Human NK cells can be divided into CD56dim and CD56bright subsets. These two types of NK cells respond to different types of stimuli, with CD56dim NK cells having direct cytotoxic ability and CD56bright NK cells having mainly an immunoregulatory function. We show that the CD16+CD56dim NK subset is characterized by sensitivity to cell death induced by activated granulocytes. We identified hydrogen peroxide (H2O2) as the major effector molecule responsible for the cytotoxic effect of granulocytes on CD56dim NK cells, because the ability of granulocytes to kill CD56dim NK cells was completely abrogated in the presence of the hydrogen peroxide scavenger catalase. When exposing NK cells to H2O2, CD56dim cells showed rapid mitochondrial depolarization and down-regulation of activating NKRs, eventually resulting in cell death, whereas CD56bright cells remained unaffected. The difference in sensitivity to H2O2 was mirrored by a difference in intracellular oxidation levels between CD56dim and CD56bright NK cells, and cell lysates from the latter subset possessed a greater ability to block H2O2-mediated oxidation. Our data may explain the preferential accumulation of CD56bright NK cells often seen in environments rich in reactive oxygen species, such as at sites of chronic inflammation and in tumors.

Natural killer cells are lymphocytes that derive from the same common lymphocyte precursor as T cells. They play an important role in the early innate response against pathogens and may also be of importance in the recognition and clearance of cancer cells (1, 2). Human NK cells can be subdivided into two subsets based on their level of CD56 expression. CD56dim NK cells are potent cytotoxic cells, express high levels of granzymes and perforin, and can also mediate Ab-dependent cellular cytotoxicity through the FcR CD16. CD56bright NK cells represent ∼5–10% of the NK cells in peripheral blood and, upon stimulation with monokines (IL-12, IL-15, and IL-18), produce high levels of cytokines such as IFN-γ (3), which play an important role in regulating immune responses.

Additional differences between CD56dim and CD56bright NK cells have been described. CD56bright NK cells preferentially accumulate in the T cell area of the lymph node, presumably through their selective expression of CCR7 and CD62L (4, 5). The developmental relationship between CD56dim and CD56bright NK cells is not clear, and these two subpopulations of NK cells may represent two developmentally distinct subsets. Alternatively, CD56bright NK cells have been considered to be immature NK precursor cells with a potential to differentiate into CD56dim NK cells (6).

There have been several reports of an increased ratio of CD56bright to CD56dim NK cells at sites of chronic inflammation, such as rheumatoid arthritis, tuberculosis, and pulmonary sarcoidosis (7, 8, 9, 10, 11, 12). It is not clear whether this is due to increased recruitment, proliferation, or generation of CD56bright cells or due to preferential death of CD56dim cells. However, there is some evidence to suggest that the latter alternative may be the case, such as the finding that CD56dim cells preferentially undergo apoptosis upon incubation with tuberculous effusions (12). In addition, tumors are often themselves sites of chronic inflammation, and circulating CD56dim NK cells in cancer patients were shown to be preferentially targeted for apoptosis, leading to low levels of NK activity (13).

Reactive species that give rise to oxidative stress, such as the reactive oxygen species (ROS)4 hydrogen peroxide (H2O2) and the reactive nitrogen species NO, are produced by activated granulocytes and macrophages during inflammation (14, 15, 16). Furthermore, H2O2 may be produced by the tumor cells themselves (17, 18, 19). This study was undertaken to address whether the decreased numbers of cytotoxic CD56dim NK cells seen during chronic inflammation may be due to selective cell death in response to exposure to ROS produced by granulocytes. Our data suggest that CD56dim cells, but not CD56bright cells, are likely to succumb to cell death primarily as a result of exposure to H2O2 produced by inflammatory cells. We further propose that the reason for the difference in sensitivity to H2O2 between CD56dim and CD56bright cells is the lower antioxidative capacity found among CD56dim cells.

Cells were stained using anti-CD3-allophycocyanin (UCHT1), anti-CD56-PE (NCAM 16.2), anti-CD16-FITC (3G8), anti-NKp46-allophycocyanin (9E2/NKp46), and anti-NKG2D-allophycocyanin (1D11). In addition, cells were labeled with 7-aminoactinomycin D (7-AAD) and/or Annexin VFITC according to the manufacturer’s protocol. Viable cells were identified using 2-fold criteria: 1) no shift in the forward scatter/side scatter (FSC/SSC) profile and 2) no staining of 7-AAD and/or annexin V. All Abs and reagents above were obtained from BD Biosciences. To discriminate between sorted NK cells and granulocytes in coculture experiments, granulocytes were labeled with 5 μM CFSE (Invitrogen Life Technologies) for 10 min at 37°C, 7.5% CO2 in RPMI 1640 serum-free medium (Invitrogen Life Technologies) and washed three times in AIM-V medium (Invitrogen Life Technologies). This enabled identification of NK cells (CFSE) and granulocytes (CFSE+) according to their fluorescence intensity in channel I (FL-1). For investigating mitochondrial depolarization, cells were incubated with 20 nM 3,3′-dihexyloxacarbocyanine iodide (DiOC6(3); Invitrogen Life Technologies) in AIM-V for 15 min. at 37°C, 7.5% CO2 in the dark before FACS analysis. A decrease in DiOC6(3) fluorescence intensity in FL-1 corresponds to mitochondrial depolarization. For analyzing intracellular levels of oxidative stress, cells were labeled with 0.25 μM 5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate, acetyl ester (CM-H2DCFDA; Invitrogen Life Technologies) or the oxidized control 5-(and-6)-carboxy-2′,7′-dichlorofluorescein diacetate (Invitrogen Life Technologies) in PBS for 15 min at 37°C, 7.5% CO2 before FACS analysis.

