CTLA4 Promotes Tyk2-STAT3–Dependent B-cell Oncogenicity

CTL–associated antigen 4 (CTLA4) is well recognized as an immune checkpoint, and has emerged as a prominent target for cancer immunotherapy (1, 2). CTLA4-blocking antibodies, along with PD1 and PD-L1–blocking antibodies, are capable of unleashing antitumor immune responses with durable cancer regression (1, 2). However, despite being one of the most potent anticancer drugs, CTLA4-blocking antibodies are unable to significantly prolong the lives of majority of the treated patients, suggesting an urgent need to further understand CTLA4 biology in cancer, thereby enabling the development of rational combinatory approaches to optimize the antitumor efficacy of CTLA4-blocking antibodies.

The mechanism by which CTLA4 dampens T-cell responses has been attributed to the fact that CTLA4 shares identical ligands, B7-1 (CD80)/B7.2 (CD86; refs. 3, 4) on antigen-presenting cells, with T-cell costimulating receptor CD28. However, whether and how CTLA4 may dampen T-cell activation through cell-intrinsic mechanism remains unknown. In addition, although it is considered expressed exclusively by T cells, there are some indications that CTLA4 is expressed by certain malignant B cells (5). If CTLA4 is consistently and highly expressed by B cells in the tumor microenvironment, it would suggest that B cells could also dampen T-cell activation by competing with CD28 for engaging B7-1 (CD80)/B7.2 (CD86) on antigen-presenting cells. However, these concepts have not been formerly tested.

A critical role of tumor-associated B cells in promoting cancer survival/resistance to therapies as well as immunosuppression has been reported (6–13). Among several mechanisms, STAT3 has been shown to mediate the cancer promoting activities of tumor-associated B cells (12, 13). STAT3 is persistently activated in diverse cancers, including many B-cell malignancies (14, 15). STAT3 is critical for upregulating the expression of numerous genes involved in cancer cell survival/proliferation, and invasion (16). A standout feature of STAT3 in cancer is that it also promotes expression of an array of immunosuppressive genes while inhibiting many Th1 immunostimulatory genes necessary for, inducing antitumor T-cell immunity (16–18). STAT3 activity in malignant B cells has been shown to inhibit the antigen presentation ability of these cells (19). STAT3 is persistently activated in diverse immune subsets in the tumor microenvironment, including myeloid cells, B cells, as well as T-cell, inducing immunosuppression and promoting tumor growth (4, 12–15, 20). Nevertheless, the upstream molecules/receptors that activate STAT3 in malignant B cells and in tumor-associated “normal” B and T cells remain to be further explored. In this study, we investigated the potential role of CTLA4 in B cells in promoting tumor progression. Our studies identified a cell-intrinsic immunosuppressive pathway for CTLA4 and an unexpected function of CTLA4 in promoting tumor cell growth and survival.

For subcutaneous tumor challenge, C57BL/6, Balb/c (The Jackson Laboratory) or athymic nude mice (NCI Frederick), were injected with 10⁵ B16 melanoma or 2.5 × 10⁵ A20 lymphoma, respectively. Athymic nu/nu mice (NCI Frederick) were engrafted with 2 × 10⁶ Ly3 human lymphoma cells subcutaneously into the flank. After tumors reached 5 to 7 mm in diameter, treatment with 250 μg/dose/mouse CTLA4 blocking antibody (BioXCell) was locally administered every other day.

Human B-cell lymphoma Ly3, U266 cells (kindly provided in 2010 by Dr. Ana Scuto, Beckman Research Institute at the Comprehensive Cancer Center at the City of Hope, Duarte, CA), Daudi, JeKo-1, SU-DHL-6, Raji and RPMI6666 cells (ATCC obtained in 2016) were cultured in IMDM or RPMI medium (Gibco), respectively, human multiple myeloma MM.1S (kindly provided in 2016 by Dr. Stephen Forman, Comprehensive Cancer Center at the City of Hope) and H929 (ATCC) were cultured in DMEM medium supplemented with 10% FBS (Sigma) and 0.05 mol/L mercaptoethanol. Mouse DC2.4 dendritic cells (kindly provided in 2008 by Dr. Marcin Kortylewski, Beckman Research Institute at the Comprehensive Cancer Center at the City of Hope), A20 B cell lymphoma (ATCC obtained in 2009), and mouse B16 melanoma (kindly provided in 2007 by Dr. Drew Pardoll, The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins School of Medicine, Baltimore, MD) were grown in RPMI1640 (Gibco) containing 10% FBS. Mouse RAW264.7 macrophages (ATCC, obtained in 2010) were cultured in DMEM supplemented with 10% FBS. Cells used in this study were routinely freshly thawed, subcultured for up to 3 weeks for desired in vitro studies or in vivo engraftment, tested for mycoplasma contamination and authenticated by RT-PCR and flow cytometry. Cell subculture was immediately amplified for long-term storage in liquid nitrogen.

