While investigating surface-associated motility in rsmA mutants of P. aeruginosa PAO1, clear differences in swarming motility between sublines originating from different laboratories were observed. Consistent with previous studies, swarming was abolished entirely in a ΔrsmA mutant made in the PAO1-N subline from Nottingham⁴⁶, whereas the same deletion made in the PAO1-L subline from Lausanne had only a modest effect on motility (Fig. 1). Moreover, swarming motility was reduced in the wild type (WT) background of PAO1-N with respect to PAO1-L WT (Fig. 1), suggesting the existence of distinct genetic element(s) that contribute to this differential behaviour.Fig. 1

Representative images of swarming motilities displayed by P. aeruginosa PAO1-L and PAO1-N WT strains and their rsmA mutant derivatives. Swarming plates were incubated at 37 °C for 16 h.

To discover component(s) within the PAO1-L genome able to restore swarming in the PAO1-N ΔrsmA mutant other than rsmA itself, random Sau3AI chromosomal DNA fragments from PAO1-L (2–4 kb) were cloned into pME6000, transformed into the ΔrsmA mutant of PAO1-N and the resulting clones (~4,000, 92% genomic coverage) screened for restoration of swarming motility. Thirteen clones partially complemented swarming motility in the PAO1-N ΔrsmA mutant (Supplementary Fig. 1), suggesting that their chromosomal DNA inserts incorporated genes absent or downregulated in PAO1-N. To isolate the genetic elements able to complement swarming when overexpressed, ORFs in the selected inserts were individually sub-cloned and seven genes were identified. Amongst these, several genes encoding hypothetical proteins (PA0285, PA2072, PA2567) were found. In addition, PA3825 and proE coding for c-di-GMP-specific phosphodiesterases^47,48 were isolated. Other genes which restored swarming motility in the PAO1-N ΔrsmA mutant were rsmN encoding the small RNA-binding protein RsmN³, which was present in three of the thirteen clones indicating a good level of coverage of the PAO1-L genome, and unexpectedly toxR, a gene coding for the regulator of ETA production ToxR. Interestingly, six of the genes found to restore swarming in PAO1-N ΔrsmA mutant showed conserved domains from the family of proteins involved in c-di-GMP metabolism and hence would be expected to have an impact on swarming through the modulation of c-di-GMP levels (Table 1 and Supplementary Fig. 1).Table 1

Genetic elements with conserved domains from the family of proteins involved in c-di-GMP metabolism that restored swarming motility in PAO1-N ΔrsmA mutant.

To determine which of the gene(s) identified was involved in the altered swarming motility of PAO1-N with respect to PAO1-L, the extent of the genetic differences between sublines was investigated using whole-genome sequencing, and the assembled chromosomes of both sublines were compared with the reference genome of PAO1-UW⁴⁹. Candidate single nucleotide polymorphisms (SNPs) as well as small insertions/deletions (INDELs) between sublines were then selected (Supplementary Table 1). Additionally, optical restriction mapping was performed to detect large scale chromosomal rearrangements, deletions and insertions^50,51. Although none of the previously identified genes was affected by the SNPs identified when comparing genomes from both sublines, sequencing analysis revealed a large deletion (~59 kb) in the PAO1-N genome that included the toxR gene (Fig. 2 and Supplementary Table 2). These results strongly suggested that ToxR might play a role in the control of swarming motility. Optical restriction mapping revealed that neither PAO1-N nor PAO1-L sublines have the large 2-Mb rrnA/rrnB chromosomal inversion reported for PAO1-UW⁴⁹ (Supplementary Fig. 2), and sequencing results indicated that both carry the 12-kb RGP42 island initially found in PAO1-DSM⁵², excluding these as possible differences between the sublines.Fig. 2

a Genetic organization of the ~59 kb deletion in P. aeruginosa PAO1-N subline. b The deletion spans from the 308^th bp of the 16 S rRNA gene PA4280.5 until the 2nd bp before the start codon of toxR.

