Letters to the Editor 213 PAX5 overexpression is not enough to reestablish the mature B-cell phenotype in classical Hodgkin

Leukemia (2014) 28, 213–216; doi:10.1038/leu.2013.211 The inability of PAX5 to reactivate the of its B-cell-related target in L428-PAX5 raises the question for the reasons of this finding. To accomplish its role as a The PAX5, consistently expressed from the transcription factor, it is essential that PAX5 can bind to the pro-B to the mature B-cell stage, plays a decisive role in B-cell regulatory region of the promoters of its target genes. Therefore, differentiation.1 In derived from mature B cells, the we performed PAX5 chromatin immunoprecipitation (ChIP) expression of PAX5 is maintained at high expression levels in followed by next generation sequencing (ChIP-Seq) to determine virtually all instances. An exception to this rule is the classical the genome-wide accessibility of PAX5-binding sites in (cHL) which—although derived in almost all the cHL cell line L428-PAX5. For comparison, the B-cell lines Raji instances from germinal center B cells—displays a variable and Namalwa were included as they both display an expression and mostly very weak PAX5 expression in the vast majority of program of mature B cells compatible with a functional binding 2 activity of PAX5. All in all, a total of 14 465 PAX5-binding sites cases as compared with normal B cells. In addition, the À 5 expression of many B-cell-specific antigens is absent or strongly (Po10 according to PeakRanger) were identified in at least one reduced in cHL, whereas genes from other hematopoietic of the conditions, 4667 of which can be assigned to a symbol 3 lineages are upregulated. Because PAX5 plays a central role in ( Gene ID) and have an adjusted P-value (Padj)o0.05 as the maintenance of the B-cell phenotype4,5 we reasoned, if determined by DESeq (Supplementary Table S3). re-constitution of PAX5 expression in a cHL-derived cell line PAX5 was able to bind to a comparable number of promoters in would be able to reactivate the PAX5-driven expression of the PAX5-expressing cHL cell line L428-PAX5 as compared B-cell genes. with the B-cell lines (2914 versus 2562; Figure 2a). However, To reestablish the PAX5 expression in the PAX5-negative only a small proportion of the PAX5-binding sites (n ¼ 809; cHL cell line L428, stable transfection with a mifepristone- Supplementary Table S3) overlap between the B-cell lines and the inducible PAX5-expression system (GeneSwitch System, Life PAX5-positive L428 cells (Figure 2a). Most importantly, well- Technologies/Invitrogen, Carlsbad, CA, USA) was performed characterized B-cell-specific PAX5 target genes such as CD19, (Supplementary Methods) and the resulting cell line was CD79A, BLK and BLNK were not bound by PAX5 in the PAX5- designated L428-PAX5. Successful PAX5 induction was tested by expressing Hodgkin cell line. PAX5 binding to genes involved in western blotting and immunocytochemistry, which showed a the B-cell signaling pathway is observed only in B-cell strong nuclear PAX5 expression over a period of time ranging lines (Figure 2a), whereas genes associated with B-cell differentia- from 48 h (Figures 1a and b) to 5 days (data not shown). In tion (TSHR, IKZF1, AICDA, LRRC8A, ADAM17, TCF3, CEBPG, BCL6, order to test if the reexpressed PAX5 is indeed able to evoke DCLRE1C and BCL3) are bound by PAX5 in both L428-PAX5 and transcriptional activity, a Dual-Luciferase Reporter assay was B-cell lines as revealed by DAVID analysis. PAX5 is further