Optimizing rhamnolipid biosynthesis: evaluating predictive methods using Pseudomonas aeruginosa mutants Authors: Ingrid Yoshimura, Jonas Contiero, Eric Déziel Published: 2026 ABSTRACT ABSTRACTRhamnolipids (RLs) are versatile biosurfactants produced by Pseudomonas aeruginosa with significant industrial potential. However, high production costs remain a barrier to large-scale use, necessitating genetic strategies to improve yields. While many genes are reported to influence RL production, studies often rely on qualitative phenotypic assays of questionable reliability. We systematically evaluated 29 P. aeruginosa PA14 mutants using traditional assays (Siegmund–Wagner blue plates, swarming motility) and validated the findings using liquid chromatography–mass spectrometry (LC/MS). We found that traditional phenotypic assays have a high misprediction rate (~35% to 38%), primarily due to confounding factors like variable flagellar function, colony spreading, and growth rates. Specifically, LC/MS quantification revealed that rpoN and pvdQ knockouts significantly increased total RL titers, whereas crc, dksA, and dspI knockouts decreased production. Notably, the increased titers in rpoN and pvdQ mutants were linked to enhanced biomass accumulation rather than higher per-cell biosynthetic rates. These findings highlight the critical necessity of using quantitative analytical methods for accurate strain screening and provide a clarified set of genetic targets for metabolic engineering aimed at optimizing RL production.IMPORTANCEThis study addresses a critical methodological flaw in biosurfactant research: the overreliance on qualitative phenotypic assays that too often lead to inaccurate conclusions. By systematically comparing traditional screening methods with liquid chromatography–mass spectrometry quantification across a collection of Pseudomonas aeruginosa mutants, we demonstrate that common assays like swarming motility and blue plates fail to accurately predict rhamnolipid (RL) production in over one-third of cases. These inaccuracies lead to the misidentification of genetic targets and may waste resources in metabolic engineering efforts. Our work provides a reliable framework for strain screening, identifies specific genes that influence RL yields, and clarifies the biological factors—such as flagellar motility and growth dynamics—that bias traditional results. These findings are essential to optimize biosurfactant production and ensure data reproducibility. INTRODUCTION Biosurfactants are microbially derived compounds increasingly explored as sustainable alternatives to conventional chemical surfactants (1). They are synthesized by microorganisms such as bacteria and offer several advantages, including high biodegradability, low toxicity, production from renewable resources, structural diversity, selectivity, and stability under extreme conditions of temperature and salinity (2–4). Among them, rhamnolipids (RLs) stand out as glycolipid biosurfactants with remarkable properties and demonstrate commercial potential. However, large-scale applications would still benefit from increased production yields (5, 6). This challenge is driving efforts to develop high-producing microbial strains. Pseudomonas aeruginosa, the main natural RL producer, regulates RL biosynthesis, especially the rhlAB biosynthetic genes, through a complex regulatory network involving three interconnected quorum sensing (QS) systems (7). Due to this high level of regulatory complexity, many genes could influence RL production directly or indirectly. For instance, many factors are involved in social swarming motility and biofilm formation—processes closely tied to RL production (8–13)—suggesting that previously poorly investigated genetic alterations may impact RL production. To explore this relationship, a literature review was conducted to list genes that have been identified to influence RL production and/or regulatory pathways, either positively or negatively. The results of this survey are summarized in Table 1. Although numerous studies have reported that knocking out specific genes leads to altered RL production, the methods commonly used to assess such changes are varied, and their reliability to reach such conclusions may be questionable (14). For instance, a frequent approach is the use of swarming motility assays on semisolid agar plates as a proxy for RL production by P. aeruginosa. However, any correlation between swarming behavior and RL levels should be expected to be inconsistent, as this phenotype is also dependent on flagellar function. In this context, it is important to evaluate the limitations and potential pitfalls associated with phenotype-based inferences and predictive approaches, particularly in screening processes. In the present study, we used 29 mutants from a nonredundant library to evaluate the reliability of various methodologies and identify their limitations in RL production