Global reprogramming of virulence and antibiotic resistance in Pseudomonas aeruginosa by a single nucleotide polymorphism in elongation factor, fusA1

Clinical isolates of the opportunistic pathogen Pseudomonas aeruginosa from patients with cystic fibrosis (CF) frequently contain mutations in the gene encoding an elongation factor, FusA1. Recent work has shown that fusA1 mutants often display elevated aminoglycoside resistance due to increased expression of the efflux pump, MexXY. However, we wondered whether these mutants might also be affected in other virulence-associ-ated phenotypes. Here, we isolated a spontaneous gentamicin-resistant fusA1 mutant (FusA1 P443L ) in which mexXY expression was increased. Proteomic and transcriptomic analyses revealed that the fusA1 mutant also exhibited discrete changes in the expression of key pathogenicity-associated genes. Most notably, the fusA1 mutant displayed greatly increased expression of the Type III secretion system (T3SS), widely considered to be the most potent virulence factor in the P. aeruginosa arsenal, and also elevated expression of the Type VI (T6) secretion machinery. This was unexpected because expression of the T3SS is usually reciprocally coordinated with T6 secretion system expression. The fusA1 mutant also displayed elevated exopolysaccharide production, dysregulated siderophore production, elevated ribosome synthesis, and transcriptomic signatures indicative of translational stress. Each of these phenotypes (and almost all of the transcriptomic and proteomic changes associated

Due to its high intrinsic resistance to antibiotics and aggressive virulence, Pseudomonas aeruginosa holds the dubious accolade of consistently occupying a "top ten" slot on lists of clinical threats across the globe. Indeed, the World Health Organization recently classified it as a top priority pathogen for which the development of new antimicrobial interventions is critical. This opportunistic, Gram-negative bacterium is ubiquitous and exhibits a particular predilection for the built environment, making encounters with the human populace commonplace. P. aeruginosa is frequently isolated from burn wounds, the respiratory tract, and the urinary tract and is the leading cause of morbidity and mortality in people with cystic fibrosis (CF) (1)(2)(3). CF is a genetic disease characterized by defective targeting or activity of the cystic fibrosis transmembrane conductance regulator. Although it is a multisystem disease affecting many organs, the most obvious manifestations of CF are associated with the respiratory tract. Here, defective mucociliary clearance causes an accumulation of thick mucus plugs within the airways. Such oxygen-limited environments provide the perfect niche for P. aeruginosa to thrive (4,5).
Chronic P. aeruginosa infection of the CF lung is associated with the transition from an active, motile lifestyle to a sessile, biofilm-like mode of growth. These are bacterial communities embedded within a self-produced extracellular polymeric matrix, composed of mannose-rich polysaccharides, extracellular DNA, and proteins (6). This matrix confers a level of protection against antibiotics and the host immune system (7). Biofilm formation is also associated with increased expression of the Type VI (T6) secretion machinery. The function of the P. aeruginosa T6 secretion system has become clearer in the last decade; it appears to play a role in killing other bacterial species (or even "non-self" P. aeruginosa strains) (8,9), especially in tightly packed biofilms where competition for the same resources is rife. Conversely, "free-swimming" planktonic cells predominate in acute infection scenarios. Here, virulence factors and motility are up-regulated, and pathogenicity is enhanced (10). Perhaps the most potent P. aeruginosa virulence determinant is the Type III (T3) secretion system, which mediates the translocation of cytotoxic effector proteins directly into the cytoplasm of neighboring host cells. These effectors subvert the function of the recipient cells, typically by disrupting the cytoskeleton, promoting cell rounding and apoptosis, and therefore assist immune evasion through preventing phagocytosis by host innate immune cells (11,12). Both the T3 and T6 secretion systems are contact-triggered injectisomes; however, they are structurally and mechanistically distinct, as are their targets, and the expression of these two systems appears to be inversely correlated (13).
In previous work, we showed that the rapidly growing planktonic cells associated with increased virulence factor production display markedly up-regulated expression of the machinery required for macromolecular synthesis, especially proteins involved in translation (14). Translation comprises four main steps; initiation, elongation, termination, and recycling. The ribosome-associated protein, elongation factor G (EF-G, encoded by fusA1 and fusA2 in P. aeruginosa), is essential for two of these steps: "elongation" and "recycling" (15). During elongation, EF-G catalyzes the translocation of charged tRNA from the A-site to the P-site and from the P-site to the E-site of the large ribosomal subunit. This is coupled with movement of the ribosome along the mRNA being translated. EF-G is comprised of five domains; domains I and II mediate GTP binding and hydrolysis, and domains III, IV, and V dock to the A-site of the ribosome. The tip of domain IV interacts with the mRNA and promotes tRNA translocation (16). This involves large structural rearrangements as the EF-G domains swivel relative to one another (17,18). The translocation process is repeated until a stop codon is encountered and release factors catalyze hydrolysis of the peptidyl-tRNA bond, thereby liberating the newly synthesized polypeptide. EF-G then coordinates with ribosome-recycling factor (a structural mimic of tRNA) to promote disassembly of the ribosomal subunits (16,19). Aminoglycoside antibiotics can disrupt both the elongation and recycling steps, thereby leading to a decrease in the overall number of ribosomes available.
Whole-genome sequence analyses have revealed that fusA1 is a hotspot for accruing mutations in P. aeruginosa isolates from patients with CF (20). These fusA1 mutants often display increased resistance to aminoglycoside antibiotics. However, little more is known about the phenotypic consequences of such fusA1 mutations on the wider physiology of the cell. In this study, we show that a spontaneous SNP in P. aeruginosa fusA1 gives rise to discrete but large-magnitude changes in key pathophysiological processes. For example, the strain containing the mutated fusA1 allele displayed selective up-regulation of genes encoding the T3 secretion system (T3SS) apparatus, the T6 secretion system (T6SS) apparatus, exopolysaccharide biosynthesis genes, and a multidrug efflux system. These findings suggest a hitherto unexpected subtlety in the chain of events linking transcription, translation, and virulence/antibiotic sensitivity in this organism.

