A Rapid Embryonic Stem Cell–Based Mouse Model for B-cell Lymphomas Driven by Epstein–Barr Virus Protein LMP1 Free

Epstein–Barr virus (EBV) is a γ herpes virus that can infect and transform human B lymphocytes. EBV causes a variety of human pathologies, ranging from nonmalignant diseases such as infectious mononucleosis to malignant diseases such as posttransplantation lymphoproliferative disorder (PTLD), AIDS-associated B-cell lymphomas, and X-linked lymphoproliferative disorder–associated B-cell lymphomas in immunocompromised hosts (1). EBV can also cause Burkitt and Hodgkin lymphomas in immunocompetent hosts (1). After a B cell is infected, EBV infection drives B-cell proliferation, but the infected cells are under tight immune surveillance by T cells and natural killer (NK) cells that develops upon EBV infection and largely eradicates infected B cells (2, 3). Subsequently, EBV infection reaches a latent stage, which can persist in a small fraction of B cells for life. Under immunosuppression, EBV can spread and cause massive B-cell expansion and malignant transformation (1). The EBV latent membrane protein 1 (LMP1) is a major contributing factor to the activation and transformation of human B cells (4). LMP1 is a transmembrane protein that functionally mimics a constitutively active B-cell CD40 coreceptor, which signals through the TNF pathway to activate downstream NF-κB, ERK, JNK, and JAK/STAT signaling pathways that promote cell growth and survival (5–7). Indeed, ectopic LMP1 expression can transform rodent fibroblasts (8).

A mouse model to study LMP1-induced immune surveillance and lymphoma has been developed based on the conditional expression of LMP1 in mouse B cells using B cell–specific CD19-driven Cre/loxP-mediated recombination to activate LMP1 expression (9, 10). In such “CD19-cre; LMP1^stopFL/+” mice, LMP1⁺ B cells were eliminated by T-cell immune responses, leading to a reduction in the number of splenic B cells compared with that of wild-type (WT) mice, a phenomenon reminiscent of the clearance of EBV-infected human B cells by the host immune system (2). However, elimination of the host T-cell immune response by crossing into backgrounds with homozygous elimination of both TCRα/β and TCRγ/δ T cells (TCRβ^−/−δ^−/−) resulted in rapid fatal LMP1⁺ B-cell expansions and B-cell lymphomas in the compound mutant mice (subsequently referred to as “CLT” mice; ref. 9). Thus, elimination of T-cell immune surveillance in mice allows ectopic LMP1 expression in mouse B cells to routinely cause robust B-cell proliferation and aggressive B-cell lymphoma (9). Notably, the LMP1⁺ lymphomas that develop in CLT mice routinely express activation-induced cytidine deaminase (AID; ref. 9), which is consistent with findings that AID expression is upregulated by LMP1 (11, 12). AID initiates mutations during immunoglobulin (Ig) gene variable region somatic hypermutation (SHM) process and DNA double-stranded breaks during the Ig heavy (IgH) class switch recombination (CSR) process (13, 14). Off-target AID activity has been implicated in certain B-cell lymphomas in mice and humans, including human Burkitt lymphoma. On this basis, a role of AID expression in CLT B-cell lymphomagenesis (9), and other B-cell lymphomagenesis (14–17), has been considered.

CLT mice provide a very attractive model for studies of EBV-related pathologies and immune surveillance (9, 10). However, this compound mutant model cannot be maintained as a pure strain due to frequent death of CLT mice from B-cell expansion/lymphoma before they reach breeding age. Moreover, the complexity of the model, which is based on multiple independently segregating alleles, makes it impractical to incorporate additional genetic modifications. Indeed, some desired mutations may reside on one of the 4 mutant chromosomes, requiring extensive back-crossing for their introduction into the model. We aimed to develop a CLT mouse model into which new mutant alleles could be easily introduced. For this purpose, we used the RAG2-deficient blastocyst complementation (RDBC) approach (18). Development of B and T cells is abrogated at the progenitor stage in RAG2-deficient mice, as RAG2 is essential for the V(D)J recombination process that assembles antigen receptor variable region exons (19). Thus, when V(D)J-competent embryonic stem (ES) cells are injected into RAG2-deficient blastocysts, all B and T cells derive from the injected ES cells, allowing use of ES cells containing other types of mutations to test effects on lymphocyte development and function. To generate the ES cell–based CLT model, we established a CLT ES cell line that was then used for RDBC, resulting in chimeric mice that died early with phenotypes indistinguishable from those of germline CLT mice. To further test the utility of the model, we also tested CLT cells in which we eliminated both copies of the AID gene, which lies on chromosome 6 along with the LMP1 knock-in (Rosa26 gene) and the TCRβ locus, via a Cas9/gRNA nuclease strategy.

