(EBF1) in Hodgkin lymphoma - Nature

Leukemia (2013) 27, 671–679 & 2013 Macmillan Publishers Limited All rights reserved 0887-6924/13 www.nature.com/leu

ORIGINAL ARTICLE

Role of early B-cell factor 1 (EBF1) in Hodgkin lymphoma V Bohle1, C Do¨ring2, M-L Hansmann2 and R Ku¨ppers1 A hallmark of classical Hodgkin lymphoma (cHL) is that the B-cell-derived Hodgkin and Reed–Sternberg (HRS) tumor cells have largely lost the B-cell-typical gene expression program. The factors causing this ‘reprogramming’ of HRS cells are only partly understood. As early B-cell factor 1 (EBF1), a major B-cell transcription factor, is downregulated in HRS cells, we analyzed whether this downregulation contributes to the lost B-cell phenotype and tested the consequences of EBF1 re-expression in cHL cell lines. EBF1 re-expression caused an upregulation of B-cell genes, such as CD19, CD79A and CD79B, although the B-cell genes FOXO1 and PAX5 remained lowly expressed. The re-expression of CD19, CD79A and CD79B occurred largely without demethylation of promoter CpG motifs of these genes. In the cHL cell line L-1236 fitness decreased after EBF1 re-expression. These data show that EBF1 has the ability to reintroduce part of the B-cell signature in cHL cell lines. Loss of EBF1 expression in HRS cells therefore contributes to their lost B-cell phenotype. Notably, in the cHL cell line KM-H2 destructive mutations were found in one allele of EBF1, indicating that genetic lesions may sometimes have a role in impairing EBF1 expression. Leukemia (2013) 27, 671–679; doi:10.1038/leu.2012.280 Keywords: B-cell phenotype; CD19; CD79; EBF1; Hodgkin lymphoma

INTRODUCTION Classical Hodgkin lymphoma (cHL) is a common malignant lymphoma in the western world. The Hodgkin and Reed– Sternberg (HRS) tumor cells are rare and usually account for only a few percent of cells in the tumor tissue, whereas the vast majority of cells in the lymphoma microenvironment represent inflammatory cells. Although HRS cells originate in nearly all cases from B cells,1–3 they have a profound lack of B-cell-typical gene expression.4 This includes cell surface markers (CD19, CD79A),5 signaling molecules (Syk, Lyn, Blk)4 and transcription factors (Oct2, BOB.1).6,7 Furthermore, HRS cells express markers of other hematopoietic lineages such as CCL17,8 GATA-3,9 Notch1,10 and inhibitor of differentiation and DNA-binding 2 (ID2).11,12 cHL is unique among lymphoid malignancies in the extent to which the lymphoma cells have lost the gene expression pattern of their normal precursor cells and have upregulated expression of non-Bcell genes.13 It has been speculated that this ‘reprogramming’ is of pathogenetic relevance for HRS cells.13 Early B-cell factor 1 (EBF1) is a central transcription factor in B cells.14–16 In B cells, it operates as a homodimer in cooperation with two other major B-cell transcription factors, that is, E2A and PAX5.14,16 EBF1 is expressed in all stages of B-cell development except plasma cells. EBF1 not only induces expression of numerous B-cell genes, but at least in murine B cells it also represses factors of other hematopoietic lineages. For example, EBF1 suppresses the myeloid gene CEBPA, the T-cell transcription factor Notch1 and the natural killer cell factor ID2.15 EBF1 also seems to be involved in epigenetic modifications such as demethylation of the CD79A promoter17 and methylation of histones.16 Conditional knockout mice have shown that EBF1 is a B lineage commitment factor and essential for the survival of pro-B cells, marginal zone and B1 B cells, whereas germinal centers were formed but not maintained without EBF1.18,19

In cHL, the function of the three central B-cell transcription factors EBF1, E2A and PAX5 is compromised. E2A is expressed in HRS cells but often at low level,11,20 and is functionally inactivated by the E2A inhibitors ID2 and activated B-cell factor 1, which are highly expressed in HRS cells.9,11,12 PAX5 is present in most cHL cases but there is large variation in the number and intensity of positive HRS cells.11,20–22 EBF1 mRNA is absent or expressed only at very low level in cHL cell lines11,20 and also weak or absent in primary cases in comparison with germinal center B cells (see genechip data from ref. Tiacci et al.23). Beside ID2 and activated B-cell factor 1, the T-cell transcription factor Notch1 might further influence this B-cell transcription factor network. Notch1 is strongly expressed in HRS cells and inhibits the expression of E2A and EBF1.10,24 In addition to unbalanced transcription factors, epigenetic features are also deregulated in HRS cells and might influence the aberrant gene expression in these cells. B-cell genes such as SYK, POU2AF1 (BOB.1, OBF1) and CD79B have methylated promoters in cHL cell lines and primary HRS cells and are therefore silenced.25,26 A genome-wide DNA methylation analysis of cHL cell lines showed that B-cell genes were preferentially found among those with hypermethylated promoters.27 In this study, we analyzed the relevance of the low or absent EBF1 expression for the phenotype of HRS cells. We wondered whether EBF1 silencing has an impact on the aberrant gene expression of HRS cells or whether its loss influences the deregulated methylation pattern of B-cell genes.

MATERIALS AND METHODS Cell culture and B-cell isolation Cell lines L-1236, L-428, KM-H2, Raji and SUP-HD1 were cultured in RPMI1640 medium with stable glutamine supplemented with 10% fetal calf serum (Biochrom AG, Berlin, Germany). Cell line HDLM-2 was cultured in

1 Institute of Cell Biology (Cancer Research), University of Duisburg-Essen, Medical School, Essen, Germany and 2Senckenberg Institute of Pathology, University of Frankfurt, Frankfurt/Main, Germany. Correspondence: Professor R Ku¨ppers, Institute of Cell Biology (Cancer Research), University of Duisburg-Essen, Medical School, Virchowstr. 173, 45122 Essen, Germany. E-mail: [email protected] Received 10 August 2012; revised 17 September 2012; accepted 18 September 2012; accepted article preview online 1 October 2012; advance online publication, 23 November 2012

EBF1 in classical Hodgkin lymphoma V Bohle et al

672 RPMI-1640 medium with 20% fetal calf serum and U-HO1 in Iscove’s modification of Dulbecco’s medium/RPMI-1640 (4:1) with 20% fetal calf serum. All media were supplemented with 1% penicillin–streptomycin (Invitrogen, Darmstadt, Germany). Germinal center B cells were isolated from human tonsils as CD77-positive cells, using the MACS system (Miltenyi Biotech, Bergisch-Gladbach, Germany).