assessments. TABLE 1Genes reported to influence rhamnolipid production in Pseudomonas aeruginosaGeneProtein function/roleStrainMethodReferenceProposed positive impact on production algRGlobal transcriptional regulator; controls alginate production, biofilm formation and virulence (15)PAO1Orcinol assay using supernatant(16) bifACyclic di-GMP phosphodiesterase, inversely regulates biofilm formation and swarming motility, promotes dispersion when overexpressed (17)PA14Swarming and biofilm impact(18) crcGlobal regulator involved in catabolite repression; modulates assimilation of certain sugars, nitrogen compounds, hydrocarbons compounds (19)PAO1Blue plate and orcinol assay(20) dipAEnables dispersion in biofilms, influences flagellar motility (21)PA14Swarming and biofilm impact(18, 22) dksAMediator of the stringent response; influences many adaptive responses, including motility/biofilm; reported as independent and hierarchically dominant over some QS and c-di-GMP signaling systems (23)PAO1Blue plate and orcinol assay(18) dspIEnoyl-CoA hydratase/isomerase producing the biofilm dispersion signal cis-2-decenoic acid (24)PA14Biofilm impact(24) gidAAffects flagella-mediated motility and biofilm formation, required for full virulence (25)PA14Blue plate and protein levels(26) ntrBTwo-component system that senses nitrogen limitation, controls small RNA NrsZ production (27)PA14Blue plate and swarming impact(28) ntrCTwo-component system that senses nitrogen limitation, controls small RNA NrsZ production (27)PAO1Blue plate and swarming impact(29) phoBResponse regulator of the PhoR–PhoB two-component system, controls genes for phosphate uptake and adaptation to phosphate limitation (30)PAO1rhlR and rhlA expression(11) ppkPolyphosphate kinase; polyP plays roles in stress responses, biofilm formation, and virulence (31)PAO1Swarming impact(32) pvdQControls pyoverdine production and swarming motility, increases virulence via pyoverdine/iron pathway (33)PAO1Swarming impact(34) rsmAGlobal post-transcriptional regulator; affects >500 genes; many involved in virulence (iron acquisition, efflux pumps, motility, etc.) (35)PAO1Orcinol assay(36)Proposed negative impact on production lonATP-dependent protease; essential for adaptation in complex environments; involved in antibiotic resistance, QS, biofilm, motility, and DNA damage response (37)PAO1Blue plate(20, 37) mexEMultidrug efflux pump (operon); contributes to antibiotic resistance; regulators may also modulate QS/virulence (38)PAO1Swarming impact and rhlA expression(39) mexTTranscriptional regulator, activates MexEF-OprN (38)PAO1 and PA14QS-regulated gene expression(40–42) mucRContributes to alginate biosynthesis by generating local c-di-GMP pools, under direct positive control of AlgR (15, 43)PA14Swarming impact(18) mvaTRepressor of rsmZ, an sRNA crucial for the Gac/Rsm system (44)PAO1rhlR expression(45) ptxRNegative regulator of RL and pyocyanin genes via RhlI and PqsABCDE, positive regulator of elastase (LasB) via LasI (46)PAO1rhlA expression(46) qscRLuxR-type quorum sensing receptor, modulates QS, acts as repressor/regulator, and reduces virulence (47)PAO1rhlI expression(48) roeAPromotes production of extracellular polysaccharide (EPS) and contributes to biofilm formation (49)PA14Swarming and biofilm impact(18) rpoNSigma factor; controls many biological pathways, including virulence, motility, and nutrient regulation (50)PAO1Blue plate(51) rsaLDirect repressor of lasI transcription (9)PAO1QS-regulated gene expression(52) wspRDiguanylate cyclase; increases intracellular c-di-GMP levels and promotes biofilm formation (53)PA14Swarming and biofilm impact(18) yfiNRequired for biofilm maintenance via c-di-GMP signaling (54)PA14Swarming and biofilm impact(18)Contradictory predicted effect on RL production gacAResponse regulator of GacS/GacA two-component system, activates sRNAs RsmY and RsmZ, governs switch between acute and chronic infection phenotypes (55)PAO1Swarming impact(56) gacSSensor histidine kinase of GacS/GacA system; senses environmental signal; controls phosphorylation of GacA; thereby influences RsmY/Z production and downstream gene expression (57)PA14Swarming impact(58) rpoSAlternative sigma factor; regulates production of several virulence factors and QS; stress response, stationary-phase regulator (59)PAO1rhlAB expression and orcinol(60) vfrGlobal virulence factor regulator; homologous to cAMP receptor protein; activates transcription of many virulence genes, including type III secretion, QS genes (61)PAO1rhlR expression(62) Genes reported to influence rhamnolipid production in Pseudomonas aeruginosa RESULTS AND DISCUSSION Assessing swarming motility as a predictive method for rhamnolipid production Over the years, several genes have been implicated in the production of RLs and/or the expression of the rhlAB biosynthetic genes across different P. aeruginosa strains, under various culture conditions and using diverse methods. In order