Death receptor Fas-mediated apoptosis not only eliminates nonspecific and autoreactive B cells but also plays a major role in antitumor immunity. However, the possible mechanisms underlying impairment of Fas-mediated induction of apoptosis during lymphomagenesis remain unknown. In this study, we employed our developed syngeneic lymphoma model to demonstrate that downregulation of Fas is required for both lymphoma development and lymphoma cell survival to evade immune cytotoxicity. CD40 signal activation significantly restored Fas expression and thereby induced apoptosis after Fas ligand treatment in both mouse and human lymphoma cells. Nevertheless, certain human lymphoma cell lines were found to be resistant to Fas-mediated apoptosis, with Livin (melanoma inhibitor of apoptosis protein; ML-IAP) identified as a driver of such resistance. High expression of Livin and low expression of Fas were associated with poor prognosis in patients with aggressive non-Hodgkin's lymphoma. Livin expression was tightly driven by bromodomain and extraterminal (BET) proteins BRD4 and BRD2, suggesting that Livin expression is epigenetically regulated in refractory lymphoma cells to protect them from Fas-mediated apoptosis. Accordingly, the combination of CD40-mediated Fas restoration with targeting of the BET proteins–Livin axis may serve as a promising immunotherapeutic strategy for refractory B-cell lymphoma.
These findings yield insights into identifying risk factors in refractory lymphoma and provide a promising therapy for tumors resistant to Fas-mediated antitumor immunity.
Despite recent advances in drug development such as the introduction of rituximab for aggressive NHL such as Burkitt lymphoma and diffuse large B-cell lymphoma (DLBCL), a subset of patients with lymphoma remains refractory or becomes unresponsive to drug treatment as a result of intrinsic or acquired resistance (1). The development of new drugs targeted to specific molecules that play a role in the genesis or maintenance of lymphoma will be facilitated by the availability of appropriate models that recapitulate NHL.
Recent progress in cancer immunotherapy has led to the development of immune checkpoint inhibitors such as antibodies that target the checkpoints mediated by the PD-1 receptor and its ligand PD-L1 or by CTLA-4 (2–4). Such inhibitors directly block the inhibitory checkpoints and thereby reactivate CTLs that are able to kill cancer cells and give rise to durable remission. Enhancement of the stimulatory immune system, such as with cancer vaccines or immunostimulatory antibodies, has also shown promise for cancer therapy (5, 6). However, the efficacy of immunotherapies appears to be limited in a certain proportion of patients due to primary and adaptive resistance (7). Identifying the molecular mechanisms by which cancer cells are intrinsically or extrinsically resistant to immunotherapy are necessary to develop more effective immunotherapeutic strategies.
Fas/FasL–induced apoptosis is a main killing mechanism by which CTLs attack cancer cells (8, 9). Fas is a member of the TNF receptor superfamily of proteins and functions as a proapoptotic death receptor when bound to its ligand, Fas ligand (FasL; ref. 10). Fas has been implicated in lymphomagenesis on the basis of its proapoptotic function as well as in autoimmune lymphoproliferative syndrome, with germline mutations in the Fas gene having been found to predispose individuals to lymphoma development (11, 12). Expression of Fas or FasL is associated with a favorable clinical outcome in patients with DLBCL (13), whereas Fas gene mutations have not frequently been detected in NHL (14). Fas-FasL–mediated immune surveillance by cytotoxic T cells is essential for prevention of DLBCL development in Blimp1 knockout and Bcl6 transgenic mice (15), suggestive of a causal relation between defective Fas-mediated apoptosis and tumorigenesis. However, possible mechanisms underlying impairment of Fas-mediated apoptosis induction during human lymphomagenesis have remained unknown.
We now describe the development of a new mouse model of mature B-cell lymphoma and show that downregulation of Fas expression is required not only for lymphomagenesis but for lymphoma maintenance. We also show that activation of CD40 signaling restores Fas expression in lymphoma cells, and we identify a molecular mechanism of resistance to Fas-mediated apoptosis. On the basis of these findings, we propose a new strategy to combat such resistance in refractory lymphoma.
