This study was carried out in accordance with the recommendations of the Research Ethics Guidelines, Health Research Ethics Board of Alberta—Cancer Committee 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 Health Research Ethics Board of Alberta—Cancer Committee.

Primary human γδTc and αβTc cultures were established and maintained as described (9). Briefly, peripheral blood mononuclear cells were isolated from healthy donor blood using density gradient separation (Lymphoprep, Stem Cell Technologies, Vancouver, BC, Canada) and cultured at 1 × 10⁶ cells/ml in RPMI complete medium containing 1 μg/ml concanavalin A (Sigma-Aldrich, Oakville, ON, Canada), 10% fetal bovine serum (FBS), 1× MEM NEAA, 10 mM HEPES, 1 mM sodium pyruvate (all Invitrogen, Burlington, ON, Canada) plus 10 ng/ml recombinant human IL-2 and IL-4 (Miltenyi Biotec, Auburn, CA, USA). Every 3–4 days, cells were counted and densities adjusted to 1 × 10⁶ cells/ml by addition of fresh medium and cytokines. Cells were maintained in a humidified atmosphere at 37°C with 5% CO₂. After 1 week, αβTc were depleted after labeling with anti-TCRαβ PE antibodies (BioLegend, San Diego, CA, USA) and anti-PE microbeads (Miltenyi Biotec), filtering through a 50 μm Cell Trics filter (Partec, Görlitz, Germany) and running through an LD depletion column (Miltenyi Biotec). The flow-through contained γδTc, which were further cultured in RPMI complete medium plus cytokines (as above) at 37°C with 5% CO₂. For some experiments, the positively selected αβTc were also recovered and maintained. For cytotoxicity experiments, γδTc cultures were used at the end of the culture period (day 19–21), as they were most differentiated and therefore most cytotoxic at this time. Blocking assays were also typically done at this time, to mimic conditions used for cytotoxicity assays. Some experiments were done at earlier time points (days 14–16), and susceptibility to mAb-induced cell death did not appear to be significantly different, although this was not tested directly. Donor cultures are identified as follows: donor number-culture number; thus, 3-2 = the second culture derived from donor 3.

Breast cancer cell lines MCF-7, T47D, and MDA-MB-231 were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and maintained as per ATCC guidelines.

Target cells were labeled with 5 μM CalAM as per the manufacturer’s instructions (Invitrogen/Thermo Fisher Scientific, Waltham, MA, USA). Cells were diluted to a density of 30,000 cells/100 μl medium for use in cytotoxicity assays.

The following antihuman anti-γδTCR mAb clones were used: B1 (BioLegend, San Diego, CA, USA), B1.1 (25) (eBioscience, San Diego, CA, USA), 5A6.E9 (26) (Thermo Fisher Scientific, Waltham, MA, USA), and 11F2 (27) (Becton Dickinson, Mississauga, ON, Canada); mouse IgG (Sigma-Aldrich, Oakville, ON, Canada) was used as a control. The anti-NKG2D antibody was purchased from BioLegend (Clone 1D11). For immobilization, mAb were diluted at 1 μg/ml in PBS, then plated at 100 μl/well and incubated overnight at 4°C. Prior to blocking assays, the plates were washed twice with PBS.

γδTc cells were resuspended at 6 × 10⁶ cells/ml in complete medium (RPMI 1640 with 10% FBS, heat-inactivated, 1× MEM NEAA, 10 mM HEPES, 1 mM sodium pyruvate, 50 U/ml penicillin–streptomycin, and 2 mM l-glutamine—all from Invitrogen) plus 20 ng/ml recombinant human IL-2 (Miltenyi Biotec, Auburn, CA, USA) where indicated. For Fc blocking experiments, 5 μl Human TruStain FcX (BioLegend, San Diego, CA, USA) were added per 600,000 cells in 100 μl and incubated for 10 min at room temperature prior to the addition of mAb. 6 μg mAb was added to 600 μl cell suspension/test in Eppendorf tubes, then 100 μl/well plated in a 96-well round-bottomed plate. After incubation at 37°C for 30 min, 100 μl complete medium only (blocking assay only), or CalAM-labeled targets were added (cytotoxicity assay). After further incubation at 37°C for 4 h, two wells/condition of resuspended cells were pooled and stained for flow cytometry (blocking assay only). Experimental controls were untreated and IgG-treated cells. For cytotoxicity assays, plates were centrifuged and supernatants transferred to fresh 96-well plates (Falcon, U bottom, low evaporation) for CalAM fluorescence detection on a fluorimeter (FLUOstar Omega, BMG labtech). Controls were CalAM-labeled target cells incubated alone (spon = spontaneous release) and 0.05% Triton-X 100 (Thermo Fisher Scientific)-treated cells (max = maximum release). Percent lysis was calculated: [(test − spon)/(max − spon)] × 100%.