to clarify the literature and reconcile often-contradictory findings, we compared several knockout mutants of the prototype strain PA14 side by side (listed in Table 1), under the same experimental conditions and using the same methodologies. As an initial indicator of differences in biosurfactant production, a swarming assay was performed. Swarming motility is a rapid and coordinated multicellular movement across a semisolid surface, driven by morphological differentiation and intercellular interactions (63). A distinctive feature of P. aeruginosa swarming is the formation of dendritic, fractal-like tendrils radiating from the inoculation site (64), a process associated with the production of RLs and its precursor 3-(3-hydroxyalkanoyloxy) alkanoic acids (65). It is therefore commonly used as an indirect measure of differential RL synthesis. As shown in Fig. 1, the lon and crc mutants exhibited no swarming under our experimental conditions. Mutants for bifA, dipA, dksA, gacA, gacS, pvdQ, rpoN, and rsmA showed severely reduced swarming motility, while rsaL and dspI displayed mild swarming compared to wild-type (WT) PA14. Mutants in genes such as algR, gidA, mexE, mexT, mucR, ntrC, phoB, ppk, ptxR, qscR, roeA, vfr, wspR, and yfiN exhibited swarming patterns similar to WT. None produced obviously enhanced swarming motility, suggesting no mutant would produce enhanced levels of RL compared to the parental background. Swarming motility assays of P. aeruginosa PA14 and various isogenic mutants. Strains were assessed after 16–20 h of incubation. Assays were performed in triplicate with the wild type included for comparison to validate the reproducibility of the results. Based on these results and following the hypothesis that swarming motility correlates with RL production, it would be predicted that crc and lon mutants are defective in the biosynthesis of this wetting agent. In P. aeruginosa, deletion of crc disrupts the rhl quorum sensing system, which would thereby impair RL synthesis since it positively regulates rhlAB. Crc, along with the RNA chaperone Hfq, normally represses the expression of the Lon protease, which targets RhlI, the enzyme responsible for producing the Rhl quorum sensing signal N-butanoyl-L-homoserine lactone (C4-HSL). In the absence of Crc, Lon levels increase, leading to RhlI degradation, reduced C4-HSL synthesis, and consequently, downregulation of RL production (20). Notably, deletion of lon or constitutive expression of rhlI restores RL levels, confirming the role of this regulatory cascade (20). Thus, the crc knockout is expected to reduce or inhibit RL production, whereas the lon knockout should not. This raises the question: is the lack of swarming motility of the lon mutant really explained by the absence of RL production? Additionally, swarming motility requires a functional flagellum (64, 65). Therefore, in addition to RL production, we need to consider whether altered flagellar regulation might explain some of the swarming phenotypes observed. Indeed, several selected mutants are involved in the regulation of the second messenger bis-(3′−5′)-cyclic dimeric guanosine monophosphate (c-di-GMP), a key regulatory molecule in bacteria typically associated with biofilm formation and a sessile lifestyle (66, 67). In general, low levels of c-di-GMP are linked to motility and virulence factor production, whereas high levels suppress flagellar activity and promote the expression of adhesins and biofilm-associated exopolysaccharides (EPSs) (18, 66). Accordingly, it is reasonable to assume that mutants with a tendency to form more biofilm may exhibit reduced flagellar activity and, consequently, diminished swarming motility. The dipA mutant (dispersion-induced phosphodiesterase A) exhibits increased biofilm formation and reduced swarming motility (68), findings that align with our observations under the tested conditions. In P. aeruginosa PA68, inactivation of dipA was associated with increased biofilm formation, likely due to compromised flagellar function (18). Specifically, DipA loss significantly inhibits flagellar motor switching and directional reversal during swimming in a MapZ-dependent manner (21). In this case, less swarming would not be a consequence of reduced RLs. Taken together, differences in swarming behavior compared to WT can be attributed to one or more of the following factors: (i) altered growth rate (e.g., some mutants may have growth defects or poor adaptation to the swarming medium); (ii) modified RL production; and (iii) different motility (e.g., flagellar defects could result in reduced spreading). Next, we verified these factors separately. Comparative growth analysis of mutant strains Considering that growth should also be considered when assessing the RL production potential of a bacterium, we next assessed whether any of the mutants exhibited growth defects in the M9DCAA culture medium and monitored their growth for 36 h with a