To investigate the role of ToxR in swarming, the impact of toxR deletion and the double deletion ΔrsmA ΔtoxR on PAO1-L swarming motility was assessed. Although deletion of toxR resulted in a slight reduction in swarming motility, supporting a link with this phenotype, no swarming deficiency was recorded for the ΔrsmA ΔtoxR double mutant in PAO1-L in contrast to the PAO1-N ΔrsmA strain (ΔrsmA and ~59-kb deletion including toxR) (Fig. 3). We therefore hypothesised that another genetic element missing in PAO1-N could be responsible for the residual swarming activity in the PAO1-L ΔrsmA ΔtoxR mutant. Hence, alternative genomic modifications explaining this divergence were explored. Among the thirteen distinct SNPs/INDELs found when comparing the two PAO1 sublines, a non-synonymous point mutation was predicted in a PAO1-N gene encoding the c-di-GMP degrading PDE BifA (Supplementary Table 1). This mutation results in tyrosine to aspartate substitution at amino acid 442 (Y442D), a position immediately preceding a glutamine (Q443) involved in substrate binding^53,54. This suggested that the Y442D substitution in PAO1-N might reduce BifA affinity for its substrate by introducing a negative charge next to the glutamine involved in c-di-GMP binding.Fig. 3

Swarming motilities of PAO1-L a and PAO1-N b WT strains and their corresponding rsmA, toxR and bifA mutants, assayed over a period of 16 h at 37 °C.

Interestingly, a PAO1 bifA mutant was previously shown to have increased intracellular c-di-GMP levels leading to a non-swarming phenotype^22,55,56. Thus, to determine whether defective BifA activity contributed to the swarming motility defect in PAO1-N ΔrsmA and played a role in the differential swarming behaviour between sublines, the motility of strains generated by introducing bifA^Y442D or bifA^D442Y point mutations in PAO1-L WT, PAO1-L ΔrsmA ΔtoxR and PAO1-N ΔrsmA were also studied. As shown in Fig. 3, the triple mutation ΔrsmA ΔtoxR bifA^Y442D resulted in swarming impairment in PAO1-L. At the same time, the reversion of Y442D to D442Y restored swarming motility in PAO1-N ΔrsmA, suggesting that bifA is the third genetic element involved in this phenotype which potentially explains motility differences between the two PAO1 sublines.

Given that the alternative sigma factor PvdS is required for optimal toxR expression⁵⁷ and that pvdS transcript levels were shown to be affected by RsmA^4,58, we sought to determine whether RsmA regulates ToxR activity partly through the control of pvdS expression therefore establishing a link between both regulators and swarming motility. To this end, transcriptional fusions of the lux operon to the toxR promoter region were constructed using the mini-CTX system⁵⁹ and introduced into the PAO1-L WT and the IPTG-inducible rsmA_ind, ΔpvdS and ΔpvdS rsmA_ind mutant chromosomes. Moreover, since toxR can be transcribed from two promoters (P1 and P2⁶⁰) we investigated the effect of a rsmA mutation on expression from the complete intergenic region upstream of toxR (P_toxR1,2-luxCDABE) as well as from each promoter separately (P_toxR1/P_toxR2 - luxCDABE) (Supplementary Fig. 3A). Additionally, since the transcription of pvdS is upregulated under iron-limiting conditions⁶¹, the toxR reporter strains were tested under iron deficiency using CAA medium.

As shown in Supplementary Fig. 3B, the non-induced PAO1-L rsmA_ind strain exhibited lower toxR expression levels than WT. Conversely, IPTG-induced expression of rsmA in PAO1-L rsmA_ind stimulated toxR expression to levels matching or above those of the WT, suggesting a positive impact of RsmA on ToxR production. Results from assays including ΔpvdS and ΔpvdS rsmA_ind strains showed a significant reduction of P2 promoter activity in the ΔpvdS mutant compared with WT strain. In agreement with previous studies, P2-driven expression was upregulated under low iron conditions⁵⁷ (Supplementary Fig. 3C). In addition, in the absence of PvdS, RsmA could no longer activate toxR expression, suggesting that the RsmA control on toxR expression is at the transcriptional level via PvdS (Supplementary Fig. 3C).

