Ex vivo characterization of acute myeloid leukemia patients undergoing hypomethylating agents and venetoclax regimen reveals a venetoclax-specific effect on non-suppressive regulatory T cells and bona fide PD-1+TIM3+ exhausted CD8+ T cells

Acute myeloid leukemia (AML) is an aggressive heterogeneous disease characterized by several alterations of the immune system prompting disease progression and treatment response. The therapies available for AML can affect lymphocyte function, limiting the efficacy of immunotherapy while hindering leukemia-specific immune reactions. Recently, the treatment based on Venetoclax (VEN), a specific B-cell lymphoma 2 (BCL-2) inhibitor, in combination with hypomethylating agents (HMAs) or low-dose cytarabine, has emerged as a promising clinical strategy in AML. To better understand the immunological effect of VEN treatment, we characterized the phenotype and immune checkpoint (IC) receptors’ expression on CD4+ and CD8+ T cells from AML patients after the first and second cycle of HMA in combination with VEN. HMA and VEN treatment significantly increased the percentage of naïve CD8+ T cells and TIM-3+ CD4+ and CD8+ T cells and reduced cytokine-secreting non-suppressive T regulatory cells (Tregs). Of note, a comparison between AML patients treated with HMA only and HMA in combination with VEN revealed the specific contribution of VEN in modulating the immune cell repertoire. Indeed, the reduction of cytokine-secreting non-suppressive Tregs, the increased TIM-3 expression on CD8+ T cells, and the reduced co-expression of PD-1 and TIM-3 on both CD4+ and CD8+ T cells are all VEN-specific. Collectively, our study shed light on immune modulation induced by VEN treatment, providing the rationale for a novel therapeutic combination of VEN and IC inhibitors in AML patients.


Introduction
Acute myeloid leukemia (AML) is an aggressive heterogeneous disease.Despite recent advances, the 5-year AML patients' overall survival (OS) is still largely unsatisfactory, reaching only 30% and dropping to 5-10% in the elderly (1).To improve AML patients' clinical outcomes, novel drugs have entered the clinical stage with favorable results (2).Although their direct mechanisms of action are defined, some bystander and/or off-target effects are not fully elucidated.In particular, the activity of novel drugs in modulating the immune cell repertoire has been poorly investigated.
The immune microenvironment of AML is characterized by a broad spectrum of alterations, including a marked dysfunction in the T cell compartment (3)(4)(5)(6)(7), leading to the creation of an immunosuppressive milieu, which favors immune escape and alters the response to therapy (8,9).Furthermore, AML cells can enhance T cell dysfunction through the expression of ligands activating immune checkpoint (IC) receptors expressed on T cells (9,10), modulating their functionality (11).These findings have provided the rationale for IC inhibitor (ICI)-based therapies in AML patients.However, despite the solid preclinical background, the early clinical results from trials addressing the impact of ICIs in AML have been globally disappointing (12).These data indicate the need for a more in-depth understanding of the complex cellular network operating within the AML immune microenvironment.Furthermore, considering that many studies have already shown that chemotherapy can be detrimental to the activity and survival of immune cells (13-16), a better characterization of the immunological effects of novel drugs is of utmost importance for fully exploiting immunotherapy strategies in AML.
The targeting of the anti-apoptotic B-cell lymphoma 2 (BCL-2) pathway has emerged as an efficacious and well-tolerated clinical strategy in AML.In particular, the remarkable results of Venetoclax (VEN), a specific BCL-2 inhibitor in combination with hypomethylating agents (HMAs) or low-dose cytarabine (17,18), have led to the drug approval for the treatment of newly diagnosed unfit-for-chemotherapy AML patients (19).It is well-known that VEN causes leukemic cell apoptosis through several characterized mechanisms (20)(21)(22)(23).However, VEN resistance can occur (24) and biomarkers able to predict response to treatment are still under investigation (25,26).Furthermore, few studies analyzed the bystander effect of VEN on immune cells (27)(28)(29).In particular, it has been shown that VEN enhances T cell-mediated activity against AML cells both in vitro and in vivo through the generation of reactive oxygen species (ROS), and in parallel, azacitidine increases AML cell susceptibility to T cell-mediated cytotoxicity through viral mimicry (28).Furthermore, in AML patients, HMA and VEN treatment modulates the phenotype of NK and T cells, IFN-g secretion by CD8 + T cells and Treg proliferation.Finally, an increased expression of perforin, CD39 and IFN-g production by T cells as a pre-treatment signature is associated with VENresistance (29).
Despite its recognized importance for therapeutic purposes, the expression of several IC receptors on T cells has not been evaluated in AML patients after VEN treatment.In this study, we aimed to characterize the immunological repertoire of AML patients treated with VEN in combination with HMA by focusing on the modulation of IC receptors' expression.

