Fas Ligand Deficiency Impairs Tumor Immunity by Promoting an Accumulation of Monocytic Myeloid-Derived Suppressor Cells

In various pathologic situations, such as chronic inflammation and cancer, the differentiation of immature myeloid cells into mature granulocytes, macrophages, and dendritic cells (DC) is blocked, resulting in their expansion and conversion into potent immunosuppressive cells [myeloid-derived suppressive cells (MDSC); ref. 1]. MDSCs consist of two major subsets, granulocytic (G-MDSC) and monocytic (M-MDSC) subpopulations, that produce immune suppressive factors [arginase I, inducible nitric oxyde synthase (iNOS), and cytokines], able to block effector T-cell functions by different mechanisms, including the differentiation of naïve CD4 T cells into TH17 or regulatory T cells (Treg) populations (2–4). However, the mechanisms that regulate MDSC accumulation and subset distribution remain poorly understood. In addition to MDSCs, the stroma of many tumor types, both in mice models and human patients, contains macrophages [tumor-associated macrophage (TAM)] that are able to suppress immunity by different ways (5–7).

T-cell responses are regulated by a balance between stimulatory and inhibitory signals that are delivered by a set of immunoregulatory receptors and their ligands, such as PD-1 and its ligand PD-L1. High levels of PD-L1 were found in the tumor environment (8, 9), and blockade of its interaction with PD-1 receptor results in tumor regression in patients with various cancer types (10). The PD-1/PD-L1 interaction may exert its inhibitory effect on antitumor immunity, in part, by inducing CD4⁺CD25⁺FoxP3⁺ Tregs with suppressive function (11, 12). In steady state, Tregs are a specialized subset of CD4 T cells that have a crucial role in maintaining peripheral tolerance. The absence of Tregs leads to autoimmunity (13), whereas increased Tregs promote tumor progression by interfering with tumor-specific T cells (14–16).

The Fas–FasL pathway is an established homeostatic mediator of immune cell apoptosis (17–19). On tumor cells, the role of the Fas–FasL pathway has been a matter of debate for a while. During tumor progression, Fas is frequently downregulated or cells are resistant to Fas-induced apoptosis, raising the possibility that loss of Fas is a way of tumor evasion. However, complete loss of Fas is rarely seen in patients with cancer (20). In addition, FasL- or Fas-deficient mice, despite the massive accumulation of immune cells, do not, in contrast to perforin-deficient mice, develop spontaneous malignancies (17, 18, 21).

In the present study, we examined, in Lewis lung carcinoma (LLC)–bearing mice, the impact of FasL on myeloid cell distribution and function. We show that the homeostasis of macrophages and DCs is FasL independent and that FasL plays a key role in regulating the suppressive environment by controlling MDSC subsets, TAM, Tregs, as well as PD-1/PD-L1 expression levels.

C57BL/6 Fasl^−/− mice and control littermates (Fasl^+/+), as described previously (17), were housed in the animal facility at Gustave Roussy. Experimental animals were age- and sex-matched and used at 8 to 10 weeks of age. All experiments were approved by the institutional guidelines. Syngeneic tumor cell lines, LLC, and B16-F10 (melanoma cell line) were purchased from ATTC, amplified and kept in liquid nitrogen. E2 leukemic cells were produced by Dr. Louache's team (INSERM U1170, University Paris Sud, Le Kremlin-Bicétre, France) according to Lavau and colleagues (22). Briefly, retroviral pseudotypes were generated in 293T cells (from the ATTC) transiently transfected with the retroviral vector MSCV/EGFP plasmid plus plasmids expressing MoMuLV env and gag/pol genes. Ly5.2 fetal liver (FL) cells were obtained from E16 pregnant mice and FL progenitors (lineage-negative) were enriched using the Midi MAX Lineage Depletion Kit (Miltenyi Biotech). For infections, virus particles were fixed onto 35-mm plates coated with Retronectin (Cambrex) before the addition of FL progenitors suspended in DMEM (Gibco-BRL) supplemented 50 ng/mL mSCF (PeproTech), 100 ng/mL FLT3-L (Celldex Therapeutics, Inc.), and 10 ng/mL IL3 (PeproTech). Infections were repeated after 24 hours. Transduced cells (10⁶) supplemented with 2 × 10⁵ congenic C57BL/6-Ly5.1 bone marrow (BM) cells were transplanted into irradiated (9,5 Gy) C57BL/6-Ly5.1 mice via the retro-orbital plexus. Tumor development was assessed by GFP and Ly5.2 expression on cell from BM at 8 weeks after transplantation. Eighty percent of BM cells were positive for both markers. Leukemic E2 cells were kept frozen in liquid nitrogen and, at the time of use for experiments, they were tested for GFP expression. We did not perform authentication of the cell lines.Mice were injected either intravenously (2 × 10⁵ cells or 3 × 10⁵ cells depending on tumor cell line) or subcutaneously (3 × 10⁵ cells), as appropriately indicated in the text. Tumor growth was monitored by measuring two opposing diameters and volume was calculated as follow: (D × d²) × 0.52. Tumors were analyzed when they reached an average volume of 327.5 ± 34.9 mm³ and 728.8 ± 153.2 mm³ for Fasl^+/+ and Fasl^−/− mice, respectively, within 15 to 23 days after tumor cell injection. GFP-positive leukemic mice were analyzed at day 15 to 20 after injection. For induction of peritonitis, mice were given intraperitoneal injection of 1 mL of 3% thioglycollate solution (Bio Merieux). At day 3 after injection, the recruited immune cells were removed from the peritoneum.

