Interleukin-15-Cultured Dendritic Cells Enhance Anti-Tumor Gamma Delta T Cell Functions through IL-15 Secretion

Dendritic cell (DC) vaccination can be an effective post-remission therapy for acute myeloid leukemia (AML). Yet, current DC vaccines do not encompass the ideal stimulatory triggers for innate gamma delta (γδ) T cell anti-tumor activity. Promoting type 1 cytotoxic γδ T cells in patients with AML is, however, most interesting, considering these unconventional T cells are primed for rapid function and exert meaningful control over AML. In this work, we demonstrate that interleukin (IL)-15 DCs have the capacity to enhance the anti-tumoral functions of γδ T cells. IL-15 DCs of healthy donors and of AML patients in remission induce the upregulation of cytotoxicity-associated and co-stimulatory molecules on the γδ T cell surface, but not of co-inhibitory molecules, incite γδ T cell proliferation and stimulate their interferon-γ production in the presence of blood cancer cells and phosphoantigens. Moreover, the innate cytotoxic capacity of γδ T cells is significantly enhanced upon interaction with IL-15 DCs, both towards leukemic cell lines and allogeneic primary AML blasts. Finally, we address soluble IL-15 secreted by IL-15 DCs as the main mechanism behind the IL-15 DC-mediated γδ T cell activation. These results indicate that the application of IL-15-secreting DC subsets could render DC-based anti-cancer vaccines more effective through, among others, the involvement of γδ T cells in the anti-leukemic immune response.

cell (DC) vaccination is subject of intense examination in the MRD setting (5)(6)(7). DC vaccine strategies aim to generate an anti-tumor immune attack, by harnessing both the innate and adaptive arms of the immune system (6,8). This dual action is of interest, considering that both arms of the immune system are unequivocally important and cooperate in the generation of an effective anti-tumor immune response (9,10). For example, it has been shown that γδ T cells and CD8 + T cells synergistically target cancer stem cells (10). γδ T cells, an unconventional T cell subset in the twilight zone between innate and adaptive immunity, have the ability to provide strong and sturdy anti-tumor responses (11,12). Contrary to αβ T cells, activation of γδ T cells does not rely on antigen presentation by major histocompatibility complexes. On the other hand, γδ T cell activation can be induced by T cell receptor-dependent recognition of phosphoantigens, such as isopentenyl pyrophosphate (IPP). This intermediate of the mevalonate pathway is commonly overexpressed in AML blasts due to aberrations contributing to transformation (13,14). Moreover, given their direct killing capability, antigen-presenting capacity and skill to mobilize other components of the immune system (8,11), it is no coincidence that a recent meta-analysis of gene expression data from over 18,000 cancers, including hematological malignancies, identified the presence of γδ T cells to be the most significant factor associated with favorable prognosis (15). In line with this, γδ T cells exert meaningful control over AML (16). Harnessing γδ T cells by a DC vaccine is therefore an interesting approach to tackle MRD and to maximize the anti-tumor potency of vaccination [recently reviewed (8)]. However, little is known about DC-mediated γδ T cell activation and the widely used DC vaccines, that are commonly referred to as interleukin (IL)-4 DCs, have been found to be largely ineffective at stimulating γδ T cells (8).
IL-4 DCs are generated by stimulating monocytes with IL-4 and granulocyte-macrophage colony-stimulating factor for 5 days, followed by a maturation step of 1-2 days with a cytokine-based cocktail (17). We previously established a novel monocyte-derived DC generation protocol with improved potency by employing IL-15 and a toll-like receptor ligand, both generating strong immunostimulatory signals (17,18). These so-called IL-15 DCs have already proven themselves superior to the classic IL-4 DCs based on their direct cytolytic activity against tumor cells and their capacity to stimulate adaptive (17) and innate (19) anti-tumor immunity. Moreover, CD8 + T cells and natural killer cells respond to the chemokine milieu created by IL-15 DCs, whereas IL-4 DCs are suboptimal in attracting effector lymphocytes (20). This is of importance considering that vaccine DCs can only perform their directing and activating functions in vivo, provided that the effector cells come in their vicinity. γδ T cells too are more prone to migrate towards IL-15 DCs (20), suggesting the possibility of interaction between both cell types.
In the current study, we characterized the IL-15 DC-mediated γδ T cell activation on both the phenotypic and functional level and investigated the potential mechanisms involved. With this, we focused on DC vaccination as a therapeutic approach to surmount MRD in AML patients.