Materials and Methods
Daudi and Raji as well as NIH3T3 cells were obtained from RIKEN Cell Bank, and Ramos and Namalwa were from JCRB Cell Bank. The human DLBCL cell lines SU-DHL6, DB, Toledo, and NU-DUL1 were obtained from ATCC, whereas WSU-DLCL2, NU-DHL1, and OCI-Ly19 were from DSMZ. All human cell lines were cultured in RPMI1640 medium (Sigma-Aldrich) supplemented with 10% FBS, penicillin (100 U/mL), and streptomycin (100 U/mL). Plat-E (kindly provided from Dr. T. Kitamura, The University of Tokyo, Tokyo, Japan) and Lenti-X 293T (Takara Bio) and NIH3T3 cells were cultured in DMEM (Nacalai Tesque) supplemented with 10% FBS, penicillin (100 U/mL), and streptomycin (100 U/mL). Cell lines are maintained at ≤ 15 passage from receipt and were characterized by short tandem repeat analysis before use. Mouse lymphoma cells were cultured in RPMI1640 supplemented with 10% FBS and 2-mercaptoethanol (55 μmol/L), penicillin (100 U/mL), and streptomycin (100 U/mL). All cells were maintained at 37°C under 5% CO2 and at 100% humidity.
All animals were maintained under specific pathogen–free conditions and handled in accordance with the guidelines of Keio University School of Medicine (Tokyo, Japan). All animal procedures and study protocols were approved by the Keio University Ethics Committee for Animal Experiments. C57BL/6 mice (6–10 weeks of age) were obtained from Charles River Japan. Cdkn2a−/− mice (B6.129-Cdkn2atm1Rdp) and λ-Myc mice [C57BL/6N-Tg(Igl-MYC)3Hm/Nci] were obtained from Mouse Models of Human Cancers Consortium (NCI-Frederick, MD). For BV6 treatment in vivo, recipient mice transplanted with BIRC7- and Fas-transduced lymphoma cells were treated with a single intraperitoneal injection of BV6 (20 mg/kg) or vehicle (1% DMSO in PBS) twice weekly. The number of GFP+ cells in peripheral blood of recipient mice at 2 weeks after the first injection was determined by flow cytometry. The mice in each group were monitored for survival.
Chemicals and reagents
Inhibitors of PI3K (LY-294002), MEK (U0126), and p38 MAPK (SB203580) were obtained from FUJIFILM Wako, and those of Ezh2 (DZNep), NFκB (BAY 11–7085), JAK3 (tofacitinib), and JNK (SP600125) were from Sigma-Aldrich. The inhibitors of G9a (UNC0638) and Bcl2 (ABT737) were from Cayman Chemical and ApexBio, respectively. Inhibitors of IAP proteins (BV6), Wnt–β-catenin signaling (ICG10), and Myc (10058-F4) were from Selleckchem. The BET inhibitors JQ1 and ARV-825 were obtained from BioVision and ChemieTek, respectively, and trichostatin A and zebularine were from Sigma-Aldrich. IL4, IL10, IL21, and BAFF were obtained from PeproTech, whereas CpG and LPS were from InvivoGen and Sigma-Aldrich, respectively. Multimeric FasL was obtained from Adipogen.
All labeled antibodies for flow cytometry are listed in Supplementary Table S1. Cells were exposed to antibodies specific for CD16/32 (BioLegend) for 10 minutes before incubation with labeled antibodies for 20 minutes on ice. They were then either sorted with a MoFlo flow cytometer (Beckman Coulter) or analyzed with Gallios (Beckman Coulter) or Attune Acoustic Focusing (Thermo Fisher Scientific) flow cytometers.
The relation between gene expression levels for BIRC7 or FAS and overall survival in patients with 159 patients with mature aggressive B-cell lymphomas (91 males and 68 females, age range of 2–92 years), including 27 cases of Burkitt lymphoma, 123 cases of DLBCL, and 9 cases of unclassified aggressive B-NHL, was analyzed with gene expression data of GSE4475 obtained from the Gene Expression Omnibus (GEO) database. Raw expression data (CEL files) were summarized and normalized with the use of the robust multi-array average algorithm and the Bioconductor package affy (http://www.bioconductor.org/packages/2.0/bioc/html/affy.html). The log-rank test was applied for analysis of survival curves. Statistical analysis was performed with JMP software (SAS Institute).
