Besides their role in destruction of altered self-cells, NK cells have been shown to potentiate T-cell responses by interacting with DC. To take advantage of NK–DC crosstalk in therapeutic DC-based vaccination for infectious diseases and cancer, it is essential to understand the biology of this crosstalk. We aimed to elucidate the in vitro mechanisms responsible for NK-cell recruitment and activation by DC during infection. To mimic bacterial infection, DC were exposed to a membrane fraction of Klebsiella pneumoniae, which triggers TLR2/4. DC matured with these bacterial fragments can actively recruit NK cells in a CCR5-dependent manner. An additional mechanism of DC-induced NK-cell recruitment is characterized by the induction of CCR7 expression on CD56dimCD16+ NK cells after physical contact with membrane fraction of K. pneumoniae-matured DC, resulting in an enhanced migratory responsiveness to the lymph node-associated chemokine CCL19. Bacterial fragment-matured DC do not only mediate NK-cell migration but also meet the prerequisites needed for augmentation of NK-cell cytotoxicity and IFN-γ production, the latter of which contributes to Th1 polarization.
NK cells are important effector cells in the innate immune response against virally infected or malignantly transformed cells and their cytotoxicity is regulated by a delicate balance of inhibitory and activating signals 1. Recent studies suggest that the interplay between NK cells and DC, the specialized antigen-presenting cell of the innate immune system 2, is critical in shaping the adaptive immune response 3. This concept originates from several lines of evidence including: the discovery of NK cells colocalizing with DC in the T-cell areas of lymph nodes 4, 5, the coupling of NK-cell recruitment to lymph nodes with the induction of more potent Th1 skewing 3, and the identification of NK-cell subpopulations with helper properties 6. Although the exact mechanisms of NK–DC interaction remain to be elucidated, increasing evidence supports the importance of bidirectional NK–DC crosstalk 7, 8.
On the one hand, NK–DC crosstalk is characterized by the capacity of activated NK cells to induce DC maturation with elevated IL-12p70 production and subsequently an increased capacity to induce Th1 and CTL responses 9. This NK-induced DC maturation depends at least in part on soluble factors such as TNF-α 10 and IFN-γ 11 as well as on engagement of the natural cytotoxicity triggering receptor 30 12. Moreover, NK cells control the quality of the adaptive immune response by natural cytotoxicity triggering receptor 30-mediated lysis of immature or inadequately matured DC 13, enabling only fully mature DC to migrate into lymph nodes and subsequently prime T cells. On the other hand, DC are able to induce NK-cell proliferation, augmentation of cytotoxicity and cytokine secretion 8. The DC-induced modulation of cytokine production by NK cells requires a strong immunological synapse between the two cell types and both DC-derived soluble factors (e.g. IL-12 and IL-18) as well as contact-dependent factors (e.g. IL-15 and NKG2D ligands) are implicated in this crosstalk 6, 14, 15. Mouse models have revealed a DC-dependent mechanism of NK-cell recruitment into the lymph nodes. One of the functions of this recruitment is to provide IFN-γ, necessary for efficient Th1 skewing 3. In humans, the mechanism for DC-dependent NK cell recruitment into inflamed lymph nodes remains to be unraveled and it is unclear whether chemokine profiles of activated DC during infection match the prerequisites necessary for NK-cell activation.
Klebsiella pneumoniae is an extracellular bacterium that induces activation of the innate immune system. In vivo mouse studies have revealed that endogenous IL-12, produced by alveolar macrophages, is a critical component of the antibacterial host defense against K. pneumoniae as blocking of IL-12 is associated with increased mortality 16. CXCL10-mediated NK as well as T-cell recruitment and IFN-γ production has been reported to be essential to control K. pneumoniae infection 16, 17. This is supported by a study on CpG triggering during K. pneumoniae infection. CpG-treated mice showed increased infiltration and activation of NK cells providing an early source of IFN-γ, which was associated with decreased mortality 18. NK cells are not only activated by other immune cells triggered by K. pneumoniae, but they are also equipped themselves with PRR suitable for direct binding of the outer membrane fraction of K. pneumoniae (OmpA), which triggers NK cells to produce IFN-γ and the anti-bactericidal peptide α-defensin 19.
