Chemokines and NK cells: regulators of development, trafficking and functions.

NK cells are innate lymphocytes capable of killing malignant or infected cells and to produce a wide array of cytokines and chemokines following activation. Chemokines, play critical roles in the regulation of NK cell tissue distribution in normal conditions as well as their rapid recruitment to the parenchyma of injured organs during inflammation, which is critical for NK cell ability to promote protective responses. In this regard, differences in chemokine receptor expression have been reported on specialized NK cell subsets with distinct effector functions and tissue distribution. Besides their role in the regulation of NK cell trafficking, chemotactic molecules can also affect NK cell effector functions by regulating their priming and their ability to kill and secrete cytokines.

NK cell differentiation, tissue distribution and function

Natural killer (NK) cells represent a discrete subset of innate lymphocytes that act as first line of defense against several microbial infections, early cellular transformation and tumor growth. NK cell prompt response is mainly related to their ability to release cytolytic mediators such as perforin and granzymes or to express ligands able to trigger death receptors on target cells; moreover they can also secrete a wide array of cytokines and chemokines to recruit and to instruct other immune cell types [1], [2]. The main site of NK cell differentiation is the bone marrow (BM) where they develop from a common lymphoid progenitor that generates also B and T lymphocytes. Notably, NK cell precursor has been found also outside the BM, thus suggesting that NK cell development can occur also in the peripheral tissue microenvironment [3], [4].

During development, NK cells have to pass through several stages of differentiation in order to progressively acquire a mature phenotype consisting of both activating and inhibitory receptors as well as adhesion molecules and chemotactic receptors, and effector ability. Among growth factors critically involved in NK cell differentiation, a crucial role is played by IL-15 and its receptor. IL-15 needs to be trans-presented by cells expressing IL-15Ralpha chain and was found on the surface of several hematopoietic or non-hematopoietic cells such as dendritic cells or stromal cells both in the BM and in peripheral tissues where it promotes the generation of NK cells from their precursors and the differentiation of immature NK cells and represents a survival signal for mature NK cells [5], [6]. In regard to human NK cell development, there are evidence that human CD34+ hematopoietic stem cells and hematopoietic progenitors, including NK cell precursors [7], are present within the BM, and that co-culture of BM-derived CD34+ cells with BM stromal cells promotes the differentiation of cytotoxic NK cells in vitro [8]; however, unlike the mouse, the identification of different stages of human NK cell differentiation within the BM is still lacking. By contrast, at least five distinct progressive stages of NK cell development have been described in the secondary lymphoid tissues such as lymph nodes (LNs) and tonsils, based on the expression of distinct set of markers and functional capability: stage 1, CD34+CD45RA+CD117−CD161−CD94−; stage 2, CD34+CD45RA+CD117+CD161+/−CD94−; stage 3, CD34−CD117+CD161+NKp46−CD94−; stage 4, CD34−CD117+/−NKp46+CD94+CD16−CD56brigh; and stage 5, CD34−CD117−NKp46+CD94+/−CD16+CD56dim. From stage 1/pro-NK cells to stage 2/pre NK cells and stage 3/immature NK (iNK) cells, these cells become committed to the NK cell lineage; from stage 3 to 5, they undergo functional maturation in that iNK cells can produce GM-CSF and potentially type 2 cytokines, while stage 4/CD56bright NK cells preferentially release IFN-gamma, and stage 5/CD56dim NK cells are the principal cytotoxic population. Within stage 4 and 5, NK cells also acquire the killer inhibitory receptors (KIR) [4], [9].

