(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Plasmids manipulate bacterial behaviour through translational regulatory crosstalk [1]

['Catriona M. A. Thompson', 'Department Of Molecular Microbiology', 'John Innes Centre', 'Colney Lane', 'Norwich', 'United Kingdom', 'School Of Biological Sciences', 'University Of East Anglia', 'Norwich Research Park', 'Norfolk']

Date: 2023-02 

Beyond their role in horizontal gene transfer, conjugative plasmids commonly encode homologues of bacterial regulators. Known plasmid regulator homologues have highly targeted effects upon the transcription of specific bacterial traits. Here, we characterise a plasmid translational regulator, RsmQ, capable of taking global regulatory control in Pseudomonas fluorescens and causing a behavioural switch from motile to sessile lifestyle. RsmQ acts as a global regulator, controlling the host proteome through direct interaction with host mRNAs and interference with the host’s translational regulatory network. This mRNA interference leads to large-scale proteomic changes in metabolic genes, key regulators, and genes involved in chemotaxis, thus controlling bacterial metabolism and motility. Moreover, comparative analyses found RsmQ to be encoded on a large number of divergent plasmids isolated from multiple bacterial host taxa, suggesting the widespread importance of RsmQ for manipulating bacterial behaviour across clinical, environmental, and agricultural niches. RsmQ is a widespread plasmid global translational regulator primarily evolved for host chromosomal control to manipulate bacterial behaviour and lifestyle.

Funding: JGM and CMAT were supported by BBSRC Responsive mode Grant BB/R018154/1 to JGM. JGM and RHL were supported by BBSRC Institute Strategic Programme Grant BBS/E/J/000PR9797 to the John Innes Centre. SP was supported by a Royal Thai Government PhD Scholarship. AP was supported by UKRI-BBSRC Grant BB/T004363/1 to JGM. JPH was supported by BB/R014884/1. MAB, SF and SMB were supported by BBSRC grants BB/R014884/1, BB/R014884/2, BB/R018154/1 and NERC grants NE/R008825/1, NE/R008825/2. EH is supported by a NERC Independent Research Fellowship NE/P017584/1. RWJ is supported by BBSRC grants BB/R014884/1 and BB/T010568/1. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability: All relevant data are within the paper and its Supporting Information files, with the following exceptions: Processed RNA-seq data is deposited in ArrayExpress (E-MTAB-11868). Experimental data for the proteomic experiments is deposited in ProteomeXchange (PXD033640). Scripts and bioinformatic analyses can be found at www.github.com/jpjh/PLASMAN_RsmQ .

Copyright: © 2023 Thompson et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Plasmids are recognised for their important role in bacterial evolution as drivers of horizontal gene transfer. Less well understood are the effects of plasmids upon bacterial behaviours by manipulating the expression of key bacterial phenotypes. Until now, examples of plasmid manipulation of their bacterial hosts were limited to highly targeted transcriptional or translational control of a few related traits. In contrast, here we describe the first plasmid global translational regulator evolved to control the bacterial behavioural switch from a motile to a sessile lifestyle and bacterial metabolism, mediated through manipulation of the bacterial proteome. Moreover, this global translational regulator is common across divergent plasmids in a wide range of bacterial host taxa, suggesting that plasmids may commonly control bacterial lifestyle in the clinic, agricultural fields, and beyond.

To understand the function of rsmQ, we explored the distribution of rsm genes on plasmids, and the effects of rsmQ on the transcriptome and proteome of SBW25, as well as on the expression of key bacterial ecological traits. Further, we biochemically characterised the interactions of RsmQ with a close structural proxy for RNA (single-stranded DNA (ssDNA)) and with the bacterial Rsm proteins. Our findings show that rsm genes are widespread on Pseudomonas plasmids and that RsmQ interacts with the resident Gac-Rsm system and the host RNA pool, binding to specific nucleotide motifs in order to post transcriptionally regulate translation. RsmQ extensively remodelled the SBW25 proteome including altering production of metabolites nutrient uptake and chemotaxis pathways, despite having only a limited impact on the SBW25 transcriptome. In turn, RsmQ translational regulation altered the expression of ecologically important bacterial behaviours, most notably motility, conjugation rate, and carbon source metabolism. Together, our findings expand the known molecular mechanisms causing PCC to include translational regulator homologues, which act in this case to extensively manipulate bacterial behaviour by altering the expression of a range of ecologically important bacterial traits. These findings have broad implications for understanding the role of plasmids in microbial communities.

