Page 81 - 2019_08-Haematologica-web

ZFP36L1 enhancer DNA methylation in MF
post-ET MF, respectively) by mechanisms that are still poorly understood and are clinically and morphologically indistinguishable from primary MF.3
MF has been intensively studied from the genetic per- spective;4,5 in fact, the modified World Health Organization (WHO) diagnostic criteria for Philadelphia chromosome-negative myeloproliferative neoplasms require the demonstration of a genetic marker of clonal hematopoiesis (JAK2V617F, CALR or MPL mutations).6 The frequency of mutations on relevant epigenetic genes (i.e., DNMT3A, EZH2 and ASXL1) suggests that MF might have an epigenetic component that, to our knowledge, remains poorly characterized.5 So far, epigenetic changes such as DNA methylation have been scarcely addressed in MF7 partly due to the limited changes in promoter DNA methylation compared to those in other hematologic malignancies, as previously published by our group.8 DNA methylation of CpG islands (CGI) (mostly on putative promoter regions) has been traditionally studied in both normal and neoplastic hematopoiesis.9,10 However, high- throughput platforms offer a wider coverage of the genome, allowing a better understanding of DNA methy- lation dynamics in regions distant from CGI.11 In this regard, enhancer regions have been characterized as potentially relevant sites of DNA methylation outside CGI.12-14 Chromatin immunoprecipitation-sequencing studies have enabled reliable mapping of genome-wide active enhancer regions based on histone modifications (e.g., H3K4me1 and H3K27Ac),15,16 allowing the identifica- tion of enhancers playing a role in dynamic transcriptional regulation during hematopoiesis.17
The present work describes a comprehensive genome- wide analysis of DNA methylation in MF patients, cou- pled with a gene expression analysis and information on functional chromatin states, compared with those of sam- ples from healthy donors.16 Focusing on potential epige- netic alterations in enhancer regions, we identified ZFP36L1 as a potential tumor suppressor gene with rele- vance for the pathogenesis of MF.
Methods
Patients’ samples and clinical data
Samples from MF patients (n=39) were bone marrow, granulo- cytes or total peripheral blood cells. The MF cohort comprised cases of primary MF (n=22), post-ET MF (n=7) and post-PV MF (n=10). Peripheral blood cells from healthy donors (n=6) were used as control samples in this study. All patients were diagnosed using the 2008 version of the WHO classification system of hema- tologic malignancies.18 Data on JAK2V617F mutation status were ret- rospectively available for all patients, whereas no data on CALR and MPL mutations were available. The patients’ data are accessi- ble from the Gene Expression Omnibus (GSE118241).
Samples and patients’ data were provided by the Biobank of the University of Navarra and were processed following standard operating procedures approved by the local Ethics & Scientific Committee. Prior to the collection of samples, all patients consent- ed to the use of their data and to the use of stored material for research purposes.
DNA methylation profiling
DNA methylation was assessed using a Human-Methylation 450K Bead-Chip kit (Illumina, Inc., San Diego, CA, USA) and the data were analyzed by Bioconductor open source software. The
analytical pipeline implemented several filters to exclude technical and biological biases and take into account the performance char- acteristics of Infinium I and Infinium II assays.19 Differentially methylated CpG were defined as previously described.13,19 Details on the experimental procedures, annotation of CpG sites, detec- tion of differentially methylated regions, and Gene Ontology analysis20 are described in the Online Supplementary Methods.
Identification of candidate genes targeted by aberrant DNA methylation in enhancers
Data on gene expression profiling from primary MF and healthy peripheral blood samples were obtained from the publicly avail- able Gene Expression Omnibus accession bank number GSE26049.21 Data were further processed using R and the open source Limma package.22 Further details are described in the Online Supplementary Methods.
Luciferase reporter assays
The CpG-free vector (pCPG-L), kindly provided by Dr. Michael Rehli,23 was used to clone the ZFP36L1 enhancer region. Luciferase experiments were performed in triplicate and the details are described in the Online Supplementary Methods. Primer sequences are available in Online Supplementary Table S1.
ZFP36L1 binding motif search
To further validate the potential relevance of the ZFP36L1 gene
in MF, the DREME motif discovery algorithm24 was used to assess enrichment of genes with the ZFP36L1 consensus binding sequence among those genes differentially expressed in MF [false discovery rate (FDR)≤ 0.05].
Overexpression of ZFP36L1
A vector containing the ZFP36L1 open reading frame was kind-
ly provided by Dr. Murphy and subcloned into a PL-SIN-GK vec- tor.25 Further details are described in the Online Supplementary Methods.
Statistical analysis
For parametric group comparisons one-way analysis of variance (ANOVA) with the Dunnet correction was used, whereas for non- parametric group comparisons the Kruskall-Wallis test with the Dunn correction was employed. Paired data were analyzed with a Friedman non-parametric test with the Dunn correction for mul- tiple comparisons, for the data with single measurements. Two- way ANOVA with the Tukey correction was used for data with multiple paired measurements. All tests were performed using Prism 7TM software (GraphPad, La Jolla, CA, USA).
Details of other experimental procedures are given in the Online Supplementary Methods.
Results
Myelofibrosis is characterized by a specific DNA methylation pattern enriched in enhancer regions
In order to provide an exhaustive analysis of the DNA methylation profile in patients with MF, we analyzed the DNA methylome of patients with primary MF, secondary MF (including post-ET/post-PV MF) and healthy donors as controls, using the Human- Methylation 450K array. The first result worth highlighting was the epigenetic similari- ty between primary and post-ET/post-PV MF. Interestingly, with a FDR<0.05, no differentially methylat- ed CpGs were found between primary and secondary MF. Furthermore, we did not identify any differentially methy-
haematologica | 2019; 104(8)
1573