Скачать или смотреть TRANSFAC for identification of composite modules- combinations of transcription factor binding sites

Dr. Alexander Kel, the CSO of geneXplain GmbH, is talking about using TRANSFAC to find composite modules (CMs).
What are the composite modules? These are combinations of transcription factor binding sites (TFBS) in promoters or enhancers of genes that control the expression in specific conditions for instance in a particular tissue, cell types or under special signal transduction events, in different cell cycle phases.

There are a few good examples of composite elements documented in the TRANSCOMPEL section of TRANSFAC database.
One example shows a chicken embryonic alpha-gobin gene promoter, where there are two transcription factors, SP1 and NFY, that bind to their neighbouring sites. They interact with each other and activate the gene in primitive erythroid cells. In all other types of adult cells, another factor - NFY binds to an adjacent site and represses that gene.
Another example is in the red serum amyloid A1 gene where there are two sites for C/EBP and NFkappaB that sit together in neighbouring sites and interact with each other and activate the gene in liver cells during acute phase response, but in other cells the same region is occupied by YY1 factor, which removes the other factors from the promoter and represses the gene.

Dr. Kel also talks about the principles of combinatorial regulation of genes by combinations of synergistically acting transcription factors. He also presents the basics of the algorithm of CMA (Composite Module Analyst) which searches for cluster of TF bindings sites in the promoters of up-regulated genes. It works like the that: in a silent mode of a gene its promoter is often occupied by a repressor. But once other factors start getting active in a cell, they can come to that promoter region and remove the repressor, after what they will start binding to that promoter region in a synergistic manner. They cooperate with each other to form so called enhanseosome and activate the gene. In our CMA algorithm to search for the composite modules we take each promoter and search for TFBS. Next, we slide a normal function through the gene promoter and search for the place where most of the sites are in a cluster and then we compute a score of the promoter of that gene which is basically sum of the scores of all the sites found in that region multiplied by the weighting normal function which slides through the promoter.
At the end we calculate the score for each of the promoters of the upregulated genes and then we put these scores on the histogram – for upregulated genes and for non-changed genes. Next, our algorithm searches for such combinations of motifs that discriminate maximally the upregulated genes from the non-changed genes and at the end we build the model which includes these motifs and their combinations, as well as the distances between them and score cutoffs for each of the motifs.

We have demonstrated how the search for the composite model works using an example of publicly available data from GEO (GSE184398) on RNA sequencing of tumor infiltrating immune cells across 12 different types of tumor. And our interest is focused on so called Tregs - T regulatory cells, which are very important targets for tumor therapy and are comparing them to the regular T-cells. It is known that one particular transcription factor – FoxP3 is a muster regulator of Tregs.
We are interested to search for FoxP3 binding sites in the promoters of up-regulated genes in Tregs (compared to regular T-cells) and to find composite modules of FoxP3 with other transcription factors that are enriched in the promoters of Treg specific genes.

The RNA-seq gene count data (TPM – transcripts per million) were downloaded from the GEO entry, combined in an Excel table and uploaded to the geneXplain platform.

Dr. Kel shows how to run various pipelines and individual methods in the geneXplain platform to search for enriched binding sites and to construct composite modules. The constructed full composite model that consists of four modules is discussed. Each module consists of two motifs (PWMs - positional weight matrices) and specifies the parameters of the motifs, such as the score cut-off and number of considered top sites, and the optimal maximal distance between the sites in the pairs.

Among the found composite modules the algorithm identified pairs of FoxP3 and FoxP1 PWMs in combinations with motifs for such factors as GATA-3, Smad-3, Snail and Tbx2. Dr.Kel shows that these factor pairs are quite relevant for the considered gene expression regulation specific for Tregs and can be confirmed by multiple publications showing the crosstalk between these transcription factors on various levels.

In summary, Dr. Kel outlines the described algorithm and gives perspectives for the use of the results of this algorithm in such modern fields of gene engineering as construction of tissue or cell type specific artificial promoters.

Get your free geneXplain platform account at https://bit.ly/3Pse3eE