carried out. To this end, a reporter construct that contains the associated with cell cycle regulation in B-cell lines and cHL, functional promoter proportion of the PAX5 target gene CD79A whereas only B-cell lines showed an additional association of (pGL3-CD79A) was transfected into L428-PAX5 cells. Reexpressed PAX5 with several metabolic processes. Although PAX5 binds in PAX5 is clearly able to activate the synthetic CD79A promoter both cHL and B-cell lines target genes involved in apoptosis and (Supplementary Figure S1), which demonstrates the suitability cell death, the genes leading to this categorization are different in of our system to study the activation of cellular PAX5 target genes both cell lines. in L428-PAX5. Our ChIP-Seq data were additionally examined by real-time To determine the impact of PAX5 reexpression, we performed DNA-PCR, which allows an independent and quantitative valida- analysis (Affymetrix Human Gene 1.0 ST Array, tion of the ChIP-Seq results for several genes known as PAX5 Affymetrix Inc., Santa Clara, CA, USA) with RNA derived from L428 targets as well as genes newly described in this manuscript. No cells with and without PAX5 overexpression as well as from B-cell binding to PAX5 target genes typically expressed in mature B cells lines as controls. Initially, special attention was given to a set of 68 such as CD19, CD79A, BLK and BLNK could be observed in the genes3 known to be differentially expressed in B-cell and Hodgkin PAX5-producing Hodgkin cell line L428-PAX5 (Figure 2b). samples. Based on these 68 genes, there was no reexpression of Although genes known to be typically expressed in cHL (ATF3,6 B-cell-related genes or reduction of Hodgkin-characteristic genes CCR7,7 GATA38 and TNFRSF89) displayed enriched PAX5 binding, detectable (Figure 1c, left panel). Real-time RT-PCR was addition- the binding of PAX5 has no or only limited impact on the ally carried out for a limited set of B-cell genes (CD19, CD79A, BLNK expression of these genes since there was no significant difference and BLK; Supplementary Figure S2). However no reactivation of between PAX5-expressing and PAX5-deficient L428 cells these genes was detectable by this quantitative assay despite (Supplementary Table S2). Interestingly, genes such as AICDA strong PAX5 expression (Figure 1a) in the mifepristone-induced and BCL6, known to be downregulated in cHL,10 were also covered L428-PAX5 cells. Extension of the bioinformatic analysis to the by PAX5 (ChIP-Seq (Supplementary Table S3) and real-time DNA- global transcriptomic scale revealed a significant PAX5-related PCR (data not shown) in L428-PAX5, which was associated with upregulation of 535 genes without statistical significance for their further downregulation. This argues for a repressive function any KEGG pathway or biological process. In contrast, the 415 of PAX5 under these conditions. downregulated genes were involved in JAK-STAT signaling, the The combined consideration of the PAX5 DNA-binding pattern regulation of cell activation/proliferation and apoptosis/cell and PAX5-associated gene expression reveals several explanations death. In essence, this clearly indicates that PAX5 is unable to for the inability of PAX5 to reactivate the B-cell phenotype in reconstitute the B-cell phenotype in Hodgkin cells despite Hodgkin cells. (i) B-cell-specific genes lack the capability of PAX5 overexpression, nuclear localization and promoter activity of PAX5. binding, a prerequisite for their (re-)activation; (ii) PAX5 binding to