Recombinant murine GM-CSF was obtained from R&D Systems. Direct conjugated antibodies used for cell labeling were purchased from Affymetrix/eBiosciences: anti-F4/80 (BM8), anti-Ly6C (HK1.4), anti-Ly6G (1A8), anti-Gr-1 (RB6-8C5), anti-CD11b (M1/70), anti-CD4 (GK1.5), anti-CD8 (53–6.7), anti-CD25 (PC61), anti-PD-1 (J43), anti PD-L1 (MIH5), anti-Foxp3, and Ki67 (clone SolA15). Fluorochrome-conjugated antibodies were used at the appropriate combinations and dilutions in staining buffer (PBS with 0.5% BSA). All samples were treated with anti-CD16/32 (2.4G2) antibody to reduce Fc-receptor binding. For Foxp3 and Ki67, staining was performed according to the manufacturer's instructions. Cell staining was examined by flow cytometry using the LSR II (BD Biosciences). Data were analyzed by FACS DIVA 7.0 or FlowJo 7.6.5 software. For functional studies, high-grade purified anti-CD3 (145–2C11), anti-CD28 (37.51), and anti-PD-L1 (MIH5) antibody and its isotype control were used (Affymetrix/eBiosciences).

Single-cell suspensions were prepared from spleen by mechanical dissociation followed by removal of red blood cells with ACK buffer. Solid tumors were dissociated into small fragments and digested with collagenase XI and type IV bovine pancreatic DNase (Sigma-Aldrich) for 45 minutes at 37°C with gentle shaking. Red blood cells were removed by ACK buffer in single-cell suspensions. For peritoneal macrophages, thioglycollate-treated mice were injected with 5 mL of PBS to recover immune cells from peritoneum. Macrophages and MDSCs were, respectively, stained with conjugated CD11bF4/80 and CD11bGr-1 and sorted on a FACS Aria. This process yielded cell-type suspension with purity >95%.

LLC-derived conditioned medium was generated by culturing 2 × 10⁵ cells/mL for 48 hours in DMEM/F12 supplemented with 10% FBS and penicillin/streptomycin. Supernatants were concentrated to a third of volume using 10-KDa molecular weight cutoff columns (Millipore).

For an evaluation of T-cell proliferation, splenocytes from either naïve mice or LLC-bearing mice were cultured in the absence or the presence of MDSCs or macrophages at different ratios. Splenocytes were stimulated with either soluble anti-CD3 (2 μg/mL) and anti-CD28 (1 μg/mL) antibodies or coated anti-CD3 antibodies (5 μg/mL), as appropriately indicated in the text, for 72 hours at 37°C. Cell proliferation was measured by either MTT, a colorimetric assay for assessing cell viability and proliferation, or BrdUrd incorporation (BrdUrd Flow Kit; BD Biosciences). For assessment of T-cell function, cytokine production was determined in the supernatants of activated splenocytes cocultured with either MDSCs or macrophages. Cytokine levels were determined using the Mouse Th1/Th2 FlowCytomix Multiplex (Affymetrix/eBiosciences) according to the manufacturer's instructions. Samples were quantified using the Flow Cytomix Pro 2.2 software (Affymetrix/eBiosciences). For PD-L1 blocking, MDSCs were treated with neutralizing PD-L1 antibody (5 μg/mL) or IgG control at 37°C for 2 hours prior to the addition of activated splenocytes.

