Cytokines enhance human Vγ9Vδ2 T‐cell TCR‐dependent and TCR‐independent effector functions

Vγ9Vδ2 T cells can recognize various molecules associated with cellular stress or transformation, providing a unique avenue for the treatment of cancers or infectious diseases. Nonetheless, Vγ9Vδ2 T‐cell‐based immunotherapies frequently achieve suboptimal efficacies in vivo. Enhancing the cytotoxic effector function of Vγ9Vδ2 T cells is one potential avenue through which the immunotherapeutic potential of this subset may be improved. We compared the use of four pro‐inflammatory cytokines on the effector phenotype and functions of in vitro expanded Vγ9Vδ2 T cells, and demonstrated TCR‐independent cytotoxicity mediated through CD26, CD16, and NKG2D, which could be further enhanced by IL‐23, IL‐18, and IL‐15 stimulation throughout expansion. This work defines promising culture conditions that could improve Vγ9Vδ2 T‐cell‐based immunotherapies and furthers our understanding of how this subset might recognize and target transformed or infected cells.


Introduction
γδ T cells expressing the Vγ9Vδ2 T-cell receptor (TCR) account for up to 10% of the total circulating lymphocyte population in healthy human adults. Vγ9Vδ2 T cells (herein referred to as Vδ2 T cells) display unique potential for use in "off the shelf" adoptive cell therapeutics due to their potent cytotoxic effector functions, Vδ2 T cells mediate effector functions by releasing preformed cytotoxic granules or secretion of pro-inflammatory cytokines such as IFN-γ and TNF-α. These effector functions are typically triggered directly upon TCR recognition of phosphoantigens (PAgs), which are intermediates of isoprenoid biosynthesis [10], but may also occur through TCR-independent mechanisms via cytotoxic surface receptors such as NKG2D, DNAM-1, or the FсγRIIIa receptor CD16 [11]. Although many of these cytotoxic surface receptors have been well defined in the context of NK cells, their expression and functional capacity on Vδ2 T cells is relatively understudied [12].
Vδ2 T cells demonstrate a remarkable sensitivity to stimulation by pro-inflammatory cytokines. We and others have previously defined a spectrum of Vδ2 T-cell phenotypes in human adults, with varying cytokine responsiveness and ex vivo cytotoxic capacity [13][14][15]. Notably, CD26 hi CD94 lo Vδ2 T cells exhibit low ex vivo expression of cytolytic molecules, but high responsiveness to cytokine stimulation [13]. Activation of these Vδ2 T cells by a combination of pAg, IL-2, and IL-23 induced transcriptional changes associated with a cytotoxic phenotype, but the cytolytic capacity of these cells has not been assessed. Other studies have explored the capacity of either IL-15 or IL-18 plus antigen stimulation of Vδ2 T cells to enhance effector phenotypes and/or TCR-dependent cytotoxic functions [16][17][18][19][20][21][22][23][24], but the relative potency of different cytokine combinations in altering Vδ2 function is unclear.
To date, we lack a comprehensive understanding of the impact of cytokine stimulation and surface receptor triggering on Vδ2 T cell-mediated cytotoxicity. Therefore, we studied how combinations of cytokines and antigen-based expansion could be harnessed to influence Vδ2 T-cell target recognition and killing through various mechanisms. We compared the effects of IL-23, IL-15, and IL-18 stimulation on effector phenotypes and TCR-dependent or TCR-independent cytotoxic functions of Vδ2 T cells expanded in vitro. Our findings identify distinct activation and functional profiles of Vδ2 T cells driven by various proinflammatory cytokine stimulation conditions and provide strong evidence to include these cytokines in Vδ2 T-cell expansion protocols for future therapeutics to ensure more effective elimination of transformed or infected cells.