Similarly to human, mouse NK cell development involves distinct stages that can be identified by multi-parameter cell surface marker analysis and that comprise functionally distinct NK cell subsets. Precursor cells committed to the NK cell lineage (pNKs) express the common beta chain of the interleukin (IL)-2 and IL-15 receptors (CD122) and lack surface proteins associated with mature hematopoietic cells (referred to as Lineage negative or Lin−), including most NK cell receptors with the exception of NKG2D and 2B4 (CD244) [3]. Generation of these cells does not require IL-15, but responsiveness to this cytokine is critical for later steps of the differentiation process [6]. As pNKs mature they progressively acquire a number of NK cell receptors, including NKR-P1C and NKp46. At this stage, the cells initiate expression of the major histocompatibility complex (MHC) class I-recognition receptors encoded by CD94 in combination with NKG2A/C/E, and the Ly49 genes, which encode a family of C-type lectin receptors. Subsequently, the cells up-regulate the alpha2/CD49b integrin (DX5) and acquire cytolytic and cytokine producing capacities. Within DX5+ NK cells, two populations with different functional competence can be identified by the level of expression of the CD11b integrin chain (CD11blow and CD11bhigh). A better characterization of functionally distinct NK cell subsets is provided by the analysis of CD27, a costimulatory molecule belonging to the TNF receptor superfamily, that is downmodulated during CD11bhigh NK cell differentiation. Thus, three distinct subpopulations that are developmentally related can be identified: CD11blow/CD27high, CD11bhigh/CD27high, that display the higher effector capacity and CD11bhigh/CD27low, that represents a subsequent stage of maturation [10] The latter population can also be identified by KLRG1 expression, a C-type lectin inhibitory receptor [11]. Recent observations suggest that CD27high to CD27low transition may occur in the spleen and is driven by splenic monocyte-presented IL-15.

Ablation of BM hematopoiesis irreversibly alters NK cell development. Nevertheless, NK cells from several extramedullary tissues display tissue specific phenotypes suggesting that maturation can be completed in organ different from BM including thymus, spleen, LNs and liver. In mice, the identification of immature NK cells in the liver that express the effector molecule TRAIL and the recent identification of thymus-derived GATA-3+ CD127+ NK cells are in agreement with the prediction of an organ-specific maturation of NK precursors [12]. Differently from BM-derived NK cells, thymic-derived NK cells preferentially home to LN, where they maintain CD127 and GATA-3 expression and display stronger cytokine production and less cytotoxic ability than their splenic counterpart. These functional characteristics are similar to the CD56hiCD16− human NK cells subset, which was also found to be CD127+ and to express Gata-3.

Mature NK cells predominantly circulate in the peripheral blood, but are also resident in several lymphoid and non-lymphoid organs, such as the spleen, tonsils, LNs, liver, lungs and intestine and uterus [13]. In all these organs, NK cells have been found in close proximity of vasculature, and in most instances in areas distinct from those of T or B cells. In the spleen, NK cells are primarily present in the red pulp where they co-localize with endothelial cells suggesting that splenic NK cells are mainly distributed within blood sinuses or vessels [14], [15], [16]. Also in mouse BM a fraction of NK cells, comprising mostly mature cells, was found to reside in sinusoids [17]. Using immunohistochemical analysis and dynamic intravital imaging of the LNs, it has been shown that NK cells reside in the medulla, in the perifollicular regions as well as in the paracortex where they co-localize with dendritic cells (DC). In a steady state condition, the density of LN NK cells is higher in the medulla than in T or B cell areas and they are mainly located within lymphatic sinuses [18].

In humans, while the majority of blood NK cells are CD56dim, the predominant NK cell population found in the LNs and tonsil secondary lymphoid tissues is CD56bright. As above mentioned, CD56bright and CD56dim NK cell populations are quite different, in that they have a distinct inhibitory and activating receptor profile, display diversity in their adhesion molecules and chemokine receptor profile and thus they have different homing capability, and most importantly, they are endowed with unique functional ability being CD56bright the major source of cytokines and the CD56dim the major cytotoxic population.

Among non lymphoid organs, NK cells are selectively enriched both in liver where they are mainly situated in the lumen of the intrahepatic sinusoids and in uterus, where they are located around endometrial glands and vessels [19], [20]. Recently, a human CD56+CD3− cell population that is predominantly located in the mucosa surrounding the lymphoid follicles of tonsils and Peyer's Patches of the ileum and appendix have been described and referred as NK-22 cells based on its ability to release IL-22 [21]. The human NK-22 cells, express the NKp44 receptor and the chemokine receptor CCR6 and seem to have important roles in protective immunity.

