Antigen cross-presentation by dendritic cell subsets: one general or all sergeants?

Abstract

Antigen cross-presentation describes the process through which dendritic cells (DCs) acquire exogenous antigens for presentation on MHC class I molecules. The ability to cross-present has been thought of as a feature of specialized DC subsets. Emerging data, however, suggest that the cross-presenting ability of each DC subset is tuned by and dependent on several factors, such as DC location and activation status, and the type of antigen and inflammatory signals. Thus, we argue that capacity of cross-presentation is not an exclusive trait of one or several distinct DC subtypes, but rather a common feature of the DC family in both mice and humans. Understanding DC subset activation and antigen-presentation pathways might yield improved tools and targets to exploit the unique cross-presenting capacity of DCs in immunotherapy.

DCs and antigen cross-presentation

DCs are professional antigen-presenting cells (APCs) that are uniquely capable of attracting and activating naïve CD4⁺ and CD8⁺ T cells. After infection or inflammation, DCs undergo a complex maturation process, and migrate to lymph nodes (LNs) where they present antigens to T cells. Immature DCs acquire exogenous antigens, which they can present on MHC class I molecules via the process of cross-presentation. Cross-presentation is thought to occur through one of two main pathways [1]. The ‘canonical’ endosome-to-cytosol pathway involves transport of exogenous antigens from endosomal vesicles into the cytosol, where they are trimmed and processed by the proteasome and subsequently loaded on MHC class I molecules in the endoplasmic reticulum, similar to endogenous antigens [2]. In the second, proteasome-independent cytosol-independent pathway, DCs use endosomal proteases to process and load captured antigens directly onto MHC class I molecules in endosomal compartments [3]. Antigen cross-presentation should not be confused with cross-priming, which also describes priming of CD8⁺ T cells and which requires additional signals. Antigen cross-presentation is crucial for the priming of CD8⁺ cytotoxic T cell responses against pathogens and tumors, but is also required for the maintenance of self-tolerance [4]. For proper differentiation between the two, it is important that experimental studies include crucial controls, for example, peptide controls, and titration of DCs or antigens.

Different DC subtypes are equipped with a diverse repertoire of antigen capture and innate sensing mechanisms. Thus, different DC subsets may differ in their capacity to acquire and process internalized antigens, produce cytokines, and activate T cells in response to distinct microorganisms. Cross-presentation capacity could be intrinsic to the specific DC subset or could be influenced by antigen or other environmental factors. For instance, steady state CD8α⁺ DCs are equipped with machinery to control endosomal maturation and acidification and are efficient in transporting antigens into the cytosol relative to CD8α⁻ DCs [5-7]. These components may be regulated or induced by environmental stimuli. Furthermore, although plasmacytoid DCs (pDCs) fail to transport cytochrome c to their cytosol, a measure of cross-presenting ability, they are able to cross-present exogenous antigens. Possibly, pDCs and other CD8α⁻ DCs may exploit different mechanisms to cross-present and/or transport antigens to the cytosol for cross-presentation.

Cross-presentation has mostly been studied in human monocyte-derived DCs (MoDCs), and in mouse spleen- or bone-marrow-derived DCs, using model proteins, such as chicken ovalbumin (OVA), which are provided as cell-associated or soluble antigens [8-13]. Such in vitro studies provide the basis for current understanding of antigen cross-presentation mechanisms. However, the cross-presenting ability of other DC subsets, such as human pDCs or CD8α⁻CD11b⁻ DCs in mice, for different antigen sources should not be ignored [14-17]. In vivo studies in which specific DC subsets are selectively depleted, for example, CD8α⁺ DCs in Batf3^−/− mice or diphtheria-toxin-based depletion studies, have provided pivotal information on the functional role of DC subsets in antigen presentation [18,19]. However, the interpretation of such depletion studies investigating cross-presentation can be complicated by incomplete deletion, depletion-associated side effects, and DC crosstalk (reviewed in [20]). Nevertheless, multiple in vivo studies have demonstrated that the CD8⁻lineage DCs [18] are indispensable for antigen cross-presentation and not pDCs [21,22] or Langerhans cells (LCs) [23]. By contrast, other in vivo studies with pDC-depleted mice have provided evidence that activated pDCs do play a role in antigen cross-presentation and CD8⁺ T cell priming [16,24]. Furthermore, in Batf3^−/− mice residual cross-presentation capacity is observed and is responsible for protection against tumors. This indicates that other DC subsets cross present, albeit less efficiently than CD8α⁺ DCs [18]. These studies support the view that, under certain circumstances, specific DC subsets are required for in vivo cross-presentation, and that other DC subsets might be dispensable. They also leave us wondering about whether or not all DCs may be potent cross-presenters in specified conditions, and if yes, what is needed to acquire these cross-presenting abilities.