Results
An SNP in fusA1 causes decreased expression of the biofilmassociated protein, CdrA At the outset of this investigation, we sought to identify potential regulators of biofilm formation in P. aeruginosa. To do this, we made a stable chromosomal reporter construct in which the promoter of the cdrAB operon (PcdrAB) was fused to a promoter-less lacZ ORF. CdrA is a biofilm-associated extracellular matrix adhesion and is known to be primarily expressed in conditions that favor biofilm formation. The construct was integrated at a neutral site in the PAO1 chromosome using the mini-CTX system (21). The resulting PcdrAB::lacZ reporter strain, hereafter referred to as EMC0, yielded vivid blue colonies when grown on M9 minimal medium agar plates containing X-Gal and glucose, but far paler colonies on medium containing X-Gal and glycerol. This suggested that PcdrAB is activated during colony growth on glucose. To identify genes that might impinge upon transcription from the cdrAB promoter, we mutagenized EMC0 by introducing the pTnMod-OGm plasposon (22). The resulting mutants were selected on plates containing gentamicin (to select for likely Tn insertion mutants) and X-Gal 1 glucose (to establish whether any of the mutants were affected in transcription from PcdrAB). Around 10,000 gentamicin-resistant mutants were screened in all. Several of these yielded a paler pigmentation than EMC0 when grown on X-Gal/glucose plates, indicating a reduction in transcription from PcdrAB (Fig. 1A). Further analysis of one of these isolates (hereafter denoted EMC1) in M9-glucose liquid cultures confirmed the diminished b-gal production. This also indicated that EMC1 had a minor growth defect in this medium (Fig. 1A). We therefore measured growth and b-gal production in a rich medium, LB. To our surprise, EMC1 exhibited an even greater growth defect in this medium (Fig. 1B).
Attempts to identify the insertion site of the plasposon in EMC1 using conventional approaches (including "randomprimed" PCR-based amplification of the regions flanking the plasposon, or "cloning out" of the plasposon as described previously (22)) failed. We therefore used whole-genome sequencing (WGS) of EMC1 (and, as a control, also of the EMC0 progenitor strain) to identify the plasposon insertion site. Remarkably, and despite the robust Gm R phenotype of the strain (Table 1), we found that EMC1 did not contain a plasposon insertion. It did, however, contain a CfiT transition at position 4,770,363 in the genome (Fig. S1). This SNP was located in the fusA1 ORF and resulted in a proline-to-leucine substitution at position 443 in the protein. The SNP was confirmed by PCR amplification of the gene followed by Sanger sequencing of the PCR product. WGS also revealed a selection of additional potential SNPs in EMC1, but these were all subsequently found by PCR/Sanger sequencing to be false positive SNP calls arising from the WGS technology. We conclude that the only difference between EMC0 and EMC1 is the SNP in fusA1.
The gentamicin resistance of EMC1 was heritable but could be somewhat unstable. In one experiment, progressive subculturing (via 1:100 dilution of the culture into fresh medium every 24 h) of EMC1 in M9-glucose minimal medium lacking gentamicin gave rise to gentamicin-sensitive (Gm S ) derivatives. After each 24-h round of subculturing, aliquots were removed, serially diluted, and plated onto nonselective M9glucose agar. A selection of 40 colonies from these plates were then retested for gentamicin resistance by plating onto M9glucose 1 Gm. Following the first round of subculturing (24-h growth), none of the 40 tested colonies were Gm S . However, after the second round of subculturing (a cumulative 48 h of growth) 1 of 40 colonies tested was gentamicin-sensitive, rising to 5 of 40 colonies after the third 24-h round of subculturing. PCR amplification and sequencing of the fusA1 gene in these Gm S derivatives revealed that they all still contained the fusA1 P443L mutation. These data suggest that whatever gives rise to the Gm R phenotype in EMC1 can be overridden by secondary mutations elsewhere. However, these bypass mutations do not arise with high frequency, and even when they did, they took time to spread through the culture (with only 1 in 8 of the isolates sampled being Gm S after 72 h of growth). Indeed, subsequent independent repeats of the experiment yielded no further Gm S isolates upon subculturing EMC1 in the absence of gentamicin.
The proline fi leucine substitution at position 443 in FusA1 affects protein stability and conformation FusA1 encodes one of the two paralogous EF-G proteins in P. aeruginosa and plays a pivotal role in protein synthesis and ribosomal recycling (23). The amino acid sequence of the two paralogues (denoted fusA1 (PA4266) and fusA2 (PA2071)) is highly conserved, with a shared identity of 84%. EF-G1B, encoded by fusA2, is thought to have greater involvement in elongation and polypeptide synthesis (23). By contrast, EF-G1A (encoded by fusA1) has a more dominant role in ribosomal recycling and association with ribosomal recycling factors (23). FusA1 is a conformationally flexible multidomain protein (24), and proline 443 sits in close proximity to the crucial GTPase "switch" regions ( Fig. 2A). The switch regions play an important role in directing GDP-GTP exchange, raising the question of whether the P443L substitution might affect the conformation of the protein. To test this, we measured the intrinsic Trp fluorescence profile of purified WT FusA1 and FusA1 P443L . FusA1 contains 6 tryptophan residues. Protein Trp fluorescence is exquisitely sensitive to the microenvironment of each Trp residue and, as such, can be a sensitive reporter of protein conformation. These analyses indicated that purified FusA1 P443L had a lower quantum yield at the Trp emission l max (332 nm) compared with the WT protein (Fig. 2B). This indicates that one or more Trp residues in the mutant protein are likely to exhibit altered solvent accessibility compared with the WT protein, possibly due to conformational differences. One of the more widely used web-based algorithms, mCSM (25), predicted that the P443L substitution should destabilize FusA1 by 0.267 kcal/ mol. Consistent with this, purified FusA1 P443L had a lower melting temperature than the WT protein (Fig. 2C). Taken together, these data indicate that the P443L substitution likely alters the conformation, stability, or dynamics of FusA1.
The Pro fi Leu substitution in FusA1 has phenotypic consequences The substantial growth defect associated with EMC1 grown in LB was complementable by expression of the WT fusA1 gene in trans on a plasmid (Fig. S2), suggesting that the WT allele is dominant. The complemented EMC1 also displayed the same MIC for gentamicin as EMC0. Compared with EMC0, EMC1 also exhibited defects in twitching motility and swimming motility, and these too could be complemented by introduction of WT fusA1 in trans (Fig. 3A). EMC1 also displayed lower secreted gelatinase activity, and poor growth on the gelatinase plates. However, the lower secreted gelatinase production and slower growth of EMC1 may be linked because complementation of EMC1 in trans with WT fusA1 restored both phenotypes. Expression of fusA1 P443L in trans in EMC0 or EMC1 had no apparent effect on the motility/gelatinase phenotypes of the strains compared with the empty vector control. This suggests that the WT fusA1 allele in EMC0 is dominant.
By contrast with the impaired motility and secreted protease phenotypes, exopolysaccharide production was increased in  EMC1 compared with EMC0, both on plate assays (Fig. 3A) and in liquid culture (Fig. S3). Expression of WT fusA1 in trans in EMC1 led to a decrease in exopolysaccharide production (compared with EMC1 containing the empty vector control), whereas expression of fusA1 P443L enhanced exopolysaccharide synthesis. Exopolysaccharides comprise the extracellular matrix, which "glues together" cells in a biofilm. Interestingly in this regard, the increased exopolysaccharide production in EMC1 was not accompanied by an increase in its biofilm-form-ing ability compared with EMC0 (data not shown). Exopolysaccharide production is often inversely correlated with expression of the T3 secretion machinery. We therefore examined whether the P443L substitution in FusA1 impacted T3 secretion. To our surprise, cultures of EMC1 overexpressed the T3SS protein, PcrV (Fig. 3B). This increased expression was due to increased transcription of the pcrV-encoding operon, because RT-PCR analyses indicated that the amount of mRNA encoding pcrV was also increased in EMC1 (Fig. 3B and suggests that one or more Trp residues resides in an altered microenvironment in FusA1 P443L . C, thermal shift data indicate that the WT protein (top) has a higher melting temperature than the FusA1 P443L protein (bottom). This suggests that the P443L substitution decreases the thermal stability of the protein.
Each melting curve was measured in triplicate.  These observations led us to test whether previously reported Gm R fusA1 point mutants also displayed increased PcrV expression. A selection of Gm R fusA1 mutants (all in a PAO1 genetic background) recently reported by Bolard et al. (20) were grown (alongside the progenitor PAO1 strain from that study) in M9-glucose, and the corresponding cell extracts were probed after SDS-PAGE resolution with anti-PcrV antibodies. None of these mutants displayed the robust PcrV expression associated with EMC1 (Fig. S5). This suggests that the Gm R phenotype conferred by these other mutated fusA1 proteins is not linked with PcrV expression.
