Chemokine Receptor Expression on Normal Blood CD56+ NK-Cells Elucidates Cell Partners That Comigrate during the Innate and Adaptive Immune Responses and Identifies a Transitional NK-Cell Population

Studies of chemokine receptors (CKR) in natural killer- (NK-) cells have already been published, but only a few gave detailed information on its differential expression on blood NK-cell subsets. We report on the expression of the inflammatory and homeostatic CKR on normal blood CD56+low CD16+ and CD56+high  CD16−/+low NK-cells. Conventional CD56+low and CD56+high NK-cells present in the normal PB do express CKR for inflammatory cytokines, although with different patterns CD56+low NK-cells are mainly CXCR1/CXCR2+ and CXCR3/CCR5−/+, whereas mostly CD56+high NK-cells are CXCR1/CXCR2− and CXCR3/CCR5+. Both NK-cell subsets have variable CXCR4 expression and are CCR4− and CCR6−. The CKR repertoire of the CD56+low NK-cells approaches to that of neutrophils, whereas the CKR repertoire of the CD56+high NK-cells mimics that of Th1+ T cells, suggesting that these cells are prepared to migrate into inflamed tissues at different phases of the immune response. In addition, we describe a subpopulation of NK-cells with intermediate levels of CD56 expression, which we named CD56+int NK-cells. These NK-cells are CXCR3/CCR5+, they have intermediate levels of expression of CD16, CD62L, CD94, and CD122, and they are CD57− and CD158a−. In view of their phenotypic features, we hypothesize that they correspond to a transitional stage, between the well-known CD56+high and CD56+low NK-cells populations.


Introduction
Natural killer-(NK-) cells were originally identified by their natural ability to kill target cells and are known for a long time as effector cells of the innate immune system, with an important role in controlling several types of tumors and infections [1]. In recent years, NK-cells have also been recognized as regulatory cells, which are able to interact with other cells of the immune system, such as dendritic cells (DC), monocytes/macrophages, and T cells, thereby influencing the innate and adaptive immune responses [2][3][4][5]. The role of their interaction with neutrophils in shaping the immune response is also being increasingly documented [6,7].
The cytotoxic activity of the NK-cells is controlled by the balance between inhibitory and activating receptors, whose ligands are self-Major Histocompatibility Complex (MHC) class I molecules and molecules expressed on stressed, viral infected, and tumor cells. They comprise, among others, the killer cell immunoglobulin-like receptors (KIR), killer cell be naturally prepared to act in different sites and at different phases of the immune response.
The exact relationship between these NK-cell subsets still remains unclear. Some studies have shown that bone marrow progenitor cells give rise to CD56 +high or CD56 +low NKcells depending on being cultured in the presence of IL-15 alone or in combination with IL-21, respectively [35,36]. However, more recent data would favor a possible maturation relationship between these NK-cell subsets and suggest that CD56 +low NK-cells originate from CD56 +high NK-cells [37][38][39][40][41][42].
Chemokines are small proteins that control a number of biological activities, including cell development, differentiation, tissue distribution, and function [43]. They act by binding chemokine receptors (CKR), a family of seventransmembrane proteins that are classified by structure according to the number and spacing of conserved cysteines into four major groups given the names CXCR, CCR, CX3CR, and XCR to which four groups of CK correspond: CXCL, CCL, CL, and CX3CL [44]. In addition, CXCL chemokines have been further subclassified into glutamic acid-leucinearginine tripeptide (ELR) positive or negative, based on the presence or absence of the ELR motif N-terminal to the first cysteine. From a functional point of view, two distinct types of CK have been considered: inflammatory/inducible CK, which are regulated by proinflammatory stimuli and dictate migration to the inflamed tissues and homeostatic/constitutive CK, which are responsible for the homing of the immune cells to the lymphoid organs and tissues. Similarly, two distinct groups of CKR have been described: those that interact mainly with inflammatory/inducible CK and have overlapping specificities and those that are relatively specific for homeostatic/constitutive CK [43,44].
We have previously characterized the immunophenotype of blood CD56 +low and CD56 +high NK-cells [29]. In order to better understand the migration pathways and cell-interactions of these NK-cell subsets and to establish the normal reference patterns for the study of the NK-cell lymphoproliferative disorders, we decided to investigate the expression of a number of CKR on these NK-cell subsets. At some point in our study, we found that blood CD56 + NKcells include a minor population of CXCR3/CCR5 + NK-cells whose levels of CD56 expression are intermediate between those observed on CD56 +low and CD56 +high NK-cells, most of which are CD16 + . These cells, from now on referred to as CD56 + int NK-cells, fail to display CD57 and KIR, and they have intermediate levels of CD62L, CD94, and CD122 expression. Based on the results presented herein and on the published data, we discuss the migration routes of the conventional CD56 +high and CD56 +low NK-cells and their relevance for the success of the immune response and hypothesize that CD56 + int NK-cells probably represent a transitional NK-cell state.