Cytokine-mediated activation of human Vδ2 T cells
As Vδ2 T cells are highly responsive to various cytokines, we aimed to characterize their relative sensitivity to stimulation with IL-18, IL-15, IL-23, IL-23 + IL-7, or TGF-β. Live CD3 + Vδ2 + lymphocytes within whole peripheral blood mononuclear cells (PBMCs) cultures were analyzed by flow cytometry after 16, 48, or 72 h of in vitro stimulation (Supporting information Fig. S1). IL-15 elicited the greatest levels of activation, with significant increases in median CD25 (12.3-fold at 16 h, 10.3-fold at 48 h, and 20.6-fold at 72 h) and CD69 expression (4.9-fold increase at 16 h, 14.0-fold at 48 h, and 11.1-fold at 72 h) across all time points when compared to an unstimulated control ( Fig. 1a and  b). Stimulation with IL-18 resulted in a modest activation of the Vδ2 subset, with median CD25 and CD69 expression upregulated across all time points compared to the unstimulated control; however, for CD25 expression, significance was only reached at the 16 h time point (CD25: sixfold increase at 16 h, 2.5-fold increase at 48 h, 2.9-fold increase at 72 h; CD69: fourfold increase at 16 h, 6.9-fold increase at 48 h, and 4.9-fold increase at 72 h).
In contrast, we found no evidence of activation following TGFβ or IL-23 stimulation ( Fig. 1a and b). Consistent with previous data, IL-7 co-stimulation was required for IL-23 to activate Vδ2 cells in vitro [13]. Upregulation of CD25 following IL-23 + IL-7 stimulation was noted across all time points in comparison to the unstimulated control (9.1-fold increase 16 h, 5.9-fold increase 48 h, 10.8-fold increase 72 h), and CD69 expression was significantly increased after stimulation for 48 or 72 h when compared to the control (4.8-fold increase at 48 h and 4.2-fold increase at 72 h).
We further assessed whether cytokine-mediated activation of Vδ2 T cells was sufficient to upregulate intracellular granzyme B (GzmB) expression. Again, we found IL-15 to have the greatest effect, with significant increases in the mean fluorescence intensity (MFI) of GzmB across all time points compared to the unstimulated control (1.8-fold at 16 h, 3.1-fold at 48 h, and 3.1-fold at 72 h) (Fig. 1c). Upregulation of GzmB was also driven by IL-23 + IL-7 stimulation at all time points compared to the unstimulated controls (1.3-fold at 16 h, 1.9-fold at 48 h, and 2.1-fold at 72 h), and both IL-18 and IL-23 stimulation amounted in small yet significant increases in GzmB levels at the 16-h time point (1.1-fold increase for both IL-18 and IL-23 stimulation) compared to the unstimulated controls.
Our group has previously identified a highly IL-23 responsive subset of Vδ2 T cells by expression of CD26 [13]. While we previously found that IL-23 responsiveness primarily occurred in combination with pAg or IL-7 co-stimulation, we wanted to confirm that the low IL-23 responsiveness in these participants was not due to a low abundance of CD26 + Vδ2 T cells. Importantly, over 50% of the total Vδ2 T cell population was CD26 + in seven of the 12 donors (Supporting information Fig. S2a). We further assessed activation of either the CD26 + or CD26 − Vδ2 T cell compartments after 48 h of stimulation with either IL-23 or IL-23 + IL-7, observing a significant increase in expression of both CD25 and CD69 in the CD26 + subset in comparison to the CD26 − subset (Supporting information Fig. S2b). These data further confirm that IL-23 responsiveness is primarily restricted to CD26 + Vδ2 T cells.
To confirm whether cytokine stimulation was acting directly on Vδ2 T cells, or indirectly through activation of other cell types within the PBMC cultures, we enriched the total γδ T-cell population through negative isolation. We found similar trends for Vδ2 T cells within both the PBMC and γδ T-cell cultures, with IL-15 and IL-23 + IL-7 inducing significant upregulation of CD25 and CD69 (Supporting information Fig. S3). Thus, Vδ2 T cells demonstrate varying responsiveness to direct stimulation with these proinflammatory cytokines.