The CLT ES cell line was derived from embryonic day 3.5 (E3.5) blastocysts resulting from a cross between CD19^cre/+; TCRβ^−/−δ^−/− and LMP1^stopFL/+; TCRβ^−/−δ^−/− mice according to an established protocol (20). The transgenic mice LMP1^stopFL used were generated on a BALB/c background, whereas other mouse strains, including CD19-cre, TCRβ^−/−, and TCRδ^−/−, were all on a C57BL/6 background (9). RAG2-deficient blastocyst injection and implantation were performed as described (18). For each RDBC injection, 6 to 8 ES cells were injected per RAG2-deficient blastocyst, and 26 to 28 injected blastocysts were implanted. Chimeric mice were assessed for extent of chimerism at 3 weeks of age by tail PCR. All of the animal experiments were performed under protocols approved by the Institutional Animal Care and Use Committee of Boston Children's Hospital.

Cas9/gRNAs targeting the AID locus were cloned into PX330 (Addgene plasmid 42230) as described (21). Multiple gRNA-targeting sequences (PAM) were used as follows: AID-A: GTAGGTCTCATGCCGTCCCT (TGG); AID-B: GCCGAAGTCCAGTGAGCAGG (AGG); AID-C: GGATTTTGAAAGCAACCTCC (TGG); and AID-D: GCGAGATGCATTTCGTATGT (TGG). The maintenance, transfection, and screening of CLT ES cells for targeting experiments were performed as described (22).

Single-cell suspensions from the spleens and livers of CLT and CLT; AID^−/− RDBC chimeras and control mice were stained with each of the following sets of anti-mouse monoclonal antibodies: anti–B220-PE-Cy5 (eBioscience Inc.) and anti–IgM-FITC (eBioscience Inc.); anti–B220-APC (eBioscience Inc.) and anti–Igκ-PE (BD Biosciences); anti–CD19-FITC (BD Biosciences) and anti-CD95/Fas-PE-Cy7 (BD Biosciences); or anti–CD8a-APC (BD Biosciences) and anti–CD4-FITC (eBioscience Inc.). FACS data were acquired using the FACSCalibur Flow Cytometer equipped with CellQuest software (BD Biosciences) and analyzed with FlowJo software (TreeStar).

Southern blotting was performed with 5 to 10 μg of genomic DNA isolated from the spleens and livers of CLT and CLT; AID^−/− RDBC chimeras and control mice or CLT ES cell clones as described previously (23). The J_H4-3 probe is a 1.6-kb HindIII/EcoRI fragment downstream of J_H4. The AID probe is a 585-bp EcoRI fragment located at 5′ arm of AID locus. A PCR fragment comprising exosc3 exon 3 was used as a loading control probe.

Total RNA was isolated with TriPure Isolation Reagent (Roche) and reverse-transcribed by M-MLV Reverse Transcriptase (Invitrogen) with oligo (dT). qRT-PCR was performed using Power SYBR Green PCR Master Mix (Applied Biosystems) on 7300 Real Time PCR System (Applied Biosystems) with AICDA-specific primers: GGGCCAAGGGACGGCATGAG and CCCGGGTCCAGGTCCCAGTC. The HPRT gene was detected in parallel and used as the internal control with primers GTCATGCCGACCCGCAGTC and GTCCTGTCCATAATCAGTCCATGAGGAATAAAC.

Cell suspensions were lysed on ice for 5 minutes with lysis buffer (50 mmol/L Tris-HCl, pH 7.4, 150 mmol/L NaCl, 1% NP40, 5 mmol/L EDTA, and 1 mmol/L PMSF) supplemented with 1× Protease Inhibitor Cocktail (Roche). After lysis of nuclei with a final concentration of 500 mmol/L NaCl and elution of genomic DNA by adding equal volume of water and centrifuging at 13,000 rpm for 10 minutes, the supernatant was boiled with 1× SDS loading buffer, separated by SDS-PAGE, and probed with an anti-AID polyclonal antibody that has been described (24). Tubulin was detected in parallel with anti–α-tubulin antibody (Sigma; T5168) and used as a loading control.

In each set of tumor transfer experiments, at least 5 × 10⁵ cells from 4 independent B-cell expansions derived from either CLT or CLT; AID^−/− RDBC chimeras were transferred via i.v. injection into immunodeficient (RAG2^−/−; γc^−/−) and immunocompetent (C57BL/6xBALB/c, F1) mice. Recipients were observed for evidence of tumor formation and further characterization performed as outlined above for analyses of primary chimeras.

