacks the myeloid markers CD14, CD13, and CD33, lacks mannose receptors, and expresses high amounts of CD123 (127, 174). Plasmacytoid DCs share a common function with LCs and interstitial DCs in having the capacity to activate CD4 and CD8 naive T cells and secrete IL-12 upon CD40L activation (53, 174). Several lines of evidence suggest that plasmacytoid DCs originate from a lymphoid precursor. First, precursor plasmacytoid DCs lack expression of myeloid antigens (127). Second, precursor plasmacytoid DCs express pre-TCR- transcripts (42, 272). Third, ectopic expression of inhibitor of DNA binding (Id)2 and Id3 inhibits the development of CD34+ progenitor cells into CD123+ precursor plasmacytoid DCs, T and B cells, but not myeloid DCs (320). Finally, precursor plasmacytoid DCs express the immunoglobulin-like transcript receptor (ILT)3, in contrast to myeloid DCs that express both ILT3 and ILT1.

Murine DCs have been widely employed by researchers investigating the roles of DCs in the generation and regulation of specific immunity. Although it is clear that differences exist between murine and human DC, it is evident that murine DCs are relevant to human DCs and provide an appropriate model for human cells in most cases. Like human DCs, murine DCs 1) originate from CD34+ bone marrow stem cells, 2) are found in blood and tissues, 3) are able to take up and degrade antigen to antigenic peptides, 4) express MHC class II molecules complexed with antigenic peptide, 5) express costimulatory molecules, 6) mature and migrate in response to danger signals, and 7) are responsive to the microenvironment with a controlled release of chemokines and cytokines. Data from murine DC studies should be interpreted with caution in cases where clear discrepancies exist between murine and human subtype counterparts.

	III. DIFFERENTIATION AND TRAFFICKING PATHWAYS

Top Previous Next References

DCs are migratory cells that traffic from one site to the next, performing specific functions at each site (273, 284, 287, 396). Bone marrow-derived DCs circulate as precursors in blood before entering tissue where they become resident immature DCs that monitor their environment (Fig. 2). Interstitial DCs and LCs are found at sites that interface with the external environment, i.e., mucosal surfaces and in the skin. In peripherial tissues, immature DCs have the ability to migrate toward inflammatory foci where they take up and process available antigens and then emigrate through the lymphatics to draining lymph nodes. There they home to T cell-rich areas and interact with T cells to initiate an immune response. It has been proposed that the origin of lymphatic-borne DCs may be blood monocytes (242, 268). Using an in vitro model of transendothelial trafficking, Randolph et al. (268) observed that monocytes matured into DCs as they migrated across endothelium from the abluminal to the luminal surface in a manner mimicking entry into the afferent lymphatics. Monocytes that remained in the subendothelial matrix became macrophages and lost migratory capability. In contrast, plasmacytoid DCs are thought to migrate directly from the blood to lymphoid tissue (22, 54).

View larger version (27K): [in this window] [in a new window]   Fig. 2. The migratory and maturation pathways of DCs. DCs are migratory cells that traffic from one site to the next, acquiring specific abilities at each site and performing specific functions in a stepwise fashion. See text for additional details. pDC, precursor DC; iDC, immature DC; mDC, mature DC.

The extravasation of DCs from the blood to peripheral tissue and the movement from peripherial tissue into lymphoid tissue requires chemoattractants called chemokines. Chemokines are peptide activators of G protein-coupled receptors expressed on leukocytes that regulate recruitment of inflammatory cells (182, 235, 278). Chemokines are differentially produced at peripheral tissue sites by endothelial cells, epithelial cells, and leukocytes in response to diverse inflammatory stimuli. Chemokines are also constitutively produced by endothelial cells or stromal cells and leukocytes within secondary lymphoid organs to regulate encounters between DC, T cells, and B cells (20, 70). The anatomic location of inflammatory chemokines within peripheral tissue and constitutive chemokines within lymphoid tissue regulates the migration of DCs initially to sites of antigen and ultimately to lymphoid tissue to initiate an immune response.

The ability of DCs to respond to inflammatory and lymphoid chemokine gradients is presumably linked to their maturation state, because as DCs mature they lose responsiveness to inflammatory chemokines and gain responsiveness to lymphoid chemokines. Both human monocytes and monocyte-derived immature DCs and murine CD34+-derived immature DCs express both CC and CXC chemokine receptors (CCR and CXCR), such as CCR1, CCR2, CCR5, and CXCR1, and respond to inflammatory chemokines such as macrophage inflammatory protein-1 (MIP-1), monocyte chemotactic protein-1 (MCP-1), and regulated on activation normal T cell expressed and secreted (RANTES) chemokine (87, 289, 315, 318, 319, 364). In addition, monocytes and immature DCs express chemoattractant receptors for cleavage products of bacterial proteins such as formyl-methionyl-leucyl-phenylalanine (fMLP), products of host complement activation, such as C5a, and lipid metabolites, such as platelet activating factor (PAF) (9, 239, 317, 319, 385). As immature DCs migrate toward increasing concentrations of inflammatory chemokines, they are also exposed to increasing concentrations of proinflammatory cytokines, such as TNF- and IL-1 and the pathogen products initiating the inflammatory response. In response to these danger signals, DCs mature, and in doing so switch the usage and expression of chemokine receptors from inflammatory to lymphoid homing receptors. The loss of inflammatory chemokine receptors is at least partly regulated by ligand-induced downregulation by autocrine secretion of MIP-1, MIP-1, and RANTES by maturing DCs (288). Maturing DCs downregulate the expression of CCR1, CCR5, and CXCR1 and upregulate the expression of CXCR4, CCR4, and, in particular, CCR7, a chemokine receptor that responds to secondary lymphoid tissue chemokine (SLC) and Epstein-Barr virus-induced ligand chemokine (ELC). SLC is produced by lymphatic endothelial cells, and both SLC and ELC are produced by stromal cells and DCs in the T cell areas of lymphoid organs (85, 87, 108, 130, 198, 247, 315, 364, 388). The anatomic distribution of SLC and ELC secretion coordinately attracts DCs first from peripheral tissue to afferent lymphatics and then to T cell areas in lymphoid tissue. CCR7 is also selectively expressed on naive T and B lymphocytes, allowing these cell types to also home to lymphoid tissue (45). CCR7 is a unique chemokine receptor, because it is resistant to ligand-induced downregulation (288). Sustained expression of CCR7 may allow DCs to perform their stepwise migration from tissue to afferent lymphatics to the lymphoid organ. The essential role of CCR7 in DC homing to lymphoid organs is supported by the observation that in CCR7-deficient mice, maturing DCs are not able to migrate to lymph nodes (107).

