Induction of ectopic Myc target JAG2 augments hypoxic growth and tumorigenesis in a human B-cell model

Jason T. Yusteina,1,2, Yen-Chun Liub, Ping Gaoc, Chunfa Jied, Anne Lec, Milena Vuica-Rossb, Wee Joo Chnge,f, Charles G. Eberhartb, P. Leif Bergsagele, and Chi V. Dangb,c,g,2

Departments of aPediatrics, bPathology, and cMedicine, dMicroarray Facility, and gSidney Kimmel Cancer Center, Johns Hopkins University School of Medicine, Baltimore, MD 21205; eComprehensive Cancer Center, Mayo Clinic, Scottsdale, AZ 85259; and fDepartment of Medicine, Yong Loo Lin School of Medicine, National University of Singapore, Singapore 545523

Edited* by Robert N. Eisenman, Fred Hutchinson Cancer Research Center, Seattle, WA, and approved December 22, 2009 (received for review February 5, 2009) Ectopic Myc expression plays a key role in human tumorigenesis, provides an experimentally tractable system to identify putative and Myc dose-dependent tumorigenesis has been well established endogenous and ectopic Myc target (6). in transgenic mice, but the Myc target genes that are dependent on The P493-6 cells were derived from human peripheral blood B Myc levels have not been well characterized. In this regard, we used cells immortalized by an Epstein–Barr viral (EBV) genome that is the human P493-6 B cells, which have a preneoplastic state depend- complemented with an EBV nuclear antigen-estrogen ent on the Epstein–Barr viral EBNA2 and a neoplastic state (EBNA2-ER) fusion protein and a tetracycline-repressible Myc with ectopic inducible Myc, to identify putative ectopic Myc target transgene (6). We selected P493-6 cells because they exist in at genes. Among the ectopic targets, JAG2 that encodes a Notch recep- least three states. With tetracycline, the P-493-6 cells withdraw tor Jagged2, was directly induced by Myc. Inhibition of Notch from the cell cycle in a state with very low Myc levels, arbitrarily signaling through RNAi targeting JAG2 or the γ-secretase Notch termed the “NO Myc” state. In the presence of tetracycline and inhibitor N-[N-(3,5-difluorophenacetyl)-L-alanyl]-(S)-phenylglycine estradiol, which activates EBNA2-ER, the cells proliferate with t-butyl ester (DAPT) preferentially inhibited the neoplastic state in induction of endogenous Myc by EBNA2, achieving a “LOW Myc” CELL BIOLOGY vitro. Furthermore, P493-6 tumorigenesis was inhibited by DAPT in nontumorigenic state that is equivalent to EBV-immortalized vivo. Ectopic expression of JAG2 did not enhance aerobic cell pro- peripheral B lymphocytes (7). In the absence of tetracycline and liferation, but increased proliferation of hypoxic cells in vitro and estradiol, ectopic Myc is induced in a “HIGH Myc” tumorigenic significantly increased in vivo tumorigenesis. Furthermore, the state that resembles human Burkitt lymphoma. This system has expression of Jagged2 in P493-6 tumors often overlapped with allowed us to delineate a subset of genes that specifically respon- regions of hypoxia. These observations suggest that Notch signal- ded to ectopic Myc expression, potentially representing target ing downstream of Myc enables cells to adapt in the tumor genes that contribute to lymphomagenesis. hypoxic microenvironment. Here, we report JAG2, which encodes the Notch ligand Jag- ged2, as one of the genes most highly up-regulated upon ectopic neoplasia | Notch | target genes | transcription | hypoxia Myc expression in the P493-6 cells. Although the role of Notch signaling has not been thoroughly dissected in B-cell lymphomas, he Myc proto-oncogene encodes a transcription factor, c-Myc Jagged2 overexpression is prevalent in the B-cell malignancy T(herein termed Myc), that is implicated in genesis of many multiple myeloma (8). Recent evidence suggests that Notch may human malignancies (1, 2). Deregulated Myc expression induces synergize with the B-cell receptor to enhance B-cell activation tumorigenesis, presumably via its downstream targets comprising (9). In this report, we document that JAG2 is a direct Myc target about 10–15% of the (2). It is therefore essential and that Jagged2 and Notch signaling participate in P493-6 lym- to determine whether the Myc target gene network is both phomagenesis. quantitatively and qualitatively different between nontumorigenic Results and Discussion and tumorigenic states. The chromosomal translocations of Burkitt lymphoma dereg- Human P493-6 B-Cell Neoplasm Model. The P493-6 cells were ulate the expression of the Myc transcription factor. Ectopic Myc engineered with a tetracycline-regulated Myc and an EBNA2- dosage effects on tumorigenesis are best exemplified by many ER fusion protein to achieve NO, LOW, and HIGH Myc states A–C transgenic mouse models, in which the targeted tissues succumb to as described previously (10). As shown in Fig. 1 , P493-6 tumor formation (3). The spectrum of hematologic neoplasms cells in the NO Myc state grew minimally with more compact depends on the dosage of Myc in bone marrow cells (4). In fact, Myc-induced hematologic transgenic tumors regressed when Myc Author contributions: J.T.Y., M.V.-R., P.L.B., and C.V.D. designed research; J.T.Y., Y.-C.L., levels are diminished (5); however, the Myc dose-dependent target P.G., A.L., M.V.