BTN3A molecules considerably improve Vγ9Vδ2T cells-based immunotherapy in acute myeloid leukemia

ABSTRACT

Given their recognized ability to kill acute myeloid leukemia (AML) blasts both in vitro and in vivo, Vγ9Vδ2 T cells are of growing interest in the design of new strategies of immunotherapy. We show that the Butyrophilin3A (BTN3A, CD277) subfamily is a critical determinant of Vγ9Vδ2 TCR-mediated recognition of human primary AML blasts ex vivo. Moreover, anti-BTN3A 20.1 agonist monoclonal antibodies (mAbs) can trigger BTN3A on AML blasts leading to further enhanced Vγ9Vδ2 T cell-mediated killing, but this mAb had no enhancing effect upon NK cell-mediated killing. We show that monocytic differentiation of primary AML blasts accounts for their AminoBisphosphonate (N-BP)-mediated sensitization to Vγ9Vδ2 T cells. In addition, anti-BTN3A 20.1 mAbs could specifically sensitize resistant blasts to Vγ9Vδ2 T cells lysis and overcome the poor effect of N-BP treatment on those blasts. We confirmed the enhancement of Vγ9Vδ2 T cells activity by anti-BTN3A 20.1 mAb using a human AML xenotransplantation mouse model. We showed that anti-BTN3A 20.1 mAb combined with Vγ9Vδ2 T cells immunotherapy could increase animal survival and decrease the leukemic burden in blood and bone marrow. These findings could be of great interest in the design of new immunotherapeutic strategies for treating AML.

KEYWORDS:

• Acute myeloid leukemia
• aminobisphosphonate
• BTN3A
• immunotherapy
• monoclonal antibody
• γδT cells

Abbreviations

Ab	=	Antibody
ADCC	=	Antibody-dependent cell cytotoxicity
AHSCT	=	Allogeneic hematopoietic stem cell transplantation
AML	=	Acute myeloid leukemia
BM	=	Bone marrow
BrHpp	=	Bromohydrine pyrophosphate
BTN	=	Butyrophilin
Chr	=	chromosome
DNAM-1	=	DNAX accessory molecule-1
GvHD	=	Graft-versus-host-disease
HV	=	Healthy volunteer
Ig	=	Immunoglobulin
IPP	=	Isopentenyl diPhosphate
mAb	=	Monoclonal antibody
MHC	=	Major histocompatibility complex
N-BP	=	Nitrogen-containing bisphosphonates
NKG2D	=	Natural killer group 2 member D
NSG	=	NOD-SCID-common-γ-chain-knockout mice
pAg	=	phosphoAntigen
RLI	=	Receptor-linked IL-15
TCR	=	T cell receptor
UPN	=	Unique patient number
ZOL	=	Zoledronate

Introduction

Allogeneic haematopoietic stem cell transplantation (AHSCT) has highlighted a major role for the immune system in the control of residual AML.¹ AHSCT still remains the most effective therapy for most AML, though its use is still restricted to selected cases and outcomes are limited by both relapse and graft-versus-host-disease (GvHD). New approaches including the development of immunotherapy strategies are being actively investigated that may improve outcomes in AHSCT. Innate immune cells, notably NK cells and γδ T cells, are of particular interest on account of their ability to kill hematological^2,3 or solid tumor cells⁴ without inducing GvHD.^5,6

The anti-leukemic role of γδ T cells was suggested by a correlation between Vδ1 T cells expansion and prognosis following AHSCT.^7,8 Aside from Vδ1 T cells, Vγ9Vδ2 T cells, the main circulating subset, account for 1–10% of human T lymphocytes. Their T cell receptors (TCR) sense non-peptidic-phosphorylated phosphoAntigens (pAgs), such as isopentenyl diphosphate (IPP), an intermediary metabolite of the human mevalonate pathway.⁹ In malignant cells, the blockade of this pathway with pharmacological agents such as N-BPs leads to IPP accumulation that triggers Vγ9Vδ2 T cells MHC-unrestricted functions.^10,11

Furthermore, we have recently shown that Vγ9Vδ2 T cells can specifically recognize and kill AML blasts in vitro and in a xenograft model and that patients with active AML have lower counts of Vγ9Vδ2 T cells.¹² Those Vγ9Vδ2 T cells have reduced expansion potential and a differentiation profile biased toward an effector memory phenotype. This phenotype can be reversed in vitro after blasts removal or after patients achieve remission,¹² supporting a model of blast-induced impairment of Vγ9Vδ2 T cells differentiation and function. Attempts to bypass this mechanism of escape to Vγ9Vδ2 T cells rely on TCR direct agonists, including bromohydrin pyrophosphate (BrHpp), a synthetic pAg, or indirect agonists, such as zoledronate (ZOL), an N-BP.¹⁰ The anti-leukemic effect of TCR agonists was supported by a trial showing partial remission in two AML patients receiving N-BP with IL-2.¹³ As Vγ9Vδ2 T cells are altered in patients,^12,14 the adoptive transfer of these cells emerges as a promising alternative.¹⁵ However, 50% of N-BP-treated blasts still failed to enhance allogeneic γδ T cells cytotoxicity in vitro, suggesting that novel-triggering agents are still warranted.¹⁶

