Pcg Methylation of the HIST1 Cluster Defines an Epigenetic Marker Of

LETTERS TO THE EDITOR PcG methylation of the HIST1 cluster deﬁnes an epigenetic marker of acute myeloid leukemia

Leukemia (2015) 29, 1202–1206; doi:10.1038/leu.2014.339 genes involved in cell surface and chromatin organization (Supplementary Figure S1B). We observed a significant over- representation of the Polycomb group target genes being Acute myeloid leukemia (AML) is characterized by the clonal differentially H3K27 tri-methylated (Figure 1a), including 10 genes expansion of hematopoietic progenitors that have acquired with known impact in AML pathogenesis. The observed variation numerous genetic and epigenetic alterations. In a substantial in EZH2 activity at specific promoters supports the hypothesis of proportion of cases (~40%), the leukemic cells lack any visible site-specific deregulation of the Polycomb group/EZH2 target chromosomal abnormality (CN-AML); this group is highly hetero- genes in AML rather than a global dysregulation.8 geneous in terms of biology and clinical outcome, which remains Unsupervised hierarchical-clustering analysis of these highly poorly understood.1 The discovery of mutational events that affect H3K27me3 variant promoter regions revealed, among different genes involved in the regulation of hematopoietic commitment/ genomic regions, nine significant clusters with homogeneous differentiation, cell cycle and more recently epigenetics, has enrichment levels (Supplementary Figure S2). Cluster #3 was helped to dissect the molecular pathogenesis and enhance the noteworthy being a robust cluster (dendrogram scale between 0.6 classification of CN-AML.2,3 Indeed, molecular profiling using and 1) comprised of 22 sequential genomic regions, all belonging targeted sequencing approaches has been shown to provide to the HIST1 gene locus located on chromosome 6p22. independent prognostic information, of potential value to guide Hierarchical clustering performed with H3K27me3 enrichment of patient management.4 the 22 HIST1 genomic regions clearly distinguished two groups of Epigenetic changes in AML have been extensively studied, first patients based on their H3K27me3 level: one group with high and looking at specific oncogenic or tumor suppressor loci, then more homogeneous H3K27me3 levels, and the other with low recently at the genome-wide level, where DNA methylation H3K27me3 levels (Figure 1b). These distinct H3K27me3 patterns profiling was used to identify potential biomarkers.5,6 Here we at the HIST1 cluster did not correlate with patient characteristics, used epigenetic profiling to gain further insights into CN-AML, such as age or gender (Figure 1b). Specificity of these differences which is much more poorly characterized compared with rarer in H3K27me3 at the HIST1 cluster, was analyzed by supervised subsets of AML with chimeric oncoproteins consequent upon clustering of H3K27me3 levels at the majority of HIST1 cluster balanced chromosomal rearrangements. As EZH2 is the histone promoter regions on chromosome 6 within our patient cohort H3 methyl transferase catalytic subunit of the Polycomb group (Figure 1c). The heatmap showed that heterogeneity in H3K27me3 complexes that has been most frequently implicated in the levels was restricted to the 22 regions of chromosome 6, as pathogenesis of human malignancies7 we analyzed 35 CN-AML previously revealed by our hierarchical clustering analysis, samples by chromatin immunoprecipitation (ChIP) coupled with whereas the other HIST1 promoter regions were homogeneous hybridization on oligonucleotide promoter arrays (Chip-chip) for across patient samples (Figure 1c and Supplementary Figure S3). genomic H3K27me3 distribution. Through this, we could We have thus identified a clustered genomic region that harbors a identify 586 highly H3K27me3 variable genomic regions across dramatic difference in H3K27me3 level within our group of CN- patients, corresponding to 461 genes (standard deviation40.6; AML patients, indicating that epigenetic deregulation can affect Supplementary Figure S1A). These H3K27me3 variable regions localized chromosomal regions in tumor cells lacking the chimeric across CN-AML were mostly observed in the promoter regions of oncoproteins.

