Pseudomonas aeruginosa GacA, a factor in multihost virulence, is also essential for biofilm formation

Authors


Abstract

We have investigated a potential role for GacA, the response regulator of the GacA/GacS two-component regulatory system, in Pseudomonas aeruginosa biofilm formation. When gacA was disrupted in strain PA14, a 10-fold reduction in biofilm formation capacity resulted relative to wild-type PA14. However, no significant difference was observed in the planktonic growth rate of PA14 gacA. Providing gacA in trans on the multicopy vector pUCP-gacA abrogated the biofilm formation defect. Scanning electron microscopy of biofilms formed by PA14 gacA revealed diffuse clusters of cells that failed to aggregate into microcolonies, implying a deficit in biofilm development or surface translocation. Motility assays revealed no decrease in PA14 gacA twitching or swimming abilities, indicating that the defect in biofilm formation is independent of flagellar-mediated attachment and solid surface translocation by pili. Autoinducer and alginate bioassays were performed similarly, and no difference in production levels was observed, indicating that this is not merely an upstream effect on either quorum sensing or alginate production. Antibiotic susceptibility profiling demonstrated that PA14 gacA biofilms have moderately decreased resistance to a range of antibiotics relative to PA14 wild type. This study establishes GacA as a new and independent regulatory element in P. aeruginosa biofilm formation.

Introduction

Biofilms are adherent microcolonies of bacteria, in which groups of cells are embedded within a complex and highly heterogeneous extracellular polymeric matrix (Costerton et al., 1995). These consortia represent a unique mode of bacterial growth that is fundamentally different from planktonic or free-swimming cells. Infections resulting from pathogenic biofilms are characterized by a chronic or recurrent nature and are highly resistant to conventional treatments (Costerton et al., 1999). The basis for the persistent nature of biofilm infections is multifactorial. One factor is that the biofilm mode of growth affords a certain degree of resistance to the host immune response (Jensen et al., 1990; 1992; 1993; Anwar et al., 1992; Meluleni et al., 1995). A second factor is that biofilms have a two to three order of magnitude decrease in susceptibility to antimicrobials compared with planktonically grown bacteria (Ceri et al., 1999; Costerton et al., 1999).

Pseudomonas aeruginosa, the archetypical opportunistic pathogen, has become the organism of choice for studying the physiological and genetic basis of biofilm formation and phenotype. The suitability of this organism as a model system for understanding biofilms is threefold: it has a propensity for developing a variety of biofilm infections ranging from cystic fibrosis chronic pulmonary infections (Lam et al., 1980; Singh et al., 2000) to implant device-related infections (Costerton et al., 1999). Secondly, the bacterium is easily manipulated in the laboratory owing to its rapid rate of growth, sparse nutritional requirements and ability to develop highly resistant biofilm structures. Finally, genetic analysis of P. aeruginosa has been simplified as a result of the availability of the genome sequence (Stover et al., 2000).

Genetic screens based on impaired attachment have identified a number of factors required for initial biofilm formation, including flagella, Clp intracellular protease and many genes of unknown function (O'Toole and Kolter, 1998a). Further studies have demonstrated that type IV pili-mediated surface motility, known as twitching, is required for microcolony aggregation, a secondary step required in biofilm maturation (O'Toole and Kolter, 1998b; O'Toole et al., 2000). Davies et al. (1998) have demonstrated further that the las quorum-sensing system, but not the rhl system, is required for the development of a mature biofilm architecture and for biofilm biocide resistance. However, Singh et al. (2000) have shown that rhl quorum-sensing signal is elevated in biofilms relative to planktonically grown cells.

Of particular interest to our laboratories are the global regulatory genes that could influence the shift to and the maintenance of biofilm growth. These regulators may serve as targets for future antimicrobial therapies. As such, regulators that play a role in the virulence of the microbe as well as in its ability to form biofilms are of particular interest. A number of these regulators have been identified. LasR, the transcriptional activator of the las quorum-sensing system, plays a role in both biofilm formation (Davies et al., 1998) and virulence (Tang et al., 1996; Rumbaugh et al., 1999a,b). Recently, polyphosphate kinase (Rashid et al., 2000) and the crc global carbon metabolism regulator (O'Toole et al., 2000) were added to this group of genes. Our goal was to determine whether other loci might also fit into a group of genes that are regulators of both biofilm formation and virulence.