Mouse care and experimental procedures with mice were performed under pathogen-free conditions in accordance with established institutional guidance and approved Institutional Animal Care and Use Committee protocols from the Research Animal Care Committees of the City of Hope.

This study was performed in accordance with the Helsinki principles and approved by the Institutional Review Board at City of Hope Medical Center (IRB14225). Informed written consent was obtained. The human tumor samples were evaluated by physicians at Department of Pathology of City of Hope. Detailed information is summarized in Tables 1 and 2.

To generate BA/F3 cell lines stably expressing human CTLA4 constructs, murine pro-B-cell line BA/F3 was grown in IL3 containing RPMI1640 medium containing 10% FBS, 10 ng/mL IL3 or 10% conditioned medium of WEHI-3B cell line. Mouse WEHI-3B cells were grown in Iscove's MDM supplemented with 5% to 10% FBS, 2 mmol/L l-glutamine, and 2.5 × 10⁻⁵ mol/L mercaptoethanol. Human CTLA-GFP constructs were introduced by electroporation. Briefly, 3.5 × 10⁶ BA/F3 cells were resuspended in 800 μL cell culture media containing 28 μg vector. Cells were pulsed with 200 V for 70 msec and subcultured.

Human B-cell lymphoma Ly3 cells with knocked down human CTLA4 expression were generated using lentiviral shRNA particles obtained from Santa Cruz Biotechnology. Cellular introduction of shRNA was carried out according to the manufacturer's instructions.

Plasmid coding for mouse CD86-mCherry was obtained from (GeneCopoeia). Plasmid encoding human CTLA4-GFP was purchased from OriGene (RG210150). Site directed mutagenesis was performed using QuickChange (Stratagene), resulting in hCTLA4 constructs hCTLA4-Y201F (5′-ctcttacaacaggggtctttgtgaaaatgccccca-3′; 5′-tgggggcattttcacaaagacccctgttgtaagag-3′) and Y218F (5′-gcaatttcagcctttttttattcccatcaatacgcgtacg-3′; 5′-cgtacgcgtattgatgggaataaaaaaaggctgaaattgc-3′).

Human CD86 gene was obtained from DNASU plasmid repository (clone: HsCD00039473). Soluble human CD86-Fc gene in pVL1393 vector was transfected into Sf9 cells with BestBac 2.0 Baculovirus Cotransfection kit (Expression Systems). High titer virus was generated and used to infect Tni cells at an MOI of 3 for protein production. Cells were harvested 48 hours postinfection, centrifuged at 4,000 rpm for 25 minutes, and the filtered supernatant was applied to a Protein A resin (GenScript). After PBS wash, protein was eluted with 0.1 mol/L glycine, pH 3.0 and immediately pH adjusted with 1 mol/L Tris-HCl pH 8.0. Concentrated eluate was applied to HiLoad 26/60 Superdex 200 column (GE Healthcare) in PBS. Peak fractions were concentrated, flash frozen, and stored at −80° C. Purity was monitored by SDS-PAGE.

Generated and purified human sCD86 was fluorescently labeled. Briefly, peptide diluted in 200-μL PBS was activated with a 1:10 dilution of 1 mol/L NaHCO₃ (20 μL), mixed with a grain of NHS coupled AlexaFluor 647 (Invitrogen) dissolved in 2 μL DMSO (Sigma), and incubated light protected at room temperature for 1 hour up to 1.5 hours. Gel filtration column was packed with G75 Sephadex (GE Healthcare) and fluorescently labeled sCD86 peptide was eluted by centrifugation for 5 minutes at 1,100 × g.