-R., and C.G.E. performed research; P.G. contributed new reagents/analytic genes have not been elucidated. Though these animal models tools; .J.T.Y., Y.-C.L., P.G., C.J., M.V.-R., W.J.C., C.G.E., P.L.B., and C.V.D. analyzed data; and show the importance of Myc dosage on tumorigenesis, they are not J.T.Y., C.J., A.L., W.J.C., and C.V.D. wrote the paper. easily amenable to molecular analysis to identify the putative Myc The authors declare no conflict of interest. * target genes that are induced in the tumorigenic state. This Direct Submission article had a prearranged editor. The critical Myc target genes that are necessary for cellular Data deposition: The data reported in this paper have been deposited in the Gene Ex- transformation and tumorigenesis are beginning to emerge, but pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo/ (accession no. GSE19703). 1 the effects of cell type and cellular context on target gene Present address: Division of Pediatric Hematology-Oncology, Texas Children’sCancer Center, Baylor College of Medicine, Houston, TX 77030. response to Myc remain poorly understood. The technical limi- 2To whom correspondence may be addressed. E-mail: [email protected] or yustein@bcm. tation has been the dearth of models of Myc-mediated tumori- edu. genesis that provide both nontumorigenic and tumorigenic states This article contains supporting information online at www.pnas.org/cgi/content/full/ in the same system. In this regard, the human B-cell line P-493-6 0901230107/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.0901230107 PNAS Early Edition | 1of6 Downloaded by guest on September 25, 2021 Fig. 1. c-Myc levels influence cell morphology, proliferation, and in P493-6 human B cells. (A) Immunoblot of Myc in NO (0.1 μg/mL tet- racycline), LOW (0.1 μg/mL tetracycline and 1 μM estradiol), or HIGH (untreated) Myc states of P493-6 cells. Tubulin serves as loading control. Relative expression levels to LOW Myc state are shown (numbers below the panel) as determined by densitometry of Western blot bands. (B) Histological appearance of P493-6 cells in NO, LOW, or HIGH Myc states. Note several mitotic cells, as well as deep basophilic cytoplasm with vacuoles in the ectopic Myc-expressing cells. (C) Proliferation of P493-6 cells with NO Myc, LOW Myc, or HIGH Myc. Experiment was repeated thrice, and a representative graph is shown (D). Unsupervised hierarchical cluster analysis of P493-6 and human B-cell malignancies using 445 gene probes with >5-fold expression changes that discriminate HIGH Myc from LOW Myc states. Analysis was performed with normal and malignant B-cell samples together with P493-6 in NO, LOW, or HIGH Myc states. Expression of the genes in the different samples is represented by the heat map where every row is a gene and every column is a sample. Red represents genes that are overexpressed, blue underexpressed, and white median expression. The diagnoses for the samples are indicated below the heat map.

chromatin and pale cytoplasm, whereas those with LOW Myc conditions revealed that signals of >10,000 probe sets are dif- grew with the appearance of lymphoblastoid cells. Untreated ferent between the HIGH and NO Myc states, and >8,000 are cells expressing HIGH Myc grew even better, with the appear- different between the LOW and NO Myc states, whereas just ance of Burkitt cells characterized by perinuclear lipid vacuoles over 3,100 are different between EBNA-2 driven LOW Myc and and frequent mitotic figures. Because P-493-6 cells were derived ectopic HIGH Myc states. The intersection of all three sets yields from mature circulating B cells, they do not express Bcl-6, which 959 probe sets whose signals are altered in common. We focused is usually found in germinal center-derived Burkitt lymphoma on genes whose profiles are different between the HIGH and cells. In this regard, we sought to molecularly characterize the LOW Myc states and different between the HIGH and NO Myc P493-6 cell line by arbitrarily selecting 445 gene probes (Table states, but not between the LOW and NO Myc states. We sur- S1) whose expression is >5-fold different between the NO and mise that this set would contain genes that could be pathologic, HIGH Myc states for clustering analysis using expression profiles because they are significantly induced in the HIGH Myc but not of 169 primary human mature B-cell malignancies and 21 normal the LOW Myc state. The 874 probe sets, as shown in the dia- B-cell samples (11). The P493-6 cells with LOW and HIGH Myc gram, represent 603 putative Myc targets that are differentially clustered together and are closest to Burkitt lymphoma (Fig. regulated only upon ectopic Myc expression (Table S2 with list of 1D), thus validating P493-6 cells as a model of Burkitt lym- 874 probe sets identified). phoma. The LOW Myc state, however, also resembles multiple The heat map in Fig. 2B displays the relative expression myeloma. The NO Myc state is closest to chronic lymphocytic intensities of this set of 603 ectopic Myc-driven genes under NO, leukemia, which has a low proliferative status. Though many LOW, and HIGH Myc states. The map reveals a set of genes that genes up-regulated in P493-6 cells are shared with Burkitt lym- are down-regulated in a stepwise manner from