In this study, we investigate the role of translational regulation in mediating PCC between Pseudomonas fluorescens SBW25 and the 425 kb conjugative plasmid pQBR103. P. fluorescens is a common, soil-dwelling, plant growth-promoting bacterium that is capable of accepting diverse plasmids, including those from the pQBR family of large conjugative plasmids [ 36 , 37 ]. Both SBW25 and the pQBR plasmids were first isolated in the 1990s from the sugar beet rhizosphere at Wytham Woods in the United Kingdom [ 36 , 38 ]. The ability of several of the pQBR plasmids to persist within P. fluorescens strains across a range of environments including in soil, on plants, and in lab media has been well documented [ 36 , 37 , 39 , 40 ]. Moreover, acquisition of pQBR103 by P. fluorescens SBW25 alters the expression of approximately 440 chromosomal genes [ 40 , 41 ]. The large-scale regulatory disruption caused by pQBR103 can be negated by a range of compensatory mutations restoring wild-type (WT) expression levels, including loss-of-function mutations affecting the bacterial TCS gacA/S. Notably, while the genetic sequence of pQBR103 encodes a range of accessory functions including mercury resistance and UV resistance, it also encodes a homologue of the widespread rsmA bacterial translational regulator gene, which we identify here as rsmQ.

The Gac-Rsm pathway is one of the best characterised translational regulatory systems in pseudomonads [ 16 – 20 ] and controls a wide variety of traits including biofilm formation [ 21 ], motility [ 22 ], quorum sensing [ 23 ], siderophore production [ 24 ], and virulence [ 12 , 25 ]. Gac-Rsm is highly conserved within the Pseudomonas genus and comparable systems exist in a wide range of bacteria [ 12 , 20 , 24 , 26 , 27 ]. Rsm proteins are able to interact directly with the bases AnGGA around the ribosome-binding site (RBS) of their target transcript [ 27 – 30 ]. Rsm proteins can both activate and repress bound mRNA transcripts, either by opening up the mRNA to allow ribosomal access to the RBS, or by making the RBS inaccessible [ 30 – 32 ]. This allows Rsm proteins to exert tight translational control over a wide range of targets to impact bacterial phenotypes [ 33 ]. The activation of Rsm proteins are regulated by the activation of the GacA/S two-component system (TCS), which in turn is activated by a complex, but largely uncharacterised, set of environmental cues. Upon activation, GacA promotes transcription of the small-regulatory RNAs RsmY and RsmZ, which leads to the sequestering of regulatory Rsm proteins away from their mRNA targets through competition for binding [ 30 , 34 ]. The number of Rsm proteins encoded by individual Pseudomonas species varies, with each protein having both unique and overlapping regulons with other Rsm proteins [ 35 ]. The large number of traits regulated by the Gac-Rsm system suggests that there could be significant effects on bacterial behaviour caused by PCC manipulating this system.

The molecular mechanisms of known PCC involve plasmid-encoded transcriptional regulators causing targeted changes to the expression of small numbers of chromosomal genes. Although transcriptional regulation is important for bacterial survival and adaptation, bacteria also rely on translational regulation to respond to changes in their environment [ 9 ]. Bacteria are able to exert this control by deploying second messenger signals [ 10 ], directly altering the ribosome [ 11 ] or impacting mRNA stability and accessibility via pathways such as Gac-Rsm [ 12 , 13 ]. Although specific targeted instances of translational control have been observed [ 14 , 15 ], it is currently unknown whether conjugative plasmids are able to manipulate global translational regulatory pathways.