Here, we review the capacities of mouse DC populations to cross-present directly cell-associated, soluble, immune-complexed and particulate antigens, and antigens derived from nonviral intruders such as bacteria or fungi in different locations and under (non)-inflammatory conditions, and we examine how these findings extrapolate to human DC subsets.

Phenotype and cross-presentation capacity of DC subsets in mice

Genetic profiling has identified a common origin of many DC subsets together with the transcription factors needed for DC lineage commitment (Box 1) [25-29]. An outstanding question is whether efficient cross-presentation is an exclusive trait of some DC subpopulations or a common feature of many or even all DCs.

Box 1. Characterization of DC subsets.

The characterization of DC subsets is an ongoing process. Characterization of migratory DC subsets in peripheral tissues and lymphoid organs is particularly complicated due to tissue-specific and inflammation-dependent expression kinetics of phenotypic markers. The use of a combination of markers (all nonexclusive when used alone) is therefore advised to study the selective characteristics of DC subsets.

Murine conventional DCs: express high levels of CD11c and are further subdivided in blood-derived resident DCs and migratory DCs. The first group resides in the spleen and LNs and is generally subdivided into CD8α⁺ and CD11b⁺ or CD4⁺.

CD8α⁺-expressing DCs: identified in the spleen and LNs, selectively express the transcription factors basic leucine zipper transcription factor, ATF-like 3 (Batf3) and interferon regulatory factor 8 (IRF8), and high levels of CD24, CD205 (DEC-205), chemokine (C motif) receptor (XCR1), and C-type lectin domain family 9A (CLEC9A). CD103 expression varies between DCs, but is mostly found on migratory CD8α⁺ DCs and may relate to an activation or developmental state [109]. Analyses of CD24⁺ DCs in CD8α-deficient mice and FLT3L-stimulated bone-marrow-derived DCs reveals that CD8α is dispensable for the characteristic functional capacities of this subset [30]. As CD8α is expressed relatively late in DC development, is has been suggested that CD24⁺CD8α⁻ cells may develop into CD8α⁺ DCs [17].

CD11b⁺ DCs: The transcription factor reticuloendotheliosis homolog B (RelB) drives the development of cDCs that lack CD8α but express CD11b, CD172a [signal regulatory protein (Sirp-α)], and DC immunoreceptor (DCIR)2, and may show expression of Dectin-1 (Clec7a). Less than 50% of CD11b/CD172a⁺ cells express CD4, but no clear discrimination has been found in the function between CD4⁺ and CD4⁻ CD11b⁺ DCs.

CD8α⁻CD11b⁻ DCs: a population of spleen DCs that may express CD24, but not CD4, CD8, and CD11b/CD172α.

Migratory DCs: differ in phenotype dependent on the microenvironment in which they reside, such as skin, intestine, or lung tissues. In skin, LCs abundantly express the C-type lectin langerin (CD207). However, later findings indicate that CD207 is also expressed by (CD103⁺) dermal DCs [34].

MoDCs: isolated from spleen are characterized either by the expression of CD11b⁺Ly6c⁺CD11c⁺MHCII⁺, or on the expression of DC-SIGN/CD209a in combination with CD11b⁺CD11c⁺ for identification.

Human conventional DCs: are CD11c⁺ and are divided according to the specific and nonoverlapping expression of CD1c (BDCA1) and CD141 (BDCA3). Recently, DCs were characterized in human LNs, tonsil, and spleen in untreated breast cancer patients: pDCs (BDCA4), LCs (Epcam⁺), CD1a⁺ DCs, CLEC9a⁺ DCs, and two populations of BDCA1⁺ DCs showing differential expression of CD206. Three of these DC subsets (LCs, CD1a⁺ and CD206⁺ DCs) are absent from cervical LNs draining the oropharynx, iliac LNs, tonsils, and spleen, suggesting that these cells specifically drain from the skin [91].

pDCs: murine pDCs express intermediate levels of CD11c, and high levels for CD45RA (B220), sialic acid-binding immunoglobulin-like lectins-H (Siglec-H) and/or mouse pDC Ag 1 [mPDCA1; bone marrow stromal cell antigen 2 (BST-2), 120g8], whereas their human equivalents lack the expression of CD11c, but rather are discriminated based on the expression of BDCA2, BDCA4, or CD123.