Indirect immunoflourescence and IHC were carried out as described previously (19) staining CD3, CD20 (BioLegend), CTLA4, c-Myc, pSTAT3 (Santa Cruz Biotechnology), Hoechst 33342 (Sigma), Ki67 (Vector), CD19, CD31 (BioLegend, BD Biosciences), pTyk2 and cleaved caspase-3 (Cell Signaling Technology). CFSE was purchased from Invitrogen and CFSE loading into cells was carried out according to the manufacturer's instructions. Imaging was carried out on a confocal microscope Zeiss LSM510 Meta.

Cell suspensions isolated from tissue were prepared as described previously (20) and stained with different combinations of fluorophore-coupled antibodies to CD3, CD4, CD8, CD19, CD28, CD62L, CD69, CD80, CD86, B220, CTLA4, phospho-Tyr705-Stat3, FoxP3, IFNγ, IL4 (BD Biosciences). Antibodies against c-Myc and pTyk2 were purchased from Cell Signaling Technology; staining was performed using a fluorescently labeled secondary antibody (Invitrogen). Fluorescence data were collected on Accuri or Fortessa flow cytometers (BD Biosciences) and analyzed using FlowJo software (Tree Star).

Whole cell lysates were prepared using RIPA lysis buffer containing 50 mmol/L Tris (pH 7.4), 150 mmol/L NaCl, 1 mmol/L EDTA, 0.5% NP-40, 1 mmol/L NaF, 15% glycerol, and 20 mmol/L β-glycerophosphate. A protease inhibitor cocktail was added fresh to the lysis buffer (Mini Protease Inhibitor Cocktail, Roche). Normalized protein amounts were subjected to electrophoretic separation by SDS-PAGE, transferred onto nitrocellulose for Western blotting, and subsequently immunodetection was performed using antibodies against STAT3, Tyk2, PY99 (Santa Cruz Biotechnology), anti-pTyr (clone 4G10, Millipore) and β-actin (Sigma). For coimmunoprecipitation, CTLA4, JAK1, JAK2, JAK3, Tyk2 antibodies (Santa Cruz Biotechnology) were used to label rProtein G agarose beads (Invitrogen), subsequently incubated for 16 hours with whole-cell lysates, subjected to electrophoretic protein separation and Western blot detection.

Nuclear extracts from cells were isolated using buffer A containing 10 mmol/L HEPES/KOH pH 7.9, 1.5 mmol/L MgCl₂, 10 mmol/L KCl and buffer C containing 20 mmol/L HEPES/KOH pH 7.9, 420 mmol/L NaCl, 1.5 mmol/L MgCl₂, 0.2 mmol/L EDTA, 25% glycerol; per 2 mL buffer, protease inhibitors at 0.2 mmol/L PMSF, 0.5 mmol/L DTT, and 1 mmol/L Na₃VO₄ were added fresh before use. Cells were washed with PBS, resuspended in buffer A, incubated on ice for 20 minutes and sedimented by centrifugation for 20 seconds at 13.2 rpm in a table-top centrifuge. Pellet was resuspended in buffer C, incubated for 30 minutes on ice and sedimented by centrifugation for 10 minutes at 13.2 rpm. Double-stranded DNA SIE oligo (5′-AGCTTCATTTCCCGTAAATCCCTA-3′/3′AGTAAAGGGCATTTAGGGATTCGA-5′ containing STAT1and STAT3 consensus binding site was radiolabeled with ³²P-ATP/³²P-CTP using Klenow enzyme (Promega). Nuclear extracts were resuspended at 10 μg with loading buffer (50 mmol/L HEPES pH 7.8, 5 mmol/L EDTA pH 8, 25 mmol/L MgCl₂ adjusted to pH 7.8 with 3 mol/L KOH) containing radiolabeled SIE-oligo and separated by PAGE electrophoresis; dried gel was exposed on X-ray film to assess STAT3 DNA binding. For supershift analysis, αSTAT3 antibody (C-20X, Santa Cruz Biotechnology) was added to nuclear extract at 1 μL/20 μL and incubated on ice for 15 minutes before loading onto PAGE for electrophoretic separation.

