A novel subset of NK cells expressing high levels of inhibitory FcγRIIB modulating antibody‐dependent function

NK cells can kill antibody‐coated target cells following engagement of FcγRIIIA, the major activating FcγR expressed by these cells. The presence of FcγRIIC (CD32C) has also been reported, but its contribution to the FcγR‐dependent effector functions of NK cells remains debated. We demonstrate here that inhibitory FcγRIIB is also expressed by a small subset of CD56+/NKp46+ NK cells and can efficiently down‐modulate their FcγR‐dependent effector function. Immunofluorescence analyses of NK cells from 52 healthy donors showed the presence of CD56bright/FcγRII− (5.2%±3.4), CD56dim/FcγRIIlo/‐ (94.1%±3.4), and CD56dim/FcγRIIbright (0.64%±0.72) cells. QRT‐PCR and protein analyses performed on isolated FcγRIIbright NK cells indicated that FcγRIIB is strongly expressed by these cells but not by FcγRIIlo/‐ cells. In addition, FcγRIIbright cells showed a weaker antibody‐dependent degranulation when incubated with IgG‐coated target cells compared with FcγRIIlo/‐ NK cells, although a strong FcγRIIIA expression was detected in both cells. Furthermore, the addition of anti‐FcγRII Fab paralleled a higher degranulation of FcγRIIbright NK cells, indicating a direct role for FcγRIIB in this down‐modulating effect. Thus, it is proposed that FcγRIIBbright NK cells represent a new NK cell compartment able to down‐modulate NK cell functions triggered by the engagement of activating FcγR.


INTRODUCTION
Human natural killer (NK) cells can kill immunoglobulin G (IgG)-coated targets through antibody-dependent cellular cytotoxicity (ADCC) following engagement of IgG Fc receptors (Fc␥R) expressed at their membrane [1,2]. The receptor responsible for mediating ADCC in NK cells, Fc␥RIIIA (CD16), is an intermediate-affinity Fc␥R that triggers NK cell cytotoxicity and cytokine production following its engagement by IgG bound to cell surface antigens [3]. Although Fc␥RIIIA is commonly thought to be the only Fc␥R expressed on NK cells, studies have demonstrated an heterogeneous expression of Fc␥RIIC (CD32C) on human NK cells from ϳ40% healthy subjects using RT-PCR and flow cytometry analyses [4 -6]. Fc␥RIIC belongs to the Fc␥RII family that consists of different structurally related 42-kDa molecules with low affinity for IgG [3,[7][8][9][10], including also the activating Fc␥RIIA and the inhibitory Fc␥RIIB [9 -12]. The FCGR2C gene results from an unequal crossover event between FCGR2A and FCGR2B genes and encodes a Fc␥ receptor exhibiting more than 99% homology with Fc␥RIIB in the extracellular domain [9,11,13]. Fc␥RII family members differ in their cellular expression, function, and ligand binding specificities [14]. Cross-linking of Fc␥RIIA results in up-regulation of intracellular Ca 2ϩ concentration, phagocytosis of opsonized particles, as well as internalization of immune complexes [15][16][17]. In contrast, engagement of Fc␥RIIB results in the inhibition of cell activation triggered via cell surface activating receptors [18,19]. Fc␥RIIB is mostly present on B cells, basophils, mast cells, and cells from the monocytic lineage [3]. However, it has been suggested that NK cells from a number of rheumatoid arthritis patients and from one healthy individual might also express Fc␥RIIB rather than Fc␥RIIC [20,21]. The role of Fc␥RIIC remains debatable, as the FCGR2C gene is often found to contain an in-frame termination codon, and this receptor has been considered as evolving into a pseudogene [11,22,23]. However, it has been demonstrated that some Fc␥RIIC splicevariants act as activating Fc␥R [24].
