NK and DC reciprocal interactions have only recently been investigated. In this study, we focused on the interplay between NK cells and DC in two models of bacterial infection. Immature monocyte-derived DC were cultured in the presence of live Escherichia coli or bacillus Calmette–Guérin. Upon exposure to either extracellular or intracellular bacteria, DC underwent maturation as assessed by the increased levels of expression of CD80, CD86, and HLA molecules and the de novo expression of CD83 and CCR7. Significant amounts of TNF-α and IL-12 were released by DCupon infection, whereas IL-2 and IL-15 were barely detectable in culture supernatants. Both infected and uninfected DC were capable of inducing in fresh autologous NK cells the expression of CD69 and HLA-DR and of inducing cell proliferation. Remarkably, however, infected DC were much stronger inducers of NK cell activation and proliferation than uninfected DC. Thus, after just 24 h of NK/DC coculture, only those NK cells that had been exposed to bacteria-infected DC had acquired the ability to lyse autologous immature DC. In addition, infected DC were more resistant to NK-mediated lysis as a consequence of the up-regulation of HLA class I molecule expression on their surface. This study suggests a regulatory circuit involving NK cells and DC in which DC-induced NK cell activationis effectively enhanced by the presence of pathogens. Activated NK cells, by limiting the supply of immature DC, may then exert a control on subsequent innate and adaptive immune responses.
NK cells are generally thought to play a major role in host defense because of their ability to sense and remove cells that have lost, or express insufficient amounts of, MHC class I molecules1. This situation frequently occurs in tumor cells or in cells infected by certain viruses. The molecular mechanisms that allow NK cells to discriminate between normal and MHC class I deficient cells have been elucidated 2, 3. NK cells express a number of inhibitory receptors specific for MHC class I molecules and the interaction of NK cells with normal autologous cells (expressing normal levels of MHC class I molecules) leads to NK cell inactivation.
However, a remarkable exception is represented by DC, which get targeted by NK cells despite significant levels of MHC class I 4–6. In addition, it has recently been demonstrated, in an in vivo model, that infusion of alloreactive NK cells (i.e. characterized by a mismatch between KIR and HLA class I) kill recipient DC, while sparing normal nonmyeloid tissues 7. This further indicates that DC have a preferential susceptibility to NK-mediated lysis. This feature is of particular interest since DC constitute potent APC, whose role is crucial for the initiation and maintenance of an immune response. Located in various peripheral tissues in an immature form, DC are particularly efficient in capturing Ag suchas pathogens and their products as well as dying cells that are present in the surrounding environment. Immature DC (iDC) also release proinflammatory cytokines and chemokines, while they are still inefficient in Ag presentation. Upon maturation, induced by different pathogen derived products and/or various cytokines, DC up-regulate a number of costimulatory molecules, express particularly high levels of MHC molecules and become highly efficient in Ag presentation 8–10. During the maturation process DC also decrease their capability to endocytose, down-regulating the surface receptors involved in antigen capture 11, and change their pattern of chemokine receptors, in order to migrate to secondary lymphoid organs 12, 13.
Recent studies have explored DC/NK cross-talk, demonstrating that this interaction can also lead to NK cell proliferation and induction of cytolytic activity 14–17. In particular, it has been shown that DC enhance NK cell activation (as measured by CD69 up-regulation or cytotoxic activity against tumor cell targets) but only in the presence of inflammatory stimuli 15. Similarly, other authors reported that the DC:NK ratio is crucial in determining the outcome of this interaction. Thus, when DC are added in surplus, DC undergo maturation and NK cells become activated; conversely, when NK cells outnumber DC, the latter are effectively killed by NK cells 16. Our group provided evidence that in addition to CD69 up-regulation and enhanced cytotoxicity, the interaction with DC promoted the expansion of resting NK cells. In addition, we showed that activated NK cells are able to kill DC by the preferential use of a defined triggering pathway initiated via NKp30, a cytotoxicity activating receptor expressed on the NK cell surface 17.
Since DC/NK interactions are likely to occur primarily during infection, in this study we analyzed the effect of live bacteria on the cross-talk between DC and NK cells. We showed that the induction of NK cell proliferation and function by DC is greatly enhanced in the presence of live bacteria. In addition, DC infected by bacteria up-regulate HLA class I molecules, become resistant to NK-mediated lysis and promptly express CCR7, thus acquiring the ability to migrate to secondary lymphoid organs in order to drive the adaptive immune response.