CD8α⁺ DCs

CD8α⁺ DCs (Box 1) [17,30] are generally thought of as the dominant, if not exclusive, cross-presenting DC subset, irrespective of antigen type (Table 1). It has been shown over a decade ago that, in a population of low-density splenocytes isolated from mice primed with OVA-loaded β2-microglobulin-deficient cells, ex vivo depletion of CD8α- ^high but not CD11b⁺ cells abrogated cross-presentation to OT-I cells [13]. The cross-presenting ability of splenic CD8α⁺ DCs has since been confirmed, not only for cell-associated antigens, but also for proteins, OVA-coated latex beads, immune complexes, and many pathogens (Table 1) [31-35]. However, in experiments using Escherichia coli and Saccharomyces cerevisiae, CD8α⁺ DCs cross-present less well than CD8α⁻ DCs [33,35], suggesting that the immunostimulatory features of distinct pathogens may determine the cross-presenting capacity of DC subsets. This is also suggested by the finding that CD8α⁺ DCs isolated from skin-draining LNs fail to cross-present intramembrane antigens (i.e., cell-associated antigens) in the K5.mOVA transgenic mouse model [36]. In this model, the transgene OVA is fused to the transmembrane domain of the transferrin receptor under the control of the K5 keratin promoter [36]. By contrast, in a model in which transgenic mice express yellow fluorescent protein linked to cytotoxic T lymphocyte (CTL) epitopes for glycoprotein B (gB) of herpes simplex virus under the rat insulin promoter, the CD8α⁺ LN DCs were instrumental in cross-tolerization of gB-specific hybridoma T cells [37]. Furthermore, CD8α⁺ DCs from mesenteric LNs poorly cross-presented intestinal soluble OVA [38]. The cross-priming function of CD8α⁺ DCs in LNs might also be dictated by immunostimulatory features of the surrounding environment. CD8α⁺ DCs in skin-draining LNs can potently cross-present OVA–Toll-like receptor (TLR)7 conjugates [39] and saponin-formulated antigens [40], emphasizing the potential of pattern-recognition receptor (PRR) ligands and adjuvants in conventional DC (cDC) function.

Table 1.

Mouse spleen.

Subsets defined by:	Cell-associated	Soluble protein	IC	Particulate	Bacteria/yeast/yarasite
CD8α+CD11b−	Yes a [9,13,17,33,44,47–50,53,58,105,107,110] b[51] a/b [11]	Yes a [5,7,9,47,53,54,38] b [110] a/b [11,57,102] a/c [40] Low a [110]	Yes a [55] Low a [54]	Yes a [5,30,53]	Yes (E. coli) [33,53] (rLM) [34] (Plasmodium) [31] (Salmonella) [32] (yeast) [35]
CD8α+CD103−	No a [41]	Low a [41]
CD8α+CD103+	Yes a [41]	Yes a [41]
CD8α−	No a [9,47,48]	Yes a [47] Low a [5,7,9,38]	Yes a [55]	No a [53] Low a [5]	Yes (yeast) [35] (Salmonella) [32]
CD8α−CD11b+	No a[13,44] b [51] Low a [49,50]	No a/b [57] Low a [54]	Yes a [54]
CD4−CD8−	Low a [33,53]	Yes a/b [102] No a [102] Low a [53]			No (Plasmodium) [31]
CD4+	No a [20,53,107]	Yes c [40] No a [40,102] a/b [57] Low a [53]		No a [30]	Yes (E. coli) [33,53] Low (Plasmodium) [31]
CD8α−CD11b−	Yes a [49,50,58]
CD8α−CD11b−CD24+	Yes a [17] b [51]	Yes a/b [57]
pDCs	No a [9,49,50,58]	Yes a [16] No a [9]		Yes a [16]	No (rLM) [34] (Plasmodium) [31]

^aNot matured.

^bMatured with TLR-L or CD40L.

^cSaponin-based formulation.