In the mouse intestine, CD3−NKp46+ cells can be divided into two main subsets based on the expression of IL7Ralpha (CD127) and NK1.1 [22]. Importantly, these subsets appear to mediate different functions because IL7Ralpha−NK1.1+ cells are similar to conventional NK cells, producing perforin and IFN-gamma, whereas IL7Ralpha+NK1.1− cells lack these effector functions. Instead, they express IL-22 and RORgamma and, in contrast to conventional NK cells, do not require IL-15 for differentiation. [22], [23], [24]. The origin and lineage of gut mucosa associated human NKp44+ cells and murine NKp46+ cells is still debated and it is still unclear if they are developmentally related to the NK cells and/or to the lymphoid tissue inducer (LTi) cells or if they represent a distinct innate immunity cell types [25], [26].

Tissue resident NK cells can perform many functions including cytotoxicity or cytokine release mostly related to maintain tissue integrity, but in some cases, i.e. uterine NK cells, they can make a more specialized function by modulating vessel behavior, promoting vascular remodeling or influencing the recruitment and activation of other leukocytes to ensure a good outcome of pregnancy.

Although, in steady-state conditions, mature NK cells can be found in some lymphoid and non lymphoid organs, following an insult such as a viral infection, inflammation or tumor growth, specialized NK cell subsets can be rapidly recruited from blood and accumulate in the parenchyma of injured organs to perform their specific function and participate to the restoration of tissue integrity.

Chemokine receptor expression by NK cells: impact on NK cell function

A large body of evidence indicate that NK cells can express several receptors for CXC, CC, C and CX3C chemokines, with great heterogeneity in the chemokine receptor repertoire among different NK cell populations and between resting versus activated NK cells [27], [28], [29].

It has been previously reported that peripheral blood human NK cells express both CXCR1 and CXCR2 as CXCL8 (IL-8) receptor [30], [31], [32] and CX3CR1 as CX3CL1 (fractalkine) receptor [33]. These observations have been further extended by Campbell et al. who provided the first evidence that distinct (CD56posCD16neg and CD56posCD16pos) peripheral blood NK cell subsets have a unique repertoire of chemokine receptors [34]. CD16pos, that are bona fide CD56low, NK cells uniformly express high levels of CXCR1 and CX3CR1, low levels of CXCR2 and CXCR3 and no detectable levels of CXCR5. By contrast, CD16neg (mostly CD56bright) NK cells express high levels of CXCR3, CCR5 and CCR7, the latter molecule mainly involved in the homing of lymphocytes to secondary lymphoid organs and low levels of CX3CR1, and are negative for CXCR1, CXCR2 and CXCR5; moreover, both NK cell subsets express high levels of CXCR4, the receptor for CXCL12 (SDF-1alpha/beta). The differences in chemokine receptor expression correlate with differences in the migratory behavior, being CD16pos NK cells more responsive to CXCL8 and CX3CL1 and the CD16neg cells to CCL19 and CCL21 as well as to CXCL10 and CXCL11. Both NK cell subsets strongly migrate in response to the ligand of CXCR4, CXCL12 [34], [35]. Besides chemokine receptors, human and mouse NK cells were also shown to express receptors for chemotactic molecules that do not belong to the chemokine superfamily, and can regulate NK cell trafficking under steady state and inflammatory conditions. In this regard, NK cells were shown to migrate in response to the pro-inflammatory plasma protein chemerin, and CD56lowCD16pos, but not CD56highCD16neg NK cells express its receptor, ChemR23. Sphingosine 1-phosphate (S1P) is a sphingo-phospholipid generated by the conversion of sphingomyelin into ceramide by sphingomyelinase, that influences lymphocyte trafficking as well as proliferation, adherence and morphogenesis [36]. Previous reports show that human NK cells express the mRNA for S1P1, while mouse NK cells express both S1P1 and S1P5 and migrate in response to S1P [37], [38]. Interestingly, the selective expression of CXCR3 on CD56high and of CX3CR1 on CD56low NK cells is very similar to their expression pattern on the two main subsets of mouse mature NK cells defined as CD27high and CD27low. Indeed, as compared to CD27low NK cells, CD27high (both CD11blow and CD11bhigh) cells have higher expression of the chemokine receptor CXCR4, selectively express CXCR3 and have much lower expression of the chemoattractant receptor S1P5 and CX3CR1 [28], [38]. Moreover, the recent identification of chemokine receptor expressing mouse NK cell subsets with distinct effector functions suggests that discrete subsets of NK cells can be defined according to their chemokine receptor expression profile [17], [39], [40].