Upon maturation, bacterial and complement receptors on DCs are also differentially regulated. Maturing DCs downregulate their responsiveness to and receptor expression for fMLP, while maintaining their responsiveness to and receptor expression of C5a (385). Yang et al. (385) proposed that the interaction of C5a with C5aR on mature DCs may participate in guiding mature DCs to B cell follicles in lymphoid tissue where naive B cells, a source of C5a, acquire antigens delivered by mature DCs.

LCs are unique in that they express CCR6 in addition to other chemokine receptors expressed by immature DCs (85, 86, 316). CCR6 is a chemokine receptor that responds to MIP-3 produced constitutively by epithelial cells in human liver and lung, induced in the crypts of inflamed tonsils and appendix in humans, and produced in noninflamed follicle-associated epithelium of murine Peyer's patches (63, 86, 155). MIP-3 and CCR6 represent a chemokine/receptor pair that has dual function in recruiting DCs. Not only does this pair recruit immature DCs to mucosal and nonmucosal sites of inflammation, but also recruits immature DCs to become sentinels in noninflamed tissues. As LCs mature, CCR6 is downregulated with a concomitant upregulation of CCR7 and homing to T cell areas of lymphoid tissue.

During the course of an inflammatory reaction, DCs produce inflammatory and lymphoid chemokines in a specific spatial and time-ordered manner (108, 284, 287). Immature DCs constitutively produce MCP-4 that binds to both CCR2 expressed on immature DCs and to CCR3, a potential marker for a subset of Th2 cells (286). In response to maturation stimuli, the production of MCP-4 by DCs is rapidly downregulated, and the inflammatory chemokines, MIP-1, MIP-1, and IL-8 are transiently induced for a few hours. Other inflammatory chemokines such as RANTES, MCP-1, and MCP-2 are also induced, but for a longer period of time. The inflammatory chemokines produced by maturing DCs function in both autocrine and paracrine modes to regulate DC trafficking. In an autocrine mode, they initially stimulate, then downregulate, cognate receptors allowing the DC to respond to other chemoattractants. In a paracrine mode, DCs sustain the inflammatory process by recruiting monocytes, immature DCs, and other inflammatory cells to the site of antigen. At later time points, when mature DCs reach T cell areas of lymphoid tissue, they produce high levels of lymphoid chemokines such as ELC, macrophage-derived chemokine (MDC), thymus and activation-regulated chemokine (TARC), pulmonary and activation-regulated chemokine (PARC), and IFN--inducible protein (IP-10) (247, 288, 346). The production of lymphoid chemokines by mature DCs in lymphoid tissue recruits T cells, augmenting the chances of DC-T cell contact. As previously discussed, ELC binds CCR7 expressed on naive T and B cells and mature DCs. MDC and TARC bind CCR4, a receptor expressed on recently activated T cells, but not on naive T cells (73, 282). PARC binds to an unidentified receptor expressed on naive T cells (3). IP-10 is a chemoattractant that mediates the migration of plasmacytoid DCs directly from blood to the inflamed lymphoid tissue. Plasmacytoid DCs express the IP-10 ligand CXCR3 that can also bind to monokine induced by IFN- (Mig) produced by mature DCs.