Subjects.
We first analyzed by flow cytometry the expression of a number of CKR on the CD56 + NK-cells in the PB of 15 adult healthy individuals (blood donors), 9 males and 6 females, aged from 19 to 54 years (median age of 38 years). After suspecting the existence of a subpopulation of CD56 + int NK-cells, these cells were further characterized using another group of 13 adult healthy individuals (blood donors), 8 males and 5 females, aged from 20 to 64 years (median age of 40 years).

Ethical Statement.
This study was approved by the Ethical Committee as part of a research project aimed to characterize the CKR on normal and neoplastic T cells and NK-cells in order to better understand the biology of the T cell and NK-cell lymphoproliferative disorders. All individuals gave informed consent to participate in the study.
Data acquisition was carried out in a FACSCalibur flow cytometer (BD) equipped with a 15 mW air-cooled 488 nm argon ion laser and a 625 nm neon diode laser, using the CellQUEST software (BD). Information on a minimum of 2 × 10 5 events was acquired and stored as FCS 2.0 data files for each staining. For data analysis the Paint-a-Gate PRO (BD) and the Infinicyt (Cytognos, Salamanca, Spain) software programs were used.
Using the first staining protocol, NK-cells were first gated based on their CD3 − /CD56 + phenotype; then, the conventional CD56 +low and CD56 +high NK-cell subsets were selected based on their levels of CD56 expression and on their differential positivity for CD16 and separately analyzed for the expression of CXCR1, CXCR2, CXCR3, CCR4, CCR5, and CCR6. Using the second staining protocol, in which the anti-CXCR3 and CCR5 mAbs used have the same fluorochrome, we were able to distinguish three populations of CD56 + NKcells: CD56 +low CXCR3/CCR5 − , CD56 + int CXCR3/CR5 + , and CD56 +high CXCR3/CR5 + . These were separately analyzed for the expression of CD16, CD56, CD57, CD62L, CD94, CD158a, and CD122.
The percentage of positive cells, the mean fluorescence intensity (MFI, expressed as arbitrary relative linear units scaled from 0 to 10,000), and the coefficient of variation of the MFI (CV, expressed as percentage) were recorded for each molecule tested.

Statistical Analysis.
For all quantitative variables under study, mean, standard deviation, median, and range values were calculated. The statistical significance of the differences observed between groups was evaluated using the Mann-Whitney -test (SPSS 10.0, SPSS, Chicago, IL, USA). values less than 0.05 were considered to be associated with statistical significance.

