Annotation of the results obtained in 2020
In 2020, as part of the project, we completed a pipeline for analyzing the role of long non-coding RNAs in the chromatin regulation. The main improvement made this year is the independent analysis of various ASOs, which allows for more reliable identification of ncRNAs that are involved in epigenetic regulation. Using this pipeline we analyzed ten epigenetic modifications and regions of variable DNA methylation. Based on these results we developed HiMoRNA database (himorna.fbras.ru) containing 5640250 significantly correlated lncRNA-genome regions pairs for 10 histone modifications and 4145 ncRNAs. To characterize each lncRNA-genomic locus pair, we also collected a series of metadata for each interaction. Our pipeline revealed 48 lncRNAs potentially involved in the regulation of at least one epigenetic modification associated with functional changes in gene expression. More than half of these RNAs showed consistent results in biological replicas and were also validated using RNA-Chromatin Interaction Data (iMARGI). We found an extremely non-uniform chromosomal distribution of loci in which epigenetic modification was correlated with lncRNA expression. In particular, our pipeline shows a strong association between FTX and DNA methylation and between JPX and H3K9me3. Both DNA methylation and H3K9me3 are repressive modifiers and are known to contribute to different stages of X inactivation. We have shown that almost all of the JPX-associated H3K9me3 peaks are located on the X chromosome, indicating that JPX can also contribute not only to the regulation of XIST per se, but also directly to the repression of the X chromosome. Investigating the mechanisms of interaction of lncRNAs with target regions, we also showed that about 30% of all knockdown lncRNAs can regulate target genes through the formation of ncRNA: RNA duplexes. Most of these lncRNAs are likely to function in the nucleus, where they can interact with target transcripts during transcription (cotranscriptionally). This observation was also confirmed by experimental data on lncRNA-chromatin interactions. LncRNA CHASERR is most likely to form lncRNA: RNA duplexes with regulated genes co-transcriptionally. We have proposed a model suggesting that a helicase CHD2, in interaction with CHASERR, is guided to the target promoters and activates genes. Conservative regions of this lncRNA were identified, which are most likely involved in the function of CHASERR. We also identified 14 lncRNAs (CTD-3131K8.2, CTD-2587H24.5, AC124789.1, TUG1, RAB30-AS1, RP13-463N16.6, LINC00886, LINC00886, ZNF37BP, FTX, EMX2OS, FGD5-AS1, RP11- 417E7.1, AC016747.3), which can form RNA: DNA triplexes with promoters of regulated genes, some of which are also involved in epigenetic regulation. In summary, we identified 17 lncRNAs potentially involved in the regulation of at least one epigenetic modification. Among these lncRNAs we found as those described in the literature as other lncRNAs, which role has not been shown earlier. Each lncRNA was knocked down again with at least two antisense oligonucleotides (ASOs) for targeted validation. In addition to the previously planned work, we have shown the role of lncRNA ecCEBPA in the regulation of DNA methylation in acute myeloid leukemia. We have proposed a model for the regulation of DNA methylation through the formation of triple helices by ecCEBPA with the promoters of regulated genes. It has also been shown that NR2F1-AS1 can function as a sponge for microRNAs targeting DEGs that respond to knockout of this lncRNA. We have also shown that chimeric transcripts, which play an important role in carcinogenesis, form a long hairpin, possibly playing a similar role as small interfering RNAs (siRNAs) and down regulating gene expression. Our study emphasizes the functional importance of lncRNAs regardless of their level of expression, localization, and conservation, including their specific role in the epigenetic regulation of gene expression. Molecular phenotyping combined with the integration of multi-OMIX data allows high-quality functional annotation of lncRNAs, including detecting their role in the onset and progression of socially significant diseases, in particular in cancerogenesis. Our results reveal new targets for drug therapy of these diseases.

 