Previous studies have shown that iron depletion stimulates surface motility in P. aeruginosa^62,63. Moreover, rhamnolipid biosynthesis is also upregulated under low iron conditions and is under the control of RhlR^30,35,64. The correlation between surface motility, iron regulation and the Rhl QS system led us to hypothesise that the mechanism by which ToxR impacts on swarming could be linked to rhl expression and hence rhamnolipid production. To investigate this, we first compared expression of the rhamnosyltransferase gene rhlA using a PrhlA chromosomal transcriptional fusion to the lux operon in the PAO1-L WT and the toxR conditional IPTG-inducible mutant strain toxR_ind using the integrative vector mini-CTX. A significant reduction in rhlA transcription in the toxR mutant was observed with respect to the WT strain (Fig. 4a). Conversely, the IPTG-induced expression of toxR restored the expression of rhlA to WT levels (Fig. 4a), indicating that ToxR has a positive impact on rhlA expression. To determine whether the reduced transcription of rhlA observed PAO1-L toxR_ind resulted in decreased rhamnolipid production, extraction and quantification of these biosurfactants were performed in both strains. Results showed that rhamnolipid yields were reduced in the toxR_ind strain when compared to WT levels and addition of IPTG stimulated rhamnolipid production (Fig. 4b), confirming the positive impact of ToxR on rhamnolipid biosynthesis.Fig. 4

In all these experiments gene expression and phenotypic analysis were performed in the PAO1-L WT and toxR_ind mutant strains with and without 1 mM IPTG induction of toxR. a Transcriptional activity of the rhlA promoter assessed at 37 °C in LB by measuring the luminescence emitted by cells containing a P_rhlA - luxCDABE transcriptional reporter integrated in the chromosome using the mini-CTX system. Values correspond to the area under the curve (AUC) derived from plotting relative light units normalized to culture density (RLU/OD₆₀₀) over time (24 h). b Rhamnolipid production measured using the Orcinol method⁸⁶ from the supernatant fraction of the three strains tested grown in LB at 37 °C for 24 h. c Effect of toxR mutation on the transcriptional activity of the rhlI promoter in LB at 37 °C for 24 h. Values correspond to the area under the curve (AUC) derived from plotting relative light units normalized to culture density (RLU/OD₆₀₀) over time (24 h). d C4-HSL UV absorbance values (214 nm) normalised to PAO1-L WT and toxR_ind mutant growth (OD₆₀₀) in LB at 37 °C for 6 h. C4-HSL was extracted from culture supernatant fractions using acidified ethyl acetate and analysed by LC-MS. Values given are averages from three different cultures ± standard deviation. Statistical differences between group means were determined by one-way ANOVA analysis using Tukey’s multiple comparisons test. (*p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.0001).

To establish whether ToxR modulates rhamnolipid production via the activation of the Rhl QS system, the effect of the toxR mutation on the expression of rhlI, coding for the C4-HSL autoinducer synthase RhlI, was assessed using a mini-CTXlux transcriptional fusion of PrhlI in PAO1-L toxR_ind. In the absence of IPTG (ToxR negative), toxR_ind showed ~25% reduction in rhlI expression compared with the WT. Conversely, the addition of IPTG led to an increase of ~25% in rhlI expression over the WT levels (Fig. 4c), suggesting a positive effect of ToxR on rhlI expression under the conditions tested. We then measured the levels of C4-HSL produced by the WT and toxR_ind strains using LC-MS. The toxR_ind strain showed reduced levels of C4-HSL and the addition of IPTG restored the levels of this molecule close to those of the WT strain (Fig. 4d), further supporting an impact of ToxR on the Rhl QS system. These results provide further insights into the mechanism by which ToxR affects swarming motility in PAO1 through the activation of rhamnolipid production via the modulation of the Rhl QS system.

Swarming motility and biofilm formation are intimately interconnected and described as inversely regulated in P. aeruginosa^21–23. Given that ToxR modulates swarming motility in PAO1, we monitored biofilm formation by the toxR conditional mutant toxR_ind and PAO1-L WT strains using Bioflux microfluidic chambers to assess whether ToxR could also affect this phenotype. Our results showed that biofilm growth was enhanced in the absence of toxR expression, while the IPTG-induced transcription of this gene decreased biofilm biomass compared to the parental PAO1-L WT strain (Fig. 5a, b), suggesting that ToxR has a negative impact on biofilm formation. To study whether changes in exopolysaccharide (EPS) biosynthetic genes expression mediated the effect of ToxR on biofilm formation, the activity of the EPS Psl and Pel biosynthetic operon promoters was determined by measuring the fluorescence from cells containing P_psl-dsRed and P_pel-dsRed reporter fusions, respectively. Results showed that the inactivation of toxR significantly increased psl and pel gene expression especially for the pel promoter fusion. Conversely, toxR complementation through the addition of IPTG reduced the expression of both EPS biosynthetic gene clusters (Fig. 5c, d). These data suggest that ToxR partially represses biofilm formation through a negative impact on EPS production.Fig. 5

