Bacterial strains and plasmids used are described in the Supplementary Table S1. P. aeruginosa was cultivated in Luria-Bertani [0.5% (w/v) yeast extract, 1% (w/v) tryptone, and 1% (w/v) sodium chloride] broth with a pH of 6.8, using 50 mM 3-(N-morpholino)propanesulfonic acid as a buffer (LB-Mops broth), PM broth with an addition of 1% sodium caseinate (casein broth) (Kostylev et al., 2019), or Pseudomonas P broth [0.14% (w/v) magnesium chloride, 2% (w/v) pancreatic digest of gelatin, and 1% (w/v) potassium sulfate]. Several types of agar were used for plating purposes, including LB agar, and skim milk agar (Sandoz et al., 2007). Typically, 100 μg/mL of gentamicin, 50 μg/mL of kanamycin, or 300 μg/mL of carboxybenzylmycin was added to the medium. Bacteria were subcultured in 14 mm Falcon tubes with fresh LB or LB-Mops broth and incubated at 37°C with shaking at 250 rpm. Escherichia coli used in the cloning process was grown in LB broth.

Our investigations utilized the P. aeruginosa PAO1 strain (Stover et al., 2000), which differs from the currently sequenced strain due to an 8 bp deletion at the mexT gene. Mutants with deletions of lasR, lasR^Q45stop, psdR, mexT, xcpA, and lasB were derived by cloning PAO1 using the suicide vector pEXG2, as previously described (Hmelo et al., 2015). The Supplementary Table S1, lists the primers used to construct the pEXG2 knockout plasmids. Reverse-transcriptase PCR of genomic DNA confirmed all deletion mutations. To compensate for the mutations, we integrated wild-type copies of genes from pUC18-mini-Tn7-Gm, as previously reported (Choi et al., 2005).

We measured promoter activity in cells cultured in LB-MOPS broth, as previously reported (Feltner et al., 2016), using lasI-gfp reporter strains in a 96-well plate. These were incubated at 37°C with continuous shaking in a Synergy H1 microplate reader (Biotec, America). GFP fluorescence (excitation at 485 nm, emission at 535 nm) and cell density (OD₆₀₀) were recorded at 2 h intervals for 28 h (refer to Supplementary Table S1 in the Supplementary material). For the detection of cyanide, we employed a cyanogenic test paper approach detailed previously (Chen et al., 2019). Pyocyanin activity was estimated using the modified protocol of O’Loughlin et al. (2013). The P. aeruginosa PAO1 strain was cultured overnight and subsequently diluted to an OD₆₀₀ of 0.01, followed by incubation in 3 mL of P broth medium for 24 h.

This study assessed swarming motility using a medium composed of nutritional broth [0.8% (w/v)] supplemented with D-glucose [0.5% (w/v)] and solidified with 0.5% (w/v) Bacto Agar. Overnight liquid broth cultures were adjusted to an optical density of 1.0 before inoculation onto plates. Growth was evaluated after 18–24 h of incubation at 37°C. After 24 h of growth in LB medium containing 1% solidified Bacto Agar at 37°C, the agar was gently detached, revealing a twitching motility zone adhered to the petri dish. A 10 min stain with 1% (wt/vol) crystal violet made the twitching zones visible, and their diameters were subsequently measured (Liu et al., 2022).

Quorum sensing signal extraction was carried out as described (Dong et al., 2008). The bacterium was cultivated in LB-Mops broth for 18 h, using 3 mL. Subsequently, an equal volume of acidified ethyl acetate (0.1% acetic acid) was combined with 3 mL of culture supernatant for signal extraction. Dried samples were reconstituted in 200 μL methanol and immediately filtered before LC-MS/MS analysis. We isolated and selected the prevalent signaling molecules for quantitative LC-MS/MS investigation. Modifications to the LC-MS procedure were made (Liu et al., 2022). HPLC was conducted using a Waters X Select HSS T3 1.8 μm phase column (2.1 × 100 mm) on the Dionex Ultimate 3000 system (Thermo Fisher Scientific). Metabolites identified included 3-oxo-C12-HSL (m/z 298.20128) and C4-HSL (m/z 172.09682). Data evaluation was executed with TraceFinder and Thermo Xcalibur, both from Thermo Fisher Scientific.

Bacteria grown on LB broth milk agar plates were incubated for 30 h at 37°C. The amount of extracellular protease produced was quantified by measuring the area of the proteolytic zone surrounding the bacterial colony.

Under these conditions, the gentamicin-resistance marker is neutral. We carried out competitive studies using P. aeruginosa strains both with and without this marker. The proportion of the competitor bearing the gentamicin-resistance marker was set at 1%, 10%, 50%, or 90% in casein broth for 48 h. To ascertain cell yields, plate counting, both with and without 10 μg/mL gentamicin, was employed to gauge proportional abundances after 48 h of incubation. The relative fitness was calculated by dividing the ratio of mutant to wild type after 48 h by the ratio at the time of inoculation.