Global consequences of the fusA1 P443L mutation on the proteome Given that multiple phenotypes were affected by the fusA1 P443L mutation in EMC1 and that these phenotypes were not all modulated in the manner expected from previous studies (e.g. exopolysaccharide production and T3 secretion both being up-regulated instead of inversely regulated), this suggested that the P443L mutation may lead to global dysregula-tion in EMC1. To investigate this further, cultures of EMC0, EMC1, and EMC1 complemented with WT fusA1 expressed from pUCP20 in trans (hereafter EMC1*) were grown to late exponential phase in M9 minimal medium 1 glucose and were prepared for iTRAQ-based proteomic analysis. To establish whether any of the observed changes in protein profile were also underpinned by transcriptional changes, samples were also harvested for RNA sequencing (from cultures grown in the same conditions).
The proteomic analysis resolved 3506 proteins (of a total of 5570 predicted ORFs encoded by P. aeruginosa PAO1). Principal components analysis of the data revealed that the proteome of EMC1 was distinct from that of EMC0 and that the proteomic changes giving rise to this segregation could be largely reversed by expression of WT fusA1 in trans in EMC1* (Fig. S6 (A-C) and Tables S1 and S2). Proteins were considered to be significantly modulated if they exhibited a log 2 -fold change (FC) . 1 (or , 21, if down-regulated) with a false discovery rate (FDR)-adjusted p value of 0.01. Based on these criteria, 128 proteins were up-regulated in EMC1 compared with EMC0, and 166 proteins were down-regulated. The 20 most highly up-regulated proteins are shown in Table 2. Remarkably, Global impact of fusA1 mutation and consistent with the earlier phenotypic analyses, over half (12 of 20) of these proteins were involved in T3 secretion. Similarly, and also consistent with our earlier observations, PelA, involved in the biosynthesis of exopolysaccharide, was also upregulated. A list of the top 20 down-regulated proteins is shown in Table 2. The situation here is more ambiguous, because the majority (13 of 20) of these proteins currently have no assigned function.
To obtain a more functional overview of the data, the 128 statistically significantly up-regulated proteins in EMC1 were analyzed using STRING to identify clusters of associated proteins. STRING is a database of known and predicted physical and functional protein-protein interactions (26). Inspection of the STRING output revealed that the modulated proteins fell into several distinct functional clusters involved in a variety of pathophysiological processes. The most obvious cluster comprised proteins associated with the T3SS (Fig. 4). The likely driver behind this was the ;2-fold up-regulation of ExsA, which is the master regulator (activator) of T3SS expression (27). Also consistent with the phenotypic data, a cluster of proteins (PelA, PelF, and PslG) involved in exopolysaccharide bio-synthesis were up-regulated in EMC1. The data also revealed a probable explanation for the enhanced gentamicin resistance of EMC1; ArmZ, a major activator of the mexXY aminoglycoside efflux pump, was up-regulated, as was expression of MexXY (28,29). Unexpectedly, we also noted that a selection of proteins involved in biosynthesis of the siderophore, pyochelin, were significantly up-regulated. This was somewhat surprising because (i) proteins associated with the biosynthesis of the other major siderophore in P. aeruginosa, pyoverdine, were unaffected; (ii) the abundance of iron-uptake master regulators, Fur and PvdS, as well as the pyochelin-specific regulator PchR were unchanged in EMC1 compared with EMC0; and (iii) phenotypic analyses revealed that cultures of EMC1 produce less secreted siderophore(s) than EMC0 or EMC1* (Fig. S7). It therefore seems that in EMC1, the biosynthetic pathway for pyochelin is expressed, but the siderophore is not secreted. This dysregulation of iron homeostasis in EMC1 may have additional consequences. A link between iron availability and T3 secretion has been documented in several bacterial genera, including Bordetella, Salmonella, Shigella, Edwardsiella, Vibrio, and Yersinia (30-35). To investigate further whether iron availability impacts upon expression of the T3SS in EMC1, we examined whether supplementation of the growth medium with additional iron had any effect on PcrV expression. The additional iron had little effect on expression of PcrV in EMC0 or EMC1*. However, the addition of excess iron to the EMC1 cultures led to increased PcrV expression (Fig. S7). It is therefore possible that the dysregulation of iron homeostasis in EMC1 is another factor that impacts expression of the T3SS.
A number of bacteriophage-like R2-type pyocin gene products were also up-regulated. R-type pyocins cause membrane depolarization and inhibit active transport in closely related species to reduce bacterial competition (36). Another obvious functionally related grouping was comprised of proteins associated with sulfur metabolism (including CysA, CysD, SsuD, TauA, and TauB). However, perhaps the most notable cluster of up-regulated metabolic proteins were those involved in polyamine catabolism (PauA4, PauA5, PauB1, PauB3, and PauB4). The genes encoding these proteins are strongly induced by the polyamines, putrescine and spermidine, and are distributed across the chromosome (37). These gene products encode probable g-glutamylpolyamine synthases involved in converting putrescine into g-aminobutyric acid, and their collective up-regulation indicates that putrescine and/or spermidine may be more abundant in cultures of EMC1. Consistent with this notion, spermidine is known to also increase the expression of the T3SS proteins (38).
We also carried out a STRING-based analysis of the proteins that were down-regulated in EMC1. Fewer clusters were apparent compared with the up-regulated STRING map. This likely reflects the large number of uncharacterized proteins in the data set; with little information to functionally link these proteins, clusters will inevitably be more sparsely populated. However, some patterns were apparent. From the STRING map (Fig. S8), a small cluster comprising ExaB, nitrite reductase (NirS), and azurin (Azu) is discernable. ExaB is of interest because it was the protein with the greatest FC (Table 2). ExaB is a cytochrome c 550 involved in the breakdown of ethanol to aldehyde (39). Azurin transfers electrons from another c-type cytochrome, cytochrome c 551 , to cytochrome oxidase. Cytochrome c 551 is also the electron donor of nitrite reductase (NirS) and plays a role in dissimilative denitrification (40). These findings suggest that elements of the bacterial electron transport chain may be affected in EMC1. The ExaB/NirS/Azu cluster was linked to an adjacent group of proteins with roles in disulfide bond formation (DsbA, DsbD2, a probable GSH peroxidase (PA1287), TrxA, and three other Trx-like proteins (PA0941, PA3664, and PA0950)) and protein folding (PpiA and PpiC2). These proteins are closely linked to the electron transfer chain and cellular redox status.
Several other highly down-regulated proteins were linked to the T6SS. PA5086 and PA5088 are predicted to encode T6SS lipase immunity proteins, protecting the producing cell from lipase-inflicted self-harm (41). PA5088 is operonic with an encoded T6SS phospholipase D effector (tleB5) and with vgrG5, although neither of these proteins were detected in our experiments. VgrG proteins are secreted by the T6SS to form complexes that perforate host cell membranes, and the genomic position of vgrG5 adjacent to PA5088 makes it likely that its expression was also reduced. The down-regulation of T6SS-associated proteins observed here is consistent with the known inverse regulation of T6 and T3 secretion (13). Another discernible cluster of proteins was related to cell motility and chemotaxis, including CheR1, PA0177, PctC, PA1464, and the twitching motility protein, PilH. The flagellar motor protein, FliY, and an uncharacterized protein, PA2781, were also downregulated. PA2781 is likely operonic with PA2780 (bswR), a bacterial swarming regulator (undetected in the proteomic analysis) and is predicted to interact with FlgB, a structural component of the bacterial flagellum. This suggests that PA2781 may also affect flagella-mediated motility. Collectively, these observations are consistent with the decreased twitching and swimming motility associated with EMC1 (Fig. 3).
Global consequences of the fusA1 P443L mutation on the transcriptome Given the role of FusA1 in translation and given the apparently very selective consequences of the fusA1 P443L mutation on the proteome, we wondered whether these changes might also be reflected at a transcriptomic level. To address this possibility further, we carried out an RNA-Seq analysis of the mRNA expression profiles in EMC1 and EMC1* compared with EMC0. Of the 5678 predicted RNA-encoding genes in P. aeruginosa PAO1, 5628 were represented in the RNA-Seq data set. This suggests that most ORFs are expressed at detectable levels in the conditions tested. Of these expressed genes, 657 were significantly (p , 0.01) up-regulated (FC . 1), whereas 374 were significantly down-regulated (FC . 21) when comparing EMC0 with EMC1. Far fewer transcripts were significantly modulated when comparing EMC0 with EMC1* (Fig. S9), indicating that expression of the WT fusA1 gene was able to complement many of the changes in EMC1. This was also reflected by the lower level of scatter away from the midline in the FPKM (fragments per kilobase per million) plots (Fig. S9).
The top 20 most highly up-regulated transcripts are shown in Table 3. Most (16 of 20) of these are involved in T3 secretion. This indicates that the T3SS is up-regulated (by up to 46-fold) at a transcriptional level as well as at a translational level in EMC1. These changes were reversed back toward the levels seen in EMC0 by the introduction of the WT fusA1 gene in trans (Tables S3 and S4).
