Innate lymphocyte and dendritic cell cross-talk: a key factor in the regulation of the immune response

Abstract

Dendritic cells (DC) are specialized in the presentation of antigens and the initiation of specific immune responses. They have been involved recently in supporting innate immunity by interacting with various innate lymphocytes, such as natural killer (NK), NK T or T cell receptor (TCR)-γδ cells. The functional links between innate lymphocytes and DC have been investigated widely and different studies demonstrated that reciprocal activations follow on from NK/DC interactions. The cross-talk between innate cells and DC which leads to innate lymphocyte activation and DC maturation was found to be multi-directional, involving not only cell–cell contacts but also soluble factors. The final outcome of these cellular interactions may have a dramatic impact on the quality and strength of the down-stream immune responses, mainly in the context of early responses to tumour cells and infectious agents. Interestingly, DC, NK and TCR-γδ cells also share similar functions, such as antigen uptake and presentation, as well as cytotoxic and tumoricidal activity. In addition, NK and NK T cells have the ability to kill DC. This review will focus upon the different aspects of the cross-talk between DC and innate lymphocytes and its key role in all the steps of the immune response. These cellular interactions may be particularly critical in situations where immune surveillance requires efficient early innate responses.

Introduction

Dendritic cells (DC), characterized as the most potent antigen-presenting cells (APC), represent a multi-functional population of cells (Table 1). Conventional DC subsets described in humans include myeloid DC (mDC) CD11c⁺ BDCA1⁺ and plasmacytoid DC (pDC) CD11c⁻ BDCA2⁺ (CD303⁺) whereas, in mice, the corresponding subsets are CD11c⁺ CD8a⁻CD11b⁺ and CD11c^loB220⁺Ly6C⁻CD11b⁻ respectively [1–3]. The two subsets differ in their expression of highly conserved microbial pattern recognition receptors, known as Toll-like receptors (TLR), but both are able to induce the stimulation of naive T cells. In mice, a small population of mDC can express CD8 [2]. These splenic CD8⁺ DC are able to cross-present antigen to T cells in vitro, in contrast to CD8⁻ DC [4]. No equivalent CD8⁺ DC subpopulation has been described in humans, but Langerhans cells (LC), a particular DC population in the skin of mice and humans, also display good cross-presentation capabilities [5]. In steady-state conditions, DC are in an immature stage and induce tolerogenic T cell responses [6,7]. In an inflammatory microenvironment, upon ligand recognition by TLR, maturation process occurs and DC migrate to the lymph nodes where productive immune responses are induced [8,9]. Recent reviews highlighted the role of inflammation signals on DC differentiation [10] and on the pro- or anti-inflammatory effects of DC [11,12].

Table 1.

Populations of DC in mice and humans.

DC subsets	Tissue distribution	Species	Immature phenotype	Mature phenotype
Bone marrow derived DC	Dermis, airways, intestine, thymus, spleen, liver, lymphoid tissue	Mouse	CD11c+CD8α− CD11b+MHC-II+ TLR-1–3+/−TLR-2,4–9+	CD83+ CCR7+ CD80++ CD86++ MHC-II++
		Human	CD1a+ CD14−CD11c++CD11b++ CD1c+ CD209+ MHC-II+ TLR-1, 6+,3,8++	CD40+
Plasmacytoid DC	Lymphoid organs, liver, lung, skin	Mouse	CD11c+/−B220+Ly−6C+CD11b− PDCA+ MHC-II+ TLR-2–9+
		Human	CD14−CD11c−CD123++ BDCA2+ ILT7+ MHC-II+ TLR-7,9++
CD8α+DC	Thymus, spleen, lymph node, liver	Mouse	CD8α+CD4−CD11c++ CD11b CD205++-TLR-2–4,6,8,9+
		Human	Not identified
Langerhans cells	Mucosal epithelia, epidermis	Mouse	CD8α− CD11c+ CD205++ E-cadherin+CD207+	E-cadherin +/−
		Human	CD14+/−CD11c+CD1a+ E-cadherin+CD207+CCR6+

DC, dendritic cells. ^+/−, low; ⁺⁺, high.

In addition to their role in induction of adaptive immune responses, DC also activate natural killer (NK) cells [13]. More recently, this DC activation was extended to other cell types such as NK T or T cell receptor (TCR)-γδ cells [14,15]. Moreover, certain DC subsets share common developmental pathways with NK cells, suggesting that these cells could influence each other during differentiation [16–18].

Natural killer, NK T and TCR-γδ lymphocytes constitute particular populations of lymphocytes which play important roles in innate immunity and share similar functions upon activation, such as expansion, secretion of soluble factors (cytokines, chemokines) and cytolytic activity [19–22].