The expression of chemokine receptors and the corresponding NK cell chemotactic response is also modulated upon cytokine stimulation. A significant decrease of CXCR3 expression and chemotaxis to CXCL10 was reported in human NK cells treated for 6 or 24 h with IL-2 and IL-12, used alone or in combination. IL-2 down-regulated CCR7 receptor expression on NK cells whereas induced the CCR4 and CX3CR1 molecules [41]. Stimulation for longer periods of time with IL-2 resulted in increased expression of CCR1, CCR2, CCR4, CCR5 and CCR8 [42], while expression of CXCR2 and ChemR23 decreased [43]. CXCR4 and CX3CR1 have been shown to be down-regulated through stimulation with other cytokines, including IL-15, whereas TGF-beta has been proven to be a potent inducer of CXCR4 on NK cell subsets [42], [44]. In addition, treatment with IL-18 resulted in selective induction of CCR7 expression and enhanced response to CCL21 of the CD56low NK cell subset [45].

Besides migration, other human and mouse NK cell functions can be affected by chemokines. Indeed, chemokines can activate an NK cell defense machinery that may directly counteract an infectious agent by performing cytotoxicity or by secreting pro-inflammatory cytokines that recruit and activate other effector cells [46]. Soluble and membrane-bound CX3CL1 (fractalkine) induce IFN-gamma production by NK cells and affect NK cell ability to kill tumor cells both in vitro and in [47], [48]. Enhancement of NK cell degranulation has been also demonstrated in response to other chemokines belonging to the CC family and to CXCL10 [46], [49]. In addition, CC chemokines-mediated redistribution of adhesion molecules on NK cell surface may also increase NK cell cytotoxicity by promoting effector-target cell interaction [50].

CX3CL1 has been found to be a pivotal molecule driving immune synapse formation during DC/NK cell interaction leading to efficient NK cell activation. Importantly, the absence of CX3CR1, or CX3CL1-blockade, abrogated the ability of NK cells to produce IFN-gamma when exposed to DC, underlying a key role of this chemokine during DC-mediated NK cell priming [51]. Thus, the CX3CR1/CX3CL1 axis regulates NK cell functions at different levels promoting their migration through endothelial vessels, their activation and their ability to kill or to secrete cytokines.

Chemokine receptors and NK cell development into bone marrow

Lymphocyte development in the bone marrow and thymus requires repeated changes of their migration properties, which facilitate the sequential localization of individual precursors into selected microenvironments important for their maturation.

A number of studies show the important role of CXCL12/CXCR4 axis in the regulation of the positioning, retention and homing of hematopoietic cells into the BM [52], [53], [54], [55]. CXCL12/CXCR4 chemokine/receptor pair plays a crucial role also in B cell differentiation, mainly acting at the earliest stages both in fetus and adult mice [56]. CXCL12 deficiency has been associated with severely reduced number of early B cell precursors in fetal liver during embryogenesis. Also the adult BM of irradiated mice reconstituted with CXCR4-deficient fetal cells shows a marked reduction of B cell precursor, including pre-pro B cells that correlate with their enrichment in blood as compared to control mice [57]. Correspondingly, adventitial reticular cells that express high level of CXCL12 (CXCL12 abundant reticular (CAR) cells) were found to be in strict contact with pre-pro B cells [58]. Beside BM, CXCR4 play a role also during thymocyte differentiation being CXCR4 deficiency associated to a reduced number of thymocytes, likely due to the loss of mitogenic stimuli provided by CXCL12 [59].