Patients and sample collection
The whole patient population included 27 patients, including 23 patients treated with the combination of HMAs and VEN and 4 patients with HMAs alone.The research was approved by the institutional review board of Area Vasta Emilia Centro (AVEC) Ethical Committee (approval code: 94/2016/O/Tess).Patients with AML were recruited at IRCCS Azienda Ospedaliero-Universitaria, Seràgnoli Hematology Institute in Bologna.Clinical samples and data were collected after written informed consent.According to normal clinical practice and to the study rationale, peripheral blood (PB) samples were collected from AML patients before the start of the treatment (baseline), and at the end of the first and second treatment cycle.Patients' characteristics are summarized in Table 1.PB samples were used for flow cytometry analysis or centrifuged over a Ficoll-Hypaque gradient (Lympholyte CL5020, Cedarlane, Burlington, Canada) to collect mononuclear cells (MNCs), and then cryopreserved for sample collection and storage.

Flow cytometry analysis
Fresh PB was used for multi-parametric flow cytometry analysis.The monoclonal antibodies (mAbs) used for flow cytometry are listed in Supplementary Table 1.For the surface marker staining, PB samples containing at least 0.5x10 6 white blood cells (WBCs) were labeled with dye-conjugated antibodies by incubation in the dark for 15 min at room temperature.Then, the red blood cells (RBC) were lysed, adding the FACS lysing solution (Beckton Dickinson Biosciences-BD, La Jolla, CA, USA), according to the manufacturer's instruction.

Statistical analysis
Statistical analyses were performed using GraphPad Prism software (v6.0).T-test and one-way ANOVA followed by Tukey's multiple comparison post hoc test were used for comparison of groups.P value <0.05 was considered significant.
Eighteen patients received azacitidine, while 5 patients were treated with decitabine.Furthermore, 7 patients from the first group, treated with HMA plus VEN, were consolidated with allogeneic hematopoietic stem cell transplant (HSCT).

