Background
Natural killer (NK) cells are large granular lymphocytes, which provide innate immune surveillance against virus infections and tumor cells [
1]. Direct or indirect regulation of adaptive immune responses by NK cells may also affect the outcome of autoimmune diseases [
2].
NK cells can rapidly kill tumor or pathogen-infected cells without prior sensitization and produce an array of cytokines and chemokines, which particularly influence dendritic cells (DCs). The bidirectional crosstalk between NK cells and DCs can lead to NK cell activation, DC maturation, or DC killing. Killing of tumor cells as well as most interactions between NK cells and DCs depend on cell contact and, therefore, require cell-to-cell adhesion [
3]. DNAX accessory molecule 1 (DNAM-1, CD226) is such an adhesion and co-activating receptor on NK cells. It binds to the nectin family members CD155 and CD112 [
4]. DNAM-1 thus facilitates NK cell interactions with tumor cells and CD155-expressing DCs. Recent studies have demonstrated the relevance of DNAM-1 for NK cell-mediated tumor cell elimination in the absence of NKG2D ligands, showing that DNAM-1 significantly contributed to the control of tumor growth and metastasis formation in experimental B16F10 melanoma models [
5,
6]. Polymorphisms in the DNAM-1 gene, which decrease DNAM-1 protein expression on the cell surface, are risk factors for neoplastic diseases and have been linked to several autoimmune diseases, among them multiple sclerosis (MS) [
7‐
9]. This suggests that diminished DNAM-1/CD155 interactions might not only reduce tumor cell killing but also impair the immunoregulatory capabilities of NK cells in controlling autoimmunity [
10,
11].
Most studies in MS patients reported reduced numbers of circulating NK cells, which are functionally impaired with regard to cytotoxicity and interferon-gamma (IFNγ) production [
12]. Evidence for a regulatory role of NK cells has been provided by studies with daclizumab, a humanized monoclonal antibody against the IL-2 receptor α-chain. In these studies, the expansion of peripheral and intrathecal CD56
bright NK cells correlated positively with the therapeutic response [
13,
14]. In EAE, depletion of NK cells or diminished recruitment of NK cells to the CNS exacerbated EAE in most but not all studies [
15‐
18].
The majority of data on regulatory NK cells in MS and EAE suggest that NK cells kill T cells directly or inhibit their proliferation [
19,
20]. Gross et al. demonstrated that the reduced surface expression of DNAM-1 on NK cells of MS patients correlated with the impaired cytolysis of autologous, activated, CD155-expressing CD4 T cells [
10]. The perforin-mediated killing of myelin oligodendrocyte glycoprotein (MOG)-specific T cells by NK cells was inhibited when the non-classical class 1 molecule Qa1 on T cells interacted with the NKG2A receptor on NK cells [
21]. NK cells can also be demonstrated in the CSF of MS patients and in actively demyelinating MS lesions [
10,
22], and regulatory functions of NK cells within the CNS have been postulated which require the interaction of NK cells with microglia cells [
23].
Quinoline-3-carboxamide derivatives such as laquinimod, tasquinimod, or pasquinimod have shown immunomodulatory, anti-tumor, and anti-angiogenic effects in pre-clinical animal models [
24‐
26]. So far, the modulation of myeloid cells was primarily held responsible for their immunomodulatory and anti-tumor effects [
27‐
31]. Here, we provide evidence that the aryl hydrocarbon receptor-dependent activation of NK cells is relevant for the efficacy of laquinimod to suppress melanoma cell metastases and CNS autoimmunity. The NK cell-mediated inhibition of T cell proliferation is dependent on cell-to-cell contact and requires the interaction of DNAM-1 with CD155 on dendritic cells.