Consolidating what we know about the unique migratory patterns of myeloid and plasmacytoid DCs and their patterns of IL-12 production, Patterson (255) has proposed a model delineating the contributions of myeloid and plasmacytoid DCs to the generation and regulation of an immune response to a viral infection. Myeloid precursor DCs leave the blood and home to various tissues in response to chemoattractant gradients. Immature myeloid DCs are located at sites where most pathogens enter the body. Upon exposure to virus, immature myeloid DCs bind and internalize virus and subsequently release inflammatory cytokines and chemokines that initiate the recruitment of more immature myeloid DCs and other leukocytes to the site of infection. Maturing DCs carrying antigen migrate via afferent lymphatics to T cell areas of draining lymph nodes, where they interact with and stimulate pathogen-specific T cells. Myeloid DCs initially release IL-12 and drive the generation of effector cells with a Th1 phenotype. However, within 24 h, production of IL-12 ceases and IL-4-secreting T cells are generated to dampen the Th1 response (53). Coordinately, a switch away from the generation of effectors to unpolarized memory T cells occurs. Myeloid DCs in lymph nodes produce chemokines that recruit T cells, B cells, more myeloid DCs, and plasmacytoid DCs. In response to IP-10 and Mig, plasmacytoid DCs leave the blood and enter inflamed lymph nodes through HEV. The expression of CD62L on plasmacytoid DCs allows them to migrate via HEVs (186). In lymph nodes, plasmacytoid DCs are stimulated via the expression of CD40L on recently activated T cells to mature and secrete IL-12 and maintain production of IFN- (53). The production of IFN- by plasmacytoid DCs prolongs the Th1-stimulating phase of myeloid DCs. The unique migratory properties and controlled temporal release of IL-12 by myeloid and plasmacytoid DCs provides the generation of the right kind of T helper cell at the right time.

	IV. DENDRITIC CELL-T CELL INTERACTIONS

Top Previous Next References

During the development of an adaptive immune response, the phenotype and function of DCs play an important role in initiating tolerance, memory, and polarized Th1 and Th2 differentiation. As discussed, DC subsets have been proposed as playing differing roles in defining the outcome of an immune response, although clearly some plasticity within defined subsets is possible so that each subset can exert tolerizing and polarizing influences on responding T cells (255). Important factors other than signals delivered by DCs that drive primary immune responses are concentration of antigen in the microenvironment, concentration of cytokines and other soluble factors present in the fluid phase in the vicinity of the APC-T cell interface and, of course, the genetics of the host that may limit how the interacting cells may respond. For example, studies in which APCs are replaced by molecular complexes on plastic or lipid substrates have demonstrated that T cells can be polarized by adding cytokines to the culture systems. Still, delivery of the relevant signals by DCs at the DC-T cell interface is likely the most efficient and physiologically relevant mechanism for initiating an immune response.

CD4 and CD8 T cells respond to peptide antigen displayed on MHC class II and MHC class I molecules, respectively (referred to as signal 1, see Fig. 3). Accessory molecules on DCs are required to ensure that T cells will divide and differentiate into effector cells (signal 2, Fig. 3). In the absence of sufficient costimulation, T cells exhibit anergy or undergo apoptosis. Secretion or lack of secretion of factors by DCs, particularly IL-12, are instrumental in the final differentiation of T cells into type 1 or type 2 effector T cells, respectively (signal 3, Fig. 3). The model illustrated in Figure 3 is the simplest one to explain the development of a productive immune response by CD4 T cells, but the list of membrane and secreted molecules that play roles in regulating the interaction of DCs and T cells is growing.

View larger version (32K): [in this window] [in a new window]   Fig. 3. Model for DC-CD4+ T cell interaction. DCs provide three signals to antigen-specific CD4+ T cells to initiate T cell proliferation and differentiation. DCs take up antigen and readily degrade antigens to produce antigenic peptides capable of binding to MHC class II. DCs express a high density of MHC class II/peptide complexes on their cell surface for recognition by the T cell receptor (TCR) expressed on CD4+ T cells (signal 1) and costimulatory molecules (signal 2) to stimulate CD4+ T cell proliferation. Secretion or lack of secretion of interleukin-12 (IL-12) by DCs (signal 3) is important in the final differentiation of CD4+ T cells into type 1 or type 2 effector T cells, respectively. All accessory molecules are not included. See text for additional details.

Initially, DCs were described as both nonphagocytic and nonendocytic. Macrophages are avidly phagocytic, but the antigens taken up are rapidly degraded, unless the material that is endocytosed is inert or is a microorganism with the ability to prevent phagosomal-lysosomal fusion and/or enzymatic degradation. Thus the absence of macrophage phagocytic mechanisms was troublesome for explaining the ability of DCs to take up and present peptides from complex antigens. We now know that immature DCs are avidity endocytic, whereas mature DCs have downregulated this activity (330). DCs degrade antigens within a MHC class II-rich endosomal compartment (MIIC) yet preserve sufficient peptide structure to be expressed on their cell surface bound to MHC class II molecules. DCs take up antigens by phagocytosis, utilizing membrane receptors to trigger uptake, by receptor-mediated pinocytosis in clathrin-coated pits and by fluid-phase pinocytosis. DCs can take up whole cells, including necrotic and apoptotic cells. They can also acquire antigens from live cells for presentation to cytolytic T cells (133). Receptors available to some or all DC subsets for antigen uptake include the FcRs CD32 and CD64; the high- and low-affinity IgE receptors FcRI and FcRII (CD23), respectively; the complement receptors CD11b and CD11c; a C lectin type of mannan binding receptor, DEC205 (CD205), and the scavenger receptor pair for apoptotic cells _v5 and CD36 (reviewed in Refs. 134, 191). During maturation, as endocytosis decreases, these receptors are usually downregulated. More recently, immature human interstitial DCs were shown to express the FcR CD89 (117).