The functions of NK cells are the result of activating and inhibitory signals received from their receptors (Table 2), which can sense viral infections, as well as modified self-antigens. In contrast to cytolytic T cells, these cells do not require specific antigen recognition to acquire the ability to kill target cells. Two different functional subsets of NK cells can be distinguished in humans regarding the expression of CD56 molecule, NK CD56^low-non-cytolytic and NK CD56^high-cytolytic respectively [23]. Mouse NK cells do not express CD56 and were subdivided recently according to CD27 expression [24]. After activation, which can be dependent upon DC, NK cells proliferate, produce interferon (IFN)-γ and acquire cytolytic activity.

Table 2.

Innate lymphocyte receptors.

	Species	Ligands	Function	NK NKp46+ CD3–	NK T Particular Vα TCR	γδT TCR-γδ
KIR	h	HLA-A,B,C allotypes	Activation and inhibition	+	+	+
Ly49	m	MHC class I	Activation and inhibition	+	+	Subset
NKG2A-C/CD94	h	HLA-E	Inhibition	+	?	+
	m	Qa-1	Inhibition	+	?	?
NKG2D	h	MICA/B, RAET	Activation	+	?	+
	m	RAET	Activation	+	?	?
NCR (NKp30,44)	h	Virus?	Activation	+	?	Subset
CD16	h, m	Immune complexes	Activation	+	–	–
CD161	h, m	Clr-g, Clr-b	Activation and inhibition	+	+	?
CD27	h, m	CD70	Activation	+	?	?

HLA, human leucocyte antigen; MICA/B, major histocompatibility complex class I-related; NK, natural killer; TCR, T cell receptor, MHC, major histocompatibility complex; h, human; m, mouse; RAET, retinoic acid early transcript.

Natural killer T lymphocytes are T lymphocytes expressing several NK markers (Table 2) and bearing an invariant TCR (Vα14Jα28 gene segments in mice and Vα24JαQ in humans). Both self and microbial glycolipids presented on non-polymorphic CD1d molecules can induce NK T activation [25]. Upon activation, they produce type 1 as well as type 2 cytokines and display cytotoxic activity [21]. Therefore, NK T cells are involved in a large number of immune responses such as autoimmunity [26] and immunity against viral [27], bacterial [28], fungal [29], parasitic pathogens [30] and tumours [31].

T cell receptor-γδ T cells differ from the classical αβ T lymphocytes by a TCR with unique structural and antigen-binding features [20], that endows them with independence of any APC for the recognition of foreign epitopes [32]. Each tissue has its own specific subset of TCR-γδ cells, according to the variable (V) gene used to generate the TCR [33]. Various populations of TCR-γδ cells reside in the peripheral tissues of mice, including the skin, gut and uterus. However, in humans, tissue-specific TCR-γδ cells are not as prominent and the adult human TCR-γδ cell repertoire is dominated by a polyclonal population bearing the Vγ9Vδ2 TCR. TCR-γδ cells recognize phosphorylated isoprenoid precursors and alkylamines, which are conserved in the metabolic pathways of many species including plants, pathogens and primates [34]. In the periphery, TCR-γδ cells express mainly the Vδ1 TCR [35]. TCR-γδ cells are key players in innate immunity and, arguably, the most complex and advanced cellular representative of the innate immune system [36].

However, a clear distinction between these innate lymphocyte populations is difficult, because of overlapping of some specific markers. Expression of NK1·1, NKG2D, Ly49, 2B4 and NKp46 receptors were detected on TCR-γδ cells, while non-rearranged germline TCR-δ locus transcripts were present in mature NK cells [37]. Similarly Vα14 activated NK T cells can down-regulate CD3/TCR, which renders them indistinguishable from NK cells [38].

The interactions between DC and innate lymphocytes have been documented both in mice and humans. This review will focus upon the different aspects of this cross-talk and its key role in all steps of an immune response, such as differentiation, maturation, effector stage and control.

Innate lymphocyte ability to modulate the differentiation of DC

It has been suggested that DC differentiation can be influenced by interactions with other immune cells, such as NK cells, in the inflammatory environment of rheumatoid arthritis joints [39,40]. In mice, NK cells are able to inhibit myelopoesis in autologous host [41,42]. Moreover, NK depletion resulted in increased numbers of myeloid precursors in the spleen, but not in the bone marrow, suggesting that NK cells serve as regulators of myelopoesis [41,43]. Soluble factors such as granulocyte–macrophage colony-stimulating factor (GM-CSF) and interleukin (IL)-3, which are produced mainly by resting NK, stimulate colony formation, while tumour necrosis factor (TNF)-α, secreted after NK activation, exerts inhibitory effects [44].