The role of chemokines in NK cell development is less defined. It has been demonstrated that NK cells change their chemokine expression profile during development in BM. CXCR4 is highly expressed by pNK cells but its expression progressively decreases during development. The progressive decrease of CXCR4 parallels the increased expression of CXCR3, CCR1 and CX3CR1 on mature NK cells, being the expression of CX3CR1 mainly confined to the KLRG1+ subset that poorly expresses CXCR4 and does not express CXCR3 [17], [60]. Interestingly, expression of CX3CR1 on KLRG1+ cells was shown to define a more differentiated NK cell subset that is developmentally related to KLRG1+CX3CR1− NK cells [17]. The role of CXCR4 receptor on NK cell retention in BM was analyzed by in vivo administration of CXCR4 antagonist AMD3100, that induced the reduction of immature and mature NK cells in the bone marrow and their increase in spleen and peripheral blood, indicating that CXCL12 regulates the retention of selected NK cell subsets at various stages of maturation in the BM during development [60]. This is also supported by the evidence that retention of NK cells residing in BM sinusoids, that likely represent cells that have completed their maturation in BM and are ready to exit into systemic blood circulation, is CXCR4-independent [17]. Recently, Noda et al. showed that NK cell development is severely impaired in adult CXCR4 conditionally deficient mice and this was correlated with reduced number of NK cell precursors, decreased proliferation rate of CXCR4 deficient DX5− iNK cells and with the ability of CXCL12 to promote mature NK cell generation from lymphoid precursor and immature NK cells in the presence of IL-15 [61]. In addition, using chemokine receptor deficient mice it was evidenced that also CCR5 but not CCR1 can participate to NK cell differentiation and this was associated to regulation of developing NK cell proliferation under physiological conditions [62]. In regard to CXCR4, it was demonstrated that a fraction of CAR cells co-express IL-15 and IL-15Ralpha and thus CAR cells may trans-present IL-15 and provide CXCL12 to NK cells during development. The latter hypothesis is supported by immunofluorescence studies on BM tissue sections showing that NKp46+ NK cells can be found in close proximity to CAR cells in vivo. The observation that the residual fraction of NKp46+ NK cells that develop into BM in CXCR4-deficient mice is still close to CXCL12-producing cells suggested that CXCR4 is not required for NK-CAR cell co-localization. However, a role of CXCR4 in NK cell localization in BM microenvironments during normal development cannot be excluded as CXCR4-deficiency led to a strong impairment of NK cell development and to marked reduction of maturing NK cell subsets. In addition, CXCL12 may also be presented by other BM cells, including osteoblasts.

Overall, CXCR4/CXCL12 interaction may play a dual role during NK cell development: it may directly promote NK cell differentiation by acting on NK cell precursors and more immature DX5neg NK cells. Indeed, although their CXCR4 expression is higher as compared to more mature NK cells, ex vivo migration and in vivo retention of these cells into BM is less influenced by CXCL12, while in vivo proliferation of more immature NK cell is more dependent on CXCR4 expression [60], [61]. On the other hand, CXCR4/CXCL12 interaction is required to maintain immature and mature NK cells in BM parenchyma during maturation and thus indirectly can favor NK cell response to differentiation stimuli (Fig. 1 ).

Fig. 1

Hypothetical model of CXCR4 role on NK cell development during NK cell interaction with stromal cells. During development, high expression levels of CXCR4 on immature NK cells may potentiate their response to IL-15, directly favoring their proliferation and differentiation, while CXCR4/CXCL12-mediated retention may influence stromal cell-promoted differentiation in later differentiation stages. Down-modulation of CXCR4 coincides with NK cell subset exclusion from BM-parenchyma and enrichment in sinusoids and is likely to allow responsiveness to other chemoattractant including S1P and CCL3, allowing egress from BM.

In vivo regulation of NK cell functions by chemokines

NK cells express several chemotactic receptors that are involved in the control of NK cell migration across endothelium and in the correct tissue positioning of lymphocytes, but only recently we are beginning to appreciate the contribution of chemotactic factors on NK cell trafficking and tissue distribution under homeostatic and pathological conditions.

Although the differential expression of chemokine receptors and chemotactic responsiveness of the CD27low and CD27high subsets strongly suggests a correlation with their tissue distribution, at present few chemoattractant receptors have been demonstrated to play a role in NK cell subset distribution in vivo under normal conditions. A drastic decrease in NK cell numbers has been observed in the blood, spleen, and lungs of S1P5-deficient mice, associated with increased number of NK cells in the BM and LNs. This altered distribution has suggested that S1P5 provides an egress signal to NK cells, promoting their exit from BM and LNs [38]. In addition, the observation that CX3CR1-deficient mice display a selective reduction of DX5+CD3− cell number in the lung, strongly suggests that this receptor is important for the accumulation of at least one subset of NK cells in this organ under steady state conditions [63]. Maintenance of a subset of NK cell in the liver depends on the chemokine receptor CXCR6. Indeed, the ligand for CXCR6, CXCL16, is constitutively presented by the liver sinusoidal endothelium and CXCR6+ NK cells that represent a consistent fraction of liver NK cells, are drastically reduced in CXCR6−/− mice [40]. Also the expression of CXCR4 profoundly affects NK cell subset distribution likely by contributing to the maintenance of an NK cell pool within BM, as shown by the recruitment of BM iNK and mNK cells into circulation and spleen following in vivo delivery of the CXCR4 pharmacological antagonist AMD-3100 [60].