Figure 1. Epigenetic profiling reveals a gain of H3K27me3 at the HIST1 cluster that distinguishes two subgroups of CN-AML patients. (a) The overlap between the highly H3K27me3 variable genes among patient samples, the Polycomb-regulated genes17 and the AML-associated genes (GeneCards database; http://www.genecards.org/) is illustrated with a Venn diagram. Significance of overlap was calculated using the hypergeometric test. H3K27me3 variable genes vs Polycomb target genes; P = 1.248e-11; H3K27me3 variable genes vs AML-associated genes P = 0.001. (b) Heat map showing a two-way hierarchical clustering of H3K27me3 enrichment at the most-variant clusters of promoters encompassing 22 HIST1 promoter regions. Each column represents a patient sample and each row a promoter region. The dendrogram of patient samples distinguishes two groups of patients that are not linked significantly by age (P = 0.711) or sex (P = 0.733). Boxes at the bottom represent sample annotations for age at diagnosis and gender. The blue color represents males and the pink color females. (c) Heat map representing H3K27me3-promoter enrichment of the HIST1 cluster for 35 CN-AML patients. Chromosome 6 is represented at the top of the figure and the HIST1 cluster, including the 55 histone genes in 6 (p22.2-p22.1), is indicated by a red square. Only HIST1 promoter regions are analyzed and gaps between HIST1 promoter regions 440 kbp are indicated. Each line represents a patient sample ordered as in Figure 1b and each column represents a HIST1 promoter region ordered according to the HG18 version of the human genome. The sequential order of the 11 HIST1 genes that display high H3K27me3 variation is shown at the bottom of the figure. (d) Validation of H3K27me3 enrichment by ChIP- quantitativePCR in four CN-AML patients at different HIST1 promoter regions. Data are presented as percentage of bound/input. White bars represent the GAPDH promoter. Light gray represents HIST1 promoter regions that were not variable in our ChIP-chip experiment (HIST1H1E and HIST1H4E). Dark gray represents HIST1 promoter regions that were variable in our ChIP-chip experiment. Black represents HOXD4 promoter. We analyzed two patients that showed high H3K27me3 profiles (0253 and 7159) and two that showed low H3K27me3 profiles (5191 and 6670). (e) Analyzes of H3K27me3 levels by ChIP-qPCR in 51 CN-AML, three normal bone marrow and three CD34+ sorted cord blood samples, at five HIST1 regions. Enrichment was calculated as percentage of bound/input and normalized with HOXD4 and GAPDH. Data are presented as heatmaps. Each column represents a patient sample, sorted by upward-median values for HIST1 enrichment.

Accepted article preview online 8 December 2014; advance online publication, 16 January 2015 Letters to the Editor 1203 HIST1 H3K27me3 enrichment differences were independently (HIST1H2BG-2AE, HIST1H1D, HIST1H4F, HIST1H4G and HIST1H3F-2BH) confirmed by reanalyzing four patient samples using ChIP but not in the two flanking regions, HIST1H1E and HIST1H4E,intwo followed by qPCR (ChIP-qPCR). We confirmed the difference in ‘high H3K27me3’ and in two ‘low H3K27me3’ patients (Figure 1d). H3K27me3 enrichment at the five ‘variable’ HIST1 regions In order to independently extend this observation, we analyzed the

Figure 2. High H3K27me3 HIST1 level associates with NPM1 mutation and good prognosis. (a) Comparison of expression level of four histone genes between high H3K27me3-enriched and low H3K27me3-enriched patients. Results are presented relative to HPRT (hypoxanthine- guanine phosphoribosyltransferase) expression level. Statistical significance was estimated using Mann–Whitney t-test, (***) P ⩽ 0.0001. High refers to the high H3K27me3 level patients, low refers to the low H3K27me3 level patients, nLC stands for non-leukemic cells and refers to normal bone marrow cells or CD34+ sorted cord blood cells. (b) Comparison of normalized-H3K27me3 level of five histone promoter regions (see ChIP-qPCR normalization in supplementary methods) between NPM1mut (n = 46) and NPM1wt (n = 25). (c) Co-occurrence of mutations in nine genes and H3K27me3 HIST1 level in CN-AML samples. Columns show results for each of the analyzed cases. Solid boxes indicate mutated cases. Red boxes correspond to high HIST1 H3K27me3 cases, Green boxes to low HIST1 H3K27me3 cases. (d) Allo-censored leukemia-free survival. Statistical significance was estimated using the long-rank test. (e) Multivariate analysis on LFS-allo. CI indicates confidence interval. Variables considered are HIST1 H3K27me3 level (high vs low), age at diagnosis (o56 vs ⩾ 56 years) NPM1mutation (present vs absent), FLT3- ITD (present vs absent).

