Distribution of ZDRs throughout human chromosome 14: black lines show ZDR clusters around the TSS in contrast to random quasi-uniform distibution depicted in light grey. 

Context in source publication

Context 1

... with PolII and therefore change the occupation rates of the enzyme. This analysis was carried out using public data retrieved from the NCBI Short Reads Archive (SRA) [11], particularly from Joseph et al. [9]. Reads derived from Illumina platform were divided in two groups: (i) cells treated with estradiol and (ii) untreated cells (control). Both groups were then aligned to the human chromosome 14 by the Bowtie aligner [10] with default parameters, except for --best which reports only the best alignments (those with fewer mismatches). The aligned reads were filtered out by a process called “peak calling” which identifies genome enriched areas where reads cluster together [14]. This process was performed in R by the BayesPeak package [4] and the differential occupation fold-change analysis was calculated by the DESeq package. [2] To further investigate the distribution of ZDRs in relation to genomic structure features, we searched for ZDRs within exons, introns, UTRs and intergenic regions. Although this analysis used the same ENCODE annotation file, it differed by being processed through the GenomicFeatures [5] suite so that the genes’ annotation would be subdivided into their features. Then, those features were matched against the ZDRs to return an overall percentage of ZDRs relative to each genomic feature. As stated above, the first step of our analysis was to correlate ZDRs found by Z-Catcher with the gene annotation from ENCODE . Figure 3 shows that our approach was able to reproduce literature findings by showing an overall clus- tering of ZDRs around TSSs [22], as opposed to randomized distances to each TSS. Once the ZDRs are not equally spread over the genome, this distribution suggests that the ZDRs may play a role in [17] or be dependent upon transcription events [12]. To further address this issue, we looked deeper into the distribution and plotted the exact location of ZDRs relative to their nearest TSS and its respective transcript. With this analysis, we wanted to investigate if there was any bias towards specific ZDR hotspots around or within transcripts. Indeed it is possible to observe in Figure 4 that ZDRs seem to be more concen- trated upstream of TSSs, which would corroborate the hypothesis on the Z-DNA relationship with transcription events. To date, no Z-DNA mapping approach has focused on the ZDR distribution throughout the genome in relation to its genomic features. Taking that into account, it is important to further investigate this correlation, since it may re- veal some unknown distribution pattern and may also help to elucidate Z-DNA function. Some works had suggested that the presence of ZDRs within introns would enable and guide the coupling of proteins from the ADAR1 family, which are responsible for mRNA editing [1]. These proteins are not only known to be present in Z-DNA binding sites with high affinity but also to be responsible for the deamination of adenosines to inosines (which are translated as guanines). These editing events act as a source of phenotypic variation [7] and could play an important role at modulation of the nervous system [20]. If it is found that this interaction is dependent on Z-DNA formation, an important function for ZDRs would be revealed. ZDRs found inside transcripts lie almost exclusively within introns, which account for roughly 18% of the total number of detected ZDRs (Figure 4). Con- sidering that genes are composed mostly by intronic sequences, this percentage may not represent a strict preference of the ZDRs’ distribution. Anyhow, in Table 1 it is possible to see five of the transcripts and their associated genes, which exhibit the largest number of ZDRs within introns. Such genes may be good candidates for further investigation of ADAR1 family mechanism of action, which would contribute to understand the potential role of Z-DNA guiding RNA editing enzymes. Even though we were able to show the correlation between ZDRs and TSSs, it still remained unclear if it represents only a by-product of gene transcription events or if ZDRs indeed act as gene expression regulators. To address this point, we analysed whether ZDR presence could modify RNA Polymerase II occupation of transcription start sites. Our dataset of PolII was taken from a ChIP-Seq experiment which analysed the occupation of the promoter region of the ER- α estrogen receptor of MCF-7 cells in two conditions: activated (with presence of its ligand, estradiol) and inac- tivated (controlled set). With this analysis, we were able to investigate whether the differential enrichment of the PolII tags in specific locations may correlate to the presence of ZDRs. We divided our fold-change results in two groups, one with transcripts with up-regulated occupation in the activated condition (fold-change ≥ 2) and the other with transcripts with down-regulated occupation (fold-change < 0.5). Next, we ranked the 100 topmost differentially occupied transcripts of each group and analysed their ZDRs related position. Figure 5 shows that the number of ZDRs overlapped by PolII tags at the 100 topmost differentially occupied transcripts exhibits an interesting pattern. Those ZDRs present in up-regulated genes tend to gather upstream of TSSs while those ZDRs present in down-regulated genes exhibit a weak tendency to gather downstream. This may suggest a positive correlation between ZDRs position and the activation state of genes, consider- ing that the formation of ZDRs upstream could favour gene transcription, while ZDRs downstream could inhibit it. Nevertheless, our data need additional experimental evidence and further statistical analysis to confirm these statements. In this work we developed a workflow for analysis of ZDR regions in animal cells that merges in silico data with experimental ones. The workflow uses several bioconductor packages and retrieves biological data from high throughput sequencing. Hence, one could easily correlate data from ChIP-seq and RNA-seq to ZDR regions in whole chromosomes. Our case study focused on the human chromosome 14, and the results showed that our workflow approach was able to conduct ZDR distribution analysis that corroborates previous studies. It brought as well new information on how those ZDRs spread over the chromosomal sequences. The role of Z-DNA in gene regulation has been debated for a long time. In our case study we showed that the majority of ZDRs appear upstream of the transcripts. We also showed that when accounted for internal genomic features, ZDRs tend to concentrate in introns rather than exons. Although this was expected, it showed that our approach is able to succesfully detect ZDRs’ distribution within transcripts. Hence, one could investigate in other human chromosomes or another species genome the hypothesis of ZDRs serving as anchor sites for Z-DNA binding factors such as ADAR1, which is responsible for RNA edition. The comparison of ZDRs prediction to PolII occupancy in steroid regulated genes suggests differences in ZDRs positioning in relation to TSSs. Up regulated genes seem to concentrate ZDRs upstream of TSSs as opposed to down regulated genes that tends to concentrate ZDRs downstream. However, further experimental studies and statistical investigation are still necessary to convinc- ingly correlate ZDRs to gene expression. We are presently working to assemble the R scripts developed for this workflow in a user friendly R package, where the user will be able to perform similar analysis as those previously shown. Our goal is to deliver an easy and fast way to perform basic distribution analysis associated to biological information in different kinds of genomes, allowing for an efficient computing platform for the Z-DNA biology ...