TCRVγ9 γδ T Cell Response to IL-33: A CD4 T Cell–Dependent Mechanism Free

The γδ T cells expressing the TCR Vγ9 represent 1–3% of blood mononuclear cells and are therefore the prominent γδ T cell subset in human peripheral blood. These so-called unconventional T lymphocytes respond selectively to various non–peptide phosphoantigens (PAgs), which are metabolites from two distinct biosynthesis pathways, namely the prokaryotic methyl erythritol phosphate pathway in microbial pathogens (1) and the eukaryotic mevalonate pathway in tumor cells (2). These cells can also be activated by direct treatment with aminobisphosphonates such as zoledronic acid because this upregulates the endogenous biosynthesis of PAgs in mammalian cells (3). Furthermore, recent studies have illustrated a selective activation of Vγ9 T lymphocytes by Abs targeting cell surface BTN3A1, a member of the butyrophilin receptor family (4). Surprisingly, the activation of Vγ9 T lymphocytes by these three distinct types of soluble stimuli leads to the same pattern of immune response.

In vitro and in vivo experiments have shown that in the presence of exogenous IL-2, PAg-stimulated Vγ9 T cells proliferate vigorously, secrete proinflammatory cytokines and chemokines, and exert a potent cytotoxic activity against a broad spectrum of malignancies (5, 6). Vγ9 T cells spontaneously recognize and lyse various Burkitt’s lymphomas, as well as some anaplastic large-cell lymphomas and non-Hodgkin’s lymphomas, all of which are thought to overproduce endogenous PAgs. Overproduction of endogenous PAgs might reflect the metabolic biases of cancer cells, and presumably occurs in most if not all types of tumors. Accordingly, tumor-infiltrating Vγ9 T cells have been detected in several solid and hematopoietic malignancies (7, 8). Combinations of IL-2 with either PAg or zoledronate have been used to harness these cells for therapeutic purposes in vivo, both in several preclinical studies in macaque monkeys (9, 10) and in clinical trials in cancer patients (reviewed in Ref. 11). Although these trials resulted in a significant amplification of circulating Vγ9 T cells, their stringent dependency on an exogenous supply of IL-2 has raised concerns due to its inherent toxicity for patients. This prompted us to investigate alternative protocols that could activate Vγ9 T cells selectively without requiring an exogenous supply of IL-2.

IL-33 is a cytokine that was initially designated NF from high endothelial venules (HEV), as it was found associated with chromatin in the nucleus of endothelial cells from HEVs, specialized blood vessels that mediate lymphocyte recruitment into lymphoid organs (12, 13). IL-33 is the ligand of ST2, a member of the IL-1R family belonging to the TLR/IL-1R superfamily (14). IL-33 is constitutively expressed by HEVs (12), but also by endothelial cells in many normal human tissues, by dermal fibroblasts and keratinocytes, and by epithelial cells in tissues exposed to the environment (15, 16). IL-33 is also overexpressed in some circumstances such as during mechanical or UV stress (17, 18) and in inflammatory and autoimmune diseases such as asthma (19, 20). Furthermore, IL-33 is abundantly expressed in the nucleus of endothelial cells in human tumors of the kidney, stomach, liver, and pancreas (15).

IL-33 passively released during necrosis (21) was proposed to function as an alarmin, alerting the immune system (16, 21–23). IL-33 promotes the recruitment of Th2 cells and eosinophils in vitro and in vivo to inflammatory lesions and induces the production of IL-5 and IL-13 cytokines by Th2 lymphocytes (14, 24, 25). It also strongly activates type 2 innate lymphoid cells (ILC2s) and mast cells when injected into mice (26–31). Human blood-derived basophils also respond to IL-33 by producing several proinflammatory cytokines and exhibiting enhanced adhesion, integrin expression, chemotaxis, degranulation, and increased survival (20, 27, 32, 33). IL-33 also induces the upregulation of cell-surface expression of MHC class II molecules and CD86 on dendritic cells (34). Moreover, IL-33 administration into mice also amplifies the immunoregulatory CD11b⁺Gr-1⁺ myeloid-derived suppressor cells and CD4⁺CD25^hiFoxp3⁺ regulatory T cells (35, 36).