PBMC were isolated from buffy coats from healthy blood donors (Karolinska University Hospital, Solna, Sweden) by density gradient separation through Ficoll-Paque (Amersham Biosciences). PBMC were then washed three times in PBS at 90 × g to remove platelets. NK cells were purified from PBMC by negative selection using the Human NK Isolation kit II (Miltenyi Biotec), according to the manufacturer’s protocol. The purity of total NK cells was routinely >95%. For further separation into CD16CD56bright and CD16+CD56dim subsets, CD16 microbeads (Miltenyi Biotec) were used according to the manufacturer’s protocol. The purity of CD56dim cells was routinely >98% and the purity of CD56bright cells was 47–60%. In some experiments, pure (>95%) CD56dim and CD56bright NK cell populations were isolated by cell sorting using flow cytometry. Purified NK cells were labeled with anti-CD56 and anti-CD3 Abs together with 7-AAD. The cells were then sorted into CD56dim (CD56dim, CD3, 7-AAD) and CD56bright (CD56bright, CD3, 7-AAD) NK cell populations by flow cytometry using a FACSAria (BD Biosciences). After separation, the cells were resuspended at 1 × 106 cells/ml in AIM-V medium (Invitrogen Life Technologies). Cells were treated with gamma-radiation, sodium nitroprusside (SNP; Sigma-Aldrich) or hydrogen peroxide (H2O2, 30% w/w in H2O; Sigma-Aldrich). H2O2 was diluted and SNP solubilized in PBS and used within minutes of preparation. For coculture experiments, granulocytes were isolated from the pellet, containing granulocytes and RBC, after Ficoll-Paque separation of buffy coat. RBC were removed using isotonic lysis. Granulocytes were then washed and labeled with CFSE. Granulocytes were activated by adding PMA (50 ng/ml; Sigma Aldrich) to the granulocytes for 7 min. The cells were then washed three times in AIM-V before coculture with NK cells, thus minimizing the risk of NK cell activation by contaminating PMA. Granulocytes were used at different cell concentrations in coculture experiments with autologous NK cells. In blocking experiments, catalase (1000 U/ml; Sigma-Aldrich), Nω-hydroxy-nor-l-arginine (0.5 mM; Calbiochem/Merck), or NG-monomethyl-l-arginine (0.5 mM final concentration; Calbiochem/Merck) was added to granulocytes before addition of NK cells in the coculture experiments. In all experiments, the treated cells were cultured at 37°C, 7.5% CO2. At various time points (indicated for each experiment), samples were removed, stained, and analyzed by flow cytometry.

Levels of intracellular oxidation were determined by labeling NK cells with CM-H2DCFDA. CM-H2DCFDA is cell permeable, and retention within cells is promoted by CM covalent binding to thiols such as glutathione. Once incorporated into the cell, the acetates are removed by intracellular esterases to yield H2-2′,7′-dichlorofluorescein, which upon oxidation (by various ROS, including H2O2) forms nonpermeable fluorescent dichlorofluorescein (DCF). Because there was a possibility for incorrect compensation settings between DCF (FL-1) and anti-CD56-PE (FL-2), isolated NK cells were in some experiments labeled with CM-H2DCFDA before separation into CD16CD56bright and CD16+CD56dim fractions. The cells were labeled with (0.25 μM) CM-H2DCFDA or oxidized DCFDA in PBS at 37°C for 10 min. The cells were then washed and resuspended in fresh AIM-V. The subsets were cultured in AIM-V with or without addition of H2O2 for 30 min at 37°C, 7.5% CO2, and then either left solely CM-H2DCFDA labeled, or additionally stained with anti-CD56-PE. After verifying that compensation between FL-1 and FL-2 was not interfering with analysis of the data, total NK cells were also analyzed in the manner described, performing staining for CD56 expression along with CM-H2DCFDA labeling.

To measure the total antioxidant levels of NK cell subsets, the ability of cellular lysates to inhibit H2O2-dependent oxidation of homovanillic acid (HVA; Sigma-Aldrich) was determined. HVA is oxidized by HRP (Sigma-Aldrich) and H2O2 to form a highly fluorescent compound. Purified CD16+ (CD56dim) and CD16 (CD56bright) NK cells were lysed at 10 × 106 cells/ml in 50 mM phosphate buffer containing 1 mM EDTA (Sigma- Aldrich) and 1% Triton X-100 (Merck). Eighty microliters of a 1/2 dilution of each sample (cell lysate or PBS) was mixed with 15 μl of reaction mix (30 mM HVA, 1 U/ml HRP) and 5 μl of 6 mM H2O2 (300 μM). The decreased production of oxidized substrate was measured with excitation at 355 nm and detection at 420 nm in a spectrophotometer (Victor2 1420 Multilabel Counter; Wallac/PerkinElmer). Readings were made every 180 s for 30 repeats, and the average of peak repeats was analyzed (typically seven repeats, performed in duplicate).

The statistical analysis was done using the two-tailed paired t test. Brackets in figures indicate which groups were compared and ∗, ∗∗, ∗∗∗ indicate values of p < 0.05, <0.01, and <0.001, respectively.

Several reports have shown increased CD56bright to CD56dim NK cell ratios at sites of chronic inflammation, including cancer (7, 8, 9, 10, 11, 12). Activated granulocytes, associated with chronic inflammation, have been shown to suppress T and NK cell function and viability as well as the expression of NKRs (20, 21, 22, 23). We therefore asked whether the suppressive activity of granulocytes discriminates between the two NK cell subsets, which may explain the “skewed” CD56bright to CD56dim NK cell ratio seen in chronic inflammation. To obtain CD56dim and CD56bright NK cells, purified NK cells were separated using magnetic beads based on their expression of CD16. The CD16+ cell fraction routinely consisted of >98% CD56dim NK cells, whereas the CD16 cell fraction contained only 47–60% CD56bright NK cells and were termed CD56bright-enriched (remaining cells were CD56dim NK cells). When CD56dim NK cells were cocultured for 24 h with PMA-stimulated granulocytes from autologous healthy donors, a significant decrease in viability was found at a granulocyte-to-NK ratio of 0.5:1 (p = 0.025) (Fig. 1,A). A tendency of decreased viability was also seen at a granulocyte-to-NK ratio of 0.25:1, but the difference did not reach statistical significance (p = 0.298). In contrast, the CD56bright-enriched cells proved almost totally resistant to the granulocyte-induced cell death (Fig. 1 A). It should be noted that because the purity of CD56bright-enriched cells was never >60%, the actual difference between the subsets is likely even larger than what was observed. We conclude that granulocytes can induce significant cell death of CD56dim but not CD56bright NK cells. Altogether, this finding may explain the preferential accumulation of the more resistant CD56bright NK cells in various inflammatory conditions where activated granulocytes are abundant. Next, we proceeded to investigate the mechanism responsible for this selective effect.

