Bone marrow mononuclear cells (BMMC) from BM aspirates and autologous peripheral blood mononuclear cells (PBMC) from MM patients at different stages of disease (diagnosis: MM-dia; remission: MM-rem; relapse: MM-rel) were used for the study. All experiments were performed with BM samples from MM-dia unless otherwise specified. BMMC from patients with hematological malignancies in unmaintained molecular remission, frozen human normal BMMC purchased from Stem Cell Technologies, and PBMC from healthy donors attending the local Blood Bank were used as control (Ctrl). The study was approved by institutional regulatory boards (n.176-19 December 11, 2019).

The monoclonal antibodies (mAbs) used in the study are listed in Supplemental Table I . Cell surface and intracellular flow cytometry were performed as previously reported (9). Vγ9Vδ2 T cells were identified with αTCR Vγ9 mAb conjugated with the appropriate fluorochrome (FITC, PE, APC) depending on the multicolor staining combination (see Supplemental Table I ). We have intentionally focused on Vγ9Vδ2 T cells because this is the only γδ T-cell subset directly activated by pAgs or indirectly activated by ZA stimulation (36–38). Moreover, Vδ2 chain is the only one to combine with the Vγ9 chain confirming that αTCR Vγ9 mAbs are reliable tools to identify Vγ9Vδ2 T cells (39). Cytofluorimetric analyses were performed with FACS Calibur Cell Sorter and FlowJo software (Becton Dickinson, Mountain View, CA).

Cryopreserved or freshly isolated PBMC or BMMC from MM patients and Ctrl were cultured for 7 days with 10 IU/ml IL-2, and 1 µM ZA+10 IU/ml IL2. In selected experiments, cells were cultured in the presence of αPD1 (10 μg/ml), αTIM-3 (10 μg/ml), αLAG-3 (10 μg/ml), or a combination thereof. Proliferation was evaluated by calculating total counts of viable Vγ9Vδ2 T cells on day 7 with the trypan blue staining assay and flow cytometry after gating for CD3 in combination with appropriate αVγ9 mAb. IL-17 production was evaluated in freshly isolated BMMC after incubation with PMA (50 ng/ml)/Ionomycin (1 μg/ml) for 4 hours at 37°C and 5% CO2 with brefeldin (500 ng/ml) added during the last hour. IFN-γ, and CD107 expression were evaluated as previously reported (9).

Conventional T-cell proliferation was measured by carboxyfluorescein-diacetate-succinimidyl-ester (CFSE) dilution assay. BMMC were suspended in warmed PBS at a concentration of 10×10⁶ cells/ml and labeled with 1 μM CFSE at 37°C for 15 min in the dark. After quenching with FCS for 10 minutes in dark at 37°C and washing with RPMI medium, cells were seeded at 1×10⁶ cells/ml in 96‐well flat‐bottom plate and stimulated with αCD3 (1 μg/ml - BioLegend) and αCD28 (2 μg/ml - BioLegend) antibodies for 72 h at 37°C. After 3 days, conventional T cells where harvested and identified with αCD8 and αCD4 rather than αCD3 given the down-modulation induced by αCD3/αCD28 stimulation and the lineage discrimination capacity of CD4 and CD8 expression (40). In selected experiments, the proliferation of BM CD4 and CD8 T cells with αCD3 and αCD28 was performed in the presence (BMMC) or absence of γδ T cells (BMMC-γδ-). Depletion was performed by immune magnetic separation using Anti-pan-γδ-conjugated magnetic microbeads (Miltenyi Biotec, Germany #130-050-701).