Western blot Western blot analysis to demonstrate the endogenous and exogenous EBF1 expression was performed using standard conditions and the following antibodies: anti-EBF1 (#H00001879-M01, Abnova, Heidelberg, Germany, 1:100–1:10 000; 2nd ab: #115-036-062, Jackson Immuno Research, Hamburg, Germany, 1:2000), anti-glyceraldehyde 3-phosphate dehydrogenase (#sc-31915, Santa Cruz, 1:200; 2nd ab: #sc-2350, Santa Cruz, Heidelberg, Germany, 1:10 000). EBF1 target gene expression was analyzed using the same conditions as above and the following antibodies: antiCD79A (#ab-79414, Abcam, Cambridge, UK, 1:500; 2nd ab: #711-036-152; Jackson Immuno Research, 1:10 000), anti-b-tubulin (#69126, MP Biomedicals, Eschwege, Germany, 1:200; 2nd ab: #115-036-062, Jackson Immuno Research, 1:2000).

Sequence analysis of EBF1 Sanger sequencing was performed to analyze the HL cell lines for mutations in EBF1. A 3130 Genetic Analyzer (Applied Biosystems, Darmstadt, Germany) and the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) were used. Four PCR products were generated to amplify the coding region of EBF1, of which three were sequenced directly and one (*) was cloned in pGEM-T easy before sequencing because of four different isoforms, which are also present in CD77-positive tonsillar germinal center B cells. The following primers were used for amplifying EBF1 complementary DNA (cDNA) (50 –30 , forward and reverse): TTCAAGGGGGAGGAGATTTTCC, CCGGTAGTGAATTCCGTTATTGG; AATCCAA CTTCTTCCACTTCGTCC, GGAGTAGCATGTTCCAGATAAGAG; GTCAATGTGG ATGGCCATGTCC (*), GTTGTCCACTGAACGAATTCACG (*); GGAAATCATTCT GAAGAGAGCGG, CTCTGGGACTTGTATCAGATTACTC. Cell line KM-H2 produced two different reverse transcriptase (RT)-PCR products for the first amplicon. The following primers were used to analyze exons 1 and 2 of KM-H2 at the DNA level to reveal the reason for these two amplificates (50 –30 , forward and reverse): TTCAAGGGGGAGGAGATTTTCC, CAGCAGCTG CCGCTGCC; CGGCTGCTTCTCAAAGTGAGC, GCAGACAGCTCCAGGTCC.

Lentivirus production and transduction The constructs used are based on the pGIPZ-plasmid (Thermo Scientific, Schwerte, Germany/Open Biosystems, St. Leon-Rot, Germany) in which the puromycin resistance gene and the short hairpin RNA were exchanged by EBF1 cDNA (kindly provided by Dr Andreas Braüninger). By co-transfection (GeneJuice, Merck, Darmstadt, Germany/Millipore, Schwalbach, Germany) of the modified pGIPZ, psPAX2 (Addgene plasmid 12260, Didier Trono, Lausanne, Switzerland) and pMD2.G (Addgene plasmid 12259, Didier Trono) viral particles were produced in 293T cells. The viral supernatants were harvested 3 days post transfection and titered on fresh 293T cells. B-cell lines were transduced with 8 multiplicity of infection supplemented with polybrene (final concentration 5 mg/ml). For subsequent experiments, transduced cells were isolated by fluorescence-activated cell sorting (FACSDiva; BD Biosciences, Heidelberg, Germany), gating on turbo green fluorescent protein (tGFP)-positive, propidium iodide-negative cells. Transduced cells were analyzed on a BD FACSCanto (BD Biosciences).

RNA isolation and cDNA synthesis To analyze EBF1 target gene expression in transduced cells, total RNA was prepared from 10 000 sorted tGFP-positive cells by RNeasy Micro Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol using 75 ml RLT buffer, without DNA digestion and elution in 12 ml water. Ten microliters of total RNA were reverse transcribed using the Sensiscript Reverse Transcription Kit (Qiagen) according to the manufacturer’s protocol using random hexamer primers and the RNase inhibitor RNasin Plus (Promega Corporation, Mannheim, Germany).

exponential phase of the PCR was determined in pretests. The primer sequences are as follows (50 –30 , each forward and reverse): ACTB (b-actin), AGCCTCGCCTTTGCCGATC, AGCGGGCGATATCATCATCC; GAPDH, CCACATCGCTCAGACACCATG, TGAAGGGGTCATTGATGGCAAC; EBF1, GTACCATGCTGGTCTGGAGTG, GTGTGACTTCCACAACACCAGG; CD19, CAA CCTGACCATGTCATTCCACC, CACAGGCAGAAGATCAGATAAGCC; CD79A, ATCTGGTACCCTGGGACTGC, GGACCTTGTGCATCCACAGG; CD79B, AGCCTCG GACGTTGTCACG, GATTCCGGTACCGGTCCTC; PAX5, GTCCCAGCTTCCAGTCA CAG, CGGAGACTCCTGAATACCTTCG; ID2, CTCGCATCCCACTATTGTCAGC, GAACACCGCTTATTCAGCCACAC and NOTCH1, GAATGGCGGGAAGTGTG AAGC, TGCAGGCATAGTCTGCCACG. The cycling program consists of 95 1C for 3 min, followed by 27–45 cycles of 95 1C for 15 s/60 1C for 15 s/72 1C for 20 s.

Real-time RT-PCR of EBF1 target genes in transduced cells Quantitative real-time PCR analysis was performed on an ABI Prism 7900HT Fast Real-Time PCR System (Applied Biosystems) using predesigned, intron-spanning assays (Applied Biosystems): ACTB (Hs99999903_m1), CD79B (Hs00236881_m1), ID2 (Hs00747379_m1), CD19 (Hs00174333_m1), PAX5 (Hs00277134_m1), NOTCH1 (Hs01062014_m1), FOXO1 (Hs01054576_m1), and TaqMan Universial PCR MasterMix, No AmpErase UNG (Applied Biosystems). Each cell line was transduced independently two to three times and each gene was measured in duplicates or quadruplicates.