FIGURE 1.

CD56dim NK cells selectively undergo cell death upon coculture with granulocytes in a H2O2-dependent manner. Granulocytes and NK cells were isolated from buffy coats from healthy blood donors. Granulocytes were labeled with CFSE to discriminate between granulocytes and NK cells after coculture. NK cells were separated into CD16+ (CD56dim) and CD16 (CD56bright-enriched) subsets. A, CD16+ or CD16 NK cells were cocultured for 24 h at the indicated ratios with PMA-stimulated granulocytes. Cell viability of NK cells (CFSE) was determined by flow cytometry using 7-AAD as a cell death marker. Data are presented as the percent of dead cells at 1:0, 1:0.25, and 1:0.5 NK:granulocyte ratio divided by the percent of dead cells without granulocytes (1:0 ratio). B, CD16+ NK cells were incubated with the indicated ratios of PMA-stimulated granulocytes either in the presence or absence of H2O2 scavenger (catalase). The figure shows data from one representative experiment of three experiments.

FIGURE 1.

CD56dim NK cells selectively undergo cell death upon coculture with granulocytes in a H2O2-dependent manner. Granulocytes and NK cells were isolated from buffy coats from healthy blood donors. Granulocytes were labeled with CFSE to discriminate between granulocytes and NK cells after coculture. NK cells were separated into CD16+ (CD56dim) and CD16 (CD56bright-enriched) subsets. A, CD16+ or CD16 NK cells were cocultured for 24 h at the indicated ratios with PMA-stimulated granulocytes. Cell viability of NK cells (CFSE) was determined by flow cytometry using 7-AAD as a cell death marker. Data are presented as the percent of dead cells at 1:0, 1:0.25, and 1:0.5 NK:granulocyte ratio divided by the percent of dead cells without granulocytes (1:0 ratio). B, CD16+ NK cells were incubated with the indicated ratios of PMA-stimulated granulocytes either in the presence or absence of H2O2 scavenger (catalase). The figure shows data from one representative experiment of three experiments.

Close modal

Activated granulocytes are known to produce a variety of molecules with a broad range of biological activities (see, for example, Ref. 24). An analysis of the underlying molecular mechanism behind their ability to induce cell death in the CD56dim NK cell subset was therefore undertaken. The ability of PMA-activated granulocytes to kill CD56dim NK cells was completely abrogated when catalase was present during the coculture period (Fig. 1 B). In contrast, preincubation of the granulocytes with the inducible NO synthase inhibitor NG-monomethyl-l-arginine or the arginase inhibitor Nω-hydroxy-nor-l-arginine had no or only marginal effects on the viability of this NK cell subset (data not shown). Thus, H2O2 plays a dominant role in granulocyte-mediated killing of CD56dim NK cells.

Our results clearly indicated a dominant role for H2O2 in the observed cytotoxicity mediated by granulocytes on the CD56dim NK cells. We therefore next performed a detailed analysis on the effect of H2O2 on this subset, analyzing the sensitivity of total PBMC or isolated NK cells to H2O2 as a model for exposure to ROS (Fig. 2,A). Activated granulocytes can produce H2O2 at concentrations in the range of 5–50 μM (25), and we used a similar range of concentrations for our experiments to mimic physiological levels of H2O2 exposure. With increasing concentration of H2O2, the relative proportion of CD56dim among total live NK cells decreased, as most of the CD56bright NK cells were alive whereas a large proportion of CD56dim NK cells were dead. At concentrations of 10 μM and above, the difference in proportion of live CD56dim cells between untreated (0 μM) and treated NK cells reached statistical significance. Preferential survival of CD56bright cells (p = 0.02 at 20 μM H2O2) was also seen when NK cells were separated into CD56dim and CD56bright subsets using flow cytometry-based cell sorting (>95% purity of each subset) and then exposed to H2O2 (Fig. 2 B), similar to what was found following exposure to granulocytes as detailed above. To test whether H2O2, besides inducing death of CD56dim cells, may also have a direct effect on proliferation of CD56bright cells, CFSE-labeling experiments were performed. There was no induction of proliferation by H2O2 in our model system (data not shown). Thus, CD56dim cells were sensitive whereas CD56bright cells were largely resistant to cell death induced by physiological levels of H2O2 produced by activated granulocytes or added directly to the cells.

FIGURE 2.

CD56dim NK cells undergo cell death upon exposure to low levels of H2O2. A, PBMC (×) or purified NK cells (▵) were incubated in the presence of increasing concentrations of H2O2 for 24 h and stained with anti-CD3 and anti-CD56. The percent of CD56dim cells of total live NK cells (CD56+, CD3, no FSC/SSC shift and 7-AAD) are shown. Each line represents a separate donor and experiment and the bracket surrounding the p value indicates at what H2O2 concentration differences in viability between treated and nontreated cells reach statistical significance. B, NK cells were isolated and labeled with anti-CD56, anti-CD3 Abs, and 7-AAD. Pure (>95%) CD56dim and CD56bright NK populations were then separated by flow cytometry-based cell sorting. Cells were incubated in the absence or presence of increasing concentrations of H2O2 for 24 h. Dot plots indicating the percent of dead (7-AAD+) cells in each purified subset are shown. Results from one representative experiment of three are shown.

FIGURE 2.

CD56dim NK cells undergo cell death upon exposure to low levels of H2O2. A, PBMC (×) or purified NK cells (▵) were incubated in the presence of increasing concentrations of H2O2 for 24 h and stained with anti-CD3 and anti-CD56. The percent of CD56dim cells of total live NK cells (CD56+, CD3, no FSC/SSC shift and 7-AAD) are shown. Each line represents a separate donor and experiment and the bracket surrounding the p value indicates at what H2O2 concentration differences in viability between treated and nontreated cells reach statistical significance. B, NK cells were isolated and labeled with anti-CD56, anti-CD3 Abs, and 7-AAD. Pure (>95%) CD56dim and CD56bright NK populations were then separated by flow cytometry-based cell sorting. Cells were incubated in the absence or presence of increasing concentrations of H2O2 for 24 h. Dot plots indicating the percent of dead (7-AAD+) cells in each purified subset are shown. Results from one representative experiment of three are shown.