In this study, we addressed the question whether human DC triggered with fragments of K. pneumoniae display the ability to recruit and activate NK cells, ultimately leading to Th1 polarization. Additionally, we set out to elucidate which mechanisms are responsible for these processes. To mimic a bacterial infection, DC were matured by a membrane fraction of K. pneumoniae (FMKp), which is a lysate of the outer membrane fraction of K. pneumoniae containing the DC-triggering agents LPS and outer membrane protein A (OmpA), which signal trough TLR4 and TLR2, respectively 20. For its proven additional effect on upregulation of costimulatory molecules and modulation of cytokine secretion, IFN-γ was added to the FMKp maturation cocktail 21. This study led to the identification of chemokines and cytokines responsible for NK–DC crosstalk and we describe two possible mechanisms of NK cell recruitment into the inflamed lymph node. These in vitro observations are of relevance in unravelling the in vivo biology of NK–DC crosstalk.
DC-dependent NK-cell recruitment is CCR5 mediated
To investigate whether NK cells are involved in the immune response against bacteria, we tested the ability of DC to induce NK-cell migration. A bacterial infection was mimicked by triggering monocyte-derived DC with FMKp. As a control, DC were matured with the proinflammatory mediators TNF-α and prostaglandin E2 (PGE2) in the absence of bacterial triggers. In addition to FMKp, also purified, single TLR ligands were used to investigate whether these individual pathways dominate induction of NK–DC interaction. Therefore, DC were matured with either Zymosan (TLR2 ligand) or LPS (TLR4 ligand) in the presence of IFN-γ. Moreover, the combination of both single TLR-triggering agents was used to determine a possible synergism.
IL-4/GM-CSF differentiated DC were matured for 6 h in the presence of either TNF-α/PGE2 or IFN-γ/FMKp after which the cells were extensively washed and transferred to serum-free medium without maturation stimuli. Washing was performed to avoid altered NK-cell migration due to activation by the maturation cocktail. For the TNF-α/PGE2 and IFN-γ/FMKp-matured DC, 6 h maturation irreversibly triggered maturation as evidenced by the upregulation of MHC class II and costimulatory molecules (Supporting Information Fig. 1), which was comparable to the expression of these markers detected after 48 h maturation 22. Filtered, cell-free, DC supernatant was harvested after 18 h of culture and transferred to the lower compartment of a Transwell®.
Since NK-cell migration is more reproducible at longer migration periods (Supporting Information Fig. 2), recruitment assays were performed over a period of 16 h. NK-cell migration to a CCL19 gradient was used as a positive control to normalize donor and interexperimental variability. After 16 h of NK-cell migration, the supernatant of the IFN-γ/FMKp-matured DC contained significantly more NK cells as compared with the supernatant of the TNF-α/PGE2-matured DC (Fig. 1A). We did not observe a selective enrichment of CD56bright nor CD56dim NK cells in the lower well, suggesting that there was no preferential recruitment of either NK-cell subset (data not shown). No significant difference in the number of migrated NK cells was detected when medium was present instead of the supernatant of TNF-α/PGE2-matured DC (data not shown), indicating that these DC fail to attract NK cells.
To identify the chemokines responsible for the observed NK-cell recruitment by IFN-γ/FMKp-matured DC, chemokine arrays were performed on DC supernatants. The chemiluminescence values of the chemokinespots were quantified by densitometry, normalized to control spots and plotted in XY graphs (Fig. 1B). DC maturation induced by IFN-γ/FMKp led to an enhanced (>1.5-fold increase in chemiluminescence) secretion of eleven chemokines as compared with TNF-α/PGE2-induced maturation. These chemokines included CCL5, CXCL10 and CCL19 to which NK cells can respond 23–25. Quantification with ELISA confirmed that these chemokines were indeed produced by IFN-γ/FMKp-matured DC and not by TNF-α/PGE2-matured DC (Fig. 1C). To evaluate whether the chemokine production was induced by TLR-triggering or IFN-γ stimulation, DC were matured by either FMKp or IFN-γ alone. Maturation with IFN-γ does neither induce DC to produce CCL19 nor CCL5; however, IFN-γ stimulates DC to secrete CXCL10 (Fig. 1D). FMKp is sufficient to trigger DC to produce all three cytokines, albeit in lower concentrations as compared with maturation in the presence of both FMKp and IFN-γ, evidencing their additive effect.
