Full Text (PDF)

Published OnlineFirst October 24, 2018; DOI: 10.1158/1541-7786.MCR-18-0582

Oncogenes and Tumor Suppressors Molecular Cancer Research Sleeping Beauty Screen Identiﬁes RREB1 and Other Genetic Drivers in Human B-cell Lymphoma Eric P. Rahrmann1,2, Natalie K. Wolf1, George M. Otto2, Lynn Heltemes-Harris2,3,4, Laura B. Ramsey4, Jingmin Shu2, Rebecca S. LaRue2, Michael A. Linden2,5, Susan K. Rathe2, Timothy K. Starr2,6, Michael A. Farrar2,3,4, Branden S. Moriarity2,7,8,and David A. Largaespada1,2,7,8

Abstract

Follicular lymphoma and diffuse large B-cell lymphoma human data sets revealed novel and known driver genes (DLBCL) are the most common non-Hodgkin lymphomas for B-cell development, disease, and signaling pathways: distinguishable by unique mutations, chromosomal rear- PI3K–AKT–mTOR, MAPK, NFkB, and B-cell receptor (BCR). rangements, and gene expression patterns. Here, it is dem- Finally, functional data indicate that modulating Ras- onstrated that early B-cell progenitors express 20,30-cyclic- responsive element-binding protein 1 (RREB1) expression nucleotide 30 phosphodiesterase (CNP) and that when in human DLBCL cell lines in vitro alters KRAS expression, targeted with Sleeping Beauty (SB) mutagenesis, Trp53R270H signaling, and proliferation; thus, suggesting that this proto- mutation or Pten loss gave rise to highly penetrant lymphoid oncogene is a common mechanism of RAS/MAPK hyper- diseases, predominantly follicular lymphoma and DLBCL. activation in human DLBCL. In efforts to identify the genetic drivers and signaling pathways that are functionally important in lymphomagenesis, Implications: A forward genetic screen identiﬁed new genetic SB transposon insertions were analyzed from splenomegaly drivers of human B-cell lymphoma and uncovered a RAS/ specimens of SB-mutagenized mice (n ¼ 23) and SB-muta- MAPK–activating mechanism not previously appreciated in genized mice on a Trp53R270H background (n ¼ 7) and human lymphoid disease. Overall, these data support target- identiﬁed 48 and 12 sites with statistically recurrent trans- ing the RAS/MAPK pathway as a viable therapeutic target in a poson insertion events, respectively. Comparison with subset of human patients with DLBCL.

Introduction occur predominantly in older adults, diagnosis and treatment have greatly been impacted by the genetic profiling efforts. DLBCL B-cell malignancies comprise a large family of diseases ranging is categorized into two unique molecular subtypes based on gene from highly curable Hodgkin lymphoma to the more diverse non- expression profiling: activated B-cell–like (ABC) and germinal Hodgkin lymphoma subtypes including the indolent follicular center B-cell–like (GCB; ref. 1). Transcriptomic and genomic lymphoma and the aggressive, genetically heterogeneous diffuse analyses identified recurrent genomic aberrations and signaling large B-cell lymphoma (DLBCL; ref. 1). Molecular profiling of pathway alterations unique to each subtype and common to both B-cell malignancies has identified defining genetic features for (2, 3). Mutations in genes altering B-cell receptor (BCR) signaling many of the subtypes leading to new therapeutic targets and and NFkB activation (e.g., CD79A, MALT1, and MYD88) are more increased survival-rates for some diseases (1). DLBCL, which common in ABC DLBCL, whereas mutations in genes altering histone modifications and B-cell homing (e.g., EZH2, CREBBP, and MLL2) are more common in GCB DLBC (4–6). Mutations in 1 Department of Genetics, Cell Biology, and Development, University of TP53, immunosurveillance genes (e.g., B2M, CD58), epigenetic 2 Minnesota, Minneapolis, Minnesota. Masonic Cancer Center, University of modifiers (e.g., CREBBP), and MYC copy number alteration Minnesota, Minneapolis, Minnesota. 3Lab Medicine and Pathology, University of Minnesota, Minneapolis, Minnesota. 5Center for Immunology, University of (CNA) gains occur in both subtypes (2). Whole-genome and fi Minnesota, Minneapolis, Minnesota. 4Department of Laboratory Medicine and -exome sequencing efforts have identi ed over 300 recurrently Pathology, Division of Hematopathology, University of Minnesota, Minneapolis, mutated genes in primary DLBCL samples (3, 5, 7, 8). However, Minnesota. 6Department of Ob-Gyn and Women's Health, University of there is still limited knowledge on functional impact of many of 7 Minnesota, Minneapolis, Minnesota. Department of Pediatrics, University of these mutations and genetic alterations on disease initiation and 8 Minnesota, Minneapolis, Minnesota. Center for Genome Engineering, University progression; genetically engineered mouse models (GEMM) pro- of Minnesota, Minneapolis, Minnesota. vide a platform to begin evaluating these putative targets. Note: Supplementary data for this article are available at Molecular Cancer The Sleeping Beauty (SB) somatic cell mutagenesis system has Research Online (http://mcr.aacrjournals.org/). successfully identified genetic drivers of various cancers including Corresponding Author: Eric P. Rahrmann, University of Cambridge, Robinson hepatic, intestinal, pancreatic, osteosarcoma, and T-cell (9–14). Way, Cambridge CB2 0RE, United Kingdom. Phone: 012-2373-0854; Fax: We previously reported the identification of novel genetic drivers 012-2376-9881; E-mail: [email protected] of peripheral nerve–related cancers targeting SB mutagenesis to doi: 10.1158/1541-7786.MCR-18-0582 20,30-cyclic-nucelotide 30 phosphodiesterase (Cnp)-expressing 2018 American Association for Cancer Research. cells in mice in the context of EGFR overexpression with

www.aacrjournals.org 567

Rahrmann et al.