H3K27me3 status by ChIP-qPCR, at the same five HIST1 genomic 8.92 months P = 0.0053) (Supplementary Table S2; Figure 2d). locations, in an independent cohort of 51 CN-AML patients. Interestingly, the prognostic significance of HIST1 H3K27me3 was These experiments confirmed the previously described clustered independent of the age and the presence of NPM1 or FLT3 H3K27me3 profile at the HIST1 locus, distinguishing two separate mutations in multivariate analyzes (Figure 2e). groups of patients (Figure 1e). H3K27me3 ChIP experiments using Here, we described a gain in H3K27me3 at a defined region of non-leukemic hematopoietic cells (normal bone marrow, CD34+ the HIST1 cluster as a new epigenetic alteration in CN-AML, sorted cord blood) revealed that HIST1 cluster promoters are not identifying patients with differing clinical outcome. However, normally enriched for H3K27me3 (Figure 1e). This observation was because of a bias in the studied population reflected by the confirmed by the analysis of publicly available ChIP-seq data (GEO, important proportion of patients with high WBC, validation studies GSM773041) that shows absence of the H3K27me3 mark at the are warranted to further characterize this epigenetic signature. HIST1 locus in human CD34+ hematopoietic stem cells Although mechanisms underlying this gain in H3K27me3 are (Supplementary Figure S4). Together, these results highlight that unknown, it is interesting to note that this specific gain in gain of H3K27me3 at the HIST1 locus provides an epigenetic H3K27me3 is not associated with alteration in pathways known to signature that discriminates two subgroups of CN-AML patients. influence the activity of EZH2 (Supplementary Table S1) and might CN-AML samples were split in two groups, according to their be the result of an indirect deregulation of EZH2. Noticeably, this HIST1 H3K27me3 enrichment levels (method described in specific HIST1 gene signature was not highlighted by genomic supplementary information). First, as H3K27me3 is an epigenetic expression profile data,12 and our unpublished data on the same mark of repression that is associated with poor transcription rate,9 cohort. This could be explained by the intrinsic features of the we compared the expression of four HIST1 genes (HIST1H1D, histone genes: redundancy of the five canonical-histone genes, as HIST1H2BH, HIST1H3F, HIST1H4F) by real time quantitative PCR, in well as subtle differences in their corresponding proteins that the two groups (n = 86). Expression of the four HIST1 genes was renders investigations on the function and expression of significantly higher in the ‘low H3K27me3 level’ group, as individual histone genes difficult.13 How deregulation of some compared with the ‘high H3K27me3 level’ group (P-value ⩽ 0.0001; canonical-histone genes could contribute to leukemogenesis is Figure 2a). The inverse correlation observed between H3K27me3 unclear. Growing evidence suggests that mutations in histone levels and H1ST1 expression suggests that the elevated level of genes are associated with hematological malignancies: mutations H3K27me3 might be involved in the transcriptional repression of in HIST1H3B and HIST1H1C have been found in diffuse large B-cell some of the HIST1 cluster genes in CN-AML patient blasts. lymphomas.14,15 Interestingly, focal deletion of a histone gene Next, the two groups were analyzed for clinical and molecular cluster at 6p22, overlapping with our H3K27me3 region, has been characteristics: no significant differences in median age at described in 19% of near-haploid cases of acute lymphoblastic diagnosis were observed, suggesting that H3K27me3 enrichment leukemia.16 In line with these findings, our data suggest a does not correlate with aging (Supplementary Table S1). In common neoplastic mechanism that may alter histone function addition, the two groups were not associated with a specific and regulation affecting nucleosome structure and function by French-American-British class, suggesting that H3K27me3 level at altering histone-DNA interactions, chromatin compaction or the HIST1 locus is not biased by the differentiation stage of the interactions with other effectors that bind to histones. leukemic cells (Supplementary Table S1). Patients with high In conclusion, by using epigenetic profiling we identified two H3K27me3 level had a markedly higher incidence of NPM1 subgroups of CN-AML patients that differ according to H3K27me3 mutation (89 vs 40%; P = 1.75 × 10 − 5) and a substantially lower levels at genes within the HIST1 cluster. The enrichment of incidence of WT1 mutation (0 vs 20%; P = 0.028). No significant H3K27me3 at the HIST1 cluster is associated with NPM1 mutations association was observed with FLT3 (ITD and TKD), IDH1/2, and provides a new molecular marker with potential value in DNMT3A nor ASXL1 mutations (Supplementary Table S1). We diagnosis and prognosis. This example supports the interest of compared H3K27me3 levels on 4 HIST1 promoter regions, in both using epigenetic profiling for identifying new deregulated loci NPM1-mutated and NPM1wt (in 86 AML cases). For each of the linked to pathology. four promoter regions, we observed that the higher H3K27me3 levels were restricted to NPM1-mutated AML cases (all P-values CONFLICT OF INTEREST ⩽ 0.0001; Figure 2b). These findings support the previous observations of low-level expression of some HIST1 genes in The authors declare no conflict of interest. NPM1-mutated AML10,11 and provide possible insight into specific transcriptional programs driven by NPM1 mutations. ACKNOWLEDGEMENTS Co-occurrence analysis reveals a favorable mutation profile (high incidence of NPM1 mutations and no mutations in WT1)4 We thank Anne-Marie Imbert for the help with primary cells; all the Duprez Lab associated with the high H3K27me3 signature (Figure 2c) that was personnel for helpful discussions and the onco-genomic facility at IPC for microarray analysis. This work was supported by the Fondation de France grants 2008002092 in accordance with the better outcome of the high H3K27me3 and 201200029166, by the Laurette Fugain grants ALF 08/04 and 09/08, by the level sub-group of CN-AML patients. Patients with high HIST1 Fondation pour la Recherche Médicale (FRM) grant INE201000518534 and by the H3K27me3 level had a significant longer leukemia-free survival at SIRIC leukemia program (INCa-DGOS-Inserm 6038). GT was supported by the FRM 5 years (allo-grafted patients censored, LFS-allo; 13.33 vs and Cancéropole PACA. AP was supported by Marie Curie research training