Additionally, IL-33 increases the in vitro production of IL-4 and IFN-γ by human invariant NKT and NK cells (27) and enhances the clonal expansion of invariant NKT in IL-33–treated mice (32). Moreover, IL-33 induces the in vitro cytotoxic activity of NK cells and CD8 T cells (37) and can drive the protective antiviral CD8 T cell responses in vivo (38). Thus, IL-33 is a multifaceted immune promoter of both Th1 and Th2 immunity. However, despite the extensive literature available on these bioactivities, its impact on γδ T cells has not been investigated so far. The detection of tumor-infiltrating Vγ9 T cells in a broad spectrum of malignancies that also turn out to frequently express IL-33 led us to examine the biological activity of this alarmin/cytokine on the functions of Vγ9 T cells.

In this study, we report that, in vitro, IL-33–activated CD4 T cells leads to amplifying of the whole spectrum of Vγ9 T cell responses induced by PAgs and other stimuli as effectively as the classical combination of PAg/IL-2. The cell contact of ST2-devoid Vγ9 T cells with CD4 T cells underlies the indirect ability of IL-33 to potently promote the anticancer functions of PAg-activated Vγ9 T lymphocytes, opening new perspectives for future γδ T cell–based clinical trials.

The mouse Abs conjugated to human molecules used are listed in Supplemental Table I. We used respective isotype controls.

Bromohydrin pyrophosphate (BrHPP) and recombinant human (rh)IL-33 were produced in our laboratory, and anti-BTN3A1 mAb clone 20.1 was provided by Daniel Olive (Centre d’Immunologie de Marseille–Luminy, Marseille, France). rhIL-2 was gifted from Sanofi-Aventis (Toulouse, France). CFSE was purchased from Life Technologies (Saint-Aubin, France). Brefeldin A and 4′,6′-diamidino-2-phenylindole (DAPI) were from Sigma-Aldrich (Lyon, France). ⁵¹Cr radionuclide was obtained from PerkinElmer (Courtaboeuf, France).

Human PBMC were isolated from healthy donors using density gradient centrifugation with Ficoll-Paque Plus (Thermo Fischer Scientific, Illkirch, France) and cultured at 1.5 × 10⁶ cells/ml in complete medium consisting of RPMI 1640 (Lonza, Les Mureaux, France) supplemented with 10% FCS clone I (Thermo Fischer Scientific) and 2 mM l-glutamine (Lonza), 100 μg/ml streptomycin, 100 IU/ml penicillin, and 1 mM sodium pyruvate (PAA Laboratories, Velizy-Villacoublay, France). To produce Vγ9 primary cell lines (γδ-pcl) containing >50% Vγ9 T cells, 3 μM BrHPP was added to the PBMC culture in complete medium at day 0 and rhIL-2 (IL-2 γδ-pcl; 400 IU/ml) or rhIL-33 (IL-33 γδ-pcl; 100, 500, or 1000 ng/ml) was added at day 0 and then every 3 d until 10 d. Cell concentration was maintained at 1.5 million/ml every 3 d.

Freshly isolated γδ cells (γδ, purity ≥ 90%) were obtained by negative magnetic cell sorting (Miltenyi Biotec) of freshly isolated PBMC.

Specific population depletions were performed from freshly isolated PBMC by positive magnetic cell sorting (Miltenyi Biotec) targeting CD14, CD19, CD4, CD8, CD56, CD1c, CD141, or CD304. For ILC2 isolation, PBMC were stained with lineage markers CD3, CD14, CD20, CD56, CD117, and CD127 and cells were sorted by automatic fluorescence cell sorting with a BD FACSAria III.

γδ-pcl, PBMC, and depleted PBMC were labeled with 1 μM CFSE for 8 min at 37°C and cultured (at 1.5 × 10⁵ to 3 × 10⁵ cells/ml) in 96-well plates in complete medium with either BrHPP (100 nM), isopentenyl pyrophosphate (IPP; 1 μM), zoledronate (10 μM), or anti-BTN3A 20.1 mAb (10 μg/ml) and either rhIL-33 (100, 500, or 1000 ng/ml) or rhIL-2 (100 IU/ml) for 6 d. For the specified experiments, 10 μg/ml blocking anti-CD25 mAb (R&D Systems) was added to the culture. Then, CFSE dilution in TCR Vγ9⁺ CD3⁺ cells was evaluated by flow cytometry analysis.