To establish potential functional interactions between the modulated transcripts, we carried out a STRING interactome analysis. Due to the large number of significantly modulated transcripts, only the top 200 were used to predict interactions between the encoded proteins. T3SS genes were by far the largest cluster on the STRING protein interaction map (Fig. 5). Consistent with the proteomic analysis, exsA expression was up-regulated by around 3-fold and is most likely to be a major factor driving the observed global up-regulation of the T3SS transcripts. It is worth noting that the transcriptional regulator PsrA is thought to be required for full activation of the exsCEBA operon (42). However, psrA transcripts exhibited a 2fold reduction in expression, suggesting that PsrA may not be a major requirement for T3 secretion in all conditions and that Global impact of fusA1 mutation low levels of psrA transcription do not necessarily prevent expression of the exsCEBA operon. Other distinct clusters predicted by STRING included a group of ribosome-associated proteins and, surprisingly, also a robust cluster of T6SS-associated transcripts (mostly from Hcp secretion island I (HSI-I)). Indeed, the transcript encoding the HSI-I needle protein, Hcp1, was up-regulated 13-fold, and many other T6-associated transcripts were up-regulated .4-fold. This was unexpected because the T3SS and T6SS are reciprocally regulated in most conditions, and several T6-associated proteins were downregulated in the proteomic analyses. Given that we previously noticed that Gm S mutants can potentially arise in the EMC1 population (albeit at low frequency, and very slow to spread through the population), one formal possibility that might have accounted for this observation is that growth in the absence of Gm gives rise to a mixed population of Gm S mutants (expressing the T6SS) alongside the Gm R progenitor, EMC1 (expressing the T3SS). To test this, we struck the Gm R EMC1 from frozen stock on a concentration of Gm (5 mg ml 21 ) that would prevent the growth of any spontaneous Gm S mutants. In parallel, the Gm S progenitor of EMC1 (EMC0) and, as a control, also the WT strain (PAO1) were struck out from frozen stocks in the absence of Gm. A single colony from each plate was then used to inoculate M9-glucose liquid medium containing either no Gm (PAO1, EMC0, and EMC1) or 1 mg ml 21 Gm (0.5 3 MIC Gm ) in the case of EMC0 or 5 mg ml 21 Gm (0.5 3 MIC Gm ) in the case of EMC1. In this setup, the Gm R EMC1 starter culture should contain no Gm S cells, and if these arise at all, they should only be present in the cultures grown without Gm. We then blotted the cell extracts from each culture for the HSI-I needle protein (Hcp1) and also for a T3S system protein (PcrV). As shown in Fig. S10, PcrV expression was basal in the PAO1 and EMC0 cultures, irrespective of the presence of sub-MIC Gm . However, PcrV expression was robust in both the presence and the absence of Gm in the EMC1 cultures. By contrast, Hcp1 expression was moderate to high in the PAO1 and EMC0 cultures grown without Gm but was low in the EMC1 culture grown without Gm. We conclude that in conditions that may potentially allow for the growth of Gm S bypass mutants in the EMC1 culture, elevated expression of the HSI-I T6SS protein, Hcp1, is not observed. This suggests that although the transcript encoding this protein (and, presumably, also other T6SS transcripts) is up-regulated in EMC1, translation of that transcript remains low. It is also worth noting that, consistent with . We also noted that exposure to sub-MIC aminoglycoside (here, Gm) strongly induced expression of the T6SS in EMC1 (as reported previously by Jones et al. (43)), although in the EMC1 background, this was not accompanied by a decrease in PcrV expression. Although the list of most highly up-regulated transcripts was dominated by the T3SS genes, the two genes with the highest modulation in EMC1 were an operonic pair, PA1325 and PA1326 (Table 3). PA1326 (ilvA2) encodes a threonine dehydratase involved in glycine, serine, and threonine metabolism, and the adjacent gene, PA1325, is uncharacterized. Indeed, a relatively large number of genes linked to amino acid metabolism were found to be up-regulated in this region of the genome. For example, the nearby ggt (PA1338) gene encodes a g-glutamyltranspeptidase involved in GSH catabolism (up-regulated in EMC1 2.6-fold). Adjacent to ggt is another set of up-regulated genes encoding an ABC transport system for glutamate and aspartate (aatP (PA1339, :4.0-fold), aatM (PA1340, :4.5-fold), aatQ (PA1341, :6.1-fold), and aatJ (PA1342, :3.7-fold)). Another up-regulated gene was PA3965 (:2.4-fold), encoding an AsnC-type transcriptional regulator that, although uncharacterized, is one of just two Lrp homologs encoded by PAO1. In P. aeruginosa, Lrp regulates the expression of genes involved in amino acid biosynthesis and catabolism (44). We also noted that transcripts from an uncharacterized gene cluster (PA3327-PA3336) encoding a nonribosomal peptide synthase were up-regulated. The product of the nonribosomal peptide synthase in the cluster (PA3327, :6.6-fold) Global impact of fusA1 mutation was predicted by antiSMASH (45) to generate a dipeptide core based around the condensation of serine and proline. Another large cluster (comprising 62 up-regulated transcripts in all) encoded ribosomal proteins and translation factors. Of these, 20 encoded 30S subunit proteins, and 26 encoded 50S subunit proteins. Elongation factor Tu, initiation factor-2 and peptide chain release factors, were also up-regulated. One possible reason for the overall up-regulation of ribosomal transcripts in EMC1 came from the earlier proteomic analysis, which revealed increased abundance of SmpB. This is an RNA-binding protein that is induced upon ribosomal stalling. The increased expression of SmpB indicates that the mutated FusA1 likely leads to decreased translation efficiency and is compensated for by an increase in the overall expression of the translation machinery. Several genes involved in RNA processing were also up-regulated, including the RNA 39-terminal phosphate cyclase (rtcA, :5.2-fold) and RNA ligase (rtcB, :9.3fold). A large number of transcripts were also down-regulated in EMC1, although generally, the magnitude of modulation was low (Table 3). A STRING analysis of the 200 most highly downregulated transcripts is shown in Fig. S11. Rather few distinct clusters were apparent; again, this is likely due to the predominance of transcripts from uncharacterized genes in the data set. However, a few of the identified genes are worthy of comment. First, the small untranslated RNA, rsmY (PA0527.1) was downregulated 3.8-fold. The rsmY-binding partner and antagonist, RsmA, was also down-regulated (;5.7-fold). These are potentially important observations because excess "free" (i.e. unsequestered, by rsmY) RsmA increases expression of the T3SS genes and depresses expression of the T6SS/exopolysaccharide genes (8). The expression of rsmY is thought to be regulated by the Gac signaling pathway. However, none of the other Gac/ Rsm pathway-encoding transcripts (ladS, retS, gacA, gacS) were modulated in EMC1, nor was rsmZ (another untranslated small RNA thought to act in a similar manner to rsmY). If current models are correct, a stoichiometric imbalance of rsmY/ RsmA levels should impinge reciprocally on T6SS expression and extracellular polysaccharide production (on the one hand) and T3SS expression (on the other). That we saw an increase in the expression of T3SS, T6SS, and exopolysaccharide biosynthetic genes suggests that signaling through the Rsm pathway may be dysregulated in EMC1. Second, and consistent with the proteomic observations, mexXY expression was up-regulated at a transcriptional level (:2-fold). EMC1 exhibited downregulation of pprA expression (;2.4-fold). PprA and PprB are predicted to be part of a two-component regulatory system controlling membrane permeability. PprAB expression leads to increased membrane permeability and increased sensitivity to antibiotics, including aminoglycosides (46). The decreased expression of pprA may have been enough to prevent activation of the two-component system and to decrease cell permeability, thereby also contributing toward the elevated aminoglycoside resistance in EMC1. Another factor that impinges upon mexXY-oprM expression is the extracytoplasmic s factor, SigX (47). sigX transcripts exhibited a slight decrease (;1.3-fold) in expression in EMC1. This is potentially significant because SigX normally stimulates expression of the response regulator, PprB, and Gicquel et al. (47) have also reported elevated mexXY expression in a sigX mutant. Taken together, the decreased expression of both sigX and pprA, as well as the increased expression of the mexXY transcriptional activator, ArmZ (observed in the proteomics), likely explains the elevated gentamicin resistance of EMC1 (Table 1). The transcriptomic data also revealed that, in contrast with the observed up-regulation of mexXY, another RND-family efflux pump, mexGHI-opmD, was down-regulated in EMC1. This efflux system provides resistance to a variety of xenobiotics and has recently also been associated with binding a quorum-sensing molecule, the Pseudomonas quinolone signal (48). Although the precise physiological function(s) of MexGHI-OpmD remains to be elucidated, it has no known association with gentamicin resistance (49,50). Finally, and consistent with the original aim of the study, cdrA transcription was down-regulated (;1.4-fold) in EMC1.

