Zoledronic acid renders human M1 and M2 macrophages susceptible to Vδ2⁺ γδ T cell cytotoxicity in a perforin-dependent manner

Anonymised leukocyte cones from healthy donors were obtained from the National Health Service blood transfusion unit at St. George’s Hospital, London. PBMCs were isolated by density-adjusted centrifugation using Histopaque-1077 (Sigma-Aldrich). RBCs were lysed with ammonium chloride solution and platelets removed by slow-speed centrifugation. PBMCs were resuspended at 2 × 10⁷ cells/ml of freezing medium (45% RPMI-1640, 45% FBS and 10% DMSO; all from Sigma-Aldrich) and frozen at −80 °C in Mr Frosty freezing containers (Thermo Scientific) prior to transferring them to liquid nitrogen.

All cell culture was carried out in a humidified incubator at 37 °C with 5% CO₂. To generate Mϕs, monocytes were isolated from PBMCs using CD14 microbeads according to the manufacturer’s instructions (Miltenyi Biotec). Monocytes were resuspended in serum-free medium (RPMI-1640 containing 2 mM l-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin; all from Sigma-Aldrich) at a density of 3.8 × 10⁵ cells/ml, and 200 μl, 2 or 5 ml of cell suspension added per well of 96-well, 12-well or 6-well tissue culture plates, respectively (Thermo Scientific). Monocytes were cultured for 2 h, after which time the majority of cells were adherent to the tissue culture plate. This process is known to activate monocytes and initiate the macrophage colony-stimulating factor production required for Mϕ differentiation [21]. The adherent monocytes were then cultured for 10 days in complete medium (RPMI-1640 containing 10% FBS, 2 mM l-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin), after which time the monocytes had differentiated into Mϕs, as indicated by the morphological changes and plastic adherence observed using light microscopy. M1 and M2 Mϕs were generated by adding 25 ng/ml of recombinant human IFN-γ or IL-4 (R and D Systems), respectively, on day 7. Mϕs that had not been treated with IFN-γ or IL-4 (designated M0s) were used as controls throughout. 10 μM ZA (Sigma-Aldrich) was added to the Mϕs on day 9. To generate pure populations of Vδ2⁺ T cells, PBMCs were resuspended at 2 × 10⁶ cells/ml of complete medium containing 1 μM ZA and 5 ng/ml recombinant human IL-2 (R and D Systems), and 250 μls of cell suspension added per well of 96-well round-bottomed tissue culture plates (Thermo Scientific). The cells were cultured for 9 days and fed every 2–3 days with fresh medium containing 5 ng/ml IL-2. Dead cells and non-γδ T cells were depleted sequentially using dead cell removal kits and TCRγδ negative isolation kits according to the manufacturer’s instructions (Miltenyi Biotec). The purity of Vδ2⁺CD3⁺ cells was assessed by flow cytometry using PE-conjugated mouse anti-human Vδ2 (clone 123R3; Miltenyi Biotec) and PerCP-conjugated mouse anti-human CD3 (clone SK7; Biolegend) or FITC-conjugated mouse anti-human CD3 (clone HIT3a; Becton–Dickinson). For one donor, a high percentage of Vδ1⁺CD3⁺ cells was detected post-isolation, and so Vδ1⁺ cells were depleted using allophycocyanin-conjugated recombinant human anti-Vδ1 (clone REA173; Miltenyi Biotec) and anti-allophycocyanin microbeads according to the manufacturer’s instructions (Miltenyi Biotec; supplementary Fig. 1). We speculate that this donor’s PBMCs had a particularly high percentage of Vδ1⁺ T cells prior to ZA and IL-2 stimulation and/or their Vδ1⁺ T cells underwent bystander expansion in response to ZA and IL-2.