To date, well-characterised PCCs involve plasmid-encoded transcriptional regulators that alter the expression of specific bacterial traits. For example, in Acinetobacter baumannii several multidrug resistance plasmids encode transcriptional repressors of the bacterial type VI secretion system (T6SS) [ 6 ]. Plasmid-mediated repression of the T6SS enhances plasmid horizontal transmission by ensuring that plasmid recipient cells are not killed by the donor’s T6SS apparatus [ 7 ]. Similarly, plasmid-encoded transcriptional regulators alter the expression of several chromosomal regulators of virulence associated traits in Rhodococcus equi, thus enhancing survival of both the bacterium and the plasmid in macrophages by stalling phagosomal maturation [ 8 ]. Together, these examples suggest that plasmid-encoded regulatory homologues may have important fitness consequences for the plasmid, either through horizontal replication, through conjugation to new host cells or through vertical replication within the current host cell [ 3 ].

Bacteria regulate the expression of functional traits in response to their environment, enabling colonisation of diverse ecological niches. However, control over bacterial gene regulation is not exclusively under the control of the bacterial genome [ 1 , 2 ]. The mobile genetic elements that inhabit bacterial hosts, such as conjugative plasmids, commonly encode homologues of bacterial regulators [ 3 , 4 ]. The introduction of plasmid-encoded regulator homologues into the bacterial cell can rewire the gene regulatory networks of the bacterium, potentially altering the expression of bacterial traits, a process termed plasmid-chromosome crosstalk (PCC, [ 5 ]). However, how and why plasmid-encoded regulators would manipulate the expression of bacterial traits is poorly understood.