For Western blot experiments, γδ T cells were purified by immune magnetic separation using Anti-pan-γδ-conjugated magnetic microbeads (Miltenyi Biotec, Germany #130-050-701). Purity was always > 90% by FITC-conjugated-Hapten MicroBeads staining (Miltenyi Biotec, Germany #130-050-701). After ZA stimulation, Vγ9Vδ2 T cells were the predominant population also in MM patients who did not respond to ZA stimulation ( Supplemental Figure 1 ). Cells were lysed in a MLB buffer (125 mM Tris-HCl, 750 mM NaCl, 1% v/v NP40, 10% v/v glycerol, 50 mM MgCl2, 5 mM EDTA, 25 mM NaF, 1 mM NaVO4, 10 μg/ml leupeptin, 10 μg/ml pepstatin, 10 μg/ml aprotinin, 1 mM phenylmethylsulphonyl fluoride, pH 7.5), sonicated and centrifuged at 13,000 × g for 10 min at 4°C. Twenty μg of proteins from cell lysates were subjected to Western blotting and probed with the antibodies listed in Supplemental Table II . The proteins were detected by enhanced chemiluminescence (Bio-Rad Laboratories). The band density analysis was performed using the ImageJ software (https://imagej.nih.gov/ij/) and expressed as arbitrary units. The ratio band density of each protein/band density of tubulin (as housekeeping protein) was calculated in each experimental condition. For untreated/baseline/unstimulated cells, the band density ratio was considered 1. For the other experimental conditions, the ratio was expressed as proportion towards the ratio obtained in untreated cells.

Supernatants (S/N) from Ctrl and MM BMMC stimulated for 7 days with 10 IU/ml IL2, 1 µM ZA+10 IU/ml IL2 in the presence or absence of αPD1 were collected and stored at -80°C. The concentration of human IL27 was quantified in S/N by enzyme-linked immunosorbent assay (ELISA) technology with the IL-27 Human ELISA kit (Invitrogen; Catalogue number: # BMS2085) according to manufacturer’s instructions.

The results are expressed as mean ± SE. Differences between the groups have been evaluated with the one-way analysis of variance, and the Wilcoxon–Mann–Whitney non-parametric test for paired or unpaired samples as appropriate and considered to be statistically significant for p values <0.05. Correlation analyses have been performed with the non-parametric Spearman Rank Order test with a cut-off p value <0.05.

Figure 1A shows PD-1, TIM-3, LAG-3 and CTLA-4 expression in resting PB and BM Vγ9Vδ2 T cells from Ctrl and MM patients. PD-1 and TIM-3 expression was significantly higher in BM of MM patients than in Ctrl samples. After ZA stimulation, BM MM Vγ9Vγ2 T cells further increased PD-1 (9) and TIM-3 expression ( Figure 1B ), while the increase in BM Ctrl Vγ9Vδ2 T cells was limited and significantly lower ( Figure 1B ). Cytofluorometric analysis from one representative MM shows that PD-1 and TIM-3 are co-expressed by approximately 60% of BM MM Vγ9Vδ2 T cells after ZA stimulation ( Figure 1C ). In freshly isolated Vγ9Vδ2 T cells we have previously shown that central memory (CM) Vγ9Vδ2 T cells display the highest PD-1 expression (9). After ZA stimulation, TIM-3 up-regulation was documented in all Vγ9Vδ2 T-cell subsets with CM and naïve Vγ9Vδ2 T cells showing slightly higher levels than effector memory (EM) and terminally differentiated effector memory (TEMRA) Vγ9Vδ2 T cells ( Figure 1D ). The gating strategy used to investigate TIM-3 expression in Vγ9Vδ2 T-cell subsets is shown in Supplemental Figure 2 .

Figure 2A compares the expression of immune senescence markers (33, 41, 42) in BM Ctrl and MM Vγ9Vδ2 T cells. BM MM Vγ9Vδ2 T cells showed significantly higher CD57 and CD160, and lower CD28 expression than BM Ctrl Vγ9Vδ2 T cells, even if differences were not statistically significant. The highest CD160 expression was observed in CM BM Vγ9Vδ2 T cells which is the cell subset with the highest PD-1 (9) and TIM-3 expression ( Figure 2B ). Cytofluorometric analysis of CD160 and PD-1 co-expression in BM MM Vγ9Vδ2 T cells from one representative sample is shown in Figure 2B (right panel).