Among the B7-related family, the butyrophilin3A (BTN3A) subfamily comprises three isoforms (BTN3A1, BTN3A2 and BTN3A3)¹⁷ that contain two extracellular immunoglobulin (Ig) domains (IgV and IgC) and a transmembrane domain. Only BTN3A1 and BTN3A3 contain a B30.2 cytoplasmic domain. BTN3A members are broadly expressed, notably by most immune cells and various tumors.^17,18 The anti-BTN3A 20.1 agonist monoclonal antibody (mAb) recognizes all three isoforms. Recently, the requirement of BTN3A molecules in pAg-sensing by Vγ9Vδ2 T cells was demonstrated.¹⁹ Anti-BTN3A 20.1 mAb acts indirectly, through sensitization of target cells to Vγ9Vδ2 T cells recognition.²⁰ This mAb induced the immobilization of BTN3A on the target cell surface, suggesting an involvement of the extracellular domain in Vγ9Vδ2 T cells activation. However, only the B30.2 domain of the BTN3A1 isoform can bind pAg²¹ through a positively charged pocket.^22,23 This intracellular binding induced extracellular conformational changes of BTN3A, similar to those observed with 20.1 mAb and resulting Vγ9Vδ2 T cell activation.²⁴

Together, these data open new perspectives in Vγ9Vδ2 T cells-based immunotherapies, with BTN3A appearing as an interesting target to regulate their functions. In this study, we decided to investigate how BTN3A molecules might contribute to AML cells clearance through Vγ9Vδ2 T cells-based immunotherapy.

Results

BTN3A specifically enhance Vg9Vd2 TCR-mediated lysis of primary AML blasts

BTN3A molecules are key players in Vγ9Vδ2 TCR activation.^19,20 Hence, we decided to assess whether anti-BTN3A 20.1 mAb impact on the recognition and killing of primary AML blasts by Vγ9Vδ2 T cells.

We first compared the effect of Vγ9Vδ2 TCR agonists such as BrHpp and ZOL to anti-BTN3A agonist 20.1 mAb. We observed that, for optimal stimulation conditions, ZOL and anti-BTN3A 20.1 mAb were significantly more efficient (p = 0.0170 and p < 0.0001, respectively) than BrHpp to enhance Vγ9Vδ2 T cells lysis of AML blasts (Fig. 1A).

Figure 1. BTN3A triggering enhances TCR-mediated lysis of AML blasts by Vγ9Vδ2 T cells. AML blasts were pre-incubated for 1 h with Control Isotype, anti-BTN3A 20.1 agonist mAb (10 μg/mL) or O/N with Zoledronate (ZOL) (45 μM), before addition of effector cells. Each AML blast sample was tested with at least three different Vγ9Vδ2 T cells or NK cells donors. (A) Comparison of TCR agonists. Specific lysis of primary AML blasts (n = 25) by expanded Vγ9Vδ2 T cells from HV after treatment with BrHpp, ZOL, or anti-BTN3A 20.1 mAb. Data are shown for E:T 30:1. (B, C) Effect of TCR and BTN3A targeting. Primary AML blasts lysis (n = 11) by expanded Vγ9Vδ2 T cells from HV. (B) Vγ9Vδ2 T cells were pre-incubated for 30 min with anti-TCRγ9 mAb. The blasts were treated with ZOL or anti-BTN3A agonist 20.1 mAb. (C) The blasts were treated with ZOL, agonist (20.1) or antagonist (108.5) anti-BTN3A mAb or combination and then incubated with Vγ9Vδ2 T cells. Data are shown for E:T 30:1. The lysis is expressed as percentage of specific lysis. (D, E) Effect of anti-BTN3A mAb and ZOL on NK cells lysis compared to Vγ9Vδ2 T cells. Representative data from two independent experiments. Primary AML blasts lysis (n = 3) by (D) expanded Vγ9Vδ2 T cells from HV; (E) NK cells from HV isolated and treated O/N with IL2; DAUDI and K562 were used as positive controls respectively for Vγ9Vδ2 T cells and NK cells lysis. The lysis is expressed as percentage of specific lysis. Results were expressed as mean ± SEM and statistical significance was established using the non-parametric paired Wilcoxon U-test. *p < 0.05; **0.001 < p < 0.01; ***p < 0.001.

Secondly, to dissect the role of BTN3A in Vγ9Vδ2 T cells lysis of AML blasts, we performed cytotoxicity assays involving different combinations of TCR Vγ9 blocking mAb, ZOL and anti-BTN3A mAbs with agonist (20.1 mAb) or antagonist (108.5 mAb) functions. The TCR blockade resulted in significant decreases of ZOL (p = 0.0156) and anti-BTN3A 20.1-induced lysis of AML blasts ( p = 0.0156) (Fig. 1B). This demonstrated that Vγ9Vδ2 TCR-induced lysis of AML blasts could be enhanced using agonists such as anti-BTN3A 20.1 mAb.