a Confocal images of mClover3-labelled P. aeruginosa PAO1-L WT and toxR_ind mutant ±1 mM IPTG biofilms obtained after 48 h incubation in microfluidic channels (Bioflux, Fluxion) in RPMI-1640 medium. Scale bar: 100 µM. b Quantification of biomass from confocal images of PAO1-L WT and toxR_ind mutant (±1 mM IPTG) biofilm cultures using Comstat2⁸⁹. c and d Effect of toxR mutation on EPS biosynthetic gene transcription. Expression was determined by measuring the fluorescence from wild type PAO1-L and the toxR inducible mutant toxR_ind cells (±1 mM IPTG) carrying the transcriptional fusions (c) P_psl-dsRed and (d) P_pel - dsRed in their chromosomes and grown at 37 C for 24 h in LB. Reported values are averages from three different cultures ± SD and correspond to the area under the curve (AUC) derived from plotting relative fluorescent units normalized to culture density (RLU/OD₆₀₀) over time. Statistical differences between group means were determined by one-way ANOVA analysis using Tukey’s multiple comparisons test. (*p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.0001).

Our results show that ToxR, in addition to regulating the production of exotoxin A, influences swarming and biofilm formation in P. aeruginosa. Given that both phenotypes are known to be modulated by c-di-GMP signalling, we hypothesised that ToxR might influence c-di-GMP levels impacting those traits. To test this, the intracellular levels of c-di-GMP were semi-quantified in cells transitioning from motility to sessile lifestyles by monitoring the activity of the transcriptional fusion P_cdrA-gfp(ASV)^C, which responds to the levels of this signal⁶⁵, in the PAO1-L WT and ΔtoxR mutant strain. As shown in Fig. 6 (and Supplementary Fig. 4), the toxR mutant displayed high levels of cdrA promoter activity, well above that of the WT, consistent with elevated c-di-GMP levels and supporting the hypothesis that ToxR impacts on motility/biofilm formation by reducing c-di-GMP signalling.Fig. 6

a Fluorescence output of PcdrA gfp-expressing PAO1-L and ΔtoxR cells as they transition from liquid culture to glass surface over time (3 h). Scale bar: 20 µm. b Representative combined Differential Interference Contrast (DIC) and widefield epifluorescence images of PAO1-L and ΔtoxR after 3 h exposure to a glass surface. Scale bar: 27 µm. c Graph shows the fraction of cells expressing P_cdrA-gfp(ASV)^C over time for the first 100 min after attachment to a glass surface. Dots represent mean ±1 SEM (n = 3).

An EAL domain was previously identified in ToxR (Table 1) however, given the absence of conserved residues critical for PDE activity in its sequence (Supplementary Fig. 5), this regulator was assumed to lack c-di-GMP catalytic activity⁶⁶. To verify that the degenerate EAL domain in ToxR is an enzymatically inactive PDE sequence, a 6×His-fusion ToxR protein was expressed, purified and tested for c-di-GMP degradation activity by LC-MS analysis. No c-di-GMP cleavage was detected, indicating that ToxR does not negatively regulate c-di-GMP signalling through degradation of this second messenger. To test whether c-di-GMP is a ToxR ligand, a Surface Plasmon Resonance (SPR) binding assay using purified ToxR protein was performed as previously described⁶⁷. As shown in Fig. 7, we were able to detect reproducible concentration-dependent binding of c-di-GMP by ToxR, with a calculated K_D of 2.36 ± 0.7 μΜ. This indicates that ToxR is a c-di-GMP-binding protein and suggests that ToxR may impact on swarming and biofilm formation through direct stimulation of additional enzymes able to metabolise this signalling molecule explaining the altered c-di-GMP levels in the toxR mutant.Fig. 7

Purified ToxR was tested for dinucleotide binding by Surface Plasmon Resonance (SPR) as previously described⁶⁷. a SPR sensorgrams showing affinity measurements of an increasing range of ToxR concentrations (0–5 µM) binding to biotinylated c-di-GMP. Protein binding and dissociation phases are shown. b Affinity fit for ToxR-c-di-GMP binding. The binding response for each concentration was recorded, and the K_D values for ToxR binding to c-di-GMP (2.36 ± 0,7 µM) were calculated using Biacore T200 BiaEvaluation software version 1.0 (GE Healthcare) and confirmed by GraphPad Prism. The experiment was repeated three times independently.

