hPBMC were isolated from buffy coats by Ficoll-Hypaque gradient centrifugation of EBV⁺ healthy donors after informed consents were obtained. PAM-expanded Vγ9Vδ2-T cells were prepared according to the protocol we established before (40). Briefly, hPBMC were cultured in RPMI1640 medium with 10% fetal bovine serum (FBS) in the presence of PAM from day 0 to day 3 at a concentration of 9μg/ml. Recombinant human IL-2 was added to medium from day 3 to day 14 at a concentration of 500 IU/ml. After 2 weeks, the γδ-T cells were purified by positive selection with α-TCRγ/δ MicroBead (Miltenyi Biotec).

Purified Vγ9Vδ2-T cells were cultured with IL-10^low or IL-10^high conditioned medium for 24h, RPMI 1640 with 10% FBS medium (plain medium, PM) as a control. The pretreated Vγ9Vδ2-T cells (effector cells, E) were cocultured with autologous EBV-LCL (target cells, T) at an E: T ratio of 10:1 for 4 to 6 h in the IL-10^low/high CM or PM, and then the death of target cells was analyzed with flow cytometry. Cells were stained with anti-CD3 to identify Vγ9Vδ2-T cells and propidium iodide (PI) was used to identify dead cells. The death of EBV-LCL was shown as the percentage of PI⁺ cells in the CD3^- population (40). In some experiments, neutralizing antibody against IL-10 (abcam) was added to block IL-10 mediated pathways. To confirm the suppressive role of IL-10 in the CM, recombinant hIL-10 (Peprotech) or recombinant vIL-10 (R&D systems) was added to culture medium at the indicated concentration.

EBV-secreting cell lines B95-8 and B95.8EBfaV-GFP were cultured and EBV-containing supernatants were collected for the following infection. hPBMC were incubated with EBV-containing supernatants, and then cultured in the RPMI 1640 medium containing 15% FBS with the addition of cyclosporine-A (1μg/ml) as we describe before (15).

EBV-LCL were cultured in RPMI1640 medium for 24 h. The conditioned medium (CM) was collected, centrifuged at 5000 rpm at 4°C for 10 min to remove cell debris and then frozen at −80°C in aliquots. Stored CM was passed through a 0.22-μm syringe filter (Millipore) before use. Plain medium (PM) collected from complete medium without cell incubation under the same experimental conditions served as the control for CM.

For hIL-10, the concentrations in conditioned medium were measured by ELISA. The procedures for human IL-10 ELISA kits (Biolegend, San Diego, CA, USA) were performed based on the manufacturer’s instructions. For vIL-10, the concentrations in conditioned medium were measured according to the method described before (41). The conditioned medium was concentrated by Amicon-Ultra centrifugation filters (Millipore) following the manufacturer’s instructions. Then, the concentrated conditioned medium was used for performing Western blot assay. Mouse monoclonal antibody against vIL-10 (R&D) was used as primary antibody for incubating transferred membranes at 4°C overnight. Horseradish peroxide conjugated goat anti-mouse secondary antibody (R&D) was used as secondary antibody for detecting vIL-10 levels. The bands of Western blot were quantified by “Gels” analysis tool of ImageJ. Recombinant vIL-10 was used as a standard to quantify the vIL-10 level in conditioned medium.

