Eμ-TCL1xMyc: A Novel Mouse Model for Concurrent CLL and B-Cell Lymphoma

Aggressive B-cell lymphoma is a highly heterogeneous disease entity with subtypes characterized by molecular, genetic, and epigenetic signatures (1, 2). Increased Myc activity is a major oncogenic factor in several aggressive B-cell lymphomas, including Burkitt lymphoma, diffuse large B-cell lymphoma (DLBCL), and high-grade B-cell lymphomas (HGBCL; ref. 3). Aberrant Myc expression is also observed in patients with Richter syndrome (4–6), an aggressive B-cell lymphoma arising from chronic lymphocytic leukemia (CLL) that occurs in around 5% of patients with CLL (7). Multiple studies link increased Myc expression in lymphomas to inferior survival (8–11).

Novel targeted agents such as the Bruton's tyrosine kinase (BTK) inhibitor ibrutinib and the Bcl2 inhibitor venetoclax have shown remarkable activity in CLL, but they have generally been less effective in aggressive B-cell malignancies (12, 13). The sensitivity of DLBCL to ibrutinib is largely determined by specific genetic lesions associated with DLBCL subtypes, producing overall response rates (ORR) of 40% in activated B‐cell (ABC) subtypes, but only 5% in germinal center B‐cell (GCB) DLBCL (14). Alternative treatment approaches that circumvent B-cell receptor (BCR) and apoptotic pathway-mediated resistance mechanisms are therefore of increasing clinical importance. Our group previously demonstrated efficacy of the eukaryotic nuclear export protein exportin-1 (XPO1) inhibitor selinexor in settings of ibrutinib resistance (15). Other studies further support that XPO-1 inhibition has significant antitumor activity in pretreated or relapsed/refractory patients with hematologic malignancies (16). Next-generation XPO-1 inhibitors currently in development, such as KPT-8602, exhibit similar potency but substantially lower systemic toxicities (17–19).

Genetically engineered murine models are powerful tools to investigate the biology and therapeutic responses of lymphoid malignancies (20). Commonly used models of aggressive lymphoma are largely driven by overexpression of Myc (20). Myc under the lymphoid-specific immunoglobulin heavy chain (IgH) enhancer (Eμ) gives rise to a variety of B-cell lymphomas ranging from early-onset immature Burkitt-like to late-onset DLBCLs (21). Limitations of this model include heterogeneous phenotypes and gene expression profiles and variable responsiveness to therapy (22, 23). Recent studies suggest that lymphomagenesis in Eμ-Myc mice is also determined by Myc-cooperative mutations and concomitant multigenic lesions involving Cdkn2a, Nras, and Kras, suggesting a two-hit pathogenesis (24). Combinations of Myc with other oncogenes such as Ras (25), oncogenic EBV components (26), or c-Myc-surface CD19 signaling amplification loops (27) have been shown to accelerate lymphoma development. However, the suitability of these models to mirror therapeutic responses to targeted treatments has not been adequately tested.

The transgenic Eμ-TCL1 mouse model is the most widely used preclinical in vivo platform for CLL and generally mirrors tumor-induced immune defects and therapeutic responses of aggressive unmutated human CLL (28). Occasional Richter syndrome transformation was observed in some immunocompetent Eμ-TCL1 mice as well as Eμ-TCL1 mice with conditional B-cell–specific TRP53 deficiency, leading to the occurrence of highly proliferative large blastoid cells in splenic infiltrates and blood (29); however, a model of simultaneous CLL and aggressive lymphoma does not exist. Given the poor survival seen in patients with Richter syndrome, a model for testing therapeutics in the setting of both CLL and aggressive lymphoma would be of high interest.

We crossed Eμ-TCL1 with Eμ-Myc mice to mimic the effects of Myc overexpression in the context of CLL. Eμ-TCL1xMyc mice uniformly developed a highly aggressive lymphoid disease with features of distinct CLL and lymphoma components. While mice failed to respond to BTK inhibition with ibrutinib, durable disease control was observed with the XPO1 inhibitor KPT-8602. This provides in vivo proof-of-concept that Eμ-TCL1xMyc mice can serve as a therapeutic platform to test agents simultaneously against CLL and aggressive lymphoma.