The GacA/S two-component global regulatory system has been demonstrated to be an essential virulence factor for P. aeruginosa pathogenesis independently in animal, plant, nematode and insect models of infection (Rahme et al., 1995; 1997; Mahajan-Miklos et al., 1999; Tan et al., 1999a,b; Jander et al., 2000). This regulatory system is composed of GacS, the histidine kinase sensor protein (Barta et al., 1992), and the cognate response regulator GacA, which has significant homology to the FixJ family of regulatory proteins (Rich et al., 1994). Although the signal to which GacS responds remains unknown, GacA has been shown to exhibit post-transcriptional regulatory control of genes within the Gac regulon (Blumer et al., 1999). In P. aeruginosa, GacA has been demonstrated to regulate positively the production of several virulence factors, specifically N-butyryl-l-homoserine lactone, pyocyanin, cyanide and lipase (Reimmann et al., 1997). However, studies in other microorganisms have implicated much broader ranging effects, including regulation of toxins (Barta et al., 1992; Rich et al., 1994; Kitten et al., 1998), proteases (Liao et al., 1994; Grewal et al., 1995), type III secretion (Hirano et al., 1999), alginate biosynthesis (Liao et al., 1996; Castaneda et al., 2000), secondary metabolites (Whistler et al., 1998), siderophores (Liao et al., 1996; Zhang and Normark, 1996), swarming (Kinscherf and Willis, 1999) and invasion (Johnston et al., 1996). Despite the diversity of functions regulated by GacA/GacS, the unifying theme that can be observed is that most products are extracellular and tend to play a role in the modification of the surrounding environment.

Because of the broad range of activities controlled through the gac regulon and its involvement in extracellular product formation, we explored the potential involvement of the GacA/GacS two-component regulatory system in biofilm formation in P. aeruginosa. In this study, we demonstrate that GacA plays a critical role in biofilm formation, independent of factors previously identified to function in biofilm development. Furthermore, biofilms formed by a gacA-deficient strain of P. aeruginosa PA14 displayed a reduction in resistance to several different classes of antimicrobial agents. Together, these data establish that the gacA regulon acts to mediate biofilm formation through a novel pathway. It is also exciting to identify a factor key for the pathogenesis of P. aeruginosa involved in biofilm formation, as this may suggest that biofilms play a role in several previously identified infection models.

Results

A gacA mutant strain of P. aeruginosa is impaired in biofilm formation

To determine whether gacA plays a role in biofilm formation, we compared the biofilm formation ability of a gacA mutant strain of P. aeruginosa, PA14 gacA, with two control strains, PA14 wild type and PA14 toxA. The PA14 toxA strain was used as an additional control in these studies, as it was engineered using the same GmR cassette used for the construction of PA14 gacA(Rahme et al, 1995). Each population was grown in tryptic soy broth (TSB) in a minimum biofilm eradication concentration (MBEC) device (MBEC Biofilm Technologies) with sampling over a 24 h growth period. Biofilm samples were obtained from the pegs of the MBEC device, and planktonic samples were taken directly from the growth vessel. Examination of growth rates over a 24 h time period revealed that P. aeruginosa PA14 gacA was defective in biofilm formation, as it formed biofilms at a reduced rate and with a 10-fold reduction in final cell number (Fig. 1A). This was not caused by a defect in growth, as planktonic populations of P. aeruginosa PA14 gacA proliferated at the same rates as PA14 wild type and PA14 toxA(Fig. 1B). Interestingly, there was no difference in biofilm formation or growth between PA14 wild type and PA14 toxA. This suggested that toxA probably does not play a role in biofilm formation and that the genetic manipulations carried out on both PA14 toxA and PA14 gacA did not influence the ability of the strains to form biofilms.

Figure 1.

A. Biofilm formation rates. Biofilm growth curves were performed on P. aeruginosa strains PA14 wild type (◆), PA14 toxA (▪) and PA14 gacA (▴) in the MBEC device over a 24 h period.

B. Planktonic population growth rates. Proliferation rates of the planktonic populations of P. aeruginosa strains PA14 wild type (◆), PA14 toxA (▪) and PA14 gacA (▴) in the MBEC device over a 24 h period.

To add further support to the role of gacA in biofilm formation, complementation studies were performed by transforming PA14 wild-type, PA14 toxA and PA14 gacA strains with pUCP-gacA. Biofilm growth rates of each transformed strain are shown in Fig. 2. Biofilm formation by PA14 gacA was restored when complemented in trans with pUCP-gacA; however, overexpression of gacA did not increase the biofilm formation rates of PA14 wild type or PA14 toxA(Fig. 2). Overexpression of gacA also did not significantly affect planktonic growth in any of the strains (data not shown). When transformed with only the vector control pUC181.8, strain PA14 gacA maintained its biofilm-deficient phenotype (data not shown). Biofilm formation rates of strains PA14 wild type and PA14 toxA were not affected by the presence of the control vector alone. These results suggest that gacA is required for optimal biofilm formation and may regulate genes involved in biofilm development.

Figure 2.

Biofilm growth rates of PA14 strains overexpressing gacA. The rate of biofilm growth of P. aeruginosa strains PA14 wild type (pUCP-gacA) (◆), PA14 toxA (pUCP-gacA) (▪) and PA14 gacA (pUCP-gacA) (▴) were monitored for a 24 h period in the MBEC device.