Bioscreen system. This medium is the same used for swarming assays, without agar. The results revealed that the rsaL and lon mutants display impaired growth under these conditions (Fig. S1). The inability of the lon mutant to swarm may thus be explained by its growth defect, while the rsaL mutant exhibited reduced swarming compared to the WT, which could also be partially attributed to its compromised growth. In contrast, the gacS, dipA, bifA, rsmA, and rpoS mutants demonstrated enhanced growth compared to the WT under the same conditions. Interestingly, all mutants that exhibited increased growth showed a reduced swarming motility phenotype as well. Thus, since growth alone did not account for all cases of increased or decreased swarming, we turned to a more specific colorimetric assay on agar plates to better understand whether the observed differences in swarming patterns were related to RL production or not. Siegmund–Wagner blue plate assay for rhamnolipid detection The Siegmund–Wagner blue plate method allows the direct observation of RL production around a colony by the formation of halos resulting from the interaction of the anionic biosurfactant with a cationic surfactant, causing its precipitation, thus enabling semiquantitative assessment of RL production (69). This is a widely used method for screening and considered a straightforward and reliable way to verify RL production. However, although apparently simple, we noted that interpretation of results can be challenging because the area of the RL halo zone is affected by the size of the colony; indeed, sometimes, it is not possible to conclude whether the zone is absent or hidden behind the colony (Fig. 2A), and bacteria forming smaller colonies might, misleadingly or not, produce smaller halos. To partially account for these factors, we measured the total diameter of the colony plus the RL production zone, as shown in Fig. 2B. Rhamnolipid production zones by Pseudomonas aeruginosa PA14 wild type and mutants on Siegmund–Wagner blue plates. (A) Pictures of plates after 72 h of incubation. (B) Rhamnolipid production zones represented by the mean area of the production zone, including the colony. Data were analyzed by one-way ANOVA followed by Dunnett’s post hoc test vs control (****P < 0.0001, **P < 0.01, *P < 0.05). While some mutants, including dipA, gidA, mexE, phoB, pvdQ, roeA, wspR, and yfiN, presented halo areas comparable to the WT, none displayed a significantly larger zone, thus presumably higher RL production. Among these mutants indistinguishable from WT, all except dipA and pvdQ demonstrated swarming behavior also similar to the WT, suggesting that these genes are actually not involved in RL production, at least under the tested conditions. Other mutants such as dipA and pvdQ displayed defective swarming but WT RL production on blue plates. This adds to the possibility that other factors, such as a flagellar function defect, could play a role—which likely explains the behavior of the dipA mutant—as previously discussed. The mutants algR, crc, gacA, lon, mvaT, ntrB, ntrC, rsaL, rsmA, and vfr formed smaller colonies on blue plates, which also interferes with interpretation of the results; potential growth defects in this specific condition and/or impaired motility function could be affecting colony size. Among these, only lon exhibited impaired growth in the Bioscreen assay, raising the possibility that the reduced colony size of the other mutants might be due to defective flagellar function rather than growth limitation. The dksA, ntrC, ntrB, and rsaL mutants did not form visible halos at all, indicating no RL production under the tested conditions. However, those mutants swarmed in the previous test (Fig. 1), where dksA and rsaL swarmed less than the WT, but ntrB and ntrC exhibited swarming patterns similar to WT. This observation suggests two possible hypotheses: RL production is influenced by the composition of the culture medium—which is different in blue plates—or a halo is formed, but its size is too small to be discernible, potentially remaining obscured beneath the bacterial colony. Colony size influenced halo interpretation, which could again lead to biased conclusions, in the same way as swarming motility: some colonies might spread more easily due to increased flagellar function or better adaptation to the growth medium, not necessarily due to higher RL production. To better understand the impact of flagellar function of mutants on swarming motility and blue plate results, a swimming assay was conducted to investigate potential flagellar activity defects. Swimming assay-based evaluation of flagellar motility is needed for interpretation of swarming and blue plate results Since a swarming defect or smaller colonies can result from defective flagellar motility and thus cause incorrect interpretation of results obtained on plates in the context of RL production, a swimming assay was conducted to