Most studies focused upon the role of NK cells in the maturation of immature DC differentiated in vitro in the presence of cytokines [45–47]. In contrast to IL-4 differentiated cultures, DC derived from monocyte-enriched peripherasl blood mononuclear cells, in the presence of IFN-α, require NK cells for complete maturation as shown by CD83, major histocompatibility complex (MHC) class II and co-stimulatory molecules up-regulation and priming of efficient antitumour CD8⁺ T cell responses [48].

A recent study investigated the role of NK cells in an early stage of DC differentiation from CD14⁺ monocyte precursors. Co-culturing these cells with autologous NK cells in the presence of IL-15 induced morphological and phenotypic changes associated with DC, such as down-regulation of CD14, up-regulation of CD40, CD80, CD86, DC-SIGN, DEC 205 and antigen uptake. The process requires cell–cell contact between NK and monocytes, as well as soluble factors, such as GM-CSF and CD40L (CD154), produced by NK cells. CD3⁻ KIR⁻ CD56^bright NK cells were shown to be more efficient in inducing DC differentiation when compared with CD56^dim counterparts, suggesting that the process might occur in lymph nodes. Supporting the hypothesis that this process occurs in vivo, synovial fluid from patients with rheumatoid and psoriatic arthritis (containing IL-15) was able to induce DC differentiation only in the presence of freshly isolated autologous NK cells. Therefore, NK have been proposed as crucial for maintaining inflammatory diseases by providing a milieu for monocytes to differentiate towards DC [40].

Dendritic cell-innate lymphocyte cross-talk in the activation of the immune response

The role of DC-innate lymphocytes cross-talk in the activation of the immune response is illustrated in Fig. 1. The first evidence for innate immunity stimulated by DC was provided by the work of Fernandez et al. [13], which showed a direct activation of NK cells by DC in vivo. Subsequently, NK cell activation in vitro has been documented by using a variety of mouse or human DC [19]. Most studies focused upon the interactions between NK cells and mDC and only recently upon the interaction with pDCs [49]. Monocytes-derived DC appear as the most potent stimulators of NK cell proliferation and cytotoxicity. Human dermal DC have intermediate effects and LC are able to activate NK cells only in the presence of exogenous IL-2 and IL-12 [50]. Moreover, mDC and pDC isolated from peripheral blood are both able to activate NK cells after stimulation with different TLR ligands [51,52].

Fig. 1.

Schematic representation of dendritic cell (DC)-innate lymphocyte cross-talk in the activation of the immune response. pDC, plasmacytoid DC; IL, interleukin; IFN, interferon; TNF, tumour necrosis factor; TLR, Toll-like receptors; mDC, myeloid DC; NK, natural killer; NKT, natural killer T; TCR, T cell receptor; HLA-DR, human leucocyte antigen-DR.

The DC-induced activation of resting NK cells in vitro requires direct cell contact resulting in a polarized secretion of pre-assembled stores of IL-12 by DC towards NK cells within the synapse between both cell types [53]. In synergy with IL-12, IL-15Rα expression on DC was shown to be required for NK cells priming [54]. Subsequently, the DC/NK interactions were found to be multi-directional, involving not only cell–cell contact but also soluble factors [49]. Cytokines such as IL-2, IL-18 and type I IFN play crucial roles in this cross-talk [55–57]. These results suggest that different DC subsets activate NK cells at several steps of the immune response, with the pDC promoting early activity of NK cells followed by mDC. Early release of IL-12 by pathogen-activated DC could be the link between NK/mDC and NK/pDC interactions [52]. Conversely, NK can activate DC optimally in vitro. This process requires both cell–cell contact, mediated via the receptors NKp30, KIR or NKG2A [58], and TNF-α production [47].

Similar to NK/DC cross-talk, bidirectional interactions take place between DC and NK T. Responses of NK T cells to synthetic α-GalCer presented by CD1d molecules on DC requires IL-12 production by DC and CD40/CD40 ligand-mediated cell contacts [59]. Conversely, CD40 signalling and the release of TNF-α and IFN-γ induce DC maturation [60]. Despite the expression of CD1d, NK T activation by mDC does not occur unless the DCs are activated through TLRs [61,62]. DC maturation induced by TLR stimuli is enhanced by NK T cells [63,64]. pDC stimulated through TLR-9 induce NK activation markers and licence these cells to respond to antigen loaded mDC [65]. These data suggest that NK T cells interact with DC subsets (mDC and pDC) and contribute to the outcome of immune responses after pDC activation.