All animal experiments were performed under protocols approved by The Ohio State University Institutional Laboratory Animal Care and Use Committee. Eμ-TCL1 transgenic mice developing CLL and Eμ-Myc transgenic mice developing B-cell lymphoma have been described (21, 28). c-Myc and wild-type (WT) mice were purchased from Jackson Laboratories. C57BL/6J females were crossed with Eμ-Myc hemizygous males to produce Eμ-Myc hemizygous or non-transgenic (nTG) pups. Eμ-TCL1-B6 homozygous females were crossed with Eμ-Myc hemizygous males to produce Eμ-TCL1B6 hemizygous/Eμ-Myc hemizygous (abbreviated to Eμ-TCL1xMyc or double transgenic (dTG)) or Eμ-TCL1B6 hemizygous/nTG pups. To produce genetic models of inactive BTK, Eμ-Myc mice were crossed with homozygous X-linked immunodeficiency (XID; ref. 30) and with Eμ-TCL1xXID mice (31). Predefined euthanasia criteria included lethargy, difficulty walking due to spleen size, lymph node masses ≥1.6 cm, or loss of ≥20% body weight. Survival of transgenic mice was assessed in all transgenic colony mice born within a 24-month period.

Organs were harvested from diseased single and double transgenic colony mice and diseased mice after injection of Eμ-TCL1xMyc spleen cells (meeting euthanasia criteria defined above). Tissues were fixed and processed as described in Supplementary Methods. Staining with hematoxylin and eosin (H&E; Leica) and Ki67 (Dako, clone TEC-3) IHC were described previously (32). For confirmatory bone marrow aspirate preparation, femurs were harvested from WT (n = 3) and diseased dTg (n = 3) mice. After air-drying, aspirates were fixed and stained using a modified Wright stain (Heme-3 staining set, Fisherbrand) as per the manufacturer's recommendation. Photographs were taken using an Olympus SC30 camera with an Olympus BX53 microscope.

Flow optimizations and protocols are detailed in Supplementary Methods. Antibodies are listed in Supplementary Table S1. Murine markers of B-cell development and gating strategies followed published data and technical resource publications (Supplementary Fig. S1; refs. 33–35). The gating strategy and graphs from a representative WT spleen are shown in Supplementary Fig. S2. All gates were set on fluorescence-minus-one (FMO) controls prepared for each individual experiment. B-cell phenotype and percentage comparisons in marrow, spleen, and blood were conducted in 8 Eμ-TCL1, 6 Eμ-Myc, and 9 Eμ-TCL1xMyc (all disease requiring euthanasia), and 5 WT mice matched to the age of diseased dTg mice. B-cell light-chain expression in spleen was investigated in an additional set of diseased Eμ-TCL1xMyc (n = 8) and nontransgenic (nTG) age-matched littermates (n = 3). A third set of spleen cells from 8 diseased Eμ-TCL1xMyc mice was stained with intracellular c-Myc and TCL1-A/respective isotypes after B-cell surface staining, and corrected median fluorescence intensities (MFI, MFI TCL1/Myc MINUS MFI isotype PE/AF488) were compared with those in WT spleen B cells (n = 2). Only samples with >60% viability were included into statistical analyses. All data were analyzed using KALUZA software (Becton Dickinson).

Freshly harvested spleens from diseased dTG mice meeting euthanasia criteria were processed as described in Supplementary Methods. A dose of 1 × 10⁷ viable splenocytes, prepared from a pool of diseased dTG donors, was injected into the tail vein of B6 WT recipient mice (n = 10) at approximately 2 months of age. All cell suspensions contained ≥75% CD19⁺ B cells verified by flow cytometry. Recipients were monitored by standard hematology testing and euthanized following the criteria described above.

CD19⁺CD5⁺ CLL-like, CD21⁻IgM⁻ and CD21⁻IgM⁺ cells were sorted from a spleen single-cell suspension pool of diseased dTg mice (n = 3) that met euthanasia criteria. A total of 4.1 × 10⁸ spleen cells were obtained, stained, and gated as before. Flow sorting was performed on a BD FACSAriaIII sorter in purity mode, yielding a total of 1 × 10⁶ viable CLL-like cells and 3 × 10⁶ viable CD21⁻IgM⁻ and CD21⁻IgM⁺ cells each, which were injected into tail veins of nTG littermates (n = 3/population). Recipient mice were then monitored every 7–14 days using the immunophenotyping panel and gating strategy above. Mice were euthanized once they exhibited clinical signs of overt disease, or were censored at the end of the experiment after 136 days after injection, and B-cell phenotype and light-chain expression were examined by flow cytometry.