A gacA mutant of P. aeruginosa fails to aggregate to form microcolonies and mature biofilm structures

To determine whether the biofilm formation defect of P. aeruginosa PA14 gacA was accompanied by morphological changes to the structure of the biofilms it formed, we performed scanning electron microscopy (SEM) on biofilms formed by P. aeruginosa PA14 wild type, PA14 toxA and PA14 gacA. Biofilms of each strain were grown in the MBEC device for 24 h. Interestingly, differences in the amount of growth on the pegs could be seen unaided, with the wild-type cells producing large visible biofilms compared with a significant lack of a cell mass with PA14 gacA. Visualization of the growth on the pegs by SEM was used to assess whether differences in biofilm architecture existed. Figure 3A–F shows that clear differences occurred in the biofilms formed by the three strains. PA14 gacA cells did adhere to the surface of the peg, but failed to aggregate and form microcolonies, even after 24 h of growth. We also observed that, whereas PA14 wild-type and PA14 toxA colonization of pegs was restricted for the most part to large mats of biofilms, PA14 gacA colonization was sparse but uniformly distributed on the pegs. Thus, microscopic examination of biofilms formed by P. aeruginosa PA14 gacA also suggested that gacA is required for biofilm development.

Figure 3.

High- and low-magnification scanning electron micrographs of biofilms formed by PA14 wild type (A and B), PA14 toxA (C and D) and PA14 gacA (E and F) after 24 h of growth. PA14 gacA adheres to the peg; however, it fails to aggregate to form microcolonies and develop mature biofilm structures. (A), (C) and (E), high magnification; (B), (D) and (F), low magnification.

Characterization of the biofilm formation defect of P. aeruginosa PA14 gacA

The gacA/gacS regulons of several Pseudomonas spp. have been demonstrated to influence a number of factors, including toxin and protease production and secretion, quorum sensing, alginate biosynthesis and swarming (Barta et al., 1992; Liao et al., 1994; 1996; Rich et al., 1994; Grewal et al., 1995; Kitten et al., 1998; Kinscherf and Willis, 1999; Castaneda et al., 2000). Although it is unlikely that proteases and toxins play a role in biofilm formation, there is evidence that extracellular polysaccharides, quorum sensing and surface-associated motility are required for biofilm formation and development. Thus, we wanted to determine whether the gacAmediated defect in biofilm formation acts through any of the factors previously identified as being involved in biofilm formation.

Effect of gacA on quorum sensing in P. aeruginosa Previous reports have suggested that GacA acts to enhance the transcription of lasR and thus influences autoinducer production (Reimmann et al., 1997). As such, disruption of gacA may result in decreased LasR production and, hence, autoinducer production and, therefore, may explain the biofilm formation defect of P. aeruginosa PA14 gacA. To determine whether autoinducer production is altered in strain PA14 gacA, we measured the levels of autoinducer produced by each strain. The reporter strain Escherichia coli MG4 (pKDT17) was used to measure the amount of N-3-oxododecanoyl-l-homoserine lactone (3-oxo-C12-HSL) in the supernatants of stationary phase cultures of PA14 wild type, PA14 toxA and PA14 gacA(Table 1). The level of 3-oxo-C12-HSL produced by PA14 gacA is only slightly, and not significantly, diminished relative to PA14 wild type or PA14 toxA. Likewise, because GacA has been reported to regulate N-butyryl-l-homoserine lactone (C4-HSL) production positively (Reimmann et al., 1997), we used the reporter strain E. coli DH5α (pECP61.5) to monitor the production of C4-HSL in the PA14 strains (Table 1). The production of C4-HSL was also only slightly decreased in PA14 gacA relative to the two control strains. We have also measured the timing of autoinducer production in strains PA14 wild type and PA14 gacA, and these strains do not differ dramatically in the timing of either C4-HSL or 3-oxo-C12 HSLs (S. Sandhu and D. G. Storey, unpublished observation). These results suggest that autoinducer production in strain PA14 is not dramatically influenced by a mutation in GacA.

Table 1.  Physiological properties of each PA14 derivative.
CategoryPA14 wild typePA14 toxAPA14 gacA
Motility (mm)
 Twitching (24 h)8.5 ± 1.210.5 ± 0.79.3 ± 1.0
 Twitching (48 h)17.3 ± 1.318.7 ± 1.517.7 ± 1.5
 Swarming (72 h)22 ± 7.516 ± 1158 ± 11
 Flagellar swimming (72 h)46 ± 8.646 ± 3.247 ± 6.3
Autoinducer production (Miller units)
 3-oxo-C12-HSL10 568 ± 299091 ± 378852 ± 25
 C4-HSL165.5 ± 6.17158.5 ± 3.8149.4 ± 3.3
Alginate production (mg alginate mg−1 protein)0.614 ± 0.34NA0.771 ± 0.54

Previous research has shown that, if lasR is present on a multicopy plasmid (pMJG1.7), LasR and autoinducer production is increased (A. Kirkham and D. G. Storey, unpublished data). Thus, to compensate for the disruption of a potential positive regulator of lasR, we transformed the lasR overexpression vector pMJG1.7 into PA14 wild type, PA14 toxA and PA14 gacA. We then examined the resulting biofilm and planktonic growth rates. The presence of multiple copies of lasR did not bypass the biofilm formation defect of PA14 gacA, as P. aeruginosa PA14 gacA (pMJG1.7) produced biofilms at the same decreased rate (≈10-fold lower than PA14) as PA14 gacA (data not shown). Furthermore, overexpression of lasR in strains PA14 wild type (pMJG1.7) and PA14 toxA (pMJG1.7) did not increase the biofilm formation ability of these strains. Taken together, these experiments suggest that, in strain PA14, an alteration in quorum sensing could not explain the decrease in biofilm formation that we see with PA14 gacA.