Phosphorylated-γH2AX (p-γH2AX) is an early marker of DNA damage associated to immune senescence (43). p-γH2AX expression in BM Ctrl and MM-dia Vγ9Vδ2 T cells is shown in Figure 2C (one representative experiment) and Figure 2D (pooled data). These experiments were performed on purified γδ T cells. Both Vδ1 and Vγ9Vδ2 subsets can be represented in variable proportions in freshly purified γδ T cells (day 0), whereas after ZA stimulation Vγ9Vδ2 T cells become predominant ( Supplemental Figure 1 ) and any change should be referred to these because they are the only γδ T-cell subset sensitive to ZA stimulation. In freshly isolated BM γδ T cells, p-γH2AX expression was slightly higher in MM than Ctrl, but the difference was not statistically significant. After ZA stimulation, p-γH2AX expression was significantly increased in BM MM only ( Figures 2C, D ).

IL-7 has been reported to mitigate the induction of immune senescence of conventional T cells exposed to tumor cells (44, 45). We have investigated whether exogenous IL-7 could relieve the immune dysfunction of BM MM Vγ9Vδ2 T cells, but we have not observed any beneficial effect (data not shown).

Accumulating evidences indicate that Vγ9Vδ2 T cells can exert different functions depending on the local microenvironment, including the ability to promote tumor progression via the acquisition of regulatory or pro-tumoral functions (46). Figure 2E shows the expression of CD38, CD39, and CD73 in BM Vγ9Vδ2 T cells from Ctrl and MM patients. These molecules cooperate in the induction of the immune suppressive TME in MM via adenosine production (47). Only CD38 was significantly up-regulated in MM compared with Ctrl, whereas no differences were observed in CD39 and CD73 expression. The adenosine circuitry operated by CD38, CD39, and CD73 is well known to contribute to the establishment of the immune suppressive contexture in the TME of MM (47), but our data indicate that Vγ9Vδ2 T cells are not directly involved in this immune suppressive circuitry.

Lastly, BM MM Vγ9Vδ2 T cells did not show any phenotypic and/or functional features consistent with suppressor and/or pro-tumoral functions. The proliferation of CD4+ and CD8+ T cells after αCD3/αCD28 stimulation was similar in the presence or absence of γδ T cells ( Figure 2F ). Supplementary Figure 3A shows that proliferation of BM MM CD4+ and CD8+ cells was similar or even better compared with PB Ctrl CD4+ and CD8+ cells. Unlike BM Vγ9Vδ2 T cells, CD4+ and CD8+ cell proliferation was not influenced by the disease status ( Supplementary Figure 3B ), confirming the unique BM MM Vγ9Vδ2 T-cell susceptibility to the immune suppressive TME contexture.

The expression of PD-L1, GAL-9 and IL-17 characterizes Vγ9Vδ2 T cells with pro-tumoral functions in the TME (48). As shown in Figures 2G, H , the expression of GAL-9 and cytoplasmic IL-17 was similar in BM Ctrl and MM Vγ9Vδ2 T cells except for PD-L1 expression, which was slightly increased in the former, but the difference was not statistically significant. Representative dot plots of IL-17 expression in BM MM and Ctrl Vγ9Vδ2 T cells are shown in Supplemental Figure 4 .

ICP expression and immune senescence in T cells are associated with defective intracellular TCR signaling (49, 50). Figure 3A shows the expression of selected TCR-associated molecules in purified BM γδ T cells from one representative Ctrl and MM patient on day 0 and after ZA-stimulation (day 7). As reported above, both Vδ1 and Vγ9Vδ2 cells are represented in freshly purified γδ T cells (day 0), whereas Vγ9Vδ2 T cells are predominant on day 7 and they are the only γδ T-cell subset engaged by ZA ( Supplemental Figure 1 ). Pooled data are shown in Figure 3B showing that BM MM Vγ9Vδ2 T cells had significantly lower pAKT, higher PTEN, and lower pSTAT-1 expression on day 7 compared to BM Ctrl Vγ9Vδ2 T cells.

