Distinct Screening Approaches Uncover PA14_36820 and RecA as Negative Regulators of Biofilm Phenotypes in Pseudomonas aeruginosa PA14

Pseudomonas aeruginosa is a notorious human pathogen well known for forming biofilms, communities of bacteria that protect themselves within a self-secreted matrix. Here, we sought to find genetic determinants that impacted biofilm matrix production in P. aeruginosa strains. ABSTRACT Pseudomonas aeruginosa commonly infects hospitalized patients and the lungs of individuals with cystic fibrosis. This species is known for forming biofilms, which are communities of bacterial cells held together and encapsulated by a self-produced extracellular matrix. The matrix provides extra protection to the constituent cells, making P. aeruginosa infections challenging to treat. We previously identified a gene, PA14_16550, which encodes a DNA-binding TetR-type repressor and whose deletion reduced biofilm formation. Here, we assessed the transcriptional impact of the 16550 deletion and found six differentially regulated genes. Among them, our results implicated PA14_36820 as a negative regulator of biofilm matrix production, while the remaining 5 had modest effects on swarming motility. We also screened a transposon library in a biofilm-impaired ΔamrZ Δ16550 strain for restoration of matrix production. Surprisingly, we found that disruption or deletion of recA increased biofilm matrix production, both in biofilm-impaired and wild-type strains. Because RecA functions both in recombination and in the DNA damage response, we asked which function of RecA is important with respect to biofilm formation by using point mutations in recA and lexA to specifically disable each function. Our results implied that loss of either function of RecA impacts biofilm formation, suggesting that enhanced biofilm formation may be one physiological response of P. aeruginosa cells to loss of either RecA function. IMPORTANCE Pseudomonas aeruginosa is a notorious human pathogen well known for forming biofilms, communities of bacteria that protect themselves within a self-secreted matrix. Here, we sought to find genetic determinants that impacted biofilm matrix production in P. aeruginosa strains. We identified a largely uncharacterized protein (PA14_36820) and, surprisingly, RecA, a widely conserved bacterial DNA recombination and repair protein, as negatively regulating biofilm matrix production. Because RecA has two main functions, we used specific mutations to isolate each function and found that both functions influenced matrix production. Identifying negative regulators of biofilm production may suggest future strategies to reduce the formation of treatment-resistant biofilms.

IMPORTANCE Pseudomonas aeruginosa is a notorious human pathogen well known for forming biofilms, communities of bacteria that protect themselves within a selfsecreted matrix. Here, we sought to find genetic determinants that impacted biofilm matrix production in P. aeruginosa strains. We identified a largely uncharacterized protein (PA14_36820) and, surprisingly, RecA, a widely conserved bacterial DNA recombination and repair protein, as negatively regulating biofilm matrix production. Because RecA has two main functions, we used specific mutations to isolate each function and found that both functions influenced matrix production. Identifying negative regulators of biofilm production may suggest future strategies to reduce the formation of treatment-resistant biofilms.
KEYWORDS Pseudomonas aeruginosa, RecA, biofilms, recombination P seudomonas aeruginosa is a Gram-negative opportunistic pathogen of significance to human health, especially for immunocompromised patients and individuals with cystic fibrosis (1). P. aeruginosa can be difficult to treat due to its many virulence factors and behaviors, including biofilm formation (2). Biofilms comprise a community of bacterial cells that are held together and encapsulated by a self-produced extracellular matrix. The biofilm matrix provides extra protection to P. aeruginosa cells, allowing them to tolerate harsh environmental conditions and resist antibiotics (3). Production of the biofilm matrix also is associated with biofilm phenotypes such as adhesion to biotic and abiotic surfaces in industrial (4) and medical settings (5). One readily visible phenotype associated with small group of genes. After analyzing the biofilm effects resulting from deletion or overexpression of these genes, we took a mutagenesis and visual screening approach to find genes whose deletion could reverse biofilm suppression in a D16550 strain. Surprisingly, we identified the recombination protein-encoding gene recA. Because RecA functions both in homologous recombination and in the SOS response, we genetically dissected these functions and assessed their individual contributions to biofilm phenotypes; we found that abrogation of either function increased biofilm levels.