NO to LOW to phoma, there are subsets of genes unique to P493-6 cells. HIGH Myc (Table S3). This set is distinct from the set that is Moreover, a small subset of genes (at the top of the clustering down-regulated by HIGH Myc but is apparently induced by diagram; Fig. 1D) is shared with multiple myeloma. As such, the EBNA2 (yellow to red to blue, going from NO to LOW to HIGH P493-6 cells most closely resemble Burkitt lymphoma; however, Myc). A number of known EBNA2 targets, such as IL4I1, CD84, their expression profiles indicate that they also share gene and MFN1 (Table S4) are found in this group (12). Likewise, expression patterns found in other B-cell malignancies there is a set of genes (toward the bottom of the heat map) that Immunophenotyping of HIGH Myc P493-6 cells by flow appears to be repressed by EBNA2 in the LOW Myc state, but is cytometry reveals high CD19 (surface kappa chain) and low increased between the NO and HIGH Myc states (Table S5). CD10 expression, which is characteristic of activated mature B Thus, under the condition of β-estradiol and tetracycline treat- cells. All of these markers are consistent with a Burkitt pheno- ment, the genes are not differentially expressed solely based type. However, these cells also express high CD38 and CD9, upon Myc expression levels but are likely modulated by EBNA2. markers indicative of plasma cell differentiation—perhaps Notwithstanding this caveat, we sought among the 603 genes a reflecting the peripheral blood origin of these cells. These set that is only expressed highly in the HIGH Myc state. markers indicate that the P493-6 cell system does not fully We have identified a set of putative Myc-dependent genes that recapitulate the classic Burkitt phenotype, because they have a display a distinct stepwise increase between LOW and HIGH Myc partial plasma cell phenotype. However, LOW or NO Myc P493- states (yellow – yellow – red in the heat map from NO to LOW to 6 cells have significantly diminished CD38 as well as CD9, sug- HIGH Myc states; Table S6). Among these genes, several are gesting that the P493-6 cells in the HIGH Myc state acquire a previously reported Myc targets (Table S7). Furthermore, gene set partial plasma cell phenotype. enrichment analysis (GSEA) of 167 annotated up-regulated genes unique to the HIGH Myc state reveal an overlap with the Gene Expression Profiling of NO, LOW, and HIGH Myc States. To ZHAN_MMPC_LATEVS set, which contains genes expressed in determine differences in gene expression profiles between the tonsilar plasma cells and the STEMCELL_HEMATOPOIETIC- P493-6 cells expressing ectopic (HIGH Myc), EBNA-2 driven UP set, which are genes increased in hematopoietic stem cells as (LOW Myc), or very low Myc (NO Myc) levels, we analyzed compared with differentiate brain and bone marrow cells multiple biological samples using oligonucleotide microarrays. (Table S8). Seventy-eight annotated down-regulated Myc targets As shown in Fig. 2A, gene expression profiles under the three overlap with the BASSO_GERMINAL_CENTER_CD40_UP set

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.0901230107 Yustein et al. Downloaded by guest on September 25, 2021 To determine the response of JAG2 to Myc induction, we measured protein levels of Jagged2 following removal of tetra- cycline from cells in the NO Myc state (Fig. 3B). Jagged2 protein briskly increased upon reinduction of Myc (Fig. 3B). The cor- relation between Myc and Jagged2 expression in B cells was further documented in human CB33-Myc B cells, EBV-negative human CA46, and Ramos Burkitt lymphoma cells, which have deregulated Myc expression (Fig. S2). The human Raji EBV- positive Burkitt lymphoma cell line, however, had virtually undetectable Jagged2 expression. Intriguingly, EBNA2 protein behaves like an activated intracellular Notch receptor, thereby bypassing the need for canonical Notch signaling or Jagged2 expression to sustain EBV-positive cell proliferation (13). The rapid induction of JAG2 mRNA within 3–5 h posttetracycline release in P493-6 cells is compatible with the response of a direct target gene (Fig. S2C). We have also found from array-based nuclear run-on studies that JAG2 is among genes that are robustly increased at the transcriptional level in response to Myc in the P493-6 cell system (Fig. S2D). These observations suggest that JAG2 mRNA expression could correlate with Myc status in primary human B lymphomas bearing Myc chromosomal translocations to Ig gene loci. In this regard, we used the Oncomine database (https://www.onco- mine.org/resource/login.html) to examine the relationship between Myc chromosomal translocation, Myc expression, and JAG2 expression in the study of Burkitt lymphomas by Hummel et al. (14). Among 200 lymphomas that were karyotyped, 59 samples (including 37 Burkitt specimens) had Myc chromosomal translocations. Com- parison between lymphomas with rearranged Myc versus negative CELL BIOLOGY controls reveals a highly statistically significant increase in the expression of both Myc and JAG2 in the lymphomas with Myc translocations (Fig. S3A). The link between genetic aberrations of Myc and increased JAG2 expression suggests that Myc activation of JAG2 may also occur in primary human lymphomas. In