A recent study revealed that splenic CD8α⁺CD103⁺ DCs presented cell-associated and soluble antigens more efficiently than CD8α⁺CD103⁻ DCs did [41], suggesting differences within the CD8α⁺ subset. However, most studies have not distinguished CD103⁺ from CD103⁻ CD8α⁺ DCs, making it difficult to interpret these results for DC subsets in LNs or for other types of antigen. Although CD8α is not expressed in in vitro generated Fms-related tyrosine kinase 3 ligand (FLT3-L) DCs cultures, the use of CD24 enables the identification of CD8α⁺ DC equivalents that potently cross-present cell-associated [42-44] and soluble [42,43,45] antigens, and antigen-coated latex beads [43]. Moreover, the cross-presentation of cellular [46] and soluble [46] antigens resides predominantly in the CD103⁺ DCs. The expression of CD103 itself seems, however, not equivalent to a cross-presenting phenotype [46]. In conclusion, CD8α⁺ are able to cross-present a broad spectrum of antigenic formulations at steady state conditions, but environmental factors may affect that function in the peripheral lymphoid organs, in particular in responses to nonsterile infections.

CD11b⁺ DCs

Splenic DCs that lack CD8α [9,47,48] or that express CD11b [13,37,44,49-51] or CD4 [52,53] are inefficient in cross-presenting cell-associated antigens (Table 1), soluble proteins [5,7,9,38,42,47,53,54] or antigen-coated beads [5,30,53]. Furthermore, CpG-matured CD11b⁺ DCs loaded with dying cells fail to cross-present their cargo in vivo [51]. Therefore, these DCs seem to lack cross-presenting capabilities. However, CD8α⁻ [55] and CD11b⁺ DCs do induce potent CD8⁺ T cell responses when immune complexes are used instead of cell-associated or soluble antigens [54]. Moreover, CD4⁺ DCs potently cross-present soluble antigens administered together with saponin-based adjuvants [40] and initiate CD8 T cell responses to E. coli [33,53]. DCs lacking CD8α also more efficiently cross-present OVA antigens from Salmonella typhimurium [32] and S. cerevisiae than CD8α⁺ DCs do [35]. These data emphasize the importance of specific immune activation signals to acquire the ability to cross present. The exact mechanism for the stimulation of cross-presentation by saponin-based adjuvants is not clear yet. This is of interest, because in contrast to most clinically used adjuvants, these adjuvants seem to stimulate CD8⁺ T cell responses in particular [40,56].

CD8α⁻CD11b⁻ DCs

A population of splenic DCs that express CD24, but not CD4, CD8 and CD11b/CD172α was described in 2007 that has similar cross-presenting capacity to CD8α⁺ DCs (Box 1) [57]. Forty-two percent upregulated CD8α after overnight culture [57] and >80% of CD24⁺CD8α⁻ cells expressed CD8α 4 days after transfer [17], suggesting that these cells were ‘in transit’ towards a CD8α⁺ DC phenotype. A different study isolated splenic CD11b⁻ DCs into CD8α⁺ and CD8α⁻ subsets, and found that CD11b⁻CD8α⁻ DCs were more potent cross-presenters of cell-associated antigens than were CD8α⁺ DCs under steady-state conditions [58]. However, CD24 was not included in that study, preventing direct comparison with the data of Bedoui et al. [17]. The CD11b⁻CD8α⁻ DCs preferentially internalized small particles, derived from dying cells (and were accordingly designated merocytic DCs), that are stored in non-acidic compartments with reduced lysosomal degradation for prolonged periods of time. The size of the internalized particles seems favorable for cross-presentation (0.5 and 3 mm) [59]. Indeed, CD11b⁻CD8α⁻ DCs show sustained antigen presentation to both CD4⁺ and CD8⁺ T cells with enhanced effector functions and memory formation when compared with CD8α⁺ DCs [49,50]. It will be interesting to determine whether these DCs exploit comparable mechanisms as CD8α⁺ DCs to delay antigen degradation to enhance their antigen-presenting capacity.

Notably, the cross-presenting capacity of merocytic DCs is also most increased after crosstalk with pDCs when compared to CD8α⁺ and CD11b⁺ DCs [51]. It will be interesting to dissect the role of merocytic DCs during bacterial and viral infections and whether there is a human equivalent of this DC subset.

Mouse pDCs

Relative to splenic cDCs, cross-presentation by pDCs (Box 1) remains largely understudied. The few available studies suggest that splenic murine pDCs do not cross-present cell-associated antigens [49,50,58], soluble OVA, and peptide-coated beads [16] at steady-state (Table 1). They do not seem to prime CD8⁺ T cells in response to Listeria monocytogenes [34], Plasmodium berghei [31], or after intragastric administration of protein [38]. In vitro generated FLT3-L pDCs also show low cross-presenting abilities when loaded with cell-associated [58] or soluble [45,60] antigens. By contrast, activation of pDCs by the TLR7/8 ligand R848 and to a lesser extent CpG (TLR9) leads to efficient cross-presentation of soluble antigens and antigen-coated beads [16], again emphasizing the significance of specific stimulation of DC subsets in determining their cross-presenting capacity. Thus, murine pDCs seem inefficient in cross-presenting exogenous antigens but further work is needed.

