EPCO-19. ADAPTIVE RESPONSES TO GENOME-WIDE DNA DAMAGE RESULT IN TOPOLOGIC GENOME REORGANIZATION IN GLIOBLASTOMA

In glioblastoma, treatment with radiation and chemotherapy leads to DNA-damage and most DNA breaks are faithfully repaired, but the impact on the epigenome is largely unknown. Using newly developed tools to enable these studies, we hypothesize that genome-wide DNA damage leads to local alterations in DNA-methylation, genome organization, and results in persistent gene-expression alterations near sites of repaired damage. We use patient-derived human glioblastoma stem-like cells (GSCs) as a model. DNA breaks are induced using (i) irradiation or (ii) a novel “multi-cut” CRISPR-Cas9 DNA break system followed by multi-omic profiling. With radiation, we find significant and wide-spread alterations in DNA-methylation after treating multiple glioblastoma cultures. However, it is challenging to study local alterations around sites of radiation induced damage because breaks are introduced at different sites in each cell, resulting in stochastic DNA methylation alterations. To circumvent this issue, we developed a multi-cut CRISPR-Cas9 DNA break system that targets 142 or 483 pre-defined loci. Induction of pre-mapped genome-wide cuts reproduces a similar level of toxicity as standard doses of radiation. To assess repair efficiency and confirm induction of breaks, we performed targeted sequencing of the 142 or 483 sites to allow for high coverage sequencing. To understand how DNA damage may lead to regional epigenetic and 3D chromatin organization changes, we performed HiC, Methylation-seq, ChIP-seq of the chromatin organizing factor CTCF and enhancer marker H3K27ac, as well as RNA-seq, before and after cut induction. Our findings show significant mega-base scale alterations in chromatin contacts centered around cut sites, enrichment of DNA methylation alterations at regulatory elements and altered gene-expression. The findings here provide a mechanistic view of the interplay between genome-wide DNA damage, DNA methylation and genome re-organization, and have wide ranging implications for the effect of DNA damage on the epigenome in glioblastoma.

EPCO-04. RADIOTHERAPY IS ASSOCIATED WITH GLOBAL METHYLATION ALTERATIONS IN PATIENT DERIVED GLIOBLASTOMA CELL LINES

PURPOSE/OBJECTIVE(S)

In glioblastoma, DNA methylation states are the most predictive marker of overall survival and response to therapy. Our understanding of how epigenetic states, such as DNA methylation, are “mis-repaired” after DNA damage repair is scant, hampering our ability to understand how treatment associated DNA methylation alterations may drive tumor resistance and growth.

MATERIALS AND METHODS

Three different patient derived IDH wild-type glioma stem cell (GSC) lines, in duplicates, were treated with radiation (20 Gray in 10 fractions vs. sham control) and allowed to recover prior to DNA methylation analysis with 850K methylation arrays. To analyze the methylation array data via bioinformatic methods we used RnBeads (version 2.4.0) and R (version 3.6.1) packages. We further focused our analysis to specific genomic regions, including CpG islands, promoters, gene bodies and CTCF motifs to understand how methylation alterations may differ between these and other genomic contexts following radiation.

RESULTS

There were widespread differential methylation (pre-treatment vs. radiation treatment) changes among the genomic regions examined. Interestingly, we found differential methylation changes at CTCF motifs, which play important DNA-methylation dependent roles in gene expression and chromatin architecture regulation. Hierarchical clustering, PCA and MDS analysis of DNA methylation status amongst CpG islands, promoters, gene bodies and CTCF domains revealed strong intra-sample differences, but not inter-sample differences (between GSC lines), suggesting radiation associated methylation alterations maybe loci and context dependent.

CONCLUSION

Radiation treatment is associated with wide-spread alterations of DNA methylation states in this patient derived glioblastoma model. Such alterations may drive gene expression changes or genomic architecture alterations that lead to treatment resistance, warranting further mechanistic investigation of the interplay between radiation induced DNA damage and local epigenetic state restoration following DNA damage repair.