Additionally, other TLR ligands were used to induce DC maturation and induction of chemokine production was compared with the production by IFN-γ/FMKp-matured DC (Fig. 1D). In the presence of IFN-γ, zymosan, LPS and the combination of both TLR ligands triggered DC to produce CCL19 and CXCL10, although at significantly lower amounts as compared with FMKp/IFN-γ-matured DC. Zymosan in contrast to LPS was not able to induce CCL5 production by DC. No additive effect on chemokine secretion was observed when TRL2 and TLR4 were stimulated simultaneously. To evaluate if the differences in chemokine production between different TLR-triggered DC influenced NK-cell recruitment, supernatant of zymosan/IFN-γ and LPS/IFN-γ triggered DC, matured by the same maturation and washing procedure as the IFN-γ/FMKp-matured DC were used in NK-cell migration assays. Migration data revealed that IFN-γ/FMKp-matured DC recruited more NK cells than DC matured with either FMKp or IFN-γ alone (Fig. 1E). Moreover, FMKp/IFN-γ-matured DC also recruited significantly more NK cells than zymosan/IFN-γ-matured DC (Fig. 1E). Furthermore, LPS/IFN-γ-matured DC do not differ from IFN-γ/FMKp-matured DC in their capacity to recruit NK cells although they produce lower amounts of chemokines.
To establish which chemokines are involved in NK-cell recruitment by IFN-γ/FMKp-matured DC, corresponding chemokine receptors on NK cells were blocked (CCR5 for CCL5, CXCR3 for CXCL10 and CCR7 for CCL19). Only blocking of CCR5 reduced NK-cell migration significantly (Fig. 1F). No significant reduction was observed with a CCR7-blocking antibody, which is like anti-CCR5 an IgG2a antibody and therefore serves as an isotype control.
In conclusion, DC triggered with bacterial fragments of K. pneumoniae produce many chemokines in substantial amounts, including CXCL10, CCL19 and CCL5, and are able to recruit NK cells in a CCR5-dependent manner.
DC-dependent induction of CCR7 expression on NK cells
In steady-state conditions, the CD56brightCD16− NK-cell subpopulation has been reported to express CCR7 25, which allows them to migrate into lymph nodes 26. However, in response to IL-18-producing monocytes, induction of CCR7 has also been observed on CD56dimCD16+ NK cells 6. To investigate whether IFN-γ/FMKp-matured DC also induce CCR7 expression on NK cells, both cell types were cocultured. After 24 h, NK cells were harvested and CCR7 expression was evaluated by flow cytometry. Besides CCR7 expression on CD56bright NK cells, CCR7 was also observed on the CD56dim NK-cell subpopulation after coculture with IFN-γ/FMKp-matured DC (Fig. 2A), which was significant in seven different donors (Fig. 2B; p<0.01). When NK cells were physically separated from DC by a Transwell® (pore size, 0.4 μm), CCR7 expression was still induced, albeit on a lower percentage of NK cells (Fig. 2B). This indicated that the induction of CCR7 expression was predominantly contact dependent.
To test whether this induction of CCR7 expression resulted in an increased migration toward lymph node-related chemokines, NK cells were isolated from NK–DC cocultures and tested for their capacity to migrate toward CCL19. Compared with freshly isolated NK cells, NK cells from NK–DC cocultures showed a significantly increased migratory responsiveness (p<0.01; average migration index of 1.25) toward CCL19 (Fig. 2C).
These data indicate that, upon contact with DC triggered with bacterial fragments, CCR7 expression is induced on a subpopulation of NK cells, increasing their migratory responsiveness to CCL19, which is considered to be a lymph node-homing chemokine.