For the proliferation assay using the Transwell system, magnetically purified CD4 T cells (5 × 10⁴) were placed in the upper chamber and CD4 T cell–depleted PBMC (2 × 10⁵) were placed in the lower chamber in complete medium (96-multiwell insert system; BD Biosciences) then cultured for 6 d as described above.

Cells (5 × 10⁵) from IL-33 γδ-pcl or IL-2 γδ-pcl were stained for the specified markers and analyzed by flow cytometry analysis. For intracellular cytokine and lytic granule expression, cells from IL-33 γδ-pcl or IL-2 γδ-pcl were cocultured for 4 h with Daudi cells (E:T ratio 1:1) in 96-well plates in complete medium supplemented with 10 μg/ml brefeldin A. Then, cells were fixed with PBS 2% paraformaldehyde and permeabilized with PBS containing 5% FCS and 1% saponin (Sigma-Aldrich) prior to staining for 30 min with the specified mAb for flow cytometry analysis.

IL-33 γδ-pcl or IL-2 γδ-pcl were cocultured with Daudi cells (E:T ratio 1:1) in 96-well plates in complete medium with anti-CD107a mAb (5 μg/ml) with or without anti-Fas ligand (FasL) and anti-TRAIL neutralizing Abs (10 μg/ml). After 4 h, cells were stained with anti-TCR Vγ9, anti-CD3, and anti-CD20 mAbs and then suspended in PBS plus DAPI (10 μg/ml). Analyses of CD107a expressed by Vγ9 T cells and DAPI staining of Daudi cells were performed by flow cytometry.

Cells from IL-33 γδ-pcl or IL-2 γδ-pcl were cocultured for 4 h in complete medium with Daudi cells (4000 cells/well) at the E:T ratios of 30:1, 10:1, and 3:1. Specific lysis of Daudi cells by IL-33 γδ-pcl or IL-2 γδ-pcl was measured by standard ⁵¹Cr-release assays. Lysis rates were obtained using [(experimental release − spontaneous release)/(maximum release − spontaneous release)] × 100.

Transcriptome raw data files were downloaded from the GSE27291 dataset available at the National Center for Biotechnology Information repository Gene Expression Omnibus database (39). These comprised 12 samples of TCRVγ9⁺ γδ cells highly purified (>98% purity) from PBMC: four “resting control” samples, four “6 h after BrHPP activation” samples, and four “7 d after BrHPP activation” samples. Data were log₂ transformed and analyzed with a GEO2R platform using the relative mRNA expression profile of the IL1RL1 gene normalized to that of GAPDH.

Flow cytometry acquisitions were performed on a BD LSR II cytometer (BD Biosciences) and data were analyzed and figures processed with Cytobank software (http://www.cytobank.org).

RNA was isolated using an RNeasy kit (Qiagen) and each RNA sample was treated with RQ1 RNase-free DNase (Promega). Generation of cDNA was carried out with the RevertAid first-strand cDNA synthesis kit (Thermo Scientific) according to the manufacturer’s instruction. Real-time PCR assays were carried out with the ABI Prism 7300 real-time PCR system (Applied Biosystems) using SYBR Green JumpStart Taq Ready Mix (Sigma-Aldrich) with the primers IL-2 forward, 5′-GTCACAAACAGTGCACCTAC-3′ and IL-2 reverse, 5′-GAAAGTGAATTCTGGGTCCC-3′ or GAPDH forward, 5′-AGGGCTGCTTTTAACTCTGGT-3′ and GAPDH reverse, 5′-CCCCACTTGATTTTGGAGGGA-3′. GAPDH was used as a reference gene. The amplification fold change was calculated with the ΔΔCT method.

Data are expressed as mean ± SEM, as specified in the figure legends. Differences between groups were analyzed using a paired two-tailed Student t test or, when more than two groups of samples were analyzed, an ANOVA test. Statistical analyses were performed with Prism software.