Strains, plasmids and oligonucleotides used in this study are listed in Supplementary Table 3. E. coli and P. aeruginosa strains were routinely grown at 37 °C on lysogeny broth (LB) or LB agar supplemented with antibiotics as required. For studying toxR expression, toxR reporter strains were grown under iron deficiency using CAA medium⁷⁹. IPTG was added to cultures at 1 mM final concentration to complement inducible mutant strains. Swarming assays were performed in 60 ×15 mm dishes with 8 g/L Nutrient Broth No.2 (Oxoid), 0.5% w/v Bacto agar (Difco) and 0.5% w/v D-glucose (Sigma). Plates were inoculated with 3 µL of OD₆₀₀ 1.0 ON cultures washed in fresh LB. Plates were then incubated for 16 h at 37 °C.

Biofilms were grown continuously in microfluidic channels using a Bioflux system (Fluxion). Briefly, mClover3-labelled⁸⁰P. aeruginosa PAO1-L WT and toxR_ind mutant strains were grown at 37 °C for 16 h in 2 mL Lysogeny broth (LB) broth. The cultures were diluted to OD₆₀₀ 0.05 in RPMI-1460 (Lonza) and used to seed the microfluidic channels. The biofilm was allowed to form at a flow rate of 16 μL/h (corresponding to a shear rate of 0.5 dyn/cm²) at a temperature of 37 °C for a period of 48 h in RPMI-1640 (Lonza Bioscience).

To find genetic elements in PAO1-L restoring swarming motility in P. aeruginosa PAO1-N ΔrsmA, a genomic library was generated by partially digesting the PAO1-L chromosome with Sau3AI and 2–4 kb fragments cloned into pME6000 plasmid. The resulting plasmids were transformed into P. aeruginosa PAO1-N ΔrsmA and each clone screened for restoration of swarming⁴⁶.

To catalogue single nucleotide polymorphisms (SNPs) and small insertions/deletions (INDELs) in PAO1-L and PAO1-N sublines, paired-end sequencing libraries were prepared using the Illumina Nextera kit according to the manufacturer’s protocols (Illumina), and whole-genome sequencing was performed on an Illumina MiSeq platform. Whole-genome sequence datasets were subjected to adaptor trimming using Trimmomatic version 0.30⁸⁰ and hard trimming at the 5′ end using a custom Perl script. Trimmed reads were aligned to the modified PAO1-UW sequence³ using Smalt version 0.7.0.1 (sanger.ac.uk/tool/smalt-0/) with a kmer of 12 and a step size of 2. The alignment files were sorted, and PCR duplicates were removed using Samtools⁸¹. INDEL realignment was carried out with the Genome Analysis Toolkit⁸², and SNPs and INDELs were determined using Samtools mpileup, employing varying base quality thresholds (parameter-Q). SNPs were filtered to exclude those with low coverage (<5x) and those not supported by at least two reads on each strand using a custom Perl script. A custom Perl script was used to characterise SNPs in genes, and the resulting amino acid changes. Regions with zero read coverage were detected using the “genomeCoverageBed” functionality of bedtools with the -bga option (version 2.17.0)⁸³. De novo assembly was carried out using Velvet version 1.2. with a kmer of 121 with no scaffolding and the expected coverage flag set to auto.