NK-cell activation by IFN-γ/FMKp-stimulated DC enhances NK-cell cytotoxicity and IFN-γ production
Recent data have shown that lymph node-homing NK cells assist in Th1 polarization 3, 27. Since IFN-γ/FMKp-matured DC were able to recruit NK cells, we investigated whether these DC were also able to induce IFN-γ production by NK cells. To quantify the percentage of NK cells producing IFN-γ, intracellular IFN-γ staining was performed on NK cells cultured in DC supernatant (Fig. 3A). A small subpopulation (about 2%) of NK cells accumulated IFN-γ when cultured for 16 h in the supernatant of IFN-γ/FMKp-matured DC (Fig. 3B). Among the IFN-γ-accumulating NK cells, we identified NK cells with both high and intermediate expression of CD56. Though small in number, the IFN-γ-accumulating NK-cell subpopulation was capable of producing detectable amounts of IFN-γ as measured by ELISA (Fig. 3C). Induction of IFN-γ secretion of NK cells by DC depends on the additive effect of IFN-γ and FMKp during DC maturation, since neither of these stimuli alone could induce as much IFN-γ production by NK cells as IFN-γ/FMKp-matured DC (Fig. 3D). After activation by IFN-γ/FMKp DC, the complete NK-cell population showed upregulation of the NK-cell activation markers CD69 and CD25 (Supporting Information Fig. 3). Notably, NK cells cultured in the supernatant of the TNF-α/PGE2-matured DC neither accumulated IFN-γ nor showed upregulation of activation markers.
To evaluate whether DC triggered by single TLR ligands, also induce NK-cell activation, NK cells were incubated for 16 h in the supernatant of DC matured with zymosan, LPS or the combination of both in the presence of IFN-γ. NK cells incubated with supernatant of these TLR-triggered DC showed significantly lower production of IFN-γ than NK cells incubated with supernatant of IFN-γ/FMKp-matured DC (Fig. 3D).
Two crucial soluble cytokines responsible for NK-cell activation are IL-12 and IL-18, which act interdependently in a two-signal mechanism 6. To examine whether these cytokines are involved in the observed IFN-γ production, their presence in the supernatant of IFN-γ/FMKp and TNF-α/PGE2-matured DC was determined. IFN-γ/FMKp-matured DC produced IL-12p70 as well as IL-18, albeit with a large donor variation (Fig. 3E). On the contrary, TNF-α/PGE2-matured DC did neither produce IL-12p70 nor IL-18. Since activation of NK cells by DC depends on the additive effect of IFN-γ and FMKp, IL-12 production after maturation with these triggers individually was examined. DC produced IL-12 after FMKp triggering but not after IFN-γ triggering; however, the combination of FMKp and IFN-γ had a synergistic effect on IL-12p70 production (Fig. 3F). When production of IL-12 by DC matured with other TLR2 and TLR4 ligands was evaluated, it was found that LPS, in contrast to zymosan, induced IL-12 production and also the combination of LPS and zymosan did, however, the amount produced by these DC was significantly lower than that of IFN-γ/FMKp-matured DC (Fig. 3F).
Upon physical contact with IFN-γ/FMKp-matured DC, NK cells produce slightly elevated amounts of IFN-γ as compared with NK cells cultured in DC supernatant (Supporting Information Fig. 4), indicating that predominantly soluble factors, with minor contribution of contact-dependent factors, mediate IFN-γ secretion by NK cells. This phenomenon was observed in both autologous as well as killer cell immunoglobulin-like receptor (KIR) mismatched NK–DC cocultures. A candidate cytokine responsible for this contact-dependent increase of IFN-γ production is IL-15, which can be presented in trans by DC 15, 28. Therefore, expression of IL-15 by IFN-γ/FMKp and TNF-α/PGE2-matured DC was evaluated by flow cytometry. Notably, IFN-γ/FMKp DC express more surface-bound IL-15 (Fig. 3G).
The dependency of NK-cell activation on DC-derived IL-12/IL-18 was examined by neutralizing both cytokines with the respective neutralizing antibodies. This resulted in a decreased IFN-γ production after neutralizing IL-18 and a complete block of IFN-γ secretion after neutralizing IL-12p70 (Fig. 3H).