In this line, as both ZOL and BTN3A exerted strong agonist functions, we combined them in cytotoxic assays against AML blasts. The addition of anti-BTN3A 20.1 mAb to ZOL resulted in a poor but significant ( p = 0.0420) enhancement of AML blasts lysis by Vγ9Vδ2 T cells (Fig. 1C). Conversely, the antagonist anti-BTN3A 108.5 mAb could significantly and drastically abrogate ZOL-induced lysis ( p = 0.0156) (Fig. 1C).The lack of additivity or synergic effect of ZOL and agonist 20.1 mAb-combined stimulation suggested that they had overlapping functions through BTN3A molecules to mediate the enhancement of lysis. In addition, we confirmed BTN3A key function in the recognition of N-BP-treated blasts by Vγ9Vδ2 T cells with anti-BTN3A antagonist 108.5 mAb that totally abrogated the ZOL-induced activation of Vγ9Vδ2 T cells.

In order to test whether anti-BTN3A 20.1 mAb and ZOL effects were specifically related to Vγ9Vδ2 TCR activation, we assessed their effects on NK cells-mediated AML non-MHC-restricted blasts lysis. Unlike Vγ9Vδ2 T cells (Fig. 1D), anti-BTN3A 20.1 mAb and ZOL had no effect upon NK cells-mediated AML blasts lysis (Fig. 1E). Collectively, these observations show that TCR agonist-mediated enhancement of Vγ9Vδ2 T cells lysis of AML blasts specifically involves BTN3A molecules.

Triggering BTN3A with agonist 20.1 mAb enhances Vγ9Vδ2 T cells antitumor functions

Treatment with BrHpp has been shown to enhance the contact-dependent capture of AML cell membrane via trogocytosis¹² leading us to hypothesize that the anti-BTN3A 20.1 agonist mAb might induce a similar effect. As shown in Figs. 2A and B, treatment with anti-BTN3A 20.1 mAb significantly enhanced the trogocytosis of AML blasts by Vγ9Vδ2 T cells (p = 0.0039). Finally, we observed that when Vγ9Vδ2 T cells were co-cultured with AML blasts in the presence of anti-BTN3A 20.1 agonist mAb, they showed greater cytokine production of TNFα (p = 0.0313) and IFNγ (p = 0.0313) and a greater ability to degranulate as assessed by the expression of CD107a (p = 0.0313) (Fig. 2C). Collectively, these data demonstrated a potent role of anti-BTN3A 20.1 mAb to enhance the recognition of resistant AML by Vγ9Vδ2 T cells.

Figure 2. Triggering of BTN3A with 20.1 mAb enhances Vγ9Vδ2 T cells antitumor functions toward AML blasts. (A) Representative experiment showing trogocytosis of AML blasts from patient UPN01 by Vγ9Vδ2 T cells from HV (full gray histogram) and after 1 h in co-culture with IgG1 (dashed line) or anti BTN3A 20.1 mAb (full line). (B) Vγ9Vδ2 T cells trogocytosis (PKH67 MFI ± SEM) of AML blasts (n = 10). Cumulative data at 0 and 60 min from three different experiments are provided. (C) CD107a expression, TNFα and IFNγ production by Vγ9Vδ2 T cells from HV after a 4-h co-culture with AML blasts (n = 6).

Agonist BTN3A 20.1 mAb enhances Vγ9Vδ2 T cells killing through fixation on AML blasts

Up to now, we have demonstrated that BTN3A agonist mAb enhanced Vγ9Vδ2 T cells antitumor functions. However, as BTN3A molecules are expressed both on Vγ9Vδ2 T cells and blasts cells, we do not know if the agonist molecules precisely exert their effects on the target, the effector cells or both.

We thus examined whether sensitization to killing occurred through the interaction of anti-BTN3A 20.1 mAb with AML blasts themselves, or rather as a consequence of anti-BTN3A 20.1 mAb exerting its effects upon Vγ9Vδ2 T cells. We compared the lysis of AML blasts either (i) after blasts pre-incubation with anti-BTN3A 20.1 mAb (washed before co-incubation with Vγ9Vδ2 T cells), or (ii) after simultaneous incubation of blasts, anti-BTN3A 20.1 mAb and Vγ9Vδ2 T cells, or (iii) after Vγ9Vδ2 T cells were previously incubated with anti-BTN3A 20.1 mAb (then washed before co-incubation with AML blasts). The lysis achieved with Vγ9Vδ2 T cells pre-incubated with anti-BTN3A 20.1 mAb was comparable to basal lysis (4.36 ± 2.11%). The lysis was enhanced when blasts were pre-incubated with anti-BTN3A 20.1 mAb (22.92 ± 7.64%) and further enhanced by simultaneous incubation of effector cells, target cells and mAb (35.95 ± 1.16%) (Fig. 3). Those results show that anti-BTN3A 20.1 mAb biological effect is upon AML blasts rather than directly upon BTN3A molecules expressed by Vγ9Vδ2 T cells. The anti-BTN3A 20.1 mAb agonist effect was even more pronounced when effectors, targets and mAb were simultaneously present.