MaTerials anD MeThODs ethics statement and cell Material
This study was approved by the Ethics Committee of the Antwerp University Hospital (Edegem, Belgium) under the reference number B300201419756. Primary cells were isolated from buffy coats of healthy volunteers supplied by the Red Cross Blood Transfusion Center (Mechelen, Belgium). Immune cells of AML patients ( Table 1) were obtained as residual material of the WIDEA DC vaccination trial (NCT01686334) or obtained via the division of Hematology of the Antwerp University Hospital. Informed consent was received from all patients for being included in the study. Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll density gradient centrifugation. γδ T cells were isolated using a negative (EasySep, Cologne, Germany) or positive (Miltenyi, Leiden, The Netherlands) immu nomagnetic cell selection kit for cytokine production determination and cytotoxicity assays, respectively. γδ T cells isolated with the EasySep γδ T cell isolation kit were >90% pure, whereas with the anti-TCRγ/δ microbead kit of Miltenyi a purity of >95% was routinely obtained. The Burkitt's lymphoma tumor cell line Daudi, a known target for γδ T cells, was kindly provided to us by the laboratory of Prof. Kris Thielemans (Free University of Brussels, Brussels, Belgium). The chronic myeloid leukemia cell line in blast crisis K562 was purchased from the American Type Culture Collection (Rockville, MD, USA) and the AML cell lines NB4 and THP-1 were obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (Braunschweig, Germany).

co-culture setup
PBMCs, unless stated otherwise, were cultured with IPP (122 µM; Tebu-bio, Le-Perray-en-Yvelines, France), autologous IL-15 DCs, and/or tumor cells (K562, Daudi) at an effectorto-target cell (E:T) ratio of 1:(1:) 10. Unstimulated cells were used as negative control. With regard to cultures with immune cells of AML patients (Table 1), the same culture setup was used let alone the use of peripheral blood lymphocytes (PBLs) instead of PBMCs.

Proliferation assay
Proliferation of γδ T cells was quantified by the degree of 5,6carboxyfluorescein diacetate succinimidyl ester (CFSE; Invitrogen) dye dilution. At different time points, γδ T cell proliferation was measured as the proportion of CFSE-diluted cells within the viable CD3 + γδTCR + gate (Live/Dead ® Fixable Aqua Stain, CD3-PerCP [BD], γδ TCR-APC). Unstimulated CFSE-labeled cells served as a non-dividing control.

Defining contact-and il-15 Dependency
To examine the role of cell-cell contact, IL-15 DCs were separated from γδ T cells by a 0.4-µm pore size Transwell insert (Elscolab, Kruibeke, Belgium). In specific experiments, 10 µg/mL anti-IL-15 neutralizing mAbs (R&D, Minneapolis, Canada) or corresponding isotype control mAbs were added to investigate the involvement of IL-15. IL-15 DCs were incubated for 1 hour with the above mAbs before starting co-culture to ensure adequate blocking of IL-15. The mAbs remained present during the further course of the co-culture.