Accepted article preview online 11 July 2013; advance online publication, 23 August 2013

& 2014 Macmillan Publishers Limited Leukemia (2014) 179 – 238 Letters to the Editor 214

Figure 1. Strong overexpression of PAX5 in the cell line L428-PAX5. (a) Western blot analysis of PAX5 in the induced cell line L428-PAX5, the B-cell lines Namalwa and Raji as well as the cHL cell line L428. (b) Immunocytochemistry for PAX5 in wild-type L428, L428-PAX5 (48 h-induced) and Raji cells. (c) Gene expression profile of the cell lines Namalwa and Raji (untreated), L428 and the PAX5-producing L428-PAX5 with and without 5-aza-dC/TSA treatment (5 days). The heatmap is based on the gene list published by Kuppers et al.3 with minor modifications (not including immunoglobulin genes). (M: mifepristone, A: 5-aza-dC, T: TSA).

B-cell inappropriate genes fails to downregulate their expression epigenetic modifications such as DNA promoter methylation in Hodgkin cells; (iii) PAX5 binding to genes associated with or dense histone conformation. One of the most prominent B-cell differentiation results in their further downregulation and histone modifications represents the acetylation of lysines, which (iv) other positive or negative transcriptional regulators seem to correlates positively with gene expression.12 be absent or active, respectively, in cHL. In essence, PAX5 alone Genes such as CD19, CD79A, BLK and BLNK, which represent has no impact on B-cell activation or the reduction of the Hodgkin well-known PAX5 targets are specifically hypoacetylated in cHL phenotype. With regard to the inability of a single transcription cell lines compared with B-cell lines and, vice versa, repressive factor to reconstitute the B-cell identity in Hodgkin cells, our histone marks (H3K27me3) are found in the promoter regions of results fit nicely in the concept most recently provided by typical B-cell genes in cHL cell lines.13 Although murine PAX5 can Bohle et al.11 They demonstrated that the B-cell transcription switch promoters and enhancers of its activated genes from a factor EBF1 alone is also not able to cause full restoration of the repressed state into an activated state (induction of H3K4 B-cell phenotype in Hodgkin cell lines. methylation and H3K9 acetylation, reduction of H3K27me3),14 this The ChIP-Seq data proved that the regulatory regions of mechanism does not seem to be activeforPAX5targetgenesincHL. the promoters of many B-cell-related PAX5 target genes are To eliminate a potential inhibition of PAX5 binding due not accessible to PAX5 in the induced L428-PAX5 cell line. to promoter DNA methylation and/or histone deacetylation, Impaired binding of transcription factors might be caused by a treatment with 5-aza-20-deoxycytidine (5-aza-dC) in combination

Leukemia (2014) 179 – 238 & 2014 Macmillan Publishers Limited Letters to the Editor 215

Figure 2. Identification of PAX5-binding sites in Raji, Namalwa and L428-PAX5 (induced). (a) Venn diagram to demonstrate the number of PAX5-binding sites found by ChIP-Seq in the L428-PAX5 (n ¼ 2105) and in the B-cell lines (n ¼ 1753), as well as shared binding sites (intersection; n ¼ 809). Numbers of corresponding unique gene symbols are given in brackets. KEGG pathways and biological processes identified by DAVID in each section are listed in the tables together with the Benjamini correction of the significantly (Benjamini correction o0.05) enriched terms. (b) Real-time DNA-PCR for selected PAX5-binding sites. Enrichment of the PAX5 target genes from ChIP was calculated as fold change relative to input DNA with error bars representing s.d. from duplicate readings. with Trichostatin A (TSA) was performed. Despite successful for their help with DNA dot blot analysis. We thank the Deutsche epigenetic modification by this treatment (Supplementary Forschungsgemeinschaft (SFB/Transregio 54, subprojects B04 and B06) for supporting Figure S3), an upregulation of CD19, CD79A, BLK and BLNK was this work. not detectable by real-time RT-PCR in the PAX5-positive Hodgkin cell line (data not shown). Additional global Affymetrix gene expression analysis confirmed that there is no shift of the epigenetic-modified and PAX5-producing L428-PAX5 cells toward a B-cell expression L Dimitrova1, V Seitz1, J Hecht2, D Lenze1, P Hansen2,3, program (Figure 1c, right panel). M Szczepanowski4,LMa4,5, E Oker1, A Sommerfeld1, F Jundt6, Taken together, our data clearly demonstrate that the expres- W Klapper4 and M Hummel1 sion of PAX5 with or without global DNA demethylation/histone 1Institute of Pathology, Charite´-Universita¨tsmedizin Berlin, acetylation is not sufficient to induce a B-cell phenotype in HRS Campus Benjamin Franklin, Berlin, Germany; cells. Therefore, additional repressive post-translational histone 2Berlin-Brandenburg Center for Regenerative Therapies, marks or further molecular mechanisms are active in HRS cells Charite´-Universita¨tsmedizin Berlin, Berlin, Germany; preventing the restoration of the B-cell phenotype. 3Institute for Medical Genetics and Human Genetics, Charite´-Universita¨tsmedizin Berlin, Berlin, Germany; 4Institute of Pathology, Kiel University (CAU), Kiel, Germany; CONFLICT OF INTEREST 5Department of Hematology, The First Affiliated Hospital of Zhejiang The authors declare no conflict of interest. University, Hangzhou, China and 6Department of Hematology, Oncology and Tumorimmunology, ACKNOWLEDGEMENTS Charite´-Universita¨tsmedizin Berlin, Campus Virchow-Klinikum, We thank Hedwig Lammert and Erika Berg for their excellent technical assistance, Berlin, Germany Maria Joosten and Rebecca Haberman (Johns Hopkins University, Baltimore, USA) E-mail: [email protected]

