An experimentally evolved variant of RsmA confirms its central role in the control of Pseudomonas aeruginosa social motility

Bacteria can colonize a variety of different environments by modulating their gene regulation using two-component systems. The versatile opportunistic pathogen Pseudomonas aeruginosa has been studied for its capacity to adapt to a broad range of environmental conditions. The Gac/Rsm pathway is composed of the sensor kinase GacS, that detects environmental cues, and the response regulator GacA, that modulates the expression of a specific genes. This system, through the sRNA repressors RsmY and RsmZ, negatively controls the activity of the protein RsmA, which is centrally involved in the transition from chronic to acute infections by post-transcriptionally regulating several virulence functions. RsmA positively regulates swarming motility, a social surface behaviour. Through a poorly defined mechanism, RsmA is also indirectly regulated by HptB, and a ΔhptB mutant exhibits a severe swarming defect. Since a ΔhptB mutant retains all the known functions required for that type of motility, we used an experimental evolution approach to identify elements responsible for its swarming defect. After a few passages under swarming conditions, the defect of the ΔhptB mutant was rescued by the emergence of spontaneous single nucleotide substitutions in the gacA and rsmA genes. Since GacA indirectly represses RsmA activity, it was coherent that an inactivating mutation in gacA would compensate the ΔhptB swarming defect. However, the effect of the mutation in rsmA was unexpected since RsmA promotes swarming; indeed, using expression reporters, we found that the mutation that does not abolish its activity. Instead, using electrophoretic mobility shift assays and molecular simulations, we show that this variant of RsmA is actually less amenable to titration by its cognate repressor RsmY, supporting the other phenotypes observed for this mutant. These results confirm the central role of RsmA as a regulator of swarming motility in P. aeruginosa and identify residues crucial for RsmA function in social motility. Author summary Bacteria need to readily adapt to their environment. Two-component systems (TCS) allow such adaption by triggering bacterial regulation changes through the detection of environmental cues. The opportunistic pathogen Pseudomonas aeruginosa possesses more than 60 TCS in its genome. The Gac/Rsm is a TCS extensively studied for its implication in virulence regulation. This system regulates the transition between chronic and acute bacterial infection behaviours. To acquire a better understanding of this regulation, we performed a directed experimental evolution on a swarming-deficient mutant in a poorly understood regulatory component of the Gac/Rsm pathway. We observed single nucleotide substitutions that allowed restoration of a swarming phenotype similar to the wild-type behaviour. More specifically, mutations were found in the gacA and rsmA genes. Interestingly, the observed mutation in rsmA does not result in loss of function of the protein but rather alters its susceptibility to repression by its cognate interfering sRNA. Since modification in the RNA sequence of RsmA results in the rescue of swarming motility, we confirm the central role of this posttranscriptional repressor in this social lifestyle.


57
The opportunistic pathogen Pseudomonas aeruginosa possesses more than 60 TCS in its 58 genome. The Gac/Rsm is a TCS extensively studied for its implication in virulence regulation.

59
This system regulates the transition between chronic and acute bacterial infection behaviours . To 60 acquire a better understanding of this regulation, we performed a directed experimental evolution 61 on a swarming-deficient mutant in a poorly understood regulatory component of the Gac/Rsm 62 pathway. We observed single nucleotide substitutions that allowed restoration of a swarming 63 phenotype similar to the wild-type behaviour. More specifically, mutations were found in the gacA 64 and rsmA genes. Interestingly, the observed mutation in rsmA does not result in loss of function Introduction 71 Bacteria can adapt to diverse environments using various mechanisms. They use two-component 72 systems (TCS) to rapidly modulate the expression of specific subsets of genes (1). TCS convert 73 external stimuli into an internal response that promotes adaptation to environmental cues. Some 74 bacteria exploit these systems for virulence regulation (2). Typically, TCS consists of a histidine 75 sensor kinase that responds to an external signal to trigger the autophosphorylation of an 76 intracellular histidine residue. Then, the phosphoryl group of the sensor kinase is transferred to 77 an aspartate residue located in the receiver domain of a cognate response regulator, which then 78 modulates the expression of a specific set of target genes (3). In some cases, phosphorylation of 79 the receiver domain can occur through a His-containing phosphotransfer (Hpt) protein that acts 80 as an intermediate between the membrane sensor and the response regulator (4).

82
Pseudomonas aeruginosa is an opportunistic pathogen responsible for several nosocomial 83 infections and also a major cause of morbidity and mortality among individuals with cystic fibrosis 84 (5, 6). The genome of prototypical P. aeruginosa strain PAO1 contains 63 histidine kinases, 64 85 response regulators, and three Hpt proteins (7). The Gac/Rsm pathway regulates, among others, 86 virulence-associated genes, and biofilm formation (8). This pathway regulates the transition 87 between chronic (associated with the sessile lifestyle) and acute (associated with the motile 88 lifestyle) infections (9, 10). The Gac TCS is composed of the histidine sensor kinase GacS (11) 89 and its cognate response regulator GacA. When GacS is phosphorylated, it transfers its