Discussion
In this work, we serendipitously identified an SNP in the fusA1 gene of P. aeruginosa that gave rise to increased gentamicin resistance and global changes in expression of the T3 and T6 (HSI-I) secretion systems and exopolysaccharide biosynthetic pathways. FusA1 mutations are a common feature in certain clinical P. aeruginosa isolates (such as those derived from CF sputum) and are potentially a low-cost response to exposure to sublethal concentrations of aminoglycosides in vitro (20,51,52). However, to our knowledge, the wider phenotypic consequences of such mutations have not been explored further. The current work shows that at least one such fusA1 mutation can have selective, but potentially clinically significant, effects beyond conferring antibiotic resistance, by affecting the expression of key virulence factors. Early indications that the fusA1 P443L mutant (EMC1) might be pleiotropically affected came from the unexpected observation that it had a larger growth defect (compared with the WT progenitor) in rich medium than it did in minimal medium. Our transcriptional data -showing that ribosomal gene expression is increased in EMC1-indicate that this is likely caused by the increased (but unmet, due to defective FusA1 P443L function) translational demand during rapid growth in rich medium.
The elevated gentamicin resistance associated with EMC1 and other fusA1 mutants (20) appears to be driven by increased expression of the mexXY-encoded aminoglycoside efflux pump. This, in turn, is likely due to increased expression of the cognate transcriptional activator of mexXY expression, ArmZ. In addition, EMC1 also displayed transcriptional hallmarks indicative of decreased cell envelope permeability to aminoglycosides (linked to lower sigX and pprA/B expression). It seems highly unlikely that the fusA1 mutation itself contributes directly toward gentamicin resistance because "bypass" mutants in which gentamicin sensitivity was restored could be isolated. These appear to be due to the acquisition of secondsite mutations because the fusA1 P443L mutation was retained in all cases tested.
Interestingly, post-translational modifications (such as those carried out by ErmBP or TetO) that prevent the binding of antibiotics to the ribosome or resistance-conferring mutations that block the binding of antibiotics to the ribosome also prevent induction of mexXY (53). Furthermore, although most of the antibiotics that are known to induce mexXY expression target the ribosome, not all of these are necessarily substrates of the pump (e.g. chloramphenicol). The mechanism underpinning this appears (at least in part) to be integrally linked with ArmZ expression. The armZ ORF is preceded by a short leader peptide, whose translation leads to the formation of a specific mRNA secondary structure. In this structure, a transcriptional terminator is exposed prior to the RNA polymerase reaching the armZ ORF, thereby preventing armZ (and thence mexXY) expression. However, when translation of the leader peptide is impaired, an alternative mRNA secondary structure forms in which the termination signal is occluded, allowing armZ expression (54). It requires no great leap of the imagination to infer that the impaired translation accompanying the fusA1 P443L mutation could lead to a similar occlusion of the transcriptional terminator, enabling increased armZ expression. It is possible that a similar mechanism(s) is responsible for some of the other pleiotropic (but nevertheless discrete) effects associated with the fusA1 P443L mutation.
Residue Pro-443 (P. aeruginosa numbering) is conserved in .70% of FusA1 sequences across multiple phyla (55). The P443L substitution in the FusA1 of EMC1 introduces a hydrophobic residue (leucine) in place of a "structure-breaking" one (proline). This was anticipated to have potential conformational consequences affecting the function or stability of the protein. Consistent with this, we found that purified FusA1 P443L mutant protein was somewhat less thermostable than the WT protein (DT m = 1.4°C) and displayed an altered intrinsic fluorescence profile. EF-G undergoes a series of large conformational reconfigurations during its catalytic cycle, especially upon ribosomal engagement. The most dynamic movement occurs in domain IV (responsible for binding mRNA in the A-site of the 30S ribosomal subunit), and this dictates the rotation of domains III and V (56). The P443L substitution is positioned in a loop region between an a-helix and a b-strand in domain III. This is likely to be functionally significant because this loop lies directly opposite the critical "switch-1" and "switch-2" motifs in domain I, which are essential for GTPase activity. During the GTPase reaction, the switch motifs undergo large conformational changes, bringing domain I within bonding distance of domain III (18,56,57). Indeed, Mace et al. (18) have identified a specific interaction between the switch-2 GTPase motif and residue Arg-465 in domain III of the EF-G from Thermus thermophilus (18). Given the very close three-dimensional proximity of Pro-443 to this critical residue (which corresponds to His-465 in P. aeruginosa FusA1), it is likely that mutation of Pro-443 interferes with the local protein conformation and impairs guanine nucleotide exchange. This is significant because ribosome translocation proceeds at a far slower rate in the absence of EF-G GTPase activity (58), and although less is known about the role(s) of the EF-G GTPase in ribosome recycling (the likely main function of FusA1), the elevated levels of SmpB in EMC1 do indicate increased ribosomal stalling.
Perhaps the most significant finding of the current work is that the fusA1 P443L mutant displayed elevated expression of the T3SS (at the transcriptomic and proteomic level) and elevated expression of the T6SS machinery (at the transcriptomic level). Interestingly, a selection of other fusA1 point mutants that have been previously demonstrated to display increased Gm resistance due to elevated mexXY expression did not show increased expression of the T3SS marker, PcrV. This suggests that the increased expression of the T3SS in EMC1 is not linked with mexXY expression per se. The elevated expression of T3 and T6 secretion systems in EMC1 was unexpected because these secretion are thought to be reciprocally regulated. This reciprocal regulation is linked with signaling through the Gac pathway, which controls levels of free RsmA (8). Excess free RsmA promotes T3SS expression and depresses T6SS/exopolysaccharide biosynthetic gene expression. The lower expression levels of rsmA in EMC1 are fully consistent with the observed up-regulation of T6SS gene expression but are inconsistent with the up-regulation of all five T3SS operons. However, we note that in some circumstances, the two secretion systems may be co-regulated. For example, genes encoding the T3 and T6 systems have been reported to be up-regulated in a sigX mutant (47). In this regard, we note that sigX was slightly (1.3-fold) down-regulated in EMC1.
In summary, we have shown here that a mutation in the gene encoding a ribosome-recycling factor, FusA1, can lead to largescale but discrete alterations in the physiology of P. aeruginosa. FusA1 mutants have been recently documented to confer resistance to aminoglycoside antibiotics (a phenotype that we confirm here), although their wider impact on virulence has not been reported. Our data show that one such fusA1 mutant displays greatly up-regulated expression of the T3SS machinery, as well as increased resistance to gentamicin. Taken together, our data indicate that these changes may be linked with sensing of diminished translational capacity in the fusA1 mutant, although additional work is required to confirm this and establish a mechanism. Consistent with this notion, it is worth recalling that treatment of P. aeruginosa with sub-MIC azithromycin (a macrolide targeting ribosome function) has also been shown to increase expression of the T3SS genes (59).