Large scale chromosomal rearrangements, deletions and insertions in sublines PAO1-L and PAO1-N were analysed by optical restriction mapping^50,51. Briefly, high molecular weight (HMW) DNA was isolated from single colony cultures of P. aeruginosa PAO1-L and PAO1-N using the HMW DNA Isolation Kit (Opgen Inc.) according to the manufacturer’s instructions. The quality of HMW DNA was assessed using an Argus QCard Kit (Opgen Inc.), and the concentration of the HMW DNA was adjusted using dilution buffer. DNA molecules were stretched and linearised on charged glass coverslips using a high-density channel-forming device (CFD). The restriction endonuclease BamHI was used in an Argus MapCard (Opgen Inc), and the glass coverslips were incubated in a MapCard Processor to allow digestion of the DNA molecules on the glass surface. The fragmented DNA was then stained with intercalating fluorescent dye, and the DNA molecules were imaged under a high-resolution microscope. DNA molecules were assembled into a whole-genome map using the OpGen Genome-Builder software. Optical maps were aligned to the reference P. aeruginosa subline PAO1-UW⁴⁹ (GenBank accession NC_002516) using the OpGen MapSolver software and visually inspected to locate regions of difference >2 kb, the lower limit for efficient optical mapping.

In-frame deletion mutants were obtained by two-step allelic exchange using pDM4 or pME3087 vectors. To construct P. aeruginosa PAO1-L ΔrsmA, ΔtoxR and ΔpvdS mutants, two PCR products amplifying each gene’s upstream and downstream nucleotide regions were generated using the primer pairs D1FW/D1RV and D2FW/D2RV, respectively (Supplementary Table 3). PCR products were ligated together in pBluescript II KS⁺ to create a deletion in the corresponding gene and subsequently cloned into the suicide vector pDM4 or pME3087 resulting in plasmids pZH13, pJD22 and pJD93, respectively. Following transformation into the target strain by conjugation, single crossovers were selected on chloramphenicol (Cm, 375 μg mL⁻¹) or tetracycline (Tc, 125 μg mL⁻¹). To achieve rsmA and toxR deletion, recombinants resistant to Cm were grown in LB supplemented with 10% sucrose overnight and plated on LB-10% sucrose. The carbenicillin enrichment method selected the pvdS double-crossover mutants⁸⁴. The resulting P. aeruginosa colonies were screened for the loss of antibiotic resistance by plating on LB supplemented with or without Cm/Tc. PCR and sequence analysis confirmed the in-frame deletions.

The conditional mutant IPTG-inducible toxR (toxR_ind) strain was constructed by introducing the lacI^Q repressor gene and the tac promoter transcribing the lacZ 5′ untranslated region and its ribosome binding site (RBS) directly upstream of the toxR open reading frame, resulting in constitutive transcription and translation only in the presence of IPTG. The suicide plasmid pAMP5 was constructed in several stages and is a pDM4 derivative that carries, in sequential order, the following elements: (a) a 365 bp HindIII-EcoRI fragment of the upstream region of toxR obtained by PCR using primers toxRpFW/RV (Supplementary Table 3); (b) the 2.0 kb BamHI Sm/Spc integron from pHP45Ω; (c) the 1.5 kb BamHI-EcoRI lacI^Q Ptac inducible promoter fragment of pME6032 and (d) an 0.5 kb EcoRI-XhoI fragment carrying the toxR open reading frame (obtained from pBluescript II KS⁺-based vector constructed for the ΔtoxR mutant). The mutant was obtained by double crossover of the fragment carried by pAMP5 in P. aeruginosa PAO1-L and selection of sucrose and streptomycin-resistant clones.

To generate the bifA^Y442D mutant in PAO1-L and derivative mutants as well as to revert the bifA point mutation in PAO1-N, the bifA gene was amplified from PAO1-N and PAO1-L genomes using the primer pair BifAFW/BifARS (Supplementary Table 3). The resulting PCR products were ligated with pBlueScript II KS⁺ and subsequently cloned as HindIII-EcoRI fragments into the suicide plasmid pME3087, generating the plasmids pJL103 and pJL108 and used for double crossover recombination. The resulting P. aeruginosa colonies were screened for the loss of antibiotic resistance by plating on LB supplemented with or without Tc. PCR and sequence analysis confirmed the mutations.

The promoter regions of toxR, rhlA and rhlI were amplified from the PAO1-L genome using the primer pairs ToxRp1FW/RV, ToxRp2FW/RV, ToxRp12FW/RV, RhlApFW/RV and RhlIpFW/RV (Supplementary Table 3). The resulting PCR products were ligated into the mini-CTXlux vector using the appropriate restriction enzyme to generate the transcriptional reporters pAMP11 (mini-CTX::P_toxR1 - luxCDABE), pAMP12 (mini-CTX::P_toxR2 - luxCDABE), pAMP13 (mini-CTX::P_toxR1,2 - luxCDABE), pSH30 (mini-CTX::P_rhlA - luxCDABE) and pSH32 (mini-CTX::P_rhlI - luxCDABE). Finally, the transcriptional reporters were integrated into the chromosome of P. aeruginosa PAO1 strains by conjugation with E. coli S17.1 λpir followed by selection with Tc.