HMA plus VEN modifies T cell subsets
To evaluate the effects of the treatment of HMA plus VEN on circulating immune cell repertoire, we first analyzed CD4 + and CD8 + T cell subsets in the PB of AML patients during treatment, compared to PB examined before treatment (baseline).Total T lymphocytes were slightly increased, although not significantly, after the first and second cycles of HMA plus VEN treatment, compared to baseline (Supplementary Figure 2A).Focusing on T cell subsets, we found that the percentage of CD4 + and CD8 + T cells did not change during HMA plus VEN treatment (Supplementary Figures 2B, C).A more detailed analysis of effector CD4 + T-cell subpopulations, identified as naïve/central memory/effector memory/terminally differentiated by CD45RA and CCR7 expression, did not show significant changes during HMA plus VEN treatment, compared to baseline (Figure 1A).Interestingly, regarding CD8 + T-cell subpopulations, we found a significant increase of CD8 + naïve T cells during the first and second cycles of HMA plus VEN treatment (mean ± SEM, baseline: 11.56 ± 2.07; cycle 1: 20.04 ± 2.89; cycle 2: 20.14 ± 3.18; baseline vs cycle 1, *P=0.037; baseline vs cycle 2, *P=0.034),while the memory subsets were not modified (Figure 1B left panel).Finally, we also studied the Treg population, and found that total Tregs, identified as CD4 + CD25 + CD127 low/-T cells, tended to decrease after HMA plus VEN treatment (Figure 1C left panel).A more detailed analysis of Treg subsets revealed that the HMA plus VEN only slightly affected the naive and effector Treg distribution (Supplementary Figures 2D, E).On the contrary, the percentage of cytokine-secreting non-suppressive Tregs, which includes Th17 cells, was significantly decreased after the first cycle of HMA plus VEN treatment, compared to the baseline (mean ± SEM, baseline: 4.51% ± 0.855%; cycle 1: 2.63% ± 0.34%; cycle 2: 3.28% ± 0.40%; baseline vs cycle 1, *P=0.041; Figure 1C right panel).
These data suggest that HMA plus VEN treatment induces a redistribution of immune cell subsets towards a more naïve phenotype for CD8 + T-cell subsets.On the contrary, HMA plus VEN treatment does not significantly modify the Treg compartment, while it induces a decrease in the percentage of cytokine-secreting non-suppressive Tregs.
We next analyzed the expression of IC receptors on CD8 + T cells, and we found a slight not significant increase in PD-1 expression after the second cycle of HMA plus VEN treatment (Figure 2D).Similar to what we found for CD4 + T cells, TIM-3 + CD8 + T cells resulted significantly increased after the first cycle of HMA plus VEN treatment, compared to baseline (mean ± SEM,   2E).At the same time, we found a trend toward a decrease in the PD-1 + /TIM-3 + CD8 + T cells after the first treatment cycle (Figure 2F).
To further characterize the expression of markers involved in T cells' exhausted and senescent status in AML patients treated with HMA plus VEN, we analyzed the expression of CD244 and CD57 on CD4 + and CD8 + T cells.No significant differences were found after HMA plus VEN treatment (Figures 3A, B).Regarding Tregs, the expression of CTLA-4, OX-40, ICOS, PD-1 and TIM-3 was analyzed in the effector Tregs, known to have the highest suppressive potential among Treg subsets ( 30), but we did not find any differences in their expression after treatment (Supplementary Figure 3).
Overall, these findings indicate that HMA plus VEN modify the expression of ICs on CD4 + /CD8 + T cells but not in Treg cells.

VEN specifically decreases cytokinesecreting non-suppressive Tregs and regulates TIM-3 expression on CD8 + and PD-1/TIM-3 expression on CD4 + and CD8 + T cells
To dissect the specific contribution of VEN in modulating the immune cell repertoire of AML patients treated with HMA plus VEN, we considered pair-matched samples comparing the time points before treatment (baseline considered as 1), treated only with HMA and treated with the combination of HMA plus VEN.Focusing on T lymphocyte subsets, we found that VEN did not change the distribution of CD8 + naïve T cells (Figure 4A).In contrast, VEN specifically induced a significant reduction in the percentage of cytokine-secreting non-suppressive Tregs (mean ± SEM, HMA+VEN cycle 1: 0.62% ± 0.1%; baseline vs cycle 1 HMA plus VEN, **P=0.008; Figure 4B).
Next, we analyzed the IC expression and, interestingly, in contrast to what was observed considering the entire group of patients, we found that the expression of PD-1 on CD4 + T cells was slightly decreased by VEN treatment for matched patients (Figure 4C).A similar pattern was observed for TIM-3 expression on CD4 + T cells, whose expression was not affected by VEN and seemed to decrease after HMA treatment (Figure 4D).On the contrary, we observed a significant decrease of CD4 + T cells expressing TIM-3 and PD-1 only in patients treated with the combination HMA plus VEN after one cycle of HMA (mean ± SEM, HMA+VEN cycle 1: 0.58% ± 0.14%; baseline vs cycle 1 HMA plus VEN, *P=0.025; Figure 4E).Next, we analyzed the expression of ICs on CD8 + T cells, and we found that PD-1 expression is not influenced by VEN (Figure 4F).Conversely, the expression of TIM-3 remained stable after HMA treatment only, but was significantly increased after HMA plus VEN treatment indicating a VEN-specific effect (mean ± SEM, HMA+VEN cycle 1: 1.47 ± 0.15%; *P=0.020; Figure 4G).Interestingly, VEN treatment specifically and significantly decreased the expression of PD-1/TIM-3 on CD8 + T cells (Figure 4H; mean ± SEM, HMA+VEN cycle 1: 0.68 ± 0.08%; baseline vs cycle 1 HMA plus VEN, **P=0.007).Importantly, we found a statistically significant difference between the percentage of CD8 + Tim-3 + T cells at baseline and therapy response after the first cycle (Supplementary Figure 4).In particular, the increased percentage of Tim-3 + CD8 + T cells was positively correlated with complete remission (CR) compared to the patients which did not achieve CR (Supplementary Figure 4A).The population of CD8 + T cells co-expressing PD-1 and Tim-3, strongly associated with exhaustion status, was not correlated with therapy response (Supplementary Figure 4B).
Overall, an in-depth analysis performed comparing samples collected after HMA only and at the first cycle of HMA plus VEN treatment with respect to the baseline, suggested a specific role of VEN in modulating the expression of inhibitory IC receptors on both CD4 + and CD8 + T cells.In particular, VEN induces an increase of the TIM-3 + CD8 + T cell population and a reduction of PD-1 +/ TIM-3 + CD4 + and CD8 + T cells.