Because the proliferation of PAg-activated Vγ9 T cells requires IL-2, we first tested whether IL-33 could replace IL-2 in inducing the in vitro proliferation of these cells following stimulation by PAg. Freshly isolated PBMC were thus stained with the proliferation dye CFSE and cultured for 6 d in the presence of the synthetic PAg BrHPP plus either IL-2 or different doses of IL-33. We found that the division of Vγ9 T cells (shown gated among the PBMC, Fig. 1A) occurred in PBMC cultures stimulated with BrHPP plus IL-2 and with BrHPP plus IL-33. Additionally, IL-33 promoted Vγ9 T cell proliferation in a dose-dependent manner in cells stimulated by BrHPP but not in unstimulated cells, indicating that IL-33 was not mitogenic per se for these lymphocytes. In BrHPP-stimulated Vγ9 T lymphocytes, cell division reached ∼80% in culture media with either IL-2 or IL-33 (Fig. 1B). We conducted >10 independent assays and found that an initial PBMC culture containing ∼1% Vγ9 T cells (∼0.1 million cells) typically yielded ∼85% of Vγ9 T cells (∼13 million) in the presence of 400 U/ml IL-2 and ∼50% of Vγ9 T cells (∼5 million) with 500 ng/ml IL-33 (Fig. 1C, 1D). Importantly, Vγ9 T cell proliferation was similarly promoted following stimulation by other specific stimuli such as IPP, zoledronate, or the anti-BTN3A Ab 20.1 in the presence of either IL-2 or IL-33 (Fig. 1E).

Because IL-33 was found to promote the proliferation of Vγ9 T cells, we next investigated the phenotype and functions of the IL-33–induced Vγ9 T cell progeny. Primary cell lines of Vγ9 T cells were generated by 9 d of in vitro culture of freshly isolated PBMC in complete medium supplemented with a combination of BrHPP and IL-33 (IL-33 γδ-pcl) or BrHPP and IL-2 (IL-2 γδ-pcl). After 9 d, the presence of cell surface markers classically expressed by Vγ9 T cells was analyzed in both IL-2 γδ-pcl and IL-33 γδ-pcl by labeling with mAbs followed by flow cytometry analysis. These markers included the chemokine receptors CCR5, CCR6, and CXCR3 and the costimulation receptors and cytotoxicity receptors CD16, CD28, CD161, KIR, NKG2D, NKG2A, and TRAIL. These analyses indicated that IL-2 γδ-pcl and IL-33 γδ-pcl display exactly the same profile regarding these markers, namely that both types of cell lines were uniformly positive for the expression of CCR5, CXCR3, NKG2D, CD161, and TRAIL, whereas most but not all of them expressed CD16, CD28, and NKG2A (Fig. 2A). Also, both cell lines showed almost no expression of the cytotoxicity inhibitory receptors KIR2DL-1, -2, and -3, labeled by the 1-7F9 mAb. Hence, on this basis we found no phenotypic difference between IL-2 γδ-pcl and IL-33 γδ-pcl. Additionally, the maturation switch from central memory (CD27⁺CD45RA⁻) cells to effector memory (CD27⁻CD45RA⁻) cells, which typically occurs when Vγ9 T cells freshly isolated from PBMC are cultured in the presence of PAg and IL-2 (40), was observed in both IL-2 γδ-pcl and IL-33 γδ-pcl. Therefore, in agreement with previous analyses, most of the IL-33–induced Vγ9 T cells exhibited an effector memory phenotype (CD27⁻CD45RA⁻) equivalent to that of the IL-2–induced Vγ9 T cells (Fig. 2B). Both series of analyses indicated that IL-33 induced the same phenotypic pattern over time during the 9 d culture.

We then compared the functional activity of Vγ9 T cell lines raised by either IL-2 or IL-33. For this purpose, IL-2 γδ-pcl and IL-33 γδ-pcl were cultured with or without the Daudi lymphoma cell line, and the resulting intracellular production of cytokines was analyzed by immunostaining of permeabilized cells followed by flow cytometry analysis. These assays showed that exposure to the Daudi cells induced both the IL-33 γδ-pcl and IL-2 γδ-pcl to express mainly Th1 cytokines, little IL-10 and IL-4, and no IL-17 (Fig. 2, Supplemental Fig. 1A). Typically, Daudi cells induced the production of intracellular IFN-γ in the same proportion of Vγ9 T cells from the IL-2 γδ-pcl as the IL-33 γδ-pcl, whatever the dose of IL-33 (Fig. 2C, Supplemental Fig. 1B). In these experiments, however, the proportion of Vγ9 T cells producing TNF-α was significantly higher in IL-33 γδ-pcl produced with 100 or 500 ng/ml IL-33 than in IL-2 γδ-pcl (Fig. 2C, Supplemental Fig. 1C).