Transcript amplification was determined from total RNA purified using the RNeasy Kit (Qiagen). cDNA was synthesized using the iScript cDNA Synthesis Kit (Bio-Rad). Real-time PCR was performed in triplicates using the Chromo4 Real-Time Detector (Bio-Rad). The human GAPDH housekeeping gene was used as an internal control to normalize target gene mRNA levels. Primers were obtained from SA Biosciences (human BCL2L1: PPH00082B-200, human MMP9: PPH00152E-200) or customized from Integrated DNA Technologies IDT (human IL6: hIL6 F: 5′-GTACATCCTCGACGGCATC-3′, R: 5′-CCTCTTTGCTGCTTTCACAC-3′, human IL10: hIL10 F: 5′-TGCCTAACATGCTTCGAGATC-3′, R: 5′-GTTGTCCAGCTGATCCTTCA-3′, human IFNγ: hINFG F: 5′-GAGATGACTTCGAAAAGCTGAC-3′, R: 5′-CACTTGGATGAGTTCATGT ATTGC-3′).

Statistical analyses were performed using Prism (GraphPad) software. The overall significance for each graph was calculated using the two-tailed Student t test. P values of less than 0.05 were considered statistically significant.

To date, CTLA4 regulatory functions are considered only in T cells (2). However, it has been suggested that CTLA4 is also expressed in certain malignant B cells (5). We therefore assessed CTLA4 expression in patient B-cell lymphoma biopsies. We observed considerably elevated CTLA4 expression by tumor infiltrating CD3⁺ T cells as well as in CD20⁺ cells in human B-cell lymphoma tissues (Fig. 1A, top). Compared with normal lymph node, expression of CTLA4 is significantly increased in lymph node with B-cell lymphoma (Fig. 1A, bottom). We also assessed CTLA4 expression in two main types of human NHL lymphomas, diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma (FL; Tables 1 and 2). We show that CTLA4 is detectable in both types of NHL lymphomas (DLBCL, 81% and FL, 36%; Fig. 1B).

CTLA4 is also expressed in tested cell lines derived from human B malignancies, including Ly3 (DLBCL; Fig. 1C; Supplementary Fig. S1A and S1B) and human multiple myeloma cell lines (Supplementary Fig. S1). CTLA4⁺ B-cell lymphoma cells rapidly engaged with soluble CD86 (sCD86; Fig. 1D; Supplementary Fig. S1C), allowing CD86 cellular internalization (Fig. 1E). Incubating murine RAW macrophages expressing fluorescently labeled full-length CD86-mCherry with mouse B-cell lymphoma A20 cells loaded with CFSE resulted in a CD86-mCherry⁺ A20 B-cell lymphoma population, as shown by confocal microscopy (Fig. 2A). Flow cytometric analysis validated cellular internalization of CD86-mCherry by the A20 B-cell lymphoma cells cocultured with CD86-mCherry⁺ RAW macrophages or DC2.4 dendritic cells (Fig. 2B). Because CD28 is not expressed by murine A20 B-cell lymphoma, it can be excluded from competing with CTLA4 for B7 molecule engagement and cellular internalization under the experimental conditions (Fig. 2C). Blocking CTLA4 using a CTLA4 blocking antibody resulted in considerably reduced uptake of sCD86 by human B-cell lymphoma Ly3 and Raji cells, indicating that CTLA4 contributes to CD86 cellular internalization (Fig. 2D).

To investigate the intracellular tyrosine domain(s) of CTLA4 involved in CTLA4-mediated cellular internalization of CD86, we generated cell lines stably expressing various human CTLA4 constructs, particularly those with mutated tyrosines in the cytoplasmic tail of CTLA4. Incubating the CTLA4-expressing B cell lines with human sCD86, we observed that membrane distal Y218 in CTLA4 was more critical in the ligand internalization compared to the membrane proximal Y201 (Fig. 3). Moreover, mutated CTLA4-Y201F increased sCD86 internalization (Fig. 3). However, CTLA4-Y218F affects ligand internalization in a dominant manner because ligand uptake by double-mutation Y201F/Y218 in CTLA4 was comparable with ligand uptake by single-mutation Y218F in CTLA4 (Fig. 3, bottom). These results, taken together, suggest that CTLA4 expressed on malignant B cells can interact with and internalize CD86, thereby inhibiting T-cell activation by competing with T-cell costimulating molecule CD28.