Arginase activity was measured in cell lysates (10⁶ cells), as previously described (23). For NO production, culture supernatants (100 μL) were mixed with equal volume of Greiss reagent, and nitrite concentrations were determined by comparing the absorbance values of the test samples with a standard curve generated by a serial dilution of 0.250 mmol/L sodium nitrite.

Total RNA was extracted from the samples with TRIzol solution (Invitrogen). A total of 1 μg of total RNA was converted into cDNA by using TaqMan Reverse Transcription Reagent (Applied Biosystems), and mRNA levels were quantified by the SYBR-Green qPCR method (Applied Biosystems) or TAQMAN qPCR method (Applied Biosystems). Relative expression was calculated by using the comparative C_t method (2^−ΔΔCt).

Data were analyzed with GraphPad Prism. A nonparametric Mann–Whitney test was used to compare data between Fasl^+/+ and Fasl^−/− mice. P values of <0.05 were considered statistically significant.

To investigate the role of FasL in tumor development, Fasl^+/+ and Fasl^−/− mice were injected with LLC tumor cell line, and tumor growth and mice survival were followed over a 60-day period. As shown in Fig. 1A, the mortality of Fasl^−/− LLC-bearing mice was accelerated, resulting in reduced survival rate compared with LLC-bearing Fasl^+/+ mice. In addition, tumors grew faster in Fasl^−/− mice, as indicated by the significantly increased tumor volumes, than in Fasl^+/+ mice (Fig. 1B).

Given the importance of FasL in the regulation of immune cells, we analyzed immune cells within the spleen of LLC-bearing mice at 21- to 23-day after injection. A significant increase in the total cell counts was observed in Fasl^+/+ LLC-bearing mice when compared with naïve mice (Fig. 2A). Strikingly, in Fasl^−/− LLC-bearing mice, the total spleen cell number was barely augmented when compared with Fasl^−/− naïve mice (Fig. 2A).

In search of a cellular mechanism for the FasL-mediated tumor-suppressing effect, we investigated the immunosuppressive cells, MDSCs and Tregs. MDSCs are immature myeloid cells characterized by the expression of CD11b⁺Gr-1⁺ cell markers. In the spleen of naïve Fasl^+/+, the CD11b⁺Gr-1⁺ population represented 3% of total leukocytes, whereas in FasL deficient mice, this population was significantly increased (Fig. 2B and C). A similar significant increase was seen in the absolute number, indicating that the homeostasis of this population is FasL dependent. Also, the proportions of CD11b⁺Gr-1⁺ population as well as CD11b⁺Ly6C⁺Gr-1⁻ monocytes and CD11b⁺Ly6G⁺ neutrophils were significantly amplified in the blood of naïve Fasl^−/− mice compared with controls (Supplementary Fig. S1A).

In response to tumor development, the percentages as well as the absolute cell number of CD11b⁺Gr-1⁺ cells were significantly increased in Fasl^+/+ LLC-bearing mice compared with naïve Fasl^+/+ mice (Fig. 2B and C). In contrast, in Fasl^−/− LLC-bearing mice, the small increase of CD11b⁺Gr-1⁺ cells, albeit significant in comparison to Fasl^−/− naïve mice, remained remarkably low when compared with Fasl^+/+ LLC-bearing mice (Fig. 2B and C). A similar decrease of CD11b⁺Gr-1⁺ MDSCs was observed in LLC tumor infiltrates of Fasl^−/− mice (Supplementary Fig. S1B). We checked whether macrophages (CD11b⁺F4/80⁺Gr-1⁻) or DCs could compensate for the decreased number of CD11b⁺Gr-1⁺ cells. In the spleen of naïve FasL-deficient mice, the number of macrophages (CD11b⁺F4/80⁺Gr-1⁻) was, as for DCs (19), significantly increased when compared with littermates (Supplementary Fig. S1C). However, in LLC-bearing mice, we did not observe a significant difference in the numbers of either macrophages or DCs between the two genotypes (Supplementary Fig. S1D). It is important to point out that the decrease of CD11b⁺Gr-1⁺ MDSCs cells is not restricted to the LLC tumor cell line. In B16-F10 melanoma–, as well as in E2 leukemic myeloid cell–, bearing mice, the proportion of CD11b⁺Gr-1⁺ MDSCs was reduced in the spleen and B16-tumor infiltrates (Supplementary Fig. S2A and S2B).