Chemokine
Receptors on Blood CD56 +low and CD56 +high NK-Cells. Conventional CD56 +low and CD56 +high NK-cells present in the normal PB have different CKR repertoires ( Figure 1 and Table 1).  Then, after gating for CD56 + NK-cells (first CD56/CD16 dot plot), the CD56 +low (red dots) and CD56 +high (blue dots) NK-cell populations were identified based on their typical patterns of CD56 and CD16 expression (second CD56/CD16 dot plot). Finally, these NK-cell populations were analyzed for the expression of the CKR (CKR/CD56 dot plots). The numbers above the CD56 +low and CD56 +high NK-cells inside the CKR/CD56 dot plots indicate the percentage of cells staining positively for the correspondent CKR and were obtained after gating separately for each NK-cell population (CKR/CD56 dot plots gated for CD56 +low and CD56 +high NK-cells are not shown, for simplicity).
Results are expressed as mean ± standard deviation (minimum-maximum) of the percentage of cells expressing each of the chemokine receptors analyzed within each CD56 + NK-cell population as well as mean ± standard deviation (minimum-maximum) of the mean fluorescence intensity (MFI) and coefficient of variation (CV) of expression. n.s.: not statistically significant.
In the normal PB, CD56 +low NK-cells correspond to the majority (mean of 90 ± 4%) of CD56 + NK-cells, whereas the  and CD158a molecules on CD56 +low CXCR3/CCR5 − (red dots), CD56 + int CXCR3/CCR5 + (green dots), and CD56 +high CXCR3/CCR5 + (blue dots) NK-cells in normal peripheral blood (PB). In order to obtain the dot plots showed in this figure, PB cells were stained with APC-conjugated anti-CD3, PC5-conjugated anti-CD56, PE-conjugated anti-CXCR3 + PE-conjugated anti-CCR5, and FITC-conjugated monoclonal antibodies against CD16, CD57, CD62L, CD94, CD122, or CD158a molecules. Total CD56 + cells were gated using the strategy illustrated in Figure 1. Then, using the CD56/CCR5 + CXCR3 dot plot (first dot plot), three different CD56 + NK-cell populations were identified based on the levels of expression of CD56 and CXCR3/CCR5: CD56 +low CCR5/CXCR3 − (red dots), CD56 + int CCR5/CXCR3 + (green dots), and CD56 +high CCR5/CXCR3 + (blue dots). As it can be seen in the remaining dot plots and histograms, these NK-cell populations differ on the expression of several cell surface molecules. The percentage of cells staining positively for each molecule analyzed, as well as the mean fluorescence intensity of antigen expression and its coefficient of variation, was calculated after gating separately for each NK-cell population and is shown in Table 1 (data is not shown in the figure, for simplicity).
No differences were observed between these three NKcell subsets concerning both the cell size and complexity, as evaluated by the forward (FSC) and side light scatter (SSC), respectively, except for a slightly larger size of CD56 +high NK-cells (Table 2). Nonetheless, statistically significant differences were found concerning the expression of CD56 and CD16 ( Figure 2 and Table 3) as well as of the other adhesion molecules and homing, cytokine, and killer cell receptors analyzed (Figure 2 and Table 4).
These three CD56 + NK-cell subsets also differ in the expression of other molecules (Figure 2 and Table 4).

Discussion
In the present study we show that CD56 +low and CD56 +high NK-cells that circulate in the normal blood have typical and quite different patterns of expression of receptors for inflammatory chemokines. At the same time, we identify and describe a subpopulation of CD56 + int NK-cells that could represent a transitional stage in between the conventional NK-cell subsets referred to above, based on their intermediate levels of CD56 and CD16 expression and on their patterns of chemokine (CXCR3, CCR5), cytokine (CD122), and killer cell (CD94, CD158a) receptors and adhesion molecules (CD62L, CD57).
Differences on the CKR repertoires make the NK-cell subsets naturally able to circulate in the blood, to home into secondary lymphoid organs, or to migrate into inflamed tissues, in different circumstances and with different partners (Figure 4), in response to constitutive and inflammatory chemokines (Table 5).
In accordance, the majority of CD56 +high NK-cells are CXCR3/CCR5 + , a pattern of CKR expression that is typically observed in Th1 cells [65], while CD56 +low NK-cells do express CXCR1 and CXCR2, the only CKR specific for the ELR + CXCL chemokines involved in inflammation, thus mimicking neutrophils [66,67]. In addition, both NK-cell subsets have variable levels of CXCR4 and virtually no CCR4 and CCR6 expression.
CD56 +high NK-cells and Th1 cells, the primary cell populations responsible for IL-2, IFN-, and TNF-production in response to IL-2 or certain monokines, such as IL-12 and IL-15, are attracted together to chronically inflamed tissues in response to CCR5 (MIP-1 , MIP-1 , RANTES, and MCP-2) and CXCR3 (MIG, IP-10, and I-TAC) chemokine ligands, where they orchestrate the adaptive immune response. Some of these CK, such as RANTES and MIP-1 , also attract proinflammatory CD14 +low CD16 + monocytes, by acting as ligands for CCR1 and CCR4, as well as for CCR5 [68].
In agreement, CD56 +high NK-cells accumulate within Th1-type chronic inflammatory lesions in a wide variety of pathological conditions such as rheumatoid arthritis [69], psoriasis [70], sarcoidosis [71], and allograft rejection [72] as well as in sites of intracellular bacterial infections [73], chronic viral infections [74], and tumors [75]. Inside the inflamed tissues and imbibed in the appropriate monokine environment, CD56 +high NK-cells are able to engage with monocytes in a reciprocal fashion [76], thereby amplifying the inflammatory response and having important antitumor and antiviral effects. In the LN, they can induce the maturation of DC via IFN-and TNF-release and/or cell-cell contact-dependent mechanisms [2,3], in that way shaping the subsequent immune response. Moreover, activated NKcells can kill immature myeloid DC, which have insufficient 8 Journal of Immunology Research  Data were obtained using the gating and analysis strategies described in Figure 2, where representative dot plots of these three NK-cell subsets are presented.
Results are expressed as mean ± standard deviation (minimum-maximum) of the percentage of each NK-cell subset within total CD56 + NK-cells and as mean ± standard deviation (minimum-maximum) of the transformed side scatter (tSSC) and forward scatter (FSC) channel of each NK-cell subset.
Journal of Immunology Research 9