Modulation of antigen-induced Vδ2 T-cell proliferation and phenotype by cytokines
γδ T-cell-based immunotherapies rely on producing large numbers of cells generated by in vitro or in vivo expansion, with common protocols using aminobisphosphonate drugs and IL-2 to drive rapid Vδ2 T-cell proliferation. Therefore, we evaluated cytokine treatment's impact during in vitro Vδ2 expansion facilitated by zoledronate and IL-2. We first evaluated the proliferation of Vδ2 T cells for 20 days within cultures stimulated with zoledronate + IL-2 with or without additional stimulation with IL-18, IL-15, IL-23, or IL-23 + IL-7. Starting frequencies of Vδ2 T cells were within the normal range for all donors selected (1.14%-11.9% of the total CD3 + population). We found additional cytokine treatment had no effect on the proliferation of Vδ2 T cells in comparison to standard expansion with zoledronate and IL-2 alone ( Fig. 2a and b).
We next examined the phenotype of Vδ2 T cells after 14 days of expansion under various conditions (Supporting information Fig.  S4). Despite the robust activation of Vδ2 T cells by IL-15, GzmB expression following in vitro expansion was relatively unaffected by cytokine co-stimulation. Indeed, only the addition of IL-23 to the standard expansion protocol resulted in a significant increase in GzmB MFI (twofold increase compared to standard expansion) (Fig. 3a).
Nonetheless, GzmB expression is only one potential Vδ2 Tcell cytotoxicity mediator. CD56, which is also associated with cytolytic effector functions of γδ T cells [25,26], was also significantly upregulated on Vδ2 T cells upon expansion with IL-15, IL-23, and IL-23 + IL-7 (1.5-fold, 1.3-fold, and 1.2-fold increase, respectively, compared to standard expansion) (Fig. 3b). Similarly, heightened NKG2A/CD94 expression has recently been implicated in Vδ2 T-cell cytotoxic capacity [15] IL-23 or IL-23 + IL-7 plus antigen stimulation drove significantly greater expression of CD94 (1.2-fold increase vs. standard expansion for both conditions), and a small yet nonsignificant increase was noted for the IL-18 plus antigen stimulation condition (1.1-fold compared to standard expansion) (Fig. 3c). CD27 − effector memory Vδ2 T cells often express heightened levels of GzmB and CD16, and are associated with heightened cytotoxicity [14,27]. Here, we found that both IL-18 and IL-15 stimulation decreased expression of CD27 (1.2-fold and 3.3-fold vs. standard expansion, respectively) (Fig. 3d), indicating that such stimulation during antigeninduced proliferation could drive Vδ2 T cells toward an effector memory phenotype, potentially conferring heightened effector capacity. We found no evidence of change in the expression of the exhaustion marker PD-1 under any expansion condition (Supporting information Fig. S5a). Finally, CD26 was upregulated during expansion with IL-15 and IL-23 + IL-7 (Fig. 3e).
As Vδ2 T cells are capable of TCR-independent cytotoxicity, we also assessed the expression of a selection of cytotoxic surface receptors. The presence of CD16 on expanded Vδ2 T cells remained low and was not upregulated upon additional cytokine  Fig. S5b). We observed consistently high frequencies of 2B4, NKG2D, and DNAM-1 expressing Vδ2 + T cells after expansion both with and without additional cytokine stimulation, with only a slight decrease in DNAM-1 observed upon IL-18 mediated expansion (Supporting information Fig. S5c-e). Upon further assessment of the density of these markers via MFI, we could observe more tangible changes in expression ( Fig. 3f-h). The MFI of 2B4 on 2B4 + Vδ2 + T cells exhibited a small but nonsignificant increase for the IL-15 and IL-23 + IL-7 expansion conditions (1.3-fold and 1.1-fold, respectively, compared to standard expansion) (Fig. 3f). The IL-15, IL-18, and IL-23 + IL-7 expansion conditions resulted in an increase in NKG2D MFI that did not reach significance (medians 1.2-fold for IL-15, and 1.1-fold for IL-18 and IL-23 + IL-7 compared to standard expansion) (Fig. 3g). Finally, MFI of DNAM-1 on DNAM-1 + Vδ2 + T cells significantly increased upon IL-23 + IL-7 stimulation, and nonsignificantly increased upon IL-23 stimulation (both showing a 1.2-fold increase in MFI compared to standard expansion) (Fig. 3h).

