Division of labor, plasticity, and crosstalk between dendritic cell subsets

When is a DC subset really a subset?

In mice there are at least 6 different subsets of DCs, in the “steady state.” In discussing DC subsets, a critical question is what one really means by a “subset.” Traditionally DC biologists have used a combination of variables including developmental origins and biological properties (e.g. phenotype, function, and micronevironmental localization) to sub-divide DCs into subsets. However opinions vary as to what combination of variables is best suited for this purpose. Classification based on development can be problematic, since the developmental pathways of DCs are not yet fully understood. For instance, earlier work suggesting that splenic CD8α⁺ and CD8α− DCs develop from distinct precursors supported their classification into distinct subsets, but has been challenged by subsequent work that demonstrates that both of these subsets can develop from the early clonal lymphoid and myeloid precursors (10). In fact, it is increasingly clear that the birfurcation of DC development into distinct streams that give rise to distinct DC subsets occurs relatively late in DC development, since precursors that yield splenic CD11c⁺ classical DCs versus plasmacytoid DCs can be isolated from the spleens (10, 16). Classifications based on phenotype whilst useful, can also pose complications. For example, as will be discussed later, intestinal DC subsets have been defined by different groups using different phenotypic markers (e.g. CX3CR1 versus CD103 versus the conventional markers such as CD11b and CD8α [17, 18]), thus giving rise to some confusion as to exactly which subset is being studied. Furthermore, in some cases, a sub-population of DC might have only a very subtle phenotypic difference from a different sub-population, and in this case the demarcation of such sub-populations as distinct subsets becomes more arbitrary. For example, the CD8α− DC subset can be sub-divided further into CD4⁺ versus CD4− DCs, which are phenotypically and functionally very similar (11). Thus, perhaps the most informative way to classify DCs is based on their biological properties, microenvironmental localization, and phenotype. For example, as will be discussed below, CD8α ⁺ DCs and CD8α− DCs in the spleen differ with respect to their phenotype, functions and microenvironmental localization, and thus can be considered to be two different subsets.

DC subsets: Phenotype, function and microenvironmental localization [Table 1]

DCs in the spleens and lymph nodes of mice, in the steady state, are characterized by the expression of the integrin-α CD11c and class II MHC. In the spleens of mice, at least 3 subsets of DCs have been described (Table 1, 10, 11, 14):

1. CD11c^highCD8α⁺ CD11b⁻ DEC205⁺ DCs [so called “CD8α⁺ lymphoid” DCs]

2. CD11c^highCD8α⁻ CD11b⁺ DEC205⁻ DCs [so called “CD8α⁻ myeloid” DCs]. This subset can be further divided into two, based on the expression of CD4 or 33D1.

3. CD11c^intermediateCD8α^+/− CD11b⁻B220⁺ Gr-1⁺ [plasmacytoid DCs]

In the lymph nodes of mice, in addition to these 3 or 4 subsets, there are at least 2 other subsets (Table 1, 10, 11, 14):

1. CD11c^highCD8α^dullDEC205^high Langerin⁺ [Langerhans cell-derived DCs (LCDCs)]

2. CD11c^highCD8α⁻ CD11b⁺ DEC205⁺ [dermal DCs]

The CD11c^highCD8α⁺ DCs are localized in the T-cell rich areas of the spleen and lymph nodes and have some predetermined capacity to secrete abundant IL-12(p70) and stimulate Th1 responses (10, 11, 14, 19–21). In contrast, CD8α⁻ DCs are localized in the marginal zones of the spleen, and the subcapsular sinuses of the lymph nodes, do not generally secrete much IL-12(p70), but rather secrete IL-10, and have been shown to induce Th2 responses (20, 21). The intrinsic functional differences between these subsets was elegantly demonstrated in vivo, by targeting an antigen to a specific subset using antibodies that bound to a specific receptor on each of the subsets (22). With regards to cross-presentation of particulate or soluble protein antigens to CD8⁺ T cells, it appears that the CD8α⁺ DCs are more efficient (23, 24). Furthermore, these DCs are preferentially able to endocytose dying cells in vivo, and cross-present cellular antigens to both CD8⁺ T cells (25). This is consistent with the differential antigen-processing ability of CD8α⁺ versus CD8α− DC subsets (26). However, both the CD8α⁺, and the CD8α− DC subsets are capable of cross-presenting antigens, after activation via the Fc_γ receptor (27). Thus, any innate differences in the ability of the distinct DC subsets to cross-present can be overcome by particular modes of activation.

The CD11c^highCD8α^dullDEC205^high Langerin⁺ DCs are Langerhans cell derived DCs (LCDCs), which have migrated to the lymph nodes, from the skin (9–11). The CD11c^highCD8α⁻ CD11b⁺DEC205⁺ DCs represent dermal DCs which have migrated to the LN from the dermis of the skin (9–11). Plasmacytoid DCs (pDCs) can be stimulated to rapidly secrete copious amounts of interferon-α [IFN-α] (28–30). In addition to these subsets which are present in the steady state, it is now clear that TLRs are expressed in early hematopoeitic cells (31), and that TLR-mediated inflammation can influence DC development (10, 16). For example, CCR2⁺ Ly6C⁺ inflammatory monocytes develop into DCs in an inflammatory environment (16, 32–34).

In the intestine, DCs have been described in four major locations: Peyer’s patches (PP), lamina propria (LP), and the mesenteric LNs. Four different subsets of DCs have been described in the PP (18):

1. CD11c^brightCD8α⁺CD11b⁻CCR6⁻ DCs, which like their CD11c^brightCD8α⁺ counterparts in the spleen and LNs (19–21), are situated in the T cell rich interfollicular region (IFR), can be induced to secrete high amounts of IL-12p70 and stimulate Th1 responses (18, 35, 36).