Results

Plasmids encode regulatory protein homologues The ORF PQBR443, hereafter rsmQ, was identified on pQBR103 as a homologue of the chromosomal csrA/rsmA genes found widely within proteobacteria. We hypothesised that this gene could act as a mediator of PCC [4]. To identify whether carriage of an rsm homologue is peculiar to pQBR103 or is a general phenomenon across plasmids, we investigated the distribution of rsmQ homologues in the 12,084 plasmids of the COMPASS database [42]. Within this set, and consistent with previous studies [43], we detected 106 putative rsmQ homologues on 98 plasmids (0.8%), mostly isolated (92/98) from proteobacteria, particularly Pseudomonadaceae and Legionellaceae (Figs 1A and S1C). The distribution of rsm-containing plasmids was not uniform across taxa (Fisher’s exact test, p < 0.0005): approximately 20% of Pseudomonadaceae (41/196) and Piscirickettsiaceae (12/67) plasmids, and over 50% of Legionellaceae (21/29) plasmids contained rsm homologues, while no rsm homologues were detected on any of the 3,621 Enterobacteriaceae plasmids. rsm-containing plasmids were relatively large, with the smallest at 32.4 kb, sitting at the larger end of the size distribution for each taxon (S1A Fig). There was no general association between rsm-carriage and plasmid mobility across taxa, although within Legionellaceae, rsm-encoding plasmids tended to be conjugative (Fisher’s exact test Bonferroni-adjusted p-value = 0.02). Within Pseudomonadaceae, rsm-encoding plasmids tended to have proportionally more genes with predicted rsm binding sites (Kolmogorov–Smirnov test, p = 0.012, S1B Fig). Overall, these patterns suggest that plasmid carriage of rsm is not uncommon, but is taxon specific, indicating a functional role that is associated with particular groups of microorganisms. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. RsmQ is found on a wide range of conjugative plasmids. (a) Taxonomic distribution of plasmid borne csrA/rsmA homologues identified in COMPASS. (b) COMPASS plasmid diversity represented by NMDS of MASH sequence distances. Families with ≥1 plasmid with a csrA/rsmA homologue are coloured according to the legend. Plasmids encoding csrA/rsmA homologues are outlined in black. Selected plasmids from various taxa are annotated. (c) Unrooted phylogenetic tree of Pseudomonas csrA/rsmA homologues from COMPASS, with corresponding chromosomal homologues (where available) and genes from selected reference strains. Branches leading to nodes with >80% bootstrap support are coloured black, with decreasing support indicated with increasingly pale grey. P. fluorescens SBW25 csrA/rsmA homologues are labelled, as is pQBR103 rsmQ, and E. coli csrA. The blue highlight indicates a well-supported (bootstrap support 0.84) group of rsmA-like homologues. The green highlight indicates the group of related plasmid and chromosomal genes from plant pathogen Pseudomonas discussed in the text. All data and analyses are available on github (PLASMAN_RsmQ) with data taken from the COMPASS database. NMDS, non-metric multidimensional scaling. https://doi.org/10.1371/journal.pbio.3001988.g001 It is possible that rsm-encoding plasmids have recently spread horizontally across different species. If this was the case, we would expect rsm-encoding plasmids to be more similar to one another than to non-rsm-encoding plasmids within each taxon. To investigate the diversity of rsm-encoding plasmids relative to the other plasmids in COMPASS, we performed UMAP non-metric multidimensional scaling (NMDS) on pairwise MASH distances between plasmids [44–46]. Within the diversity of plasmids in COMPASS, rsm-containing plasmids are diverse and often cluster close to non-rsm-containing plasmids isolated from the same taxa (Fig 1B), suggesting that carriage of rsm regulators by plasmids is a convergent trait that has emerged several times over. Global regulatory genes may be frequently (re)acquired by plasmids from the bacterial chromosome. Alternatively, these genes may have a prolonged association with plasmids and evolve distinctly to chromosomal genes. To investigate these possibilities, we built a phylogenetic tree of the csrA/rsmA homologues from all Pseudomonas plasmids and their associated chromosomes (where available), alongside the rsm genes from 7 diverse Pseudomonas strains: Pseudomonas protegens CHA0, P. fluorescens Pf0-1, P. protegens Pf-5, P. fluorescens SBW25, Pseudomonas putida KT2440, Pseudomonas aeruginosa PAO1, and P. aeruginosa PA14 (Fig 1C). Chromosomal homologues of csrA/rsmA formed several distinct clusters (bootstrap support >80%), with 1 cluster including P. fluorescens SBW25 rsmA and the Escherichia coli homologue csrA. However, plasmid-borne csrA/rsmA homologues were more divergent than those that were chromosomally encoded (Fig 1C). Additionally, chromosomal homologues (including the P. fluorescens SBW25 genes rsmE and rsmI) formed a distinct cluster. Consistent with the phylogenetic analysis, sequence variation among chromosomal rsm homologues was significantly lower than when comparing chromosomal- with plasmid-borne rsm homologues (Wilcoxon test, Bonferroni-adjusted p < 0.0001) or when comparing plasmid-borne rsm homologues with one another (Wilcoxon test, p < 0.0001). The principal exception to this pattern was a cluster of closely related plasmid and chromosomal rsm genes from plant pathogenic Pseudomonas (green highlighted, Fig 1C). However, it is possible that some of these chromosomal variants are associated with chromosomally located mobile genetic elements, as at least one of these homologues is located on an annotated integrative conjugative element [47]. Overall, our comparative sequence analysis suggests that diverse plasmids have independently acquired rsm homologues, which then evolve and diversify as part of the plasmid mobile gene pool, distinct from their chromosomal counterparts. Although plasmid-encoded rsm homologues are widespread among plasmids [43], very little is currently known about their role in PCC or how they might impact bacterial behaviour.