Effect of gacA on motility of P. aeruginosa O'Toole and Kolter (1998b) demonstrated that twitching motility was necessary for microcolony aggregation, an initial step in biofilm formation. The GacA/GacS two-component regulatory system has been shown to regulate the solid surface translocation of Pseudomonas syringae (Kinscherf and Willis, 1999). To assess whether the PA14 gacA biofilm formation defect was mediated through a defect in solid surface translocation, twitching motility and swarming assays were performed on each PA14 strain. P. aeruginosa PA14 wild type, PA14 toxA and PA14 gacA were stab inoculated into thin agar-rich media plates, and the zones of twitching were monitored after 24 and 48 h. Zones of twitching were identical for each strain tested, indicating that there is no defect in type IV pili-mediated twitching motility in P. aeruginosa PA14 gacA(Table 1). Swarm assays similarly did not show that PA14 gacA had any impairment in its ability to swarm (Table 1). Notably, the mutation in gacA seems to enhance the ability of the bacteria to swarm (Table 1). As such, the biofilm formation defect of strain PA14 gacA is probably not mediated by a decrease in solid surface translocation.

Flagellar function has similarly been shown to be necessary for initial bacterial attachment and subsequent biofilm formation (O'Toole and Kolter, 1998a). We therefore assessed the flagellar function of each of the three strains using flagellar swim plates. No difference in flagellar function was observed among the strains, indicating that the PA14 gacA biofilm formation defect is not mediated through this pathway (Table 1).

Effect of gacA on alginate production in P. aeruginosa

Studies in other bacteria have demonstrated that alginate production is upregulated by the gacA/gacS regulon (Liao et al., 1996; Castaneda et al., 2000). To assess whether the biofilm formation defect of PA14 gacA was caused by altered alginate production, alginate bioassays were performed on each of the strains. Table 1 shows that PA14 gacA had a slight, but not significant, increase in alginate production over the wild-type strain.

Effect of gacA on biofilm antibiotic susceptibility of P. aeruginosa PA14 strains

The fundamental feature associated with biofilm growth is drastically increased resistance to antibacterial agents. As such, the disruption of a genetic factor required for biofilm formation may result in a corresponding decrease in resistance to an antibiotic challenge. To examine whether the biofilm formation defect observed in PA14 gacA resulted in an altered antimicrobial resistance profile, MBEC antimicrobial susceptibility testing was performed (Ceri et al., 1999). Minimum inhibitory concentration (MIC) and MBEC values were determined based on absorbance readings of the antibiotic challenge plate and the recovery plate respectively. Little difference in planktonic antibiotic susceptibility was observed between the genetic backgrounds. A moderate decrease in biofilm antibiotic resistance to azythromycin, chloramphenicol, erythromyin, piperacillin and tetracycline was observed for PA14 gacA relative to PA14 wild type and the PA14 toxA control (Table 2). In order to measure the number of viable biofilm cells remaining after antibiotic challenge at each concentration, a subsequent MBEC assay was performed on the two isogenic strains PA14 toxA and PA14 gacA, in which biofilm cells released by sonication into the recovery plate were serial diluted and spot plated to determine remaining viable cfu/peg (Fig. 4). The number of viable biofilm cells remaining after antibiotic exposure was significantly decreased in PA14 gacA relative to PA14 toxA for all antibiotics tested: azythromycin, chloramphenicol, erythromycin, polymyxin B and tobramycin. We postulate that the observed decrease in biofilm antibiotic resistance of strain PA14 gacA is the result of an inherent biofilm formation defect of this strain, resulting from the disruption of a key regulatory element required for biofilm formation and antibiotic resistance.

Table 2.  Antibiotic susceptibility profile of P. aeruginosa PA14 wild type, PA14 toxA and PA14 gacA.
AntibioticPA14 WTPA14 toxAPA14 gacA
MICMBECMICMBECMICMBEC
Azythromycin64≥ 102464≥ 102464512
Chloramphenicol64≥ 102464≥ 1024321024
Ciprofloxacin≤ 22≤ 22≤ 2≤ 2
Erythromycin128≥ 1024128≥ 102464512
Piperacillin16≥ 102464≥ 102416256
Tetracycline165121651216256
Tobramycin≤ 216≤ 216≤ 216
Figure 4.

Biofilm antibiotic resistance. Biofilms of P. aeruginosa PA14 toxA (hatched) and PA14 gacA (solid) were challenged for 16–20 h with various concentrations of five antibiotics and then sonicated to release individual cells. Surviving biofilm size was determined after serial dilution, and the concentration of antibiotic required to produce a 1000-fold reduction in biofilm mass (relative to control biofilms) is plotted.