While the contribution of chemokines in the regulation of NK cell tissue distribution during homeostasis has been poorly addressed so far, a number of studies indicate that selected chemokines play a critical role in the orchestration of NK cell trafficking during inflammation. It has been reported that during murine cytomegalovirus infection, NK cells migrate through a CCL3-dependent mechanism to sites of liver infection where they contribute to antiviral defense [64]. This evidence was further supported by the demonstration that CCL3-deficient mice show decreased resistance to cytomegalovirus infection that is associated with a dramatic reduction of NK cell accumulation and IFN-gamma production in the liver. The recruitment of NK cells to the liver during infection required the receptor for CCL3, CCR5, and the receptor for CCL2, CCR2 [64], [65]. IFN-gamma-dependent recruitment of NK cells to liver has also been described in a model of Concanavalin A-induced hepatitis. In this case, NK cell entry in the liver was reduced in the absence of CCR1 and involved NK cells that used CXCR3 to exit the spleen [66]. CXCR6 was also shown to play a relevant role for the maintenance into liver of a population of CXCR6+ NK cells that retained antigen-specific memory after antigen challenge [40].

A role for CCL3 in recruitment of NK cells has also been demonstrated by intrapulmonary transient transgenic expression of this chemokine that resulted in increased Klebsiella pneumoniae lung clearance associated with NK cell activation and accumulation in this organ [67]. In addition, a specific defect of NK cell recruitment to pulmonary granulomas was observed in CCR1-deficient mice [68]. Another chemokine, CCL2 was shown to regulate accumulation of NK cells in the lungs of mice with invasive aspergillosis [69]. In addition, using an in vivo model of NK cell-mediated lung tumor cell clearance in CX3CR1-deficient mice or after treatment with blocking antibodies against CX3CL1 or CX3CR1, it has been demonstrated that decreased clearance of tumor cells following perturbation of CX3CL1/CX3CR1 interaction is attributable to defective NK cell recruitment to the lung [63], [70]. Using CXCR3-knockout mice, recruitment of NK cells in the lungs has been also shown to participate to the pulmonary host defence against Bordetella bronchiseptica [71].

In the central nervous system (CNS), CXCL10 has been reported to promote innate defence mechanisms following coronavirus infection by recruiting and activating NK cells, while CCR5 deficiency results in impaired NK cell ability to limit herpes virus spread to brain [72], [73]. CX3CR1GFP/GFP mice showed a selective deficiency of NK cell recruitment in the CNS during experimental autoimmune encephalomyelitis that was accompanied by severe disease. A recent report indicates that the loss of the regulatory influence of NK cells on Th17 differentiation accounts for the severe EAE phenotype in CX3CR1GFP/GFP mice [74]. In accordance, alterations of CX3CR1+ NK cell numbers are observed in patients with multiple sclerosis.

The involvement of CXCL10 and CX3CL1 in supporting anti-tumor NK cell response in vivo has been also described. CXCL10 was shown to promote a strong anti-leukemic NK cell-mediated response via enhanced induction of cytotoxicity and expression of the T cell co-stimulatory molecule B7-H1 [75]. In addition, a link between IFN-gamma induction of CXCR3 ligand expression during the anti-tumor immune response and CD27high NK cell recruitment within the tumor mass was found to be critical for host survival [76]. In addition, CX3CL1-transfected tumor cells exhibit a reduced growth capability that is mediated by an increased recruitment and activation of NK cells [70].