Close modal

The finding of selective death of CD56dim NK cells by H2O2 prompted us to explore whether this was restricted to H2O2 exposure, or whether other cell death-inducing factors would also preferentially kill this subset. Total PBMC were exposed to H2O2, the NO donor SNP, or gamma-radiation and the percent of CD56dim cells among the total live NK cells was determined under conditions where the overall lymphocyte cell death was 50–75%. There was a clear difference in relative sensitivity to cell death of CD56dim cells depending on the stimulus used. Exposure to H2O2 resulted in a significant decrease in the percent of CD56dim NK cells (median 92.7% without and 76.3% with H2O2 exposure) in the live lymphocyte population (Fig. 3). A minor decrease in the percent of CD56dim NK cells was also seen in cells exposed to SNP (median 94.6% without and 89.2% with SNP exposure). In contrast, CD56bright cells preferentially underwent cell death as a result of exposure to gamma-radiation. These data show that the CD56dim cells are not sensitive to apoptosis in a nondiscriminate fashion, hence supporting a unique role for H2O2 in preferentially inducing cell death of this subset of NK cells.

FIGURE 3.

CD56dim NK cells display a unique susceptibility to H2O2 exposure. PBMC were isolated from donor buffy coats and cells were exposed to increasing concentrations of H2O2, the NO donor SNP, or gamma-radiation for 24 h. First, the total lymphocyte cell death was measured by FSC/SSC shift. To be able to compare the different cell death induction stimuli, the percent of CD56dim cells of live NK cells was analyzed when 50–75% total lymphocyte cell death was observed. Data from 10 (H2O2 and SNP) and 12 (gamma-radiation) independent experiments are shown.

FIGURE 3.

CD56dim NK cells display a unique susceptibility to H2O2 exposure. PBMC were isolated from donor buffy coats and cells were exposed to increasing concentrations of H2O2, the NO donor SNP, or gamma-radiation for 24 h. First, the total lymphocyte cell death was measured by FSC/SSC shift. To be able to compare the different cell death induction stimuli, the percent of CD56dim cells of live NK cells was analyzed when 50–75% total lymphocyte cell death was observed. Data from 10 (H2O2 and SNP) and 12 (gamma-radiation) independent experiments are shown.

Close modal

Recent data has shown that the CD56dim, but not the CD56bright, NK subset down-regulates NKp46 and NKG2D upon exposure to ROS (21). We therefore asked to what extent these findings were related to the high sensitivity to H2O2 observed in the CD56dim NK subset in the present study, and also whether other cell surface receptors in this subset may be down-regulated by H2O2. NK cells were separated into CD56dim- and CD56bright-enriched subsets and treated with H2O2 for a period of 15 h. Among CD56bright-enriched cells gated for CD56bright expression neither cell death nor down-regulation of cell surface receptor expression was seen following H2O2 exposure. However, among the viable CD56dim cells, we could confirm down-regulation of the NKp46 receptor (Fig. 4,A). In contrast, CD16 was not down-regulated as a result of H2O2 treatment (Fig. 4 A). Down-regulation of NKRs was not a general phenomenon, because NKG2A was expressed to the same extent on H2O2-treated and untreated cells, whether live or dead (data not shown). These data show that H2O2 treatment, besides triggering cell death of CD56dim NK cells, also results in down-regulation of specific NK cell surface receptors.

FIGURE 4.

NKp46 is down-regulated on H2O2-treated CD56dim cells before cell death in a manner that parallels mitochondrial membrane polarization. NK cells were purified from buffy coat and separated into CD16+ (CD56dim) and CD16 (CD56bright-enriched) subsets. Results from one representative experiment of three are shown. A, Cells were untreated (gray line) or treated with 20 μM H2O2 (black line) for 15 h and stained with anti-CD16 or anti-NKp46 ab, and 7-AAD. Live (no FSC/SSC shift and 7-AAD) CD16+ or CD16 cells were analyzed for expression of NKp46 or CD16. B, CD16+ NK cells (>95% CD56dim) were cultured in the absence or presence of 20 μM H2O2 for the indicated times. Cells were stained for NKp46 expression along with DiOC6 (3 ), as a marker for intact mitochondrial membrane potential. C, Cells shown in B stained with 7-AAD as a marker for cell death.

FIGURE 4.

NKp46 is down-regulated on H2O2-treated CD56dim cells before cell death in a manner that parallels mitochondrial membrane polarization. NK cells were purified from buffy coat and separated into CD16+ (CD56dim) and CD16 (CD56bright-enriched) subsets. Results from one representative experiment of three are shown. A, Cells were untreated (gray line) or treated with 20 μM H2O2 (black line) for 15 h and stained with anti-CD16 or anti-NKp46 ab, and 7-AAD. Live (no FSC/SSC shift and 7-AAD) CD16+ or CD16 cells were analyzed for expression of NKp46 or CD16. B, CD16+ NK cells (>95% CD56dim) were cultured in the absence or presence of 20 μM H2O2 for the indicated times. Cells were stained for NKp46 expression along with DiOC6 (3 ), as a marker for intact mitochondrial membrane potential. C, Cells shown in B stained with 7-AAD as a marker for cell death.

Close modal

Next, we asked whether the down-regulation of NKp46 in the CD56dim, but not the CD56bright, NK subset upon H2O2 exposure was related to the induction of cell death in these NK subsets. We therefore examined mitochondrial membrane depolarization as an early marker for cells targeted for apoptosis. Depolarization of CD56dim cell mitochondrial membrane potential could be seen starting as early as 1 h after H2O2 exposure, peaking at ∼6 h (Fig. 4,B). NKp46 expression was reduced in a manner that closely paralleled that of mitochondrial membrane depolarization (Fig. 4,B). In contrast, cell death as determined by 7-AAD staining was absent in a majority of CD56dim cells at 6 h, and was not detected to a large extent until after 15 h of H2O2 exposure (Fig. 4 C). Our results show that there was an early down-regulation of NKp46 on the CD56dim NK subset, preceding cell death.