Both Fc␥RIIIA and Fc␥RIIB play a role in the antitumor activity of therapeutic monoclonal antibodies (mAbs). An improved clinical response has been observed after rituximab treatment in lymphoma patients harboring the high IgG1 binder Fc␥RIIIA-158V, suggesting an important role of Fc␥RIIIA-positive effector cells in the clinical outcome of the patients [25,26]. In contrast, Fc␥RIIB has been shown to negatively impact on the antitumor efficacy of trastuzumab in mice xenografted with an HER-2-positive human breast carcinoma [27]. Moreover, more recent studies highlighted the in vivo implication of Fc␥RIIIA ϩ NK cells in the clinical response to this antibody, in patients with HER-2-positive primary breast cancer [28,29].
Altogether, these data raise the possibility that an expression of the inhibitory Fc␥RIIB by NK cells could be responsible for a reduced NK cell antitumor activity in patients treated with cytotoxic therapeutic mAbs. The present study was therefore performed as a first step to analyze whether Fc␥RIIB could be detected on NK cells from healthy donors and could act as a negative regulator of activating Fc␥R-dependent functions in these cells. We demonstrate herein that peripheral blood NKp46 ϩ /CD56 ϩ /CD3 -NK cells from healthy donors contain a subset of CD56 dim cells that strongly express Fc␥RIIB. These CD56 dim /Fc␥RII bright NK cells exhibit a pattern of NK-cell receptors (NKR) expression different from that of NK cells that express a low level, if any, of Fc␥RII (CD56 dim /Fc␥RII lo/-NK cells). Finally, the in vitro degranulation capacity of Fc␥RIIB bright NK cells was lower than that of the Fc␥RII lo/cells following incubation with anti-CD20coated CD20 ϩ Raji cells. The use of anti-Fc␥RII Fab fragments demonstrated that Fc␥RIIB is responsible for this reduced degranulation, indicating an active role of this receptor in the control of Fc␥RIIIA-dependent functions in NK cells.

Confocal microscopy
Freshly purified NK cells (5ϫ10 5 ) were stained with 2 g/ml primary antibodies [mouse IgG 2b APC-conjugated anti-CD56 mAb (clone NCAM16.2) in combination with mouse IgG 1 anti-Fc␥R mAbs (anti-Fc␥RII, clone KB61, or anti-Fc␥RIII, clone 3G8) for 30 min on ice. All incubations were done using PBS containing 5% human AB serum. Cells were then washed twice in PBS and incubated on ice for 30 min with biotin-conjugated goat anti-mouse (GAM)-IgG 1 (1/100 dilution, The Binding Site, Birmingham, UK) and, after two washes in PBS, Cy3-conjugated streptavidin (1/1000 dilution; Jackson Immunoresearch Laboratories, West Grove, PA, USA) was added for an additional 20 min on ice. Cells were then washed two more times in PBS, fixed with 0.5% formaldehyde containing PBS and resuspended with 15 l Mowiol mounting medium prior to spreading on a 75-m capillary gap microscope glass slide (Dako ChemMate; Dako, Glostrup, Denmark). Immunofluorescence analysis was performed with a confocal laser scanner microscope Zeiss LSM 510 (Zeiss, Oberkochen, Germany) and the Zeiss LSM Image Browser V3.1.0.99 software (Zeiss). The wavelength of the HeNe laser was set at 543 nm for Cy3 excitation and at 633 nm for APC excitation. Fluorescence emission was revealed by BP 560-610 band pass filter for Cy3 and by LP 560 long pass filter for APC. Double-staining immunofluorescence images were acquired sequentially in the red and blue channels at a resolution of 512ϫ512 pixels and analyzed using the Image J freeware (version 1.37v).