γδ T cell-Mediated cytotoxicity assays
In order to measure the tumoricidal capacity of (un)stimulated γδ T cells, a flow cytometric assay was setup as previously described with minor modifications (21)(22)(23)(24). 0.2 × 10 6 isolated γδ T cells were cultured for 24-36 hours in the presence or absence of IPP (122 µM) and/or autologous IL-15 DCs at a DC: γδ T cell ratio of 1:10 in 200 µL RPMI (Invitrogen) supplemented with 10% fetal bovine serum (FBS; Invitrogen). Subsequently, tumor cells (Daudi, K562, NB4, and THP-1) and fresh allogeneic primary AML blasts (UPNa/b) were labeled with CellTracker™ Blue CMF2HC Dye (Invitrogen) and served as "target cells". They were added to the γδ cell ± DC (co-)cultures for 4 hours at an E:T ratio of 2:1 in a final volume of 300 µL RPMI + 10% FBS. Cultures with allogeneic AML material were stained with CD33/CD34 to discriminate between AML blasts and the rest of the PBMC fraction of UPNa/b. Cell death was quantified after Annexin-V-APC (BD) and propidium iodide (PI) staining. Specific target cell killing was calculated according to the formula: % killing = 100 - statistics All flow cytometry data were acquired on a FACSAria II flow cytometer (BD) and analyzed using FlowJo (v10; Treestar, Ashland, OR, USA). For statistics and illustrations, GraphPad Prism software (v5.0; San Diego, CA, USA) was used. To validate the Gaussian distribution of data sets, they had to pass Shapiro-Wilk test for normality. Dissimilarities were predefined to be statistically significant when p < 0.05. All data are depicted as means ± SE of the mean.  Table 2; % and ΔMFI, Figure  S2 in Supplementary Material; histogram overlays). γδ T cells exposed to IL-15 DCs were found to express higher levels of markers related with γδ T cell cytotoxicity (CD16) and co-stimulatory molecules (CD80, CD86). Further upregulation of CD16, NKp30, CD80, and CD86 was observed with the combination of IL-15 DCs with IPP. In contrast, the expression of the co-inhibitory receptors and exhaustion-associated molecules BTLA and PD-1 on γδ T cells were not or only limited influenced by IL-15 DCs. The addition of IPP, however, resulted both in a significant induction and upregulation of PD-1 on γδ T cells.

soluble Factors secreted by il-15 Dcs Promote γδ T cell Proliferation
Next, we established the influence of IL-15 DCs on resting γδ T cells in full PBMC fraction through a standard 5-day proliferation assay (Figures 1A,B). IL-15 DCs induced significant γδ T cell proliferation, i.e. 25.13 ± 1.33% after 5 days. This proliferative response was doubled (54.98 ± 6.25%) in the presence of IPP. However, there was no clear proliferative reaction of the γδ T cells to IPP stimulation only after 5 days. When looking at a longer period, i.e. 8 and 12 days, IPP addition does result in γδ T cell proliferation (day 12; 33.60 ± 1.25%). Moreover, regarding these later time points, it is evident that the stimulatory effect of IL-15 DCs is relevant for the generation of high numbers of γδ T cells long term, namely 63.37 ± 8.81% γδ T cell proliferation after a 12-day co-culture ( Figure 1D). To end, a similar degree of γδ T cell proliferation was observed after 5 days, irrespective of whether IL-15 DCs were separated from the PBMCs by a transwell membrane or not. This indicating that soluble factors secreted by IL-15 DCs were responsible for the observed proliferative response ( Figure 1C).

il-15 Dcs stimulate iFn-γ Production in γδ T cells in the context of a leukemic environment
One of the desirable effector functions of γδ T cells in cancer immunotherapy is the secretion of pro-inflammatory cytokines, such as IFN-γ and TNF-α. Therefore, we measured the intracellular IFN-γ production in γδ T cells after overnight stimulation (Figures 2A,B). Stimulation of γδ T cells with IPP and/or tumor cells without DCs failed to trigger IFN-γ production in γδ T cells. IL-15 DCs alone also did not provide the necessary signals for IFN-γ production. On the other hand, when IL-15 DCs were added together with signals of malignancy (presence of IPP and tumor cells) a strong IFN-γ response was observed in γδ T cells. As shown in Figure 2A, stimulation with IL-15 DCs + IPP + tumor cell lines resulted in 13.12 ± 2.68% (for K562 cell line) and 7.35 ± 1.24% (for Daudi cell line) IFN-γ + γδ T cells. TNF-α production was absent in all situations, as well as the T helper type 2 (Th2)-promoting cytokine IL-4, the Th17-promoting cytokine IL-17 (Th17), and the anti-inflammatory cytokine IL-10 (data not shown). In analogy with the induction of γδ T cell proliferation, the stimulation of IFN-γ production in γδ T cells by IL-15 DCs was found to be contact-independent ( Figure 2C).