The selective role of chemokine/chemokine receptor interaction in NK cell migration in vivo suggests that different NK cell subsets may be independently recruited in distinct inflammatory settings, as the two major subsets of mature NK cells, CD27high and CD27low, can be characterized by the mutually exclusive expression of CXCR3 and CX3CR1. By using selective depletion and adoptive transfer experiments, Martin-Fontecha et al. have reported that DC-induced recruitment of NK cells into LN occurs in a CXCR3- but not CCR7-dependent manner. The entry of NK cells within the T cell area of secondary lymphoid organs has several implications allowing their correct priming by DC as well as an efficient T cell polarization [77], [78]. The role of CXCR3 as well as CCL5 receptors has also been emphasized within the spleen in the recruitment of red pulp-localized NK cells to the white pulp during TLR ligand stimulation in vivo [16]. Interestingly, the entry of NK cells into the splenic red pulp is Pertussis Toxin (PTX)-insensitive, thus suggesting that NK cell localization into this organ is regulated by a chemokine-independent or a PTX-insensitive chemokine receptor.

These data collected in a number of mouse models strongly support the in vivo relevance of a number of chemokine receptor-ligand interactions, including CX3CR1-CX3CL1, CXCR3-CXCL10/CXCL11, CCR5-CCL3/CCL4, CCR5-CCL5, that have been shown to mediate human NK cell chemotactic response in vitro.

Similarly to mouse, also in humans, inflammatory responses can induce drastic changes in the NK cell chemokine receptor expression pattern and in the expression of their ligands in inflamed or injured tissues, thus altering NK cell subset recruitment and redistribution. Indeed, expression of CCL1, CCL5 and CXCL10 and of chemerin was associated with recruitment of NK cells in skin biopsies of patients with psoriasis. In addition, a strong correlation between ChemR23 expression and co-localization of CD56lowNK cells with DC was observed in human biopsies of oral lichen planus in areas where chemerin is present [43]. Differently, in skin lesions, CD56highCD16− NK cells are preferentially recruited and display a CCR6+CXCR3+CCR5+ChemR23+ receptor profile.

Regulation of the functional interaction between NK cell and other immune cells by chemokines

Besides the role of NK cell-mediated cytotoxicity in the direct lysis of tumor or infected cells, NK cell cytolitic and cytokine secretion potential can affect innate and adaptive immune response by controlling the function of other immune cells in several conditions (Fig. 2 ). On the other hand, the achievement of the full effector capacity by NK cells requires various factors, including DC or macrophage-presented IL-15, IL-12 or IL-18 thus emphasizing the importance of a functional crosstalk between NK cells and immune cells. In this regard, a potent cross-talk can occur between NK cells and DCs: on one hand NK cells release high levels of cytokines including IFN-gamma and TNF-alpha which promote DC maturation, and in turn DC secrete cytokines such as IL-12, IL-18, IFN-alpha and IL-15 that induce NK cell activation and proliferation. When present at high numbers, NK cells can also kill immature DCs.

Fig. 2

Chemokines regulate NK cell cross-talk with other immune cells in lymphoid and non lymphoid tissues. NK cells functional interaction with DC, monocytes or tissue resident machrophages can be facilitated by tissue expression of chemokines that promote simultaneous recruitment of cells from circulation facilitating their communication in vivo. In addition, several in vitro studies have evidenced the role of chemokines produced by NK cells and other immune cells, including CCR5, CXCR3 and XCL1 ligands for reciprocal attraction.

Secretion of chemokines is key to the co-localization of NK cells with other hematopoietic cells in areas of inflammation. In this regard, during inflammation activated DCs can secrete several chemokines such as CCL4, CCL5, CX3CL1, CXCL8, and CXCL10 [79].