All animal studies were approved and performed in compliance with the guidelines for the use of experimental animals by the Committee on the Use of Live Animals in the Teaching and Research, the University of Hong Kong. Rag2^−/−γc^−/− mice were bred in Centre for Comparative Medicine Research of the University of Hong Kong. Humanized mice were generated according to the protocol we established before (14, 15). Rag2^−/−γc^−/− or humanized mice were inoculated with EBV-LCL expressing high or low level of IL-10 (0.1×10⁶/mouse) by subcutaneous injection to establish the EBV-associated lymphoma model. For Rag2^−/−γc^−/− mice, PAM-expanded Vγ9Vδ2-T cells (5×10⁶/mouse) with or without the addition of SA-hCD137L (5μg/mouse) were adoptively transferred intravenously into EBV-associated lymphoma murine model at indicated time. For humanized mice, PAM (5mg/kg body weight) and SA-hCD137L (15μg/mouse) were injected intraperitoneally at the indicated time. The mice treated with an equivalent volume of PBS or SA were used as controls. The tumor volume and mice survival were monitored every day and calculated at the indicated time. Mice were counted as dying when their subcutaneous tumor diameter was larger than 17 mm and thus sacrificed according to the regulation of Centre for Comparative Medicine Research of the University of Hong Kong. Otherwise, mice were monitored for 100 days before being sacrificed. The tumor tissues were reserved for immunohistochemical evaluation.

Recombinant SA-hCD137L proteins were generated as described before (37). Briefly, the DNA sequences were synthesized encoding the extracellular domain of human CD137L (a.a. 58-254) and the core streptavidin (SA; a.a. 16-133) with an N-terminal 6×His tag. The recombinant SA-hCD137L protein was expressed in E. coli by inserting the SA-hCD137L DNA fragments into the pETH expression vector and transforming into competent cells. After purifying with Ni-nitrilotriacetic acid affinity chromatography (QIAGEN, Germany), the recombinant SA-hCD137L protein was filtered a and quantitated by BCA Protein Assay Kit (Pierce, USA).

Cells were stained for surface molecules with the following antibodies: αIL10R (Miltenyi Biotec, clone REA239), αCD3 (Biolegend, clone HIT3a), αTCRγ9 (Biolegend, clone B3), αTCRVδ2 (Biolegend, clone B6), and αCD137 (Biolegend, clone 4B4-1). All samples were performed with a FACS LSR II (BD). The results were analyzed with FlowJo software.

The tumor tissues were fixed with 10% formalin for 24 h and maintained in 70% ethanol. Fixed tumor tissues were embedded in paraffin and sectioned. The tumor sections were performed immunohistochemistry staining with αIL-10 antibody (abcam) (42).

Data are shown in the form of mean ± standard error of the mean (SEM). All data were tested by Shapiro-Wilk test to verify the normality. For data that did not meet normal distribution, Mann-Whitney U test was used for analysis. For data that met normal distribution, one-way analysis of variance (ANOVA) with Bonferroni correction was used for analysis. For multiple variables, two-way ANOVA was used. Kaplan-Meier log-rank test was used for comparing survival among different groups. Two-tailed test was used for all analyses. P < 0.05 was regarded as significant.

To investigate the effects of IL-10 in TME on the antitumor activity of Vγ9Vδ2-T cells, conditioned medium (CM) was obtained by collecting the supernatant of EBV-LCL culture for modeling TME in vitro. As shown in Figure 1A, CM from EBV-LCL culture established from different donors contained distinct levels of IL-10, and vIL-10 accounted for about 9.56 ± 5.74% of total IL-10. CM collected from EBV-LCL1 and EBV-LCL6, which contained the lowest and highest concentrations of IL-10, was used as IL-10^low CM and IL-10^high CM, respectively, in the following experiments. Importantly, the cytotoxic activity of IL-10^high CM-treated Vγ9Vδ2-T cells against EBV-LCL was significantly lower than IL-10^low CM- or PM-treated Vγ9Vδ2-T cells (Figure 1B). To verify the immunosuppressive role of IL-10 in the CM, an IL-10 neutralizing mAb was applied to block IL-10 signaling during Vγ9Vδ2-T cells exposed to IL-10^high/low CM. The reduced cytotoxicity of Vγ9Vδ2-T cells against EBV-LCL was significantly abrogated when blocked with the IL-10 neutralizing mAb (Figure 1C). Furthermore, both hIL-10 and vIL-10 recombinant proteins showed dose-dependent inhibitions in the cytotoxicity of Vγ9Vδ2-T cells against EBV-LCL in the PM (Figures 1D, E). Taken together, our data indicate that the antitumor activity of Vγ9Vδ2-T cells against EBV-LCL was suppressed by both the hIL-10 and vIL-10 in the CM from EBV-LCL in vitro.

