Virulence of Pseudomonas aeruginosa involves the co-ordinate expression of a range of factors including type IV pili (tfp), the type III secretion system (TTSS) and quorum sensing. Tfp are required for twitching motility, efficient biofilm formation, and for adhesion and type III secretion (TTS)-mediated damage to mammalian cells. We describe a novel gene (fimL) that is required for tfp biogenesis and function, for TTS and for normal biofilm development in P. aeruginosa. The predicted product of fimL is homologous to the N-terminal domain of ChpA, except that its putative histidine and threonine phosphotransfer sites have been replaced with glutamine. fimL mutants resemble vfr mutants in many aspects including increased autolysis, reduced levels of surface-assembled tfp and diminished production of type III secreted effectors. Expression of vfr in trans can complement fimL mutants. vfr transcription and production is reduced in fimL mutants whereas cAMP levels are unaffected. Deletion and insertion mutants of fimL frequently revert to wild-type phenotypes suggesting that an extragenic suppressor mutation is able to overcome the loss of fimL. vfr transcription and production, as well as cAMP levels, are elevated in these revertants, while Pseudomonas quinolone signal (PQS) production is reduced. These results suggest that the site(s) of spontaneous mutation is in a gene(s) which lies upstream of vfr transcription, cAMP, production, and PQS synthesis. Our studies indicate that Vfr and FimL are components of intersecting pathways that control twitching motility, TTSS and autolysis in P. aeruginosa.
Pseudomonas aeruginosa is a ubiquitous Gram-negative bacterium found throughout the environment. This opportunistic pathogen causes serious and often life-threatening infections in immunocompromised humans such as those who are suffering from severe burns, cystic fibrosis or AIDS, who are undergoing cancer chemotherapy, or who are recovering from major surgery (Giamarellou, 2000). P. aeruginosa can infect many eukaryotes including mice, insects, nematodes and plants (Jander et al., 2000; Rahme et al., 2000; Tan and Ausubel, 2000). P. aeruginosa is equipped with a large arsenal of secreted and cell-associated virulence factors which provide nutrients for bacterial growth, enhance invasive potential and/or directly damage host tissue (Engel, 2003). The secretion of some of these virulence factors requires the type III secretion system (TTSS) that allows translocation of a set of toxins, termed effector proteins, directly into the eukaryotic host cell where they modulate host signal transduction pathways causing, among other things, cytotoxicity and disruption of the host cell cytoskeleton (Engel, 2003).
Pseudomonas aeruginosa tfp are polymers composed of the PilA protein (Sastry et al., 1985), referred to as pilin. The biosynthesis of the intact pilus on the bacterial surface and its proper function require the products of up to 40 genes arranged in multiple loci dispersed throughout the P. aeruginosa chromosome (Mattick, 2002). These genes fall into two broad categories: those that encode proteins involved in the structure and assembly of these organelles and those that encode regulatory proteins that control both the production of tfp (and other virulence determinants) and twitching motility in response to environmental stimuli. Included in the latter is Vfr (Beatson et al., 2002a), a cAMP-binding protein that has been implicated in the global regulation of numerous virulence determinants of P. aeruginosa including exotoxin A production, quorum sensing, expression of numerous proteins secreted by the type II general secretion pathway, repression of flagellar biosynthesis, tfp biogenesis and twitching motility, and expression of the TTSS apparatus and effector proteins (West et al., 1994; Albus et al., 1997; Beatson et al., 2002a; Dasgupta et al., 2002; Wolfgang et al., 2003).
Here we describe a novel P. aeruginosa gene, fimL, which is required for multiple virulence functions including tfp biogenesis, twitching motility, type III secretion (TTS)-mediated cytotoxicity towards epithelial cells in vitro and autolysis. Our studies indicate that FimL controls these activities, at least in part, through regulation of vfr expression under specific culture conditions. We have also found that fimL mutants frequently revert to wild-type phenotypes through acquisition of extragenic suppressor mutation(s) that result in elevated production of cAMP and Vfr.
Identification of P. aeruginosa transposon mutants of gene PA1822
Fifteen non-twitching transposon mutants of the cytotoxic P. aeruginosa strain PA103 were identified in a screen for loss of cytotoxicity on Manin Darby Canine Kidney (MDCK) epithelial cell monolayers (Kang et al., 1997), suggesting that pili were necessary for killing epithelial cells. Sequence analysis of inverse polymerase chain reaction (PCR)-generated regions adjacent to the transposon insertions in the non-twitching mutants revealed that 12 of the mutants contained insertions into previously characterized genes. One of the remaining three mutants, PA103 mutant I, was found to harbour a transposon insertion within a novel open reading frame (ORF).