Stimulation of human B-cell lymphoma Ly3 cells with soluble CD86, a critical factor driving B-cell lymphoma disease progression (21), resulted in immediate CTLA4 tyrosine phosphorylation and STAT3 recruitment by CTLA4 (Fig. 4A). Although the intracellular signaling pathways of CTLA4 are not well defined, a potential involvement of the JAK2 tyrosine kinase was indicated in T cells (22). We showed that sCD86 distinctly stimulated tyrosine phosphorylation of the JAK family member, Tyk2 (Fig. 4B), as well as induced Tyk2 recruitment to form a signaling complex with CTLA4 (Fig. 4C). CTLA4 ligation with CD86 resulted in STAT3 tyrosine phosphorylation (Fig. 4D), and induced the DNA-binding activity of STAT3, which is critically required for target gene transcription (Fig. 4E and F). Because STAT3 is well known for its role in promoting tumor immunosuppression and inhibiting Th1 antitumor immune responses, we assessed whether stimulation of B-cell lymphoma Ly3 cells with sCD86 would lead to expression of its known downstream immune-modulatory genes. Stimulating Ly3 cells with sCD86 resulted in induction of STAT3 downstream immunosupressive genes, such as IL10 and IL6, as well as inhibition of IFNγ expression (Fig. 4F). At the same time, CTLA4 ligation with CD86 caused upregulation of STAT3 downstream cancer-promoting genes in B lymphoma cells, such as BCL2L1 and MMP9, as assessed by RT-PCR (Fig. 4F). Moreover, we were able to demonstrate that sCD86-induced STAT3 activation was considerably decreased upon CTLA4 blockade in human B-cell lymphoma Ly3 cells (Fig. 4G). In addition, CTLA4 blockade resulted in significantly reduced expression of STAT3 target genes tested in various human B-cell lymphoma cell lines (Fig. 4H). Data shown in Fig. 3 have identified an unexpected role of CTLA4 in promoting tumor cell survival and proliferation. In addition, CTLA4 intracellular signaling through Tyk2-STAT3 promotes expression of immunosuppressive genes while inhibiting the production of Th1 immunostimulatory molecules.

Elevated JAK-STAT3 signaling in tumor cells, including many types of B lymphomas, has been demonstrated to promote tumor cell proliferation, survival, and resistance to apoptosis (14, 18, 23, 24). We therefore assessed whether CTLA4-CD86 ligation would increase B-cell lymphoma tumor cell proliferation. CFSE⁺ A20 lymphoma B cells cocultured with CD86-mCherry–expressing macrophages or dendritic cells diluted the fluorescent intensity of CFSE dye loaded into lymphoma B cells, indicating induced lymphoma cell division/proliferation by CD86. Conversely, nonproliferative CFSE^high lymphoma cells had low CD86-mCherry signal (Fig. 5A). These findings are indicative of a direct correlation between CD86 internalization and mitotic activity of lymphoma B cells in vitro.

CTLA4 antibody blockade in vivo, employed to inhibit CTLA4 interaction with CD86, significantly reduced tumor growth in a syngenic A20 B-cell lymphoma tumor model (Fig. 5B). CTLA4 antibody treatment also activated T cells (Supplementary Fig. S2). Importantly, Ki67⁺ proliferative activity was significantly reduced in tumors treated with CTLA4 blocking antibodies (Fig. 5C).

Moreover, inhibiting CTLA4 by either silencing CTLA4 in human lymphoma tumor cells or treating with CTLA4 blocking antibodies significantly reduced B-cell lymphoma tumor growth in mice lacking T cells and B cells (Fig. 5D–F). Importantly, CTLA4-blockade in human B-cell lymphoma considerably reduced activation of Janus kinase Tyk2 and recruitment of STAT3 by CTLA4 (Fig. 5G), as well as significantly diminished Ki67⁺ proliferative activity and increased tumor cell apoptosis, which was also associated with disruption of CD31⁺ tumor vasculature (Fig. 5H). We therefore show that CTLA4 ligation with CD86 promotes B-cell lymphoma tumor growth, which is associated with Tyk2-STAT3 activation induced by CTLA4. These results provided a molecular mechanism by which CD86 drives B-cell lymphoma progression.