Growth conditions
Unless otherwise stated, P. aeruginosa PAO1 strains were grown at 37°C in M9 minimal medium supplemented with 0.5% (w/v) glucose. Planktonic cultures were grown with vigorous aeration for the indicated time or to the indicated optical density. Late exponential cells were harvested at 7 h.

EMC0 and plasmid construction
EMC0 (the PcdrAB::lacZ reporter strain) was constructed by subcloning the PCR-amplified 320-bp upstream region of cdrAB in to mini-CTX-lacZ, immediately in front of the promoterless lacZ ORF. The reporter plasmid was introduced into PAO1 by electroporation, and transformants were selected for on 50 mg/ml tetracycline. Following integration of the construct into the chromosome, the mini-CTX backbone was Global impact of fusA1 mutation removed by introducing pFLP2 into the transformants through biparental conjugation from b2163 (pFLP2). Transconjugants were selected for on 250 mg/ml carbenicillin. Transformants were streaked onto LB agar supplemented with 5% (w/v) sucrose to select for derivatives that had lost pFLP2 and then onto LB agar supplemented with 50 mg/ml tetracycline to verify successful excision of the tetracycline-resistance cassette. Loss of pFLP2 was confirmed through carbenicillin sensitivity. The resulting EMC0 was confirmed by PCR (and also by the wholegenome sequence data).
The complementation vectors, pfusA1 and pfusA1 P443L , were constructed by PCR-amplifying the WT fusA1 and mutated fusA1 P443L ORFs from PAO1-and EMC1-derived genomic DNA, respectively. The amplicons were then cloned into the PstI/HindIII sites in the MCS of pUCP20 (downstream of the lac promoter on the plasmid). The resulting plasmids were introduced into the recipient strains by electroporation and selection on 250 mg/ml carbenicillin. All plasmid constructs were confirmed by sequencing.