CBIO-05. SOX2 CHROMATIN ARCHITECTURE AND ENHANCER ACTIVITY IN THE MAINTENANCE AND DEVELOPMENT OF NEURAL STEM CELLS

Gain of function mutations in isocitrate dehydrogenase I (IDH1) result in the formation of the oncometabolite 2-hydroxyglutarate (2HG) in adult lower grade gliomas. To gain insight into mechanisms of gliomagenesis, our lab previously created a tractable human cellular model of low grade astrocytoma (LGA) using the putative cell-of-origin, human neural stem cells (NSCs), engineered to express mutant IDH1 and knockdown constructs against TP53 and ATRX, the two other genetic changes that accompany the IDH mutation in these tumors. We found that transcription factor (sex determining region Y)-box 2 SOX2, which is essential to NSC multipotency, the ability to differentiate to neuroglial lineages, behaves as a tumor suppressor during glioma initiation. In this context, we showed SOX2 is transcriptionally downregulated to impair NSC multipotency, thus locking NSCs in an undifferentiated state to initiate gliomagenesis. This downregulation occurs secondary to dynamic reorganization of the topologically associating domain (TAD) of SOX2 and the loss of contact with several genomic loci with histone modifications and chromatin accessibility suggestive of being enhancers. Here we show that those putative enhancers acquire enhancer-like features simultaneous to tje TAD organizing in a way that facilitates interaction with the SOX2 promoter during the process of pluripotent stem cell differentiation into neuroectodermal lineages, suggesting a developmental role. Preliminary data suggests that disruption of the SOX2 TAD by preventing binding of the genome organizer CTCF downregulates SOX2 expression in NSCs. Targeted silencing of several regions of a putative enhancer with CRISPRi also downregulates SOX2. In human embryonic stem cells (hESCs), interfering with these CTCF binding sites biases their differentiation away from the neuroectoderm. We are currently performing CRISPRi screen against all putative enhancer loci, teratoma formation assays on hESCs lacking relevant CTCF binding, and CRISPR mediated deletion of putative enhancers. Understanding this developmental process may reveal underlying vulnerabilities in LGA.

STEM-17. LOW GRADE ASTROCYTOMA MUTATIONS COOPERATE TO DISRUPT SOX2 GENOMIC ARCHITECTURE AND BLOCK DIFFERENTIATION VIA PREVIOUSLY UNIDENTIFIED ENHANCER ELEMENTS

Neomorphic mutations in isocitrate dehydrogenase I (IDH1) result in the formation of the oncometabolite 2-hydroxyglutarate (2HG) in a significant subset of gliomas and other tumors including acute myeloid leukemias. Preclinical evidence suggests that gliomas harboring IDH1 mutations undergo widespread, long-lasting modifications to the epigenome that persist following inhibition of 2HG production. However, the exact mechanism underlying gliomagenesis remains unclear. To address difficulties in growing these tumors in culture, our group generated a model of low-grade astrocytoma in human neural stem cells (NSCs). This model, referred to as 3-Hit NSCs, suggested a block in differentiation potential underlies gliomagenesis at the cellular level. This block is rescued by restoration of expression of the transcription factor (sex determining region Y)-box 2 (SOX2), which is transcriptionally downregulated during IDH mutant gliomagenesis. We believe that these changes occur secondary to profound alterations in 3-dimensional chromatin organization around the SOX2 genomic locus. Our preliminary data suggest SOX2 expression in control NSCs depends on 3-dimensional association of its promoter to an uncharacterized, distal enhancer located 600 kb telomeric to the SOX2 gene. We believe this association is disrupted in 3-Hit NSCs due to eviction of chromatin organizer CTCF from its motifs in the SOX2 topologically associating domain (TAD). To test this hypothesis, we used CRISPR-Cas9 technology to excise CTCF motifs immediately upstream of the SOX2 promoter and in the region of the putative enhancer in control NSCs. Excision of such motifs significantly reduced SOX2 mRNA levels and impaired growth of control NSCs. We are currently working on characterizing this novel SOX2-enhancer interaction in native stem cells, as well as tumors that depend on SOX2 expression. This works aims to elucidate the core epigenetic mechanisms underlying IDH mutant gliomagenesis. Our findings will be used to improve therapy in IDH-mutant glioma.