Quantification of Fc␥RIIb transcripts by real-time PCR
Total cellular RNA was isolated from sorted CD56 dim /CD3 -/FcgRII bright and CD56 dim /CD3 -/FcgRII lo/-NK cells using the RNeasy mini kit or RNeasy micro kit (Qiagen, Valencia, CA, USA). Each sample was treated with DNase I (Qiagen). The integrity and the quantity of RNA were evaluated on a bioanalyzer-2100 (Agilent Technologies, Santa Clara, CA, USA). cDNA were then generated by reverse transcription (RT) of 10 ng total cellular RNA and linearly amplified using the Quantitect whole Transcriptome kit (Qiagen). Each cDNA sample was then used as a template for quantitative PCR reactions performed using the universal PCR master Mix (Applied-Biosystems, Foster City, CA, USA), Fc␥RIIb-specific primers and probe for real-time RT-PCR (Assay ID: Hs00269610_m1; Applied-Biosystems) on a ABI Prism 7900HT Sequence detection system (Taqman; Applied-Biosystems). Human ␤-actin (ACTB) Endogenous Control (part no. 4333762T; Applied-Biosystems) was used as internal control. The RQ manager software V3.2 (Applied-Biosystems) was used to establish the PCR cycle at which the fluorescence exceeded a set threshold, C T , for each sample. Data were then analyzed according to the 2 -⌬⌬CT method, where ⌬C T represents C T Fc␥RIIb -C T ACTB [33] For each tested donor, the ⌬⌬C T was then calculated as The n-fold difference between the amounts of Fc␥RIIb transcripts of the target (Fc␥RII bright NK cells) relative to the calibrator (Fc␥RII lo/-NK cells) was then calculated as 2 -⌬⌬CT [33].

CD107a mobilization assay
Degranulation was assessed with a CD107a mobilization assay [36,37] using NK cells and MHC class Ihuman erythroleukemic K562 target cells at different effector/target (E/T) cell ratios, or CD20 ϩ human lymphoblastoid Raji target cells (E/Tϭ15/1) coated with 5 or 500 ng/ml of a chimeric anti-human CD20 mAb (clone EMAB-6, Laboratoire Français de Fractionnement et des Biotechnologies, Lille, France) [38]. In some experiments, Fab fragments of the anti-Fc␥RII KB61 mAb (that inhibits the IgG binding to Fc␥RII) were added at various concentrations (0.1, 0.5 and 1.0 g/ml), in order to inhibit the engagement of Fc␥RII by the anti-CD20 mAb. Fab fragments of the anti-Fc␥RIIA IV.3 mAb [31] were used as negative control. PE-Cy5-conjugated anti-CD107a (clone H4A3, Becton Dickinson) mAb and monensin (5 M, Sigma) were added to the effector and target cell mixtures that were then incubated for 4 h. Monensin is added to prevent the acidification of the endosomal compartments, which could alter the fluorescence of internalized CD107a/CD107a mAb complexes [36]. Cells were then washed in PBS containing 2 mM EDTA and stained for extracellular markers [APC-conjugated anti-CD56, PE-conjugated anti-CD3 and FITC-conjugated anti-Fc␥RII (clone 3D3 from Becton Dickinson) mAbs], to evaluate CD107a expression level by different NK cell subpopulations defined on their Fc␥RII expression level. PE-Cy5 ϩ cells were scored by flow cytometry, revealing cells that have undergone degranulation. Results are presented as means Ϯ SD of positivestained cells.

Statistical analysis
Statistical analyses were performed using the Statview 5.0 software package for Windows (SAS Institute, Cary, NC, USA) for NKR phenotypic characterization and analysis of CD107a expression. Statistically significant differences were calculated using the Student paired t test. P values are indicated in the figures.