To determine whether the therapeutic effects of Vγ9Vδ2-T cells on EBV-induced B cell lymphoma were inhibited by IL-10 within the TME, EBV-LCL1 expressing low levels of IL-10 (IL-10^low LCL) and EBV-LCL6 expressing high levels of IL-10 (IL-10^high LCL) were inoculated into Rag2^-/- γc^-/- mice, respectively (Figure 2A). After 21 days, large subcutaneous tumors developed in all the mice as detected by in vivo imaging (Figures 2B, C). The expressions of IL-10 in the tumor tissues generated from IL-10^low LCL and IL-10^high LCL were detected by immunohistochemistry (Figure 2D). Consistent with our previous results (15), Vγ9Vδ2-T cell treatment constrained tumor growth and prolonged the survival of tumor-bearing mice in contrast with the mice treated with PBS as the control (Figures 2E, F). Importantly, Vγ9Vδ2-T cells showed less efficacy in controlling EBV-induced lymphoma developed from IL-10^high LCL compared with that developed from IL-10^low LCL, along with larger tumor volume and lower survival rates (Figures 2E, F). These results suggest that the decreased antitumor activity of Vγ9Vδ2-T cells against EBV-induced lymphoma may be associated with IL-10^high TME in vivo.

IL-10 mediates its biological effects mainly through a heterodimeric membrane receptor composed of IL-10R1 and IL-10R2 (43). Since IL-10R2 is shared by more than five IL-10 family cytokines (44), we investigated the expression of IL-10R1 on Vγ9Vδ2-T cells exposed to the IL-10^high CM upon γδ-TCR activation in vitro. Importantly, we found that following activation, IL-10R1⁺ Vγ9Vδ2-T cell subset expressed high levels of CD137 compared with IL-10R1^-/lo Vγ9Vδ2-T cell subset in the IL-10^high CM, indicating that CD137 could be an effective costimulatory signaling to restore the antitumor activity of Vγ9Vδ2-T cells compromised by the IL-10 in TME (Figures 3A, B).

To determine whether CD137 costimulation could provide an anti-exhaustion signal to mitigate the inhibiting effects mediated by IL-10 in TME, a recombinant SA-hCD137L protein containing a core streptavidin (SA) molecule with the extracellular domains of human CD137L (hCD137L) was generated as we reported previously (37). We found that the addition of the recombinant SA-hCD137L protein significantly inhibited the surface expression of IL-10R1 in total and CD137⁺ Vγ9Vδ2-T cells in IL-10^high CM in terms of both percentage and expression level (MFI) changes (Figures 3C–E). These results indicate that CD137 costimulation suppressed IL-10R1 expression in CD137⁺ Vγ9Vδ2-T cells, and thereby was able to reduce their sensitivity to endogenous IL-10 in the immunosuppressive TME.

To determine whether CD137 costimulation could rescue the impaired antitumor efficacy of Vγ9Vδ2-T cells in suppressive TME, the recombinant SA-hCD137L protein was added to the coculture of Vγ9Vδ2-T cells with EBV-LCL in IL-10^high CM for mimicking the tumor milieu. As shown in Figure 3F, the SA-hCD137L protein significantly increased the cytotoxicity of Vγ9Vδ2-T cells against EBV-LCL under both the immunosuppressive and normal microenvironments mimicked by the IL-10^high CM and the PM. Importantly, CD137 costimulation not only completely restored the reduced cytotoxicity of Vγ9Vδ2-T cells in the IL-10^high CM to normal levels, but also had a better effect to enhance the cytotoxic activity of Vγ9Vδ2-T cells in IL-10^high CM than that in PM (Figure 3F). These data demonstrate that CD137 engagement enables Vγ9Vδ2-T cells to withstand the hostile environment mediated by endogenous IL-10, resulting in the increase of the antitumor activity of Vγ9Vδ2-T cells in vitro.