investigate potential flagellar function defects (Fig. S2; Fig. 3). Swimming motility of P. aeruginosa PA14 and various mutants. The assay was performed on M9DCAA with 0.25% agar plates. Strains were incubated at 37°C, and the measurements were performed after 16 h. The ΔfliC mutant was included in the assay as a negative control. Data were analyzed by one-way ANOVA followed by Dunnett’s post hoc test vs control (****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05). The crc, dipA, gacS, lon, mvaT, ntrC, phoB, ptxR, rpoS, rsmA, and yfiN mutants exhibited reduced swimming compared to the WT, while dspI, mexE, mexT, and mucR showed enhanced swimming. Based on these results, it can be inferred that the reduced swarming motility observed for crc, dipA, gacS, rpoS, lon, and rsmA mutants may be attributed to flagellar defects. Among these, crc also displayed reduced RL production, and lon showed impaired growth, further contributing to their limited swarming colony size. On the other hand, mvaT and phoB demonstrated swarming similar to the WT despite reduced swimming, raising the possibility that these mutants may compensate for defective flagella by increasing RL production. Following this reasoning, the mexE, mucR, and mexT mutants that exhibited enhanced swarming might have benefited from other mechanisms, such as increased motility, and may in fact produce lower amounts of RL. However, since flagellar function also influences colony spreading on blue agar plates, and both swarming and halo assays can be biased by growth or motility differences, direct quantification of RL is necessary to validate these hypotheses. The most accurate approach for this is liquid chromatography–mass spectrometry (LC-MS) for direct measurements of molecules in liquid culture samples. Measurement of rhamnolipid production by LC-MS quantification To verify if the swarming and blue plate assay combined with a swimming assay to evaluate flagellar activity allowed for identification of altered RL-producing mutants, we performed LC-MS quantifications. LC-MS analyses are a reliable method to quantify RL levels since it allows for direct quantifications from cultures. For RL production measurements, bacteria were grown in M9DCAA. All mutants were included except rsaL and lon since they exhibited growth defects (Fig. S1). After 5 days of cultivation, biomass dry weight measurements indicated that gacA, rpoN, dspI, wspR, and bifA exhibited higher growth than the WT (Fig. 4A). Among these, only bifA displayed enhanced growth in the Bioscreen assay after 36 h (Fig. S1). The others possibly benefited from a late-phase growth advantage, as the prior assay was limited to a shorter monitoring window, or contributed to biomass beyond cells per se, such as exopolysaccharides (EPS) or polyhydroxyalkanoate (PHA) granules. Biomass and rhamnolipid production in P. aeruginosa PA14 and mutants. (A) Biomass quantification by dry mass. (B) Rhamnolipid quantification by LC-MS. (C) The strains were cultivated in tubes with liquid M9DCAA supplemented with dextrose for 5 days. Data were analyzed by one-way ANOVA followed by Dunnett’s post hoc test vs control (****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05). Regarding total RL production, LC-MS quantification revealed increased production in pvdQ and rpoN cultures and reduced levels by the crc, dksA, and dspI mutants (Fig. 4B). Although other mutants showed potential for higher RL production, the differences were not statistically significant. However, when production was normalized to cell biomass, the specific yield (RL/biomass) for rpoN and pvdQ remained comparable to that of the WT (Fig. 4C). This suggests that the increased RL accumulation observed in these mutants results primarily from greater cell growth or a prolonged productive phase, rather than an intrinsic increase in per-cell synthesis rates. From a process perspective, this profile can be advantageous, as higher titers facilitate and reduce the cost of downstream recovery and purification. Nevertheless, from a metabolic-engineering standpoint, rpoN and pvdQ do not show a real increase in biosynthetic capacity per unit of biomass. However, as noted above, it is important to note that dry biomass measurements do not distinguish between viable cells, EPS, or PHA, which may contribute to the observed differences. In addition, given the strong growth-phase dependence of RL production, time-resolved or transcriptional data would provide a more detailed mechanistic interpretation. The RpoN sigma factor is a global transcriptional regulator that enables P. aeruginosa to rapidly adjust gene expression in response to environmental changes. It governs several processes related to virulence and QS (50). Although RpoN primarily modulates the transcription of metabolic genes, it also regulates key components of the QS network, including lasI, rhlI, and pqsR (51, 70). Our results are in agreement with the