Discussion

The virulence of P. aeruginosa is multifaceted. Numerous virulence determinants are involved, and we are just beginning to realize the importance that the biofilm mode of growth plays in infections (Costerton et al., 1999). Thus, an understanding of genes that are involved in both virulence and the biofilm mode of growth may lead to new classes of antibacterial agents with efficacy against biofilms at physiologically achievable levels. GacA, part of the GacA/S two-component global regulatory system, has been shown to be involved in the virulence of P. aeruginosa in a wide range of organisms (Rahme et al., 2000), but its role in biofilm formation was previously unrecognized. In this study, we identified a role for GacA in biofilm formation and have also proceeded to examine the nature of this system in P. aeruginosa biofilm formation.

Reduced biofilm formation has been shown to result after the disruption of a number of genes involved in surface attachment and in the early stages of biofilm formation (Davies et al., 1998; O'Toole and Kolter, 1998a). Mutants with disruptions in key regulatory genes also seem to have reduced ability to form biofilms (Davies et al., 1998; O'Toole et al., 2000). Similarly, we showed that a gacA mutant has a 10-fold decrease in the ability to form biofilms (Fig. 1A). This was not a general growth defect, as PA14 gacA grows as well as wild-type cells during planktonic growth (Fig. 1B). Furthermore, direct observation and comparison of PA14 gacA with PA14 wild type revealed that the gacA mutant attached to surfaces but did not progress beyond the accumulation of a few cells (Fig. 3). In contrast, PA14 wild type and toxA formed dense multilayered biofilms. Taken together, these observations indicate that GacA plays an important role in biofilm formation.

Because the gacA/gacS regulon in other Pseudomonads has been demonstrated to affect solid surface translocation, alginate biosynthesis and autoinducer production (Kitten et al., 1998; Chancey et al., 1999; Kinscherf and Willis, 1999), and these functions have also been shown to be involved in P. aeruginosa biofilm formation (Boyd and Chakrabarty, 1995; Davies et al., 1998; O'Toole and Kolter, 1998b), we wished to assess whether the gacA defect in biofilm formation was mediated through any of these previously identified mechanisms. To examine the role of the GacA regulator on the regulation of the P. aeruginosa quorum-sensing systems, we examined autoinducer production by the three P. aeruginosa PA14 strains, as well as the effects of overexpressing LasR on the biofilm formation defect of P. aeruginosa PA14 gacA. Assays of 3-oxo-C12-HSL production showed no significant differences in either timing of production or production levels of 3-oxo-C12-HSL in PA14 gacA relative to the control strains (Table 1). Thus, our results suggested that production of 3-oxo-C12-HSL autoinducer was not dramatically altered in PA14 gacA growing planktonically.

In P. aeruginosa, the LasR–LasI−3-oxo-C12-HSL quorum-sensing system is intertwined with the RhlR–RhlI–C4-HSL quorum sensing-system. However, the role of C4-HSL in biofilms is as yet undefined. Whereas Davies et al. (1998) reported that C4-HSL is not required for biofilm formation or biocide resistance, Singh et al. (2000) have reported elevated levels of C4-HSL as being a molecular marker of biofilm growth. Thus, we examined production of C4-HSL in P. aeruginosa PA14 gacA and the two control strains using a similar bioassay. C4-HSL production is slightly but not significantly decreased in PA14 gacA(Table 1) relative to strain PA14. Again, this suggests that RhlI and C4-HSL are not altered in PA14 gacA compared with the parental strain growing planktonically. A possible explanation for the differences in autoinducer production between PAO1 and PA14 could be altered regulation or perhaps a different complement of genes between the two strains.

Reimmann et al. (1997) showed that, in strain PAO1, GacA enhanced lasR production and so influenced autoinducer production. Research in our laboratory has shown that overexpression of the LasR in P. aeruginosa acts markedly to increase the production of both 3-oxo-C12-HSL and C4-HSL (A. Kirkham and D. G. Storey, unpublished data). To overcome the potential deficit of LasR in strain PA14 gacA, we overexpressed lasR on a multiple-copy vector in PA14 gacA. We then used this strain to determine whether this could at least partially complement the biofilm formation ability of this strain. The overexpression of lasR from the vector pMJG1.7 in P. aeruginosa PA14 gacA did not bypass the biofilm formation defect of this strain (data not shown). Taken together, these studies may suggest that the biofilm formation defect caused by a disruption of the GacA system is independent of the las and the rhl quorum-sensing systems.