Trp53R270H mutation (12). Mutagenesis alone or in the context of IHC only Trp53R270H mutation was inefficient at developing peripheral The M.O.M. kit (Vector Laboratories Inc.) was used for blocking nervous system tumors (12). We describe here how these animals and antibody incubations. Primary antibodies: Ki67 (1:100; Leica developed highly penetrant (65%) lymphoid disease (follicular Biosystems), RREB1 (1:100; Sigma-Aldrich), pErk (1:100; Cell lymphoma and DLBCL). Analysis of SB-induced lymphomas Signaling Technology), pAkt (1:100; Cell Signaling Technology), identified 59 common insertion sites (CIS), of which several were Kras (1:100, Santa Cruz Biotechnology), and SB (1:100; R&D associated with signaling pathways altered in human DLBCL Systems). Corresponding biotinylated secondary antibodies formation: PI3K–AKT–mTOR, NFkB, and BCR signaling. We also (1:250; Vector Laboratories Inc.) were used followed by incuba- identified several novel proto-oncogenes and tumor suppressor tion with Vectastain ABC Kit (Vector Laboratories Inc.) and genes (TSG) for B-cell lymphoma, for example, Ras-responsive developed using peroxidase substrate kit DAB (Vector Laborato- element binding protein 1 (Rreb1) and Ambra1, respectively. ries Inc.). Slides were counterstained with hematoxylin, dehy- Furthermore, we described new roles for Rreb1, a MAPK pathway drated, cleared with xylene, and mounted with permount effector, in DLBCL maintenance and its impact on Kras expression, (Thermo Fisher Scientific). revealing an unknown mechanism for RAS activation in DLBCL. A tissue microarray containing classical Hodgkin lymphoma (cHL, n ¼ 3), low-grade follicular lymphoma (LGFL, n ¼ 10), and Materials and Methods DLBCL (n ¼ 34) was purchased from Cybri (CS20-00-002) and stained for RREB1 (above). IHC staining was quantified using the Transgenic animals following criteria by: 0, negative; 1, faint; focal, equivocal, 2, Three transgenes were used to induce SB mutagenesis: Condi- positive in a minority of cells; and 3, positive in a majority of cells. tionally expressed SB (R26SB11LSL; ref. 15), Cnp promoter–driven Samples stained with same antibody conditions by the Human cre recombinase (Cnp-Cre; ref. 16) and oncogenic transposon, Protein Atlas were also assessed with the same criteria. cHL, n ¼ 2; concatemer (T2/Onc15). Cnp-Cre;R26SB11LSL;T2/Onc15 (SB- þ low-grade non-Hodgkin lymphoma, n ¼ 7; high-grade non- mutagenized) mice underwent insertional mutagenesis in Cnp Hodgkin lymphoma, n ¼ 3. cells. Genotyping PCR was performed on phenol-chloroform– extracted mouse tail DNA (10, 16, 17). Conditionally expressed Pten (Ptenf/f) and Trp53 (Trp53R270H) allele mice were utilized Pathology (17, 18). B6.129(Cg)-Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo/J Board-certified pathologist Dr. Michael Linden (University reporter mice (The Jackson Laboratory) were utilized for lineage of Minnesota, Saint Paul, MN) evaluated hematoxylin and tracing studies. All mice were bred and cared for under the guidelines eosin (H&E)-stained tissues for red and white pulp content, of the University of Minnesota Animal Care and Use Committee. extramedullary hematopoiesis, megakaryocytes, erythroid precursors, immature granulocytes, lymphocyte size, number, fi V(D)J PCR morphology, plasmacytic differentiation, in ltration into fi One-hundred nanograms of DNA from control and SB- extra-hematopoietic tissues, mitotic gures, and necrosis. mutagenized spleens underwent PCR to assess V(D)J clonality for VHJ558/JH3, VHQ52/JH3, VH7183/JH3, and DHL/JH3 recombi- Comparative genomics nation (19). PCR for Actb served as the loading control. Whole-methylome, CNA, and transcriptomic data from 48 DLBCL human patient samples were acquired from The Cancer Flow cytometry Genome Atlas (TCGA) database (23). Methylome data were b Single-cell suspensions from bone marrow (femur and tibia), listed as values, CNA data were analyzed by GISTIC analysis, spleen, and lymph nodes were stained with the following anti- and transcriptomic data were listed as fragments per kilobase of bodies: a-IgM (Jackson ImmunoResearch), a-IgD (11–26), transcript per million (FPKM) mapped reads. CISs were ana- a-BP-1 (FG35.4), a-CD5 (53-7.3), a-CD19 (1D3), a-CD21/35 lyzed for enrichment into known pathway using Enrichr soft- (7E9), a-CD23 (B3B4), a-CD24 (M1/69), a-CD25 (PC61.5), ware (24). a-CD38 (90), a-CD43 (S7, BD Biosciences), a-CD45R (RA3- 6B2) for Hardy fractionation (20). Antibodies were obtained Cell culture þ from eBioscience unless otherwise indicated. SA-PerCP-Cy5.5 We purchased CD19 B cells (Sanguine Biosciences) and (eBioscience) was used to detect biotinylated antibodies. Cells DLBCL human cell lines, Toledo, Farage, Pfeiffer, and DB (ATCC). were assayed on a LSRII flow cytometer (BD Biosciences); data BL2 was gifted from Reuben Harris and KM-H2, Daudi, and were analyzed using FlowJo software (Treestar). Ramos were gifted from Vivian Bardwell at the University of Minnesota (Minneapolis, MN). Cell lines were cultured in com- plete media (1 RPMI1640, 10% FBS, and 1 penicillin/strep- Transposon insertion site analysis DNA-T2/Onc junctions were amplified by linker-mediated tomycin) and grown at 37 Cin5%CO2. Cell viability was PCR (LM-PCR), purified using MinElute 96 UF Plates (Qiagen), assessed utilizing a Trypan blue exclusion assay on a hemocy- and submitted for high-throughput HiSeq 2500 sequencing tometer every 24 hours for 5 days. No authentication or Myco- (Illumina) or 454 pyrosequencing (12). A total of 4 107 plasma tests were carried out. Cells from ATCC were passaged three 100-bp reads (Illumina) and 384,919 100-bp reads (454 pyro- times prior to experimental usage. sequencing) were processed and analyzed using Transposon Annotation Poisson Distribution Association Network Connec- RREB1 shRNA knockdown tivity Environment (TAPDANCE) software and gene-centric CIS Cells were transduced with RREB1 shRNA lentiviruses (Open þ analysis software (21, 22). Mouse build NCBI37/mm9 was used Biosystems) and flow sorted with the top 10% of GFP cells to map insertion cites and subsequent analyses. isolated and selected with 1 mg/mL puromycin.

568 Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research

Driver Genes Identiﬁed in DLBCL Using an SB Screen

RREB1 overexpression SB-mutagenized animals with or without Trp53R270H mutation RREB1 cDNA (Open Biosystems) was cloned into the Gateway revealed very rare peripheral nervous system tumors (12) but a Vector System (Life Technologies) and subcloned into a piggyBac highly penetrant lymphoid disease (65%, splenomegaly; Supple- (PB) transposon vector. Cells were transfected with 2 mgofRREB1 mentary Table S1). The lymphoid disease was predominantly or Gfp PB transposon and 2 mg of PB7 transposase plasmid using splenomegaly (62%) with some animals also presenting with an the NEON transfection system (Life Technologies) followed by enlarged thymus (13.8%) and enlarged mesenteric lymph nodes selection with 1 mg/mL puromycin. RREB1 expression was (22.4%; Supplementary Fig. S1C and S1D). Solid tumors were induced with an optimized doxycycline dosage. observed in various tissues with the liver (22.4%) having the highest incidence followed by the brain (oligodendroglioma, qRT-PCR astrocytoma; 3.4%) and fat pads (3.4%; Supplementary Fig. qRT-PCR analysis was carried out as described previously (12). S1D and S1E). Overall, SB-mutagenized mice had a significant miRNA samples were isolated utilizing the miRNeasy Mini Kit increase in the penetrance of lymphoid disease and solid tumor (Qiagen) and assessed miR-143, miR-145, and U6 expression (Life (75.9%) formation compared with control animals (29.5%; FET Technologies). P < 0.0001; Fig. 1B). SB mutagenesis did not significantly alter the phenotype penetrance in Trp53R270H (79.4%) animals compared R270H Immunoblotting with Trp53 controls (72.4%). fi Resolved lysates on polyvinylidene difluoride membranes were Histologic analysis of splenomegaly samples identi ed SB,high- probed with antibodies against RREB1 (1:1,000, Sigma-Aldrich), ly expressed in splenic germinal centers with diffuse positive cells in KRAS4A, KRAS4B (1:1,000, Santa Cruz Biotechnology), PTEN, surrounding marginal zone and red pulp (Fig. 1C). Pathologic AKT, pAKT, p4EBP1 (1:1,000, Cell Signaling Technology), and analysis indicated SB-mutagenized splenomegaly samples were fi ¼ fi GAPDH (1:10,000: Cell Signaling Technology). Corresponding signi cantly (FET P 0.0037) involved with lymphoma, speci - HRP-conjugated secondary antibodies (1:2,000: Vector Labora- cally follicular lymphoma and DLBCL, compared with control tories) were utilized. Blots were developed via chemilumines- splenomegaly samples (Fig. 1C; Supplementary Fig. S1F and cence and imaged on the LI-COR Odyssey. S1G; Supplementary Table S2). Moreover, only SB-mutagenized mice (n ¼ 5/17) had evidence of infiltrative DLBCL into surrounding tissues including liver, lungs, kidneys, skeletal muscle, and Statistical analysis adrenal glands (Fig. 1C; Supplementary Fig. S1F and S1G). Statistics were performed using GraphPad Software Prism To confirm tumor identity, we performed PCR-based clonality Version 6.0d for the following analyses: survival with Kaplan– analysis, allograft experiments, and in vivo lineage tracing analysis. Meier survival curve with log-rank Mantel–Cox test; phenotypes PCR-based clonality analysis of the BCR IgH locus from analyzed using Fisher exact tests (FET) and x2 tests. Nonparamet- SB-mutagenized spleens identified the presence of monoclonal ric Mann–Whitney tests with standard error of the mean were and oligoclonal populations in splenomegaly samples not pres- carried out on spleen weights, qRT-PCR, densitometry, and cell ent in the SB-mutagenized normal weight and C57BL/6 spleens proliferation assays. Correlation was done using Pearson corre- (Fig. 1D). Allograft transplants of primary splenomegaly samples lation analysis. þ gave rise to CD19 B-cell expansion in spleens of SCID/beige recipient mice (Fig. 1E). SB-expressing cells were immunophe- Results notyped for T-cell, B-cell, and macrophage markers on bone þ SB mutagenesis in Cnp cells induced B-cell lymphoma marrow, thymus, lymph node, and spleen samples from control We previously utilized Cnp–Cre to model peripheral nervous (Cnp-Cre n ¼ 6; R26SB11LSL n ¼ 5), SB induced without mutagen- system cancers in mice (12). Cnp, a phosphodiesterase, is highly- esis (Cnp-Cre;R26SB11LSL n ¼ 5), and SB-mutagenized mice (n ¼ expressed in nervous system tissues (oligodendrocytes and 12). Because the conditional SB allele contains a GFP stop cassette Schwann cells) starting at E14.5 through adulthood with minimal unless exposed to Cre-recombinase, GFP ve cells were used as a expression in other tissues including spleen (lymphocytes), liver, marker for SB expression. This analysis revealed SB expression þ heart, bone marrow stromal cells, and cultured mouse CD34 across all four tissues with lymph nodes (74.2%, n ¼ 10) and bone marrow cells (25, 26). Utilizing IHC for SB expression and a spleens (60.72%, n ¼ 17), showing the highest percentage of PCR-based excision assay for SB activity, we determined SB was recombination followed by thymus (40.3%, n ¼ 14) and bone expressed and active in numerous tissues (brain, pancreas, liver, marrow (30%, n ¼ 16; Supplementary Fig. S2A). Immunophe- testes, skeletal muscle, lungs, spleen, heart, and kidneys) in our notyping of these tissues for lineage-specific markers indicated model (Supplementary Fig. S1A and S1B). that SB expression occurred in myeloid, T cells, B cells, and the SB-mutagenized (Cnp-Cre;R26SB11LSL;T2/Onc15, n ¼ 63) and remaining supporting cells (stroma) of each tissue (Supplemen- control (Cnp-Cre;R26SB11LSL or Cnp-Cre;T2/Onc15, n ¼ 88) mice tary Fig. S2B–S2E). No significant changes in cellular distribution were aged and assessed for phenotypic alterations. SB- were observed in bone marrow and the peripheral lymph nodes. mutagenized mice had significantly reduced survival compared However, SB-mutagenized spleen and thymus samples had sig- with controls (log-rank Mantel–Cox, P < 0.0001) with median nificant changes in cell-type distribution. Spleen samples had survival of 436 day versus 605 days in control animals (Fig. 1A). significantly (, P < 0.05) reduced B-cell percentage with corre- Similarly, SB-mutagenized mice carrying a Trp53R270H point sponding significant increase in myeloid and the supporting mutation (Cnp-Cre;R26SB11LSL;T2/Onc15;Trp53R270H,n¼ 33) had stromal cells (Supplementary Fig. S2C). Thymus samples had significantly reduced survival (log-rank Mantel–Cox, P < 0.0001) significantly (, P < 0.05) reduced the T-cell percentage with with a median survival of 322 days versus 485 days compared with corresponding increases in B cells, myeloid, and supporting Trp53R270H (Cnp-Cre;R26SB11LSL;Trp53R270H or Cnp-Cre;T2/Onc15; stromal cells (Supplementary Fig. S2D). To correlate these Trp53R270H,n¼ 37) control animals (Fig. 1A). Examination of changes in cellular distribution with SB activity, we analyzed the