Enhancement of target cell lysis by cytokine treated Vδ2 T cells
Having established that cytokine plus antigen stimulation of Vδ2 T cells modulates the expression of key surface receptors and cytolytic mediators, we next explored the functional capacity of Vδ2 T cells under these conditions. Utilizing a series of lactate dehydrogenase (LDH) cytotoxicity assays, we assessed Vδ2 T cellmediated cytotoxicity of two tumor cell lines (Raji and K562), which are recognized by Vδ2 T cells to varying extents [28].
We first evaluated the capacity of IL-2 and pAg expanded Vδ2 T cells to lyse Raji target cells, which had either been left untouched or sensitized with zoledronate to induce pAg expression for Vδ2 recognition [29]. Across several effector:target (E:T) cell ratios (5:1, 2:1, 1:1, 0.5:1), we observed consistent increases in cytotoxicity against pAg-sensitized Raji target cells compared to the unprimed target cells (Supporting information Fig. S6a). Preincubation of expanded Vδ2 T cells with an inhibitory anti-BTN3A1 antibody (clone 103.2 [30]) led to a reduction in cytotoxicity of pAg primed Raji target cells at an E:T ratio of 5:1 in comparison to preincubation with isotype control (median 3.51% decrease), indicating some involvement of BTN3A1 in Vδ2 T-cell-mediated lysis of this cell line (Supporting information Fig. S6b).
Using this model for measuring direct Vδ2 T-cell-mediated lysis of zoledronate primed Raji target cells, we compared the impact of cytokine co-stimulation on the cytotoxic capacity of Vδ2 T cells. We found that IL-18, IL-15, and IL-23 stimulation all significantly increased the cytotoxic capacity of Vδ2 T cells at E:T ratios of 5:1 and 2:1 (IL-18; threefold increase at 5:1 and 2.7-fold increase at 2:1, IL-15 2.7-fold increase at 5:1 and 2.5-fold increase at 2:1, IL-23 2.0-fold increase at 5:1 and 1.5-fold increase at 2:1) in comparison to Vδ2 T cells expanded with IL-2 plus antigen alone. At E:T cell ratios of 1:1, an increase in lysis was also noticed for the IL-18 and IL-15 expanded Vδ2 T cells, but this only reached significance in the IL-23 expansion condition (Fig. 4). IL-23 + IL-7 plus antigen stimulation did not increase the cytotoxic capacity of Vδ2 T cells in this model (Supporting information Fig. S8a and b). Interestingly, we also observed similar trends in cytokine expanded Vδ2 T-cell-mediated cytotoxicity for Raji target cells which had not been primed with zoledronate for pAg expression (Supporting information Fig. S7a-c).
We next validated these findings in the K562 cell line, which is targeted by Vδ2 T cells for killing to a lesser extent than Raji cells [28,31]. Inhibition of BTN3A1 on Vδ2 T cells expanded with zoledronate + IL-2 alone at an E:T ratio of 5:1 did not inhibit lysis, indicative of a TCR-independent mechanism of recognition of K562 cells (Supporting information Fig. S6b). We compared the lysis of K562 cells by cytokine-expanded Vδ2 T cells at E:T ratios of 5:1, 2:1, 1:1, and 0.5:1 (Fig. 4d-f). Here, we observed lower levels of target cell lysis in comparison to the pAg-primed Raji target cells. IL-18-mediated expansion led to a significant increase in lysis of K562 cells at E:T ratios of 5:1, 2:1 and 1:1 in comparison to standard expansion (1.5-fold, 1.8-fold, and 2.1-fold increase, respectively) (Fig. 4d). IL-23 stimulation throughout expansion led to small but nonsignificant increases in cytotoxicity at E:T ratios of 5:1, 2:1, and 1:1 (1.2-fold, 1.5-fold, 2.2-fold increase, respectively, compared to standard expansion) (Fig. 4f). IL-15mediated expansion did not enhance the lysis of the K562 cell line in comparison to the standard expansion of zoledronate + IL-2 alone (Fig. 4e).

