Tuberculous pleuritis is a good model for the study of specific cells at the site of active Mycobacterium tuberculosis (Mtb) infection. We investigated the frequency and phenotype of NK cells in paired samples of peripheral blood and pleural fluid (PF) from patients with tuberculosis (TB) or parapneumonic infection. We demonstrated for the first time a reduction of NK cells in PF from TB with an enrichment in the CD56brightCD16− subset. In agreement, in PF NK cells we observed an increased expression of CD94, NKG2A, CD62L, and CCR7 molecules and lower expression of Bcl-2 and perforin. The activation markers CD69 and HLA-DR were also increased. The enrichment in the CD56bright subset was due to an increased susceptibility to apoptosis of CD56+CD16+ NK cells mediated by heat-labile and stable soluble factors present in tuberculous effusions and not in PF from other etiologies. Furthermore, in TB patients, Mtb-induced IFN-γ production by PF NK cells was not dependent on the presence of CD3+, CD19+, and CD14+ cells, suggesting a direct interaction of CD56bright cells with Mtb and/or the involvement of other accessory cells present at the site of Mtb infection.
Tuberculous pleuritis is caused by a severe delayed-type hypersensitivity reaction in response to the rupture of a subpleural focus of Mycobacterium tuberculosis (Mtb)3 infection. Although tuberculous pleurisy occurs in ∼10% of untreated individuals who test positive by the tuberculin test, it may also develop as a complication of primary pulmonary tuberculosis (TB) (1). The inflammatory process results in an increased pleural vascular permeability leading to the accumulation of fluid enriched in proteins and the recruitment of specific leukocytes into the pleural space (2, 3, 4, 5). Pleural fluid (PF) from TB patients usually shows a lymphocytic preponderance (1, 4, 6). Among lymphocytes, NK cells display an important number of effector functions, including recognition and lysis of infected, stressed, or transformed cells and production of immunoregulatory cytokines, particularly IFN-γ (7, 8, 9, 10). Human NK cells account for 10–20% of PBLs and are defined by the presence of the CD56 molecule lacking CD3 expression (11). Their activity is regulated by both positive and inhibitory signals from a wide range of cell surface receptors (12, 13), including members of the C-type lectin family, such as CD94 and NKG2A, -B, -C, and -D; members of the Ig superfamily, such as the killer Ig-like receptors (KIRs) and the leukocyte Ig-like receptors; as well as natural cytotoxicity receptors, cytokine receptors, and costimulatory molecules (14, 15, 16, 17). Alternatively, recent data suggest that NK cells can recognize microorganisms by themselves and be directly activated by microbially encoded molecules via several pathogen recognition receptors (18, 19).
Two subsets of NK cells have been recognized in peripheral blood (PB). More than 90% of circulating NK cells express moderate levels of CD56 and high levels of CD16 (CD56dim NK cells) that usually express KIRs and are heterogeneous in the expression of CD94 and NKG2A molecules. The minor subset of NK cells expresses high levels of CD56 (CD56bright), CD94, and NKG2A and tends to lack expression of CD16 (11). These two subsets also differ in terms of chemokine receptor and adhesion molecules expression, suggesting that they have different homing properties (20, 21). Indeed, the NK CD56bright cells have been found to be enriched in human secondary lymphoid organs (22, 23, 24) and in chronic inflammatory sites (25). In addition, the NK cell subsets show important functional differences, the CD56dim subset has superior cytotoxic potential and differential KIR repertory (26), whereas the CD56bright subset has greater ability to produce cytokines upon exposure to monokines (10, 23, 27), contact-dependent interaction with mature dendritic cells (28, 29, 30), or T cell-derived IL-2 (22, 31). However, the underlying developmental relationship between these two subsets remains controversial (32, 33, 34, 35).
Considering that tuberculous pleurisy is an intense immune response to mycobacteria that results in the clearance of organisms from the pleural space, we undertook to delineate the role of NK cells at the site of active Mtb infection. In this study we demonstrate for the first time a selective enrichment of CD56bright cells due to an increased apoptosis of CD56dimCD16+ NK cells mediated by a heat-stable and labile component(s) present in tuberculous pleural effusions. Moreover, activation of CD56bright cells and IFN-γ production are observed in tuberculous pleurisy in the absence of CD3+, CD19+, and CD14+ cells.