To investigate whether the observed DC-induced NK-cell activation by IFN-γ/FMKp-matured DC contributes to Th1 polarization, T-cell stimulation assays were performed in the presence or absence of NK cells. Therefore, IFN-γ/FMKp-matured DC were coated with Staphylococcal enterotoxin B (SEB) and cocultured for 10 days with naïve CD4+ T cells in either the presence or the absence of NK cells. When NK cells were present, a significantly higher percentage of CD4+ T cells produced IFN-γ (Fig. 4A; p<0.01), which was reproducible in four donors (Fig. 4B).
Besides cytokine production, the other major NK-cell function is cytotoxicity. To study the effect of NK–DC interaction on NK-cell killing capacity, we investigated NK-cell-mediated lysis of the Burkitt's lymphoma cell line Raji after 24 h incubation with DC supernatant. Raji cells are insensitive for lysis by naïve NK cells; however, if NK cells are preactivated, they become susceptible for NK-cell-mediated cytotoxicity. Therefore, IL-2 activated NK cells were taken along as positive control in the cytotoxicity assay. Only NK cells incubated with the supernatant of IFN-γ/FMKp-matured DC showed an increased cytotoxic activity, whereas NK cells incubated with the supernatant of TNF-α/PGE2-matured DC did not (Fig. 5A). Notably, in neither DC supernatant IL-2 was detected (data not shown), ruling out IL-2 as causative factor for augmentation of cytotoxicity. Another possible candidate that has been shown to induce NK-cell cytotoxicity is IL-12 29. Therefore, IL-12p70 was neutralized, which dose-dependently reduced the specific lysis of Raji cells to the level observed after coincubation with unstimulated NK cells (Fig. 5B and C). IL-15, presented in trans by IFN-γ/FMKp-matured DC, is another candidate cytokine responsible for augmentation of NK-cell-mediated lysis. However, blocking of this cytokine in the supernatant of DC did not result in decreased killing capacity of NK cells (data not shown).
In conclusion, DC triggered with bacterial fragments are able to induce IL-12/IL-18-mediated NK-cell activation, resulting in the production of IFN-γ and Th1 polarization. Additionally, the IL-12 produced by these DC is also responsible for augmentation of NK-cell cytotoxicity.
Under steady-state conditions, NK cells do not home to tissues and lymph nodes. However, during inflammation, it has been demonstrated that NK cells are able to migrate into inflamed tissue and draining lymph nodes 7, 23. Currently, it remains unclear what drives human NK-cell migration in vivo. Primary immune responses to bacterial fragments (FMKp) induce DC to produce high amounts of chemokines (Fig. 1B) and we demonstrate that these DC are able to recruit NK cells in a CCR5-dependent manner. It can be hypothesized that chemokines produced by DC, triggered with bacterial fragments, are responsible for NK-cell recruitment to inflamed tissue and into the inflammatory lymph node. The ability of human DC to recruit NK cells is consistent with the previous studies that report on CCL5 and CXCL10 production and NK-cell recruitment by DC either infected with Mycobacterium tuberculosis30, stimulated by oncolytic reovirusus 31 or matured with IFN-α/IFN-γ/TNF-α/IL-1β/poly-(I:C). Blocking of CXCR3 did not affect NK-cell migration to IFN-γ/FMKp-matured DC; therefore, our data suggest that CXCR3- and CCR5-dependent NK-cell recruitments are two different mechanisms by which NK cells can be attracted by DC. Both chemokine receptors CXCR3 and CCR5 are expressed by subpopulations of resting and activated NK cells and have been proven essential for NK-cell migration to various tissues in response to proinflammatory stimuli 23, 26, 32. Which NK-cell-recruitment mechanism is utilized seems to depend on the agents used in DC maturation and lack of NK-cell recruitment by TNF-α/PGE2-matured DC suggests that TLR ligands or other PRR-triggering agents are necessary triggers for DC to recruit NK cells. Triggering of different TLR could explain the difference in chemokines dominant in NK-cell recruitment and suggest that different infectious agents can induce different DC–chemokine profiles. This is supported by our data on differences in chemokine secretion between DC triggered with TLR2 and TLR4 agonists, resulting in differential NK-cell recruitment. Although the current data indicate that TLR4 signaling induces DC to secrete CCL5, it remains to be established whether this is the only pathway inducing production of this chemokine. IFN-γ/FMKp-matured DC produce fourfold higher amounts of CCL5 than LPS/IFN-γ-matured DC which could be due to differences in the concentration of the TLR-triggering agents, since the exact concentrations of LPS and OmpA in FMKp are unknown. Furthermore, at present, it remains unclear whether FMKp also contains other PRR ligands responsible for higher chemokine secretion. A second line of evidence supporting the fact that different TLR trigger induce different DC chemokine profiles is illustrated in mouse studies on differential NK- and T-cell activation and recruitment by differently matured DC 33, 34. Notably, the CXCR3- and CCR5-dependent mechanisms by which DC attract NK cells are not mutually exclusive, as it has been demonstrated that upon poly (I:C) triggering, intrasplenic NK-cell trafficking depends on synergistic actions of CXCR3 and CCR5 32.
