Back

Stochastic motility-adhesion switch within the E. coli K-12 strains revealed by gene expression clustering

Kaznadzey, A. D.; Bessonova, T. A.; Kuznetsova, U. D.; Gelfand, M. S.; Tutukina, M. N.

2025-12-15 microbiology
10.64898/2025.12.14.694247 bioRxiv
Show abstract

Biofilm formation in Escherichia coli arises from regulatory programs that can diverge even among genetically identical cultures. Exploiting spontaneous replicate-to-replicate variation in E. coli K-12 MG1655, we mapped expression modules whose behavior tracks biofilm abundance across four genetic backgrounds: wild type; strain with deleted gene for hexuronate regulator UxuR {Delta}uxuR; an uxuR translation-disrupted retaining locus-derived sRNAs ({Delta}uxuR-tr), and strain with deleted gene for global carbon regulator cAMP-CRP ({Delta}crp). In the wild type, regulator EcpR/MatA demonstrated the highest correlation with the biofilm intensity, and its role was, in particular, seemingly reinforced by stress/pH-homeostasis and c-di-GMP-linked factors. {Delta}uxuR exhibited the strongest stochasticity in biofilm formation, while in {Delta}uxuR-tr, replicate dispersion was mostly muted, indicating that uxuR-derived sRNAs buffer motility-adhesion switching. In {Delta}crp, pgaA was de-repressed due to the absence of CsrA, with expression being highly consistent with intensity of biofilm formation, alongside stress/ion-homeostasis genes. This is in line with the suggestion that CRP regulates biofilms and motility indirectly via regulation of more local transcription factors. Overall, replicate-resolved clustering converts apparent "noise" into structured regulatory patterns, which include genes whose role in biofilm formation was not yet evident, and helps identify switch nodes between planktonic and biofilm bacterial states.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.