© 2015 Macmillan Publishers Limited Leukemia (2015) 1202 – 1221 Letters to the Editor 1206 fellowship (MRTN-CT-2006-035733) and the FRM. Samples of human origin and 4 Patel JP, Gonen M, Figueroa ME, Fernandez H, Sun Z, Racevskis J et al. Prognostic associated data were obtained from the IPC/CRCM Tumor Bank (Biological Resource relevance of integrated genetic profiling in acute myeloid leukemia. N Engl J Med Centre in Oncology), that operates under authorization #AC-2007-33 granted by the 2012; 366: 1079–1089. French Ministry of Research (Ministère de la Recherche et de l’Enseignement 5 Figueroa ME, Lugthart S, Li Y, Erpelinck-Verschueren C, Deng X, Christos PJ et al. Supérieur). This paper presents independent research funded by the National DNA methylation signatures identify biologically distinct subtypes in acute Institute for Health Research (NIHR) under its Programme Grants for Applied Research myeloid leukemia. Cancer Cell 2010; 17:13–27. Programme (grant RP-PG-0108-10093, supporting DG and AI). This work was enabled 6 Deneberg S, Guardiola P, Lennartsson A, Qu Y, Gaidzik V, Blanchet O et al. ’ ’ by the NIHR Biomedical Research Centre based at Guy s and St Thomas NHS Prognostic DNA methylation patterns in cytogenetically normal acute ’ Foundation Trust and King s College London. myeloid leukemia are predefined by stem cell chromatin marks. Blood 2011; 118: 5573–5582. DISCLAIMER 7 Hock H. A complex Polycomb issue: the two faces of EZH2 in cancer. Genes Dev 2012; 26: 751–755. The views expressed are those of the authors and not necessarily those of the NHS, 8 Shih AH, Abdel-Wahab O, Patel JP, Levine RL. The role of mutations in epigenetic the NIHR or the Department of Health. regulators in myeloid malignancies. Nat Rev Cancer 2012; 12: 599–612. 9 Li B, Carey M, Workman JL. The role of chromatin during transcription. Cell 2007; G Tiberi1,5, A Pekowska2,5, C Oudin1, A Ivey3,4, A Autret1, T Prebet1, 128: 707–719. M Koubi1, F Lembo1, M-J Mozziconacci1, G Bidaut1, C Chabannon1, 10 Verhaak RG, Goudswaard CS, van Putten W, Bijl MA, Sanders MA, Hugens W et al. D Grimwade3, N Vey1, S Spicuglia2,6, B Calmels1,5 and E Duprez1,5 Mutations in nucleophosmin (NPM1) in acute myeloid leukemia (AML): 1CRCM (Epigenetic Factors in Normal and Malignant Hematopoiesis), association with other gene abnormalities and previously established gene fi Inserm, U1068; Institut Paoli-Calmettes; Aix-Marseille expression signatures and their favorable prognostic signi cance. Blood 2005; 106: 3747–3754. Université, UM105; CNRS, UMR7258, Marseille, France; 2 11 Vassiliou GS, Cooper JL, Rad R, Li J, Rice S, Uren A et al. Mutant nucleophosmin CIML, CNRS UMR6102, INSERM U631, Aix-Marseille Université, and cooperating pathways drive leukemia initiation and progression in mice. Marseille, France; Nat Genet 2011; 43: 470–475. 3 ’ Department of Medical and Molecular Genetics, King s College 12 Valk PJ, Delwel R, Lowenberg B. Gene expression profiling in acute myeloid London School of Medicine, London, UK and leukemia. Curr Opin Hematol 2005; 12:76–81. 4 GSTS Pathology, Guy’s Hospital, London, UK 13 Marzluff WF, Wagner EJ, Duronio RJ. Metabolism and regulation of E-mail: [email protected] or [email protected] canonical histone mRNAs: life without a poly(A) tail. Nat Rev Genet 2008; 9: 5These authors contributed equally to this work. 843–854. 6Present address: TAGC, INSERM UMR1090, Marseille, France. 14 Lohr JG, Stojanov P, Lawrence MS, Auclair D, Chapuy B, Sougnez C et al. Discovery and prioritization of somatic mutations in diffuse large B-cell lymphoma (DLBCL) by whole-exome sequencing. Proc Natl Acad Sci USA 2012; REFERENCES 109: 3879–3884. 1 Mrozek K, Dohner H, Bloomfield CD. Influence of new molecular prognostic 15 Morin RD, Mendez-Lago M, Mungall AJ, Goya R, Mungall KL, Corbett RD et al. markers in patients with karyotypically normal acute myeloid leukemia: recent Frequent mutation of histone-modifying genes in non-Hodgkin lymphoma. 476 – advances. Curr Opin Hematol 2007; 14: 106–114. Nature 2011; : 298 303. 2 Dohner H, Estey EH, Amadori S, Appelbaum FR, Buchner T, Burnett AK et al. 16 Holmfeldt L, Wei L, Diaz-Flores E, Walsh M, Zhang J, Ding L et al. The genomic Diagnosis and management of acute myeloid leukemia in adults: recommenda- landscape of hypodiploid acute lymphoblastic leukemia. Nat Genet 2013; 45: tions from an international expert panel, on behalf of the European LeukemiaNet. 242–252. Blood 2010; 115: 453–474. 17 Bracken AP, Dietrich N, Pasini D, Hansen KH, Helin K. Genome-wide mapping of 3 Miller CA, Wilson RK, Ley TJ. Genomic landscapes and clonality of de novo AML. Polycomb target genes unravels their roles in cell fate transitions. Genes Dev 2006; N Engl J Med 2013; 369: 1473. 20: 1123–1136.