Twitching motility and swarming are types of solid surface translocation implicated in microcolony aggregation and subsequent biofilm formation (Pratt and Kolter, 1998; O'Toole and Kolter, 1998b; Semmler et al., 1999; Kohler et al., 2000; Rashid et al., 2000; O'Toole et al., 2000). Twitching motility is believed to be mediated through the extension and contraction of type IV pili (Bradley, 1980; Semmler et al., 1999). The electron micrographs in Fig. 3 suggest that PA14 gacA lacks the ability to translocate across the surface of the pegs and so cannot form microcolonies that would allow further maturation of the biofilm. Assays revealed identical zones of twitching for PA14 gacA and the two PA14 control strains (Table 1), indicating that twitching is not altered in a gacA mutant. Interestingly, in P. aeruginosa strain PA14, we show that a mutant in GacA has enhanced swarming ability (Table 1). This result could be explained in two ways. First, it is possible that the enhanced ability to swarm may be detrimental to biofilm development, and this is the reason that a gacA mutant cannot form a biofilm. The second possibility is that a GacA-regulated mechanism of surface translocation other than twitching and swarming motility is needed for a biofilm to develop. We currently favour this second possibility.

Both flagella function and alginate production have been demonstrated to be important in biofilm formation and development (Boyd and Chakrabarty, 1995; O'Toole and Kolter, 1998a). We performed flagellar swim tests and alginate biosynthesis assays to assess whether the gacA-mediated defect was operating through either of these pathways. There was no difference in the ability of PA14 gacA to swim, relative to the two control strains (Table 1), indicating that the gacA-mediated biofilm formation pathway is independent of flagellar motility. We also carried out alginate assays on PA14 and PA14 gacA and found only a slight increase in alginate production in the gacA-mutant (Table 1). However, given the variability of this assay on strains that produce relatively low amounts of alginate, we do not think this difference is enough to account for alterations in the ability of the strains to form biofilms.

The fundamental feature associated with biofilm growth is their recalcitrant resistance, thus making them less susceptible to antimicrobial treatments than comparable planktonic bacteria (Costerton et al., 1999). Therefore, biofilm antibiotic susceptibility profiling of P. aeruginosa PA14 gacA was performed to examine whether the biofilm formation defect translated into a decrease in antimicrobial resistance. A two- to fourfold decrease in MBEC (Ceri et al., 1999) was observed to the antibiotics azythromycin, chloramphenicol, erythromyin, piperacillin and tetracycline in PA14 gacA relative to PA14 wild type and the PA14 toxA control (Table 2). This increase in antibiotic sensitivity was not as profound during planktonic growth, as shown by similar MICs. For both the MBEC and the MIC assays, we measured the concentration at which total killing takes place. Examination of cell viability, in which we measured the antibiotic concentration that results in a three log reduction in cell numbers, proved even more interesting. P. aeruginosa PA14 gacA biofilms survived antibiotic challenge with far fewer viable cells than comparable P. aeruginosa PA14 toxA biofilms exposed to the same concentration of antibiotic (Fig. 4). The number of viable biofilm cells remaining after antibiotic exposure was significantly decreased in PA14 gacA relative to PA14 toxA for all antibiotics tested: azythromycin, chloramphenicol, erythromycin, polymyxin B and tobramycin. This trend was most pronounced with the antibiotics chloramphenicol and tobramycin. With these antibiotics, there is a three- to fourfold log reduction in biofilm survival after exposure in P. aeruginosa PA14 gacA relative to PA14 toxA over all concentrations (data not shown).

The antibiotic resistance profile of P. aeruginosa PA14 gacA was somewhat surprising. Despite a 10-fold decrease in biofilm formation and final cell mass and failure to mature into a dense bacterial biofilm, this knock-out strain still demonstrated a relatively high level of antibiotic resistance. SEM analysis revealed that little biofilm architecture was developed by this mutant strain, despite normal levels of alginate production. These data imply that the biofilm matrix, although it may serve as a diffusion barrier to antibiotics, does not account for the bulk of antibiotic resistance observed in biofilms. This is not surprising, as previous studies have demonstrated that biofilm architecture acts to decrease the diffusion rate of several antibiotic classes; however, it does not act to block penetration completely (Nichols et al., 1989; Stewart, 1994; Suci et al., 1994). It is likely that the biofilm antibiotic-resistant phenotype is cumulative and contributed by multiple factors, of which reduced permeability may be only one.

From these data, we can conclude that GacA, a factor involved in multihost virulence, plays a critical role in biofilm formation. Furthermore, the GacA regulatory system may regulate an alternative pathway required for optimal biofilm formation. In strain PA14, this regulatory system seems to be independent of the las and rhl quorum-sensing systems, alginate production and swimming and twitching motility. Interestingly, swarming is enhanced in strain PA14 gacA, suggesting that, in P. aeruginosa, swarming is repressed by GacA. At present, data are not available regarding the pathway through which the GacA/GacS two-component regulatory system acts to mediate biofilm formation. It does appear that surface translocation is altered in the PA14 gacA. Identification of the regulatory cascade through which gacA acts to affect biofilm formation potential is essential to understanding the molecular and genetic basis of biofilm development and maturation. Furthermore, identifying the signal to which GacS (LemA) responds and initiates expression of the genes within the gac regulon is required in order to understand fully the role the GacA/GacS two-component regulatory system plays in P. aeruginosa biofilm formation. The identification of gacA/gacS two-component regulatory system involvement in biofilm formation and antibiotic resistance is important to both the understanding of biofilm development and in establishing an in vitro role for factors critical in vivo.