Data are presented as means ± SD. Differences between two groups were evaluated with the two-tailed unpaired Student t test. The log-rank test was applied for analysis of survival curves. A P value of < 0.05 was considered statistically significant. Statistical analysis was performed with JMP software (SAS Institute).
Establishment of a syngeneic mouse model of mature B-cell lymphoma
We previously developed a precursor B acute lymphoblastic leukemia/lymphoma (pre-B ALL/LBL) model based on transplantation with progenitor B cells derived from Cdkn2a−/− mouse bone marrow (BM), cultured ex vivo, and subjected to retroviral transduction with the Myc genes (16, 17). We also developed a cutaneous T-cell lymphoma model based on transplantation of CD4 T cells derived from Cdkn2a−/− mouse spleen and transduced with the c-Myc gene (18). Given that the CDKN2A locus is frequently deleted or silenced in NHL (see also Supplementary Table S2; refs. 19, 20), we adopted a similar approach to the development of a mature B-cell lymphoma model that allows simple gene manipulations and molecular analyses. B lymphocytes isolated from the spleen of WT or Cdkn2a−/− mice were stimulated with IL4 and antibodies to CD40 (Fig. 1A; ref. 21). The cells were then infected with a retrovirus encoding both GFP and c-Myc, which is frequently overexpressed in NHL (22). Two days after infection, the resulting GFP+ cells derived from WT or Cdkn2a−/− donors had formed spheroid-like assemblies (Fig. 1B) and expressed the GC B-cell markers GL7 and Fas as well as the differentiation markers IgM and IgD, but not IgG1 or IgE (Fig. 1C). The Cdkn2a−/− B cells transduced with MYC were transferred to sublethally irradiated WT mouse recipients by intravenous or intraperitoneal injection. About 60% of the recipients of the intraperitoneal injection developed ascites, splenomegaly, as well as enlargement of the thymus and of lymph nodes in the axilla, inguinal, and intestinal regions, and they eventually died of lymphoma (Fig. 1D and E). In contrast, recipients of the intravenous injection showed no signs of tumor development (Fig. 1E). Both loss of Cdkn2a and intraperitoneal cell injection were thus required for lymphoma development.
Both lymph nodes and spleen of the lymphoma-bearing mice were largely occupied by lymphoma cells, with substantial lymphoma cell infiltration in liver and BM (Fig. 1F; Supplementary Fig. S1A). Lymph nodes also contained many mitotic cells and showed “starry sky” patterns corresponding to macrophages containing apoptotic tumor cells, histopathologic features of human Burkitt lymphoma (23). We confirmed that c-Myc expression was detected in lymphoma cells isolated from lymph nodes of model mice (Supplementary Fig. S1B). Analysis of VDJ rearrangement of Ig heavy chain genes revealed monoclonal development of lymphoma in almost all examined cases (Supplementary Fig. S1C). Mice that received a transplant of lymphoma cells isolated from lymph nodes of initial recipient mice developed lymphoma with a markedly reduced latency (Fig. 1E). Collectively, these results indicated that we had developed a new mouse model for mature B-cell lymphoma based on intraperitoneal transplantation of Cdkn2a−/− B cells transduced with MYC.
All lymphoma cells isolated from lymph nodes of model mice showed a mature B-cell immunophenotype characterized as IgM+ and IgD+, but they were GL7dim, Fas−, IgG1−, and IgE− (Fig. 1G). Unlike λ-Myc mice, which harbor MYC transgene under the control of the Igλ light chain enhancer (24) and spontaneously develop Burkitt-type lymphoma, our model mice developed IgM+IgD+ mature B-cell lymphoma more rapidly and at a higher rate (Supplementary Fig. S1D).