RsmQ binds to specific RNA targets Despite a high degree of sequence similarity, it was unknown if RsmQ would be functionally similar to the chromosomally encoded SBW25 Rsm proteins (RsmA/E/I). Rsm proteins from Pseudomonas species interact with a conserved RNA sequence (AnGGA), with these bases interacting directly with the proteins’ conserved binding site (VHRE/D) [29,30]. To confirm whether RsmQ acts similarly, we designed a high-throughput method to examine the nucleotide binding properties of RsmQ in vitro using the ReDCaT surface plasmon resonance (SPR) system [48], which is primarily designed for examining double-stranded DNA (dsDNA)–protein interactions. Because Rsm proteins only interact with the nucleotide bases of RNA molecules, protein–nucleic acid interactions can be effectively examined using ssDNA probes. ssDNA probes containing the predicted RNA target sequence (ACGGA) and a nonspecific scrambled sequence were synthesised with the ReDCaT linker at the 3′ end, with either a linear or a hairpin secondary structure with the potential binding site at the top of the hairpin. RsmQ interacted strongly with both the minimal (GGA) and full length (ACGGA) binding sites when these were at the top of a hairpin loop. When the binding sequence was presented in a linear oligo RsmQ could interact but quickly dissociated, suggesting that the preferred binding site is open at the top of a hairpin loop (S2A Fig). No interaction was seen between RsmQ and a scrambled binding site confirming that the binding is specific to the target GGA/ACGGA sequence (Fig 2). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. RsmQ interacts with its preferred binding sequence (GGA/AnGGA) and this interaction is mediated by the VHRE/D binding site. (a) Percentage R max values for RsmQ WT (green) H43A (blue) and R44A (red) binding to ssDNA containing the binding sites shown above. Two-way ANOVA showed an effect of both binding site (p < 0.0001) and mutation (p < 0.0001) on R max . (b) Percentage R max values for WT RsmQ binding to ssDNAs containing the above binding site. Binding sites that are predicted to bind RsmQ are shown in purple. All oligos are designed as hairpins and results shown are for RsmQ at a concentration of 100 nM. Error bars represent the standard deviation of 2 replicates. Data are available in S1 Data. ssDNA, single-stranded DNA; WT, wild-type. https://doi.org/10.1371/journal.pbio.3001988.g002 Next, we tested if the RNA binding interaction was co-ordinated by the conserved VHRD/E motif at the C-terminus of RsmQ by examining the binding of 2 RsmQ mutants, in which key residues within the motif were changed to alanine residues (H43A and R44A), to the ssDNA probes. The alteration of these residues significantly reduced the efficiency of RsmQ binding to the target sequence (Fig 2A). Finally, to confirm the minimum RNA-binding sequence, a series of near-identical ssDNA nucleotides were tested containing the simple and full binding site sequences with a single base pair change in each case. RsmQ preferentially bound to the known binding sites GGA and A(N)GGA with a markedly higher affinity than to any of the alternate sequences tested and with a slight preference for ATGGA/AGGGA sequences, further supporting the hypothesis that RsmQ is a specific RNA binding protein that functions similarly to the chromosomal Rsm proteins (Fig 2B). Previous work by Duss and colleagues [49] has shown that RsmE is able to directly interact with the GGA motif of the AnGGA binding motif. Therefore, to directly compare RsmQ to its chromosomal counterparts in this assay, the interaction of RsmE to each of these oligos was also examined. Interestingly, whilst RsmQ showed an affinity for all AnGGA binding sites, as well as the simplified GGA binding site, RsmE was only able to effectively interact with the ATGGA and AGGGA motifs (S2B and S2C Fig), suggesting that RsmQ may have a stronger affinity than the chromosomal Rsm proteins for their mRNA targets.