Affymetrix genechips of transduced HL cells Cell lines L-428 and L-1236 were transduced three times each with EBF1 or control vector constructs and positive cells were sorted 8 days post infection for RNA isolation. In all, 150 ng total RNA (RNeasy Micro Kit, Qiagen) of the transduced cell lines L-428 and L-1236 were amplified by the Ambion WT Expression Kit (Applied Biosystems). Labeling and hybridization were performed using the GeneChip WT Terminal Labeling and Hybridization Kit (Affymetrix, Mu¨nchen, Germany). Washing and staining was done by the standard Affymetrix GeneChip protocol (Version 2) in the GeneChip Fluidics Station 450 (Affymetrix), the measurement was performed on the GC Scanner 3000 7G (Affymetrix). The genechip data are available through the GEO Omnibus database with accession number GSE41493.

Bisulfite sequencing of promoter regions of EBF1 target genes in transduced cells DNA was isolated from 100 000 tGFP-positive sorted cells with the Gentra Puregene Kit (Qiagen) and eluted in 25 ml elution buffer. The DNA was denatured by adding 2.5 ml 3 M NaOH and incubation at 42 1C for 30 min. Unmethylated cytosines were modified by adding 255 ml 3.9 M sodium bisulfate (pH 5.0), 15 ml hydroquinone, 2.5 ml H2O and incubation at three cycles of 3 h for 55 1C and 5 min for 95 1C, followed by 3-h incubation at 55 1C. DNA was cleaned with the DNA Clean and Concentrator Kit (Zymo Research, Freiburg, Germany) and resuspended in 25 ml Tris (10 mM)/EDTA (1 mM). The modified DNA was denatured with 2.5 ml 3 M NaOH and incubation for 15 min at 37 1C and subsequently neutralized with 13.8 ml 9 M ammonium acetate (pH 7.0). The DNA was precipitated with ethanol, resuspended in 50 ml Tris/EDTA and stored at –20 1C. The PCR primer pairs of the promoter regions of CD19, CD79B and PAX5 (a and b region) were designed with the help of MethPrimer28 and have the following sequences (50 –30 , forward and reverse): CD19-1, TATTTTGGTGTTTAGGTTGGAGTGTAGT, CAAAAATATAAACCCCTTAAAATAA AAACC; CD19-2, AAGGGGTTTATATTTTTGTGTAGAAAATAGAA, AAACACC CAACCACAACTCAAAT; CD79B-1, GTTTTGGGTTTTTTTAGATGTTTGATTT, TA CTCCCCTCTATCTATACTTACCC; CD79B-2, AGGATTTTAGTTGTGTTGTTTAA GTTGG, CTAAAAATAAAAACAAACCCCACAAAC; PAX5a, TTGGATGGTTG GGAATTTTG, CCCAAACTTTTATAAAAATTAAAAAAAA; PAX5b-1, TTTTGGAG ATTTTTTTTATTTTTATTTTTTAAT, CCCCATTAACTAAACAACCCACA; PAX5b-2, TTTGTGGGTTGTTTAGTTAATGGGG, AAAAAAACAAAAAATCCCAACCACCAA AAC. Standard PCR was performed, and the amplificates were cloned in pGEM-T easy (Promega Corporation) and sequenced with the PCR primers. The resulting sequences were analyzed with the help of BIQ Analyzer HT.29

Cell proliferation assay Semiquantitative RT-PCR of EBF1 target genes in transduced cells The primers were chosen to produce amplicons of a length of 100–200 bp, to have a highly efficient PCR. Each primer pair spans an intron to prevent amplification of genomic DNA. For each cell line and amplicon, the Leukemia (2013) 671 – 679

To analyze the fitness of transduced cells, 10 000 or 20 000 sorted cells were cultured for 2 days in 100 ml conditioned medium in a well of a 96-well plate. Subsequently, the glycolysis rate was monitored in an enzyme-linked immunosorbent assay reader by a colorimetric change of & 2013 Macmillan Publishers Limited


673 the CellTiter96 AQueous One Solution Cell Proliferation Assay (Promega Corporation). The measurement was performed in triplicates, and the experiment was repeated once.

RESULTS EBF1 expression in cHL cell lines To examine EBF1 expression levels in HRS cells, we performed a protein immunoblot for the HL cell lines HDLM-2, SUP-HD1, KMH2, L-428, L-1236 and U-HO1 in comparison with their proposed natural counterpart, that is, germinal center B cells. No EBF1 expression is seen for KM-H2, L-428 and L-1236 cells (Figure 1). HDLM-2 and SUP-HD1 have as much EBF1 as germinal center B cells if compared with the absolute amount of protein loaded on the blot, but a much lower EBF1 expression if normalized to the loading control glyceraldehyde 3-phosphate dehydrogenase. Cell line U-HO1 expresses high amounts of EBF1. Thus, in general EBF1 expression in cHL cell lines is much lower than in germinal center B cells. Owing to the lack of an antibody suitable for immunohistochemistry, we could not analyze EBF1 protein expression in primary cHL. To clarify whether mutations in EBF1 cause the lack of its expression, we sequenced the coding region of EBF1 from amplified cDNA of the cell lines L-1236, L-428, KM-H2, HDLM-2, U-HO1 and SUP-HD1. No peculiarities were seen except in KM-H2. Here the first cDNA amplicon showed two bands of different length. By analyzing this region in KM-H2 DNA, we detected two mutations affecting the donor splice site at the exon 1/intron 1 junction of one allele causing usage of an alternative donor splice site within exon 1 and a frameshift because of shortening of exon 1 by 56 bp (Figure 2). All other cell lines are wild type for EBF1 (data not shown). Therefore, mutations in EBF1 do not seem to be the cause for the reduced or absent EBF1 protein expression in general, although the destructive mutation of one allele of EBF1 in KM-H2 likely contributes to absent EBF1 protein in this line.

Figure 1. Western blot analysis of endogenous EBF1 levels of untransduced cHL cell lines. Fifteen micrograms of protein are loaded per lane. GCB, germinal center B cells.

EBF1 re-expression in cHL cell lines To re-express EBF1 in cHL cell lines, a lentiviral expression system was used. The EBF1 construct contains genetic information of tGFP and EBF1, whereas the control construct expresses only tGFP (Figure 3a). Semiquantitative RT-PCRs (Figure 3b) and protein immunoblots in the time frame of 3 to 10 days post transduction (Figure 3c) show expression of EBF1 mRNA and protein in all transduced cHL cell lines.