B.  Antigen Processing

1.  MHC class II presentation

Antigen processing by DCs occurs primarily through two major pathways: an exogenous or endosomal pathway and an endogenous or proteosomal pathway (Fig. 4). Exogenous antigens gain access to early and late acidic endosomal compartments in which proteases initiate degradation. The peptide fragments then associate with preformed MHC class II molecules within the MIIC. MHC class II - and -peptide chains are synthesized in the endoplasmic reticulum (ER), where they associate with invariant chain (Ii) (46). Ii protects the peptide-binding groove of the MHC class II heterodimer from being prematurely filled with self-proteins. The MHC class II/Ii complex is then transported from the ER through the Golgi from which vesicles deliver the complexes to the MIIC. There partial proteolytic cleavage of Ii occurs leaving a small fragment called CLIP (class II-associated invariant-chain peptide) in the peptide-binding groove of the MHC class II molecule (302). Another molecule, HLA-DM in humans or H-2M in the mouse, which have a structure similar to the MHC class II molecules, removes MHC class II-associated CLIP. This step allows endosomal antigenic peptide to take its place (241). Finally, MHC class II with the new antigenic peptide in its binding groove traverses the cytoplasm in exocytic vacuoles for display on the cell surface. Surface MHC class II molecules can be recycled from the cell surface through endocytic pathways and acquire new antigens in the MIIC. Immature DCs accumulate MHC class II and degraded peptide in lysosomal vesicles until the DCs are activated. After DC activation, the MHC class II/peptide complexes accumulate in nonlysosomal vesicles that migrate to the cell surface. It has recently been demonstrated that members of the B7 family of costimulatory molecules are embedded in the vesicular lipid along with MHC class II and are delivered to the cell surface in association with the MHC class II/peptide complexes (359).

View larger version (34K): [in this window] [in a new window]   Fig. 4. DC antigen processing. The major MHC class I pathways are depicted on the left. 1) Antigen is taken up by phagocytosis and receptor-mediated endocytosis, undergoes limited proteolysis, and by active transport enters the cytosol. The cytosolic antigens are further degraded via the proteosomal pathway, enter the endoplasmic reticulum (ER) utilizing TAP, and are bound to newly synthesized MHC class I molecules. MHC class I/peptide is subsequently carried by vesicular transport to the cell surface. 2) Endogenous proteins are similarly degraded via the proteosomal pathway, enter the ER utilizing TAP, are bound to MHC class I, and by vesicular transport reach the cell surface. The major MHC class II pathway is shown on the right. Antigen is taken up by phagocytosis, receptor-mediated endocytosis, and fluid- phase pinocytosis through the early and late endosomes, during which time some proteolysis occurs. The peptides enter the MHC class II-rich vesicular compartment (MIIC) where they are bound in the MHC class II peptide-binding groove and are then transported to the cell surface. MHC class II is synthesized in the ER where invariant chain (Ii) protects the groove from premature binding of self-peptides. Ii is further degraded into a smaller peptide (called CLIP and shown as a fragment in one of the three MHC moleculer in MIIC) to ready itself for its replacement by the antigenic peptide in MIIC. See text for additional details.

For MHC class I to display peptide antigens to CD8 T cells, DCs degrade cytoplasmic proteins in proteosomes and likely with the help of cytosolic heat shock proteins (HSPs) acting as chaperones, transport the resultant peptides via the heterodimeric transporter associated with antigen processing (TAP) into the ER (see Fig. 4, Ref. 254). In the ER, newly synthesized MHC class I chains form complexes with ₂-microglobulin (₂M) that then bind the resulting peptide antigens. Within the ER, chaperone proteins, including calnexin, calreticulin, and HSP gp96, aid in peptide binding and proper folding of the MHC class I/₂M complexes. Then tapsacin facilitates the formation of the MHC class I/₂M/peptide complexes (25, 273). Finally, MHC class I/₂M/peptide complexes are transported in exocytic vesicles to the DC plasma membrane.

It was originally thought that the proteosomal pathway processed only those proteins synthesized within APCs. It is now known that antigens can escape by poorly understood mechanisms from endocytic pathways, undergo proteosome-dependent degradation, and subsequently enter the ER via the TAP pathway to be presented in the binding groove of MHC class I molecules as is true for endogenous proteins (258, 270). This process is referred to as cross-presentation, and the resulting primary immune response is referred to as cross-priming (Fig. 4). The antigens may be particulate or soluble antigens, or come from dead and dying cells or from exosomes, which are small vesicles pinched off from the membranes of immature DCs and contain MHC molecules (191, 395). Cross-priming can lead to both productive immunity and tolerance of CD8 T cells. Cross-priming is the mechanism by which DC process neoplastic or infected self cells as well as engrafted allogeneic cells (191). DCs are the cells that are fully sufficient to perform cross-priming (184). Furthermore, at least in one in vivo murine study, this capacity seems restricted to the splenic CD8+ population even though the CD8 population was also capable of taking up the antigen (81). In any case, antigen uptake is an immature DC process and requires activation of the APC by costimulatory activation signals, such as cross-linking CD40, before cross-priming of the responder CD8 T cells can effectively occur. It has been postulated, and we shall return to this, that uptake of apoptotic cells by DC leads to tolerance induction, whereas DC uptake of necrotic cells tends to activate DCs to induce cytolytic CD8 T cells.