Migratory DCs

Conventional DCs include lymphoid-organ-resident DCs and migratory DCs. They are present in nonlymphoid organs and migrate to the draining LNs. The phenotype of migratory DCs differs depending on the surrounding microenvironment, for example, skin, intestine, or lung. Characterization of these cells is an ongoing process and studies have used different panels for discrimination, making it difficult to compare findings. LCs are the primary DCs in the epidermis. They express the C-type lectin langerin (CD207) and lack CD103. In the dermis, DCs are generally divided into CD207⁺ and CD207⁻ cells and further subdivided based on expression of CD103, CD11b, and CD326 [36,61,62]. LCs were initially reported to cross-present OVA in the K5.mOVA transgenic mouse model [63]. However, it was later recognized that CD207 is also expressed by (CD103⁺) dermal DCs (dDCs), which are in fact responsible for cross-presentation in this model [36,62]. TLR7-conjugated antigens are efficiently presented by LCs, skin-derived CD103⁻CD205⁻CD326⁻ DCs, and to a lesser extent by CD11b⁺CD205⁻CD326⁻ DCs (claimed to be resident CD11b DCs) and CD103⁺ DCs [39]. LCs are efficiently targeted in vivo for cross-presentation of anti-DEC205-coupled OVA. By contrast, LCs targeted through langerin fail to stimulate T cell proliferation [64], suggesting that cross-presentation by LCs is dependent on how antigens enter the cell and are routed to the antigen-processing machinery.

In immunological tissues in the intestinal tract (Peyer’s patches, mesenteric LNs, and lamina propria) and lung CD11c⁺, CD11b⁻/CD11b⁺, CD103⁺ and F4.80⁻ are mostly used to identify DCs [65,66]. CD11b⁺ DCs in the mesenteric LNs cross-present intragastrically administered antigen better than CD8α⁺ DCs [38]. In lung draining LNs, one study has shown that CD103⁺ DCs, but not CD11b⁺CD205^int DCs, present soluble antigens LNs [66] and antigens from infected cells [67]. Although in the latter study productive infection was not observed in CD103⁺ DCs, direct transfer of genetic material cannot be excluded. Another study has shown that CD103⁻CD11b^hi DCs efficiently capture antigens from the lungs, migrate to the draining LNs, and cross-prime CD8⁺ T cells [68]. Expression of CD11b on DCs in the periphery is promiscuous, therefore, these CD103⁻CD11b^hi cells may not only refer to migratory lung DCs, but also to inflammatory (monocyte-derived) DCs that are attracted during the course of the ongoing infection at the time of antigen exposure.

In all studies differential cross-presentation capacity is observed in subsets of migratory DCs in the skin, intestinal tract, and lungs. The limited number of these studies and the large experimental diversities (e.g., different subset-specific markers, antigens, and stimulatory/inflammatory conditions) has hampered side-by-side comparisons between migratory DC subsets.

Mouse MoDCs

Bone-marrow-derived DCs generated in vitro using granulocyte–macrophage colony-stimulating factor (GM-CSF) alone or in combination with interleukin (IL)-4 cross-present cell-associated antigens [42,58], immune complexes [69-74], PLGA [poly(lactic-co-glycolic acid)] particles [75], or polystyrene beads [76], and cross-priming is efficiently induced by targeting antigen conjugated to anti-DEC205 [69] or anti-DC-SIGN (dendritic cell-specific intercellular adhesion molecule-3-Grabbing nonintegrin) [77]. Responses to soluble protein range from none [73-75] to low [72] and high [42,69,70,77-80]. The suboptimal results with soluble antigen may be overcome by using an adjuvant, such as saponin-based adjuvants [56,81]. The administration of long peptides [82-84] might also be a promising alternative, because it circumvents the need for additional adjuvants. It has been suggested that mouse bone-marrow-derived DCs display similar morphology, phenotype, and immunostimulatory activity as blood MoDCs [85] that may differentiate from monocytes under inflammatory conditions. Inflammatory MoDCs isolated from the spleen (characterized in these studies as CD11b⁺Ly6c⁺CD11c⁺MHCII⁺) are efficiently cross-presenting soluble antigens [9,11]. These cells may relate to tumor necrosis factor (TNF)/inducible NO synthase (iNOS)-producing DCs, of which the gene expression profiles are similar to those of activated monocytes rather than cDCs [28]. New additional markers, such as Fc receptor phenotypes, might be helpful for better discrimination.