Identification of surface receptors mediating Vδ2 T-cell effector functions
The contribution of either individual or synergistic engagement of cytotoxic surface receptors on Vδ2 T-cell-mediated killing has been relatively understudied. We aimed to characterize the effector functions of surface receptors on expanded Vδ2 T cells and determine the impact cytokine stimulation has on these functions using a series of redirected cytotoxicity assays, in which crosslinking anti-receptor antibodies induce receptor-specific lysis of the Fс receptor expressing P815 murine cell line (Fig. 5a). These redirected lysis assays provide precise mechanistic data that cannot be derived from contexts wherein multiple unknown receptors can be simultaneously engaged. Using this experimental set-up, we aimed to identify receptors whose expression should be evaluated in more physiologically relevant models in the context of individual diseases or tumors.
Initially, we measured cytotoxicity through individual surface receptors after in vitro expansion of Vδ2 T cells with pAg and IL-2 alone (Fig. 5b). Unsurprisingly, we found that lysis mediated by CD3 activation occurred at the greatest levels, with a median of 51.8% target cell lysis. Baseline levels of CD3-mediated cytotoxicity by expanded Vδ2 T cells was highly variable between donors, ranging between 27.2% and 96.2%. NKG2D engagement induced the second greatest levels of lysis, with a median of 21.4% cytotoxicity, and was also highly variable, ranging between 13.8% and 79.3%. CD16 signaling drove a median of 21.4% cytotoxicity, ranging from 3.6% to 61.8%. Most interestingly, we observed that engagement of CD26 through the anti-CD26 antibody (clone 5F8 [32]) induced a moderate level of cytotoxicity on Vδ2 T cells in relation to the isotype control (median 16.7% cytotoxicity vs 2.7%, respectively), ranging from 9.6% to as much as 36.8%. Engagement through the activating anti-CD26 clone 1F7 also drove small levels of target cell lysis (median 11.0%, ranging between 4.6% and 21.2%). Engagement of DNAM-1 or 2B4 did not yield significant lysis of target cells compared to the isotype control antibody, indicating that these surface receptors may not function in isolation to mediate direct cytotoxicity.
As DNAM-1 and 2B4 mediated insufficient levels of lysis in our assays, we were subsequently interested in whether cytotoxic surface receptors could act synergistically with other receptors to enhance effector functions in a co-stimulatory manner, similar to their activity on NK cells [33][34][35][36][37]. To explore this on Vδ2 T cells, we stimulated receptors in a pairwise fashion in our redirected lysis assay. Here, we found that both CD26 (clone 5F8) and NKG2D could aid in CD3-mediated cytotoxicity, driving significant increases in redirected lysis compared to CD3 plus an isotype control alone (1.1-fold increase for both CD26 and NKG2D) (Fig. 5c  and d).