DCs are also capable of presenting antigens on CD1 molecules. CD1 molecules are a family of nonpolymorphic histocompatibility antigens associated, like MHC class I molecules, with ₂M (38, 303). CD1 molecules are present on myeloid DCs and, indeed, CD1a has frequently served as a marker for identifying these cells. Five CD1 isoforms, CD1a-e, have been described in humans, but only two CD1 homologs, CD1d1 and CD1d2, are expressed in the mouse and rat. The murine CD1d1 molecule crystal structure shows that it is constructed to bind very hydrophobic ligands, compatible with their presenting antigenic lipids (391). CD1 molecules present lipid and glycolipid antigens of both endogeous and exogenous derivation. The role of CD1d as a restricting element for endogenous lipid was shown to be relevant for self-antigen recognition by natural killer (NK)1.1+ T cells in mice (28). These / T cells have a restricted receptor binding repertoire in mice comprised of a V14-J281 TCR -chain paired with a restricted set of -chains, and in humans comprised of V 24-J  Q/V11. Upon responding to CD1/lipid antigen complexes, NK1.1+ T cells produce IL-4 or IFN- and have been implicated in recognition of infectious agents, tumors, and autoantigens. NK1.1+ T cells may express CD4 molecules or lack both CD4 and CD8 (28), and their cytokine production may be regulated by signal 3, i.e., IL-12 released by the CD1+ DCs (387). A second T cell subset restricted by CD1d on APCs has also been identified that does not express NK1.1. This T cell subset has a limited TCR repertoire and, like NK1.1+ T cells, can release either IL-4 or IFN- or express cytotoxicity (60, 370). Less is known about the TCR repertoire and its restriction to CD1a-c associated lipids in humans, but it is clear T cells can be restricted by these molecules, demonstrate limited diversity, and may bear either the / or / TCR (303).

Antigen processing and presentation by CD1 is different from that described above for MHC class I and II. CD1 molecules are synthesized in the ER and are expressed on the plasma membrane following traffic to the surface via vesicular transport. CD1 molecules are subsequently incorporated into endosomes and become associated with lipid ligands and recycle to the plasma membrane. Different CD1 homologs may associate with antigens in distinct endosomal compartments (337). In human myeloid DCs, CD1b binds to lipids that have been degraded in the deep endosomal compartments, whereas CD1a and CD1c associate with their corresponding lipid antigens in the recycling vesicular compartments of the early endocytic system. The purpose for the maintainance of these nonpolymorphic restricting elements during evolution may be to allow surveillance of normal intracellular lipid pathways, rather than for the development of protective immunity (303). However, it has been speculated that some microbes, such as Mycobacterium tuberculosis, developed the capacity to usurp CD1 molecules for their own purposes by having mycobacterial lipids presented to responder T cells to generate a granulomatous and necrotizing inflammatory response in the host. This process may allow the microbe to avoid complete eradication until it can be transmitted to the next host.

C.  Costimulation

1.  B7 family

Costimulation is required to initiate productive immune responses by T cells. The first and most important costimulatory molecules characterized were CD28 on naive T cells and the corresponding ligands, CD80 (B7-1) and CD86 (B7-2), which are upregulated on maturing APCs (69, 297). CD80 and CD86 molecules were the first described members of what is now known to be as a larger B7 subfamily; members of this subfamily belong to the immunoglobulin superfamily of proteins (see Table 2). CTLA-4 was later identified with strong homology to CD28 and is upregulated on activated T cells, binds with a higher affinity to CD80 and CD86 than CD28, and downregulates the immune response (369).

A third B7 family member is B7RP-1. B7RP-1 is expressed predominantly on B cells but is also found on macrophages, DCs, and nonlymphoid tissue cells and is the ligand for the inducible immune costimulator (ICOS) protein (175, 341, 389). ICOS is structurally related to CD28 and, like CTLA4, is upregulated on activated T cells (147). ICOS knockout mice demonstrate severely deficient T cell-dependent Th cell responses in which both the Th2-dependent antibody isotype IgG1 and the Th1-dependent antibody isotype IgG2a are decreased in the serum (88, 232, 343). Immune responses in ICOS knockout mice were characterized as defective with an absence of germinal center formation, a marked impairment of T cell IL-4 secretion, and IgE isotype switching. However, ICOS knockout mice were able to mount a Th1 cellular response, as noted by antigen-induced T cell IFN- secretion in response to immunization. The impaired T cell-dependent B cell responses suggest that after initial DC-T cell interactions in the paracortex of secondary lymphoid tissue, an important ICOS-dependent T cell-B cell interaction occurs that may finalize T cell subset differentiation. Importantly, in one study, addition of CD40 to upregulate CD40L largely repaired the ICOS knockout defect, suggesting the role of ICOS interacting with its ligand B7RP-1 on APCs is to facilitate the downstream CD40L-CD40 interaction, an interaction already known to be required for T cell-dependent B cell responses (232). The observation that in ICOS knockout mice experimental allergic encephalitis was exacerbated (88) suggested that peripheral antigen presentation by nonprofessional APCs might be important during the effector limb of immune-mediated inflammation. This suggestion is supported by the observation that the ICOS ligand B7RP-1 appears on nonlymphoid tissues. A clear role for B7RP-1 on DCs has not yet been established.