CD19⁺CD5⁺ CLL-like, CD21⁻IgM⁻ and CD21⁻IgM⁺ cells were flow sorted as above from spleen suspensions from diseased dTg mice (n = 3) meeting euthanasia criteria. Sorted cells were captured in PBS/10%FBS on ice and washed. Pellets were resuspended in TRIzol and stored at −80°C until RNA isolation using a chloroform/alcohol extraction method. Total RNA was quantified using the Invitrogen Qubit RNA HS Assay kit (Invitrogen). The Clontech SMARTer v4 kit (Takara Bio USA, Inc.) was used for global preamplification. Illumina sequencing libraries were derived from the resultant cDNA using the Illumina Nextera XT DNA Library Prep Kit following manufacturer's instructions. Libraries were sequenced using an Illumina HiSeq 4000 sequencer paired-end 150 bp protocol to approximately 12 million passed filter clusters per sample. Data processing was performed according to the CLEAR workflow (36), which identifies reliably quantifiable transcripts in low-input RNA-seq for differentially expressed gene (DEG) transcripts using gene coverage profiles. MiXCR (v3.0.5; ref. 37) was used with default parameters to identify preprocessed reads containing CDR3 regions from B-cell heavy, kappa, and lambda chains, generating a list of unique CDR3 sequences associated with their relative abundances and specific V(D)J gene usage. Global DEG analysis was performed using DESeq2 (v1.20.0; ref. 38). The counts table containing CLEAR transcripts from all replicates was used for comparisons between groups. The resultant data were then used for gene-specific analysis for a targeted list of oncogenes (1, 2, 39, 40). Raw FASTQ files and counts tables were deposited in GEO (accession no: GSE129515).

Ibrutinib and 10% β-cyclodextrin vehicle were purchased and delivered via drinking water (31). KPT-8602 (provided by Karyopharm, Inc.) drug (15 mg/kg/day) and vehicle (0.5% methyl-cellulose/1% tween-80) were administered by daily oral gavage (18) to WT mice injected with 1 × 10⁷ viable splenocytes, prepared from a pool of diseased dTG donors. KPT-8602/vehicle treatment was initiated once a WBC count ≥8 and/or ≥5% circulating CD19⁺CD5⁺ CLL were detected. Disease progression was monitored weekly. Mice were euthanized following criteria outlined above.

Survival was shown using Kaplan–Meier curves and curves were compared using a log-rank test. Differences between cell subset frequencies were compared using nonparametric Wilcoxon–Mann–Whitney and one-way ANOVA (with Dunn multiple comparison) tests assuming unequal variances or small numbers in experimental groups. Myeloblasts, promyelocytes, and monocytes in bone marrow aspirates were compared using a Wilcoxon signed-rank test. R program and GraphPad Prism were used for statistical analyses.

Immunoblot analysis confirmed TCL1 and c-Myc expression in Eμ-TCL1xMyc B cells (n = 3; Fig. 1A). Eμ-TCL1xMyc mice developed a malignancy at a younger age than single transgenic Eμ-Myc or Eμ-TCL1 mice, leading to significantly shortened survival [median survival dTG mice 46 days (range 29–74 days) vs. Eμ-Myc 118 (42–520 days) vs. Eμ-TCL1 368 (188–531 days), P < 0.0001; Fig. 1B]. Respiratory distress, inguinal, axillary, and intestinal lymphadenopathy were frequently observed as the malignancy progressed, and key characteristics of diseased dTG mice at endpoint were pronounced cervical lymphadenopathy and splenomegaly (Fig. 1C). Further examination of tissues from WT and diseased transgenic mice characterized tumors in dTG mice as a lymphoid malignancy histologically indistinguishable from lymphoma in Eμ-Myc mice, but distinct from leukemia in Eμ-TCL1 mice, with infiltration of lymph nodes, liver, spleen, and marrow (Fig. 1D). While pronounced spleen and liver infiltration with neoplastic lymphoid cells was seen in all transgenic models, marrow, and lymph nodes were only heavily involved in Myc-driven models. Bone marrow aspirates were obtained from WT and dTG mice to confirm infiltration by large homogenous lymphoid cells in dTG compared with WT mice (Supplementary Fig. S3A and S3B), leading to a significant reduction and impaired maturation of the myeloid and erythroid lineages (Supplementary Fig. S3B–S3D). Taken together, dTG mice develop an aggressive lymphoid malignancy that infiltrates bone marrow and lymphoid organs and is histologically similar but clinically different to aggressive lymphoma seen in Eμ-Myc mice.