The intracellular mediators of Vγ9 T cell cytotoxicity were also analyzed as above, and both IL-33 γδ-pcl and IL-2 γδ-pcl were found to have a high content of cytosolic perforin and granzyme B, without any significant difference between them (Fig. 2D). Their cytolytic activities against the Daudi lymphoma cell line were then investigated by monitoring the Vγ9 T cell surface expression of CD107a and measuring the lysis of Daudi cells with DAPI staining and ⁵¹Cr release. CD107a phenotyping revealed that contact with Daudi cells induced the same extent of cytolytic degranulation in Vγ9 T cells from both the IL-2 γδ-pcl and IL-33 γδ-pcl (Fig. 2E). Also, both DAPI staining and ⁵¹Cr release showed that Daudi cells were killed as efficiently by IL-2 γδ-pcl as IL-33 γδ-pcl (Fig. 2F, 2G).

Because Vγ9 T cells may also exert their cytolytic functions through cell surface expression of the TNF family death receptor ligands TRAIL and FasL, we asked whether these mechanisms were also involved in the Daudi lysis induced by the γδ cell lines. Therefore, we tested whether the blocking of FasL, TRAIL, or both by blocking Abs could modulate this cytotoxic activity. We found that none of these Abs either alone or together decreased the cytotoxic activity of IL-2 γδ-pcl or IL-33 γδ-pcl (Fig. 2F). Thus, IL-2 γδ-pcl and IL-33 γδ-pcl have the same potent cytolytic activity for Daudi cells mediated by granzyme B and perforin release and the same weak cytolitic activity against Raji, Ramos, RL, and NCEB1 cell lines (Supplemental Fig. 2). Finally, we tested other cancer cell lines and showed that IL-33 γδ-pcl is able to kill RPMI8226, a myeloma cell line known to be killed by Vγ9 T cells, whereas OVCAR and MIA PACA2 (ovarian and pancreatic cancer cell lines) were not sensitive to Vγ9 T cells (Fig. 2H).

Altogether, these functional assays demonstrate that Vγ9 T cells amplified by a combination of BrHPP and IL-33 display the same phenotype and functions as did those obtained with BrHPP and IL-2.

To determine the mechanisms underlying the Vγ9 T cell response of PBMC to IL-33, we first analyzed these lymphocytes for their cell surface expression of the IL-33 receptor ST2. Unexpectedly, however, these cells did not express ST2 at all on their surface, whether in the resting state or once stimulated with BrHPP and IL-33 (Fig. 3A). This defect was confirmed by the absence of significant mRNA expression for the ST2 gene IL1RL1 in resting and BrHPP-stimulated cells (Fig. 3B). This puzzling defect suggested that the Vγ9 T cell response of PBMC to IL-33 observed in the above experiments might not involve the γδ T cells alone. To test this hypothesis, γδ T cells were magnetically purified from freshly isolated PBMC, stained with CFSE, and cultured for 6 d in the presence of BrHPP and IL-33 prior to analyzing their proliferation by flow cytometry measurements of the CFSE dilution in gated Vγ9 T cells, as in Fig. 1A. Under these conditions, there was no CFSE dilution from the purified γδ cells when cultured with BrHPP plus IL-33, whereas they did proliferate with BrHPP plus IL-2. Indeed, this was in striking contrast with their strong proliferation to BrHPP plus IL-33 when testing with whole PBMC rather than the purified cells alone (Fig. 3C).

These results show that the Vγ9 T lymphocytes are not cell autonomous in the proliferative response observed in PBMC exposed to BrHPP plus IL-33, in line with their intrinsic lack of IL-33 receptor expression.

We next sought to identify the undefined bystander cell subset that is required for the Vγ9 T cell response in PBMC exposed to BrHPP and IL-33. Although they usually constitute a minority subset of PBMC, ILC2s represent very relevant candidates because they are paramount targets for IL-33 (16, 26, 29–31). Based on their recently described phenotype (26) (Fig. 4A), we separately isolated ILC2s (Lin⁻CD127⁺CD117^+/−CRTH2⁺CD62L⁺CD161⁺CCR6⁺CD7⁺) and Vγ9 T cells from PBMC using fluorescent cell sorting. Vγ9 T cells were stained with CFSE and then these two highly purified subsets were cocultured for 6 d with BrHPP plus either IL-2 or IL-33 prior to analysis of CFSE dilution. Results indicated that the purified Vγ9 T cells did not proliferate in response to BrHPP plus IL-33 when cocultured with ILC2s (Fig. 4B).