It can be envisioned that DC, displaying a favorable chemokine profile, are able to recruit immune effector cells including NK cells, Th1 cells and CTL and thereby creating a peripheral, tertiary lymph node 35. Next to this peripheral NK-cell recruitment, DC could facilitate NK-cell migration into the lymph nodes. This has been demonstrated in a mouse model by LPS-matured DC, which recruit NK cells into the draining lymph nodes 3. Since IFN-γ/FMKp-matured DC are active producers of CCR7, CXCR3 and CCR5 ligands, it can be anticipated that they attract NK cells during their arrival in draining lymph nodes by virtue of different mechanisms and our in vitro data demonstrate that CCR5-dependent recruitment is the most potent for these DC.
In addition to the CCR5-dependent NK-cell recruitment, we show that DC triggered by bacterial fragments have the ability to induce CCR7 expression on a subset of NK cells that increases their migratory responsiveness to CCL19. This represents a second mechanism of DC-induced NK-cell homing into inflammatory lymph nodes. The factors inducing CCR7 expression in our system have not been elucidated. Similar results have recently been obtained by Marcenaro et al., who showed an induction of CCR7 after coincubating NK cells and DC in an allogeneic setting 36. We, however, observed CCR7 induction in both allogeneic and autologous settings, indicating that multiple mechanisms may account for CCR7 induction. In line with the findings of Mailliard et al., 6 who demonstrated that IL-18 induces CCR7 expression on the same CD56dimCD16+ NK-cell subset as the IL-18-producing, IFN-γ/FMKp-matured DC (Supporting Information Fig. 5). Therefore, IL-18 could be one of the DC-dependent factors responsible for the CCR7 induction. In addition, the partial contact dependency may also point to IL-18, because it has been shown that DC-derived IL-18 is polarized and secreted in the immunologic synapse inducing local, high IL-18 concentrations 37.
Besides their NK-cell-attracting capacity, DC matured with bacterial fragments also meet the prerequisites for NK-cell activation. Using blocking antibodies, we demonstrated that both DC-derived IL-12 and IL-18 are necessary for IFN-γ secretion by a subpopulation of NK cells. Addition of rIL-12 to the supernatant of TNF-α/PGE2-matured DC did not result in NK-cell IFN-γ production (data not shown), which indicates that both IL-12 and IL-18 are required to induce IFN-γ secretion, supporting the two-signal requirement mechanism described by Mailliard et al. 6. Notably, the IFN-γ-producing NK-cell population contained both NK cells with an intermediate and with a high expression of CD56, which is remarkable since the CD56bright subpopulation has been described as the cytokine-producing population after NK–DC interaction 38. Upon physical contact with DC, NK cells produce even higher amounts of IFN-γ (Supporting Information Fig. 4), indicating that predominantly soluble factors, with minor contribution of contact-dependent factors, mediated IFN-γ secretion by NK cells. This phenomenon was observed in both autologous and Killer cell Immunoglobulin-like Receptor mismatched NK–DC cocultures. Ligation of NK-cell receptors could be responsible for this contact dependency 15; however, the formation of an immunological synapse could also be an important factor 39. Next to IL-18, other possible candidates are MICA/B 15 and in trans presented IL-15 15, 28. The functional consequence of NK-cell activation by IFN-γ/FMKp-matured DC appeared to be dual. On the one hand, NK cells produce sufficient cytokines to increase Th1 polarization that was shown previously to be completely dependent on NK-cell-derived IFN-γ 27. On the other hand, we demonstrated that IL-12 produced by IFN-γ/FMKp-matured DC also augments NK-cell cytotoxicity toward the Raji cell line. This augmentation has been documented after contact of NK cells with some bacteria or by NK–DC crosstalk during bacterial infection 11, 40.