Day 10 Mϕs in six-well tissue culture plates were washed twice in PBS (Sigma-Aldrich) and cultured for 15 min in PBS containing 0.25% trypsin (Life Technologies) and 2 mM EDTA (Sigma-Aldrich). Cells were detached by repeated pipetting and then washed in complete medium to deactivate the trypsin. Mϕs were resuspended in flow cytometry buffer (PBS with 1% BSA and 0.09% sodium azide; all from Sigma-Aldrich) containing either FITC-conjugated mouse anti-human CD64 (clone 10.1; Becton–Dickinson) or PE-conjugated mouse anti-human CD206 (clone 19.2; Becton–Dickinson). Matched isotype controls were used to determine the amount of background expression. After 10 min at room temperature, cells were washed in flow cytometry buffer and fixed in CellFIX (Becton–Dickinson). Perforin expression in Vδ2⁺ T cells was assessed in PBMCs cultured with ZA and IL-2 for 0, 1 and 9 days as described in "Cell culture". Cells were resuspended in flow cytometry buffer containing PE-conjugated mouse anti-human Vδ2 (clone 123R3; Miltenyi Biotec) and PerCP-conjugated mouse anti-human CD3 (clone SK7; Biolegend). After 10 min at room temperature, cells were washed in flow cytometry buffer and simultaneously fixed and permeabilised using Cytofix/Cytoperm (Becton–Dickinson) according to the manufacturer’s instructions. Cells were washed and resuspended in Perm/Wash buffer (Becton–Dickinson) and then labelled with FITC-conjugated mouse anti-human perforin (clone δG9; Becton–Dickinson) or matched isotype controls. After 10 min at room temperature, cells were washed in Perm/Wash buffer and resuspended in flow cytometry buffer. Samples were acquired on an LSR II flow cytometer (Becton–Dickinson) and analysed using FlowJo software. All comparatively analysed samples were acquired on the same day except for the time course of perforin expression where day 0, 1 and 9 samples were acquired on different days. The mean fluorescence intensity (MFI) values stated throughout are arithmetic means.

Detaching the Mϕs from the tissue culture plates prior to performing the cytotoxicity assays resulted in poor viability; therefore, cytotoxicity was assessed by adding Vδ2⁺ T cells directly to adherent Mϕs. Day 10 Mϕs in 12-well tissue culture plates were washed twice in PBS and then cultured for 20 min in PBS containing 1 μM carboxyfluorescein succinimidyl ester (CFSE; Life Technologies). Mϕs were washed three times in complete medium and then cultured overnight with or without 1.52 × 10⁶ autologous Vδ2⁺ T cells per well in 2 ml complete medium to obtain an E:T ratio of 2:1 based on the initial seeding density of monocytes. For some experiments Vδ2⁺ T cells were pre-treated for 2 h with or without 100 ng/ml concanamycin A (CMA; Abcam) or DMSO, then washed three times in complete medium prior to being cultured with Mϕs. Non-adherent cells were collected and adherent cells detached from the tissue culture plates as described in “Flow cytometry”. All cells were washed in PBS and then labelled with Zombie-NIR live/dead cell discrimination dye according to the manufacturer’s instructions (Biolegend). Zombie-NIR binds to amine groups on proteins, but does not penetrate an intact plasma membrane. Live cells have relatively low expression because only cell surface proteins are available for binding, whereas dead cells exhibit higher levels of expression because their compromised plasma membrane permits binding to both extracellular and intracellular proteins. After 15 min at room temperature, cells were washed in complete medium and fixed in CellFIX. Samples were acquired on an LSR II flow cytometer and analysed using FlowJo software. All comparatively analysed samples were acquired on the same day.

Day 10 Mϕs in 96-well tissue culture plates were washed three times in PBS and then cultured for 5 h with 1.52 × 10⁵ autologous Vδ2⁺ T cells per well in 200 μl complete medium to obtain an E:T ratio of 2:1 based on the initial seeding density of monocytes. Allophycocyanin-conjugated mouse anti-human CD107a (clone H4A3; Biolegend) and FITC-conjugated mouse anti-human CD107b (clone H4B4; Biolegend) or matched isotype controls were added directly to the wells at the start of the co-culture along with 1 μg/ml of monensin to neutralise intracellular acidity. Cells were then collected and labelled with PE-conjugated mouse anti-human Vδ2 (clone 123R3; Miltenyi Biotec) and PerCP-conjugated mouse anti-human CD3 (clone SK7; Biolegend) as described in “Flow cytometry”. Samples were acquired on an LSR II flow cytometer and analysed using FlowJo software. All comparatively analysed samples were acquired on the same day.