Materials and Methods
Patients with newly diagnosed moderately sized and large pleural effusions (PE) were identified at the Servicio de Tisioneunología, Hospital F. J Muñiz (Buenos Aires, Argentina). Informed consent was obtained from patients according to the ethics committee of Hospital Francisco J. Muñiz. After written informed consent for study participation was obtained, patients were evaluated by a history and physical examination; complete blood cell count; electrolyte; chest x-ray; and HIV testing. PE and blood were obtained from the patients during diagnostic thoracentesis before the initiation of chemotherapy. Exclusion criteria included a positive test for HIV or the presence of concurrent infectious diseases. A total of 28 PE were studied, 21 were tuberculous, 5 were uncomplicated parapneumonia (PNI), and 2 were malignant. Among tuberculous PE, 13 had also pulmonary disease, which was classified according to the extent and type of chest x-ray findings into moderate (n = 5), advanced (n = 6), and miliary (n = 2). The effusions were classified as exudates according to at least one of the Light criteria (3). Tuberculous effusions were defined as exudates with a positive Ziehl-Nielsen stain or Lowestein-Jensen culture of PE or pleural biopsy specimens. A PE was considered parapneumonic effusion when there was an acute febrile illness, with purulent sputum and pulmonary infiltrates, in the absence of malignancy or other diseases causing exudates and neutrophilia in PE. Malignant effusions were defined as exudates associated with a pathologic diagnosis of cancer whenever cytologic malignant cells were observed during histopathologic examination of pleura biopsy.
Thoracentesis and pleural biopsy
PE from hospitalized patients with TB (n = 21; average age, 40 years; range, 20–60 years), PNI (n = 5; range, 22–60 years), and cancer (n = 2; 62 and 68 years) was collected by therapeutic thoracentesis. Briefly, after local anesthesia of the skin and s.c. tissue, 100 ml of PF was aspirated under sterile conditions using an 18-gauge needle (Abrams needle), which minimizes contamination of the PE with PB. A specimen was subject to routine biochemical analysis, including test for total protein, glucose, lactate dehydrogenase, and differential cell counting. Bacterial cultures and cytological examinations were performed on all PF in the central laboratory of Hospital Muñiz. A second sample of the PE was dispensed into 50-ml polystyrene tubes (Corning) containing heparin to obtain mononuclear cells. Cells from closed pleural biopsy from two TB patients were also obtained. In addition, blood samples were collected from patients on the same day as thoracentesis.
PB and PF mononuclear cells were isolated by Ficoll-Hypaque gradient centrifugation and thereafter suspended in RPMI 1640 tissue culture medium (Invitrogen Life Technologies) containing gentamicin (85 μg/ml) and 10% heat-inactivated FCS (Invitrogen Life Technologies; complete medium). Leukocytes were obtained by mechanical resuspension from pleural biopsy samples. Suspended cells were washed twice in PBS and thereafter stained and subjected to flow cytometric analysis. Purity and viability were tested using trypan blue exclusion.
Isolation of NK cells from PF and PB
Purified NK or CD16-depleted cells were obtained by magnetic methods. Briefly, to obtain NK cells, PF (10–20 × 106 cells in pellet) were treated for 30 min at 4°C with purified anti-CD14, anti-CD19 (Immunotech), anti-CD3 (clone 145-2C11), and anti-CD16 (clone 3G8) Abs for depletion of CD16 cells from PB. Thereafter, goat anti-mouse IgG-coated beads (Dynal Biotech) were added. In general, one cycle of treatment was sufficient for an effective depletion (<5% for CD3+ and 1% for CD16/14/19+ cells) as assessed by flow cytometry. In some cases, lymphocytes were isolated from total mononuclear cells by Percoll gradient centrifugation (29, 30).
The gamma-irradiated Mtb H37-Rv strain used in this study was provided by Dr. J. Belisle (Colorado University, Denver, CO). Mycobacteria were suspended in PBS free of pyrogen, sonicated, and adjusted at a concentration of 1 × 108 bacteria/ml (OD600 nm = 1).