www.aacrjournals.org Mol Cancer Res; 17(2) February 2019 569

Rahrmann et al.

Figure 1. SB mutagenesis in Cnpþ cells induced B-cell lymphoma. A, Kaplan–Meier survival curve comparing SB-mutagenized (Cnp-Cre;R26SB11LSL;T2/Onc15, n ¼ 63), control (Cnp-Cre;R26SB11LSL or Cnp-Cre;T2/Onc15, n ¼ 88), SB-mutagenized mice carrying a Trp53270H point mutation (Cnp-Cre;R26SB11LSL;T2/Onc15;Trp53R270H, n ¼ 33), and Trp53R270H (Cnp-Cre;R26SB11LSL;Trp53R270H or Cnp-Cre;T2/Onc15;Trp53R270H,n¼ 37) control animals. B, Pie charts depicting macroscopic phenotypes for each genotype. C, Histologic analysis of spleen samples. Image in the top left depicts size of experimental versus control spleens. Image in the bottom right is IHC for SB on a splenomegaly samples. The following images are H&E–stained sections of spleen, liver, lung and kidney. Each image depicts evidence of lymphoma (cells of uniform size and shape, stained darkly). D, Agarose gel images of PCR reactions for assaying V(D)J recombination of the B-cell receptor (IgH locus). C, C57Bl/6 spleen; normal, mice undergoing transposition with normal spleen weights (up to 0.2 g); splenomegaly, mice undergoing transposition with spleen wet weights >0.2 g. Recombination events assessed are: VHJ558/JH3, VHQ52/JH3, VH7183/JH3, and DHL/JH3. a-Actin served as a loading control. 100-bp DNA ladder is in the far left lane. Multiple bands in each lane indicate a polyclonal population as observed in normal spleens, whereas lack of bands or a single band indicate oligoclonal populations, which are commonly observed in lymphoid disease. E, Serial transplant of primary SB-mutagenized lymphoma samples into the ﬂanks of recipient SCID/Beige mice. Spleens from primary SB-mutagenized mice (left) and allograft tumors (right) were isolated and analyzed by ﬂow cytometry for T-cell (CD4, CD8), B-cell (CD19), and macrophage (Mac1, GR1) markers.

570 Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research

Driver Genes Identiﬁed in DLBCL Using an SB Screen

GFP fraction of the total live cells (Supplementary Fig. S2B–S2E). Because TP53 mutations are prevalent in human DLBCL (34), Relative to the Cnp-Cre;T2/Onc control animals, in which all cells we also sequenced seven SB-induced lymphomas in the are GFP , there is a bias for supporting stromal cells to undergo Trp53R270H background by 454 pyrosequencing to identify recombination in all tissue types and for B cells to recombine in cooperating mutations with Trp53 mutation to drive B-cell the thymus. lymphoma. From these analyses, we identified 12 tdCIS genes Because the SB-mutagenized animals had an average age of 479 (10 human homologs) with the most significantly mutated days, we could not rule out the effects of aging on the results. genes being Pik3r1 (42.9%), Kcnj12 (42.9%), and Pls1 Therefore, we performed an additional lineage-tracing analysis on (42.9%; Fig. 2C; Supplementary Table S3). Pik3r1, a modulator 60-day-old Cnp-Cre mice bred to conditional GFP reporter mice of PI3K signaling, has been previously implicated in human on bone marrow, lymph node, and spleen samples (Supplemen- DLBCL (8). Kcnj12, a potassium channel, and Pls1,anactin- tal Figs. S3 and S4). Lineage-tracing analysis of the bone marrow binding protein, have not previously implicated in DLBCL þ demonstrated Cnp was predominantly expressed in B220 B-cell or follicular lymphoma. Collectively, Nfkb1 was the only þ precursors as only 13.9% of GFP ve cells were B220 (Supple- CIS significantly mutated in both screens. Moreover, 10 of mentary Fig. S3). Assessment of GFP expression during all stages the 59 CIS-associated genes (A23004603Rik, chr4:94961700- of B-cell development indicated that Cnp-cre was active from the 94971700, Tnfsf8, Inpp4b, C80913, Chst15, Crybg3, Mfap3l, pre-/pro–B cells (24%) to the mature follicular and marginal zone Tespa1,andTnfrsf13b,) have not been previously identified in B cells (87.5%; Supplementary Fig. S4). Collectively, these data other SB screens (Candidate Cancer Gene Database, n ¼ 69 demonstrate Cnp is expressed in many hematopoietic cells and studies, 12 tumor types; ref. 35), suggesting a specificity of these þ that targeted mutagenesis of Cnp cells preferentially induces a genes in B-cell lymphomagenesis and not cancer in general B-cell lymphoma phenotype arising from early B-cell precursors. (Supplementary Table S5). The position/orientation of the T2/Onc murine stem cell virus Identification of B-cell tumor driver mutation genes (MSCV) promoter, relative to the direction of gene transcription, To identify genetic drivers of lymphomagenesis, T2/Onc inser- can be used to predict whether T2/Onc is likely to drive or disrupt tions from 23 SB-derived splenomegaly samples were analyzed by gene transcription. Transcriptional activation may occur if the Illumina sequencing to identify CISs utilizing two unique statis- majority of transposon insertions are orientated upstream of a tical methods: TAPDANCE CIS (tdCIS; ref. 22) and gene-centric gene or translational start site with MSCV promoters in the same CIS (gCIS; ref. 21). These analyses identified 18 tdCIS- and 43 direction as gene transcription; the gene would be a putative gCIS-associated genes with 13 genes (27%) overlapping (Fig. 2A proto-oncogene (e.g., Bach2, Nfkb1, and Rreb1). Disruption of and B; Supplementary Table S3). The most common, significantly transcription may occur if the transposons land within a gene with mutated genes were Bach2 (43%, P ¼ 2.07 10 5), Ambra1 (35%, no MSCV promoter orientation or insertion site bias within the P ¼ 3.08 10 4), Rreb1 (23%, P ¼ 2.04 10 4), Arid1b (26%, locus; the gene would be a putative TSG (e.g., Ambra1 and Pten). q ¼ 2.13 10 4), and Nfkb1 (26%, P ¼ 3.90 10 4) of which Thirteen putative proto-oncogenes and 46 putative TSGs were Bach2 expression has been associated with DLBCL survival out- identified from the 59 unique CISs (56 human homologs) from come and Nfkb1 is a known effector gene in human DLBCL both screens (Supplementary Fig. S5A and S5B; Supplementary (27, 28). ARID1B, the chromatin modifier in SWI/SNF complex, Table S6). is known to be mutated in human follicular lymphoma (29). To determine whether T2/Onc insertions caused phenotypic Ambra1 is a scaffold protein involved in autophagy and cell alterations, we assessed the impact of the T2/Onc insertions on proliferation that functions as a TSG through regulation of Myc gene expression and disease-free survival. Of the 8 CIS genes but has not been previously implicated in B-cell lymphoma (30). assessed by qRT-PCR, only Rreb1 displayed significant (P ¼ Rreb1 is a transcription factor involved in MAPK and PI3K sig- 1.24183 10 8) changes in mRNA expression compared with naling through KRAS and is implicated in thyroid and bladder the wild-type spleens and SB-mutagenized splenomegaly samples cancer but not lymphomagenesis (31, 32). Genes previously that lacked T2/Onc insertions in the Rreb1 gene (Supplementary implicated in human DLBCL and/or follicular lymphoma were Fig. S5C). Eight CISs were significantly correlated with survival in also identified: Crebbp (17%, q ¼ 1.13 10 7), Malt1 (22%, P ¼ SB-mutagenized mice (Supplementary Table S7; Supplementary 3.90 10 4), and Pten (17%, P ¼ 4.18 10 2; refs. 5, 7, 8, Fig. S6). Hivep2,aMYC intron–binding transcription factor with and 33). Three of the 23 splenomegaly samples analyzed did not known TSG role in glioma (36), was the most significantly contribute to CIS calling (Fig. 2B). To identify potential drivers of correlated with reduced survival (P ¼ 0.0004 log-rank, Mantel– these individual samples, we assessed the top mutated genes for Cox test, Supplementary Fig. S6). The chr4:94961700-94971700 each splenomegaly sample (Supplementary Table S4). These CIS was also associated with significantly (P ¼ 0.0018 log-rank, three spleens had very low read counts for each of their putative Mantel–Cox test, Supplementary Table S7, Supplementary Fig. gene drivers suggesting there may be additional factors driving S6) reduced survival. This region is a predicted enhancer element lymphomagenesis: Wdtc1 (13), Pde2a (45), and Atp9b (245). In based on H3K4 mono-methylation marks (Supplementary Fig. addition to the splenomegaly samples, we also sequenced five S7; ref. 37). The closest genes to this region are Jun (242 kb) and normal weight spleens in which the animals did not display any Fggy (252 kb). Finally, Arid1b and Whsc1, known cancer-causing macroscopic or microscopic disease to identify potential early genes, were associated with a significant reduction in survival driver events (Fig. 2B). From these, we identified 3 CIS-associated (Supplementary Table S7; Supplementary Fig. S6). genes: Nfkb1 (60%, q ¼ 5.39 10 15), Kansl1 (60%, q ¼ 2.40 10 11), and Sp4 (60%, P ¼ 4.69 10 2). Nfkb1 is a CIS in the Pathways and upstream regulators of CISs splenomegaly samples. Kansl1 is not a CIS but splenomegly Enrichr (24) was used to identify significantly altered signaling samples did have T2/Onc insertions and Sp4 was exclusive to the pathways and cellular phenotypes in the 46 human homologs of nonsplenomegaly samples. the CISs from SB-mutagenized spleens (Supplementary Table S8).