Expression of EBF1 target genes in transduced HL cell lines To test whether the low or absent expression of EBF1 in HRS cells contributes to their lost B-cell phenotype, we analyzed the consequences of EBF1 re-expression in these cells on the expression of known EBF1 target genes. By semiquantitative RTPCR, we studied the positively regulated target genes CD19, CD79A, CD79B and PAX5 and the potentially negatively regulated genes NOTCH1 and ID2 in sorted tGFP-positive cells between days 3 and 10 post transduction in comparison with the negative control (Figure 4a and data not shown). A clear upregulation of CD19, CD79A and CD79B in the EBF1-transduced cell lines L-1236, L-428 and KM-H2 is seen. Notch1 showed either no change (L-428), or a minor (KM-H2) or more pronounced upregulation (L-1236), although a downregulation was expected based on studies in the mouse.15 PAX5 and ID2 did not change their expression level on EBF1 re-expression (data not shown). Although CD79A mRNA is strongly upregulated on EBF1 re-expression, CD79A protein was not detectable by immunoblot in transduced L-1236, L-428 and U-HO1 cells (data not shown). We aimed to validate and quantify these findings by real-time RT-PCR analysis of the EBF1 or control vector-transduced cell lines L-1236, L-428, KM-H2 and U-HO1. We again clearly detected upregulation of the transcripts for CD19 and CD79B, in a range from 2- to 4000-fold (Figure 4b). Again, PAX5, and ID2 did not show a regulation, and Notch1 was upregulated only in L-1236 cells. In this analysis, we also included Foxo1, as this transcription factor was described as a further EBF1 target gene.30 However, no regulation of Foxo1 was seen (Figure 4b). Thus, the quantitative RT-PCR analysis validates the semiquantitative RT-PCR results and extend them to the cHL line U-HO1. These results encouraged us to investigate the influence of EBF1 re-expression on gene expression in the cHL cell lines in a global way. We transduced cell lines L-1236 and L-428 with the EBF1 and the negative control constructs, sorted tGFP-positive cells 8 days post infection and used their RNA for analysis on the Human Gene 1.0 ST array. The unsupervised hierarchical clustering of 464 genes with a minimum standard deviation of one

Figure 2. Sequence analysis of EBF1 in KM-H2. (a) The wild-type (wt) sequence of the 30 end of exon 1 and the 50 end of intron 1 is shown in the upper diagram. The lower diagram shows the corresponding sequence of KM-H2. This cell line has a G substitution in the GT donor splice site and a G insertion 4 bp further 30 in intron 1. (b) A comparison of the protein and mRNA sequences of wt EBF1 and the sequence of the mutated EBF1 allele of KM-H2 is shown. Both sequences start with the same 26 codons. Owing to the mutated splice site at the exon 1/intron 1 junction in one allele of EBF1 in KM-H2 an internal cryptic splice site in exon 1 is used, causing a deletion of 56 bp of exon 1 (underlined) in the mRNA. This also leads to a frameshift, which causes a premature stop codon (gray TGA). Hence, a heavily truncated protein is made from this mRNA. & 2013 Macmillan Publishers Limited

Leukemia (2013) 671 – 679


674

Figure 3. EBF1 re-expression in cHL cell lines. (a) Lentiviral constructs on the basis of pGIPZ. (b) Semiquantitative RT-PCR of EBF1 re-expression in transduced HL cell lines. (c) Western blot analysis of EBF1 re-expression in transduced HL cell lines. Control, positive/negative PCR control; dx, days post transduction; ev, empty vector control; GCB, germinal center B cells; ut, untreated cells.

(Supplementary Figure 1) shows that the differences between the two cell lines were larger than between EBF1-expressing and EBF1-negative cells. However, the three replicates for each line and condition clustered together, indicating that consistent effects of EBF1 expression occurred in both lines. The regulation of genes in each cell line is mostly smaller than twofold (false discovery rate p5%). In L-1236, 34 of 298 differentially expressed genes have a regulation larger than twofold (including EBF1) and in L-428 this is true for 21 of 113 differentially expressed genes (Supplementary Table 1). The EBF1-induced genes in L-428 include, for example, CXCL9, CD40 and FOS, and in L-1236, we see upregulation of IL7, MAPKBP1 and PLXNA1 (plexin A1) (Supplementary Table 1). Several genes were also downregulated on re-expression of EBF1 (Supplementary Table 1). When focusing on genes induced by EBF1 in both cHL cell lines with at least a twofold upregulation, we identified only four genes, namely CCL22, SEMA7A, FCER2 (CD23) and CRLF2, besides EBF1 itself (Table 1). Thus, EBF1 re-expression influenced the expression level of more than a hundred genes in L-428 and L-1236 cells, including several B-cell markers. Notably, the two lines are quite heterogenous in their behavior.

post transduction by bisulfite sequencing of cloned PCR products. For comparison, we also determined the methylation pattern of the promoters of the three genes in CD77-positive germinal center B cells. As expected, the normal B cells showed unmethylated CpG promoter motifs for the three genes (Figure 5). In contrast, the CD19 and CD79B promoters of the cHL cell lines transduced with the control vector are highly methylated (Figure 5). The PAX5a promoter is in L-1236 strongly methylated but shows a heterogenous pattern in L-428 and KM-H2 cells. The PAX5b promoter is unmethylated in all cell lines. On EBF1 re-expression some CpGs change the trend of their methylation status, but overall, there is little effect (Figure 5). The exception is the PAX5a promoter of L-1236, which becomes strongly demethylated when EBF1 is re-expressed. In other promoters and cell lines, only single CpGs are demethylated, for example, CpG1 (second amplicon) in the CD19 promoter of KM-H2, CpG 5 and 6 in the CD79B promoter (second amplicon) of L-1236 and CpG 4 and 5 (first amplicon) in the CD79B promoter of KM-H2. There are also some CpGs that show increased methylation because of EBF1 expression. Hence, re-expression of EBF1 in cHL cell lines has relatively little impact on the methylation status of EBF1 target genes.