<0.001
Data were obtained using the gating and analysis strategies described in Figure 2, where representative dot plots of these three NK-cell subsets are presented.
Results are expressed as mean ± standard deviation (minimum-maximum) of the percentage of positive (+) cells within each CD56 + NK-cell population and of the mean fluorescence intensity (MFI) and coefficient of variation (CV) of CD16 and CD56 expression. n.s.: not statistically significant.   Data were obtained using the gating and analysis strategies described in Figure 2, where representative dot plots of these three NK-cell subsets are presented.
Results are expressed as mean ± standard deviation (minimum-maximum) of the percentage of positive (+) cells within each CD56 + NK-cell population, and as the mean fluorescence intensity (MFI) and coefficient of variation (CV) of CD57, CD62L, CD94, CD122, and CD158a expression in cells that stained positively for these antigens. n.s.: not statistically significant.
amounts of MHC molecules to activate T cells properly [2,3]. In addition, CD56 +high NK-cells also predominate in placenta [77], where they are involved in maternal-fetal tolerance [78,79]. In contrast, CD56 +low NK-cells, which are essentially cytotoxic, and neutrophils, which are phagocytic cells by excellence, predominate in the PB and are equipped with the CXCR1 and CXCR2 chemokine receptors, making them able to comigrate into sites of acute inflammation in response to IL-8 and other ELR motif containing chemokines and to participate in the earliest phase of the innate immune response. As for the neutrophils, migration of CD56 +low NKcells to inflamed tissues also depends on the interaction of different forms of PSGL-1 expressed on their membrane with the selectin molecules expressed on endothelial cells [31]. Curiously, cytotoxic T lymphocytes (CTL) have also been reported to express CXCR1 [80] and PSGL-1 [81].
Normally, both neutrophils, which are able to neutralize efficiently the extracellular pathogens, after opsonization by antibodies, using Fc receptors for IgG and IgA and CD56 +low NK-cells, which mediate antibody dependent cell cytotoxicity via Fc RIIIa (CD16), circulate in the blood. In that sense, it can be hypothesized that these cells are candidates to establish a bridge between the innate immune response and the antibody mediated adaptive immune response. Evidence is being accumulated in the last years for a cross-talk between neutrophils and NK-cells [6,7]. For instance, NKcells promote neutrophil recruitment to the inflamed tissues and several NK-cell derived cytokines and growth factors, such as GM-CSF, IFN-, and TNF-, act by enhancing neutrophil survival and by modulating cell surface expression of complement and Fc receptors in neutrophils [82][83][84]. On the other hand, neutrophils can stimulate the production of IFN-by NK-cells [84].
The other anatomical sites in which CD56 +low NKcells and neutrophils might be concomitantly present to modulate each other's activity and its contribution to disease are not completely elucidated. Normal liver contains mainly CD56 +low NK-cells, but these cells are different from the CD56 +low NK-cells that circulate in the blood [85]. In addition, CD56 +low NK-cells also infiltrate the liver of patients with primary biliary cirrhosis, an antibody mediated autoimmune disease, following the CXCR1/IL-8 axis [85]; curiously, hepatic infiltration by neutrophils is also found in these patients [86]. Moreover, CD56 +low NK-cells and neutrophils colocalize in the skin of patients with Sweet's syndrome, an acute febrile neutrophilic dermatosis that can follow viral infections, autoimmune diseases, and hematologic malignancies [84].
Also of note, in this study we confirm previous observations about the lack of expression on both CD56 +low and CD56 +high NK-cells of other CKR involved in homing to nonlymphoid organs and tissues including CCR4 (skin and lung), CCR6 (intestine and liver), CCR9 (small intestine), and CCR10 (skin) [45,87]. This suggests that, unlike memory/effector T cells, CD56 + NK-cells may not be divided into cutaneous versus mucosal/intestinal-homing compartments, based on CKR expression [58,88,89].