www.aacrjournals.org Mol Cancer Res; 17(2) February 2019 571

Rahrmann et al.

Figure 2. CIS analysis for cooperating networks and pathways in DLBCL formation. A, Weighted word cloud of 48 CIS-associated genes from SB mutagenesis alone. Red indicates predicted proto-oncogenes. Blue indicates predicted TSGs. B, Heatmap depicting the clonal analysis of CISs across each tumor sample. Clonality (top row) was determined by results of (Fig. 1D): yellow, oligoclonal; green, clonal; white, no data. Contribution to CIS was determined by the percentage of each tumor that contributes to the total number of CIS: dark blue, high; light blue, none. CISs are listed on the left followed immediately by the percentage of tumors that contribute to the CIS calling. The number of reads for each insertion following illumina sequencing are shown: red, >10,000 reads; orange, 5,000–10,000 reads; green, 1,000–5,000 reads; blue, 100–1,000 reads; black, <100 reads. Empty squares indicate no contribution to the CIS calling. C, Weighted word cloud of 9 CIS-associated genes from SB mutagenesis on the Trp53R270H background. Red indicates predicted proto-oncogenes. Blue indicates predicted TSGs. D, Network depicts the most signiﬁcant upstream regulators of the 48 CISs. Yellow, IPA upstream regulator; red, predicted proto-oncogene; blue, predicted TSG. E, Network depicts results from CIS cooccurrence analysis of the tdCIS genes. Red, predicted proto-oncogene, blue, predicted TSG. Statistical analysis was performed using GraphPad Prism (, P ¼ 0.0002).

572 Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research

Driver Genes Identiﬁed in DLBCL Using an SB Screen

This identified several signaling pathways significantly enriched in with gene expression (b value < 0.2) and 7 CIS genes displayed the CIS list: B-cell receptor signaling [PPP2R5E, PTEN, CBLB, hypermethylation, which is associated with gene silencing NFKB1, and MALT1: Benjamini–Hochberg (B-H) corrected P ¼ (b value > 0.8; Supplementary Table S12; Fig. 3B). RNA-seq 0.02] and NFkB(TNFRSF11A, NFKB1, and MALT1: B-H corrected transcriptomic data identified 5 CIS genes overexpressed (z score P ¼ 0.03) signaling pathways; pathways altered in human B-cell > 2) in at least 10% of samples: CHST15, CNOT2, MFHAS1, TAF8, lymphomas. One of the top significantly enriched phenotypes and TESPA1 (Supplementary Table S13; Fig. 3C). Combining was enlarged spleen (CREBBP, PTEN, TNFRSF11A, SMAP1, these analyses, we predicted several CIS genes as strong genetic NFKB1, and RUNX1: B-H corrected P ¼ 0.004) with many other drivers of DLBCL. For example, CNOT2 was generally hypo- genes implicated in B-cell proliferation (NFKB1, CREBBP, methylated, overexpressed, and observed CNA gains, which is a TNFRSF11A, BACH2, ARHGAP17, and RUNX1) and differentia- predicted scenario for a proto-oncogene. CNOT2 was also pre- tion (BACH2 and MALT1). dicted to be an oncogene based on CNA and expression profiling Ingenuity Pathway Analysis (IPA, Qiagen) of upstream tran- data from 392 DLBCL patient samples (39). HIVEP2, a predicted scriptional regulators of the CIS genes identified GFI1 (P ¼ 1.7 TSG in our screen, had a similar level of methylation and CNAs as 10 5), BCL6 (P ¼ 1.11 10 3), and EP300 (P ¼ 3.09 10 3)in 3 known TSGs (TP53, CDKN2A, and TNFAIP3), suggesting the top five significantly enriched transcriptional regulators (Fig. HIVEP2 is a TSG in human DLBCL. Furthermore, assessment of 2D). CIS genes CREBBP (P ¼ 1.85 10 2) and NFKB1 (P ¼ 4.74 the 15 cooccurring CISs against the human DLBCL data sets 10 2) were also significantly enriched. EP300, CREBBP, and BCL6 identified two pairs of cooccurring CISs that are significantly are highly mutated in human DLBCL, whereas GFI1 has not been coaltered: PTEN/MAP3K8 (Padj ¼ 0.001) and BACH2/HIVEP2 implicated (2). Enrichr analysis identified ZBTB7A (n ¼ 12, B-H (Padj ¼ 0.004; Supplementary Table S9). Collectively, this gene- corrected P ¼ 0.01) and SP1 (n ¼ 15, B-H corrected P ¼ 0.02) as centric analysis confirmed our identification of known genetic significantly enriched transcriptional regulators, both of which drivers of human DLBCL and created a guide for making hypoth- have been implicated in several human cancers including B-cell eses about genes with unknown roles (e.g., KIAA0391 and lymphomas (38). KIAA1033) in DLBCL.