Fc␥RII expression defines a novel NK cell subset in healthy donors
Freshly purified NK cells isolated by negative selection from PBMCs of healthy donors were analyzed for Fc␥RII expression by flow cytometry using the anti-Fc␥RII KB61 mAb [30]. Three subpopulations of CD56 ϩ /CD3 -NK cells could be defined based on Fc␥RII expression levels. Fig. 1A shows representative flow cytometry data obtained from two of the tested donors (nϭ52). First, CD56 bright NK cells (ϳ5% of total NK cells) were either Fc␥RII-negative or exhibited only marginal, as also seen for Fc␥RIIIA [39] (Fig. 1A, middle panels, solid arrowheads). Second, as previously reported by Metes et al. [5] and Stewart-Akers et al. [21], a large subpopulation of CD56 dim NK cells (Ͼ90% of total NK cells) that expressed either low levels (left middle panel) of Fc␥RII or not (right middle panel), hence termed CD56 dim /Fc␥RII lo/-NK cell subpopulation, was observed (Fig. 1A, middle panels, gray-filled arrowheads). Third, a small subpopulation of CD56 dim NK cells expressing high levels of Fc␥RII (CD56 dim /Fc␥RII bright ) was detected (Fig. 1A, middle panels, open arrowheads). This cell subpopulation was detected in all the individuals tested, ranging from 0.13% to 2.28% (nϭ52, median value: 0.64%). Similarly to Fc␥RIIIA (CD16), Fc␥RII was detected only on CD56 dim /CD3 -NK cells. Coexpression of the two receptors was always observed by quadruple CD56/CD3/Fc␥RIII/Fc␥RII staining (see Fig. 5A, upper panel).
To confirm that the CD56 dim /Fc␥RII bright cells observed are, indeed, NK cells, double NKp46/Fc␥RII staining was also performed, since NKp46 is considered to be a specific NK cell receptor [40]. Fig. 1A (lower panels) shows the data obtained with the same two donors. As already observed when NK cells were defined as CD56 ϩ /CD3cells, NKp46 ϩ NK cells expressing high levels of Fc␥RII (NKp46 ϩ /Fc␥RII bright ) were detected. Similar results were obtained with 10 donors.
Confocal microscopy experiments using purified NK cells also showed the expression of Fc␥RII by CD56 ϩ cells (Fig.  1B). As observed in flow cytometry analyses, NK cells that strongly expressed CD56 (CD56 bright , blue staining) were not labeled with the anti-Fc␥RII KB61 mAb. NK cells with a faint expression of CD56 were either moderately or strongly labeled with KB61 mAb (red staining), a situation most likely reflecting the CD56 dim /Fc␥RII lo/and CD56 dim /Fc␥RII bright NK cell subpopulations, respectively.
The expression of Fc␥RII on CD56 dim /CD3 -NK cells was further confirmed using two other anti-Fc␥RII mAbs (3D3 and AT10). All three antibodies allowed the detection of the CD56 dim /Fc␥RII bright subpopulation ( Fig. 2A). Of note, the staining of CD56 dim /Fc␥RII lo/cells by the AT10 mAb was weaker when compared with the two other mAbs (3D3 and KB61) ( Fig. 2A, right panel). KB61 mAb has been described previously as binding the three Fc␥RIIA/B and C isoforms with high affinities in contrast to the AT10 mAb that binds Fc␥RIIA and Fc␥RIIB strongly and Fc␥RIIC weakly [20]. Thus, the stronger binding of KB61 mAb to CD56 dim /Fc␥RII lo/-NK cells compared with AT10 may reflect the previously described expression of Fc␥RIIC by these cells [20,41].
Evidence that the detection of the Fc␥RII bright NK cell subset was not due to a nonspecific binding of the Fc region of the anti-Fc␥RII mAbs to Fc␥RIIIA was obtained from the following results. First, triple labeling of freshly purified NK cells with APC-anti-CD56, FITC-anti-CD3, and PE-labeled KB61 Fab fragments allowed the detection of the Fc␥RII bright NK cell subset (Fig. 2B). Interestingly, the Fc␥RII bright NK cells were not detected using PE-labeled Fab fragments of the IV.3 mAb, which is preferentially directed against Fc␥RIIA (Fig. 2B) [31]. Second, preincubation of freshly purified NK cells with an anti-Fc␥RIII binding site mAb (3G8) prior to triple CD56/CD3/Fc␥RII staining did not hinder the detection of Fc␥RII bright NK cells (Fig. 2C).