Expression of surface markers on CD3−CD56+ lymphocytes.
The following mAbs were used to evaluate the expression of surface markers on CD3−CD56+ on freshly isolated, cultured PB and PF or CD3+/CD14+/CD19+-depleted PF: Cy5PE-anti-human CD3 (eBioscience), PE-anti-human CD56 (Immunotech), FITC-anti-human CD56 (eBioscience), FITC-anti-CD16 (Ancell), PE-anti-CD16 (BD Pharmingen), anti-CD69 (Ancell), anti-CD62L (eBioscience), anti-CD94 (Ancell), and anti-HLA-DR (eBioscience). To evaluate the expression of NKG2A and CCR7 molecules on NK cells, PB and PF were stained using a two-step staining with purified anti-CD56 (Ancell) and FITC-anti-mouse IgG F(ab′)2 sheep Ab (Sigma-Aldrich) and direct staining with Cy5PE-anti-CD3 (eBioscience), PE-anti-NKGD2A (R&D Systems), or PE-anti-CRR7 (BD Pharmingen) Abs. Labeled isotype-matched Abs were also tested to evaluate nonspecific staining. Stained cells were analyzed in a FACScan cytometer using CellQuest software (BD Biosciences) by acquiring 50,000 events from PB and 100,000–150,000 events from PF samples. Analysis gates were set on lymphocytes according to forward and side scatter properties, excluding cell debris and apoptotic cells. Results are expressed as the percentage of positive cells.
Apoptosis of NK cells.
To determine NK cells apoptosis, control and Mtb-stimulated PB lymphocytes and CD16+-depleted PB cells (1 × 106 cells/ml) were cultured in the presence or the absence of various concentrations of cell-free PF (percentage v/v in complete medium) or recombinant human IL-18 (10 ng/ml), IL-12 (10 ng/ml), and TNF-α (10 ng/ml; all from R&D Systems) for 24–72 h. Thereafter, cells were stained for surface CD3 and CD56 or CD3 and CD16 expression and washed, and apoptosis was assessed based on the annexin V (AV)-FITC protein binding assay, according to the manufacturer’s instructions without the addition of propidium iodide (APO-AF kit; Sigma-Aldrich). The percentage of AV+ was evaluated on CD3−CD56+ and CD3−CD16+/− lymphocytes.
Measurements of IFN-γ+ and IL-10+ NK cells.
To measure the expression of intracytoplasmic IFN-γ or IL-10 in control and Mtb-stimulated CD3−CD56+ cells, PF and PB were cultured 24 h in 12 × 75-mm polystyrene, round-bottom tubes (BD Biosciences) with or without Mtb. Brefeldin A (5 μg/ml; Sigma-Aldrich) was added for the final 4 h to block cytokine secretion. Cells were washed and stained for surface CD3 and CD56 expression for 20 min at 4°C. Thereafter, the cells were fixed according to the manufacturer’s instructions (IntraPrep permeabilization reagent; Immunotech), washed with PBS supplemented with 1% FCS and 0.01% azide (PBS-FCS-azide), and suspended in 100 μl of PBS-FCS-azide. Fluorescein-conjugated Ab for IFN-γ or IL-10 (Caltag Laboratories) was added together with 100 μl of permeabilizing solution (IntraPrep), incubated for 30 min at 4°C, washed with PBS-FCS-azide, suspended in Isoflow (BD Biosciences), and analyzed by flow cytometry. Forty thousand to 100,000 events were acquired for each sample, and results are expressed as the percentage of IFN-γ- or IL-10-positive cells in the CD3−CD56+ NK cell population.
Expression of intracellular proteins on CD3−CD56+ lymphocytes.
Freshly isolated PF and PB cells were surface stained with CD3 and CD56, and intracellular perforin and Bcl-2 staining was performed as described above. FITC-conjugated anti-perforin, anti-Bcl-2, and isotype-matched mAb were used (all from Ancell).
PB and PF data from each group were analyzed using the Wilcoxon (nonparametric) paired test. Data from different groups of patients were analyzed using the Mann-Whitney (nonparametric) unpaired test. A value of p < 0.05 was considered significant.