Figure 3. The 20.1 mAb triggers BTN3A on the surface of AML blasts. Primary AML blasts, target (T) lysis (n = 2) by Effector (E) Vγ9Vδ2 T cells from HV (n = 2) with IgG1 (full black line) or (i) after T pre-incubation with anti-BTN3A 20.1 mAb (anti-BTN3A + T) (full gray line) or (ii) after simultaneous incubation of E, T and anti-BTN3A 20.1 mAb (anti-BTN3A + E + T) (dashed gray line), or (iii) after E pre-incubation with anti-BTN3A 20.1 mAb (anti-BTN3A + E) (dashed black line). T or E cells were preincubated for 90 min with anti-BTN3A 20.1 mAb (10 μg/mL), and then extensively washed before E + T co-culture. Representative data from two independent experiments are depicted. Data are shown for E:T 1:1, 10:1, 30:1. The lysis is expressed as a mean percentage of specific lysis ± SEM and compared to DAUDI lysis.

Among BTN3A molecules, primary AML blasts mainly express BTN3A2

As we know that BTN3A subfamily gathers three different isoforms, we decided to describe the expression of each one of them on primary AML blasts. First of all, BTN3A was expressed at the surface of all the primary AML blasts tested (n = 25), all states of differentiation confounded (Fig. 4A, Table S1). We assessed the level of expression of each BTN3A isoform at the transcriptional level by qRT-PCR and protein level by Western Blot. Compared to BTN3A1 (Mean ΔCT = 11.12 ± 0.34) and BTN3A3 (Mean ΔCT = 11.73 ± 0.29), BTN3A2 was the most abundant isoform both at transcriptional (Mean ΔCT = 7.67 ± 0.38; p <0.0001) (Fig. 4B) and protein levels in all primary AML blasts (Fig. 4C). BTN3A1 and BTN3A3 were expressed at the transcriptional level (Fig. 4B) but were not detected at the expected size at the protein level (Fig. 4C). We demonstrated for the first time that blasts mainly express the BTN3A2 isoform, i.e. the isoform without the B30.2 cytoplasmic domain.

Figure 4. BTN3A molecules expression on primary AML blasts. (A) BTN3A molecules surface expression on primary AML blasts (n = 25). The global BTN3A surface expression was assessed by flow cytometry analysis. The results are expressed as Median Fluorescence Intensity ratio (rMFI) compared to control isotype. (B, C) BTN3A isoform expression on primary AML blasts. (B) qRT PCR analysis of total RNA isolated from primary AML blasts (n = 24). Data were normalized using GAPDH as endogenous control (ΔCt = Ct_{target gene} – Ct_GAPDH). Statistical significance was established using the non-parametric paired Wilcoxon U-test. ***p < 0.001. (C) Western-blot analysis of total protein extracts from primary AML blasts (n = 3). BTN3A knock-down HEK293FT (sh#284; clone#30) transiently transfected with BTN3A1m, BTN3A2m, BTN3A3m are used as size controls. Extracts were loaded in 10% SDS PAGE gel and membranes were hybridized with anti-BTN3A 20.1 mAb and anti-Grb2 as a loading control.

The sensitization of AML blasts by N-BP depends on monocytic differentiation of blasts

Finally, Gunderman et al. have demonstrated that AML samples with poor monocytic differentiation were poorly responsive to Vγ9Vδ2 T cells lysis despite ZOL treatment.¹⁶ We thus assessed the level of monocytic differentiation in our samples by determining the percentage of monocytic component defined based on the SSC parameter and the level of expression of CD45 (Fig. 5A and B). First, we confirmed that monocytic differentiation positively correlated with N-BP sensitization of AML blasts (r_S = 0.8; p < 0.0001) (Fig. 5C). Then, based on the percentage of monocytic component, we segregated two groups of AML blasts corresponding to monocytic^Low AML with a monocytic component strictly below 10% (Fig. 5A) and monocytic^High AML with a monocytic component greater than 10% (Fig. 5B). The potentiation of Vγ9Vδ2 T cells lysis by N-BP-treated blasts was significantly higher in monocytic^High AML (38.25 ± 3.76%) than in monocytic^Low AML (20.53 ± 3.24%; p = 0.0029) (Fig. 5D). Altogether, these data show that monocytic component accounts for N-BP sensitivity of primary AML blasts.

Figure 5. The sensitization of AML blasts by N-BP depends on monocytic component of blasts. (A, B) The percentage of monocytic component among AML blasts (SSC^High CD45^High) (n = 20) was assessed by flow cytometry. Representative dot plot of: (A) monocytic^Low AML (UPN02) and (B) monocytic^High AML (UPN10). (C) Linear regression of Vγ9Vδ2 T cells lysis of N-BP-treated AML blasts with respect to monocytic component (n = 20). The correlation was established using the non-parametric Spearman correlation coefficient (r_s). ***p < 0.001. (D) Lysis of N-BP-treated blasts with low monocytic component (monocytic^Low AML, % of SSC^High CD45^High<10) (n = 10) compared with the lysis of N-BP-treated blasts with high monocytic component (monocytic^High AML, % of SSC^High CD45^High≥10) (n = 10) by Vγ9Vδ2 T cells. Results were expressed as mean ± SEM. Statistical significance was established using the Mann–Whitney test. 0.001 < **p < 0.01.