Discussion
In this study, we analyzed the immune cell repertoire of 23 AML patients treated with HMA plus VEN.We focused on the modulation by treatment of IC receptors' expression on T cells.Our data unravel the ability of VEN to modulate the immune cell compartment and the expression of ICs.While the HMA plus VEN treatment is becoming the backbone for AML therapy, new combinatorial drug strategies, including immunotherapies, have been considered to improve the survival of AML patients.For this reason, increased knowledge of the immunological effects of VEN treatment is fundamental for fully exploiting new immunotherapy strategies in AML.
VEN is a specific BCL-2 inhibitor.Interestingly, various T cell subsets depend on BCL-2 for their survival to varying degrees and, thus, may be influenced by the treatment with VEN (31).In particular, naïve and memory T cells require BCL-2 for survival and/or homeostasis (32,33).In our study, after HMA plus VEN treatment, we did not find a modulation of either the total percentage of CD3 + , CD4 + and CD8 + T cells nor the percentage of CD4/CD8 immune T cell subsets, except for an increase of naïve CD8 + T cells.This finding contrasts with previous works showing that BCL-2 inhibition in different settings led to a reduction of total and naïve T cells and to an increase in the proportion of memory T cells (34-36).In particular, in AML patients treated with HMA and VEN a depletion of total T, B and NK cells and an increase of the CD4 + and CD8 + T-cell frequencies with an effector memory phenotype at the expense of naïve T cells have been observed (29).However, despite a decrease in the total number of CD4 + and CD8 + T cells, their differentiation status did not change after 1 year of VEN-based therapy in patients with chronic lymphocytic leukemia (27).The discrepancy in these findings could be due to different reasons:1) the different timing of blood collection during the therapy; 2) the analysis of percentages compared to the absolute numbers, and 3) the different cell surface markers used to identify the T cell immune subsets.Furthermore, it must be considered that our particular and previously unexplored setting, i.e. the paired analysis conducted on patients treated with HMA plus VEN or HMA only, may have allowed us to distinguish a VEN-dependent effect from an HMA-dependent effect.In our hands, VEN did not alter the percentage of CD8 + naïve T cells.
Next, we found that HMA plus VEN treatment did not affect total Tregs and their subsets, according with a previous study (29).These results are corroborated by the demonstration that Tregs are relatively resistant to apoptosis induced by BCL-2 inhibition compared to other T cells.Indeed, BCL-2 is not required for Treg survival, as opposed to MCL-1, another member of the BCL-2 pathway, the loss of which leads to fatal autoimmunity in a mouse model (37).Interestingly, by analysing Tregs' subset after HMA plus VEN treatment, we found a decrease in the CD25 + /CD45RA -/ FOXP3 + T cells, which represent a cytokine-secreting and nonsuppressive population with Th17 cell potential (30).Of note, we did not observe such reduction in patients treated with HMA only, suggesting a putative-specific role of VEN.
To our knowledge, this is the first study reporting the effect of VEN treatment on this cell population.The plasticity of CD4 + T cell subsets is well-known, and in particular, the plasticity between Th17 and Tregs has been described (38).The Th17 and Treg differentiation networks play a critical role in the development of autoimmune diseases (39).Interestingly, Th17 cells are considered a double-edged sword in solid tumors, promoting at the same time tumor progression, angiogenesis, and anti-tumor response, and immune cells' recruiting (40,41).In line with this, Th17 cells can promote both pro-and anti-tumor immunity in hematological malignancies.An increase in Th17 cells was associated both with better and worse clinical outcomes in AML patients (42)(43)(44).Moreover, an increase in Th17 cells population was found in not responding relapsed/refractory AML patients, compared to responders treated with HMA and the anti-PD-1 Nivolumab (45).