Given the difficulty in introducing additional genetic alterations into the CLT background by standard breeding approaches, we sought to develop a more efficient strategy to generate such models through the use of the RDBC approach that we developed previously. For this purpose, we generated an ES cell line from CLT embryos derived from crosses between CD19^cre/+; TCRβ^−/−δ^−/− and LMP1^stopFL/+; TCRβ^−/−δ^−/− mice. We then injected the CLT ES cells into RAG2^−/− blastocysts to generate chimeric mice in which all peripheral lymphocytes must derive from the injected ES cells (Fig. 1). Our standard assay for the extent of chimerism in the RDBC chimeras made with ES cells expected to support normal B- and/or T-cell development is a standard flow cytometry measurement of the numbers of peripheral blood B and T cells (18). However, we found that 3-week-old CLT RDBC chimeras not only had no peripheral T cells, as expected due to the TCRβ^−/−δ^−/− mutations, but they also had very few circulating B cells (Supplementary Fig. S1). Yet, up to 10 chimeras from each injection had a strong contribution from the CLT ES cells (up to as much as 75% contribution) as assessed by genotyping of the CD19-cre transgene integrated in the CLT ES cells by tail PCR of 3-week-old pups. Thus, this finding, coupled with our finding of expanded LMP1⁺ B-cell populations in CLT RDBC chimera spleens (see below), indicates a potential defect in the recruitment of LMP1⁺ B cells into the peripheral blood, a possibility that had not been previously examined in germline CLT mice (9). Indeed, examination of 6- to 8-week-old CLT RDBC chimeras confirmed robust numbers of splenic CLT B cells but few B cells in the peripheral blood (Fig. 2C and Supplementary Fig. S2).

Of a cohort of 21 CLT RDBC chimeras maintained for analysis, all died or were sacrificed with extremely large spleens as determined by ultrasound by 13 weeks of age, with 50% of the animals dead or sacrificed by 7 weeks (Fig. 2A; Supplementary Table S1). A second cohort of 8 CLT RDBC control chimeras had a very similar survival curve (see below). Indeed, the survival curves of the two cohorts of CLT RDBC chimeras analyzed were remarkably similar to those of germline CLT mice (ref. 9; Supplementary Fig. S3A), with a slightly less sharp curve perhaps due to some variation in the extent of chimerism versus constant full contribution of mutant cells in germline mice. The vast majority of the CLT RDBC chimeras (19 of 21; the other 2 mice were dead before they could be analyzed) presented with splenomegaly sometimes accompanied by apparent tumor nodules, with spleen sizes reaching 27 mm on average (Fig. 2B and Supplementary Fig. S4). The dead or sacrificed CLT RDBC chimeras also, in several cases, showed hepatomegaly accompanied by tumor nodules (data not shown). FACS analysis demonstrated that the massive splenic cell expansions were of B-cell origin, as the vast majority of cells within them were CD19⁺, CD95/Fas⁺, B220^low, IgM⁺, and Igκ⁺, but CD4⁻ and CD8⁻ (Fig. 2C). Moreover, expression of CD95/Fas in these B-cell expansions confirmed prior findings that ectopic LMP1 expression is accompanied by the upregulation of CD95/Fas, which indeed has been used as a surrogate marker for LMP1 expression (5, 9, 25). Thus, the surface marker expression pattern of the splenic B-cell expansions in CLT RDBC chimeras is the same as that found in the germline CLT model (9).

To further characterize the nature of B-cell expansions in CLT RDBC chimeras, we assayed DNA from enlarged spleens and livers of these chimeras for rearrangements of the immunoglobulin heavy-chain locus (IgH) J_H region by a Southern blotting strategy that employed EcoRI-digested DNA and a J_H4-specific probe (Fig. 2D and Supplementary Fig. S5). Consistent with findings from the germline CLT mice (9), samples from chimeras terminally ill with enlarged spleens and/or livers before approximately 7 weeks of age generally had either no obvious IgH-rearranged bands or weak polyclonal rearrangements, as exemplified by chimeras #47 (38 days), #52 (38 days), #9 (45 days), #10 (45 days), and #3 (46 days; Fig. 2D and Supplementary Fig. S5A). However, chimeras that were terminally ill with enlarged spleens or livers at later ages often showed clonal IgH rearrangements, as exemplified by chimeras #4 (57 days), #7 (59 days), #44 (59 days), #16 (75 days), and #27 (94 days; Fig. 2D and Supplementary Fig. S5A). Moreover, as observed in the germline CLT model, several CLT RDBC chimeras had clonal IgH rearrangements that overlapped in size in spleen and liver samples (e.g., chimera samples #12, #7, #44, #27, and #91; Fig. 2D and Supplementary Fig. S5A), suggesting that they represent expansions of the same clonally derived B cells, most likely of splenic B-cell origin.