Predominance of CD56bright NK subset in TB pleurisy
Although tuberculous PEs are characterized by a T lymphocyte preponderance, other cells from the innate immune response, such as Tγδ lymphocytes, neutrophils, NK cells, and APCs, may be present at this inflammatory site, contributing to the mounting of a specific immune response against Mtb (2, 6, 36, 37). Considering that the early production of IFN-γ by NK cells at the inflammatory sites can activate APC for IL-12 production, shaping adaptive immunity toward a Th1 response (38, 39), we investigated the frequency and phenotype of NK cells on exudative PF and PB, comparing patients with TB and PNI.
As shown in Fig. 1,A, the frequency of CD3−CD56+ cells (NK cells) was significantly reduced in PF from all TB patients studied compared with PB (n = 21), whereas no significant differences were observed in the percentage of PB and PF NK cells in PNI patients. When the phenotype of NK subsets was evaluated, we observed that the majority of PB NK cells (>96% in all samples) expressed intermediate levels of CD56, consistent with the predominant CD56dim expression in circulating human NK cells from healthy donors (11). However, a selective enrichment of the CD56bright subset was observed in PF from either TB patients or PNI (Fig. 1, B and C). In addition, analysis of mononuclear cells obtained from TB pleural biopsy (n = 2; data not shown) showed a preponderance of CD56bright NK cells.
To better characterize the NK subsets from TB patients, we analyzed several molecules that are known to be differentially expressed (11). As shown in Fig. 2,A, the majority of PF NK cells expressed high levels of CD94, NKG2A, and CD62L molecules, and they also expressed CCR7 (n = 5), whereas lower levels of PF NK cells from TB patients expressed the CD16 molecule (n = 10; Fig. 2, A and B). Considering that the expressions of both CD69 and HLA-DR molecules on NK cells are associated with their activation (29, 35), we examined whether these two receptors were up-regulated on PF NK cells. As shown in Fig. 2,A, increased numbers of PF NK cells expressing the early activation marker CD69 (n = 10) and HLA-DR (n = 7) were observed in TB patients. Furthermore, the expression of perforin on PF NK cells was practically absent (Fig. 2 C), which is in accordance with the low cytolytic potential of the CD56bright subset and the granule-independent role of NK cells in TB (9, 26). In PNI, the percentage of cells that expressed CD94, NKG2A, CD62L, and CCR7 was as high as that in PF NK cells from TB patients (data not shown). In addition, we detected an up-regulation of HLA-DR expression in PF NK cells from PNI patients, and in agreement with a previous report (25), the CD69 activation marker was increased (data not shown).
Tuberculous PF (TB-PF) induces apoptosis of NK cells
Considering the lower levels of NK cells in tuberculous PEs we postulated that apoptosis of NK cells might be taking place at the site of Mtb infection. To test this, we first determined the expression of the antiapoptotic molecule Bcl-2 on PF and PB NK cells from TB. PF NK cells from TB patients showed a significant reduction in the expression of Bcl-2 (n = 4; p < 0.01) compared with PB NK cells, suggesting that the tuberculous pleural environment may make NK cells prone to undergo apoptosis (Fig. 3,A). However, neither PB nor PF NK cells showed differences in the level of Bcl-2 expression by CD56bright and CD56dim. Therefore, PBMC from TB patients and healthy individuals were incubated with cell-free PF from tuberculous effusions (TB-CFPF) with or without Mtb for 48 h, and AV+ cells were determined on CD3−CD56+ cells. In addition, given that cytokines present in TB-PF may induce NK cell death (32, 33), we tested whether apoptosis of NK cells could be triggered in vitro by IL-12, IL-18, and/or TNF-α and whole Mtb. Taking into account the levels of IL-12, IL-18, and TNF-α found in TB effusions (5, 40, 41), these sets of experiments were conducted in the presence of higher levels of cytokines after 24 h of incubation. As shown in Fig. 3,B, neither cytokine nor the presence of Mtb induced NK cell apoptosis as TB-CFPF did (diluted 50%, v/v). Furthermore, TB-CFPF induced NK cell apoptosis in a dose-dependent manner (Fig. 3,C). To investigate whether the apoptotic factors present in TB-CFPF were heat labile or stable, TB-CFPF was heated (-ØCFPF) for 30 min at 60°C and AV+ cells were determined on PB (n = 3) after 48 h of culture, as described above. As shown in Fig. 3 D, although both heated and nonheated CFPF (n = 6; 100%, v/v) increased NK cell apoptosis, a significant decrease was observed with heated compared with nonheated CFPF, demonstrating that the apoptotic effect is mediated by heat-labile as well as stable soluble factors present in CFPF from TB. The apoptotic effect on PB NK cells was variable among CFPF from different TB patients even though all of them induced NK cell death (NK AV+ range, 38–80; 100% v/v; 48 h).