Plasposon mutagenesis
The pTnMod-OGm plasposon (22) was introduced into EMC0 via triparental mating. EMC0 was spotted onto solid agar with a helper Escherichia coli strain (HB101 (pRK21013)) and E. coli JM109 (pTnMod-OGm) for 18 h. The mixed colony was then resuspended, and transposon mutants were selected on agar media supplemented with 5 mg/ml 21 gentamicin and 30 mg/ml 21 X-Gal. After 30 h, gentamicin-resistant colonies were transferred to fresh gentamicin-supplemented plates and grown for a further 24 h. EMC1 was isolated as a pale blue colony at this stage.

b-gal activity
Aliquots (100 ml) of planktonic culture were harvested over a 10-h growth period and frozen at 280°C in a 96-well microtiter plate. After collection was complete, the plate was defrosted for 30 min at 37°C, and 10 ml was transferred to a fresh 96-well microtiter plate. The samples were frozen at 280°C for a second time, followed by thawing at room temperature. To quantitatively measure the level of b-gal production, 100 ml of PBS containing 20 mg/ml lysozyme and 250 mg/ml 4-methylumbelliferyl-b-galactoside was added to the cells. The reaction progress was monitored every 30 s for 30 min at 37°C in a Gemini XPS fluorimeter (Molecular Devices) using an excitation wavelength of 360 nm and emission wavelength of 450 nm.
Whole-genome sequencing EMC0 and EMC1 were sequenced using Illumina technology (HiSeq 2500 platform) by MicrobesNG (Birmingham, UK). Genomic DNA libraries were prepared using Nextera XT Library Prep Kit (Illumina) following the manufacturer's protocol with the following modifications: 2 ng of DNA were used instead of 1 ng, and PCR elongation time was increased to 1 min from 30 s. DNA quantification and library preparation were carried out on a Hamilton Microlab STAR automated liquid-handling system. Pooled libraries were quantified using the Kapa Biosystems Library Quantification Kit on a Roche light cycler 96 quantitative PCR machine. Libraries were sequenced on the Illumina HiSeq system using a 250-bp paired end protocol. Reads were adapter-trimmed using Trimmomatic 0.30 with a sliding window quality cutoff of Q15. De novo assembly was performed on samples using SPAdes version 3.7, and contigs were annotated using Prokka 1.11. Genome alignment and analysis was carried out using Mauve Multiple Genome Alignment (60). This Whole Genome Shotgun project has been deposited at DDBJ/ENA/GenBank TM under accession numbers JAAGAW000000000 (EMC0) and JAAGAV000000000 (EMC1). The versions described in this paper are versions JAAGAW010000000 (EMC0) and JAAGAV010000000 (EMC1).

Motility assays
To detect twitching motility, 10 ml of 1.5% (w/v) agar-containing LB or M9 minimal medium/glucose was prepared in a 10-cm diameter Petri dish. Colonies were stabbed into the agar with a sterile toothpick, and the twitch halo was visualized after 24 h of incubation at 37°C. To assess swimming motility, 25 ml of LB or M9 minimal media/glucose containing 0.3% (w/v) Bacto agar was dispensed into a 10-cm diameter Petri dish and allowed to solidify/surface-dry at room temperature for 30 min. Bacterial cultures were normalized to an OD 600 of 1, and 3 ml of culture was dispensed as a stab column into the agar using a pipette. The halo of swimming bacteria was assessed after 8-18 h of incubation at 37°C.

Exoenzyme secretion assays
Gelatinase activity was measured on solid medium plates containing 13 g/liter nutrient broth and 30 g/liter gelatin. Bacterial cultures were normalized to an OD 600 of 1, and 5 ml was spotted onto the agar and left to soak in. Plates were incubated at 37°C overnight. Gelatinase activity was visualized by flooding the plates with saturated ammonium sulfate solution for 15 min to reveal the proteolytic halo.

Exopolysaccharide secretion
Agar plates for the detection of exopolysaccharide production were prepared using 37 g/liter brain heart infusion broth supplemented with 50 g/liter sucrose and 0.8 g/liter Congo Red. Bacterial cultures were normalized to an OD 600 of 1, and 10 ml was spotted onto the surface of the agar. The plates were incubated at 37°C for 24 h. Exopolysaccharide production was determined semi-quantitatively through the amount of red pigmentation associated with the colony.
To quantitatively assess exopolysaccharide production, bacterial cultures were subcultured into fresh growth medium supplemented with 10 mg/ml Congo Red and incubated at 37°C for 24 h on a rotating drum. The OD 600 was measured, and the cells were pelleted at 3200 3 g for 10 min at room temperature. This depletes any Congo Red bound to cell-associated polysaccharides; the greater the amount of polysaccharide sedimented, the lower the amount of Congo Red remaining in the culture supernatant. The optical density of the supernatant was measured at 495 nm. The A 495 was then normalized against the original OD 600 of the culture.

Siderophore detection
Bacterial cultures were harvested at late exponential phase. The cells were sedimented (3200 3 g for 10 min), and the presence of siderophores in the cell-free culture supernatant was measured using the colorimetric SideroTech kit (Emergen Bio Inc.). A color change occurs as ferric iron in the kit reagent binds to siderophores present in the culture supernatants.

Western blotting
Following SDS-PAGE, proteins were transferred onto a polyvinylidene difluoride membrane using a Trans-Blot Turbo Transfer System (Bio-Rad) with Trans-Blot Mini Transfer Packs. The membrane was incubated in blocking buffer (PBS containing 0.1% (v/v) Tween 20 and 5% (w/v) semi-skimmed milk powder) overnight and then washed once in wash buffer (PBS and 0.1% (v/v) Tween 20). The membrane was then incubated for 1 h with a primary anti-PcrV antibody (1:5000) or anti-ICD antibody (1:10,000) or anti-Hcp1 antibody (1:10,000) (as indicated) in wash buffer. The membrane was then washed four times in wash buffer (5 min each wash), before the addition of IRDye 680RD (LI-COR) secondary antibody. This was incubated for 45 min. The membrane was washed four times, as before, and the protein bands were detected using an Odyssey CLx imaging system (LI-COR).

Proteomic analyses
Strains were grown with good aeration at 37°C to late exponential phase (OD 600 of 0.6-0.8) in M9 minimal medium supplemented with 0.5% (w/v) glucose. To ensure retention of the complementation and empty vector plasmids, carbenicillin (250 mg/ml) was included in the starter culture, but not in the subculture used for proteomic analysis. We independently confirmed that the pUCP20 plasmid was retained by essentially all cells in the absence of selection over this sampling period (data not shown). For each strain, three biological replicates were analyzed. The proteome of a fourth EMC0 replicate was analyzed by MS but was not included in this study. Cells from a 45-ml culture were harvested at 3200 3 g for 30 min at 4°C. The cell pellet was resuspended in PBS and sedimented a second time. Pellets were then resuspended in 800 ml of lysis buffer (100 mM Tris-HCl, 50 mM NaCl, 20 mM EDTA, 10% (v/v) glycerol, 1 mM DTT, pH 7.5) containing a Complete TM protease inhibitor mixture tablet (Roche Applied Science) and sonicated (3 3 5 s at 15 A, MSE microtip) on ice. Unlysed cells and debris were pelleted at 21,000 3 g for 30 min at 4°C. The protein concentration of the supernatant was determined using the DC protein assay (Bio-Rad). LC-MS/MS was performed by the Cambridge Centre for Proteomics. Protein samples were precipitated using ice-cold acetone. The samples were digested with trypsin with a specificity allowing for a maximum of one missed cleavage, carbamidomethylation of cysteine residues as a fixed modification, and oxidation of methionine and deamidation of glutamine/asparagine as variable modifications. Mass tolerance for precursor ions was 10 ppm, and for fragment ions, it was 0.8 Da. The dried peptides were reconstituted in 100 mM triethylammonium bicarbonate and labeled using 10-plex TMT (tandem mass tag) reagents according to the manufacturer's proto-col (Thermo Scientific). Tagged peptides were fractionated into 20 fractions by reverse-phase chromatography. Each fraction was run on 2-h gradients and was identified and quantified using a high-resolution Lumos Fusion Orbitrap mass spectrometer coupled to a Dionex Ultimate 3000 RSLC nano-UPLC (Thermo Fisher Scientific). Synchronous precursor selection was used to ensure minimal TMT ratio compression (61). Proteome Discoverer (version 2.1.0.81; Thermo Fisher Scientific) was used to process raw MS data, with peptides identified using MASCOT (version 2.6.0, Matrix Science). Only high-confidence peptides defined by MASCOT with a 1% FDR were considered for peptide identification. FDRs were controlled with Percolator (version 2.05) at both the peptide-spectrum match and protein level using a strict threshold of 0.01 and a relaxed threshold of 0.05. Peptide and proteins were quantified by the Proteome Discoverer quantification module with the top three peptides used for area calculation. Proteomic data sets were analyzed with the empirical Bayes moderated t test implemented by the limma package (62). p values were corrected for multiple hypothesis testing using the Benjamini-Hochberg method (FDR 0.05). Differential expression was calculated based on normalized log 2 ratios. The MS/MS fragmentation data were searched against the National Center for Biotechnology Information (NCBI) Pseudomonas aeruginosa database (2017, 5584 entries) using the MASCOT search engine. The MS proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (63) partner repository with the data set identifier PXD017266.