Several possible mechanisms may account for the observation that CD56dim NK cells are more sensitive to H2O2 as compared with CD56bright NK cells. We hypothesized that this may be due to differences in levels of cellular antioxidants. To directly test this possibility, we stained NK cells with CM-H2DCFDA and exposed the cells to H2O2, followed by assessment of levels of DCF fluorescence as a global measurement of oxidative stress (26). CD56dim NK cells showed a higher level of intracellular oxidization when compared with the CD56bright NK cell subset even in the absence of H2O2, and this difference was even more pronounced upon exposure to H2O2. Even though the level of oxidation increased in CD56bright NK cells upon H2O2 exposure, the level did not exceed that seen in untreated CD56dim NK cells (Fig. 5). To verify that this observation was not due to preferential uptake of the probe by CD56dim NK cells, oxidized DCFDA was used as above. When the NK cells were labeled with oxidized DCFDA, there was no difference in fluorescence between the NK cell subsets. To further control for a possible artifact due to compensation settings used in the flow cytometric analysis, the two NK subsets were separated by CD16 expression into CD56dim and CD56bright-enriched fractions after labeling with CM-H2DCFDA, and were then cultured in the presence or absence of H2O2. The difference in DCF staining was still evident when comparing the enriched NK cell subsets not labeled by any other means (data not shown).

FIGURE 5.

Intracellular oxidative levels are higher and antioxidant capacity is lower in CD56dim NK cells as compared with CD56bright NK cells. A, NK cells were isolated, labeled with CM-H2DCFDA, and cultured in the presence or absence of 20 μM H2O2 for 30 min. Cells were stained for CD56 expression. DCF levels among gated CD56dim (black line) and CD56bright (gray line) cells are shown to the right. B, Purified NK cells or total PBMC were labeled with CM-H2DCFDA and cultured in the absence or presence of 20 μM H2O2 for 30 min. DCF levels among untreated and treated cells are shown. Data from eight independent experiments are shown, along with p values for the differences between the various groups. C, NK cells were purified and separated into CD16+ and CD16 fractions. Equal numbers of cells were pelleted and lysed. The ability of the lysate to prevent H2O2-dependent oxidation of substrate was determined. Oxidation is proportional to fluorescence in this assay. Results from one representative experiment of four are shown.

FIGURE 5.

Intracellular oxidative levels are higher and antioxidant capacity is lower in CD56dim NK cells as compared with CD56bright NK cells. A, NK cells were isolated, labeled with CM-H2DCFDA, and cultured in the presence or absence of 20 μM H2O2 for 30 min. Cells were stained for CD56 expression. DCF levels among gated CD56dim (black line) and CD56bright (gray line) cells are shown to the right. B, Purified NK cells or total PBMC were labeled with CM-H2DCFDA and cultured in the absence or presence of 20 μM H2O2 for 30 min. DCF levels among untreated and treated cells are shown. Data from eight independent experiments are shown, along with p values for the differences between the various groups. C, NK cells were purified and separated into CD16+ and CD16 fractions. Equal numbers of cells were pelleted and lysed. The ability of the lysate to prevent H2O2-dependent oxidation of substrate was determined. Oxidation is proportional to fluorescence in this assay. Results from one representative experiment of four are shown.

Close modal

To confirm these observations, cellular lysates from purified CD56dim- and CD56bright-enriched NK cells were analyzed for their antioxidative ability in a fluorometric assay. Cell lysates from CD56bright NK cells significantly reduced the level of HRP-mediated oxidation of the substrate to a greater extent than those of CD56dim NK cells (Fig. 5 C). Thus, the difference in sensitivity to H2O2 between CD56dim and CD56bright NK cells is likely due to decreased levels of intracellular oxidation in the CD56bright NK cells, most likely as a result of increased levels of cellular antioxidants.

In this study, we report that CD56dim NK cells are highly sensitive to activated granulocytes through a mechanism involving H2O2, while CD56bright NK cells are almost completely resistant. The H2O2-exposed CD56dim NK cells showed down-regulation of the activating NKR NKp46 at an early time point, which correlated with mitochondrial membrane depolarization. The difference in sensitivity to H2O2 between CD56dim and CD56bright NK cells may be related to the higher “baseline” levels of oxidative stress observed in CD56dim NK cells and the higher antioxidative capacity of cell lysates from this subset. Collectively, our data suggest that CD56dim NK cells present in an environment where H2O2 is produced, such as various inflammatory conditions, rapidly lose expression of receptors required for lytic function and then selectively undergo cell death.

We noticed rapid mitochondrial depolarization in the CD56dim NK cells, followed several hours later by annexin V and 7-AAD staining as well as by a shift in FSC/SSC profile consistent with cellular condensation. Cell death has classically been described as either apoptotic or necrotic, but more recent information has shown that a clear distinction between these two types of cell death is not always possible (for reviews, see Refs.27, 28, 29). NK cell death induced by physiological levels of H2O2 is characterized by late caspase activation, DNA fragmentation, cellular condensation, and maintenance of intracellular ATP (data not shown and Refs. 23 and 30), more resembling apoptotic than necrotic cell death. To address the underlying mechanism behind the preferential cell death of CD56dim NK cells, we investigated the antioxidative capacity of the two NK subsets. Indeed, we and others have previously shown that death of total NK cells, typically consisting of a majority of CD56dim cells, can be blocked by agents that scavenge H2O2, such as serotonin and catalase (31, 32, 33).

Selective down-regulation of NKp46 and NKG2D on CD56dim NK cells after exposure to ROS has recently been reported (21), although the correlation with eventual apoptosis of the CD56dim NK cells was not investigated in this system. Our results demonstrate that the decrease in NKp46 expression strongly correlates with H2O2 treatment. Down-regulation of NKp46 occurs in a manner that closely parallels that of mitochondrial membrane depolarization, several hours before cell death. Because NKp46 is down-regulated very early, this seems to be an active process that would presumably quickly limit CD56dim NK cells’ ability to lyse NKp46 ligand-expressing targets, such as tumor cells (34, 35). H2O2 exposure has been shown to inhibit NK cytotoxic ability against K562 cells (31, 33, 36), and signaling through NKp46 is very important for lysis of K562 cells by freshly isolated NK cells (35). Down-regulation of NKp46 and other natural cytotoxicity receptors has been seen in NK cells from patients suffering from acute myeloid leukemia (37), chronic hepatitis C infection (38), and active HIV infection (39). NKp46 associates with the ITAM-containing signaling molecules CD3ζ and FcεRI (40, 41, 42). CD3ζ chain has also been shown to be down-regulated in NK cells upon H2O2 exposure (33, 43), and a substance that up-regulates glutathione levels can protect from H2O2-mediated NK cell CD3ζ chain down-regulation (43). Thus, it seems possible that there is a connection between expression of CD3ζ and expression of NKp46.