Human DC subsets

Cross-presenting capacity of human blood DC subsets

cDCs and pDCs are the two main subtypes distinguished in the blood. The cDC subset can be further divided into at least two subtypes by the expression of CD11c in combination with CD1c blood dendritic cell antigen (BDCA)1 and CD141 (BDCA3) [86,87]. BDCA1⁺ DCs are presented as the human counterpart of murine CD11b⁺ DCs [25,26]. In contrast to murine CD11b⁺ DCs, blood-derived BDCA1⁺ DCs cross-present cell-associated [2,8,88-90], long peptides [91], soluble antigens [81,89,90,92,93], as well as immune complexes [8,92] (Table 2). In some studies cross-presentation was rather low in steady state conditions, but was enhanced by addition of either TLR ligands [88] or saponin-based adjuvants [2,8,81]. BDCA1⁺ DCs are less efficient in cross-presentation of cell-associated and soluble antigens than are BDCA3⁺ DCs [90].

Table 2.

Human DC subsets.

Subsets defined by:	Cell-associated	Soluble protein	IC
CD1c (BDCA1)	Yes [2] b/c [88] Low b [89] d [90] c [14]	Yes a [8] c [14] c/d [89,93] No d [99] c/d [81] Low [92] d [90] c/d [8] a [81]	Yes d [8] Low [92]
CD141 (BDCA3)	Yes d [90] c [14] b [89] b/c [88]	Yes [92] d [90] c [14,89,93,99] No/low d [89,93,99]	Yes [92]
CBDCs CD141/DNGR1e		Yes c/d [82] (long peptide) No d [82]
pDCs	Yes c [14] b [94] b/c [2,88]	Yes c [14] c/d [93] Low c/d [14] No d [90] 0/c [8]	No c/d [8]
MoDCs (IL-4/GM-CSF) In vitro	Yes Blood [12] b [10,95–98]	Yes d [108,111] a/c [8] a [81] No d [8,81] Low c/d [82] (long peptide)	Yes c/d [8] [112]

^aSaponin-based formulation.

^bCells loaded with virus (and thus intrinsic PRR activity of virus).

^cMatured with TLR-L or CD40L.

^dNot matured.

^eCBDCs, cord blood dendritic cells; DNGR, dendritic cell natural killer lectin group receptor.

In contrast to murine pDCs, human pDCs cross-present cell-associated antigens [2,14,88,94], soluble antigens [14,93], and a vaccinal lipopeptide preparation [2], but fail to cross-present immune complexes or antigens packed within saponin-based adjuvant formulations to CD8⁺ T cells [8]. However, some groups have demonstrated that soluble antigens are not efficiently cross-presented by pDCs [8,90].

Human BDCA3⁺ and murine CD8α⁺ DCs overlap both in gene expression patterns and ability to cross-present cell-associated antigens [88-90], long peptides [91], soluble proteins [89,90,92,93], and immune complexes [92]. All blood DCs cross-present poorly unless properly activated by TLR agonists [88,89,91]. BDCA3⁺ DCs generated in vitro from cord blood hematopoietic stem cells (HSCs) are able to cross-present long peptides when matured with the TLR3 ligand poly I:C [82]. The low numbers of BDCA3⁺ DCs circulating in peripheral blood limits the ex vivo modulation of these cells and requires efficient in vivo targeting strategies. Alternatively, adaptation of the protocol to produce BDCA3⁺ DCs from cord blood HSCs [82] may lead to generation of cord blood DCs enriched for this specific subset.

Studies with human MoDCs differentiated in vitro with GM-CSF and IL-4 report variable results with regard to effective cross-presentation [8,10,81,82,95-98]. Ex vivo data on cross-presenting function of (activated) monocytes is so far lacking. Furthermore, as for their murine counterparts, only a few studies have investigated the cross-presentation capacity of human LCs [15,99,100].

In conclusion, findings from mice do not always translate to the human setting. For instance, murine pDCs mostly fail to cross-present antigens but human counterparts seem to cross-present antigens and cross-prime CD8⁺ T cells efficiently. Furthermore, care should be taken when isolated DC subsets are used, for example, BDCA1⁺ DCs express low levels of CD14, which should be taken into account when using a lineage cocktail or CD14⁺ monocyte depletion. Studies suggest that most human DC subsets can cross-present exogenous antigens when the antigens are provided in an appropriate fashion.