Phenotypic characterization of the Fc␥RII bright NK cell subset
To assess whether Fc␥RII bright NK cells show a characteristic expression pattern of activating and/or inhibitory NKRs, NKR expression by these cells was compared with that of Fc␥RII lo/cells using 4-color flow cytometry analyses. Quadruple stainings (anti-CD3, anti-CD56, anti-Fc␥RII, and one of the anti-NKR mAbs) were carried out on freshly purified NK cells. NKG2D, CD161, NKp44, NKp46, NKp80, and CD158i were expressed at similar levels between the two subpopulations (data not shown). NKG2C, an activating receptor homologue to the inhibitory NKG2A receptor, was similarly expressed by the two NK cell subpopulations (Fig. 3). In contrast, CD94, the activating NKp30 and the inhibitory NKG2A receptors were expressed at significantly lower levels by Fc␥RII bright NK cells, while the inhibitory ILT-2 and the KIRs CD158a/h, CD158b1/ b2/j, and CD158e1/e2 showed a higher expression (Fig. 3).

Fc␥RII bright NK cells predominantly express high levels of the inhibitory Fc␥RIIB
To characterize the type of Fc␥RII expressed by Fc␥RII bright NK cells, NK cells were first enriched by negative selection (Ͼ70% CD56 ϩ /CD3cells) from more than 1ϫ10 9 PBMCs from a single healthy donor. Five independent experiments using different donors were performed. Isolated cells were triple-labeled with anti-CD3, anti-CD56 and anti-Fc␥RII mAbs to sort Fc␥RII lo/and Fc␥RII bright CD56 dim /CD3 -NK cells. Hydrophobic proteins were then extracted from the cell lysates of 5ϫ10 6 Fc␥RII lo/and 1 to 2.5ϫ10 5 Fc␥RII bright sorted NK cells depending on the experiment. Fc␥RII immunoprecipitation was then performed using the KB61 mAb, and Western blot analysis using specific rabbit polyclonal antibodies raised against Fc␥RIIB or Fc␥RIIA/C intracellular domains [34,35] were then carried out. Figure 4A shows the results obtained with two of the five donors. It indicates that Fc␥RIIB was strongly detected in the immunoprecipitates of the Fc␥RII bright NK cell lysates but not in those of Fc␥RII lo/- NK cells (Fig. 4A, upper panels). The same observation was made with the three other donors tested (data not shown). By contrast, only trace amount of activating Fc␥RIIA/C could be detected in Fc␥RII bright NK cells of only one of the donors tested (Fig. 4A, lower panels). A low level of activating Fc␥RIIA/C could be detected in Fc␥RII lo/-NK cell lysate, or not (Fig. 4A, lower panels).
To further assess the presence of Fc␥RIIB in Fc␥RII bright NK cells, Fc␥RII lo/and Fc␥RII bright CD56 dim /CD3 -NK cells from four other healthy donors were also FACS-sorted and Fc␥RIIb transcripts were quantified by real-time PCR. Data were analyzed according to the comparative C T method (see Materials and Methods) [33]. Fc␥RIIb transcripts were expressed by Fc␥RII bright NK cells at levels between 82 and 7ϫ10 5 times more than Fc␥RII lo/-NK cells for the four tested donors (Fig. 4B), thus confirming that Fc␥RII bright cells predominantly express Fc␥RIIB.

Fc␥RII bright NK cells exhibit reduced degranulation in Fc␥R-dependent in vitro assay
The functional activity of CD56 dim /Fc␥RII bright NK cells was then analyzed using the CD107a mobilization assay and was compared with that of CD56 dim /Fc␥RII lo/-NK cells. CD107a is a marker of intracytoplasmic cytotoxic granules that can be detected by flow cytometry using a fluorochrome-conjugated anti-CD107a mAb during granule-mediated exocytosis as it is transiently expressed at the degranulating cell membrane.
