King's College London

Research portal

Role of EZH2 in Myelodysplastic Syndromes

Student thesis: Doctoral ThesisDoctor of Philosophy

Occurrence of mutations in the Polycomb (PcG) gene; EZH2 (Enhancer of Zeste Homologue 2) represent a new class of molecular lesions associated with instability in the epigenome of patients with Myelodysplastic Syndrome (MDS). Detection of microdeletion at 7q36.1 or 7q Copy Neutral Loss of Heterozygosity [CN (LOH)] led to the identification of EZH2 mutations. EZH2 is the catalytic component of Polycomb Repressive Complex 2 (PRC2) that trimethylates lysine 27 of histone 3 (H3K27) resulting in gene silencing and recruitment of the sister complex i.e. Polycomb Repressive Complex 1 (PRC1) to target genes. Discovery of EZH2 mutations have shed light on the involvement of other PcG and PcG interacting proteins i.e. Jumonji (jmj) family of demethylases and DNA methyltrasferase 3A (DNMT3A) in MDS. Investigation of Single Nucleotide Polymorphism (SNP) array abnormalities and mutational analysis of these genes have not been ascertained and therefore I examined cytogenetic aberrations affecting twelve Polycomb (PRC1) and seventeen Jumonji genes using high density SNP 6 arrays. SNP6 data analyzed in this study was generated by our group for previous research projects. I visualised this data using CHromosome Analysis Suite (Chas) from Affymetrix and identified five PRC1 genes (BMI1, PHC1, PHC2, RING1A and RING1B) in 17/91 (19 %) patients with either Copy Number Variations (CNVs) like deletions or amplifications or CN (LOH). Interestingly, the frequency of SNP6 aberrations was high (two times) in Jumonji genes as compared to PRC1. 29/91 patients (31 %) showed either CNVs or CN (LOH) in fifteen (JMJD3, JMJD4, JMJD1B, JMJD2A, JARID2, JMJD1C, JARID1B, JMJD2C, UTX, JARID1C, JARID1A, JMJD2D, JHJD1A, JARID1D and JHJD1B) Jumonji genes. Mutational analysis of patients with SNP6 aberrations was carried out using Sanger or 454 sequencing but no mutations were detected in either the PRC1 or Jumonji genes. To elucidate changes in gene expression as a result of amplification or deletion of genomic material, qPCR was performed on 22/29 patients for thirteen Jumonji genes. Gene expression of JARID1A, JARID1C and UTX were modulated concomitant to the CNVs. Deletion of JARID1A locus was associated with reduced gene expression (p value <0.0001) in two patients while trisomy of JARID1C (n=1) and UTX (n=2) were associated with increased expression (p value <0.0001) of both the genes. Mutational analysis of PRC2 core components (SUZ12, EED, EZH1) and DNMT3A was carried out in a cohort of 61 MDS patients previously sequenced by our group for EZH2 mutations to examine their mutational overlap. 10/61 patients had heterozygous DNMT3A mutations (clone size 20-44 %) with two patients showing mutations at the R882 site. Interestingly, these mutations were seen predominantly (n= 6) in patients with monosomy 7/del 7q however only one patient had both DNMT3A (R882H) and EZH2 (V626M) mutations suggesting that there is no specific association between mutations of the two genes. In contrast, PRC2 genes were not mutated in this cohort emphasizing the importance of EZH2 mutations alone in MDS pathogenesis. Therefore I examined the functional consequences of the commonly occurring EZH2 (R690C/R690H) and DNMT3A (R882H) mutations in myeloid cell lines. To achieve this, numerous attempts were made to clone DNMT3A R882H mutation into p3XFLAG-myc-CMV-26 to allow transfection and in vitro assessment of the mutant in myeloid cells but all attempts to ligate the plasmid failed and therefore work on DNMT3A was discontinued. EZH2 (R690C/R69H) and Flag tagged wild type EZH2 were constructed in p3XFLAGmyc- CMV-26 vector using a PCR based cloning strategy and transfected into K562 cells. Western blot analysis at 72 hr post transfection, showed elevated levels of both R690C/R690H mutants and Flag tagged wild type EZH2 but no alterations in its target H3K27me3 levels. Affymetrix Human Transcriptome 2.0 gene expression profiling was used to identify modulation of gene signature as result of elevated EZH2 levels and MLLT10 gene was found to be up regulated in cells transfected with Flag-tagged wild type EZH2 (2.3 fold) as well as R690C/ R690H (3.6 – 4.6 fold) mutants. In contrast, PML (promyelocytic leukaemia) (2.16 fold) and FANCL (Fanconi Anaemia, Complementation Group L) (2.18 fold) genes were up regulated exclusively in cells over expressing the Flag tagged wild type EZH2. To compare this gene signature to gene expression changes as a result of EZH2 knock out (KO), shRNA mediated inhibition of EZH2 was carried out in myeloid cells and 95 % KO of both EZH2 and H3K27me3 levels were observed at Day 7 post transduction. Microarray gene expression profiling identified BCL2 (-2.14 fold), FLT1 (-4.03 fold), HOXA10 (-2.2 fold), CD44 (-8.2 fold), CD83 (-2.1 fold), TLSP (-3.24 fold), IFI16 (-3.11 fold) and PAG1 (-3.37 fold) inhibition in cells transduced with shRNA against EZH2 compared to the scrambled and wild type K562 cells. There were no overlapping genes in K562 cells with EZH2 KO and EZH2 mutants R690C/R690H. The differences in expression profiling could be due to the difference in H3K27me3 levels modulated by EZH2. Comparison of gene signature obtained by EZH2 KO on patient samples carrying the R690H mutation, showed contrasting results i.e. up regulation of HOXA10, FLT1, PAG1B, EZH1 and TLSP compared to patients with wild type EZH2 suggesting that EZH2 R690C/R690H mutants do not mimic the transcriptional changes seen in EZH2 KO. This strongly suggests the presence of other mechanisms to compensate for the loss of EZH2 in myeloid cells. However the results obtained here should be examined in additional other myeloid cell lines to validate the findings obtained in K562 cells.
Original languageEnglish
Awarding Institution
    Award date2015


    Download statistics

    No data available

    View graph of relations

    © 2020 King's College London | Strand | London WC2R 2LS | England | United Kingdom | Tel +44 (0)20 7836 5454