2.1 Infection with live bacteria induces DC maturation
Maturation of DC and their cytokine secretion were evaluated upon exposure to living bacteria. In agreement with previous data, costimulatory molecules, such as CD80 and CD86, and HLA class II molecules were greatly up-regulated 18, 19. In addition, a de novo expression of both CD83 and CCR7 could be consistently detected (Fig. 1). It is of note that CCR7 is a chemokine receptor essential for DC migration to secondary lymphoid organs 20. The increased expression, or the de novo appearance, of these molecules occurred rapidly and peaked between 24 and 48 h after DC infection. We also analyzed the secretion of TNF-α, IL-2, IL-12 and IL-15 in the culture supernatants under the same experimental conditions. Upon bacillus Calmette–Guérin (BCG) infection, TNF-α was released 4 h after infection although the highest concentration was reached after 12 h (Fig. 2). IL-12 started to be released after 12 h and peaked at 48 h. The release of high amounts of IL-12 by DC upon exposure to BCG is of particular interest since this cytokine is described to play a major role in the activation of NK cell cytotoxicity. Under these experimental conditions IL-2 and IL-15 were virtually undetectable in culture supernatants (Fig. 2).
2.2 Autologous DC-induced proliferation of NK cells is enhanced by BCG
We have recently shown that DC are capable of inducing proliferation of autologous NK cells 17. We further investigated whether the infection of DC with BCG had any effect on this capability. BCG was employed as infective agent because of its ability to efficiently infect DC without undergoing substantial proliferation (thus not interfering with [3H]thymidine incorporation assays). In this study, we confirmed that DC, derived from monocytes in the presence of GM-CSF and IL-4, were able to induce NK cell proliferation (Fig. 3). It is of note that both the NK cell proliferation (Fig. 3, panel A) and the number of viable NK cells recovered after 5 d of culture (panel B) were significantly increased in the presence of BCG infection of DC. Control experiments with culture containing NK cells and BCG alone did not lead to NK cell proliferation.
2.3 Effectiveness of BCG and Escherichia coli in inducing an activated NK cell phenotype
Recent data have shown that, following stimulation with LPS, DC induce the expression, in NK cells, of the early activation marker CD69 15, 16. We analyzed whether DC that had been exposed to living bacteria could induce activation markers on NK cells. Fig. 4 shows that the expression of CD69 on NK cell surface was significantly increased. In addition, a de novo expression of CD25 could be detected. This may be functionally relevant since the expression of CD25, a component of the IL-2-receptor complex, renders NK cells highly responsive to IL-2. HLA-DR was also up-regulated. It is of note that although CD69 and HLA-DR can be detected after interaction of NK cells with uninfected iDC, DC infection resulted in a greater increase of the expression of both molecules on the NK cell surface. Since the NKp30 receptor is primarily involved in DC recognition and lysis by NK cells 17, we also analyzed whether mAb-mediated masking of NKp30 could interfere with the NK cell activation induced by DC and bacteria. As shown in Fig. 4, the addition of anti-NKp30 mAb did not modify the expression of CD69, HLA-DR and CD25. Neither BCG nor E. coli could induce the expression of these activation markers in the absence of DC.
2.4 NK cells activated by DC and bacteria efficiently lyse autologous iDC
As previously shown, activated NK cells can lyse autologous iDC although they are less effective against mature DC 6–17. In addition, a short-term coculture of resting NK cells with DC that had been pulsed with LPS or with heat-killed Mycobacterium tuberculosis has been reported to result in increase of NK-mediated cytotoxicity against the Daudi target cell line 15. We further investigated whether NK cells cocultured with DC that had been infected with live bacteria could lyse autologous iDC. Fig. 5 shows that, after 24 h of coculture, only NK cells cultured in the presence of infected DC could lyse autologous iDC (panels B and C). After this time interval, uninfected iDC failed to induce NK cell cytotoxicity (panel A). Remarkably, infected DC were less susceptible to the lysis, as compared with iDC cultured in the absence of bacteria. These results were already obtained after just 24 h of NK/DC coculture.
We further analyzed whether polyclonal NK cell lines cultured for over a week in IL-2 could also discriminate between infected and uninfected autologous DC. As shown in panel D of Fig. 5, IL-2-cultured NK cells lysed uninfected iDC very efficiently, but were less effective against infected DC. Therefore, the resistance of infected DC to NK-mediated lysis does not depend upon the degree of NK cell activation, but rather reflects an intrinsic property of infected DC themselves.