In conclusion, based on our in vitro data, it can be hypothesized that in vivo a K. pneumonia infection is responsible for induction of innate immune responses characterized by DC-dependent NK-cell migration, facilitating NK–DC interactions both in the periphery and in the lymphoid tissue. The molecular program triggered by FMKp and IFN-γ does not only permit DC to recruit NK cells, but also induces the prerequisites necessary to activate NK-cell cytotoxicity and NK-cell helper function for Th1 polarization and possibly subsequent CTL induction. DC vaccination studies for the treatment of cancer have revealed that NK-cell activation is of importance for sustained anti-tumour responses 41. Since the examined DC maturation cocktails are clinically relevant DC preparations used in DC-based vaccination studies, unravelling the in vitro mechanisms of NK–DC interaction is of importance for the development of therapeutic DC vaccination and DC-targeting strategies for cancer and infectious diseases.
Materials and methods
Mononuclear cells from peripheral blood of healthy donors were isolated by density gradient separation using lymphoprep (Axis-Shield). NK cells were negatively selected by immunomagnetic cell separation (Miltenyi Biotec GmbH). CD4+ T cells were isolated using RosetteSep (StemCell Technologies). Naïve CD4+ T cells were negatively selected by depleting CD45RO+ cells with immunomagnetic cell separation (Miltenyi Biotec GmbH). The purity of isolated populations exceeded 95% as determined by flow cytometry.
Generation of DC
DC were prepared from peripheral blood-derived monocytes as described previously 42 and as control, highly pure monocyte fractions were obtained by elutriation from leukapheresis products of healthy volunteers, approved by the local medical, ethical committee of Maastricht University Medical Center+. Differentiation of monocytes was induced by 6 days of culture in AIM-V® medium (Gibco Life Technologies) containing 2000 U/mL IL-4 (Strathmann Biotech) and 400 U/mL GM-CSF (Berlex) or 50 ng/mL IL-13 (Biosource) and 500 U/mL GM-CSF. DC were matured for 6–24 h in AIM-V® containing either 1000 U/mL TNF-α (Biosource) and 18 μg/mL PGE2 (Sigma-Aldrich) or 500 U/mL IFN-γ (Strathmann Biotech), and 1 μg/mL FMKp (Pierre Fabre). For maturation with single TLR2 and TLR4 ligands, DC were matured in the presence of the same amounts of IL-4, GM-CSF and IFN-γ as IFN-γ/FMKp-matured DC supplemented with 10 μg/mL Zymosan (InvivoGen) and/or 10 μg/mL LPS (InvivoGen).
All antibodies for flow cytometry, except anti-IL-15 and CCR7 (R&D Systems), were obtained from BD Biosciences. Antibodies were used in Fluorescein isothiocyanate, Pycoerytherin, Peridinin chlorophyll protein or allophycocyanin. Cells were incubated with antibodies at proper dilutions for 30 min at room temperature. Analyses were performed on a FACSCalibur (BD Biosciences) and analyzed with BD CellQuest™ Pro Software (BD Biosciences).
Migration of NK cells was analyzed using 5.0 μm Transwell® plates (Corning Costar). The lower cabinet contained 600 μL cell-free DC supernatant or AIM-V® containing 250 ng/mL CCL19 (R&D Systems). Triplicate inserts were filled with 100 μL containing 5×105 NK cells, migration proceeded for 16 h at 37°C/5% CO2. Migrated cells were counted using the Z1™ Coulter Counter (Beckman Coulter). To block migration, NK cells were incubated for 1 h with blocking antibodies for CCR5 (BD Biosciences), CCR7 or CXCR3 (R&D Systems) preceding migration. The contact-dependent increase in NK-cell migration to 250 ng/mL CCL19 (R&D Systems) was determined after separating the NK cells from the NK–DC coculture by positive selection of DC on HLA-DR (Miltenyi Biotec GmbH).