Data in Figs. 1b, c, 3b, d and 4c were analysed by repeated measures one-way or two-way ANOVA and comparisons between means carried out using either Tukey’s or Sidak’s multiple comparison tests (GraphPad Prism 6). *, **, *** and **** were used to indicate p values of <0.05, <0.01, <0.001 and <0.0001, respectively. Gaussian distributions were assumed. Data in Fig. 2b included a three-way (3 × 2 × 2) factorial design repeated six times using cells from six different donors. The three factors were Mϕ type (M0, M1 and M2), ±ZA and ±Vδ2 cells. Data in Fig. 4b were a three-way (3 × 2 × 4) factorial design repeated five times using cells from five different donors. The three factors were Mϕ type (M0, M1 and M2), ±ZA and ±Vδ2 cells (−Vδ2, +Vδ2, +Vδ2[DMSO] and +Vδ2[CMA]). Data in Figs. 2b and 4b were analysed by three-way ANOVA and comparisons between means carried out using Fisher’s least significant difference (LSD; Genstat 18). Assumptions underlying the analysis were checked using the diagnostic plots produced by the software. LSDs at the 5, 1 and 0.1% level are depicted by black intervals, and differences in the means that were greater than this interval were deemed significant to an equivalent p value of <0.05, <0.01 and <0.001, respectively.

We differentiated monocytes from the peripheral blood of healthy donors into Mϕs and treated them with IFN-γ or IL-4 to generate M1 and M2 Mϕs, respectively. We then characterised these Mϕs based on their expression of markers for M1 Mϕs (CD64 and IL-12p70) and M2 Mϕs (CD206 and CCL18) [19, 22]. M1 Mϕs had upregulated expression of CD64, whereas M2 Mϕs had downregulated CD64 and upregulated CD206 (Fig. 1a, b). Although, statistically, we observed significantly higher levels of CD206 on M1 Mϕs compared with M0 Mϕs in terms of percentage expression, this was not consistent for all donors and not statistically significant in terms of relative MFI (Fig. 1b). Mϕs were then cultured overnight with or without LPS to measure the production of IL-12p70 and CCL18, respectively. M1 Mϕs produced more IL-12p70 than M0 and M2 Mϕs, whereas M2 Mϕs produced more CCL18 than M0 and M1 Mϕs (Fig. 1c). We also tested whether ZA—added for the last 18 h of culture—had any effect on these markers and found little or no difference between untreated and ZA-treated Mϕs (Fig. 1). Taken together, these data validate our protocol for generating M1 and M2 Mϕs and show that ZA does not alter the M1 and M2 profile of the Mϕs in this system.

To obtain sufficient cell numbers for cytotoxicity assays, we stimulated Vδ2⁺ T cell expansion prior to isolation. Vδ2⁺ T cell expansion was observed in PBMCs treated with ZA and IL-2 for 9 days, as shown by increased frequencies of Vδ2⁺CD3⁺ cells (supplementary Fig. 2). Vδ2⁺ T cells were purified by sequentially depleting dead cells and non-γδ T cells (mean ± standard deviation (SD) for the percentage of Vδ2⁺CD3⁺ cells from four donors = 97.2 ± 1.8; supplementary Fig. 2). The percentage of Vδ2⁺CD3⁺ cells at day 0 and day 9 pre-depletion of dead cells and non-γδ T cells was not assessed routinely; however, purities at day 9 post-depletion were assessed for all isolations performed in this study (mean ± SD for 14 isolations = 97.7 ± 1.8). We conducted preliminary experiments to determine the optimal E:T ratio and ZA concentration for Vδ2⁺ T cell-mediated cytotoxicity against ZA-treated Mϕs (supplementary Fig. 3). These experiments showed Vδ2⁺ T cell cytotoxicity and degranulation against Mϕs treated with 10 μM, but not 1 μM ZA (supplementary Fig. 3). Furthermore, they showed marked killing at the lowest E:T ratio of 2:1 (supplementary Fig. 3). Using the 10 μM concentration of ZA and 2:1 E:T ratio, we found that ZA had little or no effect on Mϕ viability in the absence of Vδ2⁺ T cells, and Vδ2⁺ T cells did not induce cell death in Mϕs that had not been treated with ZA (Fig. 2a, b). However, there was a marked increase in the amount of cell death in Mϕs that were pre-treated with ZA and then cultured with Vδ2⁺ T cells (Fig. 2a, b). Although, statistically, Vδ2⁺ T cell-mediated killing of ZA-treated M1 Mϕs was significantly higher than that of M0 Mϕs, the difference was relatively small and no statistically significant difference was found between M1 and M2 Mϕs (Fig. 2b). These results suggest that Vδ2⁺ T cells are cytotoxic towards ZA-treated Mϕs irrespective of their M0, M1 and M2 phenotype.

