Curation, inference, and assessment of a globally reconstructed gene regulatory network for Streptomyces coelicolor

Abstract

Streptomyces coelicolor A3(2) is a model microorganism for the study of Streptomycetes, antibiotic production, and secondary metabolism in general. Even though S. coelicolor has an outstanding variety of regulators among bacteria, little effort to globally study its transcription has been made. We manually curated 29 years of literature and databases to assemble a meta-curated experimentally-validated gene regulatory network (GRN) with 5386 genes and 9707 regulatory interactions (~ 41% of the total expected interactions). This provides the most extensive and up-to-date reconstruction available for the regulatory circuitry of this organism. Only ~ 6% (534/9707) are supported by experiments confirming the binding of the transcription factor to the upstream region of the target gene, the so-called “strong” evidence. While for the remaining interactions there is no confirmation of direct binding. To tackle network incompleteness, we performed network inference using several methods (including two proposed here) for motif identification in DNA sequences and GRN inference from transcriptomics. Further, we contrasted the structural properties and functional architecture of the networks to assess the reliability of the predictions, finding the inference from DNA sequence data to be the most trustworthy approach. Finally, we show two applications of the inferred and the curated networks. The inference allowed us to propose novel transcription factors for the key Streptomyces antibiotic regulatory proteins (SARPs). The curated network allowed us to study the conservation of the system-level components between S. coelicolor and Corynebacterium glutamicum. There we identified the basal machinery as the common signature between the two organisms. The curated networks were deposited in Abasy Atlas (https://abasy.ccg.unam.mx/) while the inferences are available as Supplementary Material.

System Principles Governing the Organization, Architecture, Dynamics, and Evolution of Gene Regulatory Networks

“…However, it is a still-going challenge that, on one hand, has been approached through a plethora of transcriptomics-based strategies ranging from mechanistic models to machine learning, all of them with modest to poor results ( Marbach et al, 2012 ). Network inference based on the identification of regulatory binding sites has performed significantly better ( Zorro-Aranda et al, 2022 ), but it requires a prior network for its application. One way to deal with this limitation is to transfer regulatory information from one organism to another ( Alkema et al, 2004 ).…”

“…Inferences based on gene expression data have also benefited from the integration of biological information. For instance, the pre-selection of transcription factors (TFs) from experimental data constrains the number of potential inferences, and the application of structural properties of biological GRNs improves the assessment of the predictions ( Zorro-Aranda et al, 2022 ). Other works have also shown improvements in the inference of regulatory networks integrating multiple omics data ( Cheng et al, 2011 ; Banf and Rhee, 2017 ) and network structure ( Castro et al, 2019 ).…”

“…Once we know the inferred networks behave as the biological ones, we can study their functional architecture and system-level components ( Freyre-Gonzalez et al, 2012 ). Although the diamond-like structure has been found across all the organisms in Abasy Atlas, the system-level conservation has been quantitatively evaluated only between E. coli and B. subtilis ( Freyre-Gonzalez et al, 2012 ), and Corynebacterium glutamicum and Streptomyces coelicolor ( Zorro-Aranda et al, 2022 ). Future work assessing the conservation across all the available organisms and the robustness of the node classification to network incompleteness would shed light on the missing interactions for incomplete networks and their hierarchical role in the global network.…”

Improving gene regulatory network inference and assessment: The importance of using network structure

“…2) In contrast, the COEX tools infer interactions between genes with correlated expression profiles: LSTrAP (Proost et al, 2017), WGCNA (Zhang and Horvath, 2005), and RNAseqNet (Proost et al, 2017) based on correlation; ARACNE (Margolin et al, 2006), C3NET (Altay and Emmert-Streib, 2010), CLR (Faith et al, 2007); and MRNET (Meyer et al, 2007) based on mutual information. 3) The HYBR (hybrid) group contemplates ANOVA (Kuffner et al, 2012) and Friedman (Zorro-Aranda et al, 2022) which are based on analysis of variance and therefore do not infer causality. However, we used a list of TFs to keep only TF-TG interactions.…”

Rhizobium etli CFN42 proteomes showed isoenzymes in free-living and symbiosis with a different transcriptional regulation inferred from a transcriptional regulatory network

“…We wondered how our inferred networks assess against known curated networks. As no curated network is available for R. etli , inspired by recent work showing that assessing using network structural properties provides results consistent with using a gold-standard ( Zorro-Aranda et al 2022 ), we performed a pairwise comparison via correlation of the normalized structural profiles of two well-curated regulatory networks, E. coli and B. subtilis , as positive control and a background of Erdös-Rényi parametrized random networks as a negative control ( Figure 6A ; “ Materials and Methods ” section).…”