CD107a (LAMP-1) mobilization assay therefore permits the evaluation of NK cell degranulation at the single cell level by flow cytometry [36,37].
First, we evaluated Fc␥RIIIA and perforin expression by Fc␥RII bright NK cells. As shown in Fig. 5A, CD56 dim /CD3 -/ Fc␥RII bright NK cells strongly express Fc␥RIIIA on their cell surface, and intracellular staining also showed that these cells expressed perforin.
Second, freshly purified human NK cells from healthy donors were cultured for 4 h in various conditions in the presence of PE-Cy5-conjugated anti-CD107a mAb and monensin. Cells were then harvested and stained with FITC-conjugated anti-Fc␥RII, PE-conjugated anti-CD3, and APC-conjugated anti-CD56 mAbs prior to flow cytometry analyses. Cells in the CD56 ϩ /CD3lymphocyte gate were analyzed for CD107a expression. Two gates were defined, delineating the two CD56 dim Fc␥RII NK cell subpopulations (R1: Fc␥RII lo/cells, R2: Fc␥RII bright cells).
In these experiments, freshly purified NK cells were incubated with either the MHC Class I -NK-sensitive K562 cells to evaluate degranulation induced by NK cell natural activity, or with CD20 ϩ Raji cells and an anti-CD20 mAb (5 or 500 ng/ml) to evaluate their Fc␥R-mediated degranulation. Figure 5B shows the results obtained in one representative experiment. CD107a was barely detected at the surface of nonstimulated NK cells after 4 h incubation (% CD107a ϩ : R1/R2 ϭ 0.7%) (Fig. 5B, upper left panel). However, when mixed with the MHC Class I -NK-sensitive K562 cells, CD107a was expressed at a similar level on Fc␥RII bright cells (R2: 6.1% CD107a ϩ ) and Fc␥RII lo/cells (R1: 6.2% CD107a ϩ ) (Fig. 5B, upper right  panel), showing that this Fc␥RII bright NK cell subpopulation exhibits a potent natural killing activity. When incubated with CD20 ϩ Raji cells and 500 ng/ml anti-CD20 mAb, CD107a was detected on both Fc␥RII lo/and Fc␥RII bright NK cells. Interestingly, the Fc␥RII bright cells showed a reduced degranulation (R2 ϭ 19.3% CD107a ϩ ) compared with Fc␥RII lo/-NK cells (R1 ϭ 29.7% CD107a ϩ ) (Fig. 5B, lower left panel). Six independent experiments were performed using different healthy donors and led to similar results. Figure 5C summarizes the results obtained in these experiments, represented as mean percentages Ϯ SD of CD107a ϩ cells amongst the two NK cell populations in presence of CD20 ϩ Raji cells and anti-CD20 mAb. A significant decrease of degranulation by CD56 dim Fc␥RII bright NK cells compared with Fc␥RII lo/cells was observed in presence of both 5 and 500 ng/ml anti-CD20 mAb.