Therefore, to identify the PBMC subset allowing the IL-33–induced Vγ9 T cell proliferation, we performed a series of PBMC depletion experiments (Fig. 5A). Namely, we tested CFSE dilutions from parallel PBMC cultures depleted of either CD4⁺, CD8⁺, CD56⁺, CD19⁺, or CD14⁺ cell subsets, and different dendritic cell subsets, including CD1c⁺, CD141⁺, or CD304⁺ cells. These assays showed that depletion of monocytes (CD14⁺ cells) from PBMC did not reduce the IL-33–induced Vγ9 T cell proliferation. The same observations were drawn from depleting separately dendritic cells, NK cells, B cells, and CD8 T cells from PBMC (Fig. 5A). In contrast, depletion of CD4 T cells fully abrogated IL-33–induced Vγ9 T cell proliferation (Fig. 5A, black bars). This result revealed a role for CD4 T cells in Vγ9 T cell proliferation in PBMC exposed to BrHPP plus IL-33 whereas depletion of CD4 T cells in BrHPP/IL-2 conditions had no effect (gray bars). This result was formally validated by observing the proliferation of purified Vγ9 T cells in response to BrHPP plus IL-33 when cocultured with purified CD4 T cells (Fig. 5B).

Taken together, these findings established that BrHPP/IL-33–induced proliferation of Vγ9 T cell requires CD4 T cells.

We next investigated the molecular determinants of the above CD4 T cell bystander activity on γδ T cell proliferation. Because IL-2 is the primary T cell growth factor, we tested whether IL-2 was involved in the bystander effect of CD4 T cells. The above Vγ9 T cell proliferation assay in PBMC induced by either BrHPP/IL-2 or BrHPP/IL-33 was measured in the presence of an IL-2–neutralizing or isotype-matched control Ab. This showed that IL-2 was involved in Vγ9 T cell proliferation induced by both conditions (Fig. 6A). Furthermore, blockade of the IL-2Rα-chain (CD25) by the neutralizing Ab confirmed the involvement of the IL-2/CD25 axis in Vγ9 T cell proliferation (Fig. 6B). Accordingly, IL-33 treatment led to the upregulation of the CD25 expression at the cell surface of Vγ9 T lymphocytes but not on CD4 T cells (Fig. 6C). To determine the partner that produced IL-2, we compared by quantitative PCR expression of IL-2 mRNA by CD4 T cells and Vγ9 T from PBMC cultured 48 h with or without the combination of IL-33 with BrHPP. We found that the two types of cells, but preferentially CD4 T cells, upregulated the expression of IL-2 mRNA when activated with IL-33 and BrHPP regarding the fold change between untreated and treated cells (Fig. 6D). However, analysis of PBMC culture supernatants did not show increased concentrations of IL-2 when cultured with IL-33 and BrHPP (Supplemental Fig. 3).

These findings suggested a bystander activity involving direct interactions between CD4 and Vγ9 T cells. To determine whether contact between the Vγ9 T and CD4 T cells was required, the previous PBMC proliferation assay was adapted to use a Transwell insert in the coculture wells. PBMC depleted of CD4 cells and stained with CFSE were deposited in the bottom chamber whereas purified CD4 T cells were placed in the upper chamber of the coculture well. After 6 d of culture in the presence of BrHPP and IL-33 or IL-2, CFSE dilution revealed that cell separation by the Transwell insert abrogated the Vγ9 T cell proliferation induced by IL-33, whereas it did not inhibit the proliferation induced by soluble IL-2 (Fig. 6E). Surprisingly, the weak proliferation of Vγ9 T cells when PBMC were cultivated with BrHPP alone was also totally inhibited when CD4 T cells were separated by the Transwell system.

Hence, contact between CD4 T cells and Vγ9 T cells when cultivated with IL-33 and BrHPP led to expression of IL-2 consumed by Vγ9 T cells to proliferate.

IL-33 has emerged as a pleiotropic immune regulator, implicated in the Th2, T regulatory, and Th1-type immune responses (41). However, despite its multiple functions on CD4 T cells, CD8 T cells, and NK cells, the bioactivity of IL-33 has never been studied on γδ T cells and particularly on human Vγ9 T cells, a T cell subset that fulfills important antitumoral functions.