Experimental procedures

Bacterial strains and media

Bacterial strains and plasmids used in these studies are listed in Table 3. P. aeruginosa strain PA14 and its toxA and gacA derivatives, PA14 toxA and PA14 gacA, respectively, were used in all biofilm formation studies (Rahme et al., 1995). Unless otherwise indicated, strains were grown in TSB (BDH) at 35°C with 95% relative humidity. All enzymes used for DNA manipulations were purchased from Gibco BRL. All plasmid constructs were maintained in E. coli JM109 using standard protocols (Ausubel et al., 1991), and then transformed into P. aeruginosa by electroporation (Smith and Iglewski, 1989). Antibiotics were added to the following concentrations: (i) E. coli, ampicillin, 100 µg ml−1; (ii) P. aeruginosa, carbenicillin, 400 µg ml−1.

Table 3.  Bacterial strains and plasmids used in this work.
Strain/plasmidRelevant characteristicsReference
Strain
 E. coli
  DH5αφ80 ΔlacZΔM15 Δ(lacZYAargF)
U169 recA1 endA1 hsdR17 (rKmK+) supE44 thid
Liss (1987)
  JM109 endA1 recA1 gyrA96 thi, hsdR17 (rKmK+)
relA1 supE44 Δ(lac-proAB) [F′, traD36, proAB, lacIqZ ΔM15]
Yanisch-Perron et al. (1985)
  MG4Δ(argF-lac) U169 zah-735::Tn10 recA56 srl::Tn λ::lasIp-lacZ Seed et al. (1995)
 P. aeruginosa
  UCB-PP PA14Human isolate; able to elicit severe disease in plant and animal models Rahme et al. (1995)
  PA14 toxAPA14 ΔtoxA::GmR Rahme et al. (1995)
  PA14 gacAPA14 ΔgacA::GmR Rahme et al. (1995)
  PAO1Prototrophic laboratory strain Holloway et al. (1979)
  PAO-JP2PAO1 ΔlasI::Tet, ΔrhlI::Tn501-2 Pearson et al. (1997)
Plasmids
 pMJG1.7PSW200 with 1.7 kb SclI–EcoR1 fragment from pMG3.9 in pUC181.8; contains lasR; ApR Gambello and Iglewski (1991)
 pUCP18Cloning vector, contains stabilizing fragment for P. aeruginosa replication; ApR Schweizer (1991)
 pUCP-gacApUCP18 containing PCR amplified gacA from PAO1 genome; ApR 
 pUC181.8Cloning vector, contains stabilizing fragment for P. aeruginosa replication; ApR Frank and Iglewski (1988)
 pECP61.5pJPP8 containing the rhlA′–lacZ fusion from pECP60, ptac-rhlR; ApR Pesci et al. (1997)
 pKDT17pTS400 with plac-lasR; ApR Pearson et al. (1997)

Biofilm and planktonic growth curves

All growth curve manipulations were performed in BioSafety laminar flow cabinets to reduce the possibility of contamination. The MBEC device (MBEC Biofilm Technologies) was used to form biofilm and planktonic populations (Ceri et al., 1999). The inoculum was formed from an overnight culture grown on solid media [tryptic soy agar (TSA) or TSA supplemented with 400 µg ml−1 carbenicillin]. The secondary culture used to inoculate the device consisted of 25 ml of a 1 × 107 cfu ml−1 dilution in TSB (supplemented with 400 µg ml−1 carbenicillin when necessary). The MBEC device was incubated at 35°C on a rocking table (Red Rocker set at speed 4.5, Hoefer Instrument) to generate the shear force necessary for biofilm formation. Biofilm samples were obtained by removing individual pegs from the lid of the device using sterile pliers. Biofilm pegs were added to sterile 0.9% saline and then sonicated using an ultrasonic cleaner (Aquasonic model 250 HT; VWR Scientific) to disrupt the biofilm, thus releasing individual component cells. Planktonic populations were sampled by removing a defined volume of batch culture from the trough. To enumerate samples, serial dilutions and spot plating were performed. Each growth curve was performed in duplicate, and the averages are shown.

Biofilm antibiotic susceptibility testing

The antibiotic susceptibility profiles of P. aeruginosa biofilm and planktonic populations were obtained according to the methods of Ceri et al. (1999). The MBEC device (MBEC Biofilm Technologies) was used to form 96 equivalent biofilms for biofilm antibiotic susceptibility profiling. Bacterial inocula were formed as described above. Samples were grown until biofilms had developed to a population size of ≈ 106 cells/peg (5–6 h after inoculation) and then briefly rinsed to eliminate residual planktonic bacteria. The biofilm lid was then transferred to the 96-well microtitre antibiotic challenge plate.