Supplementary Information accompanies this paper on the Leukemia website (http://www.nature.com/leu) Usefulness of the 2012 European CVD risk assessment model to identify patients at high risk of cardiovascular events during nilotinib therapy in chronic myeloid leukemia

Leukemia (2015) 29, 1206–1209; doi:10.1038/leu.2014.342 arm, 6.8% in the nilotinib 300 mg BID arm and 12.6% in the nilotinib 400 mg BID arm.7 A growing body of evidence suggests that these cardiovascular events may be inﬂuenced Nilotinib is a tyrosine kinase inhibitor (TKI) targeting the BCR-ABL1 by several factors including pro-atherogenic and anti-angiogenic oncoprotein with high potency and shows activity against most properties of nilotinib, propensity to increase glycemia and BCR-ABL1 kinase domain mutants.1 It is approved at 400 mg twice cholesterolemia, nilotinib dosing regimen and risk factors for – daily (BID) for chronic phase (CP)- or accelerated phase (AP)- atherosclerotic cardiovascular diseases (CVD).4 10 However, how chronic myeloid leukemia (CML) with resistance or intolerance to to precisely assess the risk of cardiovascular events during imatinib and at 300 mg BID for newly diagnosed CP-CML.2,3 nilotinib treatment remains unclear. In the general European Whereas nilotinib was initially considered as well-tolerated, population, a model updated in 2012 by the European Society of postmarketing reports raised concerns about an association Cardiology is used to assess the CVD risk and to guide between the drug and onset of arterial occlusions.4–6 In the prevention.11 It relies on CVD history, risk factors and phase 3 randomized trial ENESTnd (Evaluating Nilotinib Efﬁcacy SCORE (Systematic Coronary Risk Evaluation). Its endpoint is and Safety in Clinical Trials Newly Diagnosed Patients) the 5-year the 10-year atherosclerotic CVD mortality. We designed a rate of ischemic cardiovascular events was 2.1% in the imatinib single center retrospective study to ask whether CVD risk

Accepted article preview online 8 December 2014; advance online publication, 20 January 2015