For evolution experiments (Supplementary Figure S2C), log-phase LB-Mops cultures of both cooperator and competitor were prepared, adding the competitor to the casein broth at a ratio of 0.1 (total 100 μL). Two days post primary inoculation, once visible proliferation was observed, a 60 μL volume of the transfer inoculum was introduced daily into a 3 mL casein broth. Serial dilutions of isolates were grown on LB agar plates both with and without 10 μg/mL gentamicin to determine the percentage of protease-deficient cells. Cultures were sampled from LB agar plates containing 10 μg/mL gentamicin, and colonies were then streaked daily onto skim milk agar plates to detect protease-altering variations.

Bacteria were isolated from individual colonies and cultured in LB broth with a buffer for a duration of 24 h. Cells were collected for analysis via centrifugation. Bacteria were cultivated in LB-Mops broth, and genomic DNA was extracted using the Easy Pure Bacteria Genomic DNA Kit (TRAN). The genomic DNA of the four sublines (LasR^Q45stop-1, LasR^Q45stop-2, LasR^Q45stop-3, and LasR^Q45stop-4) was sequenced with the Illumina NovaSeq S4 PE-150 (Novogene, China). Whole Genomics Solution Corporation conducted the whole-genome sequencing and assembly. PCR testing was employed to validate the genomic sequencing results.

Results are depicted as mean ± SD. Experimental groups were evaluated using an unpaired student’s t-test (one-tailed, unequal variance). Differences with a p-value <0.05 were considered significant. Experiments were repeated to confirm their consistency.

The lasR^C133T substitution has been identified as a recurring mutation site in P. aeruginosa CF isolates (Feltner et al., 2016). This single-base substitution introduces a premature stop codon into the lasR sequence, terminating protein translation prematurely. This mutation was henceforth referred to as LasR^Q45stop. To validate the functionality of LasR^Q45stop, we generated a knock-out deletion mutant LasR, an allelic replacement mutant LasR^Q45stop, and a LasR complementation strain C-LasR^Q45stop in the laboratory model strain P. aeruginosa PAO1. Comparing the QS-associated traits of the strains mentioned, we found that the homologous LasR^Q45stop mutation in P. aeruginosa acts as a nonfunctional LasR in QS-associated phenotypes, such as elastase secretion, pyocyanin production, and 3OC12-HSL and C4-HSL levels (Figures 1A–F).

To ascertain earlier studies indicated that LasR mutants possess fitness advantages over WT cells when co-cultured in minimal medium with casein (casein broth) as the sole carbon source. However, this advantage was negated with the addition of adenosine to the casein medium (Feng et al., 2019). To determine if LasR^Q45stop exhibited superior fitness to WT, we co-cultured the gentamicin-resistant PAO1 LasR^Q45stop mutant alongside WT in casein broth. LasR^Q45stop mutants were then distinguished and counted from WT using plates with or without gentamicin resistance. The schematic diagram (Supplementary Figure S2C) illustrates that 1% competitors (LasR^Q45stop or LasR) were individually combined with 99% of WT in the initial casein broth, then passaged daily, and the frequency of mutant strains was subsequently monitored. Pyocyanin production during the initial days of the evolution process hinted at the robust growth of the PAO1 co-culture. Pyocyanin levels tapered off from day 4 to day 6, culminating in a milky culture by day 7. No noticeable alteration in the casein medium post-day 8 signified a population collapse (Figure 1G). Consistent with expectations, the proportion of LasR^Q45stop mutants surged, peaking at nearly 70% in cultures (Figure 1H). This shift in frequency suggests that the LasR^Q45stop mutant can act as a social cheater when co-cultured with WT in a casein-based medium.

LasR mutations are prevalent in P. aeruginosa isolated from chronic infections (Feltner et al., 2016). Interestingly, many LasR-null clinical isolates partially restored the QS regulon over extended infections (Feltner et al., 2016). This suggests that post-LasR-deficient evolution might confer diverse benefits. We were intrigued by the evolvability and evolutionary trajectory of the high-frequency mutant LasR^Q45stop in minimal medium with casein as the exclusive carbon and energy source. A competitive assay between the LasR^Q45stop variant and WT was conducted as described above (Supplementary Figure S2C). LasR^Q45stop variants were isolated using LB Gm-resistant plates, and their protease activity was assessed on skim milk agar plates (Supplementary Figure S2D). Variants exhibiting notable protease alterations were selected for further analysis (Supplementary Figure S2D). We identified four protease-altered LasR^Q45stop variants from these evolutionary experiments. Variants LasR^Q45stop-1, LasR^Q45stop-2, and LasR^Q45stop-4 demonstrated reduced protease activity relative to the parental strain LasR^Q45stop. In contrast, LasR^Q45stop-3 exhibited increased protease activity, approaching that of the wild-type strain (Figure 2A). We also evaluated the influence of these four isolates on virulence-associated traits, such as cell motility and pyocyanin production. Consequently, the twitching motility of LasR^Q45stop-1, LasR^Q45stop-2, and LasR^Q45stop-4 was diminished compared to the parental strain LasR^Q45stop (Figure 2B). Remarkably, LasR^Q45stop-3 amplified swarming motility, while LasR^Q45stop-1, LasR^Q45stop-2, and LasR^Q45stop-4 did not exhibit significant deviations from their parental strain LasR^Q45stop (Figure 2C). Additionally, pyocyanin production was elevated in LasR^Q45stop-3 (Figure 2D). Of note, LasR^Q45stop-4 maintained functional pyocyanin production, independent of its protease activity phenotype (Figures 2A,D).