Expression and purification of FusA1
For FusA1 purification, the WT and mutated fusA1 genes were PCR-amplified (from PAO1 and EMC1, respectively) and cloned into pET-19m. The resulting plasmids were introduced into Rosetta E. coli cells. The Rosetta cultures were grown in 1 liter of LB supplemented with 50 mg/ml carbenicillin and incubated at 37°C to an OD 600 of 0.6-0.7. Isopropyl-b-D-thiogalactopyranoside was added to the cultures to a final concentration of 1 mM, and the culture was incubated at 20°C for 24 h. Cells were centrifuged at 3430 3 g, for 20 min at 4°C. The cell pellet was resuspended in 10 ml of lysis buffer (100 mM Tris-HCl, 50 mM NaCl, 20 mM EDTA, 10% (v/v) glycerol, 1 mM DTT, pH 7.5) containing a Complete TM protease inhibitor mixture tablet (Roche Applied Science) and lysed by sonication (5 3 30 s at 13 A) on ice. The samples were then clarified by centrifugation (14,636 3 g for 30 min, 4°C), and the supernatant was passed through a 0.45-mm membrane filter (Millipore). FusA1 protein was purified from the soluble, filtered extracts by loading onto a pre-equilibrated nickel-nitrilotriacetic acid Superflow Cartridge (Qiagen) at 4°C. The column was equilibrated with protein purification buffer (50 mM sodium phosphate, 200 mM NaCl, 10 mM imidazole, 10% (v/v) glycerol, pH 8.0). The column was washed until the A 280 of the eluate was negligible and eluted in the same buffer containing 250 mM imidazole. The eluted protein was concentrated in a Vivaspin-20 and then dialyzed against 1 liter of 50 mM Tris, 100 mM NaCl, 5% (v/v) glycerol, pH 7.4. Aliquots of the protein were stored frozen at 280°C until use.

Global impact of fusA1 mutation Intrinsic tryptophan fluorescence
The concentration of appropriately diluted purified His 6-FusA1 was determined spectrophotometrically at 280 nm using a quartz cuvette (1-cm pathlength) and assuming MW = 78,738 and e = 61,310 M 21 cm 21 . EF-G protein was diluted in 2 ml of dialysis buffer to a final concentration of 0.8 mM. Intrinsic Trp fluorescence was measured in a thermostated (25°C) quartz cuvette, on an FP-8300 spectrofluorometer (JASCO) using an excitation wavelength of 295 nm and an emission wavelength of 305-400 nm (0.5-nm intervals, 100 nm/min). All spectra were recorded in triplicate and then averaged.

Thermal shift analyses
Purified WT FusA1 or FusA1 P443L (20-ml volume containing 20 mM protein) was mixed with SYPRO Orange dye and subjected to a thermal scan (25°C ! 95°C) in a Roche Lightcycler 480, and the fluorescence was measured at 483 and 568 nm.

RNA extraction and sequencing
EMC0, EMC1 (both containing the pUCP20 empty vector), and EMC1* (containing pUCP20:fusA1) were cultured in M9 minimal medium supplemented with glucose, and samples were harvested at the late exponential phase of growth (OD 600 of 0.6-0.8) into RNAlater (Ambion). The samples were incubated at 4°C for 15 min and pelleted at 21,000 3 g for 20 min at 4°C. Total RNA was extracted using the RNeasy Mini Kit (Qiagen) and digested twice with on-the-column DNase I digestion, following the manufacturer's guidelines. The concentration and purity of the resulting RNA were measured using a Nano-Drop ND-1000 spectrophotometer. The absence of contaminating proteins and organic compounds was indicated by an A 260/280 ratio of 1.8-2.0 and A 260/230 nm of 2.0-2.2. Total RNA was sent to GATC Biotech (Konstanz, Germany) for rRNA depletion and RNA sequencing using the Illumina platform (10 million reads per sample, single read, 1 3 50 bp). The RNA-Seq reads were processed using FastaQC and were mapped to the PAO1 genome and analyzed using the Tuxedo Suite package. The principal component analysis plot and the volcano plots were constructed in R. RNA-Seq data have been deposited in the ArrayExpress database at EMBL-EBI (64) under accession number E-MTAB-8690.
For the quantitative RT-PCR analyses, cDNA synthesis was required. To do this, 1 mg of RNA was combined with 50 ng of oligo(dT) 15 , 300 ng of random hexamers, and 1 ml of dNTP mix. The samples were then heated at 65°C for 5 min, followed by incubation on ice for 2 min. To this mixture, 4 ml of 53 SuperScript III First strand buffer, 1 ml of 0.1 M DTT, and 1 ml of SuperScript III reverse transcriptase (Invitrogen) was added. Samples were incubated at 25°C for 5 min, followed by 50°C for 60 min. The reaction was inactivated by heating to 70°C for 15 min. cDNA was used as a template for RT-PCR using 20-25 cycles of 98°C (10 s), 55°C (20 s), 72°C (20 s/kb).

Data availability
The data supporting the findings of this study are available within this article (and the supporting material) or can be accessed through publicly available repositories. Whole-genome sequencing data sets generated in this study are openly available in GenBank TM , DDBJ, and ENA, under the accession numbers JAAGAW000000000 (EMC0) and JAAGAV000000000 (EMC1). The versions described in this paper are versions JAAGAW010000000 (EMC0) and JAAGAV010000000 (EMC1). Proteomics data generated in the study are openly available via the ProteomeXchange Consortium using the identifier PXD017266. RNA-Seq data sets generated in this study are openly available in the ArrayExpress database under the accession number E-MTAB-8690. Conflict of interest-The authors declare that they have no conflicts of interest with the contents of this article.