γδTc were incubated for 4 h at 37°C, 5% CO₂ in serum-free and cytokine-free medium; they were then stimulated with 1 μg mAb for 1 min at 37°C. Anti-CD3 (UCHT1, BioLegend) stimulated cells were a positive control. Lysates were run on 12% SDS PAGE and Western blotting performed using a 1:400 dilution of the PathScan^® Multiplex Western Cocktail I (Cell Signaling Technology, Danvers, MA, USA) to detect phosphorylated signaling proteins.

Paired 1-tailed Student’s t-tests were performed (Figures 2E–G only) using Microsoft Excel version 15.3 (Microsoft, Redmond, WA, USA). ANOVA analysis followed by Bonferroni’s pairwise multiple comparison post hoc tests were performed using Prism 7.0 for Mac OSX (GraphPad Software, San Diego, CA, USA). Differences were considered significant when P < 0.05; degrees of significance are indicated by letters or asterisks as defined in the figure legends.

Incubation of γδTc with anti-human pan γδTCR mAb clones B1, 5A6.E9, and 11F2 reduces the cytotoxicity of human γδTc against T47D breast cancer cells by 1.6- to 2.8-fold/14–25% compared to IgG controls (Figure 1A). Yet if all three mAb perform an equivalent blocking function, we would expect the same decrease in lysis to occur in all cases. On the contrary, there was less reduction in average cytotoxicity with 5A6.E9 (12%) compared to 11F2 (22%) and B1 (25%) clones. One-way ANOVA followed by Bonferroni’s multiple comparisons test revealed significant differences in cytotoxicity for B1- and 11F2-blocked γδTc (P = 0.0023 and P = 0.0051, respectively), whereas the decrease in cytotoxicity with 5A6.E9 mAb was not significant (P = 0.1065) compared to IgG. This lead us to further investigate whether other mechanisms may be at play. Untreated controls were included in parallel to IgG and mAb treatments in most experiments, to verify similarity to IgG controls. However, statistical analyses reported herein focus on mAb versus IgG treatments; for simplicity, statistical analyses of mAb versus untreated control samples are not shown.

When we incubated γδTc alone with anti-γδTCR (B1), anti-NKG2D, or IgG control antibodies for 4.5 h, in the absence of targets, uptake of ZA viability dye indicated increased cell death in B1- but not in anti-NKG2D-treated cells (Figure 1B, gating controls in Figure S1A in Supplementary Material). Thus, the decrease in cytotoxicity observed after γδTCR “blocking” appears to be at least partially due to the untimely death of a significant proportion of effector γδTc. Importantly, this did not occur when anti-NKG2D was used, suggesting that prior interpretations implicating NKG2D in cytotoxicity were not compromised by the induction of effector cell death described here.

To further define this cell death, we extended our experiments to include AnnV staining for the detection of apoptotic cells via flow cytometry (28). We categorized cell death into early apoptotic (AnnV+/ZA−), late apoptotic (AnnV+/ZA+), and necrotic (AnnV−/ZA+) fractions. Treatment of γδTc with B1 and B1.1 anti-γδTCR blocking mAb resulted in increased apoptotic cell death compared to IgG controls, in both early (17.6 and 11.1 versus 4.6%) and late apoptotic compartments (34.2 and 33.9 versus 27.6%) in a representative example (Figure 1C; gating controls in Figure S1B in Supplementary Material); combined results for technical replicates from this experiment are also shown (n = 3, Figure 1D). One-way ANOVA followed by Bonferroni’s pairwise multiple comparison post hoc analysis revealed significant reductions in cell viability after incubation with B1 (P = 0.0003) and B1.1 (P < 0.0001) mAb compared to IgG. Surprisingly, a significant γδTc viability difference was also found comparing B1 and B1.1 (P = 0.0003), considered to be the same mAb, simply sold by different companies. The BioLegend product data sheet for B1 states “Clone B1 is also known as clone B1.1”; however, the nomenclature would suggest that B1.1 is a subclone of B1. Unfortunately, we were unable to learn anything about the generation of these clones, which is not reported in the literature. Since the effect (viability loss) was always similar in experiments carried out with both antibody clones (data not shown), yet B1.1 had a more significantly negative effect on γδTc, we used B1 for all experiments moving forward.

Induction of cell death via anti-γδTCR incubation was specific to γδTc, since the viability of αβTc expanded in parallel from the same donor was not compromised by incubation with B1, 5A6.E9, or 11F2 (Figure 1E). αβTc viability remained unchanged, while once again, the strongly significant reduction in γδTc viability after treatment with B1, 5A6.E9, and 11F2 compared to IgG was confirmed (P < 0.0001). Notably, there was also a significant difference in viability between B1 and 5A6.E9 (P = 0.0034) as well as between B1 and 11F2 (P = 0.0040), demonstrating that the detrimental effect of B1 is greater than that of the other two clones. While there was a significant difference in γδTc versus αβTc viability in the control samples, which is understandable, as these are two different cultures (albeit derived from the same donor), there was no difference in αβTc viability among antibody treatments or controls. Three independent experiments with three different donor cultures testing B1 compared to IgG and control are shown and the significant drop in γδTc viability after incubation with B1 was confirmed (Figure 1F, P = 0.02).