Anti-BTN3A 20.1 mAb can overcome the N-BP poor sensitization of monocytic^Low AML

We next determined the extent to which monocytic^Low AML blasts that were either resistant or poorly responsive to N-BP sensitization could be sensitized by anti-BTN3A 20.1 mAb. We thus decided to compare the effect of ZOL and anti-BTN3A 20.1 mAb upon monocytic^Low and monocytic^High AML samples. As shown in Figs. 6A and B, anti-BTN3A 20.1 mAb were able to enhance Vγ9Vδ2 T cell lysis of both monocytic^Low and monocytic^High AML blasts (p = 0.002). Anti-BTN3A 20.1 mAb enhancing effect was significantly higher than ZOL regarding Vγ9Vδ2 T cell lysis of ZOL-resistant or ZOL-poorly responsive monocytic^Low AML blasts (p = 0.002) (Fig. 6A). Anti-BTN3A 20.1 mAb and ZOL had a comparable enhancing effect (37.23 ± 3% and 38.25 ± 3.8%, respectively) upon monocytic^High AML blasts (Fig. 6B).

Figure 6. Anti-BTN3A 20.1 mAb can overcome N-BP poor sensitization of monocytic^Low AML. (A) Specific lysis of monocytic^Low AML blasts (n = 10) and (B) monocytic^High AML blasts (n = 10) by Vγ9Vδ2 T cells from HV after treatment with IgG1, ZOL (O/N, 45 μM), or anti-BTN3A 20.1 mAb. Data are shown for E:T 30:1. Results were expressed as mean ± SEM and statistical significance was established using the non-parametric paired Wilcoxon U-test. *p < 0.05; 0.001 <**p  <  0.01; ***p < 0.001.

Altogether, these data converge to show that BTN3A triggering with the agonist 20.1 mAb can significantly enhance the lysis of AML blasts by Vγ9Vδ2 T cells, thus overcoming their resistance to N-BP treatment.

Anti-BTN3A mAb combined with γδ T cells infusion decreases leukemic burden in AML xenografted mice models

We next asked whether the in vitro potentiating effects of anti-BTN3A 20.1 mAb against leukemia cells could be reproduced in vivo.

We first used a validated AML xenograft model based on the inoculation of luciferase-transduced-U937 cell line, to monitor the engraftment of malignant cells by bioluminescent imaging.¹²

We used RLI as an antitumor adjuvant tool combined with Vγ9Vδ2 T cells and anti-BTN3A 20.1 mAb infusions. RLI is an IL-15/IL15Rα fusion protein that has been shown to expand antitumor lymphocyte subsets including CD8⁺ and NK cells and to be an adjuvant to immune responses against murine and human cancers in vivo.^25,26 In this line, we first assessed whether RLI could boost Vγ9Vδ2 T cells proliferation and demonstrated for the first time that RLI could drive Vγ9Vδ2 T cells' expansion in vitro (Figs. S1A and B) and in vivo (Figs. S1C and D).

We confirmed that Vγ9Vδ2 T cells plus-RLI infusions decrease the tumor burden. This effect was even more striking and was associated with a significant increase of survival when anti-BTN3A 20.1 mAb was added along with Vγ9Vδ2 T cells (p = 0.0224) (Figs. 7A and B).

Figure 7. Evaluation of anti-BTN3A agonist mAb anti-leukemic activity in vivo in AML xenograft models. (A) (B) NSG mice (n = 15) were inoculated with luciferase-transfected U937 cells on Day 0. They were intravenously injected with PBS, anti-BTN3A mAb alone, RLI alone or 30 × 10⁶ Vγ9Vδ2 T cells-plus-RLI or 30 × 10⁶ Vγ9Vδ2 T cells-plus-RLI and anti-BTN3A 20.1 mAb. (A) Representative and comparative bioluminescent imaging of five groups of mice on Day 12 post-graft. Day 12 corresponds to the last bioluminescent measurement before first mice deceased. Medians of survival of each group of mice are notified below. (B) Median survival time was 21 d for the PBS control group (n = 3), 25 d for the group treated with Vγ9Vδ2 T cells and RLI (n = 3) and 29 d for the group treated with Vγ9Vδ2 T cells, RLI and anti-BTN3A 20.1 mAb (n = 3). Median survival times were compared using Log-Rank (Mantel-Cox) Test. *p < 0.05. (C, D) NSG mice (three groups of n = 6) were intravenously injected with primary AML blasts. Since the Week 8 post-graft, they were treated twice a week during two weeks with injections of 15 × 10⁶ Vγ9Vδ2 T cells-plus RLI and either anti-BTN3A mAb or IgG1. Control mice were injected with PBS. Mice were sacrificed at the Week 11 post-graft. (C) Percentage of variation of the absolute number of blasts in blood quantified by Flow Cytometry before and 4 d after the end of the treatment. (D) Absolute number of blastic cells quantified by Flow Cytometry in harvested BM. Representative and cumulative data from two independent experiments. Mann–Whitney test was used to compare differences between treatments. *p < 0.05; 0.001 < **p < 0.01; ***p < 0.001.