To uncover the genetic variations responsible for the observed differences in virulence-associated phenotypes, we conducted a genome-wide association study among the parent LasR^Q45stop and four protease-altered mutants (LasR^Q45stop-1, LasR^Q45stop-2, LasR^Q45stop-3, and LasR^Q45stop-4). This analysis identified three genes: mexT (PA2492), xcpA (PA4528), and psdR (PA4499) (Table 1). The single base indel or in-frame deletion in the xcpA gene (Table 1) was found in LasR^Q45stop-1, LasR^Q45stop-2, and LasR^Q45stop-4. Conversely, a missense variant in the mexT gene was identified in LasR^Q45stop-3 (Table 1). The LasR^Q45stop-4 subline displayed additional structural variations (SVs), culminating in the deletion of the entire mexT gene and adjacent genes spanning from PA2259 to PA2544. This range is part of the 38.3 kb deletion observed in LasR^Q45stop-4. Furthermore, we noted that all four sequenced isolates carried a nonsynonymous mutation in psdR, aligning with prior research that identified frequent psdR mutations in casein evolution experiments (Asfahl et al., 2015; Kostylev et al., 2019). Nevertheless, psdR has been demonstrated to enhance growth on casein independently of QS (Asfahl et al., 2015). We thus infer that MexT and XcpA are instrumental in the observed virulence-associated phenotypic variations in the evolved isolates.

To substantiate this, we engineered MexT-and XcpA-related mutants using LasR^Q45stop-PsdR as the progenitor strain. Our results show that LasR^Q45stop-PsdR-MexT and LasR^Q45stop-PsdR-MexT^A1044G mutants exhibited markedly elevated pyocyanin and elastase production (Figure 3), as well as augmented swarming motility (Figure 3). Deletion of xcpA from both LasR^Q45stop-PsdR and LasR^Q45stop-PsdR-MexT mutants led to reduced QS-dependent elastase phenotypes and twitching motility (Figure 3). Alterations in mexT have been highlighted as pivotal in reshaping the QS circuits and the fitness of P. aeruginosa PAO1 (Kostylev et al., 2019). XcpA is a prepilin peptidase that is part of the Xcp type II secretion system (T2SS) in P. aeruginosa. Furthermore, the Xcp T2SS is governed by the QS pathway, which P. aeruginosa utilizes for secreting major virulence factors, including LasA and LasB elastases (Li and Lee, 2019). Therefore, we conclude that the xcpA and mexT mutations underpin the evolved phenotype, at least under laboratory conditions.

Mutants devoid of the metabolic cost of QS could possess a negative frequency-dependent fitness advantage over the parental strain when co-cultured in an environment where QS function is essential for growth. In previous research, LasR-MexT mutants exhibited a competitive advantage over the parental LasR mutant (Kostylev et al., 2019). Similarly, we postulated that XcpA and LasB mutants might also harbor a competitive edge. We co-cultured each identified mutant with the parental strain, starting at a frequency of 10%, to evaluate their relative fitness in PM-caseinate. Distinguished mutants were marked as gentamicin-resistant (Gm^R). Relative fitness was ascertained by taking the ratio of mutant-to-parent bacteria at the endpoint and dividing it by the initial mutant-to-parent bacteria ratio (Chen et al., 2019). A control was set up with unlabeled and labeled parental strains, yielding a relative fitness of 1 (Figure 4A). Conversely, the relative fitness of the XcpA mutant, when pitted against LasR^Q45stop-PsdR-MexT or WT, exceeded 1, indicating a survival advantage in the XcpA mutant. However, the LasB mutant, which only lacked elastase production, exhibited a fitness comparable to the LasR^Q45stop-PsdR-MexT parent strain (Figure 4A). Additionally, varying proportions of XcpA mutants were co-cultured with their parent strains in PM-caseinate for 48 h to determine their frequency-dependent fitness. When XcpA mutants were more abundant relative to LasR^Q45stop-PsdR-MexT in the culture, their relative fitness was less than 1. However, when LasR^Q45stop-PsdR-MexT bacteria predominated over the mutant in the starting inoculum, the relative fitness for XcpA exceeded 1 (Figure 4B). The relative fitness of the mutants declines as the mutant subpopulation expands (Diggle et al., 2007). These observations underscore that XcpA mutants can effectively invade their ancestral populations under conditions that require QS.