A fourth B7 family member, PD-L1 (also called B7-HI), is constituitively expressed on DCs and binds to the programmed cell death 1 (PD-1) receptor on T cells (reviewed in Ref. 69). This interaction exerts an inhibitor function on T cell proliferation and cytokine production. A related B7 family member, PD-L2, plays a similar role in inhibition and also binds to PD-1 but seems to have a more important role in Th2 inflammatory states, whereas PD-L1 participates in Th1 inflammation (69). Finally, B7-H3 has been recently identified and cloned (58). It is highly expressed on immature DCs and is downregulated on mature DCs, in contrast to the expression of most of the other B7 family molecules. Although the ligand for B7-H3 is unknown, it is not CD28, CTLA4, ICOS, or PD-1. B7-H3 plays an enhancing role in costimulation of both CD4 and CD8 T cells and in the induction of IFN- production, another surprising function in view of its higher expression on immature DCs.

The TNF family of ligands and receptors, now totaling ~50, are also important costimulators in DC interactions with T cells (Table 2). Particularly noteworthy is the interaction of CD40 on DCs with CD40L (CD154) on T cells. CD40 was first identified as a critical B cell molecule that interacted with CD40L on T cells to allow for isotype switching. During effective DC-T cell interactions, as T cells become activated, they upregulate CD40L. CD40L can then interact with CD40 on mature DCs to trigger IL-12 release, required for Th1 polarization (50, 56). Additional TNF family members, OX40 on T cells and OX40L on DCs, also play an important complementary role for inducing T cell proliferation and cytokine production (372). OX40L knockout mice fail to generate contact hypersensitivity (59), and OX40 knockout mice demonstrate reduced CD4 T cell proliferation, IFN- production, and protection against an influenza lung infection (176). Both OX40 and OX40L knockout mice demonstrate no humoral immune response defects, and OX40 knockout mice retain primary and memory cytotoxic T cell responses. Other TNF family members listed in Table 2 have been shown to provide important signals that enhance CD8 T cell proliferation and IFN- production (4-1BBL on DCs and 4-1BB on T cells) and DC cytokine secretion and survival RANK on DCs and RANK-L on T cells (reviewed in Ref. 21).

Finally, a heterogeneous group of receptors has been described as regulating DC-T cell interactions (Table 2). The leukocyte function antigen-1 (LFA-1; or CD11a/CD18) interaction with ICAM-1 (CD54) induces an adhesive interaction between DC and T cells which when disrupted markedly reduces the proliferative response of T cells stimulated by DCs (19, 229). However, this molecular interaction might also influence the type of immune response that develops in the T cell. In experiments using T cells from DO11.10 mice that express a TCR specific to OVA peptide 323-339 and stimulated by OVA peptide on splenic DCs, it was shown that interactions of LFA-1 with intracellular adhesion molecule (ICAM)-1 and ICAM-2 were important for generating Th1 responses. Thus blocking the interaction with a combination of anti-ICAM-1 and anti-ICAM-2 shifted the in vitro Th1 immune response to a Th2 response (290). DC-SIGN (which comes from DC-specific, ICAM-3 grabbing nonintegrin) is a 44-kDa type I membrane protein with an external mannose-binding, C-type lectin domain (324). It has been postulated that the interaction of DC-SIGN on DCs with ICAM-3 on T cells is required to initiate effective interaction of the MHC class II/peptide complex on DCs with the TCR on T cell (115). The postulate is that the DC-SIGN-ICAM-3 interaction allows sufficient DC-T cell adhesion for signaling to occur. DC-SIGN also binds to ICAM-2, and this interaction seems important for DC migration across both resting and activated endothelium (116).

A signal from an APC can be transmitted to a recently activated T cell rather quickly, whereas signaling a naive T cell may require more prolonged interaction. Time lapse microscopy has enabled the study of T cells interacting with planar membranes in which fluorescence-labeled adhesion molecules and MHC-peptide complexes can freely diffuse. Thus the planar membrane mimics to some extent the activity of antigen-pulsed APCs. With the use of this system, early central aggregation of LFA-1 with ICAM-1 (within 30 s) can be observed followed within 20 min by aggregation of the MHC/peptide complexes by the TCR, constituting signal 1. The clustering of complexes at the DC-T cell interface has been referred to as the "immunological synapse" (reviewed in Ref. 124). In addition to surface molecules, signaling molecules in the T cell such as Lck, Fyn, and ZAP 70 have been observed in the immunological synapse. Both the clustering of DCs with T cells as well as the subsequent T cell proliferative response is dependent on the reorganization of the T cell actin cytoskeleton and is characterized by the accumulation of filamentous actin and other cytoskeletal proteins at the T cell-DC interface. It was recently shown that disruption of the cytoskeleton by cytochalasin D in DCs interfered with effective clustering and activation of responder T cells (6). Prolonged binding of T cells to DCs has been shown in some systems to be required for optimal antigen stimulation (22). Others have noted that the time for DC-T cell interaction determines the immunological outcome with shorter periods of interaction favoring anergy and death, intermediate periods the development of memory, and longer periods resulting in differentiation into effector cells (187, 189). In a recent study in which DCs were allowed to interact with T cells in collagen gels, T cells were observed to crawl continuously over the surface of DC surface in short 6- to 12-min encounters. The repeated encounters resulted in T cell calcium influx and activation as measured by increases in activation markers and finally proliferation (131). This study raises the question as to whether the T cell needs to remain in an immobile state adhered to a DC to develop into a memory or effector T cell as implied by studies on planar members or whether, in contrast, active migration is the norm. In the intact animal, after subcutaneous antigen inoculation, initial interactions occur between DCs and T cells in T cell areas and continue at the interface of the T cell area with the follicles (153). In these latter in vivo studies, whether the DC-T cell interactions that resulted in expansion of the T cell population reflected prolonged immobilization of T cells by DCs or whether repeated short-lived interactions occurred could not be assessed.