RESULTS
Impact of 16550 deletion on gene expression. Given the impact of 16550 deletion on biofilm colony morphology (9), and because 16550 is a known DNA-binding protein (36), we hypothesized that deletion of 16550 might elicit transcriptional changes that corresponded with decreased biofilm formation. To investigate the transcriptional impact of 16550 deletion on cells growing under laboratory biofilm colony-producing conditions, we performed transcriptomic analysis on RNA isolated from P. aeruginosa colonies grown on M6301 agar. We compared the transcriptome of the moderately hyper-biofilm-forming DamrZ strain (9,34) to its biofilm-impaired DamrZ D16550 derivative grown under identical conditions. We used the DamrZ strain background because of the strong impact of 16550 deletion on Pel production in this background and for consistency with our previous work (9). We observed that deletion of 16550 had significant effects on the expression of six genes ( Table 1, Fig. 1A). Among these, five genes, namely, PA14_20480, PA14_28600, PA14_49300, PA14_49310, and PA14_72360, were downregulated, whereas PA14_36820 was significantly upregulated. Hence, the absence of the 16550 protein, which corresponded with a marked reduction in biofilm formation, also corresponded with a specific transcriptional impact in biofilm-grown colonies.
Individual deletions of downregulated genes in D16550 cells minimally impact biofilm phenotypes. To test whether the genes most strongly regulated by 16550 impact biofilm formation under our laboratory conditions, we constructed in-frame deletions of the genes identified in our transcriptomic data set. We first considered whether the decrease in biofilm formation in the DamrZD16550 background was due to downregulation of one or more of the five downregulated genes we identified (Fig. 1A). If so, deletion of these genes (i.e., downregulation to zero) might, like the 16550 deletion, impact biofilm formation associated with DamrZ cells. Individual deletions of 20480, 28600, 49300, 49310 (which is cotranscribed with 49300), or 72360 in the DamrZ background showed little to no visual effect on the wrinkled morphology of DamrZ (Fig. 1B). Likewise, these deletions decreased neither Pel levels in colony biofilms as quantified by Congo red (Fig. 1C) nor the mass of surface-attached biofilms in static liquid cultures as quantified via crystal violet staining (Fig. 1D). For the sake of completeness, we also constructed the same deletions in the wild-type PA14 background, which does not form wrinkled colonies, to assess their impact (see Fig. S1 in the supplemental material). Deletion of 20480, 28600, 49300, and 49310 showed little to no effect on wild-type (smooth) colony morphology ( Fig. S1A and B), Pel levels ( Fig. S1C and D), or surface-attached biofilm mass crystal violet staining (Fig.  S1E). However, deletion of 72360 caused a dramatic increase in colony wrinkling that was accompanied by significantly increased Pel levels (Fig. S1C). Though at present it remains unclear how the effect of 72360 deletion relates to the decreased biofilm phenotype of D16550 strains, the positive impact of 72360 deletion on wild-type biofilm formation at least constitutes evidence that 16550-regulated genes influence biofilm phenotypes. Overall, we conclude that none of these genes, individually at least, is responsible for decreased biofilm formation by D16550 cells.
Combinatorial deletions of downregulated genes in D16550 cells minimally impact biofilm phenotypes. We next considered the possibility that two or more of the downregulated genes may act in concert to decrease biofilm matrix levels. If so, deleting two or more of these genes together might have a greater phenotypic impact. Beginning with a DamrZ strain, we sequentially added in-frame deletions of each of the downregulated genes. Neither a strain with all five downregulated genes deleted nor any of the intermediate strains with fewer deletions had a visually discernible effect on colony morphology ( Fig. 2A). Accordingly, these strains also showed no salient alterations in Pel matrix levels ( Fig. 2B) or attached biofilm mass (Fig. 2C). Having generated the quintuple mutant lacking all of the downregulated genes, we additionally deleted the sole upregulated gene, 36820, to test its impact on biofilm phenotypes. While we did not observe any changes to colony morphology ( Fig. 2A) or Pel levels ( Fig. 2B), 36820 deletion in the quintuple-mutant background had a modest but significant positive effect on surface-attached biofilm mass (Fig. 2C), hinting at a possible negative impact of the 36820 gene product on biofilm formation. 36820-associated phenotypes are consistent with a role in negative regulation of biofilm formation. A negative impact of 36820 on biofilm formation would be consistent with its status as the only upregulated gene in the DamrZ D16550 strain (Fig. 1A). If greater expression of 36820 were responsible for decreasing biofilm formation in DamrZ D16550, we would expect that deletion of 36820 might restore wrinkled morphology to DamrZ D16550 cells. However, deletion of 36820 in this strain background did not restore colony wrinkling (Fig. 3A) or increase Pel levels as judged by Congo red binding (Fig. 3B), but the deletion did significantly increase attached biofilm mass (Fig. 3C). As in the quintuple-mutant background ( Fig. 2B and C), this result was consistent with 36820 deletion having a greater impact on surface-attached biofilms than on agar-grown colony biofilms.