& 2014 Macmillan Publishers Limited Leukemia (2014) 179 – 238 Letters to the Editor 216 REFERENCES Hodgkin disease correlates with distinct dissemination of neoplastic cells in 1 Fuxa M, Busslinger M. Reporter gene insertions reveal a strictly B lymphoid- lymphoid organs. Blood 2002; 99: 1109–1116. specific expression pattern of Pax5 in support of its identity function. 8 Atayar C, Poppema S, Blokzijl T, Harms G, Boot M, van den Berg A. Expression J Immunol 2007; 178: 3031–3037. of the T-cell transcription factors, GATA-3 and T-bet, in the neoplastic cells of 2 Foss HD, Reusch R, Demel G, Lenz G, Anagnostopoulos I, Hummel M et al. Hodgkin lymphomas. Am J Pathol 2005; 166: 127–134. Frequent expression of the B-cell-specific activator in Reed-Sternberg cells 9 Falini B, Pileri S, Pizzolo G, Durkop H, Flenghi L, Stirpe F et al. CD30 (Ki-1) molecule: of classical Hodgkin’s disease provides further evidence for its B-cell origin. Blood a new cytokine receptor of the tumor necrosis factor receptor superfamily as a 1999; 94: 3108–3113. tool for diagnosis and immunotherapy. Blood 1995; 85: 1–14. 3 Kuppers R, Klein U, Schwering I, Distler V, Brauninger A, Cattoretti G et al. 10 Tiacci E, Doring C, Brune V, van Noesel CJ, Klapper W, Mechtersheimer G et al. Identification of Hodgkin and Reed-Sternberg cell-specific genes by gene Analyzing primary Hodgkin and Reed-Sternberg cells to capture the molecular and expression profiling. J Clin Invest 2003; 111: 529–537. cellular pathogenesis of classical Hodgkin lymphoma. Blood 2012; 120: 4609–4620. 4 Horcher M, Souabni A, Busslinger M. Pax5/BSAP maintains the identity of B cells in 11 Bohle V, Doring C, Hansmann ML, Kuppers R. Role of early B-cell factor 1 (EBF1) in late B lymphopoiesis. Immunity 2001; 14: 779–790. Hodgkin lymphoma. Leukemia 2013; 27: 671–679. 5 Cobaleda C, Jochum W, Busslinger M. Conversion of mature B cells into T cells by 12 Yoo CB, Jones PA. Epigenetic therapy of cancer: past, present and future. Nat Rev dedifferentiation to uncommitted progenitors. Nature 2007; 449: 473–477. Drug Discov 2006; 5: 37–50. 6 Janz M, Hummel M, Truss M, Wollert-Wulf B, Mathas S, Johrens K et al. Classical 13 Seitz V, Thomas PE, Zimmermann K, Paul U, Ehlers A, Joosten M et al. Hodgkin lymphoma is characterized by high constitutive expression of activating Classical Hodgkin’s lymphoma shows epigenetic features of abortive transcription factor 3 (ATF3), which promotes viability of Hodgkin/Reed-Sternberg differentiation. Haematologica 2011; 96: 863–870. cells. Blood 2006; 107: 2536–2539. 14 McManus S, Ebert A, Salvagiotto G, Medvedovic J, Sun Q, Tamir I et al. 7 Hopken UE, Foss HD, Meyer D, Hinz M, Leder K, Stein H et al. Up-regulation of the The transcription factor Pax5 regulates its target genes by recruiting chromatin- chemokine receptor CCR7 in classical but not in lymphocyte-predominant modifying in committed B cells. EMBO J 2011; 30: 2388–2404.