Because TB-CFPF induced NK cell apoptosis, we asked whether the apoptotic effect was specific for TB effusions. To test this, apoptosis of NK cells was determined on the same PB sample in the presence of PNI, cancer, or TB effusions, as described above. As shown in Fig. 4,A, neither PNI nor cancer effusions induced apoptosis of NK cells, whereas TB-CFPF did; thus, they were taken as one group (No TB-CFPF). Furthermore, although heated TB-CFPF induced NK apoptosis, heated No TB-CFPF did not (Fig. 4,B). To determine whether NK cell apoptosis was an early or a late event, PB were incubated with CFPF and heated CFPF (100%, v/v) from both TB and No TB for 24–72 h, and AV+ cells were determined. As shown in Fig. 4,C, the level of NK AV+ cells increased up to 48 h of incubation with tuberculous CFPF. Heated TB-CFPF also induced early apoptosis, but after 24 h of incubation was not able to increase the percentage of AV+ cells, suggesting that the increase in apoptosis is also dependent on a heat-labile factor(s) at later time points. In addition, no significant differences in NK cell apoptosis were observed between heated and pronase-treated TB-CFPF (3 U/ml for 1 h at 37°C; data not shown). Furthermore, the reduction in CD3−CD56+ among viable lymphocytes corroborates the results presented above (Fig. 4 C). Taken together, these results demonstrate that NK cell apoptosis is mediated by protein heat-labile and nonprotein heat-stable factors only present in PF from TB.
CD16+ NK cells are more susceptible to apoptosis induced by TB-CFPF
Given that TB-CFPF induced NK apoptosis, we wanted to determine whether a differential susceptibility to apoptosis from NK subsets might explain the reduction of CD56dimCD16+ cells in tuberculous pleural effusions. To address this, we first incubated PB in the presence or the absence of TB-CFPF for 48 h, and AV+ cells on CD3−CD16+ and CD3−CD16− cells were determined on gated lymphocytes. As shown in Fig. 5,A, the percentage of AV+ CD3−CD16+ cells was significantly increased compared with that of AV+ CD3−CD16− cells. Given that the CD16 expression on CD3− cells was down-regulated along the culture period in the presence of TB-CFPF (Fig. 5,B) and to avoid misevaluation of the percentage of apoptosis among CD3−CD56+ due to the loss of CD16 receptor, PB were depleted of CD16 cells before the assay, and apoptosis on CD3−CD56+ cells was determined in whole and CD16-depleted PB. As we expected, the depletion of CD16+ cells reduced apoptosis on CD3−CD56+ cells, as shown in Fig. 5, C and D. Together, these results demonstrate an increased susceptibility of CD16+ NK cells to undergo apoptosis induced by soluble factors present in TB-CFPF.
PF NK cells from TB patients produce IFN-γ independent of CD14+, CD19+, and CD3+ cells
Taking into account that CD56bright NK cells are producers of cytokines, we evaluated next their ability to produce proinflammatory (IFN-γ) or anti-inflammatory (IL-10) cytokines. To test this, paired samples of PB and PF from TB (n = 9) with or without pulmonary involvement were cultured with or without Mtb for 24 h, and intracellular IFN-γ and IL-10 expression was determined by flow cytometry. As shown in Fig. 6,A, spontaneous and Mtb-induced IFN-γ expressions were detected in PF cells, whereas neither spontaneous nor Mtb-induced IL-10 expression was observed in PB or PF NK cells (Fig. 6,A). Considering that T cell-derived IL-2 (22) and APC interaction (25, 28) allowed IFN-γ production by NK cells, TB-PF cells (n = 5) were depleted of CD3+, CD19+, and CD14+ populations, and Mtb-induced IFN-γ was determined. As shown in Fig. 6 B, IFN-γ production by PF NK cells was not affected by the depletion, indicating a T/CD19+/CD14+ cell-independent mechanism.