As a further test of 36820 as a potential negative regulator of biofilm formation, we overexpressed 36820 in the biofilm-forming DamrZ background using an arabinose-inducible pJN105 plasmid. In alignment with a negative role for 36820, and even without arabinose induction, strains bearing pJN105-36820 displayed a smooth colony morphology relative to DamrZ with the empty vector (Fig. 3D). Consistent with this visual phenotype, Pel levels were significantly decreased in the presence of plasmid-borne 36820 (Fig. 3E), as was attached biofilm mass (Fig. 3F), suggesting that 36820 is indeed a negative regulator of biofilm formation. Such a role for 36820 is also supported by other data. For example, in our own transcriptomic comparison of wild-type and DamrZ strains, 36820 stood out as one of the most downregulated genes in DamrZ (Fig. S2). Hence, restoration of 36820 expression in an DamrZ D16550 may at least partially explain the reduction in biofilm phenotype when 16550 is deleted.
Deletion of genes differentially regulated in D16550 cells impacts motility. As a final phenotypic test of strains deleted for 16550-regulated genes, we examined Microbiology Spectrum motility, as the inverse relationship between biofilm formation and motility is well studied (37,38). We tested the ability of the generated mutants to swim in 0.3% agar (Fig. 4A). As controls, we used a hypermotile (and biofilm-defective) (38) DpelA strain and a motility-defective (39) DfliC strain. Deletion of any of the five D16550-downregulated genes in a wild-type background significantly increased cell motility (as assessed by plate coverage) ( Fig. 4A and B), which was at least consistent with decreased biofilm. A notable exception was D36820, concordant with a role in negatively regulating biofilm formation, as its deletion would be expected to abet biofilm formation rather than enhancing motility. Collectively, our battery of phenotypic tests of strains with manipulations to 16550-regulated genes suggested a negative biofilm regulatory role for 36820 and individually weak but potentially positive biofilm regulatory roles for the remaining genes.
A screen for biofilm restoration uncovered RecA. Besides analyzing candidate 16550-regulated genes from transcriptomic data, we additionally took a second approach to identify genes that could reverse biofilm suppression in a DamrZ D16550 strain. We initiated a transposon mutagenesis screen to identify genes that, when inactivated in a DamrZ D16550 parental strain, would reverse the smooth colony morphology associated with 16550 deletion, restoring wrinkled colony morphology. Of approximately 3,000 transposon mutants, we recovered approximately 95 mutants with enhanced colony wrinkling, representing a hit rate of just under 3%. Surprisingly, this screen uncovered recA, encoding the broadly conserved RecA recombination protein. Transposon insertion in recA produced a wrinkled colony morphology in the DamrZ D16550 strain (Fig. 5A), with significantly higher Pel abundance than in the parental strain (Fig. 5B). An in-frame markerless deletion of recA in the same parental background (DamrZ D16550) showed the same restoration of colony wrinkling (Fig. 5A), and complementation of the recA gene at the neutral attB locus resulted in a smooth colony (Fig. 5A), indicating that recA is responsible for the biofilm phenotype.  We then asked whether the biofilm-enhancing effect of recA deletion was specific to the DamrZ D16550 background. When we deleted recA in the wild-type PA14 background, which normally displays smooth colony morphology in our colony morphology assay, we observed significant enhancement of colony wrinkling (Fig. 5A) and likewise elevated Pel levels (Fig. 5B). This result implied that the effect of recA deletion is not specific to D16550 strains, but rather that recA deletion is more generally associated with enhanced biofilm formation. In agreement with RecA being unconnected with 16550, the expression levels of recA in our transcriptomic data were not significantly influenced by deletion of 16550 (downregulated 1.38-fold; P = 0.23). Notably, we also recovered a transposon insertion in recF with increased colony wrinkling, and an in-frame deletion of recF likewise showed enhanced colony wrinkling across wild-type, DamrZ, and DamrZ D16550 strain backgrounds (Fig. S3). Our identification of RecF bolstered the notion that the loss of recombination and repair-related proteins like RecA and RecF can impact biofilm formation.  