Non-homologous end joining induced alterations in DNA methylation: A source of permanent epigenetic change

Abstract

In addition to genetic mutations, epigenetic revision plays a major role in the development and progression of cancer; specifically, inappropriate DNA methylation or demethylation of CpG residues may alter the expression of genes that promote tumorigenesis. We hypothesize that DNA repair, specifically the repair of DNA double strand breaks (DSB) by Non-Homologous End Joining (NHEJ) may play a role in this process. Using a GFP reporter system inserted into the genome of HeLa cells, we are able to induce targeted DNA damage that enables the cells, after successfully undergoing NHEJ repair, to express WT GFP. These GFP+ cells were segregated into two expression classes, one with robust expression (Bright) and the other with reduced expression (Dim). Using a DNA hypomethylating drug (AzadC) we demonstrated that the different GFP expression levels was due to differential methylation statuses of CpGs in regions on either side of the break site. Deep sequencing analysis of this area in sorted Bright and Dim populations revealed a collection of different epi-alleles that display patterns of DNA methylation following repair by NHEJ. These patterns differ between Bright and Dim cells which are hypo- and hypermethylated, respectively, and between the post-repair populations and the original, uncut cells. These data suggest that NHEJ repair facilitates a rewrite of the methylation landscape in repaired genes, elucidating a potential source for the altered methylation patterns seen in cancer cells, and understanding the mechanism by which this occurs could provide new therapeutic targets for preventing this process from contributing to tumorigenesis.

Keywords: DNA damage; DNA methylation; DNA repair; NHEJ repair.

Figure 1. Doxycycline inducible construct uses GFP as a reporter for NHEJ

(A) Reporter construct integrated into the genome of the IHN20.22 HeLa cell line. The NHEJ reporter GFP gene contains a Pem1 intron interrupted by an adenoviral exon. Two I-Sce1 restriction sites allow the homing endonuclease to cut the DNA and excise the adenoviral exon to produce wild-type GFP following repair by NHEJ. (B) Generation of GFP positive cells following repair. Cells were induced with dox for 24 hours and then the percentage of the population expressing GFP was analyzed using FACS. The circles on the “+Dox” plot indicate two separate GFP positive cell populations with differing expression levels of GFP (C) Time course analysis. The percentage of GFP positive cells was analyzed by FACS over the course of 9 days, either following a 24-hour induction or with continuous exposure to doxycycline. The uncut cell line with no dox exposure was also analyzed to asses basal levels of GFP expression. (D) The onset of WT GFP expression in a single cell during the 72 hours following a 24 hour Dox induction was observed using live-cell imaging. The arrows indicate time progression and the black circle indicates the GFP negative cell that is GFP positive in subsequent images.

Figure 2. The effect of induced hypomethylation on the distribution of cells in Bright and Dim GFP expression classes

(A) Histogram of GFP+ cells after induction with dox with gating for Bright and Dim populations. (B) Hypomethylation by 5’Aza-2’-deoxycitidine. IHN20.22 cells were induced with dox for 24 hours and then treated with a daily dose of 1μM AzadC for 48 hours. The percentage of cells with high (Bright) and low (Dim) GFP expression with and without AzadC treatment were quantified using FACS histograms of GFP positive cells. The second graph is an overlay of the histogram with (red) and without (blue) treatment with AzadC. (C) Characterization of the effect of hypomethylation by AzadC on the GFP expression level in post-repair cells. The first graph compares the GFP expression level over the course of 7 days with and without AzadC treatment. Cells were induced with dox for 24 hours and then given a daily dose of 1μM AzadC. The percentage of Bright cells was measured using FACS on days 1, 2, 4 and 7 following initiation of AzadC treatments. (D) The distribution of cells in each expression class was observed in the days following NHEJ repair. As indicated by the arrow on the last graph, the cells were treated with 5μM AzadC on day 25 and then analyzed by FACS on day 26, 27, 28, and 30. (E) Comparison of the effect of AzadC at different concentrations. Cells were induced with dox for 24 hours and then given a daily dose of the indicated concentration of AzadC for 2 days. After a 48-hour recovery, the percentage of Bright cells was determined using FACS. The fold increase in the percentage of bright cells is shown.

Figure 3. (A, B) Dim (A) and Bright (B) populations were sorted using FACS and then the sorted populations were treated with a daily dose of AzadC for 48 hours

The level of GFP expression was measured using FACS. The blue line in (B) indicates the median of the Bright peak. (C) Live cell imaging was used to observe the origination and propagation of cells with Dim and Bright expression of GFP during the 72 hours following a 24 hour dox induction. The arrow indicates the progression of time

Figure 4. Bisulfite Sequencing of sorted Dim and Bright cells

Next-Generation Sequencing of bisulfite converted DNA was used to determine the methylation patterns of the area around the site of repair in sorted Dim and Bright cell populations as well as uncut IHN20.22 cells

Figure 5. Qualitative DNA methylation profiles discriminate BRIGHT molecules from DIMS and UNCUT molecules

Epialleles profiles obtained from the analysis of the methylation of each amplicons were subjected to alpha and beta diversity (Qiime). (A) Profile composition of relative sample (BRIGHT, DIMS and UNCUT) grouped by number of methylated CpGs for the five regions adjacent to the DSB. “0 (red color = un-methylated)”, “1 (blue = mono-methylated)”, “2 (orange = di-methylated)” and “3 (green = tri-methylated)” represent the percent to class of methylation. (B) Principal coordinate analysis of BRIGHT, DIMS and UNCUT. In the X and Y axes are represented, respectively, the first and the second components (PC1 and PC2) with the amount of variance in the samples explained by these components, included in brackets. The first principle component represents the highest variance, and the total variance of the samples is the cumulative sum of that described by each of the axes.

Figure 6. Regions adjacent to DSB give discriminate REC from UNCUT and BRIGHT from DIMS

Epiallelesprofiles obtained from the analysis of the methylation of each amplicons were subjected to alpha and beta diversity (Qiime). (A) Shannon diversity index between BRIGHT, DIMS and UNCUT. In the X and Y axes are represented, respectively, “number of sequences for sample” and “Rarefaction measure (species richness)” for Shannon Index. (B) Gain or loss of methylation in BRIGHT and DIMS compared to the UNCUT.