hypothesis that RpoN functions as a repressor of the rhl QS system by binding to the promoters of rhlI, rhlA, and rhlR. Furthermore, RpoN downregulates the expression of gacA, encoding for another major global regulator of virulence factors (36, 51). Nevertheless, other studies have reported that RpoN can act as an activator of QS (60, 70), suggesting that its regulatory role may be context-dependent, which is consistent with previous findings that culture conditions strongly influence the expression of QS-regulated genes (71). On the other side, the protein PvdQ is involved in the maturation of pyoverdine, a siderophore important for iron acquisition in P. aeruginosa (72, 73). Moreover, PvdQ plays a significant role in QS regulation by hydrolyzing acyl-homoserine lactones (AHLs), particularly N-3-oxo-dodecanoyl-L-homoserine lactone (3-oxo-C12-HSL), thereby influencing the expression of AHL-dependent genes such as rhlAB (33, 74). A study explored the use of PvdQ to reduce the virulence of P. aeruginosa and demonstrated that treatment with this enzyme inhibited the activity of a rhlA promoter-luciferase fusion, suggesting that PvdQ-mediated degradation of QS signals reduces rhlAB expression (75). pvdQ mutants also displayed increased biofilm formation and reduced swarming motility likely due to signal accumulation (34, 76), and in our study, we found a similar increase in RL production under these conditions. Another work found decreased swarming but no difference in RL production (33). These findings further highlight that the regulation of swarming motility is not directly linked to RL production, but rather to the broader coordination of multicellular behavior mediated by several factors, including QS. The importance of signal concentration in swarming regulation is a possibility from observations that pvdQ mutants, that should be accumulating higher levels of 3-oxo-C12-HSL signal, display impaired swarming motility (Fig. 1) (37). This finding aligns with previous work, which demonstrated that QS signals are required for swarming: las system mutations reduce swarming, while rhl mutations abolish it entirely (64). These observations suggest that an optimal balance of QS signals is critical, and deficiency or excess of these molecules can both disrupt multicellular coordination. Taken together, PvdQ’s role in swarming motility is not merely a consequence of altered RL levels but likely involves fine-tuned control of QS signal turnover, which is essential for proper spatial and temporal coordination of collective motility. In a different manner, the dksA mutant produced decreased levels of RL and showed less swarming compared to the WT. Jude et al. suggested that DksA inhibits RL synthesis by repressing C4-HSL production (77). Nonetheless, DksA appears to support basal-level translation of rhlAB during the stationary phase. Transcriptional fusion assays showed that lasB and rhlAB are transcribed normally in the dksA mutant, but elastase and RL levels were markedly reduced. This indicates that DksA acts post-transcriptionally, possibly by influencing mRNA translation or protein stability (77). Clearly, the correlation between swarming behavior and RL levels is not always consistent. Despite the established role of RLs in facilitating swarming, discrepancies are observed between swarming ability and actual RL production levels measured in liquid cultures, which could lead to questioning the accuracy of the swarming assay as a screening method for factors influencing RL production. For instance, for the gacS mutant, impaired flagellar function reduces both swimming and swarming motility. Nonetheless, RL production in cultures remains comparable to that of the wild type, as confirmed by LC-MS, indicating that reduced swarming is not linked to an altered RL biosynthesis ability in this case. Similarly, bifA mutants exhibit diminished swarming motility, likely due to impaired flagellar reversal mechanisms (17). A reduced frequency of flagellar reversals limits the cell’s ability to change direction efficiently, which is essential for the flexible and coordinated movements required during complex swarming behavior (17). However, swimming remains unaffected, and RL levels are unchanged. Interestingly, Liu et al. demonstrated that reduction in c-di-GMP—specifically through activation of the phosphodiesterase NbdA—triggers a shift from biofilm formation to increased RL production (78). Therefore, decreasing c-di-GMP may act as a molecular switch that favors biosurfactant synthesis, suggesting a promising strategy to enhance RL production by modulating intracellular c-di-GMP levels. In this context, the knockout of genes such as yfiN, roeA, mucR, and wspR would be associated with lower c-di-GMP levels and may favor to enhance RL production, consistent with altered surface-associated behavior. Conversely, knocking out bifA and dipA was expected to maintain higher c-di-GMP levels and favor