Disabling homologous recombination increases biofilm levels. RecA is a ubiquitous protein that is present in most bacteria and plays a key role in two cell processes, namely, homologous recombination (HR) and the DNA damage response (SOS response) (40,41). During HR, RecA promotes strand exchange by promoting invasion of homologous double-stranded DNA (41). Upon DNA damage, RecA activates the SOS response by functioning as a coprotease to promote cleavage of LexA, a transcriptional repressor that normally holds the SOS response in an off state. Because RecA has two distinct functions, we wondered whether loss of one or both of these functions was associated with elevated biofilm production. We first turned our attention to the HR function of RecA. To specifically inactivate its HR function, we constructed a version of recA encoding a previously described amino acid substitution (with an Asn-to-Asp change at position 303; RecA N303D ) (40). This substitution was used in E. coli to specifically disable HR while not interfering with its role in the SOS response, as a strain bearing recA N303D was not as sensitive to UV irradiation as SOS mutants were (40). Because RecA N303D has not to our knowledge been reported in P. aeruginosa, we constructed strains containing an allelic replacement, substituting recA N303D for the wild-type gene at the native locus (Fig. 6A). We assessed loss of HR by measuring transformation efficiency with a nonreplicative plasmid, pEXG2, containing a 1.2-kb region of homology with the PA14 chromosome at the ptsN locus (chosen for being unrelated to this study), which requires HR for integration. As a control for the efficiency of the mating process, we used a different nonreplicative plasmid, pCTX-1, which integrates via site-specific recombination. As we expected, strains carrying recA N303D displayed markedly impaired integration of pEXG2 (efficiencies of 1.18 Â 10 27 to 3.17 Â 10 27 ,

PA14_36820 and RecA Lower Biofilm in PA14
Microbiology Spectrum compared to 0.1 Â 10 25 to 7.8 Â 10 25 for the wild type), whereas integration of pCTX-1 was unaffected (Fig. 6B). Importantly, the recA N303D strain was less sensitive to UV than the recA deletion (Fig. S4A), indicating that SOS activation was not abolished. To test the role of HR in the enhanced-biofilm phenotype seen in DamrZ D16550 DrecA (Fig. 5A), we examined the colony morphology of recA N303D strains in both the wild-type (Fig. 6C) and DamrZ D16550 (Fig. S4B) backgrounds. In both cases, the HR-disabled strains showed colony wrinkling (Fig. 6C, Fig. S4B) and Pel levels (Fig. 6D, Fig. S4C) that were visibly enhanced relative to levels in their recA 1 counterparts. These data suggested an association between disabled HR and the enhanced biofilm phenotype of DrecA strains.
Disabling the SOS response increases biofilm levels. We then examined the impact of the second function of RecA, in activating the SOS response, on biofilm formation. To specifically disable the SOS response, we constructed a previously reported version of lexA encoding an amino acid substitution that disables cleavage of LexA (LexA S125A ) (42). A lexA S125A mutant in P. aeruginosa had been previously reported to be hypersensitive to UV irradiation and was used to define the SOS regulon in this species (42). We confirmed that this mutation did not substantially affect HR (Fig. 7A), but it did affect UV sensitivity (Fig.  S4A). We then examined the colony morphology of lexA S125A strains in the wild-type (Fig. 7B) and DamrZ D16550 ( Fig. S4B and C) strain backgrounds. As in HR-disabled strains, SOS-disabled strains showed greater colony wrinkling (Fig. 7B, Fig. S4B) and greater Pel abundance (Fig. 7C, Fig. S4C) than lexA 1 strains, with similar effects to the full recA deletion. To confirm that both functions are important for the increase in biofilm formation in DrecA, we introduced both point mutations in the same background (PA14 lexA S125A recA N303D ). The double mutant displayed increased colony wrinkling (Fig. 7B) and Pel matrix levels, but combining both mutations did not have an additive effect on Pel levels (Fig. 7C). Collectively, our results argued that loss of either function of RecA, HR or SOS activation, can contribute to the increased biofilm formation observed in strains lacking recA.