Finally, we addressed whether an association could be established between the expression of IFN-γ and the severity of the pulmonary disease or the etiology. Thus, paired samples of PB and PF NK cells from five moderate and two miliary TB patients and two PNI were evaluated. As shown in Fig. 7, the severity of pulmonary disease in TB patients appeared to be inversely correlated with spontaneous IFN-γ expressed by PF NK cells, but no differences were observed in Mtb-induced IFN-γ expression. In particular, Mtb induced IFN-γ expression in almost all PF CD56bright NK cells from all TB patients analyzed independent of pulmonary involvement. In addition, PF NK cells from PNI patients produced less spontaneous IFN-γ, and Mtb did not induce its expression. Although the IFN-γ production was evaluated in only two PNI effusions, the different behaviors may be ascribed not only to the presence of Mtb, but also to the microenvironment of both effusions. Taken together, these results suggest an in vivo priming of NK cells at the site of Mtb infection involving specific recognition of mycobacterial Ags.
A protective immune response against Mtb depends on IFN-γ production by CD4+ and CD8+ T cells (36, 42). However, the early production of IFN-γ by cells of the innate immune response, such as NK cells, within the inflammatory sites can regulate innate resistance by activating phagocytic cells and priming APC for IL-12 production, thus shifting adaptive immunity toward a Th1 response (38, 39). Although the role of NK cells in mycobacterial infections is unclear, in vitro studies have demonstrated that NK cells from PB may contribute to a protective immunity through the production of IFN-γ to maintain the frequency of Mtb-responsive CD8+ IFN-γ+ and to expand the cytotoxic activity (43). In addition, NK cells may mediate early killing of intracellular Mtb via alternative apoptotic pathways (9).
In this study we have demonstrated that NK cells are reduced at the site of Mtb infection with an enrichment in the CD56bright subset. Although in PNI this NK subset was also predominant, the frequency of total NK cells at inflammatory site was the same as that in PB as previously reported (25), suggesting that the nature of the infection, chronic or acute, may influence the level of NK cells present in PF. In contrast to PNI, the loss of CD16+ NK cells was more pronounced in PF from TB patients along with undetectable perforin expression, suggesting that these cells do not have the capacity to lyse Mtb-infected target cells. To encounter possible explanations for the accumulation of the CD56bright NK subset, we hypothesized that cytokines present in PF may induce a differentiation or apoptosis of NK cells at the site of tuberculous infection. However, in our hands, neither IL-12 nor TNF-α induced the differentiation from CD16+ to CD16− NK cells (data not shown), which is in accordance with the report by Dalbeth et al. (25). In addition, neither whole Mtb nor IL-12, IL-18, and/or TNF-α stimulation induced apoptosis of NK cells. Surprisingly, TB-PF per se induced an increased apoptosis of CD56dimCD16+ cells, which may explain the predominance of the CD56bright population. Indeed, this effect was not dependent on the type and time course of effusion, because neither cancer (chronic) nor parapneumonic (acute) induced apoptosis of NK cells. Moreover, the apoptotic effect was partially lost by heating PF, suggesting that immune mediators and/or heat-labile and stable Mtb Ags are capable of inducing CD56dimCD16+ apoptosis. In this context, the activation state, ligand-induced apoptosis via CD16 signaling (44, 45), and the loss of Bcl-2 expression that we have found in PF NK cells together with soluble Fas ligand and other TNF family ligands present in TB-CFPF may be factors related to the increased apoptosis of CD56dimCD16+ cells from TB (46, 47, 48). In contrast, CD94 expression in almost all CD56bright populations observed in TB-PF may explain their survival, because its expression correlates with the survival of CD8+ T and NK cells (49). Thus, TB-PF has a pro- or antiapoptotic effect depending on the cytokine milieu, the cell populations present at the site of infection, and the bacterial Ags. In fact, neutrophil apoptosis is dependent on Mtb contact (2), whereas in T cells and mononuclear phagocytes, it is not (46). Therefore, although whole Mtb does not induce apoptosis of NK cells, we cannot exclude the involvement of soluble protein and nonprotein Mtb Ags released into tuberculous effusions, which could also induce apoptosis. This is a matter of current investigation.