2.5 The resistance of infected DC to NK-mediated cytotoxicity is due to the up-regulation of HLA class I molecules
As shown in previous studies, the different susceptibility of iDC versus mature DC is primarily related to differences in surface expression of HLA class I molecules 6. In order to determine whether the early resistance of infected DC to NK-mediated lysis reflected a rapid up-regulation of HLA class I molecules, we comparatively analyzed, on a quantitative basis, the surface density of HLA class I on DC cultured alone or in the presence of either BCG or E. coli. Thus, we evaluated by flow cytometry the cell size and the expression of HLA class I molecules by direct immunofluorescence. By the simultaneous evaluation of cell size and fluorescence intensity we could calculate the number of HLA class I molecules per μ2 of DC cell surface. Forty-eight h after infection with BCG or E. coli, DC increased the number of HLA class I molecules per μ2 by approximately 10-fold (Fig. 6). In view of the high expression of HLA class I in infected DC, it appears conceivable that resistance to NK-mediated lysis could be a distinct consequence of this phenomenon. On the other hand, another possible explanation could be that BCG and E. coli could down-regulate the expression of the ligands for triggering receptors of NK cells. In order to discriminate between these two possibilities, we performed cytolytic tests in the presence of mAb (IgM isotype) specific for HLA class I molecules. As shown in Fig. 6, upon mAb-mediated masking of HLA class I molecules, a sharp increase of cytolytic activity against infected DC could be detected. This clearly indicates that the resistance of infected DC to NK-mediated lysis is due to increased inhibitory interactions occurring between HLA class I on DC and inhibitory receptors expressed on NK cells. The anti-HLA class I mAb-induced restoration of the cytolytic activity and the inhibition of this activity by anti-NKp30 mAb (Fig. 6) clearly indicate that infected DC also express levels of ligands for this triggering NK receptor sufficient to induce NK cell activation.
In the present study, we analyzed the effect of the interaction between NK cells and DC that had been exposed to live bacteria. We show that bacterial infection of DC can lead to a particularly rapid NK cell activation. These activated NK cells efficiently lyse autologous iDC. On the other hand, DC exposed to bacteria become resistant to NK-mediated cytotoxicity because of a marked up-regulation of HLA class I molecules.
Limited information is available on the direct effect of living bacteria on DC, in particular in the course of their interaction with NK cells, which constitute another major player of innate immunity. The latter information is rather important because it may shed light on cellular interactions occurring during innate responses against pathogens in the course of infection. To this end, we analyzed E. coli, i.e. a classical gram-negative extracellular bacterium, and BCG, a typical intracellular bacterium capable of efficiently infecting DC 21–23. Our present data clearly demonstrate that exposure of DC to either pathogen could rapidly induce DC maturation and expression of functionally important surface molecules, including CD80 and CD86 coreceptors, HLA molecules and CCR7. Thus, DC acquire rapidly the ability to efficiently function as professional APC and to migrate to secondary lymphoid organs, where they can interact with T cells and evoke a prompt adaptive immune response against infecting bacteria. In addition, DC that had encountered bacteria could also rapidly and markedly potentiate an important effector arm of innate immunity by inducing a rapid activation of NK cells.
One may ask how DC and NK cells can meet each other since, under normal conditions, NK cells do not home to tissues, lymphatic vessels and lymph nodes. However, in the course of inflammatory responses, several chemokines are released that address NK cells to inflamed tissues and lymph nodes 24–26. Thus, it is conceivable that the phenomena described in the present study may indeed occur in vivo during bacterial infections. In previous studies, we and others showed that NK cells can efficiently kill autologous iDC 4–6. In this context, the arrival of NK cells in inflamed tissues and their encounter with iDC may appear paradoxical, as it would lead to depletion of APC. Nevertheless, our present data demonstrate that DC exposed to bacteria become highly resistant to NK-mediated lysis. Therefore, while they can positively influence NK cells, they are not damaged by NK cells themselves and are also allowed to migrate to secondary lymphoid organs. In turn, NK cells undergo both activation and proliferation, display a rapid increase in their cytolytic activity and may release large amounts of cytokines, including TNF-α, GM-CSF and IFN-γ 25, which may further amplify the inflammatory response.