Deletion of RecA increases biofilm formation independently of cellular c-di-GMP levels. Because c-di-GMP is an important second messenger in biofilm formation, we finally asked whether deletion of recA impacted c-di-GMP levels. Deletion of 16550 from a DamrZ strain was previously reported to result in substantially lower c-di-GMP levels (9). Is the increased colony wrinkling and matrix production in DamrZ D16550 DrecA (Fig. 5A and B) associated with increased c-di-GMP? Interestingly, deletion of recA did not have any substantial effect on c-di-GMP levels relative to those in DamrZ D16550, with the DrecA strain trending toward lower [c-di-GMP] (Fig. 8). The fact that both DamrZ D16550 DrecA and DamrZ D16550 exhibited similar c-di-GMP levels but had distinct phenotypes suggested that the biofilm effects of recA deletion may be mediated via pathways other than c-di-GMP signaling.

DISCUSSION
In this study, transcriptomic analysis revealed that deletion of 16550 significantly changed the expression of only a small set of six genes (Table 1). Constructing individual deletions of these genes and assessing the colony morphology of the resulting strains to test their potential impact on biofilm formation suggested that two of these genes (primarily 36820 but also 72360) appear to have an individual impact on biofilm formation. Interestingly, we also found that the genes each impacted motility in a manner consistent with their putative roles in biofilm formation, namely, 36820 negatively impacted biofilm formation and the remaining genes positively impacted biofilm formation. The strongest role we identified was for 36820, as its ectopic expression lowered biofilm levels (Fig. 3D to F), and its deletion increased biofilm formation under some conditions ( Fig. 2C and 3C). Moreover, 36820 was strongly downregulated when amrZ was deleted ( Fig. S2A and B). Curiously, and in contrast to our findings, its ortholog in PAO1, PA2146, has been characterized as a gene that is required for biofilm formation and resistance to tobramycin, as its inactivation suppresses biofilm formation and renders biofilms susceptible to tobramycin (43). This discrepancy may be due to variations between PAO1 and PA14 or different methods of biofilm formation (colonies versus continuous flow reactors). In any case, the relatively small set of genes regulated by 16550 is consistent with 16550 controlling PA14_36820 and RecA Lower Biofilm in PA14 Microbiology Spectrum specific aspects of P. aeruginosa physiology, and further exploring other functions of these genes, especially 36820, will be an interesting topic for future work.
We also found here that reversal of biofilm suppression in a D16550 strain could be achieved by disruption of genes that were not under 16550 control. Using a mutagenesis and visual screening approach (9) to find genes whose deletion could reverse biofilm suppression in a D16550 strain, we unexpectedly identified the recombination protein-encoding gene recA. Because RecA functions both in homologous recombination and in the SOS response, we genetically dissected these functions and separately assessed their contributions to biofilm phenotypes. Our data argued that both functions of RecA contribute to the increased biofilm formation observed in a recA deletion mutant. How deletion of  Fig. 6D. Error bars show standard deviations for five replicates. P values (Student's t test) versus PA14: a, P = 0.00067; b, P = 0.02; c, P = 0.001; d, P = 0.0002; ns, not significant (P = 0.07). Note that the PA14 colony in panel B is identical to those in Fig. 1B, 3A, and 6C and that the DrecA colony is identical to that in Fig. 6C, as these are representative of common controls in one large experiment shown across multiple figures.

PA14_36820 and RecA Lower Biofilm in PA14
Microbiology Spectrum recA results in increased biofilm formation remains unanswered, although our data suggest that it does not involve alterations in c-di-GMP levels. Moreover, because RecA has two distinct functions and disabling either of these functions separately increased biofilm formation, loss of each function may have different downstream effects.