Cooccurring CISs Heterozygous loss of Pten is sufficient to drive B-cell We performed cooccurrence analysis to identify CIS lymphomagenesis þ genes that were mutated together at a higher frequency than Targeting Cnp cells with SB mutagenesis and/or Trp53R270H expected by chance (22). Cooccurrence analysis of the tdCIS mutations gave rise to a high prevalence of B-cell lymphoma from SB-mutagenized mice identified 15 pairs of cooccurring and lineage tracing analysis identified early B-cell precursors to CISs (co-CIS; Fig. 2E; Supplementary Table S9). Twelve co-CISs express Cnp. However, Cnp has not been previously identified involved at least one 1 that regulates transcription. Several as marker for cells that functionally develop B-cell lymphoma. genes were co-CISs with Bach2, the most mutated single CIS To further explore the functional role of Cnp in B-cell lym- (43% of tumors), including the most significant pair of co-CISs phomagenesis and validate our SB screen, we crossed Cnp-Cre with Hivep2 and Chr4:94961700.Moreover,Bach2/Hivep2 and mice to animals harboring a conditional allele of Pten (40). Bach2/Chr4:94961700 co-CISs were associated with a signifi- PTEN is a known cancer-causing gene in numerous human cant reduction in disease-free survival (Supplementary Table cancers including B-cell lymphomas and is a putative CIS TSG S9). Bach2 alone was not significantly associated with survival, from our screen. whereas Hivep2 and Chr4:94961700 did. Interestingly, the three One-hundred percent of Cnp-Cre;Ptenf/f mice (n ¼ 8, median tumors with insertions in Chr4:94961700 CIS also had inser- survival 102 days, log-rank Mantel–Cox P < 0.0001) succumbed tions in Hivep2 and Bach2, suggesting a potential cooperative to paralysis-related deaths with peripheral nerve hyperplasia signaling network. and neurofibroma formation and enlarged cervical lymph nodes þ (Fig. 4A and B). Conversely, 100% of Cnp-Cre;Ptenf/ mice Comparative genomics of CISs in cancer, including lymphoma (n ¼ 16, median survival 323 days, log-rank Mantel–Cox To determine whether the CIS-associated genes were frequently P < 0.0001) succumbed to lymphoma-related deaths with mutated in human cancer, we queried the Catalogue of Somatic enlarged cervical lymph nodes but no nervous system phenotype Mutations in Cancer (COSMIC) Cancer Gene Census, which (Fig. 4A and B). Enlarged cervical lymph nodes possessed follic- annotates known cancer-causing genes (33). Eleven of the 54 ular lymphoma histologic features and were predominantly þ þ þ þ CIS-associated genes with a human homolog are in Cancer Gene B220 CD19 CD21 CD35 B cells by immunophenotyping Census (P ¼ 2.86 10 7, hypergeometric test). To assess the (Fig. 4C and D; Supplementary Table S2). Western blot analysis relevance of CIS genes to human lymphoma, we queried methy- for members of the PI3K/PTEN/AKT pathway indicated that Pten þ lome, SNPs, RNA sequencing (RNA-seq) transcriptomic, and expression was maintained in the Cnp-Cre;Ptenf/ animals with whole-genome/-exome sequencing data of human DLBCL sam- increased signaling through the downstream effector p4ebp1 ples from TCGA (n ¼ 48 samples; Fig. 3; Supplementary Tables (Fig. 4E and F). Collectively, these data suggest Cnp is expressed S10–S13). CREBBP was the most mutated gene in the list (n ¼ 6/ from cells that are prone to B-cell lymphoma formation and 48 samples) followed by AMBRA1 (n ¼ 3/48) and VANGL1 (n ¼ that Pten haploinsufficiency is sufficient to generate a follicular 3/48; Supplementary Table S10). The SNP data identified CIS lymphoma–like phenotype. genes with a tendency toward CNA gains (n ¼ 6) and CNA losses (n ¼ 14) in more than 20% of the samples (Supplementary Table Rreb1 is a predicted proto-oncogene enhancing signaling S11; Fig. 3A). Methylome data, which is predictive of gene through KRAS/MAPK expression, identified 38 CIS genes differentially methylated. RREB1 is a transcription factor and proto-oncogene in Thirty-one CISs displayed hypomethylation, which is associated solid tumor cancers (41). RREB1 operates in a feed-forward

www.aacrjournals.org Mol Cancer Res; 17(2) February 2019 573

Rahrmann et al.

Figure 3. CIS comparative analysis to human DLBCL samples. The graphs depict data from TCGA on CNA data via GISTIC scores (A), methylome data via b values (B), and RNA-seq expression by FPKM (C). Data were acquired from TCGA database. Each data point represents an individual sample n ¼ 48 DLBCL patient samples. Methylome data is represented as the b value in which values above 0.8 indicate hypermethylation events, whereas values below 0.2 indicate hypomethylation events.

574 Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research

Driver Genes Identiﬁed in DLBCL Using an SB Screen

Figure 4. Pten haploinsufficiency is sufficient to driver B-cell lymphoma. A, Kaplan–Meier curve for Cnp-Cre;Ptenf/f mice (n ¼ 8), Cnp-Cre;Ptenf/þ mice (n ¼ 16), and control (Cnp-Cre;R26SB11LSL or Cnp-Cre;T2/Onc15, n ¼ 88) mice. B, Images taken at time of necropsy of lymphoid tissues and peripheral nerves affected by Pten loss. Pie charts indicate percentage distribution of phenotypes for each genotype. C, Representative H&E images of lymph nodes from Cnp-Cre;Ptenf/þ mice displaying follicular lymphoma features. D, Flow cytometry analysis of B-cell content in peripheral lymph nodes of two Cnp-Cre;Ptenf/þ animals. E, Western blot analysis on lysates from lymph nodes of control animal (F2062) and Cnp-Cre;Ptenf/þ animals for Pten, pAkt, Akt, p4ebp1, and Gapdh. F, Densitometry quantification of the Western blot analysis in D. www.aacrjournals.org Mol Cancer Res; 17(2) February 2019 575

Rahrmann et al.

576 Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research

Driver Genes Identiﬁed in DLBCL Using an SB Screen

loop promoting KRAS expression and activation by inhibition LGFL and DLBCL samples compared with Hodgkin lymphoma of miR-143/145 expression to potentiate MAPK and PI3K signal- samples. ing (31, 41). KRAS has been implicated as a driver in human To determine whether RREB1 influences KRAS expression in DLBCL (5). DLBCL, TCGA RNA-seq data for RREB1 and KRAS expression In this study, Rreb1 is a predicted proto-oncogene based on on 48 DLBCL samples were analyzed (Fig. 6D). There was a the orientation and position of T2/Onc insertions resulting in significant positive correlation between KRAS and RREB1 Rreb1 overexpression (Fig. 5A). mRNA fusion transcripts mRNA expression (Pearson r ¼ 0.6092, P < 0.0001; Supple- between the MSCV promoter/splice donor in T2/Onc and mentary Table S14). We performed the similar correlative Rreb1 were identified in tumors with T2/Onc insertions in analysis on 15 other cancers. Significant correlations with the Rreb1 locus and two tumors where LM-PCR did not RREB1 and KRAS expression occurred in 10 of 15 cancers identify T2/Onc insertions (Fig. 5B). Similar findings were assessed including pancreatic (r ¼ 0.4717) and colorectal observed in a SB osteosarcoma screen (42). Tumors contain- cancer (r ¼ 0.1448), where the RREB1/KRAS feedback loop ing T2/Onc-Rreb1 fusion transcripts demonstrated significantly has been described (31, 41). Collectively, these data suggest increased Rreb1 mRNA expression (ANOVA multiple compar- RREB1 is a putative proto-oncogene in DLBCL and potentially isons, P < 0.0001) and increased protein levels by IHC and other cancers via a KRAS/RREB1 feed-forward loop. Western blot analysis, suggesting Rreb1 functions as a proto- oncogene (Fig. 5C–E). RREB1 To determine the functional impact of T2/Onc-Rreb1 fusion Modulating expression alters KRAS isoform usage and transcripts on Rreb1 signaling, we assessed miR-143/145 proliferation (qRT-PCR) and Kras expression (immunoblot and IHC) and In humans, alternative splicing gives rise to 10 unique RREB1 its immediate downstream effectors, pErk and pAkt (IHC). transcripts that encode for nine distinct protein products, whereas qRT-PCR analysis demonstrated a significant reduction in KRAS has four unique transcripts encoding four distinct protein mir143 (Mann–Whitney: P ¼ 0.0004, P ¼ 0.0039) and miR- products (Ensembl release 88; ref. 43). Quantitative PCR of 145 (Mann–Whitney: P ¼ 0.0056, P ¼ 0.0268) in splenomegaly human lymphoma cell lines indicated that the DLBCL cell lines samples containing T2/Onc-Rreb1 fusions, compared with nor- expressed the least amount of RREB1 compared with the Hodgkin mal weight spleens and splenomegaly samples lacking the lymphoma line KM-H2 and the Burkitt lymphoma cell lines (Raji, fusion, respectively (Fig. 5C). IHC analysis demonstrated Daudi, BL2, Ramos; Fig. 7A). To see whether these transcript levels reflected the protein levels, we performed immunoblotting on increased staining for Rreb1, pAkt, and pErk in T2/Onc-Rreb1 þ tumors compared with normal weight spleens. Western blot normal CD19 B cells and human DLBCL, Burkitt lymphoma, analysis of the tumors demonstrated a significant increase in and Hodgkin lymphoma cell lines for RREB1 and KRAS to better Rreb1 and an increase in Kras protein compared with controls understand which isoforms are differentially expressed in normal (Fig. 5F and G). Collectively, these data demonstrated T2/Onc– and malignant B cells. Several RREB1 isoforms were expressed driven Rreb1 expression functionally impacted miR-143, (Fig. 7B). The high molecular weight RREB1 isoforms were present þ fi miR-145, and Kras expression and downstream signaling effec- in all cell lines but absent in CD19 -puri ed B cells (Fig. 7B). KRAS expression was also notably different between the cell lines tors, pErk and pAkt. þ (Fig. 7B). CD19 -purified B cells expressed two KRAS isoforms (KRAS-4A, 24 kDa; KRAS-4B, 21 kDa) at similar levels, common The proto-oncogene RREB1 is highly expressed in human in many human tissues, whereas 3/4 DLBCL cell lines had more DLBCL KRAS-4B than KRAS-4A protein (44). TCGA data on RREB1 genomic and transcriptomic altera- To interrogate the role of RREB1 as a proto-oncogene, we tions indicate that patient survival was not significantly impact- utilized shRNA knockdown and cDNA overexpression con- ed by RREB1 alterations, but displayed a trend towards reduced structs in three human B-cell lymphoma cell lines (Pfeiffer, survival (Fig. 6A). As RREB1 functions as a transcription factor, DB, and BL2). Pfeiffer and DB cells (DLBCL) were transduced we performed RREB1 antibody staining on a human tissue withthreeuniqueshRNAs.EffectiveRREB1 shRNAs significant- microarray (TMA) comprised of cHL, LGFL, and DLBCLs to ly reduced the mRNA expression by qRT-PCR in both Pfeiffer determine whether the genomic and transcriptomic alterations and DB cell lines (Fig. 7C) and decreased RREB1 protein affect RREB1 protein levels (Fig. 6B). Fifty-three percent (N ¼ by approximately 75% in DB cells (Fig. 7D). Effective shRNAs 18/34) of DLBCL samples expressed RREB1 in the majority of significantly decreased proliferation in Pfeiffer and DB cells cells (Fig. 6C). In general, RREB1 staining was more intense in compared with parental cells and noneffective shRNAs (Fig. 7E