RsmQ interacts with the host Rsm system Notwithstanding the evidence for direct regulation of translation by RsmQ binding to mRNA, the large remainder of differentially regulated proteins without predicted Rsm binding sites suggests an indirect mechanism by which RsmQ regulates the abundance of these proteins. Given that RsmQ closely mimics the RNA binding characteristics of host Rsm proteins (Fig 2), we next investigated whether RsmQ interacts with other elements of the host Rsm regulatory pathway. The activity of host Rsm proteins is controlled by the noncoding RNAs (ncRNAs) RsmY/Z, which act as protein sponges, sequestering Rsm proteins away from their target mRNAs [30,61]. To test RsmQ binding to the ncRNAs RsmY and RsmZ, we copied the individual stem loops of each ncRNA into ssDNA oligos of approximately 25 bp in length and attached them to the ReDCaT linker. The oligos were modelled to determine the location of the binding site in both the ncRNA and in the case of the ssDNA sequence, to determine if this was located at the top of a stem loop. Strong binding to several ssDNA probes was observed, in each case contingent on the presence of at least a GGA sequence at the top of a stem loop, with RsmY 1–25 and RsmZ 26–50 having AnGGA motifs present (Fig 5A). These data suggest that RsmQ interacts with the Gac-Rsm regulatory system by binding to the host ncRNAs RsmZ and RsmY. This would lead to either an increase in RsmQ target translation as RsmQ is titrated away from its targets or an increase in RsmA/I/E binding to target mRNAs due to a reduction in available RsmZ/Y binding sites. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. RsmQ can both homo- and heterodimerise. (a) Percentage R max values for RsmQ binding to portions of the ncRNAs RsmY (pink) and RsmZ (blue) showing preferential binding to ssDNAs that contained the binding site in a hairpin loop. Error bars show the standard deviation of 2 replicates and oligos containing a AnGGA binding site are indicated with a (*), with those that have the full AnGGA binding site at the top of a stem loop (+) and with the shorter GGA at the top of a stem loop (-) indicated. (b) AlphaFold model of RsmQ suggests that it forms homodimers (monomers shown in contrasting colours), with the RNA-binding region highlighted in marine (B5). (c) Quantitative bacterial-2-hybrid β-galactosidase assays showing interactions between pUT18c and pKNT25 fusions are shown for RsmA (A), RsmE (E), RsmI (I) and RsmQ (Q). Results were analysed by a one-way ANOVA (p < 0.0001) with Tukey’s multiple comparisons against the empty plasmid control (-:-) indicated (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Error bars represent the standard deviation of 2 biological replicates. Additional controls are shown in S7 Fig. (d) Representative image of qualitative β-galactosidase assays on agar plates. pKT25 fusions are shown in rows and pUT18c fusions in columns, with the indicated protein/empty vector present in each case. Data are available in S4 Data. ssDNA, single-stranded DNA. https://doi.org/10.1371/journal.pbio.3001988.g005 Rsm proteins have been seen to homodimerize and are regularly found as homodimers within the cell [27,28,62]. With the exception of RsmN from P. aeruginosa, they generally have a conserved tertiary structure [28]. Furthermore, an AlphaFold [63] model of RsmQ was shown to be highly similar to the crystal structures of SBW25 host Rsm proteins (Fig 5B) [62]. We hypothesised therefore that RsmQ may also interfere with regulation by forming heterodimers with host Rsm proteins. To test this, we expressed rsmQ and the SBW25 host rsm genes heterologously in E. coli using the Bacterial Adenylate Cyclase-Based Two-Hybrid (BACTH) system. The BATCH system allows for the screening of interacting protein partners through a cAMP cascade that is activated by bringing together the 2 adenylate cyclase domains fused to the proteins of interest [64]. Interestingly, with the exception of RsmI, we saw evidence of homo- and heterodimerisation within and between the Rsm proteins. Both RsmE and RsmQ homodimerised, and heterodimerisation was observed between all pairwise combinations of RsmA, RsmE, and RsmQ (Fig 5C and 5D). These results therefore support 2 indirect mechanisms for RsmQ regulation of the SBW25 proteome in addition to direct mRNA binding: either by sequestering ncRNAs or directly interfering with the activity of host Rsm regulators. As well as direct interaction, we considered that RsmQ may be interfering with the regulation of other chromosomally encoded Rsm proteins. To determine this, we examined the expression of each of the rsm genes at 3 points within the growth cycle (Mid-log, Late-log, and Early-stationary). Interestingly, although each rsm gene appears to be expressed at a different point within the cycle, there was no effect of plasmid carriage or rsmQ on expression levels (S5 Fig).