Fas downregulation is required for development of mouse mature B-cell lymphoma
Analysis of various cell surface markers in our lymphoma model revealed that Fas expression was lower in lymphoma cells than in the pretransplanted MYC-transduced Cdkn2a−/− B cells (Fig. 2A). Essentially all lymphoma cells were negative for Fas expression, whereas pretransplanted cells were positive for Fas (Fig. 2B). Lymphomas derived from λ-Myc mice also did not express Fas (Fig. 2B). Furthermore, lymphoma cells derived from both lymphoma models did not contain Fas mRNA, in contrast to pretransplanted cells (Fig. 2C). These results thus suggested that Fas expression is downregulated during mature B-cell lymphomagenesis. Given that Fas is a death receptor that plays a role in removal of unconventional GC B cells (25) and that all mice that received a secondary transplant of lymphoma cells with low Fas expression developed lymphomas (Fig. 1E), we hypothesized that Fas downregulation is a critical event of lymphomagenesis. To test this hypothesis, we infected Cdkn2a−/− B cells with virus vectors for both MYC and either an shRNA specific for Fas (shFas #1 or #2) or a control shRNA (shCtrl) and then injected the cells intraperitoneally into recipient mice (Fig. 2D and E). All mice transplanted with the Fas-depleted MYC–transduced B cells developed lymphoma, and they died more rapidly compared with those transplanted with the corresponding shCtrl-transduced cells (Fig. 2F). Similar histopathology of lymph nodes and the liver was apparent for mice with lymphomas derived from the Fas-depleted or control cells (Supplementary Fig. S2A). We further found that all mice transplanted intravenously with lymphoma cells with low Fas expression developed lymphoma and died with a latency similar to that apparent for mice transplanted intraperitoneally (Supplementary Fig. S2B). These results suggested that Fas plays a gatekeeping role in the development of mature B-cell lymphoma.
Fas downregulation is required for maintenance of lymphoma cell survival
To determine whether Fas downregulation is also necessary for the survival of established lymphoma cells, we first isolated lymphoma cells for primary cell culture. We then infected the cultured lymphoma cells with a retrovirus encoding Fas or with the corresponding empty virus (Fig. 3A and B). Although only Fas transduction itself induced apoptosis (26), FasL treatment greatly enhanced this effect (Fig. 3C), suggesting that Fas downregulation is required for the survival of lymphoma cells ex vivo. To examine further the role of Fas expression in lymphoma cell survival in vivo, we injected the Fas-transduced lymphoma cells into recipient mice (Fig. 3A). The mice transplanted with the Fas-transduced lymphoma cells manifested a significantly longer survival compared with those that received lymphoma cells infected with the empty vector (Fig. 3D).
Cytotoxic CD8 T cells express FasL, through which they elicit Fas-mediated apoptosis in target cells (27, 28). To determine whether CD8 T cells are responsible for killing Fas-transduced lymphoma cells in recipient mice, we injected mice with a mAb to CD8 to deplete CD8 T cells before and after transplantation (Fig. 3E; Supplementary Fig. S2C). We found that the population of lymphoma cells in the spleen of mice treated with anti-CD8 was larger than that in the spleen of mice treated with an isotype control antibody (Fig. 3F). Furthermore, the mice treated with anti-CD8 died more rapidly than did those treated with the isotype control (Fig. 3G). On the other hand, mice treated with anti-NK1.1 to deplete natural killer cells, which also express FasL, did not show a significant difference in survival time compared with those treated with an isotype control (Supplementary Fig. S2D). These results suggested that depletion of CD8 T cells in recipient mice confers an immune-compromised microenvironment in which Fas-transduced lymphoma cells to avoid apoptosis.
Activation of CD40 signaling restores Fas expression and confers sensitivity to FasL-induced apoptosis in mouse and human lymphoma cells
Our results indicated the importance of Fas downregulation during lymphomagenesis. To examine whether Fas expression is epigenetically silenced in lymphoma cells, we first performed bisulfite genomic sequencing, but we found that the promoter region of the Fas gene was largely unmethylated (Fig. 4A). We also treated mouse lymphoma cells with the DNA methyltransferase inhibitor zebularine, the histone deacetylase inhibitor trichostatin A, or the histone methyltransferase inhibitors UNC0638 or DZNep, but none of these agents had a substantial effect on Fas expression (Supplementary Fig. S3A), suggesting that Fas downregulation is not mediated epigenetically. Given that Fas expression has been shown to be regulated by multiple factors related to B-cell activation and differentiation (10), we next examined whether B-cell activators might restore Fas expression in lymphoma cells. Mouse lymphoma cells were thus treated with IL4, IL10, IL21, B-cell–activating factor (BAFF), an agonistic antibody to CD40, the CpG dinucleotide, or LPS for 48 hours. Among these agents, only anti-CD40 substantially increased Fas expression (Fig. 4B; Supplementary Fig. S3B). A similar effect by anti-CD40 was also observed in λ-Myc mouse lymphoma cells (Supplementary Fig. S3C). Stimulation with CD40 ligand (CD40L) by coculture with CD40L-expressing NIH3T3 cells also restored Fas expression in mouse lymphoma cells (Supplementary Fig. S3D).