addition to the evidence that Myc directly transactivates JAG2, we found the functional activation of the Notch pathway via detection Fig. 2. Microarray gene expression analysis of P493-6 cells reveals an ectopic ICD Myc target gene set. (A) Venn diagram of differentially expressed genes of the intracellular N , following Myc activation. Jagged2 is a ligand reveals 874 putative pathologic Myc target gene probe sets. Top left circle for the Notch family of receptors (15). Upon ligand-receptor binding, represents probe sets differentially expressed between NO and HIGH Myc the Notch receptors undergo a series of γ-secretase-mediated pro- states, upper right circle encompasses probe sets differentially expressed teolytic cleavages that releases an intracellular portion of the Notch between LOW Myc and HIGH Myc, and lower middle circle represents the receptor, termed NICD (16, 17), which subsequently translocates to sets differentially expressed for NO and LOW Myc. Replicate biological the nucleus and regulates gene expression (18), including the up- samples under each condition were analyzed. (B) Heat map of 874 probe sets regulation of target genes such as Hes-1, Hey-1, Nrarp, and Deltex from A comparing expression of genes in NO, LOW, and HIGH Myc states. (19). We document that upon tetracycline release and subsequent Myc induction, there is increased formation of the Notch1 intra- ICD (Table S8). In addition, we have been able to further validate, with cellular domain N . Ectopic overexpression of JAG2 in P493-6 cells (P493-JAG2) further increased the amount of NICD with Myc quantitative real-time PCR, the differential expression of several C of these genes up-regulated only upon ectopic Myc expression induction (Fig. 3 ). (Fig. S1). Among these genes is JAG2, which encodes the Notch Jagged2 Involvement in Cell Proliferation and Tumorigenesis. Having ligand Jagged2 and is highly up-regulated in the HIGH Myc state. established that JAG2 behaves as a direct Myc target gene, we sought to determine the role of JAG2 in P493-6 cell proliferation HIGH Myc State Increases NOTCH Signaling Components with JAG2 as and tumorigenesis. We performed both gain- and loss-of-function a Direct c-Myc Target. Given our observation that JAG2 is induced studies by overexpressing JAG2 or using interference RNA, more than 5-fold only upon ectopic Myc expression, we focused respectively. Fig. 4A documents that siRNA knockdown of JAG2 on the Notch signaling cascade. We found from replicated bio- decreased proliferation rate when compared with scrambled oli- logical experiments that several members of the Notch pathway gonucleotide controls (siCont). Overexpression of Jagged2 rescued are also up-regulated upon ectopic HIGH Myc expression, the diminished growth rate of P493-6 cells caused by siRNA (Fig. including MFNG, MIB1, Jagged-1, and Notch1. 4B), indicating that the knockdown effect was not due to off-target We sought to determine whether JAG2 is a direct Myc target. siRNA effects. These findings suggest that JAG2 is necessary for fi Because JAG2 is not expressed in the Myc-ER broblast system, we maximal P493-6 cell proliferation in vitro. Previous studies have are unable to determine whether JAG2 induction by 4-hydrox- shown that inhibition of the γ-secretase Presenillin-1 reduces the ytamoxifen would still occur with cycloheximide treatment—a production of the Notch intracellular domain, and subsequent hallmark of some direct Myc target genes. Thus, we evaluated downstream signaling, often resulting in apoptosis (20). To further JAG2 regulatory regions for potential Myc binding sites and found study JAG2 loss-of-function in vitro and in vivo, we used the E boxes within the promoter region and intron 2 of the JAG2 γ-secretase inhibitor (GSI) DAPT, which blocks Notch signaling by genomic sequence. Chromatin immunoprecipitation (ChIP) preventing the processing of the Notch1 receptor to form NICD,but documented that Myc bound both regions containing the non- not the levels of Myc expression. There was a dose-dependent canonical and canonical binding sites (Fig. 3A). inhibition of cell proliferation of the HIGH Myc cells by DAPT as

Yustein et al. PNAS Early Edition | 3of6 Downloaded by guest on September 25, 2021 Fig. 3. Myc binds to the JAG2 gene, induces Jagged2, and activates Notch. (A) Chromatin immunoprecipitation (ChIP) assay for Myc binding to JAG2 in P493- 6 B cells (Left). Anti-human growth factor (HGF) antibody serves as a nonspecific antibody control. Schematic diagram of JAG2 gene highlighting consensus Myc binding sites (Right). Canonical E-box was noted in intron 2 (CACGTG) and predicted Myc-Max binding site within promoter region (CACGCG). (B) Immunoblot of Jagged2 and Myc upon tetracycline release of P493-6 cells. P493-6 cells were treated with 0.1 μg/mL of tetracycline for 72 h, washed twice in PBS, and then placed in RPMI with 10% serum for the noted times. Twenty-five micrograms of whole-cell lysate protein were run on a 10% SDS/PAGE gel. Actin serves as a sample loading control. (C) Immunoblot of Notch intracellular domain (NICD) on removal of tetracycline from P493-6 cells or P493-6 cells overexpressing Jagged2 (P493-JAG2). Tubulin serves as a sample loading control.