A critical role of the tumor-associated B cells in cancer has been demonstrated in previous pioneering studies (6–11). The oncogenic effects of tumor-associated B cells are contributed by STAT3 activity (12, 13). We therefore examined the possibility that CTLA4 is expressed by tumor-associated CD19⁺ B cells and that signaling via Tyk2-STAT3 is operative in the tumor-associated B cells, thereby promoting tumor growth. Flow cytometry analysis of tumor-infiltrating B cells showed that CTLA4 was expressed by the B cells enriched from B16 tumors (Supplementary Fig. S3). Treating B16 melanoma tumor-bearing mice with CTLA4 antibodies significantly inhibited tumor growth (Fig. 6A). Expression of pTyk2, pStat3 and c-Myc by tumor-associated CD19⁺ B cells was decreased upon CTLA4 blockade in vivo as assessed by flow cytometry (Fig. 6B). The decrease in c-Myc expression in B16 melanoma infiltrating CD19⁺ B cells upon administration of CTLA4 blocking antibody was confirmed by confocal microscopy (Fig. 6C). Furthermore, CTLA4 blockade improved the infiltration of activated CD8⁺CD69⁺ T cells into tumor tissue and induced the downregulation of CD62L by CD3⁺ T cells in the tumor environment (Fig. 6D and E; Supplementary Fig. S2).

Moreover, CTLA4 blockade treatment resulted in activation of CD19⁺ B cells (nonmalignant) in tumor-draining lymph nodes in the A20 subcutaneous tumor-bearing mice (Fig. 6F, top). Notably, the tumor-promoting CD5⁺CD19⁺ B-cell population (13) was considerably decreased upon CTLA4 blockade in vivo (Fig. 6F, bottom). Our results with B16 melanoma and A20 lymphoma show that in addition to suppressing T-cell activation, CTLA4 signaling also negatively impacts tumor-associated B-cell antitumor activity.

Although our studies focused on the role of CTLA4 in B cells in cancer, they shed light on fundamental functions of CTLA4 in B cells. By internalizing CD86 expressed on antigen-presenting cells, CTLA4 in B cells can downmodulate T-cell Th1 immune responses. Our study has identified a novel cell-intrinsic pathway by CTLA4 to suppress Th1 immunity through STAT3. During normal physiology, inhibition of Th1 immunity is a prerequisite of wound healing, which involves cell proliferation, resistance to apoptosis, and angiogenesis. The processes of wound healing are the same as those in cancer. STAT3 is known to regulate wound healing and its persistent activation is critical for oncogenesis. Our results reveal that CTLA4 not only is critical for downmodulating immune responses but also promotes cell proliferation, survival, and angiogenesis. STAT3 activation in tumor-associated immune cells, including B cells promotes production of growth factors and other mediators to enhance tumor cell growth (12–14).

We show that upon engagement with CD86, CTLA4 recruits and activatesTyk2, which is reminiscent of the interaction between a cytokine receptor and JAK. Through both genetic silencing and antibody blockade, our work suggests that CTLA4 is a target in B-cell lymphoma tumor cells and in tumor-associated B cells for cancer therapy. However, the potency of the antitumor effects by anti-CTLA4 antibody therapy, compared with CTLA4 gene silencing, in the B-cell lymphoma xenograft tumor model in the absence of T cells and B cells is not dramatic. This could be due to the fact that CTLA4 is also expressed in the cell cytoplasm (5) in addition to cell surface expression. Our results further suggest that CTLA4 blockade in conjunction with STAT3 inhibition should increase CTLA4 immunotherapy, and CTLA4 blockade treatment for B-cell lymphoma has the added advantage of directly inhibiting tumor cell growth/resistance to apoptosis.

No potential conflicts of interest were disclosed.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Conception and design: A. Herrmann, H. Lee, K. Jenkins, T. Blankenstein, H. Yu

Development of methodology: A. Herrmann

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Herrmann, C. Lahtz, J.Y. Song, W.C. Chan, C. Yue, T. Look, R. Mülfarth, W. Li, J. Williams, L.E. Budde

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Herrmann, T. Nagao, W.C. Chan, C. Yue, W. Li

Writing, review, and/or revision of the manuscript: A. Herrmann, T. Nagao, J.Y. Song, L.E. Budde, S. Forman, L. Kwak, H. Yu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Herrmann, J.Y. Song

Study supervision: A. Herrmann, H. Yu

We thank the dedication of staff members at the flow-cytometry core and light microscopy core at the Beckman Research Institute at City of Hope Comprehensive Cancer Center for their technical assistance. We also acknowledge the contribution of staff members at the animal facilities at City of Hope.

This work was supported by R01CA122976, R01CA146092, P50CA107399, the Tim Nesvig Lymphoma Society, V Foundation Translational Research Grant, and by the National Cancer Institute of the NIH under grant number P30CA033572.