As a new and unexpected finding, our study shows that there is a difference in “baseline” oxidative levels between CD56dim and CD56bright NK cells even without exogenous exposure to H2O2, as measured by CM-H2DCFDA conversion to DCF, and by the ability to block H2O2-mediated oxidation of HVA. We would thus propose that a small oxidative insult would be enough to push CD56dim NK cells into oxidative stress, whereas CD56bright NK cells have a buffer zone that allows them to absorb some level of oxidative insult without harm. It is not clear what form of ROS H2O2 is converted to during the in vitro culture, but it seems likely that the cells are not only exposed to H2O2, because H2O2 and O2 do not oxidize H2-2′,7′-dichlorofluorescein directly, whereas peroxyl alcohol, NO2, carbonate (CO3) and OH· radicals do, as does peroxynitrite (26). Because NO does not preferentially induce cell death of CD56dim NK cells (Fig. 3), it is likely that the oxidative stress resulting from H2O2 exposure acts on one or several specific substrates, which remain to be identified.

In addition, the measurement of antioxidant ability of cellular lysates from CD56dim and CD56bright NK cells lends further support to the notion that differences in antioxidative function could explain the difference in sensitivity to H2O2 between these two NK subsets. Recently, expression of genes in CD56dimCD16+ and CD56brightCD16 NK cell subsets was analyzed using a gene array approach (44). Several genes found to be differentially expressed between CD56dim and CD56bright NK cells may be candidate antioxidants, as inferred from this study. In particular, paraoxonase 2, which has antioxidant ability (45), and lysozyme, which has been shown to be protective against oxidative stress by H2O2 and other oxidants (46), are two factors that were up-regulated at least 6-fold in CD56bright NK cells (paraoxonase 2 expression was below the limit of detection in CD56dim cells). Furthermore, glutathione peroxidase 1, an enzyme that catalyzes the reduction of H2O2 and organic hydroperoxides using glutathione, was found to be expressed to a higher extent in CD56bright cells (2.5-fold).

In conclusion, our data support a model where CD56dim NK cells have a high and selective sensitivity to H2O2 exposure due to low antioxidant ability. It is important to note, however, that our data do not distinguish between an effect on CD56dim NK cells mediated by “oxidative damage” caused by the externally added H2O2 on cellular components or, alternatively, by H2O2-triggering signaling events in the cell leading to receptor down-regulation and eventual cell death. H2O2 at low levels is known to act as an important “second messenger” during lymphocyte activation (see, for example, Ref. 47), although the exact molecular pathway involved has not been conclusively determined. H2O2 has been suggested to have both proliferative and apoptotic effects on T cells through various signaling pathways (48, 49, 50). Higher levels of antioxidant ability in the CD56dim NK cells could thus either have a direct effect through oxidative damage, or alternatively act through indirect activation of signaling pathways triggering NK cell death. Studies in progress will determine which one is the most important.

Our findings may be of a direct biological relevance to explain functional consequences in pathological conditions of inflammation, where activated granulocytes, monocytes, or myeloid-derived suppressor cells accumulate, leading to enhanced ROS levels. In these environments, our data suggest that CD56dim NK cells would rapidly undergo mitochondrial membrane depolarization, accompanied by down-regulation of activating NKRs such as NKp46, and eventually undergo cell death. This may explain why CD56bright NK cells, with their relative insensitivity to low levels of H2O2, accumulate at disproportional levels at the cost of CD56dim NK cells during chronic inflammation (8, 9, 10, 11, 12). In addition, our data may explain the poor lytic function of NK cells frequently found in the tumor microenvironment (13).

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was supported by grants to R.K. from the Swedish Cancer Society, the Cancer Society of Stockholm, the European Union, the Karolinska Institutet, an “ALF-Project” grant from the Stockholm City Council, and National Institutes of Health Grant CA102280. K.-J.M. was supported by the Royal Swedish Academy of Sciences, the Cancer Society of Stockholm, the Swedish Children’s Cancer Foundation, and the Tobias Foundation.

4

Abbreviations used in this paper: ROS, reactive oxygen species; 7-AAD, 7-aminoactinomycin D; FSC, forward scatter; SSC, side scatter; CM-H2DCFDA, 5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate, acetyl ester; SNP, sodium nitroprusside; DCF, fluorescent dichlorofluorescein; HVA, homovanillic acid; DiOC6(3), 3,3′-dihexyloxacarbocyanine iodide.

