Final Report Summary - PROMOTER PREDICTIONS (Bioinformatic analysis of transcription regulation: a modeling approach)
The main goal of the project was to develop novel bioinformatic methods and generate fundamental understanding necessary for analyzing transcription regulation. Specifically, the project improved methods for TSS predictions, and gained quantitative understanding of the mechanisms of transcription initiation and promoter specificity. The project also modeled dynamics of gene expression regulation, in particular those involved in defense against bacterial viruses and antibiotic (microcine) production. The main results are briefly summarized below:
We first systematically determined specificities of the promoter elements outside of the canonical -35 and -10 box – in particular the so called -15 element, which was previously not included in the transcription start site searches. We established that strengths of sigma 70 promoter elements complement each other, so as to achieve a sufficient level of transcription activity; we related this result with a recently proposed `mix-and-match' model of promoter recognition. We used the new alignment to improve the information-theory method for promoter recognition, which significantly (~50%) reduces the number of false positives. We furthermore systematically investigated the importance of different sequence elements in determining the promoter specificity.
Furthermore, we investigated kinetic properties of E. coli genomic segments. While we find that RNA polymerase DNA-binding domains are designed to reduce the number of poised promoters, their number is still significant in E. coli intergenic regions, which significantly contributes to false positives in promoter searches. We also found examples of a significant underrepresentation of the regulatory elements in genomic sequences. Based on the notion of the significant binding score deviations from the random ensemble, we developed a new method for detecting direct targets (target genes) of a given regulator, which is based on the Kolmogorov-Smirnov procedure, leading to a significantly higher prediction accuracy.
We also developed a new procedure for detecting promoters transcribed by bacteriophage encoded sigma factors (including those from ECF sigma subfamily). Contrary to the current paradigm, by analyzing bacteriophage and canonical (SigmaE and SigmaW) ECF sigmas, we find quantitative and qualitative evidence of strong mix-and-matching in this subfamily.
As an example of the dynamics of gene expression, we modeled CRISPR transcript processing, and showed that this system functions as a strong linear amplifier. Based on this analysis, we proposed a synthetic gene circuit [4] that can produce a large amount of product, from small amounts of potentially toxic substrate. As another example of the dynamics of gene regulation, we modeled transcription regulation of mccA and mccB [7], whose products have a crucial role in microcine C (McC) synthesis.
Overall, the project resulted in 9 papers that are published in the leading international journals (the average impact factor of 4.1) four papers submitted for publication, and one manuscript in preparation, where the fellow is the first or the senior author on 11 of these papers. In addition, the fellow was recently promoted to an Associate Professor at the Faculty of Biology, University of Belgrade, where he also initiated his group, and is currently mentoring three PhD students (for more details see www.bio.bg.ac.rs/Marko_Djordjevic_web_site/).
We first systematically determined specificities of the promoter elements outside of the canonical -35 and -10 box – in particular the so called -15 element, which was previously not included in the transcription start site searches. We established that strengths of sigma 70 promoter elements complement each other, so as to achieve a sufficient level of transcription activity; we related this result with a recently proposed `mix-and-match' model of promoter recognition. We used the new alignment to improve the information-theory method for promoter recognition, which significantly (~50%) reduces the number of false positives. We furthermore systematically investigated the importance of different sequence elements in determining the promoter specificity.
Furthermore, we investigated kinetic properties of E. coli genomic segments. While we find that RNA polymerase DNA-binding domains are designed to reduce the number of poised promoters, their number is still significant in E. coli intergenic regions, which significantly contributes to false positives in promoter searches. We also found examples of a significant underrepresentation of the regulatory elements in genomic sequences. Based on the notion of the significant binding score deviations from the random ensemble, we developed a new method for detecting direct targets (target genes) of a given regulator, which is based on the Kolmogorov-Smirnov procedure, leading to a significantly higher prediction accuracy.
We also developed a new procedure for detecting promoters transcribed by bacteriophage encoded sigma factors (including those from ECF sigma subfamily). Contrary to the current paradigm, by analyzing bacteriophage and canonical (SigmaE and SigmaW) ECF sigmas, we find quantitative and qualitative evidence of strong mix-and-matching in this subfamily.
As an example of the dynamics of gene expression, we modeled CRISPR transcript processing, and showed that this system functions as a strong linear amplifier. Based on this analysis, we proposed a synthetic gene circuit [4] that can produce a large amount of product, from small amounts of potentially toxic substrate. As another example of the dynamics of gene regulation, we modeled transcription regulation of mccA and mccB [7], whose products have a crucial role in microcine C (McC) synthesis.
Overall, the project resulted in 9 papers that are published in the leading international journals (the average impact factor of 4.1) four papers submitted for publication, and one manuscript in preparation, where the fellow is the first or the senior author on 11 of these papers. In addition, the fellow was recently promoted to an Associate Professor at the Faculty of Biology, University of Belgrade, where he also initiated his group, and is currently mentoring three PhD students (for more details see www.bio.bg.ac.rs/Marko_Djordjevic_web_site/).