Analysis of methylation pattern in EBF1-transduced HL cell lines As HRS cells and cHL cell lines are known to have highly methylated promoters of many B-cell genes,25–27,31 and EBF1 seems to be involved in DNA demethylation,17 we studied whether enforced expression of EBF1 in the cHL cell lines L-1236, L-428 and KM-H2 by lentiviral transduction influences the DNA methylation status of the EBF1 target genes CD19, CD79B and PAX5 (a and b promoter). The cell lines were analyzed 9 days

Analysis of fitness of EBF1-transduced HL cell lines It has been speculated that the loss of the B-cell gene expression program may be a survival strategy of HRS cells with a nonfunctional B-cell receptor.3 As some B-cell genes are re-expressed on enforced EBF1 expression in HL cell lines, we wondered whether these changes would be sufficient to cause a reduced survival of the cells. Therefore, the fitness of the EBF1 reexpressing L-1236, L-428, KM-H2, U-HO1 was tested. We sorted

Leukemia (2013) 671 – 679

& 2013 Macmillan Publishers Limited


675

Figure 4. Expression of EBF1 target genes after EBF1 transduction. (a) Semiquantitative RT-PCR of positively and negatively regulated EBF1 target genes. b-Actin is used for normalization. For the cell lines L-1236 and KM-H2, two independent semiquantitative RT-PCRs are shown. (b) Real-time RT-PCR of positively and negatively regulated EBF1 target genes. Typical results of different transductions are summarized in this figure. Corresponding b-actin and empty vector controls are used for DDCt calculation. Control, positive/negative PCR control; ev, empty vector control; ut, untreated cells. Table 1.

Common differentially expressed genes among the EBF1-transduced cell lines L-1236 and L-428

Gene symbol CCL22 SEMA7A FCER2 CRLF2

FC (L-1236)

FC (L-428)

Transcript ID

2.9 2.6 2.5 2.3

2.0 2.5 5.1 2.5

7996022 7990345 8033420 8171105

Gene name Chemokine (C-C motif) ligand 22 Semaphorin 7A, GPI membrane anchor (John Milton Hagen blood group) Fc fragment of IgE, low-affinity II, receptor for (CD23) Cytokine receptor-like factor 2

Abbreviations: EBF1, early B-cell factor 1; FC, fold change; IgE, immunoglobulin E. All genes are significant with a false discovery rate of p5% and both have a minimum regulation of twofold.

tGFP-positive cells 3 or 5 days post transduction, let them recover for 2 days and performed a cell proliferation assay in comparison with control-transduced cells. The fitness (glycolysis rate) of EBF1transduced L-1236 cells is weaker than of control-transduced cells (Figure 6). We did not detect this effect in L-428, KM-H2 and U-HO1 cells (data not shown). The approach used may, however, underestimate the effect of EBF1 on survival, because it has been described that apoptotic cells lose green fluorescent protein fluorescence.32 DISCUSSION One of the most peculiar features of HRS cells in cHL is that these mature B-cell-derived tumor cells have largely lost their B-celltypical gene expression pattern. It has been speculated that this consistent and dramatic ‘reprogramming’ of the HRS cells is of & 2013 Macmillan Publishers Limited

pathogenetic relevance, for example, by promoting the survival of HRS precursor cells with crippled B-cell receptor.4 Hence, it is important to understand the factors that have essential roles in the loss of the B-cell gene expression program of HRS cells. A number of factors have been identified that contribute to this downregulation of B-cell genes in HRS cells. These include inhibition of E2A by ID2 and activated B-cell factor 1, downregulation of transcription factors Oct2, PU.1 and BOB1, aberrant expression of the T-cell transcription factor Notch1, constitutive activity of STAT5a and epigenetic silencing of many B-cell genes.33 In this study, we addressed the issue whether also the low or undetectable level of EBF1 contributes to the lost B-cell phenotype of HRS cells. The enforced expression of EBF1 in cHL cell lines indeed partially reconstituted the B-cell status of these cells, which was shown by semiquantitative RT-PCR and real-time RT-PCR. Leukemia (2013) 671 – 679


676

Figure 5. Methylation pattern of EBF1-transduced cells in comparison with control cells. Methylation status of each sequenced CpG is shown. Each block represents one PCR amplicon and each row one single clone. Red boxes symbolize methylated CpGs, blue ones unmethylated CpGs and white ones could not be evaluated by the program BiQ Analyzer HT. All CpGs of the first CD19 amplicon belong to the promoter region, whereas CpG 1 to 6 of the second CD19 amplicon are part of the promoter and CpGs 7 to 11 are part of exon 1. In case of CD79B, all CpGs of the first amplicon and CpG 1 to 5 of the second amplicon are of the promoter. CpGs 6 to 14 are part of exon 1 and CpGs 15 and 16 (which could not be evaluated) are part of intron 1. The PAX5a amplicon lies completely in the promoter region. The first amplicon of PAX5b and CpG 1 to 12 of the second amplicon are part of the promoter, whereas CpGs 13 to 36 are in exon 1. Leukemia (2013) 671 – 679



677

Figure 6. Cell proliferation assay of EBF1- and control-transduced cells. Sorted EBF1- and control vector-transduced tGFP-positive L-1236 cells were cultured for 2 days. Subsequently, the glycolysis rate was monitored in an enzyme-linked immunosorbent assay (ELISA) reader by a colorimetric change of the CellTiter96 AQueous One Solution Cell Proliferation Assay (Promega Corporation), which contains a tetrazolium compound [3-(4,5-dimethylthiazol-2-yl)-5-(3carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt; MTS]. The difference is significant with a P-value of 5.4 10 6 calculated with the paired two-sample t-test. One of two repetitions with similar results are shown.