To examine whether such restoration of Fas expression confers sensitivity to Fas-dependent apoptosis, we treated lymphoma cells with FasL after their exposure to anti-CD40. Assay of cell viability and Annexin V staining revealed that FasL significantly increased the level of apoptosis among the anti-CD40–treated cells (Fig. 4C and D). Activation of caspase-3, which acts downstream of Fas in the apoptotic signaling pathway, was also detected in anti-CD40–treated lymphoma cells to FasL (Supplementary Fig. S3E). These results indicated that Fas expression restored by activation of CD40 signaling sensitizes lymphoma cells to apoptosis.
We next examined Fas expression in human NHL cell lines including four Burkitt lymphoma (Ramos, Raji, Daudi, and Namalwa) and seven DLBCL (DB, Toledo, SU-DHL6, OCI-Ly19, WSU-DLCL2, NU-DHL1, and NU-DUL1) lines (Supplementary Table S3). Consistent with previous findings that FAS mutation is not frequent in NHL (14, 29), we confirmed that all 11 cell lines are WT for FAS. Nine and two of these 11 cell lines showed low and high levels of Fas expression, respectively, with the latter two cell lines being originally sensitive to FasL-induced apoptosis (Fig. 4E; Supplementary Fig. S4A). Furthermore, eight of the nine cell lines that expressed Fas at a low level also expressed CD40 and manifested upregulation of Fas expression in response to CD40L treatment (culture with CD40L-expressing NIH3T3 cells), with only OCI-Ly19 lacking CD40 expression and not undergoing such upregulation of Fas expression (Fig. 4F; Supplementary Fig. S4B). To examine whether upregulation of Fas expression conferred sensitivity to apoptosis in human lymphoma cell lines, we treated these latter eight human cell lines with FasL after exposure to CD40L. Four cell lines (Ramos, Namalwa, Toledo, and NU-DUL1) showed marked sensitivity to Fas-mediated apoptosis, whereas WSU-DLCL2 and SU-DHL6 were partially resistant and Daudi and DB were completely resistant (Fig. 4G; Supplementary Fig. S4C). These findings suggested that Fas expression was downregulated but could be restored by CD40 activation in most human lymphoma cell lines, with such restoration of Fas expression conferring sensitivity to Fas-mediated apoptosis in about half of the cell lines examined.
Role of Livin in resistance to apoptosis in human lymphoma cells with Fas expression upregulated by CD40 activation
To identify factors that contribute to resistance to Fas-mediated apoptosis after upregulation of Fas expression, we examined the expression of antiapoptotic proteins including Flip, Bcl2 family members (Bcl2, Bcl-xL, Bcl2A1, and MCL1), and Survivin family members [Survivin, Livin (also known as ML-IAP), cIAP1, cIAP2, and XIAP] in human NHL cell lines. Among these various proteins, Livin was found to be expressed at a high level in all resistant (completely or partially) cell lines examined (Daudi, DB, WSU-DLCL2, and SU-DHL6), whereas cIAP2 was expressed at a high level in the resistant DLBCL cell lines (Fig. 5A). To determine the functional role of Livin and cIAP2 in lymphoma cells, we infected NU-DUL1 cells (which express Livin and cIAP2 at low levels) with retroviruses encoding these human proteins (Fig. 5B). The cells expressing Livin, but not those expressing cIAP2, were significantly more resistant to FasL-induced apoptosis after treatment with CD40L (Fig. 5C). Analysis of gene expression profiling data for NHL with GSE4475 obtained from the GEO database (30) revealed that a high expression level of the Livin gene (BIRC7) was significantly associated with poor prognosis (Fig. 5D). Of note, among patients with a high level of BIRC7 expression, overall survival was poorer in those with low FAS expression than in those with high FAS expression (Fig. 5E). Together, these results suggested that Livin plays a major role in suppression of Fas-mediated apoptosis in lymphoma cells.