1
Allavena, P., G. Damia, T. Colombo, D. Maggioni, M. D’Incalci, A. Mantovani.
1989
. Lymphokine-activated killer (LAK) and monocyte-mediated cytotoxicity on tumor cell lines resistant to antitumor agents.
Cell. Immunol.
120
:
250
-258.
2
Oshimi, K., Y. Oshimi, M. Akutsu, Y. Takei, H. Saito, M. Okada, H. Mizoguchi.
1986
. Cytotoxicity of interleukin 2-activated lymphocytes for leukemia and lymphoma cells.
Blood
68
:
938
-948.
3
Cooper, M. A., T. A. Fehniger, S. C. Turner, K. S. Chen, B. A. Ghaheri, T. Ghayur, W. E. Carson, M. A. Caligiuri.
2001
. Human natural killer cells: a unique innate immunoregulatory role for the CD56bright subset.
Blood
97
:
3146
-3151.
4
Frey, M., N. B. Packianathan, T. A. Fehniger, M. E. Ross, W. C. Wang, C. C. Stewart, M. A. Caligiuri, S. S. Evans.
1998
. Differential expression and function of L-selectin on CD56bright and CD56dim natural killer cell subsets.
J. Immunol.
161
:
400
-408.
5
Berahovich, R. D., N. L. Lai, Z. Wei, L. L. Lanier, T. J. Schall.
2006
. Evidence for NK cell subsets based on chemokine receptor expression.
J. Immunol.
177
:
7833
-7840.
6
Nagler, A., L. L. Lanier, S. Cwirla, J. H. Phillips.
1989
. Comparative studies of human FcRIII-positive and negative natural killer cells.
J. Immunol.
143
:
3183
-3191.
7
Batoni, G., S. Esin, F. Favilli, M. Pardini, D. Bottai, G. Maisetta, W. Florio, M. Campa.
2005
. Human CD56bright and CD56dim natural killer cell subsets respond differentially to direct stimulation with Mycobacterium bovis bacillus Calmette-Guérin.
Scand. J. Immunol.
62
:
498
-506.
8
Dalbeth, N., M. F. Callan.
2002
. A subset of natural killer cells is greatly expanded within inflamed joints.
Arthritis Rheum.
46
:
1763
-1772.
9
Dalbeth, N., R. Gundle, R. J. Davies, Y. C. Lee, A. J. McMichael, M. F. Callan.
2004
. CD56bright NK cells are enriched at inflammatory sites and can engage with monocytes in a reciprocal program of activation.
J. Immunol.
173
:
6418
-6426.
10
Katchar, K., K. Soderstrom, J. Wahlstrom, A. Eklund, J. Grunewald.
2005
. Characterisation of natural killer cells and CD56+ T-cells in sarcoidosis patients.
Eur. Respir. J.
26
:
77
-85.
11
Pridgeon, C., G. P. Lennon, L. Pazmany, R. N. Thompson, S. E. Christmas, R. J. Moots.
2003
. Natural killer cells in the synovial fluid of rheumatoid arthritis patients exhibit a CD56bright, CD94bright, CD158negative phenotype.
Rheumatology
42
:
870
-878.
12
Schierloh, P., N. Yokobori, M. Aleman, R. M. Musella, M. Beigier-Bompadre, M. A. Saab, L. Alves, E. Abbate, S. S. de la Barrera, M. C. Sasiain.
2005
. Increased susceptibility to apoptosis of CD56dimCD16+ NK cells induces the enrichment of IFN-γ-producing CD56bright cells in tuberculous pleurisy.
J. Immunol.
175
:
6852
-6860.
13
Bauernhofer, T., I. Kuss, B. Henderson, A. S. Baum, T. L. Whiteside.
2003
. Preferential apoptosis of CD56dim natural killer cell subset in patients with cancer.
Eur. J. Immunol.
33
:
119
-124.
14
Klebanoff, S. J..
1980
. Oxygen metabolism and the toxic properties of phagocytes.
Ann. Intern. Med.
93
:
480
-489.
15
MacMicking, J., Q. W. Xie, C. Nathan.
1997
. Nitric oxide and macrophage function.
Annu. Rev. Immunol.
15
:
323
-350.
16
Ohshima, H., M. Tatemichi, T. Sawa.
2003
. Chemical basis of inflammation-induced carcinogenesis.
Arch. Biochem. Biophys.
417
:
3
-11.
17
Brar, S. S., T. P. Kennedy, A. B. Sturrock, T. P. Huecksteadt, M. T. Quinn, A. R. Whorton, J. R. Hoidal.
2002
. An NAD(P)H oxidase regulates growth and transcription in melanoma cells.
Am. J. Physiol.
282
:
C1212
-C1224.
18
del Bello, B., A. Paolicchi, M. Comporti, A. Pompella, E. Maellaro.
1999
. Hydrogen peroxide produced during γ-glutamyl transpeptidase activity is involved in prevention of apoptosis and maintenance of proliferation in U937 cells.
FASEB J.
13
:
69
-79.
19
Szatrowski, T. P., C. F. Nathan.
1991
. Production of large amounts of hydrogen peroxide by human tumor cells.
Cancer Res.
51
:
794
-798.
20
Berg, L., J. Ronnelid, L. Klareskog, A. Bucht.
2000
. Down-regulation of the T cell receptor CD3 ζ chain in rheumatoid arthritis (RA) and its influence on T cell responsiveness.
Clin. Exp. Immunol.
120
:
174
-182.
21
Romero, A. I., F. B. Thoren, M. Brune, K. Hellstrand.
2006
. NKp46 and NKG2D receptor expression in NK cells with CD56dim and CD56bright phenotype: regulation by histamine and reactive oxygen species.
Br. J. Haematol.
132
:
91
-98.
22
Schmielau, J., O. J. Finn.
2001
. Activated granulocytes and granulocyte-derived hydrogen peroxide are the underlying mechanism of suppression of T-cell function in advanced cancer patients.
Cancer Res.
61
:
4756
-4760.
23
Thoren, F. B., A. I. Romero, K. Hellstrand.
2006
. Oxygen radicals induce poly(ADP-ribose) polymerase-dependent cell death in cytotoxic lymphocytes.
J. Immunol.
176
:
7301
-7307.
24
Sampson, A. P..
2000
. The role of eosinophils and neutrophils in inflammation.
Clin Exp Allergy
30
: (Suppl. 1):
22
-27.
25
Test, S. T., S. J. Weiss.
1984
. Quantitative and temporal characterization of the extracellular H2O2 pool generated by human neutrophils.
J. Biol. Chem.
259
:
399
-405.
26
Halliwell, B., M. Whiteman.
2004
. Measuring reactive species and oxidative damage in vivo and in cell culture: how should you do it and what do the results mean?.
Br. J. Pharmacol.
142
:
231
-255.
27
Hetz, C. A., V. Torres, A. F. Quest.
2005
. Beyond apoptosis: nonapoptotic cell death in physiology and disease.
Biochem. Cell. Biol.
83
:
579
-588.
28
Lockshin, R. A., Z. Zakeri.
2004
. Caspase-independent cell death?.
Oncogene