Furthermore, to assess the role of Fc␥RII in the Fc␥Rdependent degranulation assay, NK cells were also incubated in presence of CD20 ϩ Raji cells, the anti-CD20 mAb (500 ng/ml), and different concentrations of anti-Fc␥RII KB61 Fab fragments (0.1, 0.5, or 1.0 g/ml). KB61 Fab fragments were added at the beginning of the incubation in order to inhibit Fc␥RII engagement by the anti-CD20 mAb. Four independent experiments were performed. Figure 5B (lower right panel) shows the results obtained in one experiment. Degranulation of Fc␥RII lo/-NK cells (R1) was not affected by the addition of KB61 Fab fragments [29.7% (Fig. 5B, lower left panel) vs. 30.9% (Fig. 5B, lower right panel) CD107a ϩ cells in the absence or in the presence of KB61 Fab fragments, respectively], indicating that Fc␥RII expressed by Fc␥RII lo/-NK cells do not play a significant role in the Fc␥R-dependent degranulation of these cells. By contrast, the degranulation of Fc␥RII bright NK cells (R2) was significantly increased when KB61 Fab fragments was added [19.3% (Fig. 5B, lower left panel) vs. 29.8% (Fig. 5B, lower right panel) CD107a ϩ cells in the absence or in the presence of KB61 Fab fragments, respectively], leading to a percentage of CD107a-positive cells comparable to that of Fc␥RII lo/-NK cells. By contrast to KB61 Fab fragments, the addition of Fab fragments of the anti-Fc␥RIIA IV.3 mAb [31] that binds only marginally Fc␥RIIB did not modify the degranulation profile of cells from the two NK cell subsets induced by the anti-CD20 mAb (data not shown). Figure 5D summarizes the results obtained with the 4 tested individuals using different concentrations of Fab fragments of the KB61 anti-Fc␥RII mAb [0.1, 0.5, and 1.0 g/ml]. A dose-dependent effect of KB61 Fab on Fc␥RII bright NK cell degranulation was observed, while Fc␥RII lo/-NK cells degranulation was not significantly affected. Thus, all of these experiments demonstrate the functional inhibitory activity of Fc␥RIIB in the Fc␥R-dependent degranulation of Fc␥RII bright NK cells.

DISCUSSION
The observation that human NK cells from certain healthy individuals express Fc␥RII (CD32) is barely taken into account in NK cell studies, despite the fact that this observation was reported more than 10 years ago [4,5]. We confirm in the present study the low level of Fc␥RII expression by a CD56 dim / CD3subset of NK cells (termed herein Fc␥RII lo/-NK cells) from ϳ40% healthy donors. Fc␥RII could be detected at the surface of Fc␥RII lo/-NK cells using anti-Fc␥RII KB61 or 3D3 mAbs. Of note, the use of the AT10 mAb reproducibly led to a much lower staining of these cells. This particular binding pattern likely reflects the previously described predominant expression of the Fc␥RIIC by these NK cells, since AT10 mAb has been proposed as binding weakly the Fc␥RIIC but not the other Fc␥RII (IIA and IIB) [20]. In contrast, the KB61 mAb, strongly binds the three Fc␥RII (IIA, IIB, and IIC) [20]. RT-PCR analyses confirmed the presence of Fc␥RIIc1 and Fc␥RIIc3 transcripts in healthy donors exhibiting a KB61 staining of their Fc␥RII lo/-NK cell subset (data not shown).
Interestingly, a small subset of NK cells that strongly express Fc␥RIIB (termed Fc␥RII bright NK cells) differing from the Fc␥RII lo/-NK cell subset, and comprising less than 2% (median valueϮSD: 0.64%Ϯ0.72) of total CD56 ϩ /CD3 -NK cells, could be defined when analyzing a large number of cells by flow cytometry with the anti-Fc␥RII KB61 mAb. NKp46, which is described as being a specific NK cell receptor [40], and NKRs commonly expressed by resting peripheral blood NK cells were also found expressed on these Fc␥RII bright cells, confirming that this subpopulation does, indeed, correspond to NK cells. Furthermore, the strong expression of Fc␥RII by this cell subpopulation was confirmed using two other anti-Fc␥RII mAbs (AT10 and 3D3). The detection of Fc␥RII on these NK cells was not due to a nonspecific binding of the Fc region of the anti-Fc␥RII mAbs to Fc␥RIIIA since anti-Fc␥RII KB61 Fab fragments also stained these cells. In addition, preincubation of freshly purified NK cells with purified anti-Fc␥RIII (clone 3G8, directed against the Fc␥RIII IgG binding site) did not alter its detection.