To characterize the model biologically, we first compared B-cell immunophenotypes and relative frequencies (percentages) from marrow, spleen, lymph node, and blood in WT and diseased transgenic mice, using established murine markers of B-cell development as shown in Supplementary Fig. S1 (33–35). Because of the similarity between spleen and lymph nodes, results are only shown for spleen. The gating strategy and representative flow plots are shown in Supplementary Fig. S2. We first compared broad B-cell maturation differences by evaluating IgM/IgD expression (Fig. 2A) among CD5⁻B220⁺ B cells, that is, after exclusion of CD5⁺ B cells with a CLL phenotype, which was expected to occur in mice carrying the TCL1 transgene. Figure 2B shows representative WT spleen IgM and IgD FMO controls. Diseased mice expressing the Myc transgene (both Eμ-Myc and dTG mice) had a distinct maturation block at early B-cell development stages, altering the presence and distribution of more mature B cells in secondary lymphoid organs and blood (Fig. 2C). As this arrest appeared to accumulate cells at the transitional B-cell stage (i.e., when B cells are developmentally intermediate between immature bone marrow and mature naïve peripheral B cells), we next compared the relative frequencies and phenotype of transitional B cells (i.e., CD5⁻CD93⁺B220⁺ cells), again after exclusion of CD5⁺ B cells with a CLL phenotype. Eμ-TCL1xMyc mice displayed a significant enrichment in all examined organs compared with WT mice and parental strains, even in peripheral blood, where such cells are usually transient and rare (Fig. 2D). However, this enriched population lacked the usual CD23 and IgM expression patterns that typically identify physiologic transitional B cells. Instead, two distinct IgM⁺ and IgM⁻ subpopulations were noticeable (Fig. 2E). Overall, our results show that B-cell–specific expression of TCL1 and Myc leads to the development of an immature atypical and heterogeneous B-cell population indicative of a B-cell malignancy, preventing normal physiologic B-cell maturation.

To further explore the B-cell maturation block in Eμ-TCL1xMyc mice, we next investigated the distribution and frequencies of CD93⁻ cells, that is, B cells that were not described above. This revealed a significantly expanded CD19⁺CD5⁺B220⁺ CLL-like population in diseased Eμ-TCL1xMyc mice that was absent in both WT and diseased Eμ-Myc mice (Fig. 3A and B, top). CD19⁺CD5⁺B220⁺ CLL-like cells were, however, less expanded than in diseased Eμ-TCL1 mice and infiltrated predominantly blood and spleen. CD19⁺CD5⁺ cells were also detected among B220^low cells in all examined organs (Fig. 3A and B, bottom), again at lower frequencies than in diseased Eμ-TCL1 mice. Among the remaining non-CLL maturing B cells (B220⁺CD19⁺CD5⁻), the frequencies of regular follicular and marginal zone/marginal zone progenitor cells were substantially reduced in diseased Myc-expressing models, further confirming the maturation arrest described above (Fig. 3C and D left panel for spleen, Supplementary Fig. S4 for marrow and blood). Instead, atypical B cells lacking CD21 expression were substantially increased in diseased dTG mice, suggestive of an immature malignant B-cell population. These B220⁺CD21⁻ cells exhibited again both an IgM⁺ and an IgM⁻ population in dTg mice, pointing toward heterogeneity within the malignant population, whereas diseased single-transgenic Eμ-Myc mice only had a single population with IgM^low/int expression (Fig. 3D, left and E). The heterogeneity of the dTG lymphoma population was further supported by differences in IgD expression (Fig. 3D right): while IgM⁻ cells also lacked IgD expression, indicative of immaturity of this population, IgM⁺ cells dimly co-expressed IgD, suggesting a more mature phenotype.

We next examined differences in light chain expression in CLL, CD21⁻IgM⁻ and CD21⁻IgM⁺ populations in 8 additional dTG mice and compared this to light chain usage in B cells from age-matched nTG littermates (n = 3). B cells from nTG littermates acquired light chain expression with increasing B-cell maturation with kappa usage predominance as expected (41) (Fig. 3F top row). Light chain expression was absent in most cells in CD21⁻IgM⁻ lymphoma populations in diseased dTG mice, confirming their immature phenotype, while CD21⁻IgM⁺ cells and CD19⁺CD5⁺ CLL populations expressed predominantly kappa light chains, confirming a more mature phenotype (Fig. 3F bottom row). All malignancy components showed higher protein expression of Myc and TCL1A than WT B cells (Supplementary Fig. S5A), with highest TCL1A expression in CD19⁺CD5⁺ CLL cells, and highest Myc expression in CD21⁻IgM⁻ and CD21⁻IgM⁺ cells. Similar patterns were seen on a molecular level (Supplementary Fig. S5B). Altogether, these findings suggest that dTG mice develop a malignancy with features of both CLL and lymphoma, with the lymphoma population consisting of heterogeneous subpopulation components characterized by differential IgM, light chain and oncogene expression.