Methods
Animals
C57BL/6-Ahr
tm1.2Arte and Itgax-DTR/EGFP mice [
32] were purchased from Taconic and the Jackson Laboratory, respectively. CD155-deficient mice were kindly provided by Prof. Bernhardt. Rag1
−/− [
33], myelin oligodendrocyte glycoprotein (MOG)-specific T cell receptor (TCR) transgenic mice (also referred to as 2D2 mice [
34]), MOG-specific Ig heavy-chain knock-in mice on a C57BL/6 background (also referred to as Th mice [
35]), and C57BL/6 mice were bred at the animal facility of the University Medical Center Göttingen under SPF conditions. All animals were housed in a temperature-controlled environment with 12-h light/dark cycles and food and water ad libitum. All animal procedures were approved by the Lower Saxony Federal State Office for Consumer Protection and Food Safety (LAVES), Germany.
EAE induction
For EAE induction, Th/+ animals were immunized s.c. with 50 μg MOG
35–55 per animal emulsified in CFA substituted with 5 mg/ml H37Ra (DIFCO, Detroit, MI, USA). Four hundred nanograms of pertussis toxin (PTX) per animal (List Biological Laboratories, Campbell, CA, USA) were injected i.p. twice on day 0 and day 2 relative to immunization. EAE was induced in Rag1
−/− mice by the adoptive transfer of 10 Mio MOG-specific 2D2 T cell blasts i.p., and animals were immunized s.c. with 10 μg MOG
35–55 in CFA the following day. EAE was scored daily as previously described [
36].
Cell depletion
NK cells were depleted in Th/+ mice or Rag1−/− mice by i.p. injections of 300 μg aNK1.1 antibody (clone PK136) per animal, and control mice received i.p injections of 300 μg C1.18.4 per animal (both Bio X Cell, West Lebanon, NH, USA). Depletion started at days − 2 and − 1 relative to immunization in Th/+ mice or adoptive T cell transfer into Rag1−/− mice and continued every other day thereafter. Depletion efficiency in the blood was evaluated by flow cytometry prior to MOG35–55 immunization. DCs were depleted in Itgax-DTR/EGFP mice by injection of 100 ng diphtheria toxin per animal (Sigma Aldrich) 24 h prior to laquinimod or vehicle treatment. DC depletion efficiency was analyzed by flow cytometry in the spleens at day 3.
Laquinimod therapy
Mice were treated with the quinoline-3-carboxamide laquinimod or vehicle (H2O) by oral gavage on a daily basis. All EAE-induced animals received 25 mg/kg laquinimod starting at the day of EAE induction. Laquinimod (50 mg/kg) treatment was initiated in animals three days prior to the i.v. injection of 500,000 B16F10 melanoma cells (preventive treatment) or 11 days after the i.v. injection of 200,000 B16F10 melanoma cells (therapeutic treatment).
T cell culture
Spleen cells from MOG TCR transgenic 2D2 mice were expanded in RPMI 1640 supplemented with 10% FCS, 50 IU/ml penicillin, 50 μg/ml streptomycin (P/S), 2 mM L-glutamine, 1 mM sodium pyruvate, and 50 μM β2-mercaptoethanol (complete medium) by plate-bound αCD3 (clone 145C11, 4 μg/ml) and freely available αCD28 (clone PV-1, 1 μg/ml, both Bio X Cell) antibodies in the presence of rm IL-12 (1 ng/ml, R&D) and rm IL-2 (2 ng/ml, R&D). Before the adoptive transfer into recipient Rag1−/− mice, cells were restimulated using 30 Gy irradiated syngenic splenocytes and 20 μg/ml MOG35–55 for 3 days.
B16F10 cells
B16F10 melanoma cells (ATCC® CRL-6475™) were previously subcloned from the C57BL/6J-derived melanoma cell line B16 and selected for their ability to form pulmonary tumor nodules [
37]. Cells were maintained in complete medium and split at 50% confluence.
Bone marrow-derived dendritic cells
For the generation of bone marrow-derived dendritic cells, femurs were flushed and single-cell suspensions were cultured with 25 ng/ml rm GM-CSF (PeproTech, Rocky Hill, USA) in complete medium for 7 days and then MACS sorted.