NK cells circulate in the blood, but are largely excluded from lymph nodes under -steady state conditions. However, some circulating NK cells express lymph node homing receptors, CD62L and CCR7, and high CD56-expressing noncytolytic NK cells have been identified in human peripheral lymph nodes (21, 22). In contrast to the PB counterpart, where CD56bright NK cells express the CCR7 and CD62L molecules, CD56bright NK cells found in the secondary lymphoid organs lose the expression of these homing molecules (24), consistent with their migratory pattern. However, in this study we show that PF CD56bright cells express CD62L but only a minor proportion express CCR7, suggesting that cytokines present in the PF may modulate their expression (22) or that PF CD56bright cells maintain the capacity to migrate to lymph nodes (38). Indeed, CD56brightCD16− NK cells from inflamed lymph nodes acquire CD69 and HLA-DR activation markers (24), as do CD56bright cells from TB pleurisy. Together our results indicate that once they arrive at the site of infection, PB CD56+ CD16+ cells are more susceptible to undergoing apoptosis, leading to an enrichment in activated CD56bright cells that maintain homing receptors for recruitment to lymph nodes (23, 24, 38, 50).
Because NK cells from TB effusions were activated, we evaluated whether these cells were capable of producing IFN-γ and IL-10, which may play a role in the pathogenesis of TB. In this study we have demonstrated that IFN-γ was produced spontaneously by PF CD56bright cells from both TB and PNI patients, although to a lesser extent in the latter. In addition, the spontaneous production of IFN-γ appears to be associated with the severity of the pulmonary disease in TB patients. However, Mtb-induced IFN-γ was only detected in CD56bright cells from TB effusions and showed no differences when considering the severity of the disease. Neither PF nor PB NK cells from TB and PNI infections produced IL-10. Given that CD56bright cells recognized Mtb without the presence of T, B, and CD14+ cells, it suggests a direct mechanism for Mtb recognition and/or the involvement of other accessory cells, such as CD14− cells present in TB pleurisy. In fact, the Ag receptors involved in signaling for IFN-γ production by PF NK cells in TB pleurisy remain a subject of ongoing investigation.
In conclusion, in this study we demonstrate that when peripheral NK cells extravasate to the site of TB infection, the CD56dimCD16+ subset has an increased susceptibility to undergo apoptosis induced by heat-stable/labile mediators present in tuberculous effusions, leading to an enrichment in CD56bright cells. These NK cells do not have the capacity of lysing Mtb-infected target cells, but are larger producers of IFN-γ upon Mtb contact, contributing to regulation of the immune response toward the Th1 profile. The IFN-γ production is not dependent on T/B/CD14 cells present in PF, however; other APC may be involved. Moreover, because CD56bright cells retain homing receptors for lymph nodes, they maintain their capacity to migrate to them. Therefore, these findings may explain why tuberculous pleurisy results in the clearance of Mtb and resolves without treatment.
We thank the medical staff of División de Tisioneumonología, Hospital F. J. Muñiz, for their great help in providing clinical samples from patients. We are also grateful to Dr. Christiane Dosne de Pasqualini for critical manuscript review.
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.
This work was supported by grants from Agencia Nacional de Promoción Científica y Tecnológica (05-14060) and Consejo Nacional de Investigaciones Cientificas y Tecnologicas.
Abbreviations used in this paper: Mtb, Mycobacterium tuberculosis; AV, annexin V; KIR, killer Ig-like receptor; PB, peripheral blood; PE, pleural effusion; PF, pleural fluid; PNI, parapneumonic infection; TB, tuberculosis; TB-CFPF, cell-free PF from tuberculous effusion; TB-PF, tuberculous PF.