We next performed RNA sequencing of flow-sorted malignant populations obtained from 3 diseased dTG mice to compare CLL-like, CD21⁻IgM⁻ and CD21⁻IgM⁺ B-cell populations. Post-alignment QC results are summarized in Supplementary Table S2. We first quantified the percentage of reads with heavy, kappa, and lambda chains compared to the number of total sequencing reads. The CD21⁻IgM⁻ showed the lowest and the CLL-like population the highest percentage of reads that aligned to the V(D)J region, indicating different amounts of BCRs in malignancy components (Fig. 4A; Supplementary Table S3). The percentage of aligned reads was significantly different between CD21⁻IgM⁻ and CD21⁻IgM⁺ (P = 0.012) and between CD21⁻IgM⁻ and CLL-like (P = 0.046) cells, but not between CD21⁻IgM⁺ and CLL-like populations (P = 0.13), suggesting a greater degree of similarity of BCR-expressing cells between CD21⁻IgM⁺ and CLL-like populations. The comparison of the top 20 most abundant CDR3 sequences and the frequency of remaining CDR3 sequences, jointly denotes as “others”, demonstrated a high degree of diversity in CD21⁻IgM⁻ cells, decreasing diversity in CD21⁻IgM⁺ cells and the lowest diversity in CLL-like populations, with light chains being more frequently used than heavy chains (Supplementary Fig. S6A; Supplementary Table S4).

To examine differences in V gene usage, BCR reads were binned by specific heavy- (H) and light-chain V genes. Each V gene was normalized by the total number of reads corresponding to that chain. A low number and similar gene usage was seen across all populations for lambda (L) V chains (Supplementary Fig. S6B; Supplementary Table S4). Kappa chain genes in contrast were highly diverse in CD21⁻IgM⁻ cells, and decreased in diversity in CD21⁻IgM⁺ and CLL-like populations, concurrent with the emergence of fewer and predominant clones (Fig. 4B; Supplementary Table S4). While replicate mice largely shared the same degree of diversity for each population, there was very little overlap of distinct kappa chain V genes between different malignant populations, indicative of limited clonal relationships. Very similar patterns were observed in heavy chain genes (Supplementary Fig. S6C; Supplementary Table S4). However, distinct differences were seen in transcriptome profiles of CLL-like, CD21⁻IgM⁻ and CD21⁻IgM⁺ populations (Fig. 4C; Supplementary Table S5). A number of oncogenes implicated in CLL and lymphoma pathogenesis (1–3, 39, 40) were solely upregulated in the CLL population, whereas HES1 and EZH2 were only overexpressed in CD21⁻IgM⁻ and CD21⁻IgM⁺ populations. Moreover, transcriptome profiles were very similar in CD21⁻IgM⁻ and CD21⁻IgM⁺ populations, with the exception of NOTCH2, where greater similarities existed between CD21⁻IgM⁺ and CLL. Altogether, these molecular patterns confirm differences in B-cell maturation and suggest limited clonal relationships between all malignancy components in the context of differentially expressed genetic profiles.

