Vicisitudes de un galán de teatro (1928-1936)

We also performed the same set of experiments on bisulphite-modified DNA derived from IV.10 (yr.3 and 4) that has been immunoprecipitated with H3K27me3 repressive chromatin mark. Based on the proportion of methylated CpGs shown in Figure 4.17, DNA methylation appears to coincide with H3K27me3 promoter deposition in all sampling time-points of IV.10 suggesting that the two events are not incompatible. This observation is consistent with the fact that gene promoters that are enriched with H3K27me3 are more likely to be methylated during differentiation and carcinogenesis and this can be attributed to the recruitment of PRC1 and PRC2 complexes that methylate H3K27 and silence CpG island associated genes (Statham et al., 2012, Rose and Klose, 2014). Although the association between PRC2 and Tet1 could explain why a subset of CpG islands that are occupied by PRC2 are not subject to DNA methylation (Neri et al., 2013).

4.8 Discussion

The intra- and inter-individual variations in GATA2 expression profiles seen within and between p.T354M GATA2-mutated family members described in Chapter 3 fostered our investigation focusing on the molecular mechanisms underlying monoallelic GATA2 expression. We hypothesised that these changes in allelic expression could be mediated by a combination of regulatory and transient epigenetic mechanisms that include changes in DNA methylation and chromatin mark deposition.

The reduction in GATA2 expression due to allele-specific fluctuations in expression implied that putative cis-acting regulatory mechanisms might be involved. Indeed, the identification of a heterozygous GATA2 promoter 2 SNP residing within the 5’UTR in our symptomatic patient vs. asymptomatic carriers constituted a marker to distinguish between mutant and WT alleles and a first step in establishing the mechanisms governing the observed ASE. Intriguingly, this SNP can also alter (generate/remove) CpG methylation sites within GATA2 promoter region, resembling a previous study showing that a CpG-SNP (rs12041331) reinforces PEAR1 enhancer activity in platelet formation through allele-specific DNA methylation (Izzi et al., 2016). Here, we also showed allele-specific DNA methylation as a regulatory mechanism contributing to the silencing of the WT GATA2 allele in our symptomatic patient’s earlier disease time-points. One would postulate promoter methylation to have an impact on TF binding affinity; increased promoter methylation would lead to lower TF binding affinity thereby inhibiting patterns of gene Figure 4.17 Linking DNA methylation and H3K27me3 promoter deposition. Bisulphite-specific PCR followed by cloning and Sanger sequencing covering GATA2 second promoter SNP [C/A] region overlapping a CpG island was performed to assess DNA methylation patterns of H3K27me3 ChIP-enriched DNA across 2 time-points of our symptomatic patient IV.10 (yr.3 with monoallelic GATA2 expression) and (yr.4 with biallelic GATA2 expression). Each row represents a separate clone. Black circles denote methylated CpGs while white circles denote unmethylated CpGs.

We did not validate allele-specific TF binding occupancy based on the PROMO in silico prediction tool in our patient samples. However, our findings are in agreement with a previous study by Celton and colleagues who attributed reduced GATA2 expression levels in normal karyotype sporadic AMLs (NK-AMLs) to promoter hypermethylation and SNPs acting as loss-of-function mutations, highlighting the importance of epigenetic alterations in modulating gene expression (Celton et al., 2014).

Another notable observation was made following allele-specific ChIP revealing an enrichment of H3K4me3 on the GATA2 promoter mutant allele compared to the WT allele at diagnosis which was reversed at later follow-up, correlating with reactivation of the WT allele expression. This interpretation is in line with a study by Stern and colleagues who showed mutant TERT promoter allele to exhibit H3K4me3 in various cancer cell lines whilst the WT allele retained the H3K27me3 mark of gene silencing (Stern et al., 2015). We also validated that H3K4me3 blocks de novo DNA methylation by showing that DNA methylation and H3K4me3 promoter deposition are mutually exclusive in our patient samples. From a translational perspective, this altered GATA2 allelic expression can be reversed therapeutically by exposure to specific epigenetic inhibitors and/or demethylating agents (e.g. 5-azacytidine). Unfortunately, treating our symptomatic patient cells with KDM5 inhibitor, a drug that inhibits the KDM5 family of histone demethylases, stabilising H3K4me3 levels and could have the potential therefore of reactivating the expression of the silenced WT GATA2 allele, has proven to be challenging due to the short life of our cells in culture and the scarcity of material available.

Moreover, with the high frequency of promotor 2 SNP (MAF 39%), there would be a reasonable chance that this SNP could be utilised similarly in other GATA2-mutated families. Indeed, we should not rule out the possibility that this promoter 2 SNP (rs1806462 [C/A]), by creating an

extra CpG methylation site within the WT allele, could play a direct role in its silencing. To test the contribution of this SNP, we analysed the haplotype of 12 p.Thr354Met GATA2-mutated individuals (10 affected members and 2 asymptomatic carriers) from the three families published by Hahn et al. (2011) and we observed two individuals (one symptomatic and one asymptomatic carrier) from two different families heterozygous [C/A] for the SNP (rs1806462). In both cases, the SNP reference allele [C] was in cis with the GATA2 mutation, the opposite of what we detected in our family. Moreover, apart from our symptomatic patient (IV.10), no other family members (including the two deceased MDS/AML cousins IV.1 and IV.6) were heterozygous for these GATA2 promoter SNPs and given the stable improvement in clinical parameters at IV.10 later time-points (yr.4 and 6), we reasoned that while this CpG-SNP plays a role in GATA2 monoallelic expression, there is no evidence to establish a correlation between the haplotype rs1806462A_-GATA2T354M _{and the progression of disease/symptomatic status. This SNP was} therefore used merely as a vehicle to distinguish between mutant and WT alleles in our DNA methylation and ChIP experiments.

Collectively, our findings from this chapter propose that allele-specific expression of GATA2 mutant allele is driven by dynamic epigenetic reprogramming; increased DNA methylation linked with lower H3K4me3 promoter deposition on the WT allele and vice versa for the mutant allele (Figure 4.18), adding another layer of complexity to the (epi)genetic basis of familial MDS/AML and contributing towards the observed reduced penetrance phenotype seen in certain inherited

Figure 4.18 Epigenetic dysregulation accounting for the mono-/biallelic GATA2 expression status observed across the 4 time-points studied in our symptomatic patient (IV.10) and its correlation with disease symptoms, using promoter 2 SNP as a means of distinguishing between mutant and WT alleles.