Increased transcription of the B-cell genes CD19, CD79A and CD79B was observed in all cell lines analyzed. For CD79A, we tested also for protein expression by immunoblot analysis, but no protein was detectable. This may on the one hand be due to instability of the protein in the absence of immunoglobulin expression to allow assembly of a stable B-cell receptor. On the other hand, the re-expression of CD79A mRNA may be restricted to only a fraction of the HRS cells, so that not sufficient protein is made to enable detection by immunoblot. Absence of EBF1 target gene protein expression after short-term expression of transfected EBF1 gene was also seen by Hertel et al.,20 who showed activity of a co-transfected CD19 reporter gene but no expression of endogenous CD19 by flow cytometric analysis.20 PAX5, a direct EBF1 target, and FOXO1, which is at least in murine B cells positively regulated by EBF1, were not re-expressed on EBF1 expression in the HL cell lines. That EBF1 does not cause a full reappearance of the B-cell program in the HL cell lines is certainly not surprising, because most B-cell genes are positively regulated by numerous transcription factors, so that enforced expression of EBF1 alone is likely not sufficient to reactivate all these genes. For example, CD79A expression is presumably coregulated by Ets, EBF1, IKAROS and Sp1,34 CD79B by Sp1, Ets, OCT, IKAROS and EBF1,35,36 and PAX5 by EBF1, STAT5, SPI1, IRF4, IRF8 and NF-kB.37,38 We also studied the influence of EBF1 re-expression on the transcript levels of ID2 and NOTCH1, because these suppressive factors for B-cell genes were expected to be downregulated by EBF1, based on studies in the mouse.15 However, no downregulation was observed. For Notch1, there was even an upregulation of transcript levels seen in L-1236 cells. However, Notch1 expression might not be completely aberrant in human B cells, as its expression has recently been described in human germinal center B cells.39 The retained expression of Notch1 and ID2 in HL cell lines with enforced expression of EBF1 is likely a further factor why only a moderate re-expression of typical EBF1 target genes was observed. Notch1 functions as an inhibitor of E2A, EBF1 and their target genes in cHL cell lines,24 and ID2 prevents DNA binding of E2A.11,12 As E2A is hence repressed in HRS cells and as two-thirds of the EBF1 target genes are coregulated by E2A,14 the retained ID2 expression presumably impaired a strong upregulation of many EBF1 target genes in the EBF1-expressing HL cell lines. By gene chip analysis, we searched for further genes showing changes in gene expression on re-expression of EBF1. The microarray studies generally confirmed the quantitative RT-PCR results, especially for CD79A and NOTCH1 in L-1236 (data not shown). As the quantitative RT-PCR is more sensitive than the & 2013 Macmillan Publishers Limited

microarray, some of the analyzed genes disappear in the background of the chip or cannot be considered as significant such as CD79A in L-428 and CD79B in L-1236 (data not shown). Interestingly, over 100 and 200 genes in L-428 and L-1236, respectively, showed a significant regulation (false discovery rate p5%), albeit mostly at a low fold-change. Nevertheless, 34 genes in L-1236 and 21 genes in L-428 showed a more than twofold change in transcript levels on EBF1 re-expression. The overlap of the regulated genes was rather small in the two cell lines, likely reflecting the already known heterogeneity of HRS cells and cHL cell lines.40,41 Focusing on commonly regulated genes, we identified CCL22 (MDC, chemokine (C-C motif) ligand 22/ macrophage-derived chemokine), SEMA7A (CD108, semaphorin 7A, GPI membrane anchor), FCER2 (CD23, low-affinity immunoglobulin epsilon Fc receptor) and CRLF2 (cytokine receptor-like factor 2) as direct or indirect EBF1 target genes in HRS cells. CCL22 attracts activated T cells and is normally expressed by activated B lymphocytes and dendritic cells.42 It is also expressed and secreted by HRS cells and might have an influence on the typical cHL microenvironment.43,44 The upregulation of CCL22 in HRS cells on EBF1 re-expression indicates that this chemokine is a target of EBF1. Sema7a has a critical role in inflammatory immune responses.45 EBF1 positively regulates expression of Sema7a in murine B cells.16,18 Its function in B cells is still unknown. The FCeR CD23 is expressed by naive B cells and other hematopoietic cells. Whether primary HRS cells express CD23 is controversial,46–49 but we observed in our analysis a medium to strong expression in the HL cell lines, which is further increased by EBF1. Thus, CD23 is a further B-cell molecule that is influenced in its expression by the low EBF1 levels in HRS cells. Finally, CRLF2, the expression of which is downregulated by EBF1, builds together with the IL7Ra chain the receptor for thymic stromal lymphopoietin, which has a role in early B-cell development.50 Studies in the mouse indicated that EBF1 influences the methylation status of target genes. However, in the studies presented here, there was hardly any demethylation of the promoters of the EBF1 target genes CD19, CD79B and PAX5 detectable, with exception of the PAX5a promoter in L-1236 cells, which showed a strong demethylation on EBF1 re-expression. Perhaps, the epigenetic structure of these gene promoters is too stable to be affected in HRS cells by several days of re-expression of EBF1. The fact that we nevertheless observed increased transcription of the CD19, CD79B and PAX5 genes could mean that low level transcription of these genes is possible even in the presence of methylated CpGs in their promoter regions. Alternatively, but not mutually exclusively, the increased expression of the three genes may stem from only a small fraction of HRS cells that opened the gene loci on EBF1 expression, whereas most cells keep a methylated and silent structure of the promoters. We also tested whether enforced re-expression of EBF1 affected the fitness of HL cell lines. However, three of the lines analyzed did not show any effect, and only L-1236 cells showed a moderate, but significant reduced fitness (Figure 6). The experimental setup may underestimate the effect of EBF1 on cell survival, because we had to sort the transduced cells as tGFP-positive cells before the cell proliferation assay, and it is known that dead cells lose green fluorescent protein expression. Hence, cells that died early after EBF1 re-expression before sorting were excluded from the measurement. Nevertheless, as we did not observe massive cell death specifically in the EBF1-transduced cells before sorting, and as only one of four cHL lines tested showed any effect of EBF1 silencing on survival, it seems that EBF1 re-expression is not sufficient to compromise the viability of the HRS cells. In summary, EBF1 has the ability to reconstitute part of the B-cell status of HRS cells but is not capable to repress aberrantly expressed factors of other hematopoietic lineages such as Notch1 and ID2. EBF1 has little influence on the methylation pattern of its Leukemia (2013) 671 – 679


678 target genes. Overall, the downregulation of EBF1 expression in HRS cell contributes to the lost B-cell phenotype and hence participates in the ‘reprogramming’ of these tumor cells. In rare instances, as seen here for KM-H2, genetic lesions in the EBF1 gene may contribute to loss of EBF1 expression. CONFLICT OF INTEREST The authors declare no conflict of interest.

ACKNOWLEDGEMENTS We thank Gwen Lorenz for expert technical assistance, Klaus Lennartz for cell sorting and Jens Stanelle for helpful discussions. This work was supported by the Deutsche Krebshilfe (108687).