biofilm formation (18), thereby presumably reducing RL synthesis. However, in all these mutants, RL production remained similar to the WT strain. In conclusion, the coordinated control of motility and secondary metabolism by multiple QS systems underscores the evolutionary plasticity of P. aeruginosa. This signaling crosstalk ensures a seamless transition between individual and social lifestyles where RL production represents a key metabolic output of this regulatory network. Understanding this integrated regulation is essential to unraveling how P. aeruginosa effectively adapts. Limitations and integration of phenotypic prediction methods In this study, we wanted to verify the reliability of classical methods often used in the literature to screen for factors leading to modified RL production or strains with different RL production ability. Based on our results, summarized in Table S1, we were able to reach some conclusions on whether swarming motility and Siegmund–Wagner blue plate assays may serve as reliable predictors of RL production in P. aeruginosa mutants (Fig. 5). This comparison not only assessed the predictive power of these phenotypic methods against known mutant profiles but also provided insight into the biological factors that contribute to inconsistencies and misinterpretations in RL production data in the literature. To this end, both assays were examined for their accuracy and limitations in reflecting the actual biosynthetic capacity of different mutants. It is important to emphasize that this analysis was conducted under specific experimental conditions (M9DCAA medium, tube-based cultivation, and endpoint quantification after 72 h). Therefore, we cannot assume that the predictive performance would be the same under all conditions. Accuracy of swarming motility vs Siegmund–Wagner assays in predicting rhamnolipid production and underlying causes of misprediction. The swarming assay yielded a 65% rate of correct prediction, confirming that there is some correlation between swarming motility and RL production. The correct predictions were associated with mutations in genes such as algR, dksA, dspI, gidA, mexE, mexT, mucR, mvaT, ntrC, phoB, ppk, ptxR, qscR, roeA, vfr, wspR, and yfiN. However, 35% of predictions were incorrect, which were attributed to three main biological factors: (i) failure in the flagellar rotation mechanism or bacterial communication (15%), linked to mutations in bifA, gacA, pvdQ, and rpoN; (ii) impaired motility, associated with disruptions in crc, dipA, gacS, rpoS, and rsmA; and (iii) loss of motility in the case of the rsmA mutation. The Siegmund–Wagner blue plate assay demonstrated 62% accuracy in predicting RL production, with correct predictions involving the same set of functional genes identified above. Nonetheless, 38% of cases were misclassified due to (i) impaired growth, observed in mutants of algR, gacA, ntrC, and vfr; (ii) unfavorable conditions for RL production in the assay, involving dksA, gacS, pvdQ, rpoN, and rpoS; and (iii) impaired motility in the rsmA mutant. Conclusion Our results reveal that swarming motility and SW blue plate phenotypic assays do not offer a sufficiently reliable prediction of RL production. A considerable proportion of incorrect predictions can occur when relying solely on these methods. While this is not unexpected about swarming, a phenotype also relying on flagellar function, we were surprised by such low reliability for the colorimetric SW blue plate assay. Use of this method should always consider the actual growth and size of the colony formed on the agar surface as influencing the analysis. The observed inconsistencies emphasize the complexity of RL biosynthesis regulation, which includes factors such as QS, stress response, and cellular metabolism. Therefore, while these assays can serve as preliminary screening tools, they must be complemented by direct, quantitative methods, such as LC-MS, to ensure accurate and robust assessment of RL levels. MATERIALS AND METHODS Strains The wild-type strain P. aeruginosa PA14 and 29 isogenic knockout mutants in genes previously found or proposed to influence RL production were investigated and are listed in Table 2. TABLE 2Strains used in this studyP. aeruginosa strainLab collection numberRelevant genotype or descriptionReferencePA14ED14Clinical isolate from a human burn patient UCBPP-PA14(79)PA14 ntrB::MrT7ED5052PA14 derivative, ntrB::MrT7 transposon insertion(80)PA14 ntrC::MrT7ED2403PA14 derivative, ntrC::MrT7 transposon insertion(80)PA14 phoB::MrT7ED284PA14 derivative, phoB::MrT7 transposon insertion(80)PA14 gacA::MrT7ED1799PA14 derivative, gacA::MrT7 transposon insertion(80)PA14 gacS::MrT7ED1360PA14 derivative, gacS::MrT7 transposon insertion(80)PA14_54640::MrT7ED1841PA14 derivative, dspI::MrT7 transposon insertion(80)PA14_56790::MrT7ED588PA14 derivative, bifA::MrT7 transposon insertion(80)PA14_66320::MrT7ED589PA14 derivative, dipA::MrT7 transposon