One possibility is that LexA may repress a gene involved in biofilm suppression. In this scenario, deletion of recA or the presence of LexA S125A would cause LexA to constitutively repress that gene, making DrecA or lexA S125A strains tend toward biofilm production. Similar logic appears to be present in Acinetobacter baumannii, where deletion of recA increases biofilm formation via bfmR, a global biofilm regulator, through a protein called UmuDAb. While A. baumannii lacks LexA (44)(45)(46), UmuDab is a known repressor (47,48) that appears unique to this species (49). However, unlike the repressor function of LexA, Ching et al. proposed that UmuDab might be an activator of bfmR (44).
Consistent with the colony wrinkling associated with deletion of recA, we also identified a transposon insertion in recF in our morphology screen as having increased colony wrinkling (Fig. S3). RecF, along with other proteins, is a recombination mediator protein that facilitates RecA loading on gapped DNA substrates, leading to the formation of the RecA nucleoprotein filament and its subsequent activation (50)(51)(52). Deletion of recF in the DamrZ D16550 and wild-type backgrounds caused a similar enhancement to colony wrinkling as did deletion of recA (Fig. S3). We speculate that impairment of RecA loading and subsequent activity in a recF mutant is the cause of its elevated biofilm formation.
High intracellular levels of c-di-GMP are typically associated with biofilm formation (33). However, our data suggest that deletion of recA increases biofilm formation independently of cellular c-di-GMP levels, as its deletion did not have any apparent effect on c-di-GMP levels relative to DamrZ D16550. One possible interpretation of this result is that deletion of recA causes changes to a local c-di-GMP pool while leaving overall cellular c-di-GMP levels unchanged, as has been previously reported in certain other cases (53). It is more likely that the biofilm effects of recA deletion might be mediated via a presently unknown pathway that does not involve c-di-GMP signaling.
The relationship between RecA and biofilm formation does not appear to be uniform across different bacterial organisms. Deletion of recA increases biofilm formation in E. coli (54). Conversely, a RecA-deficient mutant produced lower-density biofilms in Streptococcus mutans (55). Gotoh et al. reported that deletion of recA or disabling cleavage of LexA (LexA S125A ) did not increase biofilm mass production in comparison to wild type in P. aeruginosa PAO1 (56), a finding echoed by Chellappa et al. (57). A possible explanation for this difference is the different EPS produced by these common model strains. PAO1 produces an additional EPS, Psl, which has an important role in PAO1 biofilm formation, whereas PA14 does not produce Psl and instead relies only on Pel (26,58,59). Collectively, it seems that deletion of recA does not have a unanimous effect on biofilm formation. Further study is required to elucidate the mechanisms by which deficiencies in RecA function can stimulate biofilm formation in P. aeruginosa PA14.

MATERIALS AND METHODS
Strains and growth conditions. P. aeruginosa PA14 and E. coli SM10 were grown in Luria Bertani (LB)-Lennox broth (10 g/liter tryptone, 5 g/liter yeast extract, 5 g/liter NaCl) with shaking or on solid medium (fortified with 1.5% Bacto agar) at 37°C. To select for antibiotic markers, 75 mg/mL or 20 mg/mL gentamicin or 25 mg/mL tetracycline was added to media as appropriate. P. aeruginosa was selected after mating with E. coli by addition of 25 mg/mL irgasan to media or by use of Vogel-Bonner minimal medium (VBMM), which contains citrate as its sole carbon source and does not support growth of E. coli (60). The strains used in this study are listed in Table 2 and Table S1 in the supplemental material. All mutations in P. aeruginosa were markerless in-frame deletions or point mutations generated using allelic replacement with the pEXG2 vector, with counterselection on LB plates containing 6% sucrose or on no-salt LB plates containing 15% sucrose (60). All mutations were screened by diagnostic PCR (deletions) or sequencing (point mutations). Gene complementation was conducted using mini-CTX-1 integrants at the chromosomal attB locus. Plasmids and primers used in this study are listed in Tables S2 and S3, respectively, and modes of strain construction are listed in Text S1.