REFERENCES 1 Ku¨ppers R, Rajewsky K, Zhao M, Simons G, Laumann R, Fischer R et al. Hodgkin disease: Hodgkin and Reed-Sternberg cells picked from histological sections show clonal immunoglobulin gene rearrangements and appear to be derived from B cells at various stages of development. Proc Natl Acad Sci USA 1994; 91: 10962–10966. 2 Mu¨schen M, Rajewsky K, Braüninger A, Baur AS, Oudejans JJ, Roers A et al. Rare occurrence of classical Hodgkin’s disease as a T cell lymphoma. J Exp Med 2000; 191: 387–394. 3 Kanzler H, Ku¨ppers R, Hansmann ML, Rajewsky K. Hodgkin and Reed-Sternberg cells in Hodgkin’s disease represent the outgrowth of a dominant tumor clone derived from (crippled) germinal center B cells. J Exp Med 1996; 184: 1495–1505. 4 Schwering I, Braüninger A, Klein U, Jungnickel B, Tinguely M, Diehl V et al. Loss of the B-lineage-specific gene expression program in Hodgkin and Reed-Sternberg cells of Hodgkin lymphoma. Blood 2003; 101: 1505–1512. 5 Watanabe K, Yamashita Y, Nakayama A, Hasegawa Y, Kojima H, Nagasawa T et al. Varied B-cell immunophenotypes of Hodgkin/Reed-Sternberg cells in classic Hodgkin’s disease. Histopathology 2000; 36: 353–361. 6 Stein H, Marafioti T, Foss HD, Laumen H, Hummel M, Anagnostopoulos I et al. Down-regulation of BOB.1/OBF.1 and Oct2 in classical Hodgkin disease but not in lymphocyte predominant Hodgkin disease correlates with immunoglobulin transcription. Blood 2001; 97: 496–501. 7 Re D, Muschen M, Ahmadi T, Wickenhauser C, Staratschek-Jox A, Holtick U et al. Oct-2 and Bob-1 deficiency in Hodgkin and Reed Sternberg cells. Cancer Res 2001; 61: 2080–2084. 8 van den Berg A, Visser L, Poppema S. High expression of the CC chemokine TARC in Reed-Sternberg cells. A possible explanation for the characteristic T-cell infiltratein Hodgkin’s lymphoma. Am J Pathol 1999; 154: 1685–1691. 9 Ku¨ppers R, Klein U, Schwering I, Distler V, Braüninger A, Cattoretti G et al. Identification of Hodgkin and Reed-Sternberg cell-specific genes by gene expression profiling. J Clin Invest 2003; 111: 529–537. 10 Jundt F, Anagnostopoulos I, Forster R, Mathas S, Stein H, Dorken B. Activated Notch1 signaling promotes tumor cell proliferation and survival in Hodgkin and anaplastic large cell lymphoma. Blood 2002; 99: 3398–3403. 11 Mathas S, Janz M, Hummel F, Hummel M, Wollert-Wulf B, Lusatis S et al. Intrinsic inhibition of transcription factor E2A by HLH proteins ABF-1 and Id2 mediates reprogramming of neoplastic B cells in Hodgkin lymphoma. Nat Immunol 2006; 7: 207–215. 12 Renne C, Martin-Subero JI, Eickernjager M, Hansmann ML, Ku¨ppers R, Siebert R et al. Aberrant expression of ID2, a suppressor of B-cell-specific gene expression, in Hodgkin’s lymphoma. Am J Pathol 2006; 169: 655–664. 13 Ku¨ppers R. The biology of Hodgkin’s lymphoma. Nat Rev Cancer 2009; 9: 15–27. 14 Lin YC, Jhunjhunwala S, Benner C, Heinz S, Welinder E, Mansson R et al. A global network of transcription factors, involving E2A, EBF1 and Foxo1, that orchestrates B cell fate. Nat Immunol 2010; 11: 635–643. 15 Pongubala JM, Northrup DL, Lancki DW, Medina KL, Treiber T, Bertolino E et al. Transcription factor EBF restricts alternative lineage options and promotes B cell fate commitment independently of Pax5. Nat Immunol 2008; 9: 203–215. 16 Treiber T, Mandel EM, Pott S, Gyory I, Firner S, Liu ET et al. Early B cell factor 1 regulates B cell gene networks by activation, repression, and transcriptionindependent poising of chromatin. Immunity 2010; 32: 714–725. 17 Maier H, Ostraat R, Gao H, Fields S, Shinton SA, Medina KL et al. Early B cell factor cooperates with Runx1 and mediates epigenetic changes associated with mb-1 transcription. Nat Immunol 2004; 5: 1069–1077. 18 Vilagos B, Hoffmann M, Souabni A, Sun Q, Werner B, Medvedovic J et al. Essential role of EBF1 in the generation and function of distinct mature B cell types. J Exp Med 2012; 209: 775–792.