In this study, we have shown that the in vitro culture of human PBMC with IL-33 leads to the proliferation of PAg-activated Vγ9 T cells, allowing their amplification as well as their complete spectrum of functions. This surprising response from T cells, which in fact lack the IL-33 receptor ST2, actually relies on a bystander population of CD4 T lymphocytes. However, ILC2s, well known as high-affinity targets of IL-33, were not implicated in this process because cocultures of pure Vγ9 T cells with pure CD4 T cells were highly reactive to the BrHPP/IL-33 combination. CD4 T cells have already been described as bystander helper cells for the proliferation of Vγ9 T cells activated by Mycobacterium tuberculosis (42). These authors showed that Vγ9 T cell proliferation was induced by IL-2 produced in the medium by CD4 T cells upon contact with APCs loaded with M. tuberculosis. IL-2 production in these conditions was described as dependent on interaction through costimulatory molecules (43, 44). In our experiments, CD4 T cell contact with Vγ9 T cells was required for bioactivity of BrHPP alone or in combination with IL-33. Moreover, the IL-2/CD25 axis was shown as essential for the bioactivity of the combination BrHPP/IL-33 on Vγ9 T cells. Thus, in our model Vγ9 T cells activated by BrHPP could act as an APC activating CD4 T cells to produce IL-2, which is amplified by IL-33 as shown by the quantitative PCR test. Indeed, to our knowledge we showed for the first time that proliferation of Vγ9 T cells can be activated by a PAg without exogenous IL-2 but owing to a contact with CD4 T cells. Furthermore, because full-blown TCR/CD28 coactivation of Vγ9 T cells can induce autocrine IL-2 signaling to promote their survival and proliferation (45, 46), the CD4 T cell contact might provide γδ lymphocytes with costimulatory B7/CD28 (47), CD40/CD40L (48, 49), or related interactions that are able to trigger their autocrine IL-2/CD25 axis. Regarding the upregulation of IL-2 mRNA in CD4 T cells and Vγ9 T cells after incubation with IL-33, we can conclude that IL-33 must enable the contact between these two cell types, amplifying the IL-2 production. However, IL-33 did not appear to induce IL-2 release in the medium. With Vγ9 T cells being in contact with CD4 T cells, they must capture IL-2 freshly produced by themselves and CD4 T cells before its release in the medium. Upregulation of CD25 at the surface of Vγ9 T cells in the condition BrHPP/IL-33, which can be due to the costimulation (50), could favor this rapid capture avoiding excess of IL-2 in the environment, which is deleterious in the case of immunotherapies.

Moreover, our study shows that besides promoting proliferation, IL-33 is also able to induce the phenotypic maturation into memory Vγ9 T cells, which are able to express only Th1-type cytokines and that display no type of functional polarization other than cytotoxicity. Such a functionalization has already shown its virtues for cancer immunotherapies. It typically occurs in the context of γδ T cell stimulation in the presence of an exogenous supply of IL-2, and in our study might to some extent reflect the impact of the IL-2/CD25 axis discussed above. The sole difference was the superiority of IL-33 over IL-2 in inducing TNF-α expression from Vγ9 T cells. For therapeutic purposes, the consequences of this superiority are two-faced, however. On the one hand, a higher secretion of TNF-α is interesting for γδ cell–based immunotherapies of highly aggressive cancers that are sensitive to necroptosis, such as glioblastoma (51), anaplastic thyroid and adrenocortical carcinomas (52), among others. On the other hand, however, TNF-α contributes physiologically to many well-known pathological processes, rendering its overproduction potentially deleterious for patients.

Our study has cast the scene for future preclinical investigations of the hitherto unsuspected potential of IL-33/γδ T cell–based cancer therapeutics as alternatives to their IL-2 counterparts. Indeed, the abundant expression of IL-33 by endothelial cells from various human tumor tissues (15) reminds us that in patients this cytokine already affects immune infiltrates of tumors, for example by promoting their antimetastatic activity (53).

M.P., J.-J. F., C.D., J.-P.G., and C.C. are inventers on a patent for the method for inducing Il-2–free proliferation of γδ T cells (Publication No. WO2014072446-A1). The other authors have no financial conflicts of interest.