In vitro NK cell assays
Splenocyte suspensions of laquinimod- or vehicle-treated mice were enriched for NK cells by magnetic-activated cell sorting (MACS™) using the untouched NK cell isolation kit (Miltenyi Biotec, Bergisch Gladbach, Germany) and further purified to a purity > 99% using a FACSAria II cell sorter (BD Biosciences). CD4+ T cells were purified from healthy 2D2 mice by MACS with the untouched CD4 T cell isolation kit (Miltenyi Biotec) and CFSE labeled according to the manufacturer’s instructions (CellTrace™, Life Technologies, Carlsbad, USA). Bone marrow-derived DCs (bm DCs) were MACS purified with CD11c microbeads (Miltenyi Biotec) on day 7.
Splenic NK cells from laquinimod- or vehicle-treated mice were MACS sorted with the untouched NK cell isolation kit, separated into CD27 and CD11b single-positive NK cells by cell sorting and cultured at different effector/target ratios with 5000 B16F10 melanoma cells in the presence of 5000 U/ml rh IL-2 for 48 h. The tumor-lytic capacity was assessed by the crystal violet assay.
MACS-purified bmDCs were stimulated with 50 ng/ml LPS (Sigma Aldrich, St. Louis, USA) overnight, washed carefully, and co-cultured with MACS-sorted, CFSE-labeled 2D2 T cells and purified NK cells derived from the spleens of laquinimod- or vehicle-treated animals. Fifty thousand DCs, 50,000 NK cells, and 100,000 2D2 T cells/well were cultured for 72 h with or without 20 μg/ml MOG35–55, and the CFSE profile was analyzed by flow cytometry. In some experiments, NK cells and DCs were separated by a transwell membrane (Corning, Corning, USA) or T cells were stimulated in the absence of DCs with plate-bound αCD3 (145-2C11, 4 μg/ml) and soluble αCD28 (PV-1, 1 μg/ml) antibodies.
Purified splenic NK cells from laquinimod- or vehicle-treated animals were stimulated overnight with 1 ng/ml IL-12 and 25 ng/ml IL-18 (both PeproTech, Rocky Hill, USA), and IFNγ concentrations in the supernatant were determined by ELISA (R&D, Minneapolis, USA). Alternatively, IFNγ was analyzed in purified NK cells by intracellular cytokine staining, which were stimulated with IL-12 and IL-18 for 18 h, the final 6 h in the presence of GolgiStop™ (BD Biosciences, Franklin Lakes, USA).
Untouched mouse NK cells purified by MACS and cell sorting were cultivated in RPMI 1640 supplemented with 10% FCS, P/S, 2 mM L-glutamine, 1 mM sodium pyruvate, 50 μM β2-mercaptoethanol, and 1000 U/ml rh IL-2 for 48 h in the presence of 10 μM laquinimod.
Untouched human NK cells were purified by MACS and cell sorting and cultured in RPMI 1640 supplemented with 10% FCS, 2 mM L-glutamine, 1 mM sodium pyruvate, 50 μM β2-mercaptoethanol, and 50 U/ml rh IL-2 (Novartis, Nuremberg, Germany) for 120 h in the presence of 15 μM laquinimod.
In vivo tumor assay
For the induction of lung metastases, animals treated with laquinimod (50 mg/kg) or vehicle were i.v. transferred into the tail veins with single-cell suspensions of B16F10 cells. Lungs were obtained on day 10 or day 19, weighted, and immersion fixed in 4% PFA. The tumor nodules were counted on both sides of each lung lobe with the aid of a dissection microscope.