Discussion
Herein, we identify approaches to enhance Vδ2 T cell effector functions and characterize the impact of cross-linking cytotoxic surface receptors on Vδ2 T-cell-mediated elimination of target cells. We explored the diversity in activation and functional profiles of Vδ2 T cells upon alternate cytokine stimulation in vitro either with or without additional antigen-based stimulation, with the goal of identifying candidates to improve upon conventional expansion protocols for Vδ2 T cell-based immunotherapies, which have had suboptimal clinical outcomes.
Vδ2 T cell responsiveness to cytokine-mediated stimulation was highly variable, with IL-23/IL-7 driving extremely high activation levels and increases in GzmB. Despite being highly activating as a direct stimulant and slightly enhancing effector phenotype upon pAg-mediated expansion, IL-23/IL-7 plus antigen stimulation failed to improve the cytotoxic capacity of Vδ2 T cells. In contrast, IL-15 stimulation of Vδ2 T cells was not only highly activating but could also drive enhanced effector phenotypes and increased lysis of Raji target cells when provided alongside antigen stimulation. IL-18 or IL-23 both drove little to no direct activation, however when pAg stimulation was added, these cytokines resulted in an enhanced cytotoxic effector phenotype, heightened lysis of Raji and K562 target cell lines, and increased levels of TCR-independent activation in redirected lysis assays.
While our work confirms and expands most of the findings of a similar study by Aehnlich et al. [18], some slight differences in observed phenotypes after in vitro expansion with IL-15 stimulation in our results are most likely due to altered timing and concentration of cytokine stimulations, and a shorter period of expansion. Furthermore, variations in CD56, CD16, NKG2D, and DNAM-1 expression after in vitro expansion with IL-18 in our findings and those of Murday et al. [20] could potentially be explained by variations in culturing procedures, with the latter study stimulating directly with the pAg isopentenyl pyrophosphate.
Although we could identify clear trends in cytokine responsiveness, we saw a high degree of inter-donor heterogeneity in our results. Likely, proportions of Vδ2 T cell subsets between donors (as defined by Wragg et al.) are the main driver of this variation, and we speculate that increased pre-expansion levels of the MAIT-like CD26 hi CD94 lo subset, which express greater levels of chemokine and cytokine receptors such as IL-18Rα and IL-23R, confer heightened activation levels in comparison to the CD26 − NK cell-like Vδ2 T cell subsets [13]. Indeed, we observed significant differences in activation between the CD26 + and CD26 − Vδ2 T cell compartments upon either IL-23 or IL-23/IL7 stimulation, highlighting the importance of CD26 expression for such cytokine responsiveness. Further investigations into Vδ2 T-cell CD26/CD94 subset-specific cytokine and antigen responsiveness would be beneficial to the field, as clarifying these differences may aid in future disease-tailored Vδ2 T-cell-based immunotherapies.
The presence of several cytotoxic surface receptors on Vδ2 T cells allows for wide-ranging TCR-independent effector functions, conferring an increased potential for clinical applications. The contribution of these surface receptors to Vδ2 T-cell-mediated cytotoxicity has been relatively unexplored. We highlight that surface receptors such as NKG2D, CD16, and even CD26 can facilitate substantial amounts of target cell lysis, although CD3-dependent cytotoxicity occurs most efficiently. In contrast, 2B4 and DNAM-1 mediated insufficient levels of cytotoxic effector