	V. ROLE OF DENDRITIC CELLS IN LINKING INNATE AND ACQUIRED IMMUNITY

Top Previous Next References

DCs are members of the innate immune system, deployed throughout the body to sample the environment and determine whether a host response is needed and, if so, what kind of response. Infectious agents pose a threat to the normal host, who is generally well-equipped to resist these attacks. On infectious challenge of the nonimmune host, innate immunity expresses itself quickly with appropriate defenses, often in the absence of a detectible inflammatory response. An example of a regional innate response is mucociliary clearance in the lung, which daily exerts effective antimicrobial protection. Nevertheless, the ability of the lung to also respond with an adaptive immune response is essential, as has been learned from examples of multiple opportunistic pulmonary infections in victims of the human immunodeficiency virus (HIV). In 1992, Polly Matzinger proposed that it was the innate immune system that recognized danger and delivered nonspecific signals to specific T and B cells, stimulating them as to clonally divide and differentiate into effector lymphocytes and antibody-secreting cells, respectively (231, 234). Thus the older idea that the immune system learns a complicated set of rules early in ontogeny about how to recognize self and respond only to "not self" was exchanged for a simpler model. The newer model is that professional APCs respond to the environment and carry an antigenic message to responder T cells to instruct them to either develop tolerance or productive immunity. Current experimental evidence supports this notion. If a foreign or self-antigen is not dangerous, tolerance is the expected outcome, because DCs will not mature and, therefore, will not deliver a second signal. A corollary is that if a danger signal is given when self-antigens are processed, then an autoimmune response might develop.

As the initiator of T cell responses, DCs respond to danger signals and modify their function to generate an adaptive immune response. They use receptors to respond to the environment, first to take up, process, and present antigens and also to receive the danger signal. The "danger signal" receptors activate DCs causing them to engage intracellular machinery that 1) allows MHC class II to associate with immunogenic peptides in the proper endosomal compartment, 2) facilitates accumulation of MHC class II/peptide complexes in vacuoles together with costimulatory molecules that are subsequently coexpressed in domains on the cells surface, and 3) releases cytokines that further modulate the immune response (152, 166, 359).

Danger signals may arise from endogenous processes, particularly those that result in cell necrosis and tissue destruction (Table 3). Examples of specific signals of this type are cellular HSPs, matrix degradation products such as hyaluronan, and cellular cytokines and cell surface ligands such as TNF-, IL-1, and CD40L (4, 347). The normal turnover of cells is accomplished through programmed senescence whereby cells die via apoptosis, and neither inflammation nor development of adaptive immunity is a desirable outcome. On the other hand, when tissue is injured, particularly when necrosis results, inflammation is expected to initiate an appropriate repair response. HSPs are particularly important in signaling the host that something is amiss. Apoptotic cells do not express HSPs to stimulate an immune response. In contrast, before or as they die, necrotic cells release HSPs into the microenvironment (24). HSPs bind to DC receptors, induce DC maturation, and stimulate migration of DCs into secondary lymphoid tissue. The endogenous HSPs, HSP70, HSP90, and HSP96, all bind to CD91, which is present on DCs (23, 29, 30, 309).

Danger signals also derive from foreign substances. Microorganisms display motifs referred to as pathogen-associated molecular patterns (PAMPs) that stimulate DCs to undergo maturation. Microbial products are among the most common exogenous danger signals, and specific components include LPS from gram-negative organisms, peptidoglycans, and lipoteichoic acids from gram-positive organisms, microbial DNA which is rich in CpG motifs, microbial HSPs, and double-stranded viral RNA (see Table 3, Refs. 4, 14, 112). The most important receptors on DCs that recognize microbial products and transmit the message to initiate adaptive immunity are the toll-like receptors (TLRs), a family of highly conserved molecules initially described in Drosophila. TLRs are type I integral membrane receptors with extracellular leucine-rich regions and intracellular domains homologous to the signaling domain of IL-1R. Specific microbial ligands have been identified for vertebrate TLRs 2, 4, and 9 (see Table 3, Refs. 4, 135, 349). Upon interaction with their agonists, TLRs 2, 4, and 9 signal through the myeloid differentiation factor 88 (MyD88), IL-1 receptor-associated kinase (IRAK), TNF receptor-associated factor (TRAF), and the NF-B/Rel proteins, which finally results in NF-B translocation to the nucleus and transcriptional activation (10, 166). In a study examining the expression of TLRs 1-5 on human leukocytes, human DCs expressed all five, and TLR 3 was exclusively expressed on DCs (245). TLR 1-5 mRNAs were downregulated during the maturation of DCs. Furthermore, DCs responded to LPS via TLR 4, secreting TNF- but only while they were immature (367). These results suggested that only immature DCs are fully capable of responding to microbial products. Thus, after DCs mature, they would have only a short-lived ability to influence the naive T cell within secondary lymphoid organs. Thereafter, DCs would become unable to secrete polarizing cytokines such as IL-12 and undergo senescence, limiting the time frame within which they could stimulate immunity. In summary, DCs link innate and adaptive immunity by receiving danger signals that render them capable of maturing and inducing productive immunity rather than tolerance.