23
:
2766
-2773.
29
Proskuryakov, S. Y., A. G. Konoplyannikov, V. L. Gabai.
2003
. Necrosis: a specific form of programmed cell death?.
Exp. Cell. Res.
283
:
1
-16.
30
Saito, Y., K. Nishio, Y. Ogawa, J. Kimata, T. Kinumi, Y. Yoshida, N. Noguchi, E. Niki.
2006
. Turning point in apoptosis/necrosis induced by hydrogen peroxide.
Free Radic. Res.
40
:
619
-630.
31
Betten, A., C. Dahlgren, S. Hermodsson, K. Hellstrand.
2001
. Serotonin protects NK cells against oxidatively induced functional inhibition and apoptosis.
J. Leukocyte Biol.
70
:
65
-72.
32
Hansson, M., A. Asea, U. Ersson, S. Hermodsson, K. Hellstrand.
1996
. Induction of apoptosis in NK cells by monocyte-derived reactive oxygen metabolites.
J. Immunol.
156
:
42
-47.
33
Kono, K., F. Salazar-Onfray, M. Petersson, J. Hansson, G. Masucci, K. Wasserman, T. Nakazawa, P. Anderson, R. Kiessling.
1996
. Hydrogen peroxide secreted by tumor-derived macrophages down-modulates signal-transducing ζ molecules and inhibits tumor-specific T cell-and natural killer cell-mediated cytotoxicity.
Eur. J. Immunol.
26
:
1308
-1313.
34
Pessino, A., S. Sivori, C. Bottino, A. Malaspina, L. Morelli, L. Moretta, R. Biassoni, A. Moretta.
1998
. Molecular cloning of NKp46: a novel member of the immunoglobulin superfamily involved in triggering of natural cytotoxicity.
J. Exp. Med.
188
:
953
-960.
35
Sivori, S., D. Pende, C. Bottino, E. Marcenaro, A. Pessino, R. Biassoni, L. Moretta, A. Moretta.
1999
. NKp46 is the major triggering receptor involved in the natural cytotoxicity of fresh or cultured human NK cells: correlation between surface density of NKp46 and natural cytotoxicity against autologous, allogeneic or xenogeneic target cells.
Eur. J. Immunol.
29
:
1656
-1666.
36
Betten, A., C. Dahlgren, U. H. Mellqvist, S. Hermodsson, K. Hellstrand.
2004
. Oxygen radical-induced natural killer cell dysfunction: role of myeloperoxidase and regulation by serotonin.
J. Leukocyte Biol.
75
:
1111
-1115.
37
Costello, R. T., S. Sivori, E. Marcenaro, M. Lafage-Pochitaloff, M. J. Mozziconacci, D. Reviron, J. A. Gastaut, D. Pende, D. Olive, A. Moretta.
2002
. Defective expression and function of natural killer cell-triggering receptors in patients with acute myeloid leukemia.
Blood
99
:
3661
-3667.
38
Nattermann, J., G. Feldmann, G. Ahlenstiel, B. Langhans, T. Sauerbruch, U. Spengler.
2006
. Surface expression and cytolytic function of natural killer cell receptors is altered in chronic hepatitis C.
Gut
55
:
869
-877.
39
De Maria, A., M. Fogli, P. Costa, G. Murdaca, F. Puppo, D. Mavilio, A. Moretta, L. Moretta.
2003
. The impaired NK cell cytolytic function in viremic HIV-1 infection is associated with a reduced surface expression of natural cytotoxicity receptors (NKp46, NKp30 and NKp44).
Eur. J. Immunol.
33
:
2410
-2418.
40
Cantoni, C., C. Bottino, M. Vitale, A. Pessino, R. Augugliaro, A. Malaspina, S. Parolini, L. Moretta, A. Moretta, R. Biassoni.
1999
. NKp44, a triggering receptor involved in tumor cell lysis by activated human natural killer cells, is a novel member of the immunoglobulin superfamily.
J. Exp. Med.
189
:
787
-796.
41
Letourneur, O., I. C. Kennedy, A. T. Brini, J. R. Ortaldo, J. J. O’Shea, J. P. Kinet.
1991
. Characterization of the family of dimers associated with Fc receptors (FcεRI and FcγRIII).
J. Immunol.
147
:
2652
-2656.
42
Vitale, M., C. Bottino, S. Sivori, L. Sanseverino, R. Castriconi, E. Marcenaro, R. Augugliaro, L. Moretta, A. Moretta.
1998
. NKp44, a novel triggering surface molecule specifically expressed by activated natural killer cells, is involved in non-major histocompatibility complex-restricted tumor cell lysis.
J. Exp. Med.
187
:
2065
-2072.
43
Corsi, M. M., H. H. Maes, K. Wasserman, A. Fulgenzi, G. Gaja, M. E. Ferrero.
1998
. Protection by L-2-oxothiazolidine-4-carboxylic acid of hydrogen peroxide-induced CD3ζ and CD16ζ chain down-regulation in human peripheral blood lymphocytes and lymphokine-activated killer cells.
Biochem. Pharmacol.
56
:
657
-662.
44
Hanna, J., P. Bechtel, Y. Zhai, F. Youssef, K. McLachlan, O. Mandelboim.
2004
. Novel insights on human NK cells’ immunological modalities revealed by gene expression profiling.
J. Immunol.
173
:
6547
-6563.
45
Ng, C. J., D. J. Wadleigh, A. Gangopadhyay, S. Hama, V. R. Grijalva, M. Navab, A. M. Fogelman, S. T. Reddy.
2001
. Paraoxonase-2 is a ubiquitously expressed protein with antioxidant properties and is capable of preventing cell-mediated oxidative modification of low density lipoprotein.
J. Biol. Chem.
276
:
44444
-44449.
46
Liu, H., F. Zheng, Q. Cao, B. Ren, L. Zhu, G. Striker, H. Vlassara.
2006
. Amelioration of oxidant stress by the defensin lysozyme.
Am. J. Physiol.
290
:
E824
-E832.
47
Reth, M..
2002
. Hydrogen peroxide as second messenger in lymphocyte activation.
Nat. Immunol.
3
:
1129
-1134.
48
Devadas, S., L. Zaritskaya, S. G. Rhee, L. Oberley, M. S. Williams.
2002
. Discrete generation of superoxide and hydrogen peroxide by T cell receptor stimulation: selective regulation of mitogen-activated protein kinase activation and Fas ligand expression.
J. Exp. Med.
195
:
59
-70.
49
Gulow, K., M. Kaminski, K. Darvas, D. Suss, M. Li-Weber, P. H. Krammer.
2005
. HIV-1 trans-activator of transcription substitutes for oxidative signaling in activation-induced T cell death.
J. Immunol.
174
:
5249
-5260.
50
Kwon, J., S. Devadas, M. S. Williams.
2003
. T cell receptor-stimulated generation of hydrogen peroxide inhibits MEK-ERK activation and lck serine phosphorylation.
Free Radic. Biol. Med.
35
:
406
-417.