functions when cross-linked alone, causing us to investigate whether these surface receptors instead perform co-stimulatory roles. Multiple NK cell and Vδ2 T-cell focused studies have evaluated whether parallel activation through surface receptors might heighten the target cell lysis [33][34][35][36][37][38][39][40]. Here, we extend such investigations to concurrently assess pairwise co-stimulatory relationships between CD3, CD26, NKG2D, 2B4, and DNAM-1 on in vitro expanded Vδ2 T cells. CD3 activation could be enhanced upon additional signaling through CD26 and NKG2D. NKG2D-mediated cytotoxicity was complemented by concurrent CD26, CD16, and 2B4 signaling. Additionally, CD16 and CD26 were found to act in synergy, similar to Madueño et al., who found CD26 + NK cells mediated greater levels of CD16-dependent lysis than CD26 − NK cells, indicating a co-operative role for these receptors [41]. We found little evidence that 2B4 and DNAM-1 were contributing co-stimulatory signals to enhance cytotoxic functions mediated through other surface receptors. Considering both the insufficient baseline and costimulatory effector functions of both 2B4 and DNAM-1, the constitutive expression of these surface receptors on expanded Vδ2 T cells prompts further investigation into their role in this subset.
Given that stimulation with IL-15, IL-18, or IL-23 during antigen-induced proliferation heightened the cytotoxic capacity of Vδ2 T cells against Raji target cells, whereas only the IL-18 and IL-23 expansion conditions increased lysis of K562 cells, we looked to whether such activation enhanced receptor-mediated lysis. In redirected lysis cytotoxicity assays, providing IL-18 and IL-23 stimulation throughout expansion led to increased lysis through multiple surface receptors. In contrast, we did not observe a significant increase in receptor-mediated cytotoxicity in the IL-15 expansion condition.
One major caveat to this study is that most donors included had low CD16 + Vδ2 T cell frequencies ex vivo, which could not be increased upon antigen-mediated activation or additional cytokine stimulation. Although we were able to identify reasonable degrees of cytotoxic effector and co-stimulatory functions mediated by CD16, we hypothesize that the addition of donors with heightened proportions of CD16 + Vδ2 T cell subsets would reveal a more substantial role for CD16.
Adoptive transfer of expanded Vδ2 T cells for the treatment of cancers has delivered some promising results within clinical trials [4][5][6][7][8][9]. However, there is considerable speculation that the cytotoxic capacity of the Vδ2 T cell subset will need to be improved to obtain optimal results [7]. This study identifies that stimulation with either IL-18, IL-15, or IL-23 during antigen-mediated proliferation of Vδ2 T cells in vitro drives heightened effector phenotypes and improves cytotoxic capacity, often through a range of surface receptors. We offer strong evidence in favor of the inclusion of these pro-inflammatory cytokines in Vδ2 T cell expansion protocols for future therapeutics to ensure more effective elimination of transformed or infected cells. By exploring the synergistic capacity of surface receptors on Vδ2 T cells, we propose that synchronized engagement of multiple receptors by their ligands may allow for the most efficient killing, and that cytokine stimulation during pAg-induced proliferation may heighten cytotoxicity through one or more of these receptors. The results obtained here provide new knowledge into the mechanisms underlying the cytotoxic capacity of Vδ2 T cells, which may be exploited to optimize future immunotherapies for improved clinical outcomes.