Figure 5. T2/Onc insertions drive Rreb1 expression in SB-mutagenized splenomegaly. A, Schematic of T2/Onc insertions (arrows) within the Rreb1 locus. Direction of arrows indicates orientation of the MSCV 50LTR relative to the direction of gene transcription (left-to-right). Asterisk () indicates which exons the primers amplify. B, Agarose gel image of PCR analysis of T2/Onc-Rreb1 mRNA fusion transcripts from SB-mutagenized spleen samples. C, Bar graph depicts qRT-PCR analysis of Rreb1, miR-143,andmiR-145 expression in wild-type (n ¼ 4) and splenomegaly samples without insertions (n ¼ 3) and those with T2/Onc-Rreb1 fusions (n ¼ 5). Bar height indicates mean and error bars show SEM. P values were calculated with the one-way ANOVA with multiple comparisons (Tukey). , P < 0.0001. D, Western blot analysis for Rreb1, Kras, and Gapdh expression in SB-mutagenized spleen samples. M194 is a normal weight wild-type control spleen. Asterisk () represents spleens containing T2/Onc-Rreb1 fusion transcripts. E, Bar graph depicts densitometry analysis of the Western blot analysis in D. Analysis performed using ImageJ software. F, IHC analysis of Rreb1, Kras, pErk, and pAkt expression in normal weight spleens from control and splenomegaly samples from SB-mutagenized spleens containing the T2/Onc-Rreb1 fusion transcripts. G, Quantiﬁcation of the staining in F using criteria outlined in the Materials and Methods section to score TMA. Statistical analyses were performed using the GraphPad Prism Version 6.0d. Error bars represent the SEM (, P < 0.05; , P < 0.001).

www.aacrjournals.org Mol Cancer Res; 17(2) February 2019 577

Rahrmann et al.

Figure 6. RREB1 expression in human DLBCL. A, Kaplan–Meier survival curve of 48 patients with DLBCL analyzed on the basis of the presence or absence of RREB1 mutations and CNAs. Log-rank Mantel–Cox analysis P ¼ 0.3111. B, IHC analysis of RREB1 expression on TMA for human lymphoma samples. Images depict the four grades of staining observed in human DLBCL (details in Materials and Methods). C, Bar graph depicts the quantiﬁcation of the staining from (A); cHL, classical Hodgkin lymphoma n ¼ 5; LGFL, low-grade follicular lymphoma n ¼ 13; DLBCL n ¼ 41. Bar height represents the mean with SEM. D, Graph depicts RNA-seq FPKM values for KRAS plotted against RREB1 FPKM values for 48 DLBCL samples from TCGA database. Pearson coefﬁcient correlation analysis. Statistical analyses were performed using the GraphPad Prism Version 6.0d.

and F). RREB1 shRNA knockdown also reduced KRAS-4B and Trp53R270H mutation or Pten loss gave rise to highly pene- increased KRAS-4A expression in DB cells. Overexpression of trant lymphoid diseases, predominantly follicular lymphoma full-length RREB1 cDNA in DB and BL2 cells altered the KRAS and DLBCL. Analysis of SB-mutagenized splenomegaly isoform expression but did not impact proliferation (Supple- samples on wild-type and Trp53R270H-mutant backgrounds mentary Fig. S8). Collectively, these functional data suggest identified 59 CIS-associated genes and several co-CISs, sug- RREB1 expression significantly impacts human DLBCL prolif- gesting that specific, ordered cooperating genetic mutations eration and influences the amount and isoform expression are required for tumor development. CIS genes were also of KRAS. identified in known signaling pathways altered in human DLBCL including NFkB, PI3K–AKT–mTOR, and BCR signaling. Finally, we identified a new role for the putative proto- Discussion oncogene, Rreb1, and elucidated a mechanism by which In this study, we identified that early B-cell progenitors KRAS signaling is altered in DLBCL by upregulation of express Cnp and that when targeted with SB mutagenesis, RREB1 expression.

578 Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research

Driver Genes Identiﬁed in DLBCL Using an SB Screen

Figure 7. RREB1 expression influences Kras expression and human B-cell lymphoma proliferation. A, Bar graph depicts qRT-PCR analysis of RREB1 expression relative to ACTIN expression in the KM-H2 (Hodgkin lymphoma), Burkitt lymphoma cell lines (Raji, Daudi, BL2, and Ramos), and DLBCL cell lines (DB and Pfeiffer Farage). B, Western blot analysis of RREB1, KRAS, and GAPDH expression in a panel of human DLBCL cell lines (Toledo, Farage, DB, and Pfieffer) and KM-H2 (Hodgkin lymphoma) and Burkitt lymphoma cell lines (Ramos and Daudi). CD19 cells served as a normal control. C, Bar graph depicts qRT-PCR analysis of RREB1 expression relative to ACTIN expression in the DLBCL cell lines DB and Pfeiffer parental cell line and three derivatives exposed to three unique shRNAs targeting RREB1. DB control n ¼ 3 (parental and noneffective shRNAs), DB knockdown n ¼ 2 effective RREB1-shRNAs in triplicate. Pfeiffer control n ¼ 2 (parental and noneffective shRNA), Pfeiffer knockdown n ¼ 3 effective RREB1-shRNAs in triplicate. Student t test, , P < 0.0001. Bar height indicates mean and error bars show SEM. D, Western blot analysis for RREB1, KRAS, and GAPDH expression in the human DLBCL cell line DB targeted with shRNAs against RREB1. E, Graph depicts viable cell counts for modified DB cell lines over the course of 13 days. Control n ¼ 3 (parental and noneffective shRNAs) in duplicate, knockdown n ¼ 2 unique RREB1-shRNAs cell lines in duplicate. Bars indicate SE of the mean. Statistical analyses: one-way ANOVA multiple comparisons test 95% confidence interval. F, Graph depicts viable cell counts for Pfeiffer-modified cell lines over the course of 13 days. Control n ¼ 1 in duplicate, knockdown n ¼ 3 unique RREB1-shRNAs run in duplicate. Error bars indicate SEM. Statistical analyses: one-way ANOVA multiple comparisons test 95% confidence interval. Statistical analyses were performed using the GraphPad Prism Version 6.0d. www.aacrjournals.org Mol Cancer Res; 17(2) February 2019 579

Rahrmann et al.