ZAP-70 and CD3-ζ chain are other TCR-associated molecules defectively expressed in T cells from the TME of mice and humans (51). ZAP-70 expression was significantly lower in resting BM MM Vγ9Vδ2 T cells compared with PB and BM Ctrl Vγ9Vδ2 T cells, but also with PB MM Vγ9Vδ2 T cells ( Figure 3C ), further confirming the striking difference between circulating vs TME-resident Vγ9Vδ2 T cells. Representative dot plots are shown in Figure 3D . Paired analyses of Vγ9Vδ2+ and CD3+ Vγ9Vδ2- cells showed that the mean ZAP-70 expression was also significantly down-regulated in BM CD3+ Vγ9Vδ2- T cells of MM patients with a wide range of expression in individual samples ( Supplemental Figure 5 ). A slight increase was observed after ZA stimulation in Vγ9Vδ2 T cells from 3 MM patients with low ZAP-70 expression at baseline, but values remained inferior to Ctrl values ( Figure 3E ). Unlike ZAP-70, the proportion and MFI of CD3-ζ chain expression were not different in PB and BM Ctrl and MM Vγ9Vδ2 T cells ( Supplemental Figure 6 ).

It has been reported that TIM-3 up-regulation is involved in the acquired resistance to PD-1 blockade (31, 52, 53). Thus, we have investigated whether TIM-3 was involved in the incomplete recovery of BM MM Vγ9Vδ2 T cells after ZA stimulation and single PD-1 blockade. Figure 4A shows that both TIM-3 expression and MFI values were significantly up-regulated in BM MM Vγ9Vδ2 T cells in the presence of αPD-1, whereas PD-1 expression was slightly down-regulated after ZA stimulation in the presence of αTIM-3, but the decrease was not statistically significant. Representative cytofluorometric analyses of increased TIM-3 up-regulation and PD-1 down-regulation are shown in Figure 4A (right panel) and Figure 4B (right panel).

Next, we investigated whether dual PD1-/TIM-3 blockade was more effective than single blockade. We evaluated the proliferation ( Figure 4C ), IFN-γ production ( Figure 4D ) and CD107 expression ( Figure 4E ) in BM MM Vγ9Vδ2 T cells after ZA stimulation in the presence of αPD-1, αTIM-3, and the combination thereof. Representative cytofluorometric analyses of increased IFN-γ and CD107 expression in BM MM Vγ9Vδ2 T cells after dual blockade are shown in Figure 4D (right panel) and Figure 4E (right panel). Our results indicate that dual blockade PD-1/TIM-3 blockade is more effective than single PD-1 or TIM-3 blockade in MM-dia to mitigate BM MM Vγ9Vδ2 T-cell dysfunctions.

Dual PD-1/TIM-3 blockade was also associated with a partial recovery of TCR-associated alterations. Data from one representative Ctrl and MM are shown in Figure 5A , while pooled data from 2 paired experiments are shown in Figure 5B . αPD-1 partially normalized pAKT and PTEN expression, whereas αTIM-3 partially normalized pJAK1 and pSTAT1 expression. No antagonist, additive or synergistic effect was observed suggesting that αPD-1 and αTIM-3 target mutually exclusive TCR-associated molecules in BM MM Vγ9Vδ2 T cells. Supplementary Figure 8 shows pooled data from unpaired experiments after αPD-1 treatment only.