Perforin has been shown previously to play a role in γδ T cell cytotoxicity towards tumour cell lines [23, 24]; therefore, we tested whether perforin contributes to Vδ2⁺ T cell cytotoxicity towards ZA-treated Mϕs. We measured perforin expression by Vδ2⁺ T cells before, during and after expansion with ZA and IL-2. We found that, although resting Vδ2⁺ T cells expressed little or no perforin, it was markedly upregulated after 1 day of culture with ZA and IL-2 (Fig. 3a, b). After 9 days of culture with ZA and IL-2, perforin was downregulated but still expressed by Vδ2⁺ T cells (Fig. 3a, b). We also measured perforin expression in expanded and isolated Vδ2⁺ T cells from six donors and found consistent expression in terms of percentage expression (mean ± SD = 1.2 ± 0.5 vs. 23.9 ± 7.6 for isotype and test, respectively) and MFI (mean ± SD = 227.8 ± 15.2 vs. 490.7 ± 104.6 for isotype and test, respectively). To determine if Vδ2⁺ T cells release perforin when cultured with ZA-treated Mϕs, we measured the mobilisation of lysosomal-associated membrane protein 1 and 2 (i.e. CD107a and CD107b) to the surface of Vδ2⁺ T cells. CD107a—and to a lesser extent CD107b—was expressed on Vδ2⁺ T cells that were isolated from PBMCs after 9 days of culture with ZA and IL-2 (Fig. 3c). This may represent residual CD107 expression from the monocyte-dependent degranulation that is induced when PBMCs are exposed to ZA [20]. Vδ2⁺ T cells upregulated expression of CD107a and CD107b on their cell surface when cultured with ZA-treated Mϕs compared with untreated Mϕs (Fig. 3c, d). These data suggest that Vδ2⁺ T cells express perforin and degranulate in response to ZA-treated Mϕs, thus implicating a role for perforin in Vδ2⁺ T cell cytotoxicity towards ZA-treated Mϕs.

To explore further the potential role of perforin in Vδ2⁺ T cell cytotoxicity towards ZA-treated Mϕs, we repeated the cytotoxicity assays shown in Fig. 2, but this time pre-treated Vδ2⁺ T cells with the H⁺-ATPase inhibitor CMA. CMA blocks acidification of cytolytic granules, which inhibits perforin-mediated but not Fas ligand-mediated cytotoxicity [25]. We found that pre-treating Vδ2⁺ T cells with CMA reduced their cytotoxicity towards Mϕs compared with DMSO controls (Fig. 4a, b). We calculated the percentage inhibition for Vδ2⁺ T cell cytotoxicity towards M0, M1 and M2 Mϕs and found that Vδ2⁺ T cell cytotoxicity towards M0 Mϕs was more sensitive to CMA than towards M1 Mϕs (Fig. 4c). To determine whether CMA had an effect on Vδ2⁺ T cell viability, we applied a gate to CFSE⁻ cells and calculated the percentage of Zombie-NIR^low cells (supplementary Fig. 4a). There was a discernible reduction in Vδ2⁺ T cell viability in the presence of ZA-treated Mϕs compared with untreated Mϕs; however, there was little or no difference in Vδ2⁺ T cell viability between the CMA and DMSO treatment groups (supplementary Fig. 4b). These findings suggest that Vδ2⁺ T cell cytotoxicity towards ZA-treated Mϕs is sensitive—at least in part—to CMA, thus implicating a role for perforin.