Comparative genomic analysis reliably identified known common in human DLBCL, Lohr and colleagues have identi- genetic drivers of DLBCL formation (e.g., CREBBP,and fied rare KRASG13D mutations in human DLBCL and RREB1 MALT1). We also identified potential new TSGs (e.g., HIVEP2, CNA gains do occur in human DLBCL with a subset of samples and PKN2) and proto-oncogenes (e.g., CNOT2,andRREB1)in significantly overexpressing RREB1 transcripts that significantly DLBCL based on CNA, methylome, transcriptomic data, and correlate with increased KRAS expression (5). Currently, no consistent recurrent alterations in our SB screen. However, we RREB1 inhibitors exist, but there is a selective irreversible identified genes with CNA gains (e.g., TNFRSF11A,and inhibitor targeting the KRASG12C mutation (50). Our SB CNOT2) and overexpression in human DLBCLs predicted to mouse model of DLBCL provides a platform to further delve be disrupted by SB-induced mutagenesis. This discrepancy may into the Rreb1/Kras connection to understand the extent the reflect differences in mouse and human DLBCL formation RAS/MAPK signaling pathway has in DLBCL formation and and/or the cell of origin targeted in the screen. Generally, we maintenance with potential exploitation in therapeutic testing, identified few annotated DLBCL oncogenes and TSGs in our such as MEK inhibitors. It seems likely that RREB1 expression CIS lists. However, assessment of the T2/Onc insertion profiles at high level is a common source of RAS/MAPK activation on genes that define a variety of human lymphoma subtypes in DLBCL. indicated that insertions were present below the CIS threshold Overall, using a SB forward genetic screen, we identified 59 in several genes (ARID1A, BCL2, MLL3,andMYD88; Supple- candidate driver genes promoting B-cell lymphomagenesis. More- mentary Table S15; Supplementary Fig. S10), which with a over, we determined a new role for the proto-oncogene RREB1 in larger number of animals in the study would likely be signif- DLBCL and KRAS isoform usage. Further functional testing of icant. Importantly, many CIS genes contributed to signal trans- additional CIS genes may reveal new genetic pathways to target for duction pathways that are routinely activated in human DLBCL treatment of DLBCL. (e.g., NFkB signaling, PI3K–AKT–mTOR;ref.2).Furtherexper- imental evidence is required to determine the impact that the Disclosure of Potential Conflicts of Interest CIS genes may have on human DLBCL development. M.A. Farrar reports receiving a commercial research grant from Merck. RREB1 expression alone may not be predictive of oncogenic D.A. Largaespada is the co-founder/co-owner of NeoClone Biotechnologies, Inc., capacity but rather the specific isoform(s) expressed. From our Discovery Genomics, Inc., and B-MoGen Biotechnologies, Inc., is a consultant for analyses, we observed differences in RREB1 isoform expression Surrogen, Inc., and reports receiving funding from Genentech, Inc. B.S. Moriarity is þ between CD19 B cells and seven lymphoma cell lines. Moreover the co-founder and the chief scientificofficer for B-MoGen Biotechnologies. fl in our SB model, these RREB1 isoform differences were associated No potential con icts of interest were disclosed by the other authors. with differences in Kras expression. It is likely there are differences in the kinetics of RREB1 isoforms binding to and suppressing the Authors' Contributions miR-143/145 miRNA cluster to perform oncogenic functions on Conception and design: E.P. Rahrmann, D.A. Largaespada Kras signaling. Therefore, to further determine the importance of Development of methodology: E.P. Rahrmann, D.A. Largaespada RREB1 in DLBCL subtypes, and by extension, other cancers, a Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E.P. Rahrmann, N.K. Wolf, G.M. Otto, L.H. Harris, proteomic analysis may be warranted. L.B. Ramsey, J. Shu, T.K. Starr, B.S. Moriarity KRAS has two alternative versions of exon 4 generating two Analysis and interpretation of data (e.g., statistical analysis, biostatistics, isoforms differing at the carboxy terminus: the canonical KRAS- computational analysis): E.P. Rahrmann, L.H. Harris, L.B. Ramsey, J. Shu, 4A, which can promote apoptosis and KRAS-4B, which has an R.S. LaRue, M.A. Linden, S.K. Rathe, T.K. Starr, M.A. Farrar, D.A. Largaespada anitapoptotic role (45, 46). The amino acid changes modify KRAS Writing, review, and/or revision of the manuscript: E.P. Rahrmann, membrane localization altering KRAS-induced Raf-1 signaling L.B. Ramsey, M.A. Linden, S.K. Rathe, T.K. Starr, M.A. Farrar, D.A. Largaespada Other (central pathology review of microscopic data): M.A. Linden (46). Both isoforms are coexpressed in many human tissues, but the 4A/4B ratio is altered in human colorectal cancer with reduced 4A and increased 4B (47). We observed a similar phenomenon in Acknowledgments the DLBCL cell line DB, RREB1 cDNA overexpression increased TheauthorswouldliketothanktheBiomedicalGenomicsCenteratthe Kras-4B expression, whereas shRNA knockdown increased Kras- University of Minnesota (Minneapolis, MN) for performing the Illumina deep sequencing. We also acknowledge the following shared resources of 4A and significantly reduced proliferation. Collectively, these data fi the Masonic Cancer Center at the University of Minnesota: The Mouse indicate RREB1 dosage alters KRAS isoform usage and signi cant- Genetics Laboratory, Biostatistics and Bioinformatics, Flow Cytometry ly impacts DLBCL cellular proliferation. Further studies assessing Resource, and Comparative Pathology. We thank the Minnesota Super- the role of each RREB1 and KRAS isoform on oncogenic trans- computing Institute for computational resources. We thank the Research formation are warranted. Animal Resources at the University of Minnesota, specifically Alwan Aliye, Three signaling pathways were significantly enriched for for his technical support in mouse maintenance. This work received funding from the American Cancer Society Research Professor Award with CISs: NFkB(n ¼ 7/59), BCR (n ¼ 7/59), and PI3K– – ¼ and NIH-NCI CA113636 (to D.A. Largaespada) the NIH-NINDS-P50 AKT mTOR (n 11/59) signaling. Each pathway is involved N5057531, and the Margaret Harvey Schering Trust. M.A. Farrar was funded in human DLBCL with current targeted therapeutic efforts by NIH R01 CA151845 and CA154998. T.K. Starr was supported by funding underway (48). Pharmacologic inhibition of the PI3K–AKT– from the NIH NCI (5R00CA151672-04) and the Masonic Cancer Center NIH mTOR pathway in cells with activating mutations in PI3K/ support grant (P30-CA77598). AKT/mTOR pathway genes reduced the proliferation and caused apoptosis (8, 49). Because of the importance of the The costs of publication of this article were defrayed in part by the payment of PI3K–AKT–mTOR pathway in DLBCL maintenance, it is pos- page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. sible RREB1 overexpression is an alternative mechanism to enhance PI3K–AKT–mTOR pathway signaling via KRAS. Received June 12, 2018; revised August 13, 2018; accepted October 15, 2018; Although neither RREB1- nor KRAS-activating mutations are published first October 24, 2018.