We next confirmed our data in a xenotransplantation model using primary AML blasts.²⁷ Blasts from UPN09 were highly sensitized to Vγ9Vδ2 T cells lysis after anti-BTN3A 20.1 mAb treatment in vitro (Fig. S2A). Blasts from UPN09 highly engrafted in BM and blood of NSG mice, thus recapitulating human disease (Fig. S2B). We asked whether anti-BTN3A 20.1 mAb and Vγ9Vδ2 T cells-immunotherapy could decrease the disease progression in this model. We showed a significant decrease of circulating blasts in the anti-BTN3A 20.1 mAb treated group (37.38 ± 39.89) compared to IgG1(388.1 ± 212.3; p = 0.0379) and untreated group (1208 ± 817; p = 0.0022) (Fig. 7C). Similarly, anti-BTN3A 20.1 mAb treated group had a significantly lower absolute number of blasts in the BM (43.4 ± 39.28) than the IgG1 (161.5 ± 55.72; p = 0.0411) and untreated groups (7715 ± 1819; p = 0.022) (Fig. 7D).

These data highlight the potent anti-leukemic effect exerted by anti-BTN3A 20.1 mAb combined to Vγ9Vδ2 T cells immunotherapy in vivo.

Discussion

In this study, we show that TCR activation by agonists such as N-BP involves BTN3A molecules and that agonist anti-BTN3A 20.1 mAb triggers BTN3A on blasts to specifically enhance Vγ9Vδ2 T cell-mediated lysis. Finally, we demonstrate the enhancement of Vγ9Vδ2 T cell anti-leukemic effects, using anti-BTN3A 20.1 mAb in vivo, in xenotransplantation models.

We demonstrated in this study the critical role of BTN3A subfamily in N-BP-mediated sensitization of blasts to Vγ9Vδ2 T cells lysis, using primary AML samples from patients. Indeed, treatment with the newly described antagonist anti-BTN3A 108.5 mAb totally abrogated Vγ9Vδ2 T cells lysis of N-BP treated blasts.

Conversely, triggering of BTN3A with the agonist 20.1 mAb induced the TCR-mediated activation of Vγ9Vδ2 T cells. This suggests that BTN3A molecules could interact with Vγ9Vδ2 TCR and the literature rather favors an indirect engagement of the TCR mediated by BTN3A²². BTN3A 20.1 mAb and N-BP-combined treatment of AML blasts did not provide a synergic or an additive enhancing effect upon Vγ9Vδ2 T cells lysis of blasts. This is consistent with the hypothesis of overlapping effects between N-BP and anti-BTN3A 20.1 mAb. In this line, biochemical, biophysical and FRAP (fluorescence recovery after photobleaching) experiments have shown that pAg and BTN3A 20.1 mAb induced the same conformational activating changes of BTN3A molecules.^24,20

We confirmed anti-BTN3A 20.1 mAb's indirect effect, mainly on primary AML blasts.²⁰ But, interestingly, the simultaneous incubation of Vγ9Vδ2 T cells with anti-BTN3A 20.1 mAb and blasts resulted in even more enhanced cytotoxicity. This suggests a concomitant triggering of BTN3A by anti-BTN3A 20.1 mAb on both effector and target cells, which could be of interest in the perspective of a novel immunotherapy. In addition, as previously demonstrated,²⁰ this observed effect is probably not ADCC-dependent as anti-BTN3A 20.1 mAb did not have any biological effect on NK cells lysis of AML blasts.

Although N-BP could be a promising therapy in hematologic malignancies,^13,28 our data confirm that some primary AML blasts with poor monocytic differentiation are poorly sensitized by N-BP to Vγ9Vδ2 T cells lysis.¹⁶ These monocytic^Low AML were shown to have poorly active mevalonate pathway.¹⁶ This could account for a poor pAg accumulation in these blasts upon N-BP treatment that would fail to activate Vγ9Vδ2 T cells.²⁹ We show that anti-BTN3A 20.1 mAb can sensitize N-BP-poorly sensitized primary AML blasts to Vγ9Vδ2T cells lysis, notably monocytic^Low AML. The sensitization of these resistant blasts to Vγ9Vδ2 T cells lysis with anti-BTN3A 20.1 mAb points out a potential major role of this mAb to circumvent intrinsic resistance mechanisms in blasts. In addition, this observation underlies that different ways are involved in anti-BTN3A 20.1 mAb and N-BP-mediated activation. Indeed, whereas N-BP-mediated activation of Vγ9Vδ2 T cells requires the expression of the three BTN3A isoforms¹⁹ and putative additional partners,³⁰ anti-BTN3A 20.1 mAb can trigger each one of the three BTN3A isoforms.²⁰

BTN3A2 was the most expressed isoform in primary AML blasts. As BTN3A2 lacks the intracellular B30.2 domain, it has been hypothesized to be a decoy receptor via the binding of a putative ligand.³¹ Its predominant expression by AML blasts could constitute an immune escape mechanism to Vγ9Vδ2 T cells recognition. However, we assume that anti-BTN3A 20.1 mAb could by-pass N-BP resistance via the triggering of BTN3A2, through mechanisms involving decreased mobility²⁰ and a multimerization of BTN3A molecules.²⁴ These 20.1 mAb-induced extracellular conformational changes of BTN3A molecules might require transmembrane or intracellular partners such as cytolinker proteins.¹⁹ In addition, as BTN3A molecules might form heterodimeric complexes,²⁴ the 20.1 mAb binding to BTN3A2 might recruit BTN3A1 and BTN3A3 and transduce an intracellular signaling through their intracellular domain.