activation of γδ T cells by il-15 Dcs
is Dependent on soluble il-15 T cell activation. We previously performed a cDNA microarray analysis of (im)mature IL-15 DCs and "gold-standard" IL-4 DCs of three independent healthy controls (20 Figure 3A). To clarify the involvement of this pleiotropic cytokine, IL-15 effects were canceled out using neutralizing mAbs (Figures 3B,C). IL-15 DC-mediated γδ T cell proliferation was reduced by approximately 60% upon IL-15 neutralization. Concerning IFN-γ production, blocking IL-15 significantly reduced the ability of γδ T cells to produce IFN-γ upon stimulation with IL-15 DCs in a malign environment.

il-15 Dcs Potentiate γδ T cells From healthy Donors to Kill leukemic cell lines and allogeneic Primary aMl Blasts
After establishing the stimulatory effect of IL-15 DCs on γδ T cell proliferation and pro-inflammatory cytokine production, we looked at their ability to improve γδ T cell lysis of a panel of leukemic tumor cell lines and primary AML blasts. As shown in Figure 4,    Figure S4 in Supplementary Material). At the outset, we looked at the cytotoxic phenotype of γδ T cells of AML patients (after activation) ( Table 3; % and ΔMFI, Figure S3 in Supplementary Material; histogram overlays). The expression level of CD16, both basally as well as after exposure to IL-15 DCs, was similar to that of healthy donors. Interestingly, whereas IL-15 DCs of healthy volunteers did not affect the expression of NKG2D, we noticed a trend towards upregulation of this marker on γδ T cells of AML patients upon co-culture with IL-15 DCs. Of note, UPN2 had a low basal expression of both markers, but showed a similar reaction to activation as compared to γδ T cells from healthy donors. In UPN4, baseline CD16 and NKG2D expression was already high, making it difficult to observe any additional activation. Strikingly, in all patients a robust, more than 2-fold, induction of NKp30 expression on the γδ T cell surface was observed upon IL-15 DC stimulation (21.73% in IL-15-DC-stimulated γδ T cells versus 9.5% in unstimulated γδ T cells).

Functionality of il-15 Dc-activated γδ T cells of aMl Patients in remission
Finally, we aimed to examine the functionality of γδ T cells of AML remission patients upon IL-15 DC stimulation. The percentage of proliferated γδ T cells after 5-day co-culture with IL-15 DCs was 30.57 ± 1.33% in AML patients (Figure 5A), comparing favorably to the proliferative response in healthy donors. With regard to IFN-γ production by γδ T cells (Figure 5B), we observed an enhanced IFN-γ production, i.e. 15-20% intracellular IFN-γ expression, upon stimulation with IL-15 DCs in a leukemic environment in AML patients that were brought into complete remission after induction/consolidation chemotherapy (UPN1 and UPN2). However, in the other two patients (UPN3 and UPN4), who had only completed induction chemotherapy at the time of the blood draw, IL-15-DC stimulation on its own already led to a noticeable IFN-γ production by γδ T cells but with a diminished response to stimulation with additional IPP and tumor cells.