To determine if the reduced MDSC accumulation observed in Fasl^−/− LLC-bearing mice was associated with an abnormal production of proinflammatory cytokines, such as IFNγ, GM-CSF, and TNFα, previously shown to be involved in MDSC accumulation (3), we measured their production in the sera- and anti-CD3/CD28-activated splenocytes. Interestingly, the production of IFNγ, GM-CSF, and TNFα was similar in Fasl^+/+ and Fasl^−/− mice (Fig. 2D). Thus, in the absence of FasL, the proinflammatory conditions generated by tumor injection could not support the MDSC accumulation, suggesting a unique role for FasL.

In addition to MDSCs, Tregs are well-known immunosuppressive population of CD4 T cells in the tumor environment. We found that, in contrast to MDSCs, Tregs were significantly increased in Fasl^−/− LLC-bearing mice. As shown in Fig. 2E, the increase of the CD25^HighFoxP3⁺ proportion in Fasl^−/− mice, albeit weak, was significant when compared with control mice. The increase of Tregs is induced by the tumor environment, as naïve Fasl^−/− mice exhibited a similar proportion of CD25^HighFoxP3⁺ as control mice (Fig. 2E).

MDSCs are divided into monocytic (M-MDSC, Gr-1^low) and granulocytic (G-MDSC, Gr-1^high) subsets based on the differential expression of Ly6C and Ly6G antigens, respectively, along with the brightness of Gr-1 expression. In addition, the G-MDSC subset is the prevalent population of MDSCs in different tumor models (24, 25). In the spleen of Fasl^+/+ LLC-bearing mice, the proportion of the Gr-1^HighLy6G⁺ G-MDSC subset was higher than the Gr-1^lowLy6C⁺ M-MDSC proportion (Fig. 3A). This leads to the expected ratio of 60% of G-MDSCs within the CD11b⁺Gr-1⁺ population (Fig. 3B and C). Strikingly, in Fasl^−/− LLC-bearing mice, although both MDSC subsets were reduced, this reduction was significantly pronounced for the G-MDSC subset, leading to an inverted ratio of MDSC subsets as illustrated in Fig. 3A–C. The same MDSC subset distribution was observed in tumor infiltrates from Fasl^−/− mice (Supplementary Fig. S3A and S3B).

We next evaluated the capacity of sorted CD11b⁺Gr-1⁺ MDSCs from the spleen of LLC-bearing mice to suppress in vitro the proliferation of splenocytes from normal mice stimulated with anti-CD3/CD28 antibodies. We found that sorted MDSCs from either Fasl^+/+ or Fasl^−/− mice were devoid of suppressive activity, as shown by the absence of inhibitory effect on T-cell proliferation (Fig. 4A). However, when they were treated for 4 days with GM-CSF, both FasL-positive and FasL-negative MDSCs acquired a potent suppressive activity as shown by the inhibition of proliferative response of anti-CD3/CD28-activated splenocytes as well as IFNγ production (Fig. 4A and C) and by the production of IL10 and TNFα (Fig. 4B). Interestingly, MDSCs from Fasl^−/− mice produced higher amount of IL10 and TNFα and much lower level of IFNγ compared with control MDSCs, but they displayed similar arginase activity as MDSCs from FasL^+/+ mice (Supplementary Fig. S4). Furthermore, the production of IFNγ by activated T cells was at least 2-fold lower in the presence of FasL-deficient MDSCs as compared with FasL-positive MDSCs (Fig. 4C); this is in line with the pronounced inhibition of T-cell proliferation (Fig. 4A). This suppressive activity was amplified when MDSCs were cultured with GM-CSF and conditioned medium from LLC cells (CM) as shown by a further reduced amount of IFNγ and decreased T-cell proliferation (Fig. 4A and C), indicating that additional tumor-derived factors emphasize the immunosuppressive function of MDSCs. Indeed, sorted MDSCs from tumor infiltrates of Fasl^+/+ or Fasl^−/− mice were directly able to significantly suppress proliferative response of activated T cells (Fig. 4D).