For P_pel - dsRed and P_psl-dsRed transcriptional reporter construction, the promoter regions of pelA and pslA genes were amplified by PCR from P. aeruginosa PAO1-L chromosomal DNA using primer pairs PpelAdsRedFW/RV and PpslAdsRedFW/RV, respectively. The reporter dsRed-express2 gene was amplified using primer pairs dsRedFW/RW from pCMVDsRed-Express2 vector and cloned into pBluescript II KS⁺, resulting in plasmid pSC581. Then dsRed-express2 cut from pSC581 with HindIII/EcoRI, and the KpnI/HindIII cut pelA or pslA promoter regions were ligated into pUCP18, resulting in plasmids pSC855 and pSC856.

Clones with active fusions were selected and analysed for bioluminescence or fluorescence output activity over growth in CAA or LB media at 37 °C using a 96-well plate TECAN Genios Pro multifunction microplate reader.

P. aeruginosa strains were grown for 6 h (early stationary phase) in LB broth at 37 °C with vigorous shaking (200 rpm). QS molecules were extracted from 5 mL culture by adding 5 mL of acidified ethyl acetate⁸⁵. Sample separation was performed on an Agilent 1200 series HPLC with an Ascentis Express C18 150 ×2.1 mm internal diameter, 2.7 mm particle size, maintained at 50 °C. The mobile phase consisted of formic acid 0.1% (v/v) in water and formic acid 0.1% (v/v) in acetonitrile run as a gradient over 20 min at a flow rate of 0.3 ml min⁻¹. C4-HSL identification was performed using a Bruker HCT Plus ion trap LC-mass spectrometer and Hystar software (Bruker). Retention times and MS/MS peak spectra were matched to 10 mM synthetic C4-HSL standard. For rhamnolipids quantification strains were grown at 37 °C in 50-ml Erlenmeyer flasks containing 10 ml of M9 medium supplemented with glycerol (2% vol/vol), glutamate (0.05%), and Triton X-100 (0.05%) with shaking for 18 h. Rhamnolipids were extracted with three volumes of diethyl ether from culture supernatants filtered through a 0.22-μm-pore-size membrane. Extracts were extracted once with 20 mM HCl, and the ether phase was evaporated to dryness. The residue was dissolved in water. Rhamnose content in each sample was determined by duplicate oricinol assays compared to rhamnose standards. Rhamnolipid was determined by the relation that 1.0 mg of rhamnose corresponds to 2.5 mg of rhamnolipid^11,86.

The pBAD TOPO TA expression system (Invitrogen) was used to produce His-tagged ToxR protein within host E. coli BL21 (DE3) cells. The toxR gene was amplified from PAO1-L genome using primers ToxRtopoFW/RV and cloned into pBAD TOPO vector according to manufacturer’s instructions. E. coli BL21 (DE3) cells carrying the resulting pAMP14 vector were cultured overnight in LB at 37 °C with vigorous shaking (200 rpm), diluted 1:100 in Terrific Broth (TB) and grown to OD₆₀₀ 0.4 at 37 °C before protein expression was induced overnight with 0.2% L-Arabinose at 18 °C. Cells were then lysed by sonication and centrifuged at 15,000 × g for 1 h. ToxR was purified from the supernatant by NTA-Ni chromatography. HiTrap chelating HP columns of 1 mL (GE healthcare, life sciences) were equilibrated with 10 volumes of washing buffer (20 mM HEPES pH 7.5, 250 mM NaCl, 2 mM MgCl₂, and 2.5% (v/v) glycerol pH 6.8) and loaded with cell lysate. Following protein immobilisation, the column was washed with 10 volumes of buffer containing 50 mM imidazole before the protein was eluted using 500 mM imidazole buffer in a single step elution. Due to ToxR instability, it was required that later assays were performed within a few hours after protein purification.