We next confirmed that the Eμ-TCL1xMyc malignancy was amenable to adoptive transfer by injecting pooled Eμ-TCL1xMyc spleen cells from diseased donors into B6 syngeneic WT mice (n = 10). This established a circulating malignancy 8–10 days after injection, as evidenced by increased WBC counts and/or percentages of CD19⁺CD5⁺ cells. Median survival was 38 days (range 37–42 days), and tumor histology was similar to that in dTG mice (Fig. 5A). To assess the malignant and transformation potential of different components of the dTG malignancy in vivo, we flow-sorted CD19⁺CD5⁺ CLL-like cells and CD21⁻IgM⁻ and CD21⁻IgM⁺ lymphoma cells from a pool of diseased dTg mice and injected them separately into nTG littermates (n = 3/population). For key characteristics of recipient mice and diseases after injection see Supplementary Table S6. All mice injected with more mature CD21⁻IgM⁺ cells died after 58–69 days with signs of fully developed lymphoid malignancy. Two of 3 mice injected with immature CD21⁻IgM⁻ cells uniformly developed lymphoid malignancy and were euthanized in poor overall condition after 58 days, with one mouse (C1457) being censored as a failed engrafter 136 days postinjection. Recipients of CLL cells survived considerably longer, potentially as a result of the lower cell dose they had received. A B220^low/+CD19⁺CD5⁺ CLL like population was detected in blood of all CLL-cell recipients, with varying spleen involvement and the least infiltration in marrow (Fig. 5B; Supplementary Fig. S7). The phenotype of remaining and relatively reduced B220⁺CD5⁻ maturing B cells was CD21^intIgM^−/int, consistent with a population of nonmalignant follicular B cells, and there was no evidence of B220⁺ cells with CD21/IgM expression patterns seen in dTG mice (Supplementary Fig. S8). Mice with disease after injection of immature CD21⁻IgM⁻ lymphoma cells developed a B220⁺CD5⁻CD19^intCD21^intIgM^−/int population, especially in spleen, representing follicular B cells, and a B220⁺CD5^intCD19⁺CD21⁻IgM⁻ population identical to the phenotype of engrafted lymphoma cells (Fig. 5C, left). CD21⁻IgM⁺ cells and a frank CD19⁺CD5⁺ CLL-like population were not detected (Supplementary Figs. S7 and S8). All recipients of CD21⁻IgM⁺ cells lacked CD19⁺CD5⁺ CLL-like cells, but developed a B220⁺CD5^−/intCD19^int/⁺ population (Fig. 5C, right; Supplementary Figs. S7 and S8). This could be further divided into CD21^intIgM^−/int follicular B cells, CD21⁻IgM⁻ lymphoma cells, and a considerably larger population of CD21⁻IgM⁺ lymphoma cells. The CLL-like population in CLL recipients and nonmalignant follicular B cells in all donor groups preferentially expressed kappa light chains (Fig. 5D). Kappa also remained the predominantly expressed light chain in all lymphoma populations in CD21⁻IgM⁺ recipients. In contrast, CD21⁻IgM⁻ lymphoma cells in CD21⁻IgM⁻ recipients largely lacked light chain expression. Altogether, although based on small numbers, these results demonstrate that recipients of a distinct component of the dTG malignancy develop a malignancy reflecting the characteristics of injected disease, providing in vivo support of the molecular patterns described above.

To characterize the dependence of a Myc-driven malignancy on BTK-mediated B-cell receptor signaling in vivo, Eμ-Myc mice were crossed with XID and Eμ-TCL1xXID mice. In contrast to the Eμ-TCL1 model, where crossing of TCL1 with XID mice with inactive BTK prolongs survival and reduces CLL burden (31), genetic BTK inactivity did not improve survival in Myc-expressing mice, suggesting Myc involvement in mediation of resistance to BTK inhibition [median survival Eμ-Myc mice (n = 109) 118 days (range 42–520) vs. Eμ-MycxXID mice (n = 37) 78 days (50–123), P = 1.0; Eμ-TCL1xMyc (n = 68) 46 days (29–74) vs. Eμ-TCL1xXIDxMyc (n = 38) 52 days (36–73), P = 0.172; Fig. 6A]. Intrinsic resistance to BTK inhibition was then confirmed pharmacologically. Compared with vehicle, treatment with ibrutinib from the date of weaning did not improve survival in either myc-carrying transgenic strains (P = 0.83 and P = 0.88, respectively; Fig. 6B and C). XPO-1 inhibitors have successfully been used in ibrutinib-refractory mice and cell lines (15, 18), partly by inhibition of activation of downstream B-cell receptor targets and suppression of BTK gene expression (32). As a proof of concept, we therefore treated WT mice with evidence of circulating disease after injection of spleen cells from a pool of fully diseased dTG donors with XPO-1 inhibitor KPT-8602 (n = 7) or vehicle (n = 10). Median survival in vehicle-treated mice was 25 days (range 17–29 days), whereas all KTP-8602 treated mice survived until day 75 when the study was terminated (P < 0.0001; Fig. 6D). After initiation of KPT-8602, WBC counts decreased throughout treatment, while continuing to rise in mice receiving vehicle (Fig. 6E, left), mostly driven by increasing percentages of circulating CD19⁺CD5⁺ CLL cells (Fig. 6E, right). Spleen weights were significantly lower (P = 0.0025) in KPT-8602 than in vehicle-treated mice (Fig. 6F). Lymph nodes and spleens were effaced with diffusely proliferating Ki67⁺ neoplastic lymphocytes in vehicle treated mice (Fig. 6G). This experiment not only confirms the suitability of the adoptive transplant Eμ-TCL1xMyc mouse model for therapeutic exploration, but provides additional support for the evaluation of XPO-1 inhibitors in aggressive lymphomas.