Both immune and tumor cells express PD-L1 and its interaction with PD-1 receptor represents a major obstacle to antitumor immunity. We found significantly higher level of PD-L1 expression by splenocytes from Fasl^−/− mice compared with controls (Fig. 5A). Interestingly enough, when PD-L1 expression was gated in the MDCS population, we found a significant increased percentage of the PD-L1-expressing M-MDSC subset in Fasl^−/− mice relative to control mice. Although a similar percentage of PD-L1-positive G-MDSC subsets was observed in both genotypes (Fig. 5B). In the same way, we compared PD-1 expression on CD8 T cells and found a significant higher percentage of PD-1-expressing CD8 T cells in Fasl^−/− mice compared with control mice, although similar numbers of CD8 T cells were observed (Fig. 5C). To test the functional consequence of PD-L1 and PD-1 overexpression in MDSC-mediated T-cell suppression, PD-L1 expression was blocked on ex vivo MDSCs by using neutralizing anti-PD-L1 monoclonal antibody. Interestingly, PD-L1 blockade on MDSCs abrogated the suppressive activity of MDSCs. As shown in Fig. 5D, the ability of FasL-positive and FasL-negative GM-CSF/CM–treated MDSCs to inhibit T-cell proliferation was removed after blocking with anti-PD-L1, but not with IgG control. Thus, PD-L1 expression on MDSCs is involved in mediating the suppressive action of MDSCs, at least in part, as we were not able to completely restore the proliferative T-cell response after PD-L1 blockade.

A study by Movahedi and colleagues. suggested that the M-MDSC subset might be a progenitor of TAM (26). In Fasl^−/− mice, we observed a prevalence of M-MDSCs. We therefore investigated TAM (CD11b⁺F4/80⁺Ly6C^+/−Gr-1⁻) in established subcutaneous LLC tumor from Fasl^+/+ and Fasl^−/− mice. Tumors from Fasl^−/− were significantly enriched with CD11b⁺F4/80⁺Ly6C^+/−Gr-1⁻ macrophages compared with those from Fasl^+/+ mice (Fig. 6A). We therefore evaluated the potential suppressive function of sorted CD11b⁺F4/80⁺ macrophages by different ways. Gene expression analysis showed that FasL-deficient macrophages overexpressed IL6 and IL10 immunosuppressor genes and iNOS compared with FasL-proficient macrophages. The IL10 gene overexpression was correlated with enhanced production of IL10 and reduced IFNγ production (Fig. 6B and C). However, nitrite production (NO) and arginase activity levels were similar in macrophages from both genotypes (Fig. 6D). Also, inhibition of T-cell responses was similar with macrophages from both genotypes as demonstrated by the low levels of the proliferative response as well as IFNγ and GM-CSF production of anti-CD3–activated splenocytes (Fig. 6E and F). It is worth noting that the accumulation of suppressive macrophages was not restricted to the tumor microenvironment, as thioglycollate-induced inflammatory peritonitis resulted in a significant increased proportion of suppressive macrophages in the peritoneum of Fasl^−/− mice compared with control mice (Supplementary Fig. S5A and S5B). Together, these results suggested that in an inflammatory environment FasL might control suppressive cells and affect their suppressive activity.

FasL, a cytokine with apoptotic and proinflammatory functions, is implicated in the development of autoimmunity. However, the degree to which FasL contributes to malignancies remains poorly explored, because FasL-deficient mice do not develop over age spontaneous tumors (17). In the present work, we showed that FasL deficiency generated an immunosuppressive environment, as M-MDSCs, Tregs, and PD-1/PD-L1 expression were significantly enhanced. This phenotype led to an accelerated tumor growth and reduced survival rate of the mice.

FasL signaling is crucial for maintaining the homeostasis of myeloid cells, including the CD11b⁺Gr-1⁺ immature cells (19). In cancers, the differentiation of immature CD11b⁺Gr-1⁺ myeloid cells is blocked, leading to their accumulation and conversion into potent immunosuppressive cells, therefore, promoting tumor growth. The accumulation of MDSCs could be attributed to defect in death signals, an excess of proinflammatory factors produced by tumors, host cells or both or to aberrant regulation of myelopoiesis. In the current work, we showed that FasL-mediated cell death does not interfere with the regulation of the MDSC level. In Fasl^−/− LLC-bearing mice, as well as in mice bearing B16-F10 melanoma or E2-myeloid leukemia cells, MDSCs, in both lymphoid organs and tumor sites, were not overnumbered compared with control mice. These results indicate that other death signaling pathways regulate MDSC accumulation, which is in line with a recent report demonstrating the implication of intrinsic apoptotic pathway in MDSC accumulation (27). However, our results contrast with the study by Sinha and colleagues, suggesting that FasL-mediated apoptosis regulates the MDSC level in LLC-bearing gld mice (28). The discrepancy could be attributed to the use of tumor models with different genetic background (C57BL/6 vs. Balb/c mice).