DiscUssiOn
The clinical implementation of γδ T cells in cell-based cancer immunotherapy holds encouraging perspectives regarding the treatment of (advanced) malignancies (12,25). This prospect is not a remote possibility anymore, evinced by the burst of several new biotech companies aiming at bringing γδ T cells into the clinic, including Gadeta (Utrecht, The Netherlands), GammaDelta Therapeutics (London, UK), Incysus (Hamilton, Bermuda), and Lymphact (Coimbra, Portugal). In this study, we aimed at harnessing γδ T cells in the anti-tumor immune response of DC vaccination, a promising and active immunotherapeutic modality, amongst others, used in the MRD setting of AML (6,7). We recently showed that WT1 mRNA-electroporated conventional IL-4 DCs prevent or delay relapse in 43% of AML patients in remission after chemotherapy (7). However, since the IL-4 DC vaccine only weakly recruits γδ T cells (20) and generally fails to potentiate γδ T cell functions (8), this missing interaction could aid to optimize DC vaccine immunopotency, promoting cytotoxic and long-lasting anti-tumor immunity. Here, we show that our immunogenic IL-15 DC vaccine (18,20) is able to activate γδ T cells (of AML patients in remission) and that this is attributable to a large extent to the secretion of IL-15.
First, we looked at the phenotypic features of γδ T cells upon interaction with IL-15-DCs. This has highlighted that γδ T cells of AML patients are prone to IL-15 DC stimulation, particularly in the presence of IPP. The latter is evidenced by the marked upregulation of NKp30 on the γδ T cell surface. This natural cytotoxicity receptor has been shown to be the main contributor to TCR-independent leukemia cell recognition by γδ T cells, even enabling resistant primary lymphocytic leukemia targeting (26). Interestingly, CD80 and CD86, specific antigen-presenting (APC) cell markers involved in co-stimulatory signaling to αβ T cells, are clearly upregulated on γδ T cells after interaction with IL-15 DCs. This could be of importance, considering γδ T cells have been shown to adopt an APC phenotype upon activation and to effectively (cross-)present peptide antigens (27,28). Moreover, the upregulation of CD16 on the cell membrane enables antibody-dependent cellular cytotoxicity and the efficient "licensing" of γδ T-APCs (29). Subsequently, we looked at negative co-stimulatory markers associated with T cell exhaustion, i.e. BTLA and PD-1 (30). These inhibitory molecules were not, or only to a limited extent, induced by IL-15 DCs on γδ T cells. On the other hand, we are the first to show that IPP causes PD-1 expression. This is of importance in view of the increased IPP expression by cancerous cells and the compelling evidence on the stimulatory effect of IPP for γδ T cell activation and expansion (8). Interestingly, γδ T cells are able to surmount (in part) the inhibitory effects mediated by PD-1 by means of TCR triggering (31). Indeed, sensitization of tumor cells with zoledronate, a nitrogen-containing bisphosphonate resulting in the accumulation of IPP through the inhibition of farnesyl pyrophosphate synthase (14), overcomes the inhibitory effect of PD-1. Therefore, the relationship between IPP signaling and PD-1 expression/functioning warrants further investigation. Besides, DC therapy in concert with anti-PD-1 therapeutics is a combination to be investigated, possibly rendering AML cells more susceptible to immune attack. This especially considering the expression of PD-1 ligands is a well-recognized mechanism by which AML cells inhibit anti-leukemic immune cell responses (32,33). The results of a clinical trial combining anti-PD-1 therapeutics with a DC-based AML vaccine are therefore awaited (34).
To date, the capacity of circulating DCs to trigger γδ T cell activation is largely unknown. Notwithstanding, we and others  have shown that autologous ex vivo generated IL-4 DCs, used routinely for clinical studies, are inefficient in mobilizing γδ T cells (20) and unable to induce γδ T cell proliferation and effector functions, and that additional/alternative signals are required (35). In this study we provide evidence that IL-15 DCs are able to induce autologous γδ T cell proliferation and a Th1like polarization profile and that these features were conserved in AML patients who are in complete remission. Perhaps even more important, IL-15 DCs are able to significantly upgrade γδ T cell cytotoxicity against leukemic cell lines and primary AML blasts. This makes the IL-15 DC vaccine an all-round activator of the cytotoxic immune effector response, to wit γδ T cells, NK cells (19) and conventional T cells (17). The interesting observation that γδ T cells from AML patients before consolidation chemotherapy exhibited a different functional profile with regard to IFN-γ production as compared to that of patients after a consolidation regimen needs to be confirmed in a larger cohort of AML remission patients. This might highlight the importance of timing of administration of γδ T cell-activating immunotherapeutic strategies in AML (36). Future work will