Undiluted cell-free supernatants (24 h of maturation of 106 cells/mL) were analyzed on the human chemokine antibody array I (RayBiotech) according to the manufacturer's instructions. Chemokine blots were photographed with a LAS-3000 imaging system (Fujifilm) and spots were quantified using AIDA Array Analysis Software (ImaGenes). Quantitated chemiluminescence values of individual spots were calculated using the Chemokine Antibody Array C Series 2000 Analysis Tool (RayBiotech).
Cytokine and chemokine secretion by DC
Quantification of IL-18, IL-12, CCL5, CXCL10 and CCL19 in 24 h maturation supernatants was performed using ELISA (MBL International and R&D Systems) according to the manufacturer's instructions. Absorbance was measured at 450 nm using a microtiterplate reader (BioRad).
Cytokine production by NK cells
Intracellular IFN-γ was determined after 16 h NK–DC coculture in the presence of 1 μg/mL Brefeldin A (BD Biosciences). After extracellular staining, cells were fixed, permeabilized using PermWash (BD Biosciences) and stained with anti-human IFN-γ (BD Biosciences). Cells were fixed in 1% PFA/PBS and analyzed by flow cytometry. Quantification of IFN-γ production was performed using IFN-γ ELISA kit (Sanquin) in 24 h cell-free supernatant. NK-cell activation was blocked by preincubating cell-free DC supernatant for 1 h with 0.5 μg/ml anti-IL-12 (R&D Systems) and 4 μg/mL anti-IL-18 (MBL International).
For NK–DC contact dependency, NK cells (105 cells/well) were separated from DC (2×105 cells/well) by culturing the cells in different compartments of a 0.4 μm Transwell® plate (Corning Costar). Quantification of IFN-γ production was performed using the IFN-γ ELISA kit (Sanquin) in 24 h cell-free supernatant.
T helper cell priming
DC (4×104 cells/well) were cocultured for 24 h in the presence or absence of NK cells (5×104 cells/well), before they were coated with 1 ng/mL SEB (Sigma-Aldrich) for 1 h. SEB-coated DC were washed and placed in culture with CD45RA+CD45RO−CD4+ T cells (105 cells/well). On day 4, 20 U/mL rhIL-2 (Proleukin, Chiron Benelux BV) was added to the cultures. At day 10, the expanded Th cells were washed, plated in 96-well plates (105 cells/well) and stimulated with PMA/ionomycin (BD Biosciences) for 4 h in the presence of 1 μg/mL Brefeldin A (BD Biosciences). Intracellular IFN-γ staining was analyzed by flow cytometry.
NK-cell cytotoxicity assay
NK-cell cytotoxicity was analyzed by a flow cytometry-based kill assay as described previously 43. Briefly, NK cells were incubated for 24 h with either DC supernatant, medium alone or medium containing 1000 U/mL IL-2 (Proleukin). Target cells (Raji) were labeled with 3′-dioctadecyloxacarbocyanine according to the manufacturer's instructions (Sigma). Target cells (104 cells) were incubated with effectors at various effector:target (E:T) ratios for 12 h, each ratio in triplicate. Percentages of killed target cells (PI+ 3′-dioctadecyloxacarbocyanine+) were determined by flow cytometry. Percentage-specific lysis was calculated:
The statistical significance of differences between experimental samples was determined using Student's t-test for paired samples or ANOVA for multiple samples. Significance was accepted at the p<0.05 (*) and p<0.01 (**) levels. Data were analyzed using Prism software version 5.00 (GraphPad Software).
The authors thank Dr. Cloosen, S., Wetzels, R., Pekelharing, E. P. and Matos, C. for their technical assistance and Dr. Buurman, W. for critical reading of the manuscript. This work was supported by SenterNovem (Project: IS055002 to W. T. V. G. and G. M. J. B.) and Associazione Italiana Per la Ricerca Sul Cancro (AIRC) project IG 5624.
Conflict of interest: Van Elssen, C. is employed by PharmaCell BV and Libon, C. is employed by Pierre Fabre institute. The other authors declare no financial or commercial conflict of interest.