To explore potential roles of AID expression in CLT lymphomagenesis, we generated CLT mice in an AID-deficient background by multiple rounds of breeding. However, because the AICDA gene, LMP1 knock-in allele (Rosa26 gene), and the TCRβ locus all lie on chromosome 6 of the mouse genome, we obtained compound mutant mice at only very low frequency through sister chromatid crossovers during meiosis in the germline model. By this approach, at least 2 years of breeding was required to generate just 5 CLT; AID^−/− mice, 8 CLT; AID^+/− and 2 CLT; AID^+/+ control mice. Although both groups of mice developed CLT B-cell expansions and, in some cases, clonal B-cell lymphomas, the CLT; AID^−/− cohort was too small for in-depth comparison of onset times and other aspects of the expansions/tumors (Supplementary Fig. S3), whereas the control cohort of CLT; AID^+/− and CLT; AID^+/+ mice appeared similar in all respects to the CLT model reported previously (ref. 9; Supplementary Fig. S3).

To more rapidly generate larger cohorts of AID^−/− CLT mice for analysis, we employed CRISPR/Cas9-mediated genome editing (21) to delete both AID alleles in CLT ES cells (Fig. 3A and B), and then used the AID^−/− CLT ES cells along with CLT controls for RDBC. By this approach, we generated a cohort of 16 CLT; AID^−/− RDBC chimeras (based on two RDBC injections of two independent AID^−/− CLT ES cell clones). At the same time, we generated a cohort of 8 control CLT RDBC chimeras for comparison. The WT and AID^−/− CLT RDBC cohorts showed similar survival curves (Fig. 4A). Likewise, 7 of 8 CLT and 15 of 16 CLT; AID^−/− RDBC chimeras presented with splenomegaly, as determined by ultrasound, between 5–13 weeks and 5–19 weeks of age, and with average spleen sizes of 26 mm and 29 mm, respectively (Fig. 4B and Supplementary Fig. S4). Hepatomegaly and tumor nodules within the liver were also occasionally observed in both CLT and CLT; AID^−/− RDBC chimeras (data not shown). Analysis of expanded CLT splenic B-cell populations from these CLT; AID^−/− RDBC chimeras confirmed the lack of AID expression at both RNA and protein levels (Fig. 3C and D). Further analyses also showed that CLT; AID^−/− B-cell expansions and tumors had the same surface marker expression (e.g., Fig. 4C) and overall patterns of polyclonal or clonal J_H rearrangements as those of CLT RDBC chimeras (Fig. 4D and Supplementary Fig. S5). Taken together, these analyses demonstrate that CLT; AID^−/− RDBC chimeras develop B-cell expansions and tumors over the same general time period with the same general characteristics as those of CLT RDBC chimeras and CLT germline mice.

To confirm that the B-cell expansions in both CLT and CLT; AID^−/− RDBC chimeras contain transformed cells, we tested the transferability of expansions from 4 CLT and 4 CLT; AID^−/− RDBC chimeras into immunodeficient RAG2^−/−; common γ chain^−/− (RAG2^−/−; γc^−/−) mice that lack B, T, and NK cells or into immunocompetent (C57BL/6xBALB/c, F1) recipients. Many immunodeficient recipients injected with B-cell expansions from CLT or CLT; AID^−/− RDBC chimeras, respectively, developed enlarged spleens and in some cases, livers, with distinct tumor nodules (Fig. 5A). None of the immunocompetent hosts died over this time period and upon sacrifice did not have enlarged spleens or livers (Fig. 5A). Southern blotting analysis of IgH rearrangements on genomic DNA isolated from both enlarged spleens and livers from the immunodeficient recipients revealed the same or a subset of clonal rearrangements found in the liver donor cells (Fig. 5B). These findings demonstrate that, as in germline CLT mice, clonally transformed B-cell clones grow out in both CLT and CLT; AID^−/− RDBC mice, as demonstrated by their transfer into immunodeficient hosts. Also, like LMP1⁺ B-cell lymphomas from germline CLT mice, LMP1⁺ B-cell tumors from CLT and CLT; AID^−/− RDBC mice do not appear to be transferable into immunocompetent hosts.