DCs regulate primary immune responses by directing antigen-specific T cells to die or to become anergic, memory, or effector T cells. In addition, the cytokines synthesized or the lytic machinery generated define the type of effector T cell, i.e., Th1, Th2, cytotoxic, or regulatory T cell. Lanzavecchia and Sallusto (188) have proposed a linear differentiation model for T cells during priming based on the persistence of the DC-T cell interaction in lymphoid organs. According to the model, the length of time that T cells and DCs interact defines effector function, homing, and survival of responder T cells with only the fittest T cells maintaining a DC-T cell interaction and surviving to become memory T cells. Excessive stimulation causes responding naive T cells to proliferate and develop effector function, but many of these responders soon die (157), some within the lymphoid organ and others after they migrate to tissue sites. Some effector T cells survive and persist as effector memory cells as the primary response wanes, although the survival signals these cells receive are not certain. Inadequate stimulation, either through a poor fit of the TCR for the DC's MHC/peptide complex (159, 310) or lack of costimulation as a result of a low level of DC costimulatory molecules or both, would lead to T cell anergy or programmed cell death. Finally, as the model continues, a second type of memory T cell, a central memory T cell, develops if the strength of DC stimulation falls somewhere intermediate between that required for generating anergy and that required for polarizing effector T cells (285). Both types of memory T lymphocytes survive for years ready to respond to cognate antigen displayed on APCs. Central memory T cells circulate between blood and lymphoid organs and respond to MHC/peptide complexes by further expansion and differentiation into effector cells, which subsequently migrate to relevant target sites. In contrast, effector memory T cells circulate between blood and peripheral tissues and, thus, can quickly respond to antigen by displaying immediate effector function (285). Each memory T cell type has distinct markers that allow them to migrate to the appropriate tissue site and to be distinguished from naive and effector T cells generated directly in the primary immune response.

Memory cells may be preserved through the persistence of antigen, but antigen is not required for persistent CD8 T cell memory (192). In one study, memory CD8 T cells were shown to persist, depending on the balance of IL-15 with IL-2 (181). DCs secrete IL-15 (32, 161, 163), which could contribute to persistence of memory CD8 T cells; however, to date the role for DCs, if any, in persistence of CD8 T cell memory is incompletely understood. Mechanisms for maintenance of CD4 T cell memory are currently a focus for study. Van Essen et al. (363) demonstrated that CD4 T cell memory depended on DCs to process and present antigen. The source of the antigen was thought to be antigen-antibody complexes on follicular dendritic cells (FDCs). This study also demonstrated a requirement for B cells to maintain memory development, likely to facilitate the development of FDCs and to secrete the complex-forming antibody.

Effector T cell differentiation from naive T cells requires prolonged TCR contact with MHC class II/peptide complexes and costimulation to induce proliferation and cytokine secretion to finally polarize the T cells (148, 188). DCs are ideal cells for performing this function, because expression of MHC class II/peptide complexes persist on mature DCs for over 100 h (52). T cells with high avidity interactions with DCs are the most likely to successfully compete for a long-term interaction with antigen-bearing APCs and, therefore, to repeatedly divide yielding daughter T cells capable of polarizing into effector cells. DCs that secrete IL-12 induce Th1 polarization (137), but, as has already been discussed, DCs are capable of producing IL-12 for only a short time (186). After 8-12 h, DCs exhaust their ability to produce IL-12 and subsequently activate proliferating T cells toward either a Th2 response or a regulatory T cell response. Interestingly, stimulation of Th1 subset polarization requires less time for the DC-T cells interaction than Th2 subset polarization, because there is an additional time requirement to induce demethylation of the IL-4 and IL-13 genes that is required for Th2 cell generation (149). If DCs emigrate into lymphoid tissue incapable of making IL-12, continued stimulation of naive T cells will induce the Th2 subset. In the absence of IL-12, the ability of responding naive T cells to make small amounts of IL-4 favors the development of Th2 T cells by autocrine stimulation. Most pathogens induce DC IL-12 synthesis and secretion while other factors such as PGE₂, IL-10, and selected microbes, such as hyphal forms of Candida albicans, inhibit IL-12 secretion (90; see Table 4 for a list of IL-12 inducing and suppressing stimuli). Because of the transient IL-12 production by antigen-presenting DCs, the continual influx into the regional responding lymphoid organs of fresh IL-12-secreting DCs from the peripheral site is required to drive a large pool of responders to become Th1 cells. Alternatively, as the antigen-specific T cells become activated, their expression of CD40L may also continue to enhance IL-12 secretion by DCs to levels needed to maintain a Th1 response. However, a CD40L signal is not successful in driving DCs to make IL-12 unless the DCs received a prior microbial signal (138, 313). As discussed, plasmacytoid DCs that carry antigens from systemic infected sites may also directly enter the T cell area of the lymph node by HEVs and continue to drive a Th1 response. In any case, once an infection clears and the microbial stimulus is gone, the ability for DCs to continue to make IL-12 would not be expected to continue.

Tolerance is the specific inability of a host to respond to antigens and is generated both centrally and peripherally. Central tolerance mechanisms occur in the thymus for T cells and in the bone marrow for B cells (1). T cells that might inadvertently respond to DCs carrying self-peptides are deleted during ontogeny in the thymus. T cells that fail to respond to stimuli in the thymus die from neglect, while T cells that recognize MHC/peptides with high avidity undergo apoptosis and are deleted; this latter process is called