insertion(80)PA14 crc::MrT7ED1370PA14 derivative, crc::MrT7 transposon insertion(80)PA14 rpoS::MrT7ED2180PA14 derivative, rpoS::MrT7 transposon insertion(80)PA14 dksA::MrT7ED1090PA14 derivative, dksA::MrT7 transposon insertion(80)PA14 rsaL::MrT7ED1865PA14 derivative, rsaL::MrT7 transposon insertion(80)PA14 lon::MrT7ED5051PA14 derivative, lon::MrT7 transposon insertion(80)PA14 mexE::MrT7ED1517PA14 derivative, mexE::MrT7 transposon insertion(80)PA14 mexT::MrT7ED1516PA14 derivative, mexT::MrT7 transposon insertion(80)PA14 rsmA::MrT7ED282PA14 derivative, rsmA::MrT7 transposon insertion(80)PA14 ∆qscRED4724PA14 derivativeLab collectionPA14 ΔmvaTED1480PA14 derivativeLab collectionPA14 49890::MrT7ED3643PA14 derivative, yfiN::MrT7 transposon insertion(80)PA14_50060::MrT7ED596PA14 derivative, roeA::MrT7 transposon insertion(80)PA14_42220::MrT7ED600PA14 derivative, mucR::MrT7 transposon insertion(80)PA14 16500 wspR::MrT7ED578PA14 derivative, wspR::MrT7 transposon insertion(80)PA14 gidA::MrT7ED1086PA14 derivative, gidA::MrT7 transposon insertion(80)PA14 ppk::MrT7ED1866PA14 derivative, ppk::MrT7 transposon insertion(80)PA14 rpoN::MrT7ED1421PA14 derivative, rpoN::MrT7 transposon insertion(80)PA14 vfr::MrT7ED1087PA14 derivative, vfr::MrT7 transposon insertion(80)PA14 algR::MrT7ED1095PA14 derivative, algR::MrT7 transposon insertion(80)PA14 ptxR::MrT7ED283PA14 derivative, ptxR::MrT7 transposon insertion(80)PA14 pvdQ::MrT7ED1782PA14 derivative, pvdQ::MrT7 transposon insertion(80)PA14 ∆fliCED3956PA14 ∆fliC(81) Strains used in this study Swarming medium Swarming motility assays were performed using modified M9 medium with dextrose and casamino acids (M9DCAA) (g·L−1): NH4Cl, 1.068; K2HPO4, 2.99; Na2HPO4•7H2O, 1.70; NaCl, 0.50; Casamino Acids, 5.0; MgSO4 1 M, 1,000 µL; CaCl2 1 M, 1,000 µL; dextrose 1.1 M, 10.0 mL; and solidified with agar, 5 g·L−1 Bacto agar (Difco). The plates containing 20 mL of medium were dried under laminar flow for 75 min, then inoculated with 5.0 µL of suspension prepared in phosphate-buffered saline (PBS) (OD600 = 3.0) of the appropriate bacterium and incubated at 37°C (82). Bioscreen test Cells were grown overnight in TSB (Difco) at 37°C. Cultures were then inoculated in M9DCAA at an OD600 of 0.05 in a final volume of 200 µL in a Honeycomb plate. The plate was incubated in a Bioscreen C reader (Growth Curves USA) at 37°C for 36 h with OD600 measurements taken every hour, with 10 s of shaking prior to each reading. Siegmund–Wagner blue agar plates The medium composition (g·L−1) consisted of Na2HPO4, 0.9; KH2PO4, 0.7; CaCl2•2H2O, 0.1; MgSO4•7H2O, 0.4; NaNO3, 2.0; tryptone peptone, 1.0; glycerol, 20.0; cetyltrimethyl ammonium bromide, 0.2; methylene blue, 0.01; agar, 20; and trace elements (2.0 mL/L). The trace element solution consisted of (g·L−1) FeSO4•7H2O, 2.0; MnSO4•H2O, 1.5; and (NH4)6Mo7O24•4H2O, 0.6 (69). The pH was adjusted to 7.0 with NaOH solution. The plates were inoculated with 3.0 µL PBS suspensions at an OD600 = 3.0. The plates were incubated at 37°C for 24 h and then maintained at 30°C for 72 h before measurement of the halo zone. Swimming medium The medium consisted of tryptic soy broth (TSB), 25 g·L−1, and agar, 2.5 g·L−1. The plates containing 20 mL of medium were dried under laminar flow for 10 min, then inoculated with PBS suspensions at OD600 = 3.0 using a pipet tip. The plates were incubated at 37°C for 16 h. Colony diameters were measured to calculate the motility area. Cultivation conditions and LC-MS quantification of rhamnolipids Five milliliters of M9DCAA medium was inoculated at an initial OD600 = 0.05 from overnight TSB cultures. Cultures were incubated at 37°C for 120 h in a TC-7 roller drum (New Brunswick Scientific) at 240 rpm. At the end of the incubation period, 1 mL of culture was collected and centrifuged at 10,000 rpm for 5 min. The supernatant was used for RL quantification using LC-MS, whereas the pellet was dried and weighed to determine biomass dry weight. For LC-MS analysis, the supernatant was diluted in acetonitrile supplemented with 7.9 ppm 16-hydroxyhexadecanoic acid as an internal standard. Quantification of mono-RL (Rha-C10-C10) and di-RL (Rha-Rha-C10-C10) was performed based on normalized peak areas relative to the internal standard. LC-MS measurements were performed on a Waters Quattro Premier XE triple quadrupole mass spectrometer coupled to a Waters 2795 HPLC system (Waters, Mississauga, ON). Chromatographic separation was achieved on a Phenomenex Kinetex C8 column (2.6 µm, 100 × 4.6 mm) under isocratic elution with water and acetonitrile containing 1% formic acid, as previously described (83). Statistical analysis All experiments were performed with at least three independent replicates. Statistical analyses were performed using GraphPad Prism. Data were analyzed by ordinary one-way ANOVA followed by Dunnett’s post hoc test for multiple comparisons. Differences were considered statistically significant at P < 0.05.