Previously we have demonstrated that Vγ9Vδ2-T cells could control EBV-inducing lymphoma (15), and their antitumor activity in controlling EBV-induced lymphoma developed from IL-10^high LCL was lower than that developed from IL-10^low LCL (Figures 2E, F). To further elucidate the roles of CD137 costimulation in the compromised antitumor activity of Vγ9Vδ2-T cells in IL-10^high TME in vivo, EGFP-expressing IL-10^high LCL was inoculated in Rag2^-/- γc^-/- mice, s.c. (Figure 4A). Twenty-one days later, mice bearing subcutaneous tumors were randomly divided into three groups as detected by in vivo imaging (Figure 4B). No significant differences were found in fluorescent density from tumor cells among the three groups after 21 days of tumor cell inoculation (Figure 4C). PAM-expanded Vγ9Vδ2-T cells were adoptively transferred to one group of the tumor-bearing mice with the recombinant SA-hCD137L protein weekly from day 21 to day 42. The other two groups of mice were adoptively transferred with Vγ9Vδ2-T cells in the presence of PBS or SA as the controls. Importantly, Vγ9Vδ2-T cells in combination with SA-hCD137L treatment significantly limited tumor growth (Figure 4D) and improved mouse survival (Figure 4E) compared to treatments of Vγ9Vδ2-T cells with PBS or SA. These data indicate that the costimulation of CD137 efficiently enhanced the antitumor activity of Vγ9Vδ2-T cells in the highly immunosuppressive microenvironment mediated by IL-10 in vivo.

Previously we had demonstrated that PAM could expand Vγ9Vδ2-T cells in vivo to control EBV-induced lymphoma in humanized mice with functional hPBMC (15). We then investigated the role of CD137 costimulation on the therapeutic effect of PAM in controlling EBV-induced lymphoma with IL-10^high TME in humanized mice. EBV-induced lymphoma with IL-10^high TME model was generated by inoculation s.c. of IL-10^high EBV-LCL in humanized mice (Figure 5A) (15). All humanized mice developed subcutaneous tumors after IL-10^high EBV-LCL inoculation for 28 days with similar fluorescent density from tumor cells as detected by in vivo imaging (Figures 5B, C). PAM, SA-hCD137L, or the combination of these two agents were injected intraperitoneally (i.p.) at days 28, 35, 42, and 49 after IL-10^high EBV-LCL inoculation (Figure 5A). PBS- and SA-treated mice were controls. As a result, PAM administration alone decreased the tumor volume significantly and extended the survival of the tumor-bearing humanized mice compared with the treatment with PBS, SA, or SA-hCD137L protein alone, respectively (Figures 5D, E). Importantly, the combination treatment of PAM with SA-hCD137L was more potent than PAM alone to control the development of EBV-induced lymphoma with IL-10^high TME in humanized mice, in terms of tumor growth and survival (Figures 5D, E).

Humanized mice reconstituted with Vγ9Vδ2-T-cell-depleted hPBMC were also used to confirm whether the effect of SA-hCD137L costimulation on the control of EBV-induced lymphoma with IL-10^high TME was mediated by Vγ9Vδ2-T cells (Figure 5A). As shown in Figures 5D, E, there were no therapeutic effects by the combination treatment of PAM with SA-hCD137L in humanized mice reconstituted with Vγ9Vδ2-T-cell-depleted hPBMC. These data demonstrate that the recombinant SA-hCD137L protein had a synergistic effect with PAM to overcome the barriers of IL-10^high TME in vivo and this synergistic effect was mainly mediated by Vγ9Vδ2-T cells.