We here report a new double transgenic Eμ-TCL1xMyc–driven model of aggressive B-cell lymphoma that offers several advantages over existing lymphoma models. These include high penetrance of very aggressive disease, reproducible and consistent histology and immunophenotype, amenability to adoptive transfer, uniformly shortened survival duration, and possibility of studying this lymphoma in the context of coexistent CLL. Clinical responses to ibrutinib in DLBCL are largely determined by subtype and the presence of additional BCR mutations (14). Importantly, a recent phase III study reported that the addition of ibrutinib to standard immunochemotherapy did not improve event-free survival and increased toxicity in patients with DLBCL, suggesting that BCR pathway inhibition may not be a viable treatment strategy in this setting (42).

Extensive profiling studies have demonstrated that molecular alterations as well as genetic and epigenetic signatures can be used to classify different lymphoma subtypes and to predict responses to treatment (1). Similar strategies have been employed to characterize the patterns of genetic evolution leading from CLL to lymphoma transformation, and to characterize the genetic landscape in patients with CLL and clonally related lymphomas compared with CLL and de novo DLBCL (5, 6). The Eμ-TCL1xMyc mouse model reported here is based on the malignant potential of just two known oncogenes, TCL1 and Myc, which might be an oversimplification of the complex and dynamic landscape of genetic and epigenetic alterations. However, Myc dysregulation occurs in several subtypes of aggressive B-cell lymphomas, and has also been described in independent cohorts of RS patients (4–6), highlighting Myc as a major pathogenic factor. Lymphomas characterized by Myc in combination with BCL2 and/or BCL6 rearrangements are currently classified as HGBCL (3), and recent studies highlight the poor outcome of patients with HGBCL (8, 9). To further characterize the molecular signatures in the three malignancy components developing in Eμ-TCL1xMyc mice, we performed RNA sequencing and employed a candidate-gene approach to compare differential expression of oncogenes with implications in DLBCL, CLL, and RS (1, 2, 39, 40). Typical CLL-related oncogenes such as IRF4, Rb1, RelA, and CD79b were solely upregulated in the CLL population, whereas HES1, a regulator of Notch signaling, and EZH2, a molecular marker of the EZB genetic subtype (2), were only overexpressed in CD21⁻IgM⁻ and CD21⁻IgM⁺ lymphoma populations. While candidate genes were similarly expressed in the two lymphoma populations, we observed some overlap between the CLL and the IgM⁺ population for NOTCH2 expression. NOTCH2 was recently found to be a marker of the BN2 DLBCL subtype (2), and also controls nonautonomous Wnt signaling in CLL (43). Together, these molecular fingerprints confirm distinct properties of different malignant populations developing in Eμ-TCL1xMyc mice, although further studies will be needed for a comprehensive molecular characterization.

Moreover, the mechanisms by which TCL1 and Myc co-expression accelerate disease and lead to a phenotype of concurrent CLL and heterogeneous lymphoma are unknown and should be further explored. We observed higher TCL1 expression in the CLL and higher Myc expression in the lymphoma populations at both the protein and transcriptional levels, suggesting that the degree of disease expansion or the predominance of a disease component is primarily driven by one of the two oncogenes. However, TCL1 is a known activator of the PI3k–Akt (PKβ) oncogenic pathway (44) and also interacts with a variety of regulators of cell survival such as AP1 and Atm affecting NFκB and IKBα (45). Interactions between TCL1 and Myc and the mechanisms leading to changes in proliferation, cell cycle, and apoptosis are currently not understood, and might be different in the distinct malignancy components in our studies.

An important question is whether Eμ-TCL1xMyc mice can be proposed as a model of a specific aggressive B-cell lymphoma. To determine clonality patterns, and as a method to assess whether this model is of a lymphoma transformed from a CLL clone or two distinct co-occurring malignancies, we conducted V(D)J sequencing and adoptive transfer experiments of flow-sorted malignancy components. V(D)J sequencing confirmed substantial differences in the B-cell maturation stage, that is, the amount of cells expressing a complete BCR, and the diversity of V gene usage, with CD21⁻IgM⁻ cells being the least mature but most diverse and CLL-like cells being the most mature but least diverse population. Clear clonality patterns were absent, demonstrating that this is not a model of clonally-related RS. Taken together, though, our molecular data indicate that Eμ-TCL1xMyc mice are a compound model of aggressive lymphoma and CLL, mimicking at least one therapeutic response of patients with lymphoma subtypes unresponsive to BCR-targeted agents to XPO-1 inhibition.