Antibiotic challenge plates were constructed such that multiple antibiotics were tested simultaneously in each assay using 96-well microtitre plates. Antibiotics used in the challenge plate were serially diluted in cation-adjusted Mueller–Hinton broth (CA-MHB). Doubling dilutions were performed to generate a concentration gradient ranging from 1024 µg ml−1 to 2 µg ml−1. Both a growth control lane and a sterility control lane were also used to confirm growth and the absence of contamination. Biofilms were challenged for 16–20 h at 35°C with constant shear force. After antibiotic challenge, the biofilm lid was briefly rinsed in microtitre plates and then transferred to a recovery microtitre plate containing CA-MHB. Biofilms were disrupted to release individual component cells into the recovery media by sonication in an ultrasonic cleaner for 5 min. Biofilm size was measured directly after antibiotic challenge. Each sample of the 96-well microtitre plate was serially diluted in 0.9% saline to determine exact cfu remaining in the biofilm after antibiotic challenge. Alternatively, recovery plates were incubated overnight to allow for the growth of any remaining bacteria. Minimal biofilm eradication concentrations (MBECs) are defined as the minimum concentration of antibiotic that prevents growth in the recovery plate. The antibiotic challenge plate was read similarly to determine the presence or absence of growth. The MIC refers to the minimum concentration of antibiotic that prevents planktonic growth.

Assays for autoinducer production

To determine whether there was a difference in autoinducer production between P. aeruginosa PA14 wild type, PA14 toxA and PA14 gacA, bioassays were performed using E. coli reporter strains. To quantify accurately the levels of autoinducer being produced in each strain examined, the liquid culture assay of Pearson et al. (1994) was used. P. aeruginosa strains were grown in either LB broth or PTSB media, and E. coli strains were grown overnight in A medium supplemented with 100 µg ml−1 ampicillin.

To detect 3-oxo-C12-HSL, the A medium supplemented with P. aeruginosa supernatants was inoculated with 3-oxo-C12-HSL reporter E. coli MG4 (pKDT17) (lasB′–lacZ) to an A540 of 0.1 and grown for 5–6 h, at which point the A600 was measured as an indication of growth. β-Galactosidase procedures were then carried out as described by Miller (1972). To measure C4-HSL levels, the medium was inoculated with C4-HSL reporter E. coli DH5α (pECP61.5) (rhlA′–lacZ) to an A540 of 0.08 and grown at 37°C to an OD of 0.3. IPTG (1 mM) was then added, and the cells were grown for an additional hour to induce activity further (Pearson et al., 1997). β-Galactosidase activity was measured as described previously (Miller 1972).

Twitching motility assays

To assess twitching motility of P. aeruginosa PA14 wild type, PA14 toxA and PA14 gacA, zones of twitching were measured and compared. On very thin LB or TSA plates (< 2 mm thick), each of the three PA14 derivative strains was inoculated using a stab loop. Bacterial proliferation between the agar and the surface of the plate was measured as the zone of twitching. Twitching zones were measured for each strain after 24 h and 48 h. To facilitate visualization of twitching zones, cells were stained with Coomassie brilliant blue G-250. After the defined incubation period, the agar was carefully removed, and 8 ml of Coomassie brilliant blue G-250 was added and incubated for 2 min to stain the cells in the adherent zone of twitching. The surface of the Petri dish was then rinsed twice with methanol to remove excess stain. Blue-stained zones representing the zones of twitching were measured. Each assay was performed in triplicate, and the average results are shown.

Flagellar swim plates and swarming assays

Flagellar swimming and swarming assays were performed as described previously by Kohler et al. (2000). Swim plates were incubated at room temperature, and swarm plates were incubated at 37°C. All plates were grown for 72 h. Each assay was performed in duplicate, and the average results are shown.

Alginate biosynthesis assays

The alginate bioassay was performed using the modified carbazole assay described by May and Chakrabarty (1994). Each assay was performed in duplicate, and the average results are shown.

Preparation of specimens for scanning electron microscopy (SEM)

Biofilm samples were fixed to MBEC device pegs for SEM as follows. Eight samples, representing pegs found in a single column of the MBEC device, of each strain were fixed during each procedure. Fixation of samples was performed using 96-well microtitre plates. Each solution (200 µl) was added to each well in a column of the microtitre plate. The biofilm samples were fixed to the peg by incubation in a 5% glutaraldehyde–cacodylic acid buffer for 2 h at room temperature. After fixation, samples were washed for 10 min in 0.1 M cacodylic acid. This wash was repeated a total of five separate times. The samples were then washed in double-distilled water to remove the cacodylic acid. As before, five separate 10 min washes were performed. The samples were then progressively dehydrated using increasing concentrations of ethanol. Samples were incubated for 20 min at each of the following concentrations of ethanol: 20%, 30%, 50% and 70%. Samples were then air dried. Individual pegs were removed using sterile pliers and mounted to SEM pins. Samples were coated and visualized.

Acknowledgements

This work was supported by CIHR grants mt-15680 and mop-43901 to D.G.S. and an NSERC grant to H.C. M.P. was supported by an NSERC studentship and an AHFMR studentship. Technical assistance was gratefully received from Sabrina Sandhu, Amanda Kirkham, David Erickson, Jaime MacDonald, Dolina Vollman, Liz Midellemiss and Carol Stremick. The authors thank Tracey Hunt for her review of this paper.

Ancillary