compared with DMSO vehicle-treated controls (Fig. 4C). It is Because our studies demonstrate the necessity of JAG2 remarkable that the slower growth of the LOW Myc cells was expression and for full tumorigenic proportionally much less inhibited by DAPT, suggesting that the potential, we sought to determine whether JAG2 gain of function augmented expression of Jagged2 in the HIGH Myc state sensitizes would increase the growth and tumorigenic potential of P493-6 the HIGH Myc cells to DAPT inhibition. Similar findings were cells. Although JAG2 overexpression (Fig. 5A) in P493-6 cells did obtained using a second γ-secretase inhibitor, L685458. These not affect in vitro cell proliferation (Fig. 5B), the in vivo xenograft observations further underscore the potential role of JAG2 as a growth of P493-6 JAG2 cells is significantly increased with an pathological Myc target gene, which stimulates the Notch signaling average P493-6 JAG2 tumor size nearly 50% larger than the con- pathway in the HIGH Myc state. trol P493-6 cells (Fig. 5C). From these observations, we surmise Because anaplastic medulloblastoma is a Myc-driven cancer that JAG2 overexpression provides an in vivo advantage through distinguishable from classic medulloblastoma by high Myc the Notch signaling pathway in the tumor microenvironment. expression (21), we sought to determine the level of Jagged2 It has been reported that the hypoxia inducible factor 1 (HIF-1) expression in human primary medulloblastoma tissue micro- could augment Notch signaling during development (23, 24). arrays (Fig. S3B). We document that with high Myc expression in However, the role of Notch in the hypoxic tumor microenviron- anaplastic medulloblastoma, there is a higher level of Jagged2 ment has not been addressed. In this regard, we determined expression as compared with classic medulloblastoma (Fig. S3B). whether JAG2 gain-of-function activity could be unmasked under The increased Jagged2 expression in high Myc-expressing B cells, hypoxic conditions in vitro. P493-6 cells have diminished pro- Burkitt lymphoma cell lines, and anaplastic medulloblastoma liferation rate in hypoxia (1% oxygen) as compared with non- (Fig. S2B) suggests that high Myc expression in different tumor hypoxic (20% oxygen) conditions. However, ectopic expression of types can be associated with elevated Jagged2 expression. JAG2 could increase the proliferation of hypoxic P493-6 cells We next determined whether inhibition of the Notch signaling (Fig. 5D), suggesting that JAG2 expression provides a growth cascade would inhibit P493-6 tumorigenesis in vivo. We have advantage in hypoxia, particularly for in vivo tumorigenesis. By previously shown that P493-6 cells form tumors in SCID mice fluorescent immunohistochemistry, we further document the (22). These tumors are dependent on Myc, such that doxycycline expression of Jagged2 in regions of hypoxia in P493-6 tumors, (a tetracycline analog) oral administration diminishes Myc levels suggesting that JAG2 may be an important component of and inhibit tumor formation (Fig. 4D). Treatment of P493-6 cells tumorigenesis under in vivo hypoxic conditions (Fig. S3C). with thrice weekly s.c. injections of DAPT for 2 weeks led to a Our studies to identify the downstream Notch and HIF1 targets significant decrease in tumor volume when compared with control under hypoxic conditions alone have not revealed dramatic alter- vehicle-treated mice (Fig. 4D). Furthermore, we document that ations of Notch target genes in JAG2-overexpressing cells, except upon induction of Myc expression, DAPT inhibits Notch activa- for the known Notch target gene, HEY1. HEY1 is induced >3-fold tion, resulting in decreased activated NICD levels (Fig. 4E). These by hypoxia in JAG2-overexpressing P493-6 cells (Fig. S4A). observations support the hypothesis that JAG2 is an ectopic Myc However, upon comparison of in vitro versus in vivo expression target and that inhibition of the Notch signaling pathway dimin- profiles for P493-6 cells, we have observed that mRNA levels for ishes P493-6 tumorigenesis. Notch signaling components or targets, such as and

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.0901230107 Yustein et al. Downloaded by guest on September 25, 2021 Fig. 5. Overexpression of JAG2 in P493-6 cells enhances tumorigenicity and growth under hypoxic conditions. (A) Immunoblot of Jagged2 for parentalP493- 6-expressing empty vector (WT) and polyclonal P493-6 JAG2-overexpressing cells grown in the absence of tetracycline. Ten micrograms of whole-cell lysate pro- teins were used for each lane. Cells were grown in normal media conditions without tetracycline. Film was exposed for only ≈5 seconds. Also shown are immunoblots for Myc and tubulin that serves as a loading control. (B)P493-6 parental and JAG2 overexpressing cells in vitro proliferation curve. Experiment was performed three times with similar results. (C) P493-6 parental and JAG2 overexpressing cells tumor volume comparison. 20 × 106 cells were injected s.c. CELL BIOLOGY Five SCID mice per condition were used, and experiment was repeated with similar results. Tumor volumes shown as mean ± SEM. (D)GrowthassayforP493- 6 parental and JAG2 overexpressing cells under hypoxic conditions (1% oxygen) for the indicated times. Cell numbers, mean ± SD, were counted in triplicates and averaged. Experiment was repeated and produced similar results.