2. CD11c^brightCD8α⁻CD11b⁺CCR6⁻ DCs, which like their CD11c^brightCD8α⁺ counterparts in the spleen and LNs (19–21), are situated in the subepithelial dome (SED) and can be induced to secrete abundant IL-10, and to stimulate Th2 responses (18, 35, 36).

3. CD11c^brightCD8α⁻CD11b⁻CCR6⁺ which are located in both the IFR and SED and stimulate Th1 responses (18, 36).

4. CD11c^intermediateCD8α^+/− CD11b⁻B220⁺ Gr-1⁺ plasmacytoid DCs. However unlike pDCs from other tissues, PP pDCs appear incapable of producing significant levels of type I IFNs (37), perhaps via three regulatory factors associated with mucosal tissues, PGE(2), IL-10, and TGFβ (37).

In the LP, recent work shows two major subsets of DCs: CD11c^brightCD8α^dullCD11b⁺ DCs which stimulate Th17 responses (38), and CD11c^brightCD8α⁺CD11b⁻ whose function is unclear at present (38, 39). In addition, pDCs have been described in the LP (39). LP DCs have been observed to extend dendrites between epithelial cells into the intestinal lumen (40).

In addition to this “traditional” way of classifying DCs, recent work has used the chemokine receptor CX3CR1 (the receptor of CX3CL1, fractalkine) and the α integrin CD103 to classify DCs (18, 41–45). Thus mice, in which the gene for green fluorescent protein is “knocked in” to the CX3CR1 locus has revealed large numbers of CX3CR1⁺ DCs in the LP (41). The correlation between CX3CR1, CD103 and the “traditionally defined” subsets is at present murky, but our recent work suggests that CD11c^brightCD8α^dullCD11b⁺ DCs in the LP are CX3CR1⁺ and mostly CD103 intermediate (38). However, a minor subset within the CD11c^brightCD8α⁻CD11b⁺ DCs in the LP express very high levels of CD103. In contrast, CD11c^brightCD8α⁺CD11b⁻ DCs in the LP are CX3CR1- and CD103^bright (38). The functional properties of these various DC subsets are only now beginning to be appreciated. Recent work from several groups suggest that CD103 expressing DCs in the MLN (45), and DCs in the LP some of which express CD103 (44), induce T regulatory cells. This seems to be dependent on a mechanism involving retinoic acid (44–47). In the MLN, in addition to the CD11c^brightCD8α⁺CD11b⁻ and CD11c^brightCD8α⁻CD11b⁺ two additional DC subsets can be defined using the differential expression of CD4 and DEC-205 (18). However, the functional significance of these remains unknown.

Plasticity in response to microbial stimuli

The ability to respond rapidly to invading pathogens is a hallmark of DC function. It has been known for sometime that DCs cultured in vitro can be “conditioned” by microbial stimuli, or cytokines (10–14, 48). DCs like many other immune cells express a variety of pathogen recognition receptors (PRRs), such as TLRs non-TLRs (which includes C-type lectin like receptors or CLRs, NOD-like receptors or NLRs and RIG-I like receptors or RLRs), which allows them to sense a myriad of microbial stimuli and initiate an innate response (11–15). In fact, recent work suggests that direct activation of DCs by pathogen derived stimuli via TLRs is necessary to promote full DC activation and stimulation of Th1 or Th2 cells (11–15). Inflammatory mediators originating from nonhematopoietic tissues or from radioresistant hematopoietic cells seem neither sufficient nor required for full DC activation in vivo (49). Most work has focused mainly on TLRs, but emerging evidence suggests a role for non-TLRs also in modulating DC function. Numerous recent articles (11–15, 50) have reviewed the roles played by TLRs, C-type lectins, NOD-like recptors and RIG-I and MDA-5 in this process, so we will not discuss this further here.

Plasticity in response to tissue derived signals

Although direct sensing of pathogen stimuli seems to be critical for the elicitation of full DC activation and effective polarization of T helper responses (49), a growing body of evidence suggests that cytokines and chemokines in the local microenvironment can also modulate DC function and influence the immune response. Early in vitro experiments suggest that DCs cultured with IL-10 or TGF-β, induce Th2 or T regulatory responses while DCs cultured with IFN-γ induce potent Th1 responses (reviewed in 11–15, 48). These observations raised the possibility that the microenvironment potently influences DC function. In particular, emerging evidence suggests that stromal cells in the tissue microenvironment can condition DCs in a microenvironment-specific manner. Consistent with this, Peyer’s patch or respiratory tract DCs were shown to prime Th2 responses, while total spleen DCs primed Th1/Th0 responses (reviewed in 11–15, 48).

Recent studies have begun to provide direct evidence for the role of stromal cells in conditioning the function of DCs. For example, recent work suggests that splenic or endothelial stromal cells can program the differentiation of DCs into cells that induce T regulatory cells (51, 52). In the case of Th2 differentiation, thymic stromal lymphopoietin (TSLP), an epithelial cell derived cytokine strongly activates DCs, to stimulate inflammatory Th2 cells, via an OX40-L dependent mechanism (53). In the intestine, under steady state conditions, TSLP is produced by intestinal epithelial cells and influences DCs to prime Th2 responses (54). In addition, over expression of TSLP produced by keratinocytes or airway epithelial cells correlates with Th2 diseases such as atopic dermatitis and asthma (55). In contrast to these effects on Th2 responses, in the human thymus, TSLP expressed by Hassal’s corpuscles is known to instruct DCs to select T regulatory cells (56). These observations may explain classical observations that the route of antigen delivery is a crucial determinant of the type of immune response; thus inhaled antigens induce a Th2 response, while antigens injected subcutaneously induce a Th1 response.