Transposon mutagenesis and biofilm colony morphology screen. Transposon mutagenesis and biofilm colony morphology screening were performed as previously described (9). In brief, P. aeruginosa PA14 DamrZD16550 (MTC1381) was mated with E. coli SM10 pBT24 (MTC33). We then screened for transposon mutants that had restored colony wrinkling as an indication of biofilm formation.
Biofilm assay. Colony morphology of P. aeruginosa strains was assessed on M6301 agar medium containing 1% agar, 100 mM KH 2 PO 4 , 15.14 mM (NH 4 ) 2 SO 4 , and 0.36 mM FeSO 4 ÁH 2 O (pH balanced to 7.0 using 10 M KOH) (61). After autoclaving, 1 mM MgSO 4 , 0.5% glycerol, and 0.2% Casamino Acids (BD Bacto, USA) were added. M6301 agar was made fresh before each experiment; 40 mL was poured in each plate and allowed to solidify overnight. P. aeruginosa cultures were grown overnight with shaking in 3 mL liquid LB at 37°C and then normalized to an optical density at 600 nm (OD 600 ) of 1. Droplets (2 mL) of the normalized For polysaccharide content estimation, colonies were collected on day 6 from the agar and homogenized for at least 20 s with an Argos rotary pestle (Cole-Parmer, USA) in 1 mL of sterile phosphate-buffered saline (PBS). After allowing flakes generated from resuspending the colonies to settle, 100 mL was removed from the resuspension to measure the OD 600 using a Biotek Synergy HT plate reader. The remaining suspension was pelleted using a microcentrifuge at 13,000 Â g for 3 min at room temperature. After discarding the supernatant, the pellets were resuspended in 40 mg mL 21 Congo red and agitated on a GyroMixer XL (GeneMate, VWR) for 90 min at room temperature. The samples were then centrifuged again at 13,000 Â g for 3 min at room temperature, and the OD of the supernatants was measured at 490 nm. A standard curve was generated by measuring the OD 490 of Congo red at 40, 20, 10, 5, 2, 1, and 0.5 mg mL 21 . PBS (1Â) was used as a blank and as the diluent for the standard curve.
To maximize consistency across the data presented in this study, many of the colony morphology and Congo red binding data shown were taken from the same large experiment that had common controls. Hence, the colony morphology and Congo red binding data for strain PA14 are repeated in Fig. 1B, 3A, 6C, and 7B; DamrZ data are repeated in Fig. 1B, 3A, and 6C; DamrZ D16550 data are repeated in Fig. 1B and 3A; and DrecA data appear in Fig. 6C and 7B.
Crystal violet biofilm formation assay. The crystal violet biofilm formation assay was conducted as previously described (62). In brief, strains were grown overnight in LB medium. The following day, their OD 600 was measured, and cells were normalized to an OD 600 of 0.1 and incubated in M630 medium at 37°C for 48 h. After incubating the strains for 48 h, their OD 600 was measured, and unattached cells were removed from the wells by decanting. A 96-well plate was then submerged in a small tub of deionized water 3 to 4 times and shaken to remove any excess water. Two hundred microliters of a 0.1% solution of crystal violet in water was used to stain each well in the 96-well plate. After staining the cells at room temperature for 15 min, the plate was rinsed 3 to 4 times with deionized water by submerging the plate in a small tub of water (the water was changed between washes). The plate was then dried for 20 to 30 min, and 30% acetic acid in water was added to each well to solubilize the crystal violet. To quantify the biofilm, the absorbance was measured in a plate reader at 550 nm using 30% acetic acid in water as the blank.
Motility assay. Swimming motility was assessed as previously described (63). In brief, strains were inoculated at the center of LB plates with 0.3% agar and incubated overnight at 37°C. To determine the percent surface coverage, the plate was photographed with a Canon EOS digital camera. The image was converted to grayscale and the brightness and contrast were adjusted using Photoshop software (Adobe, Mountain View, CA) to provide appropriate contrast between the agar surface and the bacterial colony. The colony was specifically selected using the "magic wand" tool in Photoshop, and the number of pixels contained in the selection (i.e., proportional to the area of the colony) was divided by the number of pixels in the entire plate (i.e., the area of the plate).