Flow cytometry
Single-cell suspensions were incubated for 15 min in Fc-blocking buffer and stained with the following anti-mouse antibodies (if not stated otherwise, all from BioLegend, San Diego, USA, or eBioscience, San Diego, USA): αCD3 (145-2C11), αCD4 (RM4-5), αCD8 (53-6.7), αCD11b (M1/70), αCD11c (N418), αCD16 (R&D Systems, Wiesbaden, Germany, 275003), αCD19 (eBio1D3), αCD27 (LG.3A10), αCD40 (3/23), αCD45 (30-F11), αCD80 (16-10A1), αCD86 (GL-1), αCD112 (R&D, 829038), αCD155 (TX56), αAHR (4MEJJ), αDNAM-1 (10E5), αNKG2D (CX5), αTIGIT (1G9), αLy49C/F/I/H (14B11), αLy49C/I, αLy49D, αMHC class II (M5/114), αTRAIL (N2B2), αNK1.1 (PK-136), αNkp46 (29A1.4), and αIFNγ (XMG1.2).
Anti-human antibodies were as follows: αCD3 (SK7), αCD16 (3G8), αCD14 (Beckman Coulter, Brea, USA, RMO52), αCD19 (HIB19), αCD56 (BD, NCAM16.2), αCD69 (FN50), αCCR7 (TG8), αDNAM-1 (11A8), αNKG2D (1D11), αNKp30, αNKp46 (9E2), and αTIGIT (MBSA43).
For intracellular staining, cells were fixed after surface staining for 45 min and permeabilized for 30 min with permeabilization buffer/antibody using the BD intracellular staining kit.
All flow cytometry data were acquired on a BD FACSCanto™ II and analyzed with FlowJo Tree Star software.
Statistical analysis
Statistics were calculated using GraphPad Prism 5.01. Data were tested for normal distribution with the help of the Kolmogorov-Smirnov test with Dallal-Wilkinson-Lillie for p value. To compare two experimental groups, unpaired t tests were used for parametric data and Mann-Whitney U tests for non-parametric data. To compare three or more groups, one-way ANOVA with Bonferroni or Dunnett’s post-test was performed for parametric data and the Kruskal-Wallis test with Dunn’s post-test was applied for non-parametric data. Survival analysis was calculated with the log-rank test. All statistical analyses of EAE scores in Rag1−/− and Th/+ mice after NK cell depletion were performed using two-way ANOVA with Tukey’s multiple comparison test. Statistical significance was defined as p < 0.05. Data are presented as mean ± SEM, if not otherwise stated.
Discussion
The bidirectional crosstalk between DCs and NK cells can influence the development of adaptive immune responses, thus providing exciting possibilities for the treatment of autoimmune diseases. Laquinimod, a prototypic quinoline-3-carboxamide, modulates important functions of both innate immune cell players. It reduces CD11c
high MHCII
+ DCs [
29], which has been addressed in previous studies and activates NK cells. Both responses depend on the presence of the AhR. In the present study, we report that the AhR-mediated activation of NK cells augments anti-tumor immunity and is relevant for maintaining the efficacy of laquinimod to ameliorate CNS autoimmunity.
Laquinimod’s activating effect on NK cells and down-modulatory effect on DCs is compatible with an AhR agonistic activity [
39], and recent studies have shown that the AhR is involved in NK cell activation in mice and humans. NK cells from AhR-deficient mice had poorer cytotoxic activity against RMA-S lymphoma cells as compared to NK cells from wild-type mice [
39], treating animals with the potent AhR ligand 6-formylindolo (3,2-b) carbazole (FICZ) enhanced NK cell control of RMA-S tumors. DNAM-1, which is consistently upregulated on AhR-competent NK cells by laquinimod, is an adhesion molecule, originally shown to influence NK and T cell cytotoxicity upon interaction with its ligands, CD112 and CD155 [
43]. In this regard, DNAM-1 appears to be relevant for the control of poorly immunogenic B16F10 melanoma metastases [
5,
44]. Correspondingly, laquinimod-derived NK cells killed in vitro B16F10 melanoma cells better than NK cells from vehicle-treated controls
. Likewise, they reduced the number of pulmonary metastases in vivo, albeit only when the treatment started at incipient stages. Our results are in line with the current paradigm that increasing NK cell activity is preferentially beneficial against circulating tumor cells but less efficient once the tumor cells have extravasated and formed solid tumors [
45].