CLT mice provide an extremely attractive model for studies of EBV-related pathologies and for studies of immune surveillance and approaches to cancer immunotherapy (9, 10, 26, 27). Extension of this model for such experiments often would involve introduction of additional inactivating or activating mutations of other genes into the CLT background. Because of the difficulty of doing such experiments by germline breeding due to the great genetic complexity of the model, we have developed and now documented the effectiveness of a rapid RDBC-based model that utilizes CLT ES cells. The CLT RDBC chimeras obtained develop the same types of PTLD and B-cell lymphomas, with very similar kinetics to the germline CLT mice. The RDBC approach enables very efficient generation of large cohorts of CLT mice for analysis. Furthermore, we proved the feasibility and efficiency of the approach for studies of other genes of interest in the CLT RDBC model by knocking out both copies of the AID gene in CLT ES cells via a CRISPR/Cas9 approach and then using the resultant AID^−/− CLT ES cells for RDBC.

In the development of this new approach to generate CLT mouse tumor models, we have also obtained several new findings related to the CLT model. First, our studies show that, despite the expansion of LMP1⁺ B cells in the spleen, there are essentially none in the peripheral blood of CLT RDBC chimeras. While the basis for this finding requires more studies, one possibility would be that this phenomenon is related to known effects of CD40 or LMP1 activation on B-cell mobility/homing and molecules related to that process (5, 6, 28, 29). Most B-cell tumors that arise in CLT mice express AID, suggesting a possible role of AID expression in the tumorigenesis process (9, 11, 12). However, our current studies show that LMP1⁺ B-cell expansions and tumors develop in the AID^−/− CLT model essentially identically to their development in the CLT model. In addition, our preliminary experiments show similar characteristics of CLT and CLT; AID^−/− tumors following transplant into immunodeficient and immunocompetent hosts. In contrast, AID is required to promote germinal center (GC)–derived and post–GC-derived B-cell lymphomas that arise in transgenic mouse B-cell lymphoma models based on deregulation of BCL6 expression (30). In the latter model, AID expression likely initiates the IgH to c-myc translocations frequently found in these tumors and/or may activate oncogene via off-target SHM (30). So far, such types of oncogenic translocations or mutations have not been found in CLT tumors, consistent with the lack of a requisite role for AID in CLT B-cell lymphomagenesis.

Going forward, this ES cell–based B-cell tumor model approach will allow ready knockout or enforced expression of genes of interest in CLT ES cells to test their functions in the CLT mouse tumor model much more rapidly and efficiently than would be possible with the germline CLT model. Thus, the CLT RDBC chimera B-cell proliferation and lymphoma model should greatly facilitate future analyses of genetic alterations that cooperate with LMP1 to promote lymphomagenesis, genetic mechanisms that contribute to immune recognition of LMP1-driven lymphomas, and genetic mechanisms that contribute to immune escape of LMP1-driven lymphomas, while also potentially providing a model to develop immunotherapies that promote destruction of LMP1 immune-escape variants. Finally, beyond the CLT B-cell lymphoma model, our general ES cell–based RDBC chimera approach to tumor modeling might also be applied to other mouse tumor models that involve complex genetic modifications.

No potential conflicts of interest were disclosed.

Conception and design: Z. Ba, F.-L. Meng, M. Gostissa, P.-Y. Huang, Q. Ke, Z. Wang, M.N. Dao, Y. Fujiwara, K. Rajewsky, B. Zhang, F.W. Alt

Development of methodology: Z. Ba, F.-L. Meng, M. Gostissa, M.N. Dao, Y. Fujiwara, F.W. Alt

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Z. Ba, F.-L. Meng, M. Gostissa, P.-Y. Huang, Q. Ke, Z. Wang, Y. Fujiwara, B. Zhang, F.W. Alt

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Z. Ba, F.-L. Meng, M. Gostissa, M.N. Dao, B. Zhang, F.W. Alt

Writing, review, and/or revision of the manuscript: Z. Ba, M. Gostissa, P.-Y. Huang, M.N. Dao, K. Rajewsky, B. Zhang, F.W. Alt

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P.-Y. Huang, Q. Ke, M.N. Dao

Study supervision: B. Zhang, F.W. Alt

The authors thank Dr. Kefei Yu (Michigan State University) for providing the unpublished AID-Deficient CH12 mouse B cell lymphoma cell line used in our studies.

F.W. Alt was supported by NIH grant R01 CA098285 and by a Leukemia & Lymphoma Society SCOR. B. Zhang was supported by Dana-Farber Cancer Institute Faculty Startup Funds and an American Society of Hematology Scholar Award. K. Rajewsky was initially supported by NIH grant CA098285 and then by the European Research Council (ERC Advanced Grant ERC-AG-LS6). F.-L. Meng is a Lymphoma Research Foundation postdoctoral fellow.

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.