BET family proteins regulate Livin expression in human lymphoma cells
To explore the mechanism underlying the high expression of Livin in lymphoma cells resistant to Fas-mediated apoptosis, we first examined the roles of Wnt–β-catenin signaling, Myc and hypoxia inducible factor-1 α (HIF1α), which have been reported to induce Livin expression (31–33). However, inhibitors of neither Wnt–β-catenin signaling, Myc nor HIF1α suppressed Livin expression in DB cells (Supplementary Fig. S5A). Furthermore, any inhibitors for the major cancer signal pathways did not suppress Livin expression (Supplementary Fig. S5A). It has been reported that the BET family protein BRD4 drives expression of oncogenic factors such as Myc in lymphoma cells through binding to acetylated histones of chromatin at enhancers and promoters of the corresponding gene loci (34). We found that treatment of DB cells with the BET inhibitor JQ1 attenuated Livin expression in a concentration-dependent manner (Fig. 6A). Of note, Livin was downregulated by JQ1 at concentrations lower than those effective for downregulation of c-Myc and Bcl2, both of which are known targets of BRD4. Downregulation of Livin by JQ1 was commonly detected in other lymphoma cell lines (Supplementary Fig. S5B). Furthermore, JQ1 treatment significantly reduced the amount of Livin gene BIRC7 mRNA (Fig. 6B). The another BET inhibitor ARV-825, a hetero-bifunctional proteolysis targeting chimera that induces degradation of BET family proteins (35), suppressed expression of BRD4, BRD3, and BRD2, and showed reduction of Livin expression (Fig. 6C). To determine whether the BET proteins are responsible for such regulation of Livin expression, we introduced shRNAs that target BRD4, BRD3, or BRD2 into lymphoma cells. The abundance of Livin was greatly decreased in cells expressing either shBRD4 or shBRD2 compared with those expressing either a control shRNA or shBRD3 with the exception that shBRD2 #1 did not inhibit Livin expression in SU-DHL6 cells (Fig. 6D; Supplementary Fig. S5C).
To further determine whether BRD4 and BRD2 directly interact with the promoter and enhancer of BIRC7 locus, we subjected DB cells to ChIP with antibodies to BRD4 or BRD2 followed by quantitative PCR (qPCR) analysis with primers targeted to the positions in the range from 10,742 bp to 255 bp away from the transcription start site (TSS) for BIRC7 (Fig. 6E). The binding of both BRD4 and BRD2 was significantly enriched at positions –7902, –5168, and –255 bp relative to the TSS for BIRC7 (Fig. 6F and G). Interestingly only BRD2 showed binding at the most distal region (−10742 bp). We confirmed that JQ1 treatment inhibited BRD4 and BRD2 interaction with their binding positions (Fig. 6F and G). Furthermore, acetylated lysine 27 of histone H3 (H3K27ac), a marker of enhancer activity, was also significantly enriched at the common binding regions of BRD4 and BRD2 (Fig. 6H). Collectively, these results suggested that BET family proteins BRD4 and BRD2, but not BRD3, enhance transcription of BIRC7 through binding to its enhancer/promoter and thereby increases Livin expression in lymphoma cells.
Livin as a therapeutic target for lymphoma cells resistant to restoration of Fas-mediated apoptosis by CD40 activation
To confirm the role of Livin in lymphoma cell lines resistant to Fas-mediated apoptosis, we introduced shRNAs that target Livin in Daudi, DB, and WSU-DLCL2 cell lines (Fig. 7A). Consecutive exposure of these Livin-depleted cells to CD40L and FasL revealed them to be markedly more sensitive to Fas-mediated apoptosis compared with cells expressing a control shRNA (Fig. 7B), suggesting that Livin is a potential therapeutic target for lymphoma cells with resistance to restoration of Fas-mediated apoptosis.