Supplementary Information accompanies this paper on the Leukemia website (http://www.nature.com/leu)

VPREB1 deletions occur independent of lambda light chain rearrangement in childhood acute lymphoblastic leukemia

Leukemia (2014) 28, 216–220; doi:10.1038/leu.2013.223 hypodiploidy). The TARGET (COG) cohort was analyzed with the Affymetrix GeneChip Human Mapping 500K SNP Array (Affymetrix, Santa Clara, CA, USA). We obtained additional data from a cohort of B-cell acute lymphoblastic leukemia (B-ALL) remains one of the best infant, standard and high-risk B-ALL patients treated on Total genetically characterized cancers. With the introduction of single- Therapies XI-XV and analyzed by investigators at St Jude Children’s nucleotide polymorphism (SNP) microarray technology, there has Research Hospital (SJCRH, N ¼ 265).3,12 The SJCRH cohort included been a recent explosion in genomic investigations and discovery of leukemia and germline CEL files run on the Affymetrix Genome- recurrent alterations in B-ALL, such as PAX5, IKZF1, JAK2 and CRLF2.1 Wide Human SNP Array 6.0 and the Affymetrix GeneChip Human Deletions of the VPREB1 gene, a component of the surrogate light Mapping 500K Array. Previously published paired gene expression chain of the pre-B-cell receptor (pre-BCR), have been observed in data were obtained for both the TARGET and SJCRH samples when childhood B-ALL,2–7 and have been suggested to result from available (GeneChip U133 Plus 2.0 Array, recombination activating gene (RAG) activation and variable Affymetrix).4,12 We also reviewed our published SNP array data (joining) diversity (V(D)J)recombinationbasedonVPREB1’s location on 27 (BL) samples.13 In addition, kappa- in the immunoglobulin lambda (IGL@).3,6,7 In a recent study of expressing B-cells, lambda-expressing B-cells and control relapsed patients with ETV6-RUNX1 translocations, focal VPREB1 monocytes were separated from the whole blood of 10 healthy deletions were presumed to be involved in leukemogenesis but were volunteers. DNA was isolated from the 25 Utah B-ALL samples, not the focus of the reported findings.8 Another study explored 11 B-ALL cell lines, 5 TARGET B-ALL samples and the healthy molecular alterations and outcome in Down syndrome B-ALL and volunteers (RecoverAll Total Nucleic Acid Isolation Kit, Ambion, observed VPREB1 deletions in 18% of their patients, but this deletion Austin, TX, USA; see Supplementary Information). was again not the focus of their investigation.9 Taking advantage of SNP array data were analyzed with Nexus Copy Number 6.1 publicly available microarray data sets and performing additional (BioDiscovery, Inc., El Segundo, CA, USA). Gene expression levels experiments, we found that VPREB1 deletionsarenotpartofnormal were normalized with GC-RMA methods in Partek Genomics Suite V(D)J recombination as they frequently do not involve the VJ (Partek, Inc., St Louis, MO, USA) and plotted on a log2 scale. Each junction nor follow the ordered model of V(D)J recombination. cohort was normalized to itself, and trends of expression change We received Institutional Review Board approval to study by number of allele copies were tested by the method of formalin-fixed paraffin-embedded bone marrow aspirate clots from Jonckheere-Terpstra, as well as two-way comparisons via analysis 25 B-ALL patients treated at Primary Children’s Medical Center of variance and Kruskal–Wallis tests. at the University of Utah. We also obtained the leukemia and VPREB1 is located within the IGL@ locus among the variable germline CEL files from the previously published Therapeutically immunoglobulin (Ig) segments, upstream from the VJ junction. Applicable Research to Generate Effective Treatments (TARGET) V(D)J recombination joins light chain variable (V) Ig segments with Initiative cohort of the National Cancer Institute (NCI), which joining (J) Ig segments by deleting intervening Ig segments (light included high-risk ALL patients treated on the Children’s Oncology chains do not contain diversity (D) Ig segments). Therefore, Group (COG) P9906 trial (N ¼ 221).4,10–12 The P9906 patients physiologic VPREB1 deletions would be expected within a larger demonstrated high-risk features (older age, high white blood cell continuous deletion that contains nearby Ig segments and count, overt central nervous system or testicular involvement) stretches to the 30 border at the VJ junction. However, the without specific prognostic cytogenetic features (ETV6-RUNX1 majority of the VPREB1 deletions observed in the B-ALL microarray translocation, trisomy of 4 and 10, BCR-ABL1 translocation, data sets were focal and did not extend to the VJ junction, and

Accepted article preview online 24 July 2013; advance online publication, 23 August 2013

Leukemia (2014) 179 – 238 & 2014 Macmillan Publishers Limited