Isolation of PBMC from whole blood
Fresh whole blood samples were collected from healthy human donors recruited at either the University of Melbourne (n = 21, male and female) (University of Melbourne Ethics Review Committee #11395), or through the Australian Red Cross Service (n = 5, no demographic information available). PBMC were isolated from whole blood using Ficoll-Paque gradient density centrifugation (Cytiva/GE Healthcare). Isolated PBMC were resuspended in RPMI 1640 medium (no glutamine) (Gibco/Life Technologies) supplemented with 10% fetal calf serum (FCS) (Sigma-Aldrich) and penicillin/streptomycin/L-glutamate (Gibco/Life Technologies) (RF10) for immediate use, or resuspended in freeze solution (90% FCS, 10% dimethyl sulfoxide [DMSO] (Sigma-Aldrich)) for cryopreservation at −80°C.

Enrichment of γδ T cells from human PBMC
Cryopreserved PBMC from donors recruited through the Australian Red Cross Service were thawed in RF10. γδ T cells were enriched from PBMC by negative selection using a Human TCRγ/δ+ T cell Isolation Kit (Miltenyi Biotec) according to the manufacturer's instructions and resuspended in RF10 for immediate use.

In vitro stimulation and flow cytometric analysis of Vδ2 T cells in whole PBMC or enriched γδ T cells
Frozen whole PBMC from donors recruited through the University of Melbourne were thawed in RF10, or γδ T cells isolated as described above were placed in 96-well plates at a concentration of 1 × 10 6 cells/mL in RF10. The cells were either left unstimulated or stimulated with either 10 ng/mL IL-15 (PeproTech), 50 ng/mL IL-18 (InvivoGen), 50 ng/mL IL-23 (R&D Systems), or 50 ng/mL IL-23 plus 50 ng/mL IL-7 (Sinobiological), or 1.5 ng/mL TGF-β (InvivoGen) and incubated at 37°C in 5% CO 2 . Concentrations for cytokine stimulations were selected in accordance with previously published work [13,20,24,[42][43][44]. After 16 h, 48 h, or 72 h of stimulation, cells were collected and washed in phosphate buffered saline (PBS) for flow cytometric analysis.

In vitro expansion of Vδ2 T cells
Freshy isolated PBMC were resuspended in RF10 at a concentration of 2 × 10 6 cells/mL and placed in 24-well plates. PBMC were supplemented with 15 μM zoledronic acid monohydrate (Sigma) and stimulated with 1 × 10 2 IU/mL IL-2 (PeproTech) alone, or a combination of 1 × 10 2 IU/mL IL-2 plus either 10 ng/mL IL-15, 50 ng/mL IL-18, 50 ng/mL IL-23, or 50 ng/mL IL-23 plus 50 ng/mL IL-7. Expansions were placed in the incubator in 5% CO 2 at 37°C. Every 2-3 days, expansions were washed, resuspended in fresh media at 2 × 10 6 cells/mL, and restimulated with cytokines. For proliferation index and curves, expansions were washed, and total numbers counted using a CELL-DYN Emerald cell counter on days 6,8,10,12,14,18, and 20 of culture. On days 12 or 13, Vδ2 T cells were collected for direct or redirect lysis cytotoxicity assays, and on day 14 cells were collected for phenotypical analysis by flow cytometry.

Direct lysis cytotoxicity assay using in vitro expanded Vδ2 T cells
On day 12 or 13, Vδ2 T cells expanded as described above were collected, washed, and resuspended in RF10. Direct lysis cytotoxicity assays were carried out using a CytoTox 96 Non-Radioactive Cytotoxicity Assay Kit (Promega) and set up according to the manufacturer's instructions. Raji (ATCC) or K562 (ATCC) cells were used as targets, and effector Vδ2 T cells expanded under differ-ent cytokine conditions were added at indicated effector:target (E:T) ratios. For blocking assays, 5 μg/mL of either the inhibitory anti-BTN3A1 antibody (clone 103.2) [30] or IgG1 κ isotype control (MOPC-21; BioLegend) was incubated with Vδ2 T cells at 5% CO 2 at 37°C for 30 min prior to use in LDH assays.

Statistics
Statistical analyses were carried out using GraphPad Prism v8. All two group comparisons were completed using nonparametric Wilcoxon matched pairs signed rank tests. Friedman paired, nonparametric multiple comparison tests were used for comparison of three or more groups. For all t-tests, p values < 0.05 were determined as significant, otherwise ns. Flow cytometry data were analyzed in FlowJo v10.2 (TreeStar).
acknowledge the Melbourne Cytometry Platform for provision of flow cytometry services. The following reagents were obtained through the NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH: Anti-Human CD26 Monoclonal (1F7) from Dr. Chikao Morimoto, Anti-Human CD26 Monoclonal (5F8) from Dr. Chikao Morimoto. We are extremely grateful to A. Kristensen, T. Amarasena, J. Nguyen, L. Burmas, and C. Batten for technical assistance. Supported by fellowship and program grants from the Australian National Health and Medical Research Council. Open access publishing facilitated by The University of Melbourne, as part of the Wiley -The University of Melbourne agreement via the Council of Australian University Librarians.

Conflict of interest:
MR and APU are inventors on a patent regarding methods of inhibiting or activating γδ T cells (WO2020257871) and on filed patents regarding BTN-mediated activation of γδ T cells. All other authors declare no commercial or financial conflict of interest. Ethics approval: All participants within this study provided written informed consent prior to the collection of samples, and all procedures were conducted with the approval of the relevant Institute's ethics committee.

Data availability statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request.