We next examined the effect of BV6, a small-molecule mimetic of the endogenous IAP antagonist Smac (36) that promotes autoubiquitylation and subsequent proteasomal degradation of IAP proteins including Livin. The abundance of Livin was markedly reduced by BV6 treatment in lymphoma cells (Fig. 7C). All four cell lines resistant to Fas-mediated apoptosis showed a greater loss of viability on exposure to CD40L, FasL, and BV6 than on treatment with CD40L and either FasL or BV6 alone (Fig. 7D). Given that BV6 also targets cIAP1 and cIAP2, we examined the effects of shRNAs for these proteins in lymphoma cells. We found that depletion of cIAP2 sensitized DB cells to Fas-mediated apoptosis, whereas depletion of cIAP1 did not (Supplementary Fig. S6A). Furthermore, depletion of Flip, a protein that was shown to inhibit Fas-mediated apoptosis in other types of lymphoma (37–39), resulted in partial sensitization of DB cells to apoptosis (Supplementary Fig. S6A). Given that depletion of Livin appeared more effective at sensitizing DB cells to Fas-mediated apoptosis (Fig. 7B) than that of cIAP2 (Supplementary Fig. S6A) and that forced expression of Livin, but not that of cIAP2, conferred resistance to such apoptosis (Fig. 5C), Livin may play the predominant role in resistance to Fas-mediated apoptosis. In addition, lymphoma cell lines showed marked sensitivity to the combination of CD40L, FasL, and the BET inhibitor JQ1 (which suppressed Livin expression; Supplementary Fig. S6B), whereas lymphoma cells did not show synergistic sensitivity to the combination of CD40L, FasL, and an inhibitor of Bcl2 (ABT737), a general antiapoptotic protein (Supplementary Fig. S6C). Collectively, these results suggested that targeting of Livin has the potential to overcome resistance to Fas-mediated apoptosis in lymphoma cells.
We next investigated a potential of Livin as a therapeutic target for lymphoma in vivo. We obtained mouse lymphoma cells resistant to Fas-mediated apoptosis by retroviral transduction with BIRC7 and Fas (Fig. 7E; Supplementary Fig. S6D and S6E). We confirmed that mice transplanted with BIRC7- and Fas-transduced lymphoma cells had a shorter survival compared with those transplanted with Fas-transduced cells (Fig. 7F). Although Livin expression conferred resistance to Fas-mediated apoptosis in mouse lymphoma cells, as was the case for human lymphoma cells, combination treatment with BV6 and FasL resulted in a significantly greater loss of viability in BIRC7- and Fas-transduced lymphoma cells compared with the effects of FasL or BV6 alone (Fig. 7G). We next administered BV6 (20 mg/kg) or vehicle intraperitoneally twice weekly in mice transplanted with BIRC7- and Fas-transduced lymphoma cells (Fig. 7H). BV6 treatment significantly extended the survival of mice transplanted with BIRC7- and Fas-transduced lymphoma cells (Fig. 7I). Collectively, these results also suggested that targeting of Livin can be an effective approach for lymphomas resistant to Fas-mediated cytotoxicity.
We have here established a new syngeneic mouse model of mature B cell lymphoma and shown that induction and maintenance of this disease requires downregulation of Fas expression. This finding supports evidence that immune surveillance based on induction of Fas-FasL–dependent apoptosis by cytotoxic T cells is important for suppression of mouse DLBCL (15). Whereas initial steps in lymphomagenesis may thus include oncogenic events such as overexpression of MYC and suppression of Cdkn2a expression, the further development of lymphoma appears to require that these cells undergo Fas downregulation to overcome immune defense. We found that activation of CD40 signaling restored Fas expression in these cells. GC B-cell proliferation is normally dependent in part on activation of CD40 signaling (40), but this dependency may be lost as a result of oncogenic events during lymphomagenesis, possibly leading to downregulation of Fas expression. We extended our mouse findings to human lymphoma by showing that most NHL cell lines manifested a low level of Fas expression that was increased in response to CD40 activation. Similar to mouse lymphomagenesis, CD40-mediated signaling that regulates the proliferation and survival of human GC B cells may be rendered obsolete by oncogenic events, resulting in downregulation of Fas during lymphoma development (Supplementary Fig. S7).