compared with parental P493-6 tumors (Fig. S4B). These findings are supported by the recent findings that heterozygosity for hypoxia inducible factor 1α decreases the incidence of thymic lymphomas, which are associated with decreased Notch signaling Fig. 4. JAG2 participates in the proliferation of P493-6 B-cell line. (A)P493-6 and diminished expression of downstream target genes NRARP proliferation upon knockdown of JAG2 (siJAG2) versus control mock electro- porated cells or those treated with control oligonucleotides (siCont). Experiments and Deltex (25). The detailed events downstream of JAG2 in the were performed three times with similar results. A representative experiment is Notch signaling pathway are clearly important to delineate for a shown. Immunoblot for Jagged2 after transient knockdown with JAG2 siRNA richer understanding of the Myc, JAG2, and Notch connection in (siJAG2) smartpool or control oligonucleotides (siCont). We loaded 25 μgofcel- additional studies, which are beyond the scope of the current lular obtained at 72 h postelectroporation with siRNA (final concen- report. The Notch signaling cascade is involved in the development tration 100 nm) per well. Actin serves as a sample loading control. (B) P493-6 and of several hematological malignancies, particularly T-cell leuke- P493-6 JAG2 cell proliferation upon knockdown of JAG2 (siJAG2) versus siCont. ± mias (26) and multiple myeloma (8). In addition to previous studies Relative growth is shown based on absorbance values as mean SD (n =3).Day4P showing that endogenous Myc is a downstream target of Notch and value using the two-tailed Student’s t test was 0.02 for siJAG2-treated P493-6 fi that Myc and Notch could collaborate in T-cell leukemogenesis JAG2 versus P493-6 WT (control) cells. No signi cance difference was found fi between these cells when treated with siCont. (Inset) Immunoblots for Jagged2 (27, 28), our ndings provide evidence that ectopic Myc trans- and c-Myc, with actin serving as loading control. A total of 30 μgofcellularpro- activation of JAG2 contributes to P493-6 cell tumorigenesis. Our teins obtained at 96 h postelectroporation with siRNA were loaded per well. (C)In observation, via the Oncomine database, of a statistically sig- vitro growth of ectopic Myc-expressing (HIGH Myc) and EBNA2-driven Myc (LOW nificant correlation between Myc chromosomal translocations and Myc)-expressing P493-6 cells treated with vehicle (DMSO; dimethyl sulfoxide) mRNA levels of Myc and JAG2 among 200 primary human lym- control or the γ-secretase inhibitor DAPT at 5 μMand10μM. (D)Invivovolumeof × phomas (14) suggests that Myc could also induce JAG2 expression tumor xenografts in SCID mice treated with 3 /week s.c. DAPT injections over in human Burkitt or Burkitt-like lymphomas. 2 weeks (designated by arrows). DAPT SC treatments started on day +3 posttumor injections of 25 × 106 cells/mouse. Five mice per condition were used, and EBV immortalization of lymphocytes involves the activation of experiment was repeated with similar results. Tumor volumes determined by the Notch pathway by EBNA2, which interacts with the Notch caliper measurements are shown as mean ± SEM. (E) Immunoblot of Notch cofactor CBF1 to stimulate target genes (13). When EBNA2-ER intracellular domain (NICD) to evaluate Notch inhibition via use of the gamma- is activated in P493-6 cells by estradiol, EBNA2 direct activation secretase inhibitor DAPT. P493-6 cells were treated with 0.1 μg/mL of tetracycline of Notch target genes would not require ligand-activated Notch, for72h,washedtwiceinPBS,andthenplacedinRPMIwith10%serumwithDAPT and hence EBNA2-ER-driven P493-6 cells would be resistant to or control DMSO (solvent) for 24 h. Actin serves as a sample loading control. γ-secretase inhibitors, as we have documented. A recent study reports that selected B-cell lines are sensitive to the GSIs, sug- gesting a therapeutic role for GSIs in B lymphoid malignancies MAML2, increased significantly in the tumor xenograft samples. (20). Interestingly, Jagged2 levels are increased in samples from Furthermore, P493-6 JAG2 tumors have increased expression of multiple myeloma patients (8), which often have Myc over- Notch downstream targets genes NRARP and Deltex, when expression (8, 11). Our demonstration of an increased expression