580 Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research

Driver Genes Identiﬁed in DLBCL Using an SB Screen

References 1. Swerdlow SH, Campo E, Pileri SA, Harris NL, Stein H, Siebert R, et al. The 21. Brett BT, Berquam-Vrieze KE, Nannapaneni K, Huang J, Scheetz TE, Dupuy 2016 revision of the World Health Organization classification of lymphoid AJ. Novel molecular and computational methods improve the accuracy of neoplasms. Blood 2016;127:2375–90. insertion site analysis in sleeping beauty-induced tumors. PLoS One 2. Shaffer AL, Young RM, Staudt LM. Pathogenesis of human B cell lympho- 2011;6:e24668. mas. Annu Rev Immunol 2012;30:565–610. 22. Sarver AL, Erdman J, Starr T, Largaespada DA, Silverstein KA. TAPDANCE: 3. Zhang J, Reddy A, Love C, Moffitt AB, Rajagopalan D, Lepp€a S, et al. an automated tool to identify and annotate transposon insertion CISs and Integrative genetic and clinical analysis through whole exome sequencing associations between CISs from next generation sequence data. BMC in 1001 diffuse large B cell lymphoma (DLBCL) patients reveals novel Bioinformatics 2012;13:154. disease drivers and risk groups. Am Soc Hematol 2016;128:1087. 23. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBio 4. Davis RE, Ngo VN, Lenz G, Tolar P, Young RM, Romesser PB, et al. Chronic cancer genomics portal: an open platform for exploring multidimensional active B-cell-receptor signalling in diffuse large B-cell lymphoma. Nature cancer genomics data. Cancer Discov 2012;2:401–4. 2010;463:88–92. 24. Kuleshov MV, Jones MR, Rouillard AD, Fernandez NF, Duan Q, Wang Z, 5. Lohr JG, Stojanov P, Lawrence MS, Auclair D, Chapuy B, Sougnez C, et al. et al. Enrichr: a comprehensive gene set enrichment analysis web server Discovery and prioritization of somatic mutations in diffuse large B-cell 2016 update. Nucleic Acids Res 2016;44:W90–7. lymphoma (DLBCL) by whole-exome sequencing. Proc Natl Acad Sci U S A 25. Goolsby J, Marty MC, Heletz D, Chiappelli J, Tashko G, Yarnell D, et al. 2012;109:3879–84. Hematopoietic progenitors express neural genes. Proc Natl Acad Sci U S A 6. Pasqualucci L, Dominguez-Sola D, Chiarenza A, Fabbri G, Grunn A, 2003;100:14926–31. Trifonov V, et al. Inactivating mutations of acetyltransferase genes in B-cell 26.CaoQ,DingP,LuJ,DheenST,MoochhalaS,LingEA.20,30cyclic lymphoma. Nature 2011;471:189–95. nucleotide 30phosphodiesterase cells derived from transplanted mar- 7. Pasqualucci L, Trifonov V, Fabbri G, Ma J, Rossi D, Chiarenza A, et al. row stromal cells and host tissue contribute to perineurial compart- Analysis of the coding genome of diffuse large B-cell lymphoma. Nat Genet ment formation in injured rat spinal cord. J Neurosci Res 2007;85: 2011;43:830–7. 116–30. 8. Zhang J, Grubor V, Love CL, Banerjee A, Richards KL, Mieczkowski PA, et al. 27. Ichikawa S, Fukuhara N, Katsushima H, Takahashi T, Yamamoto J, Genetic heterogeneity of diffuse large B-cell lymphoma. Proc Natl Acad Sci Yokoyama H, et al. Association between BACH2 expression and U S A 2013;110:1398–403. clinical prognosis in diffuse large Bcell lymphoma. Cancer Sci 2014; 9. Berquam-Vrieze KE, Nannapaneni K, Brett BT, Holmfeldt L, Ma J, 105:437–44. Zagorodna O, et al. Cell of origin strongly influences genetic selection in 28. Davis RE, Brown KD, Siebenlist U, Staudt LM. Constitutive nuclear factor a mouse model of T-ALL. Blood 2011;118:4646–56. kB activity is required for survival of activated B cell–like diffuse large B cell 10. Keng VW, Villanueva A, Chiang DY, Dupuy AJ, Ryan BJ, Matise I, et al. A lymphoma cells. J Exp Med 2001;194:1861–74. conditional transposon-based insertional mutagenesis screen for genes 29. Pasqualucci L, Khiabanian H, Fangazio M, Vasishtha M, Messina M, associated with mouse hepatocellular carcinoma. Nat Biotechnol 2009;27: Holmes AB, et al. Genetics of follicular lymphoma transformation. 264–74. Cell Rep 2014;6:130–40. 11. Perez-Mancera PA, Tuveson DA, Rust AG, Biankin AV, Wessels LFA, 30. Cianfanelli V, Fuoco C, Lorente M, Salazar M, Quondamatteo F, Gherardini Wood SA, et al. Abstract A64: The deubiquitinase USP9X suppresses PF, et al. AMBRA1 links autophagy to cell proliferation and tumorigenesis pancreatic ductal adenocarcinoma. [abstract]. In: Proceedings of the by promoting c-Myc dephosphorylation and degradation. Nat Cell Biol AACR Special Conference on Pancreatic Cancer: Progress and Chal- 2015;17:20–30. lenges; 2012 Jun 18–21; Lake Tahoe, NV. Philadelphia (PA): AACR; 31. Kent OA, Fox-Talbot K, Halushka MK. RREB1 repressed miR-143/145 2012. Abstract nr A64. modulates KRAS signaling through downregulation of multiple targets. 12. Rahrmann EP, Watson AL, Keng VW, Choi K, Moriarity BS, Beckmann DA, Oncogene 2012;32:2576–85. et al. Forward genetic screen for malignant peripheral nerve sheath tumor 32. Nitz MD, Harding MA, Smith SC, Thomas S, Theodorescu D. RREB1 formation identifies new genes and pathways driving tumorigenesis. transcription factor splice variants in urologic cancer. Am J Pathol 2011; Nat Genet 2013;45:756–66. 179:477–86. 13. Starr TK, Allaei R, Silverstein KA, Staggs RA, Sarver AL, Bergemann TL, et al. 33. Forbes SA, Beare D, Gunasekaran P, Leung K, Bindal N, Boutselakis H, et al. A transposon-based genetic screen in mice identifies genes altered in COSMIC: exploring the world's knowledge of somatic mutations in human colorectal cancer. Science 2009;323:1747–50. cancer. Nucleic Acids Res 2015;43:D805–11. 14. Moriarity BS, Otto GM, Rahrmann EP, Rathe SK, Wolf NK, Weg MT, et al. A 34. Leroy K, Haioun C, Lepage E, Le Metayer N, Berger F, Labouyrie E, et al. Sleeping Beauty forward genetic screen identifies new genes and pathways p53 gene mutations are associated with poor survival in low and low- driving osteosarcoma development and metastasis. Nat Genet 2015;47: intermediate risk diffuse large B-cell lymphomas. Ann Oncol 2002; 615–24. 13:1108–15. 15. Dupuy AJ, Rogers LM, Kim J, Nannapaneni K, Starr TK, Liu P, et al. A 35. Abbott KL, Nyre ET, Abrahante J, Ho YY, Vogel RI, Starr TK. The Candidate modified sleeping beauty transposon system that can be used to Cancer Gene Database: a database of cancer driver genes from forward model a wide variety of human cancers in mice. Cancer Res 2009; genetic screens in mice. Nucleic Acids Res 2015;43:D844–8. 69:8150–6. 36. Sun L, Chen X, Jin X, Huang Q, Wang W, Zhi D, et al. Identification and 16. Lappe-Siefke C, Goebbels S, Gravel M, Nicksch E, Lee J, Braun PE, et al. characterization of human MIBP1 gene in glioma cell differentiation. J Mol Disruption of Cnp1 uncouples oligodendroglial functions in axonal sup- Neurosci 2014;52:294–301. port and myelination. Nat Genet 2003;33:366–74. 37. Rosenbloom KR, Sloan CA, Malladi VS, Dreszer TR, Learned K, Kirkup VM, 17. de Vries A, Flores ER, Miranda B, Hsieh HM, van Oostrom CT, Sage J, et al. ENCODE data in the UCSC Genome Browser: year 5 update. Nucleic et al. Targeted point mutations of p53 lead to dominant-negative Acids Res 2012;41:D56–63. inhibition of wild-type p53 function. Proc Natl Acad Sci U S A 2002; 38. Cubedo E, Gentles AJ, Huang C, Natkunam Y, Bhatt S, Lu X, et al. 99:2948–53. Identification of LMO2 transcriptome and interactome in diffuse large 18. Xiao A. Somatic induction of Pten loss in a preclinical astrocytoma model B-cell lymphoma. Blood 2012;119:5478–91. reveals major roles in disease progression and avenues for target discovery 39. Dias LM, Thodima V, Friedman J, Ma C, Guttapalli A, Mendiratta G, and validation. Cancer Res 2005;65:5172–80. et al. Cross-platform assessment of genomic imbalance confirms the 19. Reynaud D, Demarco IA, Reddy KL, Schjerven H, Bertolino E, Chen Z, clinical relevance of genomic complexity and reveals loci with potential et al. Regulation of B cell fate commitment and immunoglobulin pathogenic roles in diffuse large B-Cell lymphoma. Leuk Lymphoma heavy-chain gene rearrangements by Ikaros. Nat Immunol 2008;9: 2016;57:899–908. 927–36. 40. Lenz G, Wright GW, Emre NC, Kohlhammer H, Dave SS, Davis RE, et al. 20. Hardy RR. Resolution and characterization of pro-B and pre-pro-B cell Molecular subtypes of diffuse large B-cell lymphoma arise by distinct stages in normal mouse bone marrow. J Exp Med 1991;173:1213–25. genetic pathways. Proc Natl Acad Sci U S A 2008;105:13520–5.

www.aacrjournals.org Mol Cancer Res; 17(2) February 2019 581

Rahrmann et al.

41. Kent OA, Chivukula RR, Mullendore M, Wentzel EA, Feldmann G, Lee KH, 46. Voice JK, Klemke RL, Le A, Jackson JH. Four human Ras homologs differ in et al. Repression of the miR-143/145 cluster by oncogenic Ras initiates a their abilities to activate Raf-1, induce transformation, and stimulate cell tumor-promoting feed-forward pathway. Genes Dev 2010;24:2754–9. motility. J Biol Chem 1999;274:17164–70. 42. Temiz NA, Moriarity BS, Wolf NK, Riordan JD, Dupuy AJ, Largaespada DA, 47. Abubaker J, Bavi P, Al-Haqawi W, Sultana M, Al-Harbi S, Al-Sanea N, et al. et al. RNA sequencing of Sleeping Beauty transposon-induced tumors Prognostic signiﬁcance of alterations in KRAS isoforms KRAS-4A/4B and detects transposon-RNA fusions in forward genetic cancer screens. Genome KRAS mutations in colorectal carcinoma. J Pathol 2009;219:435–45. Res 2016;26:119–29. 48. Camicia R, Winkler HC, Hassa PO. Novel drug targets for personalized 43. Yates A, Akanni W, Amode MR, Barrell D, Billis K, Carvalho-Silva D, et al. precision medicine in relapsed/refractory diffuse large B-cell lymphoma: a Ensembl 2016. Nucleic Acids Res 2015;44:D710–6. comprehensive review. Mol Cancer 2015;14:207. 44. Plowman S, Berry R, Bader S, Luo F, Arends M, Harrison D, et al. K-ras 4A 49. Abubaker J, Bavi P, Al-Harbi S, Siraj AK, Al-Dayel F, Uddin S, et al. PIK3CA and 4B are co-expressed widely in human tissues, and their ratio is altered mutations are mutually exclusive with PTEN loss in diffuse large B-cell in sporadic colorectal cancer. J Exp Clin Cancer Res 2006;25:259–67. lymphoma. Leukemia 2007;21:2368–70. 45. Plowman SJ, Arends MJ, Brownstein DG, Luo F, Devenney PS, Rose L, et al. 50. Hunter JC, Gurbani D, Ficarro SB, Carrasco MA, Lim SM, Choi HG, et al. The K-Ras 4A isoform promotes apoptosis but does not affect either In situ selectivity proﬁling and crystal structure of SML-8-73-1, an active site lifespan or spontaneous tumor incidence in aging mice. Exp Cell Res inhibitor of oncogenic K-Ras G12C. Proc Natl Acad Sci U S A 2014;111: 2006;312:16–26. 8895–900.

582 Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research

Sleeping Beauty Screen Identifies RREB1 and Other Genetic Drivers in Human B-cell Lymphoma

Eric P. Rahrmann, Natalie K. Wolf, George M. Otto, et al.

Mol Cancer Res 2019;17:567-582. Published OnlineFirst October 24, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-18-0582

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2018/10/24/1541-7786.MCR-18-0582.DC1

Cited articles This article cites 49 articles, 18 of which you can access for free at: http://mcr.aacrjournals.org/content/17/2/567.full#ref-list-1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/17/2/567. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.