Next, we looked for possible intersections between the intracellular pathways triggered by αPD-1 and αTIM-3. Previous work from Zhu C. et al. (54) has reported a cross-talk between TIM-3 and PD-1 mediated by the IL-27/pSTAT1/T-bet axis. BM MM Vγ9Vδ2 T cells showed the lowest T-bet ( Figure 6A ), and the highest IL-27R expression ( Figure 6B ). This pattern has recently been reported in severely exhausted T cells from the BM of patients with AML in relapse after allogeneic transplantation (55). αPD-1 treatment did not increase T-bet and/or IL-27R expression in ZA-stimulated BM Vγ9Vδ2 T cells ( Figures 6C, D ). Moreover, low IL-27 levels were detected in the supernatants of BM MM Vγ9Vδ2 T cells which were not modified by αPD-1 ( Figure 6E ).

The PI3K/Akt axis is another intracellular signalling pathway connecting PD-1 and TIM-3 in tumor-infiltrating lymphocytes from patients with head and neck cancer (53). In these cells, TIM-3 up-regulation induced by αPD-1 can be abrogated with LY294002, a broad PI3K inhibitor (53). Thus, we evaluated whether αPD-1-induced TIM-3 up-regulation in BM MM Vγ9Vδ2 T cells could be inhibited by single pSTAT-1 inhibition with fludarabine monophosphate (FAMP) (56), single PI3K inhibition with LY294002, or the combination thereof. Results shown in Figure 6F indicate that these pathways are not druggable to prevent αPD-1-induced TIM-3 up-regulation in BM MM Vγ9Vδ2 T cells.

Next, we investigated whether the ICP/ICP-L immune suppressive circuitry was influenced by the disease status. PD-1 expression was significantly higher in MM-rel than in MM-dia, while MM-rem showed intermediate values. By contrast, no differences were observed in TIM-3 expression between MM-dia, MM-rem, and MM rel ( Figure 7A ). We investigated whether αPD-1 treatment induced TIM-3 up-regulation also in MM-rem and MM-rel. Figure 7B shows that TIM-3 was up-regulated in MM-dia only, but not in MM-rem and MM-rel.

The effect of single or dual PD-1/TIM-3 blockade on ZA-induced proliferation in BM MM Vγ9Vδ2 T cells in MM-dia, MM-rem and MM-rel is shown in Figure 7C . BM Vγ9Vδ2 T cells from MM-rem were the only ones to reach normal proliferation values with single PD-1 or TIM-3 blockade, the former being slightly more effective than the latter. Dual PD-1/TIM-3 blockade was not superior to single blockade in MM-rem. By contrast dual PD-1/TIM-3 blockade was more effective than single blockade in MM-dia, the only clinical setting in which αPD-1 induces TIM-3 up-regulation. BM Vγ9Vδ2 T cells from MM-rel showed the worst anergy to single and dual blockade, even if TIM-3 expression was similar to MM-dia and MM-rem and was not up-regulated by αPD-1 ( Figures 7A, B ).

These findings prompted us to investigate the expression of additional ICP on BM MM Vγ9Vδ2 T cells in MM-rel. Figure 7D shows that LAG-3 expression was similar in resting (day 0) BM Vγ9Vδ2 T cells from MM-dia, MM-rem, and MM-rel. After ZA stimulation, LAG-3 expression was slightly increased in MM-dia, unmodified in MM-rem, and increased in MM-rel, even if the differences was not statistically significant. Next, we determined which PD-1/TIM-3/LAG-3 combination was more effective to mitigate the anergy of BM Vγ9Vδ2 T cells in MM-rel. Results shown in Figure 7E indicate that dual PD-1/LAG-3 blockade was more effective than dual PD-1/TIM-3, dual TIM-3/LAG-3, and even triple PD-1/TIM-3/LAG-3 blockade, but still inferior to that reached in MM-rem after single PD-1 or TIM-3 blockade, or MM-dia after dual PD-1/TIM-3 blockade.

These data confirm that the relapse is the most challenging setting, and immune-based strategies should be delivered in remission, when the immune suppressive TME commitment is partially relieved.