Regarding the mechanism by which infected DC induce a rapid increase of cytolytic activity in NK cells, a likely explanation is the production of large amounts of IL-12 by DC following infection. On the other hand, NK cell proliferation is likely to be sustained by other lymphokines such as IL-2 and IL-15 25, 27, 28. Although these cytokines were detected at extremely low levels in the supernatants derived from DC cultured in the presence of bacteria, it is not possible to exclude their role in DC-mediated NK cell expansion. Indeed, cellular synapses occurring between the two cell types during their interactions are likely to be crucial, by greatly increasing the concentration of DC-released soluble mediators.
Our data also suggest that the induction of an activated phenotype in NK cells is not mediated via NKp30. Therefore, this triggering receptor, which plays a major role in the recognition and lysis of DC, does not appear to play any substantial role in the activating signal delivered by DC to NK cells.
The significant cross-talk between infected DC and NK cells could also explain why NK cells are essential for effective BCG immune therapy 29. Thus, the strong increase in NK cell cytotoxicity that is rapidly induced by DC infected with BCG could represent a major mechanism in the still-unclear antitumor effect of BCG therapy.
A relevant question related to our present data is why bacterial infection should lead to a rapid induction of NK cell activation and cytotoxicity. Thus, although NK cell recruitment and activation result in production of cytokines and chemokines that may contribute to the defense against bacterial spreading, the cytolytic activity of NK cells does not exert any direct effect against bacteria. We propose a physiopathologic mechanism in which activated NK cells could play a role in the homeostasis of the immune response during bacterial infections. Thus, whereas activated NK cells are inefficient in killing infected DC (as discussed above), they can efficiently lyse uninfected DC. Accordingly, the presence of activated NK cells in tissues and lymph nodes that are inflamed because of bacterial infection may limit an overwhelming recruitment of iDC at a stage in which bacteria have already been eliminated and infection has been controlled. Therefore, the ability of NK cells to discriminate between infected and uninfected DC may suggest that NK cells play an important regulatory role by selectively editing APC during bacterial infection and thus switching off an excessive immune response 26. This mechanism may be particularly useful in preventing tissue damage.
In conclusion, this study provides a further piece of information on the complex cross-talk occurring between two major players of innate immunity and suggests a possible novel mechanism by which NK cells could co-operate in defense against bacteria.
4 Materials and methods
4.1 Generation of DC
Whole blood from donors in our laboratory and leukocyte concentrates served as sources of PBMC, isolated by density gradient centrifugation on Ficoll-Paque (Pharmacia, Uppsala, Sweden). Positive selection for CD14+ PBMC was performed using anti-CD14 MicroBeads, MS+/RS+ columns and a MiniMACS separator (Miltenyi Biotec, Bergisch-Gladbach, Germany). Monocyte-derived DC were generated from CD14+ PBMC. The CD14+ PBMC (5×105/ml) were plated in 24-well plates with RPMI-1640, plus 10% FCS (Cambrex, Verviers, Belgium); rhIL-4 (R&D Systems, Minneapolis, MN) and rhGM-CSF (Immunex, Seattle, WA) were added to a final concentration of 500 and 1000 U/ml, respectively, at day 0, 2 and 4 in 500 μl of fresh medium/well. On day 5, the floating iDC were transferred to new plates at 3×105 cells/ml and half of the medium was replaced with fresh medium containing BCG (Pasteur Merieux, Lyon, France) or E. coli (DH5α). Bacteria were added to DC cultures at a multiplicity of infection (MoI) of 1 (1 DC per 1 bacterial body). Preliminary experiments showed that a higher MoI with BCG could affect DC viability.
4.2 Isolation and culture of NK cells
NK cells were negatively selected by the NK cell isolation kit (Miltenyi Biotec, Bergisch-Gladbach, Germany). The percentage of NK cells in the isolated population was then evaluated using FITC-conjugated anti-CD3 mAb and PE-conjugated anti-CD56 mAb (Coulter Scientific, Miami, FL) and flow cytometry. Recombinant IL-2 (100 IU/ml; Proleukin, Chiron, Milano, Italy) and PHA (1 μg/ml) were added in order to obtain a polyclonal NK cell population. Resting NK/DC cocultures were performed in RPMI 1640 + 10% FCS in 24-well plates. Live bacteria (BCG or E. coli) were added to DCcultures and NK cells were added to DC cultures 24 h after the infection. After additional 24 h, cells were collected and analyzed.
4.3 Cytokine assays
To detect the production of cytokines by DC cultures upon infection with BCG, cells were cultured in RPMI-1640 plus 10% FCS in 24-well plates. The supernatants of the cultures were collected at different intervals of time and assayed in commercial ELISA kits for TNF-α, IL-2 and IL-12 (BioSource, Camarillo, CA, or Endogen, Woburn, MA) and for IL-15 (R&D Systems GmbH, Wiesbaden-Nordenstadt, Germany).