Yustein et al. PNAS Early Edition | 5of6 Downloaded by guest on September 25, 2021 of JAG2 in the higher grade, HIGH-Myc-expressing anaplastic function to provide a hypoxic growth advantage independent of medulloblastoma versus classic medulloblastoma further sug- Notch target genes (30). Further evidence suggests that NICD gests the potential importance of the direct Myc target gene enhances HIF1 recruitment to its target gene promoters, JAG2 in tumorigenesis. resulting in increased HIF1 activity in vitro (23). HIF function Because P493-6 cells were derived from mature B lymphocytes would further increase the tumorigenicity of P493-6 as we have that express cell surface markers indicative of Burkitt lymphoma as previously observed with P493-6 cells overexpressing a stabilized well as CD9, a marker of plasma cell differentiation, the response of HIF-1 mutant (22). JAG2 to HIGH Myc could have resulted from a more mature B cellular state that partially mimics multiple myeloma cells. Intrigu- Materials and Methods ingly, a large study showed that Burkitt lymphomas have frequent ChemicalsandreagentswerepurchasedfromSigma-Aldrich.SmartpoolsiRNAs expression of multiple myeloma 1/IRF4 (MUM1), suggesting a were purchased from Dharmacon, Inc. and used according to manufacturer’s partial plasma cell phenotype in a subset of Burkitt lymphomas (29). fi instructions. Burkitt lymphoma cell lines were obtained from ATCC. Lymphoid Thus, in addition to de ning Jagged2 as a direct Myc target gene in cell lines and P493-6 human B-cell line was cultured as described (31). Chro- P493-6 cells, we also provide the proof of concept through these matin immunoprecipitation, immunoblotting, and RNA isolation from P493-6 functional studies that JAG2 is candidate pathological target gene cells was as described (31). Immunohistochemistry for hypoxia was performed that deserves further study in other systems. using pimonidazole hydrochloride (Hypoxyprobe) from Natural Pharmacia Our JAG2 loss-of-function studies suggest that Jagged2 is International and according to instructions. Goat anti-Jagged2 antibody necessary for the full growth and tumorigenic potential of P493-6 (Santa Cruz) was used according to instructions. Tissue microarrays were fi cells. However, Jagged2 overexpression is insuf cient to increase stained with the anti-Jagged2 antibody (1:400) using Dako Biotin Blocking P493-6 cell proliferation under nonhypoxic conditions. Intrigu- System. Microarray analyses were performed with Affymetrix GeneChip ingly, a hypoxic in vitro growth advantage of Jagged2 over- Human 133 2.0 Arrays. Gene expression analysis for gene clustering was per- expressing P493-6 cells correlated with an increase in the formed as described (11). The Oncomine database is accessible at https://www. tumorigenicity of the P493-6 JAG2 cells in vivo. Immunohis- oncomine.org/resource/login.html. Full materials and methods are found in SI tochemistry of P493-6 tumors further showed abundant expres- Materials and Methods. sion of JAG2 in hypoxic regions. These observations agree with a previous study suggesting that Notch signaling is increased in ACKNOWLEDGMENTS. WethankD.Eick(Munich,Germany)fortheP493-6cells hypoxia through the direct interaction between the intracellular and R. Dalla-Favera (New York, NY) for the CB33 and CB33-Myc cell lines. This Notch receptor domain NICD and HIF-1 (24). We speculate that work was supported by a Maryland TEDCO Grant 2008-MSCRFE-0156-00 and ICD National Institutes of Health Grants R01 CA57341 and R01 CA51497 (to C.V.D.) increased N levels through Jagged2 expression could recip- and the Leukemia Lymphoma Society (C.V.D). J.T.Y is supported by National rocally inhibit factor-inhibiting HIF (FIH) and augment HIF Institutes of Health Institutional Training Grant T32CA60441.