UV light sensitivity assay. Strains were grown overnight in liquid LB at 37°C. The following day, the overnight-grown cultures were normalized to an OD 600 of 1 and were serially diluted. Then, 10 mL was spotted on a plain LB agar plate and exposed to 10 mJ of UV light (approximately 10 s) using a Stratalinker UV cross-linker 2400 (Stratagene, La Jolla, CA).
Quantification of intracellular cyclic di-GMP. P. aeruginosa PA14 and derivative strains were grown overnight in 3 mL LB liquid shaking culture at 37°C. Cultures were further spotted on M6301 agar as described above. After 3 days of growth at 25°C, colonies were collected for cyclic di-GMP extraction. Using a sterile spatula, the colonies were collected and homogenized with 1 mL of sterile PBS in 1.5-mL microcentrifuge tubes for at least 20 s with an Argos rotary pestle. Cyclic di-GMP was extracted from either 2 or 10 OD units of cells (1 OD unit = 1 mL of cell suspension at an OD 600 of 1.0) and quantified by high-performance liquid chromatography (HPLC) as previously described (64) with a few modifications. The HPLC gradient was modified to the following: 0 to 2 min, 1% solvent B; 7 min, 10% B; 12 min, 15% B; 17 min, 18% B; 22 min, 20% B; 27 min, 22% B; 32 min, 24% B; 37 min, 26% B. This gradient resulted in the elution of c-di-GMP at approximately 6.4 to 6.5 min. The concentration of c-di-GMP in each sample was calculated from a standard curve prepared using commercially available pure c-di-GMP. In some cases, the c-di-GMP concentration in the extract was normalized to the total amount of protein from the same sample (calculated using the Bradford method with bovine serum albumin standards).
Transformation efficiency assay. The transformation efficiency of P. aeruginosa PA14 and derivative strains was assessed by mating with E. coli SM10 containing pEXG2 bearing approximately 1.2 kb of homology to the ptsN locus, whose integration into the host chromosome requires homologous recombination. P. aeruginosa PA14 and Escherichia coli SM10 pEXG2-ptsN were mated as previously described (60). In brief, 0.5 mL of the P. aeruginosa recipient strain and 1.5 mL of the E. coli donor strain were placed in separate 2-mL microcentrifuge tubes and centrifuged at 10,000 Â g for 5 min at room temperature. After discarding the supernatant, each pellet was resuspended in 50 mL of LB. Both 50-mL cell mixtures were then combined and placed on a prewarmed LB agar plate and incubated overnight at 30°C. Using a sterile spatula, the mating matrix was collected and resuspended in 1 mL of LB. The OD 600 of the mating matrix of all the strains tested was normalized to an OD 600 of 10 and further serially diluted. Dilutions (100 mL) were spread on LB-agar plates and incubated overnight at 37°C before manual enumeration of CFU.
RNA isolation and sequencing. Total RNA was isolated from homogenized colonies grown in quadruplicate on M6301-1% agar plates for 3 days at 25°C using the New England Biolabs Monarch total RNA miniprep kit. Quality-control steps, rRNA depletion, Illumina library preparation, and 150-bp paired-end high-throughput Illumina sequencing were performed by Novogene (Beijing, China). Sequence mapping and analysis were performed at the Oklahoma University Health Sciences Center Laboratory for Molecular Biology and Cytometry Research using CLC software. Complete lists of differentially regulated genes for PA14 versus DamrZ and for DamrZ versus DamrZ D16550 comparisons are provided in Data Set S1.
Statistical comparisons. All pairwise comparisons were statistically analyzed using two-tailed Student's t tests, assuming equal variance.
Data availability. The full transcriptomic data sets, including the raw sequence reads, have been deposited in the Gene Expression Omnibus at NCBI under accession number GSE226104.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.9 MB. SUPPLEMENTAL FILE 2, XLSX file, 1.2 MB.

ACKNOWLEDGMENTS
We thank members of the Cabeen lab and our departmental colleagues for helpful discussions. We also appreciate Steve Hartson and Simon Underhill for their technical assistance with HPLC experiments. This research was supported by the Oklahoma Center for the Advancement of Science and Technology (grant HR20-048 to M.T.C.).