RsmQ causes phenotypic changes in SBW25 Given the large-scale changes that RsmQ caused to the SBW25 proteome, we hypothesised that these altered protein abundances would in turn affect bacterial behaviour. To test this, we quantified the impact of RsmQ on ecologically important traits normally controlled by the Gac-Rsm regulatory system. Specifically, we initially quantified swarming motility and production of exopolysaccharide/adhesin (measured using an indirect Congo red binding assay [65]) by SBW25 in the presence and absence of rsmQ. To examine the direct impact of rsmQ on chromosomally encoded genes, rsmQ was expressed under an inducible promoter on a multicopy plasmid, in the absence of pQBR103Km. Overexpression of rsmQ led to a complete loss of swarming motility and a significant increase in Congo red binding (Fig 6A and 6C). This suggests that rsmQ shifts SBW25 towards a more sessile lifestyle as characterised by reduced flagellar motility and increased production of attachment factors and/or extracellular polysaccharides associated with biofilm formation. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 6. Motility and biofilm formation are impacted by RsmQ. (a) The 48-h swarming motility assays for SBW25 containing pME6032 +/- rsmQ. (b) The 72-h swarming motility assays for SBW25 cells either plasmid free (-) or carrying pQBR103Km (+) or pQBR103Km-ΔrsmQ grown on 0.5% agar plates with media as indicated. Quantification of swarming of 4 biological replicates in triplicate can be found in S5 Fig. (c) Congo red absorbance (A 490 ) of SBW25 strains after 48 h (light blue, pink bars) or 72 h (green, red, dark blue bars). Independently performed one-way ANOVA results show statistically significant differences for both overexpression (p < 0.0001) and deletion (p < 0.0001). Statistical significance from Tukey’s multiple comparisons is indicated (p < 0.0001, ****). Data are available in S5 Data. WT, wild-type. https://doi.org/10.1371/journal.pbio.3001988.g006 To test if rsmQ had similar effects on SBW25 behaviour when encoded on pQBR103, we repeated the experiments using SBW25 with or without pQBR103Km ±rsmQ. Acquisition of pQBR103Km caused reduced swarming motility and Congo red binding relative to plasmid-free SBW25. However, deletion of rsmQ only partially ameliorated the reduction in swarming motility (Figs 6B and S6) and had no effect on Congo red binding (Fig 6C). The expression of Rsm proteins is normally tightly controlled by the cell; these results suggest that at high concentrations RsmQ can override the native cellular control to cause drastic phenotypic changes that are not observed at the native level. This is also consistent with our proteomic data (Fig 3B), which showed little or no impact of RsmQ on the abundance of structural biofilm or motility proteins such as flagella and adhesins, but a significant impact on chemotaxis pathways. We hypothesised that the role of RsmQ may be in the perception and uptake of specific nutrients, and therefore any phenotypic changes may be carbon source dependent. To test this, we examined the effect of the nutrient environment on swarming motility phenotypes and observed that pQBR103Km carriage strongly effected swarming motility on poorer carbon sources (S7 Fig), with the loss of rsmQ leading to a small restoration of swarming on some C-sources but not others, again suggesting rsmQ is involved in manipulating the cellular perception of the environment. To examine the impact of the wider Gac-Rsm pathway on the plasmid-associated biofilm phenotype, Congo red binding was measured for plasmid-free rsm mutants, as well as those carrying pQBR103Km ±rsmQ (S8 Fig). Interestingly, the plasmid mediated increase in Congo red binding was maintained across all Δrsm backgrounds, suggesting that no single Rsm protein is responsible for plasmid-mediated biofilm formation nor is RsmQ directly involved. Interestingly, however, we saw no significant difference in biofilm formation between ΔgacS and ΔgacS +pQBR103Km ±ΔrsmQ suggesting that the plasmid biofilm phenotype proceeds via Gac-Rsm system dysregulation, potentially mediated by other plasmid-encoded regulatory proteins.