A mutually co-stimulatory relationship was also found between TCR-γδ cells and DC. Activation of TCR-γδ cells by phosphoantigens induces a significant up-regulation of CD86 and MHC class I molecules and the acquisition of functional features typical of mature/activated DC [66]. Cell–cell contacts are required for stimulation with aminophoshonate, but not with phoshomonoesters. In return, DCs induce CD25 and CD69 up-regulation on TCR-γδ cells, as well as IFN-γ and TNF-α secretion by TCR-γδ cells [66]. Moreover, TCR-γδ cells enhance DC maturation induced by TLR-2 and TLR-4 ligands [67,68]. Another study showed that freshly isolated TCR-γδ cells were as efficient as those activated by phosphoantigens in inducing maturation of DCs and IL-12 production by DC [69].

Dendritic cell-innate lymphocytes cross-talk is exploited to achieve reciprocal full activation, with DC maturation being a sensor that links innate immune responses and adaptive ones (Fig. 1). The final outcome of these cellular interactions may have a dramatic impact on the quality and strength of the down-stream immune responses.

Dendritic cell-innate lymphocyte cross-talk in the induction and effector phases of adaptive immune responses

Dendritic cells and innate lymphocytes in effector phases of the immune responses are summarized in Fig. 2. DC maturation by innate lymphocytes is relevant in early responses to tumour cells and certain viruses, when the absence of inflammation and pathogen-associated molecules does not result in DC maturation and subsequent antigen presentation [51,70]. In addition, the interaction between mature, but not immature DC, results in innate lymphocyte proliferation, IFN-γ production and cytolytic activity against tumours, virus or pathogen-infected cells [13,45]. Therefore, these interactions play an important role in the induction of early immune responses to tumours, pathogens and certain viruses, before an adaptive immune response can be developed. Moreover, this cross-talk bridges innate and adaptive immunity, as it affects the magnitude and polarization of T cell-mediated responses.

Fig. 2.

Schematic representation of dendritic cell (DC)-innate lymphocyte cross-talk in the effector phase of the immune response. IFN, interferon; IL, interleukin; TLR, Toll-like receptors; MHC, major histocompatibility complex.

In lymph nodes, NK cells may play an important role in the initiation of specific T cell responses by influencing DC maturation. Indeed, under in vitro conditions where DCs are activated suboptimally with type-I IFN, NK cells licence DC to prime T cell responses [48]. Consequently, mature DCs activate NK to produce IFN-γ, which is required for T helper 1 (Th1) polarization [71,72]. In some cases, T cell-mediated tumour rejection is dependent upon DC activation by NK cells [73]. Furthermore, Adam et al. [74] reported that NK cell–DC cross-talk may bypass the Th arm in CTL induction against tumours expressing NKG2D ligands. A critical role for NKG2D-mediated NK–DC interaction in generating robust CD8⁺ T cell immunity against intracellular parasites was also documented [75].

Antigen-specific, IFN-γ-producing CD4⁺ and CD8⁺ T cells could also be induced down-stream of interactions between NK T and antigen-capturing DC. These data suggest that DC activated by a population of innate lymphocytes, such as NK T cells, can activate another group of innate lymphocytes, such as NK cells, to induce anti-tumour effects [59,63]. TCR-γδ are also activated by DC infected in vitro by the bacillus Calmette–Guérin, as indicated by elevated CD69 expression on their cell surface, IFN-γ secretion and cytotoxic activity. Consequently, DC-stimulated TCR-γδ cells help DCs to prime a significantly stronger anti-mycobacterial CD8 T cell response [14]. TCR-γδ cells activated by the synthetic phosphoantigen bromohydrin pyrophosphate induce the production of IL-12 by DC, an effect involving IFN-γ production. The relevance of this finding to DC function was demonstrated by the increased production of IFN-γ by alloreactive T cells when stimulated in a mixed leucocyte reaction with DC preincubated in the presence of activated TCR-γδ cells. These data suggest that TCR-γδ cell activation results in DC maturation and enhanced TCR-αβ T cell responses [69].

Moreover, DC and innate lymphocytes share similar functions. As well as their APC function, DC may acquire cytotoxic properties [76,77], for example in response to stimulation with type I IFN [78]. In fact, human DC can induce in vitro growth arrest and apoptosis of tumour cells [77,79,80]. DC able to kill tumour cells have also been described in rat models [81]. Cytotoxic DC activity seems to be independent of Fas-associated death domain but dependent upon caspase-8 [77], whereas the inhibiting effect of DC on tumour proliferation is likely to be caspase-8 independent and does not require cell contact [80].