1. Meyer N, Penn LZ (2008) Reflecting on 25 years with MYC. Nat Rev Cancer 8:976–990. 17. Shimizu K, et al. (2002) Integrity of intracellular domain of Notch ligand is 2. Eilers M, Eisenman RN (2008) Myc’s broad reach. Genes Dev 22:2755–2766. indispensable for cleavage required for release of the Notch2 intracellular domain. 3. Murphy DJ, et al. (2008) Distinct thresholds govern Myc’s biological output in vivo. EMBO J 21:294–302. Cancer Cell 14:447–457. 18. Ehebauer M, Hayward P, Martinez-Arias A (2006) Notch signaling pathway. Sci STKE 4. Smith DP, Bath ML, Metcalf D, Harris AW, Cory S (2006) MYC levels govern 2006:cm7. hematopoietic tumor type and latency in transgenic mice. Blood 108:653–661. 19. Iso T, Kedes L, Hamamori Y (2003) HES and HERP families: Multiple effectors of the – 5. Felsher DW, Bishop JM (1999) Reversible tumorigenesis by MYC in hematopoietic Notch signaling pathway. J Cell Physiol 194:237 255. lineages. Mol Cell 4:199–207. 20. Kogoshi H, Sato T, Koyama T, Nara N, Tohda S (2007) Gamma-secretase inhibitors – 6. Pajic A, et al. (2000) Cell cycle activation by c-myc in a Burkitt lymphoma model cell suppress the growth of leukemia and lymphoma cells. Oncol Rep 18:77 80. 21. Stearns D, et al. (2006) c-myc overexpression causes anaplasia in medulloblastoma. line. Int J Cancer 87:787–793. Cancer Res 66:673–681. 7. Kaiser C, et al. (1999) The proto-oncogene c-myc is a direct target gene of Epstein-Barr 22. Gao P, et al. (2007) HIF-dependent antitumorigenic effect of antioxidants in vivo. virus nuclear antigen 2. J Virol 73:4481–4484. Cancer Cell 12:230–238. 8. Houde C, et al. (2004) Overexpression of the NOTCH ligand JAG2 in malignant plasma 23. Zheng X, et al. (2008) Interaction with factor inhibiting HIF-1 defines an additional – cells from multiple myeloma patients and cell lines. Blood 104:3697 3704. mode of cross-coupling between the Notch and hypoxia signaling pathways. Proc 9. Thomas M, et al. (2007) Notch activity synergizes with B-cell-receptor and CD40 Natl Acad Sci USA 105:3368–3373. – signaling to enhance B-cell activation. Blood 109:3342 3350. 24. Gustafsson MV, et al. (2005) Hypoxia requires notch signaling to maintain the 10. Lombardi L, Newcomb EW, Dalla-Favera R (1987) Pathogenesis of Burkitt lymphoma: undifferentiated cell state. Dev Cell 9:617–628. Expression of an activated c-myc oncogene causes the tumorigenic conversion of EBV- 25. Bertout JA, et al. (2009) Heterozygosity for hypoxia inducible factor 1alpha decreases infected human B lymphoblasts. Cell 49:161–170. the incidence of thymic lymphomas in a p53 mutant mouse model. Cancer Res 69: 11. Chesi M, et al. (2008) AID-dependent activation of a MYC transgene induces multiple 3213–3220. myeloma in a conditional mouse model of post-germinal center malignancies. Cancer 26. Chiang MY, et al. (2008) Leukemia-associated NOTCH1 alleles are weak tumor Cell 13:167–180. initiators but accelerate K-ras-initiated leukemia. J Clin Invest 118:3181–3194. 12. Spender LC, et al. (2006) Cell target genes of Epstein-Barr virus transcription factor 27. Weng AP, et al. (2006) c-Myc is an important direct target of Notch1 in T-cell acute – EBNA-2: Induction of the p55alpha regulatory subunit of PI3-kinase and its role in lymphoblastic leukemia/lymphoma. Genes Dev 20:2096 2109. survival of EREB2.5 cells. J Gen Virol 87:2859–2867. 28. Palomero T, et al. (2006) NOTCH1 directly regulates c-MYC and activates a feed- 13. Hayward SD (2004) Viral interactions with the Notch pathway. Semin Cancer Biol 14: forward-loop transcriptional network promoting leukemic cell growth. Proc Natl Acad Sci USA 103:18261–18266. 387–396. 29. Gualco G, et al. (2009) Frequent expression of multiple myeloma 1/interferon 14. Hummel M, et al.; Molecular Mechanisms in Malignant Lymphomas Network Project regulatory factor 4 in Burkitt lymphoma. Hum Pathol 40:565–571. of the Deutsche Krebshilfe A biologic definition of Burkitt’s lymphoma from 30. Coleman ML, et al. (2007) Asparaginyl hydroxylation of the Notch fi – transcriptional and genomic pro ling. N Engl J Med 354:2419 2430. domain by factor inhibiting hypoxia-inducible factor. J Biol Chem 282: 15. Luo B, Aster JC, Hasserjian RP, Kuo F, Sklar J (1997) Isolation and functional analysis of 24027–24038. a cDNA for human Jagged2, a gene encoding a ligand for the Notch1 receptor. Mol 31. Kim JW, Gao P, Liu YC, Semenza GL, Dang CV (2007) Hypoxia-inducible factor 1 and Cell Biol 17:6057–6067. dysregulated c-Myc cooperatively induce vascular endothelial growth factor and 16. Fortini ME (2002) Gamma-secretase-mediated proteolysis in cell-surface-receptor metabolic switches hexokinase 2 and pyruvate dehydrogenase kinase 1. Mol Cell Biol signalling. Nat Rev Mol Cell Biol 3:673–684. 27:7381–7393.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.0901230107 Yustein et al. Downloaded by guest on September 25, 2021