Leukemia (2013) 671 – 679

19 Gyo¨ry I, Boller S, Nechanitzky R, Mandel E, Pott S, Liu E et al. Transcription factor Ebf1 regulates differentiation stage-specific signaling, proliferation, and survival of B cells. Genes Dev 2012; 26: 668–682. 20 Hertel CB, Zhou XG, Hamilton-Dutoit SJ, Junker S. Loss of B cell identity correlates with loss of B cell-specific transcription factors in Hodgkin/Reed-Sternberg cells of classical Hodgkin lymphoma. Oncogene 2002; 21: 4908–4920. 21 Krenacs L, Himmelmann AW, Quintanilla-Martinez L, Fest T, Riva A, Wellmann A et al. Transcription factor B-cell-specific activator protein (BSAP) is differentially expressed in B cells and in subsets of B-cell lymphomas. Blood 1998; 92: 1308–1316. 22 Foss HD, Reusch R, Demel G, Lenz G, Anagnostopoulos I, Hummel M et al. Frequent expression of the B-cell-specific activator protein in Reed-Sternberg cells of classical Hodgkin’s disease provides further evidence for its B-cell origin. Blood 1999; 94: 3108–3113. 23 Tiacci E, Do¨ring C, Brune V, van Noesel CJ, Klapper W, Mechtersheimer G et al. Analyzing primary Hodgkin and Reed-Sternberg cells to capture the molecular and cellular pathogenesis of classical Hodgkin lymphoma. Blood 2012, ealy online Sept. 5. PMID: 22955914. 24 Jundt F, Acikgoz O, Kwon SH, Schwarzer R, Anagnostopoulos I, Wiesner B et al. Aberrant expression of Notch1 interferes with the B-lymphoid phenotype of neoplastic B cells in classical Hodgkin lymphoma. Leukemia 2008; 22: 1587–1594. 25 Doerr JR, Malone CS, Fike FM, Gordon MS, Soghomonian SV, Thomas RK et al. Patterned CpG methylation of silenced B cell gene promoters in classical Hodgkin lymphoma-derived and primary effusion lymphoma cell lines. J Mol Biol 2005; 350: 631–640. 26 Ushmorov A, Leithauser F, Sakk O, Weinhausel A, Popov SW, Moller P et al. Epigenetic processes play a major role in B-cell-specific gene silencing in classical Hodgkin lymphoma. Blood 2006; 107: 2493–2500. 27 Ammerpohl O, Haake A, Pellissery S, Giefing M, Richter J, Balint B et al. Arraybased DNA methylation analysis in classical Hodgkin lymphoma reveals new insights into the mechanisms underlying silencing of B cell-specific genes. Leukemia 2012; 26: 185–188. 28 Li LC, Dahiya R. MethPrimer: designing primers for methylation PCRs. Bioinformatics 2002; 18: 1427–1431. 29 Lutsik P, Feuerbach L, Arand J, Lengauer T, Walter J, Bock C. BiQ Analyzer HT: locus-specific analysis of DNA methylation by high-throughput bisulfite sequencing. Nucleic Acids Res 2011; 39(Web Server issue): W551–W556. 30 Zandi S, Mansson R, Tsapogas P, Zetterblad J, Bryder D, Sigvardsson M. EBF1 is essential for B-lineage priming and establishment of a transcription factor network in common lymphoid progenitors. J Immunol 2008; 181: 3364–3372. 31 Ushmorov A, Ritz O, Hummel M, Leithauser F, Moller P, Stein H et al. Epigenetic silencing of the immunoglobulin heavy-chain gene in classical Hodgkin lymphoma-derived cell lines contributes to the loss of immunoglobulin expression. Blood 2004; 104: 3326–3334. 32 Lowder M, Unge A, Maraha N, Jansson JK, Swiggett J, Oliver JD. Effect of starvation and the viable-but-nonculturable state on green fluorescent protein (GFP) fluorescence in GFP-tagged Pseudomonas fluorescens A506. Appl Environ Microbiol 2000; 66: 3160–3165. 33 Ku¨ppers R, Braüninger A. Reprogramming of the tumour B-cell phenotype in Hodgkin lymphoma. Trends Immunol 2006; 27: 203–205. 34 Leduc I, Cogne M. Regulatory elements of the mb-1 gene encoding the Ig-alpha component of the human B-cell antigen receptor. Mol Immunol 1996; 33: 1277–12786. 35 Akerblad P, Rosberg M, Leanderson T, Sigvardsson M. The B29 (immunoglobulin beta-chain) gene is a genetic target for early B-cell factor. Mol Cell Biol 1999; 19: 392–401. 36 Thompson AA, Wood Jr. WJ, Gilly MJ, Damore MA, Omori SA, Wall R. The promoter and 50 flanking sequences controlling human B29 gene expression. Blood 1996; 87: 666–673. 37 Decker T, Pasca di Magliano M, McManus S, Sun Q, Bonifer C, Tagoh H et al. Stepwise activation of enhancer and promoter regions of the B cell commitment gene Pax5 in early lymphopoiesis. Immunity 2009; 30: 508–520. 38 Hirokawa S, Sato H, Kato I, Kudo A. EBF-regulating Pax5 transcription is enhanced by STAT5 in the early stage of B cells. Eur J Immunol 2003; 33: 1824–1829. 39 Yoon SO, Zhang X, Berner P, Blom B, Choi YS. Notch ligands expressed by follicular dendritic cells protect germinal center B cells from apoptosis. J Immunol 2009; 183: 352–358. 40 Drexler HG. Recent results on the biology of Hodgkin and Reed-Sternberg cells. I. Biopsy material. Leuk Lymphoma 1992; 8: 283–313. 41 Drexler HG. Recent results on the biology of Hodgkin and Reed-Sternberg cells. II. Continuous cell lines. Leuk Lymphoma 1993; 9: 1–25. 42 Schaniel C, Pardali E, Sallusto F, Speletas M, Ruedl C, Shimizu T et al. Activated murine B lymphocytes and dendritic cells produce a novel CC chemokine which acts selectively on activated T cells. J Exp Med 1998; 188: 451–463.



679 43 Maggio EM, Van Den Berg A, Visser L, Diepstra A, Kluiver J, Emmens R et al. Common and differential chemokine expression patterns in rs cells of NLP, EBV positive and negative classical Hodgkin lymphomas. Int J Cancer 2002; 99: 665–672. 44 Poppema S. Immunobiology and pathophysiology of Hodgkin lymphomas. Hematology Am Soc Hematol Educ Program 2005 231–238. 45 Suzuki K, Okuno T, Yamamoto M, Pasterkamp RJ, Takegahara N, Takamatsu H et al. Semaphorin 7A initiates T-cell-mediated inflammatory responses through alpha1beta1 integrin. Nature 2007; 446: 680–684. 46 Angel CA, Warford A, Campbell AC, Pringle JH, Lauder I. The immunohistology of Hodgkin’s disease--Reed-Sternberg cells and their variants. J Pathol 1987; 153: 21–30.

47 Jarrett RF, Armstrong A, Wilkins BS, Jones DB. Immunohistochemical determination of CD23 expression in Hodgkin’s disease using paraffin sections. J Pathol 1991; 164: 345–346. 48 Salama ME, Rajan Mariappan M, Inamdar K, Tripp SR, Perkins SL. The value of CD23 expression as an additional marker in distinguishing mediastinal (thymic) large B-cell lymphoma from Hodgkin lymphoma. Int J Surg Pathol 2010; 18: 121–128. 49 Rowlands DC, Hansel TT, Crocker J. Immunohistochemical determination of CD23 expression in Hodgkin’s disease using paraffin sections. J Pathol 1990; 160: 239–243. 50 Pandey A, Ozaki K, Baumann H, Levin SD, Puel A, Farr AG et al. Cloning of a receptor subunit required for signaling by thymic stromal lymphopoietin. Nat Immunol 2000; 1: 59–64.

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


Leukemia (2013) 671 – 679