In the context of developing novel immunotherapies, we have adopted a preclinical xenograft mouse model employing human leukemia cells. We show that the addition of anti-BTN3A 20.1 mAb to allogeneic Vγ9Vδ2 T cells and cytokine infusions could decrease leukemic burden and increase the survival probability of U937-xenografted NSG mice. Hence, we confirmed our data in a more clinically applicable model consisting in the xenotransplantation of primary AML blast in NSG mice.²⁷ In this model, anti-BTN3A 20.1 mAb combined to Vγ9Vδ2 T cells-plus cytokines could decrease the leukemic progression in BM and blood. Accordingly, these models highlight the potential utility of anti-BTN3A 20.1 mAb in Vγ9Vδ2 T cell-based immunotherapy of AML.

In conclusion, our data demonstrate that Vγ9Vδ2 T cells lysis of primary AML blasts involves TCR and BTN3A-mediated activation. We showed that agonist anti-BTN3A 20.1 mAb can circumvent primary AML blasts resistance to allogeneic Vγ9Vδ2 T cells lysis. Anti-BTN3A 20.1 mAb effect is confirmed in vivo, using a NSG human AML-xenografted mice model. This study highlights the potential therapeutic utility of anti-BTN3A 20.1 mAb, combined with Vγ9Vδ2 T cells infusions, cytokine or chemotherapy, for the treatment of relapsed or refractory AML.

Material and methods

Patients

Written informed consent was obtained from the 25 AML patients included in this study, in accordance with the Declaration of Helsinki. Clinical data are detailed in Table S1. The study was approved by the local institutional review boards of the Institut-Paoli-Calmettes. Peripheral blood mononuclear cells (PBMCs) from 20 healthy volunteers (HV) were provided by the local Blood Bank (EFS). PBMCs from AML patients and HV were isolated by density gradient (Lymphoprep, Abcys) and frozen until use. Samples with more than 80% of blasts were selected for the study.

Reagents and antibodies

BrHpp was from Innate Pharma. ZOL was from Novartis. Recombinant human (Rh) IL-2 and rhIL-15 were from BD Biosciences. Receptor Linked IL-15 (RLI, CYP0150)²⁶ was provided by Cytune Pharma. The mAbs used are listed in Table S2.

Generation of anti-human BTN3A mAbs

Anti-BTN3A mAbs (clone 20.1 and 108.5) targeting the three BTN3A isoforms (Fig. S3) were generated as previously described¹⁷ and labeled using Alexa Fluor® 647 Protein Labeling Kit (Life Technologies).

Cell culture

Effector γδ -T cells were established as previously described.³² PBMC from HV were stimulated with ZOL (1 μM) and rhIL-2 (200 IU/mL) at Day 0. From Day 5, rhIL-2 was renewed every two days and cells were kept at 1.5 × 10⁶/mL for 15 d. NK cells were isolated using EasySep Human NK cell enrichment kit (StemCell Technologies). The purity of γδ -T cells and NK cells, assessed by flow cytometry, was greater than 80%. The Burkitt lymphoma cell line, DAUDI, the human AML cell line, U937, and the erythroleukemic cell line, K562, were obtained from the American Type Culture Collection and cultured (0.5 × 10⁶/mL) in RPMI 1640 medium with 10% FCS. BTN3A Knock-down HEK293FT cells (sh#284; clone#30) were provided by E.Scotet (Inserm U892, Nantes), cultured and transfected with BTN3A1, BTN3A2, BTN3A3 mutated cDNA-containing plasmids, as described.²⁰

Flow cytometry

2 × 10⁵ PBMC were incubated at 4°C for 20 min with specified mAb before analysis on LSRFortessa (Becton Dickinson) using DIVA software (BD bioscience). For analysis of CD107 expression, γδ-T cells were incubated at 37°C with anti-CD107a and Golgi stop, with or without anti-BTN3A 20.1 mAb. After 4 h, cells were collected and analyzed by flow cytometry. To study cytokine production, cells were permeabilized with Permwash (BD bioscience) to allow intracellular staining. Trogocytosis of PKH67 (Sigma Aldrich) labeled blasts by γδ-T cells was quantified as previously described.³³

Chromium release assay

0.5 × 10⁶ target cells were incubated with 50 μCi of ⁵¹Cr (Perkin-Elmer) for AML blasts and 10 μCi for DAUDI and K562 cells for 90 min and mixed with effector cells in a Effector:Target (E:T) ratio of 30:1 ± 15:1 ± 10:1 ± 1:1. After 4 h of-incubation at 37°C, 50 μL supernatant of each sample was transferred in LUMA plates. Radioactivity was determined by a gamma counter. The percentage of specific lysis was expressed as the mean of triplicate, using the formula [(experimental – spontaneous release/total –spontaneous release) × 100]. DAUDI and K562 cell lines were used as respective positive controls for NK cells and γδ-T cells cytotoxicity.