We have demonstrated a rational mechanism by which mouse and human NHL cells that express CD40 may undergo Fas restoration in response to CD40 activation, consistent with previous studies indicating that CD40 ligation induces Fas in certain human lymphoma cell lines (41, 42). On the other hands, about half of human lymphoma cell lines manifested resistance to Fas-mediated apoptosis even after Fas induction by CD40 activation. We identified the IAP protein Livin as a factor that drives this resistance. Livin contains both a BIR (baculovirus IAP repeat) motif that binds to caspases 3, 7, and 9 and thereby inhibits their activity, as well as a RING finger motif that catalyzes the ubiquitylation of the endogenous IAP antagonist Smac and thereby promotes its degradation. Livin thus inhibits apoptosis induced by members of the TNF receptor superfamily and by chemotherapeutic agents (43, 44). We further identified the BET family proteins BRD4 and BRD2 as upstream regulators of Livin in the resistant cells. BRD4 was shown to interact with promoter regions and super enhancers of oncogenes such as those for Myc and Bcl2 and thereby to contribute to tumorigenesis (45–47). Here we show that BRD4 and BRD2 regulate Livin transcription through binding to the upstream regions of BIRC7 locus where is overlapped by H3K27ac modification. Given that BET proteins bind to acetylated lysine residues of histones, increased acetylation of chromatin at the BIRC7 locus may trigger BRD4 and BRD2 bindings and the consequent upregulation of Livin expression in lymphoma cells (Supplementary Fig. S7).
We showed that the IAP inhibitor BV6 and the BET inhibitor JQ1 induced marked downregulation of Livin expression and thereby sensitized human lymphoma cell lines to Fas-mediated apoptosis. BV6 also showed similar effects on mouse lymphoma cells expressing Livin in the mouse model. BV6 is a bivalent IAP antagonist that promotes activation and dimerization of Livin and thereby leads to its ubiquitin-dependent degradation. BV6 and another bivalent IAP antagonist, birinapant, have shown promising activity in preclinical models of multiple cancers (48, 49). Several BET inhibitors have also been developed and examined in clinical trials (50). Targeting of Livin or BRD4 and BRD2 is thus a potential practical therapeutic strategy for patients with Livin-positive lymphoma.
We have now identified BET protein–driven Livin overexpression as a mechanism of resistance to immunotherapy at least in CD40+ lymphoma. Given that Fas-mediated apoptosis is one of the main killing mechanisms by CTL, Livin might contribute to resistance to CTL-mediated immunotherapy. Our study thus provides a new insight into improved immunotherapy for refractory B-cell lymphoma as well as other types of tumor resistant to immunotherapy.
Disclosure of Potential Conflicts of Interest
E. Sugihara reports grants from Japan Society for the Promotion of Science, Japanese Society of Hematology, and grants from Japan Leukemia Research Fund during the conduct of the study. T. Yaguchi reports grants from Ono Pharmaceutical Co. Ltd., Carna Biosciences, Inc., Kowa Company, Ltd., and nonfinancial support from JSR Corporation outside the submitted work. Y. Kawakami reports grants and personal fees from Ono Pharmaceutical Co. and Bristol-Myers Squibb, personal fees from MSD, AstraZeneca, Chugai, Taiho Pharma, grants from CarnaBioSciences, and grants and nonfinancial support from JSR outside the submitted work. No potential conflicts of interest were disclosed by the other authors.
E. Sugihara: Conceptualization, resources, data curation, funding acquisition, validation, visualization, methodology, writing-original draft, writing-review and editing. N. Hashimoto: Conceptualization, resources, investigation, methodology. S. Osuka: Data curation, formal analysis, writing-review and editing. T. Shimizu: Resources, investigation, writing-review and editing. S. Ueno: Investigation, methodology. S. Okazaki: Investigation, methodology. T. Yaguchi: Conceptualization, supervision, writing-review and editing. Y. Kawakami: Conceptualization, supervision, writing-review and editing. K. Kosaki: Data curation, writing-review and editing. T.-A. Sato: Supervision, writing-review and editing. S. Okamoto: Supervision, writing-review and editing. H. Saya: Supervision, funding acquisition, project administration.
We thank T. Kitamura (The University of Tokyo, Tokyo, Japan) for providing the vectors pMXs-IRES-GFP and -puro as well as Plat-E cells; S. Yamanaka (Kyoto University, Kyoto, Japan) for providing pLenti6/UbC/V5-DEST-Slc7a1; M. Yonezawa, I. Ishimatsu, S. Hayashi, N. Kamoshita, and K. Sonoda for technical assistance; and K. Arai, M. Sato, and M. Kobori for help with preparation of the manuscript. This work was supported by Japan Society for the Promotion of Science KAKENHI grants 15K14384 (to H. Saya) and 15K06840 (to E. Sugihara) as well as by a Keio University Grant-in-Aid for Encouragement of Young Medical Scientists (to E. Sugihara).
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.