To assess cleavage of c-di-GMP by a possible PDE activity of ToxR, purified ToxR (0.7 mg) was incubated with c-di-GMP (100 μM) (Biolog Life Science Institute) for 2 h at 30 °C. The remaining c-di-GMP in the samples was detected by liquid chromatography-mass spectrometry (LC-MS) analysis using the TSQ Quantum Access Max system (Thermo Scientific) in positive-ion mode, as previously described⁸⁷. The areas of the major MS-MS fragments of c-di-GMP (m/z 152 and 540) were measured, and a standard curve was established using defined concentrations of c-di-GMP to allow sample quantification. Moreover, the main degradation product of c-di-GMP cleavage, 5′-phosphoguanylyl-(3′→5′)-guanosine (pGpG), was also quantified in the samples by measuring the area of the major fragment (m/z = 152) of the isolated precursor ion (m/z = 709) corresponding to pGpG as determined using a standard (Biolog).

To test whether c-di-GMP is in fact a ToxR ligand, a Surface Plasmon Resonance (SPR) binding assay using purified ToxR protein was performed⁶⁷. Briefly, a Biacore T200 system (GE Healthcare) equipped with a Streptavidin SA sensor chip (GE healthcare), containing four flow cells with a SA pre-immobilized to a carboxymethylated dextran matrix each, was used. Flow cell one (FC1) and flow cell three (FC3) were kept blank for reference subtraction. The chip was washed three times with 1 M NaCl and 50 mM NaOH to remove any unconjugated streptavidin. Biotinylated c-di-GMP (100 nM) (BioLog) was immobilised on FC2 and FC4 of the streptavidin chip at a 50 RU immobilisation level with a flow rate of 5 μL/min. Soluble ToxR protein was prepared in SPR buffer (10 mM HEPES, 500 mM NaCl, 0.1% (v/v) Tween 20, 2 mM MgCl₂, pH 6.8). Samples were injected with a flow rate of 5 μL/min over the reference and c-di-GMP cells for 60 s, followed by buffer flow for 60 s. The chip was washed at the end of each cycle with 1 M NaCl. An increasing range of protein concentrations (78.125 nM, 156.25 nM, 312.5 nM, 625 nM, 1.25 µM, 2.5 µM, 5.0 µM) was used, with replicates for each protein concentration included as appropriate. All sensorgrams were analysed using Biacore T200 BiaEvaluation software version 1.0 (GE Healthcare). The experiment was repeated three times independently.

P. aeruginosa PAO1-L WT and ΔtoxR mutant strains carrying the plasmid P_cdrA-gfp(ASV)^C 65 were used to report c-di-GMP levels in cells transitioning from liquid culture to a glass surface over time (3 h). For time-lapse imaging, a bespoke multimode microscope (Cairn Ltd)⁸⁸, based in an inverted Eclipse Ti microscope (Nikon), was used. The microscope was fitted with an environmental chamber (Okolab) to regulate temperature and relative humidity. Differential Interference Contrast (DIC) and widefield epifluorescence imaging were carried out using a single channel white MonoLED light source (Cairn Ltd) and a 40× objective (Nikon, CFI Plan Fluor 40×/1.3). Images were acquired using an Orca-Flash 4.0 digital CMOS camera (Hamamatsu) every 30 sec per hour. Experiments were conducted using a 35 mm glass-bottom dish (Cellvis). The fraction of cells expressing P_cdrA-gfp(ASV)^C was calculated by counting objects in each frame in both DIC and epifluorescence mode. Images were processed in ImageJ (NIH), and an automatic threshold was applied using the Yen method to each data set. Bacterial tracks were processed through a custom single-cell tracking algorithm⁸⁸ to determine the average fluorescence (F values) of tracks corresponding to P_cdrA-gfp(ASV)^C expression levels.

Biofilms expressing the fluorescent protein mClover3 were visualised under a Confocal Laser Scanning Microscope (LSM700, Carl Zeiss) using 488 nm laser. At least 5 replicate Z-stack biofilm images were taken randomly, and biomass was quantified using Comstat2 plugin⁸⁹ in ImageJ software.

One-way ANOVA analysis using Tukey’s posthoc multiple comparisons tests were applied to determine whether mutants’ responses differed significantly from that of the parental strain (p < 0.05) when compared with the variations within the replicates (n = 3) using GraphPad Prism 8.0 (GraphPad Software, Inc., San Diego, CA).