To determine whether stimulation with anti-γδTCR mAb clones differentially activates γδTc, we stimulated effector γδTc for 1 min with anti-γδTCR mAb and used anti-CD3 stimulation with the UCHT1 clone as a positive control for activation. While clones B1 and B1.1 did not cause phosphorylation of signaling proteins AKT, ERK1/2, or S6 above that of the unstimulated control, clones 11F2 and 5A6.E9 elicited a phosphorylation pattern similar to that obtained with anti-CD3 stimulation (UCHT1) suggesting active signaling (Figure S2 in Supplementary Material). Rab11 served as an internal loading control. Clones B1 and B1.1 behaved as blocking antibodies should, by not inducing activation; however, a lack of survival signaling through phosphoAKT cannot account for the cell death observed in B1-treated γδTc, since the untreated control cells in our blocking experiments did not die to the same extent, although their activation patterns are quite similar (Figure S2 in Supplementary Material, compare lanes 1–3).

We tested whether exogenous IL-2 impacts the viability of γδTc in the presence of B1, 5A6.E9, or 11F2 mAb in a 4.5 h assay. Representative dot plots show that incubation with mAb resulted in an increase in both early apoptotic (AnnV+/ZA−) and late apoptotic (AnnV+/ZA+) γδTc, even in the absence of exogenous IL-2 (Figure 2A; gating shown in Figure S1C in Supplementary Material). Two-way ANOVA followed by Bonferroni’s multiple comparison post hoc statistical analysis revealed significant loss of γδTc viability due to IL-2 only in control (P < 0.0074) and B1-treated groups (P = 0.0042, Figure 2B). In contrast, no IL-2-induced differences in viability were observed in cell populations treated with IgG, 5A6.E9, or 11F2 mAb; however, both B1 and 11F2 mAb decreased γδTc viability significantly compared to IgG controls. Notably, 5A6.E9 treatment did not elicit significant loss of cell viability compared to IgG in this experiment. In the presence of IL-2, 5A6.E9-treated cells were more viable than those incubated with B1 (P = 0.0016) or 11F2 (P = 0.0067); the decrease in γδTc viability after incubation with 11F2 compared to 5A6.E9 in the absence of IL-2 was also deemed significant (P = 0.0163). Importantly, the difference in average γδTc viability between IgG- and B1-treated cells could account for a large proportion (if not all) of the decrease in cytotoxicity associated with the use of this mAb in blocking experiments. The cell death associated with mAb incubation was mostly apoptotic cell death (Figure 2C), with significant increases in IL-2-dependent cell death observed in control and B1-treated γδTc. Statistical analysis performed on cell death data shown in Figure 2C revealed the same results as in the reciprocal live cell data in Figure 2B.

We compiled the results for three independent experiments and statistical analysis thereof confirmed significant loss of cell viability (P = 0.0308) in B1- but not in 11F2- or 5A6.E9-treated γδTc compared to IgG in the presence of IL-2 (Figure 2D). To further demonstrate this effect, we show the impact of IL-2 on B1-treated cells from all three experiments (Figure 2E); a synergy was observed between B1 and IL-2 that resulted in significant loss of cell viability compared to B1 treatment in the absence of IL-2. This loss was evident in all three cultures but was most pronounced in the experiment performed with Donor culture 3-2 (P = 0.0006). Looking more closely at 5A6.E9 treatments of these three cultures (Figure 2F) revealed a small but significant difference with IL-2 in culture 1-2 (60.3 ± 1.0 versus 58.8 ± 0.8% live cells, without or with IL-2, respectively, P = 0.039), and sensitivity of 4-1 to IL-2 combined with this clone (P = 0.0048). Culture 4-1 viability was also decreased significantly with combined 11F2 and IL-2 treatment (Figure 2G, P = 0.029). The full experiment done with culture 4-1 is shown in Figure S3B in Supplementary Material. No significant difference in γδTc viability was observed among the three mAb-treated groups in the presence of IL-2 (Figure S3B in Supplementary Material, P > 0.9999). In the absence of IL-2, 5A6.E9 elicited the least degree of γδTc cell death and was significantly better than both B1 (P = 0.0026) and 11F2 (P = 0011), which were equally detrimental (Figure S3B in Supplementary Material). Donor culture 4-1 was 78.8% Vδ2+ (Figure S3A and Table S1 in Supplementary Material), whereas 1-2 with 14.9% and 3-2 with 50.8% Vδ2+ (Table S1 in Supplementary Material) exhibited little to no difference in viability upon incubation with 5A6.E9 or 11F2, with or without IL-2 (Figures 2F,G).