also need to reveal if patients would benefit of the addition of IPP to the vaccine or if there is sufficient IPP present on the leukemic residual cells to enhance γδ T cell activation. Seminal work of Gundermann et al. has already shown that if AML cells are pretreated with zoledronate, they display a significantly augmented predisposition to γδ T cell-mediated cytotoxicity. The latter suggests a potential benefit of the addition of IPP to the vaccine or the combination with a nitrogen-containing bisphosphonate (14,37). This is supported by our current data showing an enhanced proliferative response and IFN-γ production by γδ T cells to IL-15 DC stimulation in combination with IPP. At the same time, no apparent effect could be detected of the addition of a phosphoantigen as regards γδ T cell cytotoxicity.
Of note, phosphoantigens, such as IPP, activate γδ T cells in a TCR-mediated manner, whereby the expression of ubiquitous butyrophilin (BTN) proteins CD277/BTN3A on stimulator cells were shown to be indispensable (38)(39)(40)(41). In accordance with other data showing that BTN3 molecules are widely expressed by immune cells, some tumor cell lines and even monocyte-derived DCs (42), it can be assumed that the presentation of IPP to γδ T cells was not hampered in our experiments. Of note, since BTN3A is a determining factor of γδ T cell-recognition of AML cells and agonistic BTN3A mAbs have been show to circumvent primary AML blast resistance to allogeneic γδ T cell lysis, combination therapy is here too something to consider (43).
Finally, focusing on the IL-15 DC-mediated γδ T cell activation itself, the stimulatory effect seemed to be mainly attributable to the secretion of IL-15 by the IL-15 DC vaccine. IL-15 is a well-documented regulator of homeostasis and activator of both innate and adaptive immunity (44). Its non-redundant role is reflected by the fact that CD8 + memory T cells and NK cells are absent in IL-15-deficient environments (45,46), and the lack of IFN-γ-producing γδ T cells in IL-15 −/− mice (47). Moreover, the anti-tumor effect of IL-15 on the immune system has been well-documented in experimental systems (48)(49)(50). Concerning IL-15-mediated γδ T cell activation, our findings are in line with data describing γδ T cell proliferation in rhesus macaques after continuous administration of IL-15 (51) and an increase in absolute γδ T cell counts, accompanied by an upsurge in proliferating γδ T cells, in the first-in-human trial of recombinant IL-15 in cancer patients (52). Previously, we have shown that IL-15 supports γδ T cell proliferation ex vivo and that the addition of IL-15 to γδ T cell cultures results in a more pronounced Th1 polarization and an increased cytotoxic capacity of γδ T cells (53). In addition, pediatric thymic tissue has been used to unravel the process that drives human γδ T cell differentiation toward anti-tumor lymphocytes. They have confirmed that IL-15 signaling is sufficient to guide human γδ T cells along the Th1 pathway, coinciding with a strong killing capacity of leukemia cells (54). Moreover, the lack of IL-15 production by M. tuberculosis-infected DCs leads to deficient γδ T cell effector functions, impeding the conversion of central memory γδ T cells into effector memory cells (55). Although IL-15 is a promising immunotherapeutic candidate, and currently tested in clinical trials in AML (National Cancer Institute Trial IDs; 2016LS056 and 2016LS058), the stability and side effects of recombinant IL-15 are some of the bottlenecks to be overcome (44). The targeted release of IL-15 by IL-15 DCs might therefore offer an important advantage over the systemic delivery of recombinant IL-15, namely avoiding substantial toxicity (52).
In summary, we have shown that the IL-15 DCs are able to harness γδ T cells in their anti-tumoral activity via secretion of soluble IL-15. These data support the implementation of IL-15expressing DCs into future clinical trials.

eThics sTaTeMenT
This study was carried out in accordance with the recommendations of the Ethics Committee of the Antwerp Univer sity Hospital (https://www.uza.be/ethics-committee-uza) with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the Ethics Committee of the Antwerp University Hospital. aUThOr cOnTriBUTiOns HVA, ZB, ES, and VVT conceived and designed the experiments. HVA and HDR performed the experiments. HVA analyzed the data. HVA, SA, ES, and VVT wrote the paper. All authors have read and approved the manuscript in its current form.

acKnOWleDgMenTs
The authors would like to thank Dr. Stef Meers (AZ Klina, Brasschaat, Belgium), Dr. Christine Schuermans, and Dr. Jan Lemmens (GZA hospitals, Wilrijk, Belgium) for providing primary AML samples. Furthermore, the authors wish to acknowledge Dr. Ronald Malfait (Antwerp University Hospital, Edegem, Belgium) for providing excellent technical assistance.

FUnDing
This work was supported by grant G.0399.14 from the Fund for Scientific Research Flanders (FWO Vlaanderen) and by a Methusalem grant from the University of Antwerp. HVA is supported by a PhD grant of the FWO.