XPO-1 inhibitors have demonstrated significant preclinical activity in various hematologic malignancies, even in the setting of ibrutinib resistance (15–19). As a result, several phase Ib/II clinical studies are currently being conducted in various malignancies, including CLL and aggressive lymphoma (46–48). Another clinical study is currently under way to examine the combination of selinexor with ibrutinib in patients with relapsed/refractory CLL or aggressive lymphoma (49). Eμ-TCL1xMyc mice with circulating disease after adoptive transfer of Eμ-TCL1xMyc cells showed rapid and lasting responses to treatment with the XPO-1 inhibitor KPT-8602. These findings provide further evidence that XPO-1 inhibition is a valid therapeutic strategy to pursue for patients with resistance to BCR pathway inhibitors, and proof-of-principle that the Eμ-TCL1xMyc model is a potential tool to optimize treatment strategies aimed at XPO-inhibition, especially in combination with other agents. Taken together with data derived from patient cohorts and ongoing clinical studies, future studies using the murine model described here have the potential to identify pathways of diagnostic and therapeutic relevance. This is urgently needed as novel agents targeting the B-cell receptor pathway are mostly unsuccessful in patients with aggressive lymphoma. Altogether, we expect that the Eμ-TCL1xMyc mouse model will prove a useful and predictable resource to further study the disease biology of aggressive lymphomas as well as the activity of novel therapeutic strategies in this setting.

K.A. Rogers is a consultant/advisory board member for Acerta. R. Bundschuh is an employee of American Physical Society. J.A. Woyach reports receiving commercial research grants from Janssen, Morphosys, Karyopharm, AbbVie, Loxo, and Verastem, and is a consultant/advisory board member for Pharmacyclics and Janssen. No potential conflicts of interest were disclosed by the other authors.

Conception and design: F. Lucas, K.A. Rogers, J.C. Byrd, R. Lapalombella, J.A. Woyach

Development of methodology: F. Lucas, K.A. Rogers, B.K. Harrington, V.M. Goettl, J.A. Woyach

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Lucas, K.A. Rogers, B.K. Harrington, J. Breitbach, V.M. Goettl, Z.A. Hing, P. Kanga, R. Wasmuth, D.K. White, P. Yan, J.A. Woyach

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Lucas, K.A. Rogers, B.K. Harrington, A. Pan, L. Yu, R. Bundschuh, P. Yan, J.A. Woyach

Writing, review, and/or revision of the manuscript: F. Lucas, K.A. Rogers, B.K. Harrington, A. Pan, L. Yu, R. Bundschuh, Z.A. Hing, D. Sampath, P. Yan, J.C. Byrd, R. Lapalombella, J.A. Woyach

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F. Lucas, K.A. Rogers, R. Mantel, L.L. Smith, R. Wasmuth, D.K. White, J.A. Woyach

Study supervision: R. Lapalombella, J.A. Woyach

Dr. Dalia El-Gamal provided initial technical assistance. Alan Flechtner, Florinda Jaynes, and The Ohio State University Comparative Pathology & Mouse Phenotyping Shared Resource (funded by a Cancer Center Support Grant P30 CA016058) supported pathology studies. Flow sorting was supported by Jennifer Mele. Transcriptomic library generation and sequencing were supported by Xi Chen and The Ohio State University Genomics Shared Resource (supported in part by Cancer Center Support grant P30 CA016058). Computational resources were provided by Ohio Supercomputer Center. The authors thank Karyopharm Therapeutics for providing KPT-8602. This work was supported by NCI grants R35 CA197734 (to J.C. Byrd), R01 CA214046 (to R. Lapalombella and J.C. Byrd), K23 CA178183 (to J.A. Woyach), R01 CA177292 (to J.C. Byrd and J.A. Woyach), and R01 CA192928 (to R. Lapalombella, J.C. Byrd, and J.A. Woyach); the Leukemia and Lymphoma Society; a Research Scholar grant (129863-RSG-16-158-01-CDD) from the American Cancer Society; and The OSU Comprehensive Cancer Center NCIP30 CA 016058 and Pelotonia funds.

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.