Western blot analysis of cleaved Caspase 3 in total cell lysates did not reveal clear differences among IL-2-treated and untreated cells or mAb treatments over 4.5 h; however, this is a less sensitive method than flow cytometry, thus it was not surprising that we were unable to detect ~10% differences (data not shown).

We found no differences in γδTc viability when comparing the use of soluble and immobilized mAb, all in the absence of IL-2; however, B1 treatment resulted in a significant loss of viability compared to IgG-treated γδTc in both cases in the representative example shown (soluble, P < 0.0005; coated, P < 0.0001; Figure 2H). When we included 5A6.E9 and 11F2 in this experiment, again no differences in viability were observed using soluble or immobilized mAb, but significant viability losses compared to IgG were noted (Figure S4 in Supplementary Material). When we compiled results for control, IgG and B1 treatments from four independent experiments, no significant differences were identified (Figure 2I). This is likely due to inter-donor variation, suggesting that variability in cell viability among donors is greater than that observed after treatment with soluble or immobilized mAb.

We tested whether the Fc receptor might be involved in apoptosis induced by mAb treatment, but found that there were no differences in viabilities of γδTc that were or were not pretreated with an Fc blocking reagent prior to mAb incubation (Figures 2J,K). This was true whether IL-2 was present (Figure 2J) or not (Figure 2K). Significant decreases in viability were noted for mAb-treated cells (Figures 2J,K), confirming previous results (Figures 2A–I).

To understand the kinetics of apoptosis, we performed a time course experiment that revealed a highly significant decrease in live cells after 3.5 h of stimulation with B1 that extended to 4.5 h (P < 0.0001, Figure S5A in Supplementary Material); of note, there was a significant difference in viability, compared to IgG-treated cells, at the 30-min time point (P = 0.043), but this was no longer evident at 1.5 and 2.5 h post mAb treatment. A graph showing the percentage of dead cells over time is also shown (Figure S5B in Supplementary Material). Since cytotoxicity experiments are usually conducted over a minimum of 4 h, this loss of effector viability should be taken into account.

Since the presence of IL-2 during blocking mAb treatment caused unwanted γδTc death, we investigated whether IL-2 is necessary for assessment of γδTc cytotoxicity against breast tumor targets. In the representative example shown, IL-2 did not significantly enhance γδTc cytotoxicity against T47D at any effector:target ratio tested (Figure 3A). This was also true for the individual and compiled results of four independent experiments testing γδTc cytotoxicity against T47D (Figure 3B). We further confirmed these results in cytotoxicity assays against MCF-7 (Figures 3C,D) and MDA-MB-231 (Figures 3E,F) targets. In each case at all ratios tested, there was no significant difference in cytotoxicity of γδTc against breast tumor targets in the presence or absence of IL-2 (Figure 3).

Taking all of our previous results into account, we designed parallel blocking-only and blocking plus cytotoxicity assays, in the absence of IL-2, to assess the involvement of γδTCR in killing T47D breast tumor targets, while considering anti-γδTCR mAb-induced effector cell death. Lysis of T47D targets was reduced dramatically upon treatment of γδTc with anti-γδTCR mAb; however, incubation with 5A6.E9 did not reduce lysis to quite the same extent, as significant differences in cytotoxicity observed after treatment with B1 (P = 0.0072) and 11F2 (P = 0.0128) versus 5A6.E9 were evident (Figure 4A). In line with these data, 5A6.E9 also induced significantly less cell death in these donor γδTc compared to B1, as did 11F2 (Figure 4B, both P < 0.0001 versus B1). Looking at the percentage of reduction in lysis or live cells compared to IgG, it is evident that at least half of the decrease in percent lysis attributed to γδTCR blocking by B1 is due to γδTc death, as opposed to blocking interactions between effectors and targets. In this experiment, 5A6.E9 and 11F2 appear to be the better clones for blocking, as γδTc viability was only reduced by 14 and 23%, respectively, compared to IgG, while they still caused 63 and 87% reductions in lysis of T47D (Figures 4A,B). A compilation of data from three independent experiments confirmed a decrease in percent lysis of T47D upon treatment with all three anti-γδTCR mAb clones; however, there was considerable variability in observed cytotoxicity of these different donor cultures against T47D breast cancer targets, rendering differences among treatment groups insignificant (Figure 4C). In contrast, significant reductions in γδTc viability remained evident, despite inter-donor variation, after treatment with B1 and 11F2, but not 5A6.E9 mAb (Figure 4D). These data confirm results shown in Figures 2A–D.