Factors influencing cross-presenting capacity

The capacity to cross-present exogenous antigens may not be restricted to a specialized DC subset. Rather, it seems that a cross-presentation program can be initiated in most if not all DC subsets. Factors emerging as important for the modulation of the cross-presentation activity of specific DC subsets are: (i) type of antigen; (ii) presence of DC stimulatory factors, which can be altered by pathogens or adjuvants; and (iii) timing and phase of the immune response.

Type of antigen

DCs encounter antigens in many shapes and sizes, derived from various sources. The ability of DCs to handle these different antigen types is largely determined by the repertoire of antigen uptake receptors, and the ability to engulf antigens through receptor-independent processes, such as (micro)pinocytosis (Figure 1). The current paradigm of superior cross-presentation by murine CD8α⁺ DCs is mainly derived from the preferential use of specific antigen types such as cell-associated antigens, a restricted number of soluble model antigens or bead-bound antigens. CD8α⁺ DCs have in a few studies been found to perform worse than CD8α⁻ DCs in cross-presentation of immune complexes [54,55]. The same has been observed for presentation of yeast [35] and Salmonella-derived [32] antigens. In some cases, the results may have been affected by the disregard of additional markers to distinguish CD11b⁺ cells from merocytic cells, which are even more potent cross-presenters of cell-associated materials in direct comparison analyses. This may be especially true for studies in which DCs are isolated from FLT3-ligand-treated mice because CD8α⁻CD11b⁻ DCs are preferentially enriched in these animals [17,50].

Figure 1.

Decisive factors in dendritic cell (DC) cross-presentation. The ability of DCs to cross-present antigens is not just an intrinsic property of the specific DC subset, but is also determined by: (i) type of antigen; (ii) presence of DC stimulatory factors; and (iii) timing and phase of the immune response. DCs encounter antigens of many origins, shapes, and sizes. The ability of DCs to internalize soluble antigens, immune complexes, dying cells, or whole pathogens is largely determined by the repertoire of antigen uptake receptors (e.g., Fc receptors, CD36, and C-type lectins) and the ability to engulf antigens through receptor-independent processes. The dynamic endocytic receptor expression is, in turn, affected by ligation of pattern-recognition receptors [PRRs; e.g., Toll-like receptors (TLRs), NOD-like receptors (NLRs), RIG-like helicases (RLHs)] recognizing a particular group of pathogens expressing a unique set of pathogen- or danger-associated molecular patterns (PAMPs/DAMPs). These stimuli are also able to modulate the intracellular mechanisms of cross-presentation, emphasizing the significance of different adjuvants used in different studies. In addition, the surrounding cellular and soluble factors in the micro-milieu can significantly alter the cross-presenting potential of DCs. The effects of all these modulating factors are concordantly dependent on the timing relative to antigen processing by the DCs and thereby affect the outcome for T cell activation in different phases of the immune response.

Different DC subsets express a distinct endocytic receptor repertoire that is dynamically regulated by stimuli. DCs exploit their endocytic receptors, such as the mannose receptor (recognizing OVA), to transport antigens into early endosomes in which subsequently the cross-presentation machinery will be recruited. The neonatal Fc receptor (FcRn) is essential for uptake (and thus presentation) of immune complexes by CD11b⁺ DCs. The same is true for OVA-coated beads, which are only cross-presented when first opsonized with IgG [54]. Similarly, human blood BDCA3⁺ DCs induce strong T cell activation upon internalization of soluble protein, indicative for a potent cross-presentation machinery, whereas pDCs hardly induce T cell activation [90]. The absence of cross-presentation of soluble protein by pDCs is not unexpected because an earlier study showed that pDCs inefficiently take up soluble proteins in a receptor-independent fashion. By contrast, pDCs efficiently internalize and present antigens from immune complexes to CD4⁺ T cells [101]. When antigens are targeted trough the C-type lectin receptor dendritic cell immunoreceptor [DCIR; C-type lectin domain family 4 member A (CLEC4a)], all human DC subsets tested, including ex vivo generated DCs, skin-isolated LCs, and blood myeloid DCs and pDCs, are able to cross-present antigens and activate CD8⁺ T cells [15]. Thus, virtually all DCs have the machinery to cross-present antigens, provided that the antigen is offered to the DC in a suitable format under the appropriate conditions.