Last year, Taieb et al. [82] described a new subset of murine DC expressing NK markers and producing large amounts of IFN. These cells killed cells lacking self-MHC molecules in a TNF-related apoptosis-inducing ligand-dependent manner and they have been named ‘IFN-producing killer dendritic cells (IKDC)’[83]. They have been described as a distinct cell population because they differ in their developmental origin, because Rag2^−/−IL2rg^−/− mice lack canonical NK cells but possess functional IKDC in the spleen [82]. However, very recently several groups have shown that these cells were, in fact, activated NK cells [84–86].

Other types of DC-bearing NK cell receptors have been identified, such as bitypic cells expressing both CD11c and the NK cell marker DX5 in the context of lymphocytic choriomeningitis virus infection [87]. In a mouse autoimmune diabetes model, treatment with CD40L induces the presence of this bitypical NK/DC regulatory cell population [88]. These cells can kill NK sensitive target cells and present ovalbumin antigen [88].

On the other hand, activated NK cells could become potent APC [89]. After activation, NK cells can up-regulate MHC class II, CD80, and CD86 molecules and acquire independent unique mechanisms of antigen capture and presentation, involving activating receptors such as NKp46, NKp30 and NKG2D [89]. NK cells may also acquire functional APC-like properties after target-cell killing [89]. Studies on the T cell-activating potential of human NK cells in different clinical conditions revealed that a proinflammatory, but not immunosuppressive, microenvironment can up-regulate T cell-activating molecules on NK cells [89]. Even a subtype of TCR-γδ cells (Vδ2⁺ T cells) upon microbial activation can display antigen and provide enough co-stimulatory signals to induce a strong naive αβ T cell proliferation and differentiation [90]. TCR-γδ cells also express DC activation markers, such as MHC class II, CD80 and CD86 [91]. Expression of MHC class II and ligands for T cell co-stimulatory molecules is not a guarantee for APC capability as eosinophils, for example, express significant levels of MHC class II and CD86 on their surface after activation but cannot process antigen [92]. The APC function of TCR-γδ cells seems to be associated with the expression of CCR7, allowing their migration to lymph nodes. This expression is early, but transient [93], suggesting that the APC function of TCR-γδ cells may be more effective in the early stage of antimicrobial immune processes. The mechanisms controlling antigen uptake and processing in these cells are, however, still unknown and because in vivo studies are understandably difficult to perform in humans, there is no evidence that these cells function as APC.

Dendritic cell-innate lymphocytes in the control of immune responses

Recently, a novel function was assigned to NK cells, that involves them in ‘DC editing’. This process is mediated by NK aggression of normal immature DC. Only DCs undergoing optimal maturation become refractory to NK killing and are allowed to prime Th1 cells after migration to the lymph nodes [94]. In bacterial infections, interactions between NK cells and DC result in the rapid induction of NK cell activation and in the lysis of uninfected DC [95].

Among NK receptors, NKp30 seems to play an important role in the maturation and control of DC apoptosis, whereas the up-regulation of human leucocyte antigen-E (HLA-E) expression on DC protects them from NK lysis through the CD94/NKG2A receptor [96]. The interactions betweenNK cells and DC via NKp30 differ according to the ratio between NK cells and DC. A low NK/DC (1:5) ratio results in DC activation, whereas at a high (5:1) ratio, NK cells kill immature DC [45,47]. In contrast, pDC seem to display an intrinsic resistance to lysis by NK cells but exposure of pDC to IL-3 increases their susceptibility to NK cell cytotoxicity [49]. Until now, DC killing by NK cells in vivo has been demonstrated only in murine transgenic [97] or transplantation models, with no evidence so far that DC killing occurs in vivo under normal physiological conditions.

Like NK cells, NK T cells could also, under certain conditions, kill both immature and mature DC. These cells expressing inhibitory NK receptors are restricted by HLA-E molecules and are able to kill most NK-susceptible tumour cell lines [98]. Interestingly, in Leishmania infantum infections, immature DC up-regulate CD1d and are killed efficiently by NK T cells, whereas they are resistant to NK cell-mediated lysis because of an up-regulation of HLA-E expression which protects them through the inhibitory receptor CD94/NKG2A [99].

In conclusion, there is accumulating evidence that DC- innate lymphocyte interactions are multi-directional, playing important roles in differentiation, maturation, effector phases and control of immune responses. After activation, innate lymphocytes are able to induce DC maturation and therefore to shape both innate immune reactions in inflamed tissues and adaptive immune responses in lymph nodes. These cellular interactions may be particularly critical in situations where immune surveillance requires efficient early innate responses.