RsmQ influences plasmid conjugation rate in poor nutrient conditions Next, we tested whether the phenotypes influenced by rsmQ had an impact on the ability of SBW25 to colonise the plant rhizosphere. The rhizosphere is a complex, heterogenous environment, with diverse nutrient sources and physical habitats. Successful colonisation involves the coordinated deployment of several distinct ecological traits. A series of competitive wheat colonisation assays were performed between plasmid-free SBW25 cells and either SBW25 plasmid free, SBW25 +pQBR103Km or SBW25 +pQBR103Km-ΔrsmQ. After 7 days, significantly fewer colony-forming units were recovered for both plasmid carrying strains compared with the plasmid free competitor (Fig 7A). However, no significant difference was seen between strains with or without rsmQ. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. RsmQ exerts a conditional effect on plasmid conjugation rate. (a) Competitive rhizosphere colonisation assay comparing the fitness of plasmid-free SBW25 with SBW25 (-, green), carrying either pQBR103Km (WT, red) or pQBR103Km-ΔrsmQ (ΔrsmQ, blue). Mann–Whitney U tests comparing plasmid-carrying cells to plasmid free with significance indicated (p < 0.001 ***). (b) Relative fitness on solid agar of plasmid-free SBW25 with SBW25 carrying either pQBR103Km (red) or pQBR103Km-ΔrsmQ (blue) on KB, M9 GC, or M9 Pyr. Significant differences were observed between plasmid carriers and plasmid-free strains in KB and M9 GC media (indicated above as follows: p < 0.05*, p < 0.001**, p < 0.0001***, p < 0.00001****) when compared using a one-way ANOVA with Tukey’s multiple comparisons; however, no differences were observed when comparing SBW25 +pQBR103Km ±ΔrsmQ. (c) Conjugation rate for SBW25 carrying either pQBR103Km (red) or pQBR103Km-ΔrsmQ (blue) on KB, M9 GC, or M9 Pyr. Two-way ANOVA analysis with Sidak’s multiple comparisons showed both a significant media (p < 0.0001) and plasmid (p < 0.01) effect and a significant difference between pQBR103Km and pQBR103Km-ΔrsmQ on M9 Pyr (p < 0.01**). All assays were performed with 10 biological and 2 technical replicates. Data containing a conjugation rate count of 0 were excluded from the dataset. Data are available in S6 Data. https://doi.org/10.1371/journal.pbio.3001988.g007 Rsm proteins were first characterised by their involvement in carbon storage and metabolism (e.g., CsrA in E. coli) and the regulation of secondary metabolism [66,67]. To dissect the ecological relevance of RsmQ more closely, we examined bacterial fitness and conjugation rates in liquid media and solid agar, for strains grown in KB, M9 glucose + casamino acids (M9 GC) and M9 pyruvate (M9 Pyr). No significant differences were observed in either fitness or conjugation rate in liquid culture (S9 Fig). Conversely, we saw several key differences between strains grown on a solid surface. Plasmid carriage incurred an rsmQ-independent fitness cost on both nutrient-rich agars (KB and M9 GC) but not on nutrient-poor M9 Pyr (Figs 7B and S10) compared to a plasmid-free background. Furthermore, we saw a major increase in plasmid conjugation rate for cells grown on M9 Pyr compared to the rich nutrient sources (Fig 7C). Strikingly, conjugation rate was conditionally dependent on rsmQ, with a significantly (p < 0.05) lower rate seen for pQBR103Km–ΔrsmQ compared to pQBR103Km (Fig 7C) on M9 Pyr.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001988

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/