By specifically addressing the expression of inhibitory and stimulatory IC receptors, we found a significant increase in TIM-3 + CD4 + T cells after the first cycle of treatment.However, when we paired patients to dissect the contribution of VEN, this finding was not confirmed.On the contrary, we found a VEN-specific increased expression of TIM-3 on CD8 + T cells after the first cycle.As previously mentioned, TIM-3 expression has been associated with exhaustion and dysfunction of T cells, and with immune evasion in AML (46).Interestingly, AML cells can enhance the expression of inhibitory receptors, including TIM-3, thus determining CD4 + T cell exhaustion in vitro (47), and TIM-3 expression on immune cells has been correlated to both a better and a poor prognosis in AML patients (48-50).Besides its expression on immune cells, TIM-3 has been found in AML leukemia stem cells but not in their normal counterparts (51, 52), and increased levels of Galectin-9, a TIM-3 ligand, were found in AML patients' serum (53).However, despite its implications in cell survival and disease progression, the role of TIM-3 on AML cells is still unknown and the blockade of TIM-3 as a single agent did not show a substantial clinical benefit.However, our finding showing the increased TIM-3 expression on CD8 + T cells after VEN treatment could support on-going clinical trials based on the use of TIM-3 inhibitor (NCT03066648, NCT04150029, NCT04623216) in combination with VEN.
Of note, we found a significant VEN-dependent decrease of PD-1 + TIM-3 + CD4 + and CD8 + T cells after the first treatment cycle.PD-1 + TIM-3 + double-positive T cells have been indicated as prognostic in different settings.The exhausted T cell population PD-1 high TIM-3 + was shown to be functionally deficient and was associated with AML relapse after allogeneic HSCT (54).A higher proportion of PD-1 + TIM-3 + CD3 + T cells in the bone marrow and of PD-1 + TIM-3 + CD4 + T cells in the PB of non-complete remission versus CR patients after first-cycle chemotherapy were observed (55).In mouse model, the PD-1 and TIM-3 co-expression on CD8 + T cells was increased during AML progression (56).Within the CD8 + PD-1 + T cells, the TIM-3 expression identifies the most dysfunctional T-cell population in human chronic viral infections (57,58).This finding could indicate a beneficial effect of VEN on functional T-cell response confirming a previous study showing that VEN, leading to increased ROS generation, enhances T cell-mediated cytotoxicity against AML cells in vitro and in vivo (28).
In conclusion, our study had new evidence in support of the ability of VEN to modulate immune cell composition and phenotype.In particular, we identified two significant VENspecific effects 1) a decrease of cytokine-secreting non-suppressive Tregs with Th17 cell potential, the clinical significance of which deserves to be explored; 2) a decrease of the PD-1 + /TIM-3 + CD4 + and CD8 + T cells, which may have a beneficial effect.Furthermore, we also found an enhanced expression of TIM-3 on CD8 + T cells.Understanding the immunological microenvironment under the action of the selective BCL-2 inhibition can potentially reveal both novel mechanisms of resistance and new treatment combinations.

3 4
FIGURE 3 Expression of CD244 and CD57 on CD8 + T cells during HMA plus VEN treatment.All the samples were analyzed by flow cytometry and data are represented as mean ± SEM. (A) Percentage of CD8 + CD244 + T cells before treatment (n=12), after first (n=18), and second (n=12) cycle of HMA plus VEN.(B) Percentage of CD8 + CD57 + T cells before treatment (n=12), after first (n=18), and second (n=12) cycle of HMA plus VEN.

TABLE 1
Patient characteristics.