Deep mutational scanning of an S. pneumoniae FMN riboswitch reveals robustness during mouse infection but diverging adaptive landscapes in response to targeting antibiotics

RNA-based control of gene expression is common across all domains of life, yet shows extraordinary variety in terms of environmental cues recognized, mechanisms of action, required protein factors, and magnitude of gene expression changes. The fitness benefit conferred by RNA-mediated regulation is often subtle and challenging to measure and may be highly influenced by the environment. Yet selective forces are sufficient to generate and maintain complex RNA structures that allow rapid and robust response to stimuli. In this work we perform deep mutational scanning to measure the relative fitness conferred by all single mutants of an FMN riboswitch to the opportunistic pathogen Streptococcus pneumoniae under seven different conditions. We find that in standard culture conditions the RNA sequence is not under selective pressure, however the dynamic conditions encountered during mouse lung infection reveal an sequence that is under weak purifying selection, but robust to strongly deleterious mutations. Furthermore, assessment of fitness in the presence of two different targeting antibiotics exposes diverging adaptive landscapes that reveal suites of mutations conferring either constitutive activation or repression of gene expression. Analysis of these landscapes highlights a complex relationship between antibiotic concentration and the fitness benefit conferred by individual mutations as well as illuminates the mechanism of action for this RNA with lessons broadly applicable to interpretation of bacterial transcriptomics.