Determination of BTN3A isoforms expression by quantitative RT-PCR (qRT-PCR)

cDNA was obtained after TRIzol extraction (Life Technologies) and M-MLV mediated reverse transcription(Life Technologies). qRT-PCR was performed with the Applied Bio-systems 7900HT Fast Real-Time PCR system. The Ct (threshold cycle) was calculated for each assay (Sequence Detection System Software 2.3, Applied Biosystems). Data were normalized using GAPDH as endogenous control (ΔCt  =  Ct_{target gene}  –  Ct_GAPDH). Higher ΔCt means lower expressions of analyzed genes. Probes were BTN3A1 (Hs01063368_m1), BTN3A2 (Hs00389328_m1), BTN3A3 (Hs00757230_m1) and GAPDH (Hs99999905_m1) Taqman gene expression Assays (Applied Biosystems).

Western blot

AML blasts (5–20 × 10⁶) were washed in cold PBS, and lysed in 110 μL of ice-cold TIT buffer (50 mM Tris pH 7.4, 200 µM NaF, 2.5 mM EDTA pH 7.4, 150 mM NaCl, 1% Triton X-100, 1% Nonidet-P40, 5% glycerol) containing protease inhibitors (Roche Applied Science) and 1 mM Na3VO4. Proteins in cell lysates were quantified according to the manufacturer's instructions (Pierce™ BCA Protein Assay Kit). Proteins were resolved by SDS-PAGE 10%, followed by western blotting with primary anti-BTN3A 20.1 and anti-GrB2 Ab (SantaCruz technology) and detection with peroxidase-conjugated anti-mouse IgG1 and anti-rabbit IgG Ab (Jackson Laboratory). Immuno-reactive bands were detected using enhanced chemo-luminescent reagents (Pierce).

NSG mouse/human AML model

NOD-SCID-common-γ-chain-knockout mice (NSG) were purchased from Charles River France or home-bred and maintained under pathogen-free conditions. All animal procedures were in accordance with protocols approved by the local Committee for Animal Experiments. Healthy 6-8-weeks-old female mice (n = 15) received 0.2 × 10⁶ luciferase transduced-U937 cells on Day 0, as described,¹² and were randomly assigned in five groups to receive intravenous injections of post-expansion-purified Vγ9Vδ2 T cells on Day 1 (30 × 10⁶, 98.8% CD3 + TCRVδ2+; Unité de Thérapie Cellulaire et Génique de Nantes) alone or combined to anti-BTN3A 20.1 mAb (250 μg) on Day 2 and 6. As RLI allows the proliferation of Vγ9Vδ2 T cells (Fig. S1), RLI (2 μg) was administered on Day 1, 2, 8 and 9 post-graft. Control mice were infused with PBS, RLI or anti-BTN3A 20.1 mAb. Mice were imaged with a Photon Imager (Biospace Lab, Paris, France) at different days. For the Primary AML blasts model, NSG mice received sublethal irradiation (2 Gy). On Day 1, they received 1 × 10⁶ of CD3-depleted blasts (CD3 depletion Kit, Miltenyi Biotec) isolated from UPN09. Treatments comprising intravenous infusions of expanded Vγ9Vδ2 T cells (15 × 10⁶, 90% CD3 + TCRVδ2+), RLI (2 μg) and either anti-BTN3A 20.1 mAb (250 μg) or IgG1 (250 μg), were administered twice a week, from 8 to 10 weeks post-graft. The peripheral blood number of blasts was quantified by Flow Cytometry (Fig. S2B) and expressed as percentage of variation: [(Absolute number of blasts/μL of blood after treatment (W10) – Absolute number of blasts/μL of blood before treatment (W8))/Absolute number of blasts/μL of blood before treatment (W8)] × 100. Mice were sacrificed at week 11th post-graft and the number of blasts was quantified in their bone marrow (BM).

Statistics

Results are expressed as median ± SEM. Statistical analysis was performed using Spearman correlation, Wilcoxon test and Mann–Whitney t test. p values <0.05 were considered significant. Analyses were performed using GraphPad Prism program.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Author contributions

AB designed research, performed experimental work, analyzed and interpreted data and drafted the paper; ALR, FO, CC, AG, LP performed experimental work and analyzed data; EM contributed to draft the paper; JGD designed research, performed experimental work and analyzed data. NV, YC and ES contributed to the design of the project research. RC designed research, performed experimental work, analyzed and interpreted data and contributed to draft the paper. DO designed research, contributed to the analysis and interpretation of data and to draft the paper.

Acknowledgments

We thank the CRCM animal core, cytometry and immunomonitoring facilities for their technical assistance. We acknowledge P.Livrati, S.Just-Landi and S.Pastor for technical support.

Funding

This study was supported by research grant from Fondation ARC to AB and CC, ANR GDStress, Fondation pour la Recherche Médicale (FRMDEQ20140329534), and INCa Translationnel.