147
148 Given the emergence of non-synonymous mutations in the gacA and rsmA genes (clones C4 and 149 C2, respectively) after repeated passages of the hptB isogenic strain to select for a recovered 150 swarming phenotype, we verified whether the acquired mutations were responsible for the 151 rescue. We first looked at the activity of GacA in the C4 clone. As shown in figure 1B, 152 inactivation of gacA in a hptB strain restores the swarming phenotype. Accordingly, swarming of 153 clone C4 is similar to that of gacA and hptBgacA mutants. Furthermore, we verified whether 7 155 rsmY and rsmZ ( Fig. 2 and Fig. S1). In agreement with a loss of GacA function, the transcription 156 of both sRNAs is significantly lower in the evolved clone when compared to its parental hptB 157 background. Clone C4 shows no significant difference in expression of rsmY and rsmZ with the 158 gacA and hptBgacA mutants. Thus, these results demonstrate that clone C4 acquired an 159 inactivating mutation in the GacA regulator, allowing for the rescue of the swarming defect in the 160 hptB parental strain. 165 1C). However, swarming was rescued in clone C2 (Fig. 1C). This was surprising given that the 166 over-expression of RsmA in the hptB background also results in a rescue of the swarming 167 phenotype (Fig. S2). Thus, we considered two possible explanations for these results: 175 We observed that a plasmid-borne rsmA R31S can restore expression of rsmY in rsmA, although 176 incompletely ( Fig. 3 and Fig. S3). Thus, RsmA R31S is functional, but with a somewhat altered 177 activity, indicating that the mutation in clone C2 does not result in abrogation of the activity of the 178 protein. We then looked at the translation of hcnA mRNA transcripts, known to be directly 179 repressed by RsmA (25, 26). Interestingly, we observed that the translation of hcnA is lower in 180 the C2 clone compared to the hptBrsmA mutant, but similar to the wild-type strain ( Fig. 4A and   181 Fig.S4A). Thus, to further understand the impact of the observed non-synonymous mutation in 9 212 a RsmY sRNA molecule (29, 30). The RsmY sRNA has seven GGA sites where RsmA can bind. 213 The second, fifth and seventh binding sites are the most determinant for RsmA-RsmY complex 214 formation (29). At tested RsmA concentrations where protein:RNA interactions are detected, both 215 wild-type and RsmA R31S at the concentration of 0.05 µM formed complex 1 (Fig. 6B ). However, 216 at higher concentrations, RsmA R31S can only form complex 2. In contrast, the wild-type protein 217 was able to shift RsmY, forming complex 3, which was not observed for RsmA R31S at the same 218 concentration (Fig. 6A and Fig. S5). 223 concentration) to determine the ratio of RNA for each section (Fig. S6). In section A, complex 3 224 was found and could be quantified (Fig. S6A) while the first and second complexes were 225 quantified in section B (Fig. S6B). Unbound RNAs were quantified in section C (Fig. S6C). These 226 data demonstrate that RsmA R31S does not bind as well as wild-type RsmA to RsmY, which 227 validates our in silico model and explains the results we observed for our in vivo assays.  (Fig. S7). Residue R44 is involved in RNA interaction and R31 277 most likely plays an important accessory role impacting binding affinity and/or ligand 278 discrimination since it appears to stabilize the U 88 A 89 nucleotide pair located downstream of the 279 conserved GGA motif in RsmY (Fig. 5). This positively charged residue is solvent-exposed and 280 its interaction with RNA is mediated by two hydrogen bonds involving its terminal guanidinium 281 moiety (40, 41). In contrast, the mutation in C2 introduces a serine instead of an arginine, leading 282 to a small and polar residue unable to maintain these interactions, therefore affecting RNA 283 binding affinity and/or discrimination. 284 285 We confirmed that RsmY affinity for RsmA with the R31S substitution is modified due to different 286 RsmY mobility shift when interacting with either RsmA or RsmA R31S in EMSA experiments (Fig.   287 6). The most probable interpretation is that the loss of interaction between the R31 residue and 288 RNA reduces affinity for some binding sites, notably with GGAUA, such as in RsmY, as opposed 289 to other sites. The complexes between sRNA and RsmA are the result of the multiple RsmA 290 molecule binding to the different GGA motifs of RsmY (29). Even though both RsmA and 291 RsmA R31S are capable of interacting with RsmY and forming Complex 1, only the wild-type protein 292 can form Complex 3 which most likely represents a higher capacity to bind RsmY molecules 293 compared to RsmA R31S . Indeed, this was confirmed when we looked at the radioactivity signal of 294 the protein-RNA complexes; clearly, RsmA R31S does not bind as well as WT RsmA to RsmY (Fig.   295 S6A). Affinity between RsmA R31S and RsmY shows that the inhibition by RsmY is not as efficient, 297 4B), but different than rsmA and hptBrsmA.

299
Our results indicate that RsmA inhibits, directly or indirectly, an unknown repressor impacting 300 swarming motility. However, this repressing factor is still unknown, and it is thus not yet possible 301 to test its activity with RsmA and RsmA R31S in vitro. RsmY has many RsmA binding sites and 302 displays secondary structures with multiple stem-loops with RsmA binding sites (29). Depending 303 on the different mRNAs that are controlled by RsmA, the number of available binding sites and/or 304 secondary structures could probably affect the binding capacity between the protein and target 305 RNAs. The RsmA:mRNA interaction of these RsmA-controlled mRNAs implicated in swarming 306 motility could be less impacted by the mutation R31S than RsmY, due to their sequence and 307 structure, and then explain the rescue of swarming in the hptB background. Also, additional 308 elements such as chaperone Hfq, which can bind some mRNA transcripts that associate with 309 RsmA, could contribute to RNA binding and were missing in our in vitro experiments (42). 310 However, our data strongly support the fact that a modified RsmA binds RsmY less efficiently, 311 impacting its inhibitory effect, which further explains the swarming rescue of the C2 clone.