Stimulatory factors regulating DC cross-presentation

Each DC subset contains a restricted set of (inducible) PRRs that may have developed to react to a particular group of pathogens expressing a unique set of pathogen-associated molecular patterns. The recognition of such patterns by PRRs enhances peptide loading onto MHC class I molecules by the recruitment of the cross-presentation machinery to the endosomes [80].

How cross-presentation compares across antigens from diverse pathogens is unknown. The function of distinct human DC subsets in human LNs upon pathogenic infection is particularly unclear. It is uncertain whether the immune stimulatory features or other properties of pathogens are decisive in determining the functional characteristics of the DCs. The answer might not always be found in cell-activating molecules. Coexpression of LLO, a lysosome-disrupting hemolysin from L. monocytogenes, enhances cross-presentation of OVA E. coli [33], illustrating that specific proteins/enzymes enable access of antigens to the cytosol and thereby improve the cross-presenting capacity of DCs. This means that care should be taken in generalizing the capacities of DCs in immune responses against groups (intra- or extracellular) of pathogens.

A large set of TLR ligands are known to act as adjuvants and stimulate cross-presentation [16,45,82,89,102], illustrating that the antigen-presenting capability of different DC subsets is dependent on the stimulatory agent used. The timing of stimulation relative to antigen encounter is another crucial factor. TLR stimulation shuts down the ability to internalize certain antigen uptake receptors [103,104] and may impair cross-presentation [41,105], and thus limit the clinical efficacy of otherwise potent adjuvants [106].

DCs do not work alone in combating infections, because the surrounding environment can alter DC cross-presentation and cross-priming potential. For example, type I interferons improve DC capacity to store and process exogenous antigens, leading to enhanced cross-presentation and activation of antigen-specific CD8⁺ T cell responses [107,108]. Other soluble factors or immune cells in the microenvironment can affect the immunological outcome. DCs in the gut are exposed to a completely different micro-milieu than DCs in the skin or to DCs that circulate in the blood. Therefore, the location and the micro-milieu largely determine the cross-presenting potential of DCs. BDCA3⁺ DCs in blood only cross-present when matured by R848 [14] or poly I:C stimulation, whereas BDCA3⁺ DCs that have differentiated further and migrated to the skin are already cross-presenting without additional stimulation [99]. A challenge is to identify checkpoints that regulate DC function.

Timing of antigen entry and stage of immune response

The timing of antigen entry can affect DC cross-presenting capacity. For example, DCs harvested from mice 18 h after antigen challenge showed cross-presentation activity in the CD8α⁺ and dDCs subsets, whereas CD8α⁺ DCs were unable to prime CTLs when isolated 36 h after stimulation [39]. Similarly, the capacity of DCs in brachial LNs at day 2 (primary infection site) and axillary LNs at day 6 (secondary site) to cross-present herpes simplex virus-1 antigens to gBT-I cells ex vivo was examined. Only CD8α⁺ DCs presented antigens at both time points, whereas migratory CD103⁺ DCs had cross-presentation capacity only 6 days post-infection [36]. Immunization in the presence of saponin-based adjuvants led to a dominant role for CD8α⁺ DCs after 12 h, whereas migratory DCs joined in and efficiently cross-presented antigens after 24 h, and even after 48 h [40]. Thus, different DC subsets contribute to cross-priming at different time points during the initiation of an immune response.

The phase of the immune response should also be considered when studying cross-presentation. The finding that pDCs inefficiently take up soluble proteins, but do take up protein immune complexes suggests that pDCs are not first in line for MHC class I presentation of soluble proteins, but play a role in the second line of defense when the immune system has generated specific antibodies against this protein. In the initial phase of the immune response, pDCs produce large amounts of type I interferons that not only signal foreign invaders but also trigger antigen cross-presentation by cDCs [108]. This may be a general phenomenon that does not solely account for pDCs or Fc receptors. Therefore, in addition to the antigen type and stimulatory environment, timing and phase of the immune response are crucial for determining effective cross-presentation.

Concluding remarks

Available data suggest that all the classically characterized DC subsets have the ability to cross-present exogenous antigens. The type of antigen and presence and timing of inflammatory signals and other components of the microenvironment that program DC differentiation and activation are decisive in determining which DC subsets become dominant and sometimes indispensable for cross-presentation. Understanding the functional reprogramming of distinct DC subsets under different inflammatory conditions should provide further insight into the plasticity and exclusivity of functional specializations of DC subsets, including their ability to cross-present exogenous antigens. Knowledge of how to manipulate antigen-presentation pathways will be instrumental for vaccine adjuvant development and should ultimately lead to effective vaccination strategies against cancer and infectious diseases.