Contrary to the variability of Fc␥RII expression observed at the surface of Fc␥RII lo/-NK cells in the healthy donors tested, the Fc␥RII bright NK cell subset was detected in all of the donors (nϭ52), irrespective of the level of Fc␥RII expression   [31]] to the Fc␥RII bright NK cell subset also indicated a weak or absent Fc␥RIIA expression by this NK cell subset. Thus, the binding pattern of AT10 mAb and Fab fragments of the IV.3 mAb observed in the immunofluorescence experi- in the antibody-dependent degranulation assay performed with six donors. NK cells were mixed with CD20 ϩ Raji cells (E/T: 15/1) and 500 or 5 ng/ml anti-CD20 mAb, or not. (D) Mean % CD107a ϩ cells (ϮSD) obtained in the antibody-dependent degranulation assay performed with four of the six donors tested (C) in the presence of KB61 Fab fragments. NK cells were incubated with CD20 ϩ Raji cells (E/T: 15/1), 500 ng/ml anti-CD20 mAb, and KB61 Fab fragments (0.1, 0.5, and 1.0 g/ml) or not. Statistically significant differences were calculated using the Student paired t test.
ments suggested an expression of the inhibitory Fc␥RIIB by Fc␥RII bright cells. It was confirmed by immunoprecipitation/ Western-blot and real-time semiquantitative PCR experiments carried out on FACS-sorted NK cells showing that Fc␥RII bright NK cells, but not Fc␥RII lo/-NK cells, strongly express the inhibitory Fc␥RIIB. A marginal expression of the activating Fc␥RIIA/C at the protein level was detected in only one of the five donors tested.
Finally, the Fc␥R-dependent CD107a mobilization assay showed a reduced degranulation of the Fc␥RII bright subpopulation compared with that detected in the Fc␥RII lo/subpopulation. This occurred despite the fact that Fc␥RII bright NK cells possess an antibody-dependent cytotoxic arsenal (strong expression of Fc␥RIIIA and positive intracellular staining of perforin). Moreover, the possibility of a reduced activation potential of these Fc␥RII bright NK cells can be excluded, since these cells showed a good degranulation ability when incubated with the MHC Class I -NK-sensitive K562 cells. When a saturating quantity of Fab fragments of anti-Fc␥RII mAb (clone KB61, directed against the Fc␥RII IgG binding site) was added in the Fc␥R-dependent degranulation assay, Fc␥RII bright NK cells degranulation was significantly increased and reached a level comparable to Fc␥RII lo/-NK cells. These results demonstrate that the Fc␥RIIB inhibitory function is responsible for the reduced degranulation of the Fc␥RII bright NK cells. Of note, the fact that Fc␥RII lo/-NK cells showed the same level of degranulation in the presence of KB61 Fab or not illustrates the marginal contribution of Fc␥RIIC in Fc␥Rdependent degranulation by Fc␥RII lo/-NK cells in these experimental conditions.
In summary, we have shown that Fc␥RII expression levels by human circulating CD56 ϩ /CD3 -NK cells from healthy donors define three functionally different subsets, the previously described CD56 bright /Fc␥RIIand CD56 dim /Fc␥RII lo/-NK cells [5,21], and a newly defined CD56 dim /Fc␥RII bright small subpopulation of NK cells expressing high levels of a functionally active inhibitory Fc␥RIIB. In addition, this subset also shows an increase in the expression of the inhibitory ILT-2 and all KIR tested, suggesting an inhibitory phenotype of these cells. The low number of these Fc␥RIIB bright NK cells in the peripheral blood of healthy donors raises the question of whether these cells play a significant role in controlling NK cell functions,. A local increase of the numbers of Fc␥RIIB at sites of inflammation, infections or tumors might have important consequences on the whole NK cell compartment capacity, through the down-modulation of the different functions triggered by the engagement of activating Fc␥RIIIA, from ADCC to the release of proinflammatory molecules. Evaluation of this CD56 dim / Fc␥RIIB bright NK cell subset in various pathologies might help to better understand its physiological role.