4.4 Proliferation assay
NK cells (5×104) were incubated with DC at different ratios for 5 days in RPMI-1640 plus 10% FCS in 96-well round bottom microtiter plates. 1 μCi [3H]tymidine was added per well overnight, harvested with a Harvester Mach IIIM (Tomtec, Hamden, CT) and counted in a 1450 MicroBeta TriLux (Wallac, Turku, Finland).
4.5 Flow cytometric analysis
Analysis of DC and NK cell surface markers was performed using the following mAb in immunofluorescence assays: PE-conjugated anti-CD86 (HA5.2B7, IgG2b), FITC-conjugated anti-CD80 (MAB104, IgG1), PE-conjugated anti-CD1a (BL6, IgG1), FITC-conjugated anti-CD83 (HB15A, IgG2b), FITC-conjugated anti-CD25 (1HT44H3, IgG2a) and anti-CD16 (3G8, IgG1) PE-conjugated anti-CD56 (N901, IgG1) and anti-CD69 (TP1.55.3, IgG2b (all from Immunotech-Coulter Company, Marseille, France); FITC-conjugated anti-HLA-DR (TU36, IgG2b) and PE-conjugated anti-CD11c (B-ly6, IgG1) and anti-CCR7 (2H4, IgM) (BD Pharmingen, San Diego, CA); and FITC-conjugated anti-CD40 (BE-1, IgG1) (Ancell Corporation, Bayport, MN). Anti-CD14 (63D3, IgG1) was kindly provided by D. Vercelli, (HSR-DIBIT, Milan) and anti HLA-DR (D1.12, IgG2a) was kindly provided by G. Frumento, Genoa, Italy. The direct immunofluorescence procedure was performed by diluting fluorochrome-labeled mAb with 1 mg/ml human γ-globulin (human therapy grade from Biotest srl, Milan, Italy), in order to block non-specific Fc-receptor binding. Cells were then washed and the flow cytometric analysis performed. Indirect immunofluorescence assays were performed as follows: nonspecific binding sites on cells were saturated with human γ-globulin and then the relevant mAb was added and incubated for 30 min at 4°C. After extensive washing, FITC-conjugated isotype-specific goat anti-mouse antibodies (GAM) (Southern Biotechnology, Birmingham, AL) were added and incubated for 30 min at 4°C. Negative controls included directly labeled or unlabeled isotype-matched irrelevant mAb. Cells were then washed and analyzed by flow cytometry. The percentages and "mean fluorescence channels" of positive cells were obtained from the cytometer workstation-elaborated histograms. To calculate the number of molecules expressed on the surface of DC, saturating dilutions of FITC-labeled mAb were used. The number of Ag present on the cell surface was calculated on the basis of the mean fluorescence channel and the FITC:protein ratios of mAb, as previously described 6. Briefly, a set of fluorescent microbeads (Coulter Corp., Miami, FL) with a known content of molecular equivalent of soluble fluorescein (MESF) was used as standards. Finally, cell size was calculated using flow cytometry and a set of microbeads with known diameters (5, 7, 10 and 20 mm; Coulter Scientific, Miami, FL).
4.6 51Cr release assay
To evaluate the cytolytic activity of NK cells against DC we used autologous iDC and DC cultured with BCG or E. coli for 24 h or 48 h as indicated. Briefly, 106 target cells were incubated with 100 μCi of Na251CrO4 for 60 min at 37°C and then extensively washed. Supernatants were collected and radioactivity counted on a gamma-counter (Beckman, Milan, Italy). "Specific" 51Cr release is calculated on the basis of the formula [(sample release – spontaneous release) / (total release – spontaneous release)]. Assays were performedin triplicate at the indicated effector:target ratios. In order to analyze the role of NK cell activating and inhibitory receptors and HLA class I and in the lysis, anti-HLA class I mAb (A6136, IgM) (kindly provided by A. Moretta, University of Genoa) and anti-NKp30 mAb (F252, IgM) (provided by D. Pende (Istituto Nazionale per la Ricerca sul Cancro, Genoa) were added in saturating amounts in some experiments.
This work was supported by grants awarded by the Associazione Italiana per la Ricerca sul Cancro (A. I. R. C.), by the Ministero della Salute, Ricerca Finalizzata, Italy, and by the Fondazione Compagnia di San Paolo, Torino, Italy.