In vivo, interaction between NK cells and DC has been suggested to occur in the spleen during viral or bacterial infection [80]. Following infection with Listeria monocytogenes, CCR2-mediated signals promote emigration of monocytes from bone marrow into the blood stream and eventually to infected tissues where monocytes differentiate into TNF and iNOS producing DCs (Tip-DCs). In this regard Auffray et al. showed that recruitment of monocytes to the spleen during Lm infection involves another chemokine receptor, namely CX3CR1 [81]. During infection, DC cells where shown to promote a marked CD11b+ and NK cells reorganization in clusters positioned in the T cell areas of the spleen, with the CD11b+ cells organized centrally and the NK cells localized peripherally [82]. NK cells were the predominant producers of IFN-gamma, inducing differentiation of CD11b+ monocytes into Tip-DC. Among a number of chemokine receptors analyzed, mice deficient in CCR5, the receptor for CCL3-5, had a consistent but partial defect in NK cell clustering while myeloid cell localization were little affected by the lack of this receptor. NK cell interaction with monocytes in spleen may occur also during steady state. Indeed, Soderquest et al. recently evidenced that splenic monocytes drive transition of CD11bhighCD27high NK cell into CD11bhighCD27low in a IL-15-dependent manner [83]. Beside spleen, several chemokines participates in the recruitment of immune cells into LNs that represent critical sites for NK cell interaction with other immune cells such as DCs and for their ability to modulate DC-dependent Th1 cell polarization during inflammation promoted by adjuvants or LPS-activated DCs [77], [78]. NK cell entry was shown to be CXCR3-dependent and this correlated with a rapid and transient expression of CXCL9 in the luminal side of reactive nodes. Subsequent interaction with DC occurs to promote NK cell priming and may be driven by DC-produced chemokines including another CXCR3 ligand, CXCL10. Intravital imaging studies suggested that NK cell priming during inflammation does not require the formation of stable contact between NK cells and DCs during time and paralleled a broader distribution of NK cells that were more located in the deep T cell zone as compared to steady state [84].

In the course of inflammation, NK cells are found in close proximity to DC also in non-lymphoid tissues likely resulting in reciprocal activation. In oral lichen planus, a pathological condition known to be characterized by elevated DC tissue accumulation, expression of the receptor for chemerin ChemR23, was evidenced on CD3−CD56+ NK cells and on CD123+/BDCA+ and DC-SIGN+ DC that were close to each other in proximity of chemerin-expressing endothelial cells lining blood vessels [43].

Also during psoriasis NK cells and DC were found in close proximity, but although chemerin and CCL5 and CXCL10 expression were associated with NK cell accumulation in diseased skin, this did not correlate with their interaction with DCs.

In vitro experiments evidenced other chemokine candidates for DC–NK cell interaction by showing that DCs triggered with bacterial fragments is able to attract CCR5+ NK cells by producing CCL5. Contact with activated DC promotes IL-12/IL-18-mediated NK-cell activation, resulting in the production of IFN-gamma and CCR7 expression by a subpopulation of CD56dimCD16+ NK cells [79]. During an immune response, also NK cells produce chemokines in response to adhesive, cytokine or activating receptor stimuli, including IL-8, CCL2-5 and XCL-1. NK cell-derived chemokines can promote the recruitment of various leukocytes in inflamed tissues eventually leading to reciprocal activation.

Besides DC, other immune cells may be influenced by NK cell in peripheral tissues. Co-localization studies revealed that NK cells resided in close proximity to CNS microglial cells during experimental autoimmune encephalomyelitis and this correlated with microglial cell killing, thus resulting in NK cell-mediated suppression of Th17 differentiation. Furthermore, CNS NK cells readily produce CCL3 while microglia produced CCL2 likely resulting in reciprocal chemoattraction [74]. Recent evidences describe a cross-talk also between NK cells and neutrophils. Human NK cells may influence the survival and activation of neutrophils under co-culture conditions via the release of IFN-gamma or granulocyte macrophage colony-stimulating factor (GM-CSF) plus IFN-gamma, while neutrophil production of ROS reduces survival and activation of NK cells. It has been also suggested that the ROS preferentially affect the CD56low NK cell subset thus indirectly favoring CD56high NK cell-mediated responses. Thus, beside their ability to regulate neutrophil recruitment, chemokines production by NK cells could also affect NK cell–neutrophil encounter in inflamed tissues contributing to the pathogenesis of inflammatory diseases [85].

Conclusions

In the steady state, NK cells are present at a high frequency in the circulation, ready to extravasate to tissues under inflammatory conditions. The recent findings reviewed herein, highlight that NK cells respond to several chemoattractants regulating the maintenance of resting NK cells in the circulation or site of maturation, or the recruitment of activated NK cells into the sites of diseases and inflammation. In these locations, NK cells can play an important role as active participants in directing DC maturation and T cell response polarization and/or as cytotoxic effector cells. Involvement of chemokines in the regulation of DC-mediated NK cell priming and effector functions has also been documented and should be taken into account when analyzing the role of chemokines in NK cell-dependent immune responses