MDSCs were the cells primary affected by FasL deficiency, as their number was reduced and population structure was modified. The reduced number of MDSCs in Fasl^−/− mice was not associated with a defect of proinflammatory cytokine production, which suggests that FasL might control the differentiation of immature myeloid cells into macrophages or DCs. Indeed, tumor infiltrates from Fasl^−/− mice were significantly enriched with macrophages compared with control mice. Further investigations are needed to identify how FasL affects MDSC maturation. Nevertheless, our results are in agreement with reports demonstrating that another member of the TNF super family, TNFα, affects MDSC differentiation and accumulation in chronic inflammation (29). The mechanisms that regulate MDSC subsets are still poorly known. Recently, Youn and colleagues. showed that G-MDSCs were derived from the highly proliferative M-MDSC pool that acquired morphologic and phenotypic features of G-MDSCs (24). Here, we showed that in the absence of FasL, the MDSC population was skewed toward the M-MDSC subset (Fig. 3) with enhanced suppressive activity. These data raise the possibility that the development and/or survival of the G-MDSC subset are dependent on the FasL signaling pathway. Alternatively, and according to the work by Youn and colleagues, FasL might control the process of M-MDSC conversion into G-MDSC. Furthermore, our results provide the evidence that suppression is principally found in the monocytic component of the MDSC pool, which is in line with a very recent report by Haverkamp and colleagues showing that selective loss of the G-MDSC subset did not alter tumor incidence (30). We believe, however, that the role of FasL in the MDSC compartment requires further investigations to decipher whether the impact of FasL is intrinsic or extrinsic and how it controls MDSC subset regulation in tumor-induced inflammation.

Our data showed that FasL affects other immunosuppressive elements, such as Tregs and the PD-1/PD-L1 axis. In the absence of FasL, PD-L1 expression was increased on myeloid and nonmyeloid cells (data not shown) and blocking PD-L1 on ex vivo MDSCs enhances MDSC-mediated T-cell proliferation, which is in correlation with recent studies (9, 31). In addition, MDSCs are not the only source of PD-L1 to turn down T-cell responses in vivo. PD-L1 expressed on APC has been shown to play a role in the induction and maintenance of peripheral suppressive Tregs (32). We found an increased proportion of Tregs in the absence of FasL that cannot be attributed to a defect of FasL-mediated T-cell homeostasis, as this process is FasL independent (18), but Bcl-2 dependent (33). Thus, we could conceive that the increase of PD-L1 level on Gr-1–negative myeloid cell might contribute to the induction of peripheral Tregs. In tumor environment, upregulation of PD-L1 expression depends on proinflammatory cytokines, mainly IFNγ, and on the hypoxia signaling pathway (9, 34). It will be interesting to examine how FasL affects PD-L1 promoter regulation by investigating molecules such as NF-kb, JAK/STAT, and IRF known to bind to PD-L1 promoter.

MDSCs and PD-L1 levels are enhanced in patients with lung cancer and other cancer types, including hematologic malignancies, with a role in disease progression and/or drug resistance. Different drugs and immune checkpoint blockers, such as sunitinib and anti-PD-L1 antibody that potentially interfere with MDSC and PD-L1 activities, respectively, are under investigation or already available in lung cancer and other malignancies (35–38). As our results provide a relationship between FasL signaling pathway and the developing immunosuppressive elements, MDSCs and PD-L1, targeting FasL may potentially represent an interesting opportunity to enhance these therapeutic strategies in cancer and possibly in other pathologic conditions.

No potential conflicts of interest were disclosed.

Conception and design: S. Karray, B. Samah, F. Louache

Development of methodology: S. Karray, S. Buart, B. Samah, Y. Zhang, F. Louache

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Peyvandi, Y. Zhang, L. Durrieu, M. Polrot, K. Benihoud

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Karray, S. Peyvandi, S. Buart, B. Samah, F. Louache

Writing, review, and/or revision of the manuscript: S. Karray, S. Peyvandi, S. Buart, K. Benihoud, F. Louache

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Karray, M. Vétizou

Study supervision: S. Karray, S. Chouaib

The authors thank P. Rameau and Y. Lecluse from cytometry department and A. Noel and V. Parietti; M. Chopin from Animal Facility Staff of Gustave Roussy Campus and of Université Paris Diderot, Sorbonne Paris Cité, respectively.

S. Karray is an investigator of Centre Nationale de la Recherche Scientifique. This work was supported by INSERM, l'INCa (Cancéropole Grand-Est, convention 208-033). S. Peyvandi is a recipient of PhD grant from the University Paris-Sud and Association de Recherche contre le Cancer.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.