Of special interest is also the identification of a new population of NK-cells expressing intermediate levels of CD56 that        Figure 4: Diagram illustrating the complex relationship established between NK-cells and the other cells of the innate-dendritic cells (DC), monocytes, macrophages, neutrophils, and adaptive (T cells) immune system, whose homing to lymphoid organs and recruitment to inflamed tissues are mediated by the interaction of homeostatic chemokines constitutively expressed on locally resident cells and inflammatory chemokines, with the correspondent chemokine receptors. CCR7 expression on CD56 +high NK-cells, mature DC, and naïve T cells allows these cells to migrate into the lymph nodes, in response to CCL19 (ELC) and CCL21 (SLC) produced locally. CXCR3/CCR5 expression on CD56 +high NK-cells permits these cells to migrate into inflamed tissues, together with CCR5 + proinflammatory monocytes, CCR5/CXCR3 + Th1 cells, and CCR5/CXCR3 + cytotoxic T lymphocytes (CTL). CXCR1/CXCR2 expression on CD56 +low NK-cells, neutrophils, and CTL permits these cells to migrate into inflamed tissues in response to CXCL8 (IL-8), where they interact together and with activated macrophages. CD56 + int NK-cells are transitional NK-cells, whose properties are intermediate between those of CD56 +high and CD56 +low NK-cells. we designed as CD56 + int NK-cells. Similar to CD56 +high NKcells, most of the CD56 + int NK-cells are KIR − and CD57 − ; however, the majority of them display the CD16 molecule, a marker of CD56 +low NK-cells, and they have intermediate levels of CD62L, CD94, and CD122 expression. Due to the fact that the identification of the NK-cell populations by flow cytometry is usually based only on CD56 and CD16 expression, CD56 + int NK-cells are being considered together with CD56 +low NK-cells on routine blood analysis.
The fact that CD56 + int NK-cells have phenotypic features intermediate between those of conventional CD56 +low and CD56 +high NK-cells would suggest that they could represent a transitional NK-cell maturation stage.
In line with this hypothesis, evidence for the existence of transitional NK-cell populations with phenotypic features similar to those of the CD56 + int NK-cells described herein has also been provided in other studies [90,91]. In accordance, Yu et al. described a CD56 +low CD94 +high NKcell subset expressing CD2, CD62L, CD56, KIR, granzymes, and perforin, producing IFN-in response to monokines, and exhibiting CD94-mediated redirected killing at levels intermediate between those observed in CD56 +low CD94 +low and CD56 +high CD94 +high NK-cells [90]. In addition, Juelke el al. reported on a CD56 +low CD62L + NK-cell subset with the ability to produce IFN-and the capacity to kill [91]. Finally, when studying the differentiation of CD56 +high CD94/NKG2A + into CD56 +low CD94/NKG2A − NK-cells, Béziat et al. found a transitional CD56 +low CD94/NKG2A + NK-cell subset, expressing intermediate levels of CD62L, granzyme-K, CD27, and CD57, among other molecules [92]. Given the immunophenotypic similarities, the CD56 + int NKcell population described herein, which comprises less than 10% of the CD56 + NK-cells in the PB from normal healthy individuals, probably corresponds to a subpopulation of the CD56 +low CD94 +high NK-cell subset described by Yu et al., which accounts for half of the circulating CD56 + NK-cells [90].
Another potential interest of identifying normal NKcells with intermediate phenotypic features relies on data interpretation in clinical settings. For instance, overrepresentation of CD56 + int NK-cells in the PB from patients with NK-cell lymphocytosis may erroneously be interpreted as phenotypically aberrant (and thus potentially neoplastic) NK-cells. Thus, the knowledge about the immunophenotype of the NK-cell populations that circulate in normal PB, as well as in the PB from patients with inflammatory and infectious conditions [93,94], is essential to a better understanding of the phenotypic heterogeneity of the expanded NK-cell populations observed in patients with CLPD-NK [95], thereby contributing to distinguishing nonclonal from clonal NKcell proliferations and reactive from neoplastic conditions [96,97].

Conclusions
Differences in the CKR expression on CD56 +low and CD56 +high NK-cells may determine their ability to be recruited into inflamed tissues and colocalize with other cells at sites of inflammation, which is crucial for the success of the immune response. In addition, the phenotypic heterogeneity of the conventional CD56 +low and CD56 +high NKcells may be largely due to the presence of transitional NK-cell populations, which may be preferentially expanded in some pathological conditions. Further investigations in this area will help to better understand the terminal differentiation of the NK-cells and the maturation relationship between the NK-cell subsets, their circulation through the body, and their participation in the immune response. In addition, they will give an important contribution to establish phenotypic criteria to differentiate reactive and neoplastic NK-cell proliferations, as well as to better identify the normal cell counterparts from which the neoplastic NK-cells originate.