Publications

---

Annotation of the results obtained in 2018
In 2018, we developed statistical methods for the analysis of the role of long non-coding RNAs in the establishing and maintenance of the cell chromatin profile. We also implemented these methods into analysis pipelines. First of all, we developed a probabilistic model of the interaction of RNA and DNA with the formation of triplexes according to the Hoogsteen rules. As a basic algorithm for predicting the interaction of RNA and DNA with the formation of triplexes, we used Triplexator. In our current work, this algorithm has been substantially improved. A probabilistic model of the distribution of numerical values (scores), determined by the original algorithm, has been developed. This model allows one to take into account the lengths of interacting RNA and DNA regions, their nucleotide composition, and the specifics triplexes determines by Triplexator. The resulting model allows for genome- and transcriptome-wide comparisons, which in turn allows the selection of candidate RNAs, most likely forming a real triplex. We have previously shown that, although Triplexator outperform other software on the experimental data, it still produces a lot of false-positive predictions. In order to compensate for this, in addition to the work planned for this year, we performed tests on how the in corporation of the RNA secondary structure might affect triplex prediction. In more details, RNA regions that are not paired potentially are more likely to bind DNA. Predictions were validated by experimental data on genome-wide binding of lncRNA MEG3. We have shown that the use of only unpaired RNA segments obtained by RNAplfold (Vienna) does indeed increase the specificity of triplex prediction, without reducing sensitivity. Further, we developed a beta-version of the software for lncRNA search that could potentially affect the chromatin state by forming triplexes with the target regions. The software pipeline consists of the following main blocks: (1) a standardized search for histone modification sites in dozens of samples; (2) normalization of lncRNA expression data in the same samples; (3) correlation of the numerical value of the histone modification peak and expression of lncRNA; (4) prediction of the triplex formation between candidate DNA-RNA pairs and evaluation of its statistical significance. The beta-version of software was tested on the H3K27me3 histone mark, since it is known to be established partially by lncRNA MEG3, which also can binds DNA forming a triplex. In addition, we analyzed experimental data on the interaction of RNA with chromatin obtained in a co-laboration, as well as public data from GRID-seq. We showed that although the absolute number of interchromosomal contacts (RNA is bound to the DNA of an another chromosome) is small, the relative frequencies of the triplexes in such contacts are quite high. We also found several RNAs that form triplexes with a large number of regions of the genome, which has not been reported for these RNAs before. Also, we performed a pilot analysis of the data obtained by FANTOM6 consortium. This data reflects changes of expression in response to various RNA knockdown. We analyzed the promoters of genes that significantly differentially expressed in response to knockdown, and identified a number of RNAs, forming triplexes with the promoters that are differentially expressed in knockdown. These data indicate a possible mechanism for the regulation of gene expression using this RNA, yet, further experimental validation in needed.

 

Publications

---

Annotation of the results obtained in 2019
In 2019, we finalized the development of statistical methods for analyzing the role of long non-coding RNAs in the formation of chromatin profile, as well as the software pipeline incorporating these methods. The pipeline was improved and refined: sampling was refined, most of the calculations were vectorized to increase the processing speed, methods for statistical processing of knockdown data were incorporated, RNA and DNA colocalization data was included, and DNA methylation data was also processed. Using this pipeline, 10 epigenetic modifications and variable DNA methylation regions were analyzed. The results of the correlation analysis between lncRNA expression and the presence of histone modification at a specific locus were processed more deeply: correlated histone ChIP-seq peaks were associated with their regulated genes. In turn, the obtained lists of genes correlated with specific lncRNAs were split by the sign of correlation. We suggest that the correlation sign may indicate the functional role of the ncRNA-modification-gene correlation: a positive correlation means that lncRNA can be involved in establishing of a histone modification; a negative correlation means that ncRNA can be involved in its removal. Next, we incorporated the functional analysis using lncRNA knockdown data obtained by FANTOM6 consortium. For each knockdown lncRNA, the lists of differentially expressed genes were split into those that increase and decrease their expression. For the lncRNAs present in FANTOM6 knockdowns, the significance of overlapping the list of genes having modification peaks, correlating with lncRNAs of a certain sign with the list of genes from knockdown experiments that change the expression in a certain direction was predicted. Thus, 4 groups of ncRNAs were distinguished according to their functional role: lncRNAs that put repressive marks, lncRNAs that remove repressive labels, lncRNAs that put activator marks, and ncRNAs that remove activator marks. An analysis of the colocalization of lncRNA with correlated ChIP-seq histone peaks was also performed using iMARGi data. To evaluate the significance of colocalization of genomic markups, the GenometricCorr package was used. As a result, for each modification, a collection of lncRNAs were obtained that were potentially involved in targeting histone modifiers to specific genomic loci and were classified into one of 4 groups according to the possible type of influence on the histone modification. A list of lncRNAs that potentially affect methylation at specific locations in the genome has also been obtained. In total, more than 50 lncRNAs were identified to be potentially involved in changes in histone modifications and DNA methylation. The obtained lncRNAs were tested for the possibility of the formation of lncRNA-RNA duplexes and lncRNA-DNA triplexes using the ASSA, Triplexator, and TDF programs. Among the lncRNAs found by our pipeline, there were those previously reported in the literature as well as lncRNAs, functional role of which has not been shown earlier. The functional role of one lncRNA was experimentally validated.

 

Publications

---