Of note, besides shifting the NK cell subset distribution towards CD27
+ NK cells, laquinimod activates and improves the effector functions of both CD27 and CD11b single-positive NK cells. To strengthen effector functions of CD27, single-positive NK cells might be particularly relevant for MS, since CD27 single-positive NK cells and their human CD56
bright NK cell counterparts are the dominant intrathecal and lymph node NK cell population [
46]. Furthermore, a number of studies suggested that not only the immune-regulatory functions of NK cells might be impaired in MS but also their differentiation from the more immature CD56
bright to the more mature CD56
dim stage.
Recent data have expanded our knowledge on the function of DNAM-1 in NK cell biology and demonstrate its relevance for immune synapse formation, stable target cell conjugates, NK cell education, and memory differentiation [
47‐
49]. In addition, DNAM-1 expression separates NK cells into two functional subsets [
41], with DNAM-1
+ NK cells producing higher amounts of cytokines, having better proliferation capabilities and providing better control of lung melanoma metastases compared to DNAM-1
− NK cells. Interestingly, DNAM-1 receptor expression on NK cells is decreased in both cancer and MS patients and associated with impaired NK cell functions [
50]. Therefore, increasing the number of DNAM-1
+ NK cells might not only have therapeutic implications in a variety of malignant human tumors, expressing DNAM-1 ligands [
51,
52], but also for restoring the NK cell-mediated control in autoimmune diseases.
The contribution of NK cells in MS pathogenesis still remains controversial, but temporal correlations between reduced NK cell numbers with decreased cytotoxicity and MS relapses have been established. These correlations suggest that NK cells may play a regulatory role in MS by a number of possible mechanisms: First, NK cells can inhibit cytokine secretion or kill T cells in a Fas-dependent or granzyme-dependent manner [
13,
53,
54]. Second, DNAM-1 CD155 interactions between NK cell and activated T cells might contribute to T cell cytolysis [
10]. Third, NK cells can ameliorate Th17-driven autoimmune responses by interacting with DCs through IFNγ and IL-27, which directs Tr1 T cell differentiation [
55] or by interacting with microglia cells within the CNS [
23]. Fourth, our data uncover an additional mechanism underlying how NK cells might inhibit autoreactive T cells, which requires cell-to-cell contact with CD155
+ DCs to suppress antigen-specific T cell proliferation. This interaction might impair antigen presentation by DCs, since DNAM-1 dose-dependently downregulates MHC class II expression without changing the expression of co-stimulatory molecules. The clinical relevance of this interaction is underscored by the impaired sustainability of laquinimod to suppress CNS autoimmunity in the absence of NK cells.
ADCC effector functions of NK cells are important for the therapeutic efficacy of monoclonal antibodies in treating human disorders [
56]. NK cells also lysed oligodendrocytes or MOG-transfected tumor cells in vitro in the presence of MOG-specific, patient-derived serum antibodies. We therefore analyzed whether NK cell activation is safe in the setting of pathogenic MOG-specific antibodies. We did not observe a clinical difference in Th/+ mice with or without NK cells, and laquinimod is more effectively protective in NK cell-competent than NK cell-deficient mice. Accordingly, the major receptor for ADCC in NK cells remains unchanged in response to laquinimod therapy.
Acknowledgements
The authors wish to express their gratitude to Heidi Brodmerkel, Olga Kowatsch, Brigitte Maruschak, Elke Pralle and Katja Schulz for excellent technical assistance. We are indebted to Cynthia Bunker for language editing. We thank Dr. Hayardeny for her constructive and stimulating discussions as well as Dr. Kaye for his support.