Diversity and evolution of the transcriptional regulatory networks of Pseudomonas strains revealed using machine learning

Abstract

The genus Pseudomonas consists of diverse and ecologically significant species that form close associations with both plants and animals. This genus is widely studied due to the clinically relevant Pseudomonas aeruginosa, model plant pathogen Pseudomonas syringae, and non-pathogenic, industrially relevant Pseudomonas putida. The different metabolic and physiological capabilities of these species are enabled by their unique genetic makeup as well as varying regulatory mechanisms. To study the transcriptional basis for the diversity of the three species, we applied independent component analysis to strain-specific RNA-seq datasets to identify independently modulated gene sets (iModulons) and their condition-specific activity levels. We then mapped iModulons across strains based on their similarity in orthologous gene membership. Through comparison of iModulon gene membership and activities, we find that: (i) iModulons reveal shared and unique regulatory modalities across strains; (ii) unique adaptations in common functions, such as translation and pyoverdine production/uptake, manifest through both differential iModulon gene membership and condition-specific activation states in each strain; (iii) iModulons facilitate comparison of stress responses at the systems level; and (iv) iModulons highlight unique virulence factor enrichment and host-specific adaptations in human and plant pathogens. Altogether, comparing the modularized transcriptomes of the three strains provides unique and comprehensive insights into their differential evolution. ImportanceClosely related bacterial species often have vastly different metabolic and physiological capabilities, yet the regulatory mechanisms underlying these adaptations remain poorly understood. Here, we compare the transcriptional regulatory networks of three representative Pseudomonas strains through cross-strain iModulon analysis. By comparing both iModulon gene composition and activity across strains, we identify conserved regulatory modules alongside lineage-specific adaptations in functions associated with virulence, translation, iron acquisition, motility, and stress responses. Our results demonstrate that iModulons provide a genome-scale framework for comparing transcriptional regulation across closely related organisms, revealing regulatory innovations that are not apparent from genome comparisons alone. This work establishes a scalable approach for studying the evolution of bacterial transcriptional regulatory networks and the regulatory basis of niche specialization.