In order to isolate the genomic region containing the site of transposon insertion of PA103 mutant I for sequence analysis, a PA103 genomic fragment able to complement the twitching defect of PA103 mutant I was identified from a cosmid library by screening for the ability to restore twitching motility. Of 1000 cosmids tested, pCos5-9B was the sole cosmid able to rescue the twitching motility defect of PA103 mutant I (data not shown). Southern blot hybridization, using a probe corresponding to the PA103 mutant I inverse PCR product, was used to identify a 3.3 kb KpnI fragment of pCos5-9B that contained the site of transposon insertion. The complete nucleotide sequence of the PA103 3.3 kb KpnI fragment from pCos5-9B was determined for both strands (GenBank Accession No. AF083252). The sequence for the entire genome of P. aeruginosaPAO1 has since been determined (Stover et al., 2000) and the sequence of the corresponding regions are almost identical to that determined here for PA103.
Several putative ORFs were identified in the region, including PA1822 which contained the site of transposon insertion in PA103 mutant I (at nucleotide 792 of PA1822; Fig. 1A). PA1822 is a 1686 nucleotide ORF predicted to encode a 61 kDa protein of 562 amino acids. In a separate screen of a mTn5-Tc transposon library of PAO1293 for mutants defective in twitching motility (Beatson et al., 2002a), we identified a transposon mutant (31E6) that also contained an insertion in PA1822 (at nucleotide 891; Fig. 1A). Because of its role in twitching motility and tfp biogenesis (see below), we refer to PA1822 as fimL.
To confirm the predicted ORF for fimL (PA1822), we tested the ability of Escherichia coli to synthesize the predicted fimL gene product. Plasmids pJC207+ and pJC207– containing the fimL gene cloned in both directions relative to the T7 promoter in pGEM-7 and the vector control were transformed into E. coli strain BL21(DE3) which expresses T7 polymerase under control of the lac promoter. When transcription was induced with isopropyl β- d-thiogalactoside (IPTG), only the clone pJC207+ which contains the full-length fimL gene in the same direction relative to the T7 promoter produced an approximately 60 kDa protein (Fig. 1B) which is consistent with the predicted molecular weight (61 kDa) of the fimL gene product.
fimL is part of a cluster of three genes
Two ORFs which are predicted to be transcribed in the same direction flank the fimL gene, suggesting that it may be part of a small operon (Fig. 1A). The upstream ORF (PA1821) is predicted to encode a 270-amino-acid protein (30 kDa) which has been annotated in the PAO1 genome sequence to be a probable enoyl-CoA hydratase/isomerase. This family of proteins includes biosynthetic enzymes involved in fatty acid elongation. blastp analysis indicates that this protein has highest identity (74–80%) to homologues from other Pseudomonas species whereas the next closest homologues (46–48% identity) are mammalian peroxisomal and mitochondrial enzymes rather than enoyl-CoA hydratases from other Gram-negative bacteria. The stop codon of PA1821 is separated from the putative start codon of fimL by 40 nucleotides. The third ORF (PA1823; nudC) in this gene cluster has a start codon that overlaps the fimL stop codon. PA1823 (nudC) encodes a putative 278-amino-acid protein (31 kDa) that is predicted by swiss-prot to be a NADH pyrophosphatase and as such belongs to the NudC subfamily of the Nudix hydrolase family, which is a class of proteins that catalyse the hydrolysis of nucleoside diphosphates from a variety of substrates (Bessman et al., 1996).
blastp analysis shows that P. aeruginosa FimL is homologous (35% identity, 56% similarity) to a 576-amino-acid hypothetical protein from Microbulbifer degredans (ZP_00065238). The FimL homologue of M. degredans shares a similar genetic arrangement to that of P. aeruginosa fimL. In both bacteria, the gene encoding the FimL homologue is followed by a gene encoding a NudC homologue, although in M. degredans a gene encoding a PA1821 homologue is not located upstream.
The location of fimL upstream of nudC raises the possibility that the twitching motility defects observed in PA103 mutant I and PAO1293 mutant 31E6 could result from polar effects of fimL transposon insertions on nudC transcription. We undertook several approaches to address this possibility. First, we constructed an isogenic in-frame deletion mutant of fimL and an EZ::TN<TET-1> insertion mutant of nudC in both PA103 and PAO1 (Fig. 1A). Twitching motility stab assays (in which twitching motility results in colony expansion at the agar/plastic interface) demonstrate that these fimL deletion mutants of PA103 and PAO1 behaved identically to that observed with the original transposon insertion mutants of PA103 (mutant I), and PAO1293 (mutant 31E6) (Fig. 2A; data not shown) whereas the nudC mutants showed no defects in twitching motility (data not shown). These observations indicate that the observed twitching motility defects of fimL mutants result from the loss of fimL and not from polar influences on nudC expression nor from a secondary mutation elsewhere in the genome.
Next we determined that FimL alone was sufficient to complement the twitching defects observed in fimL mutants. An 8 kb BamHI fragment containing fimL and its surrounding ORFs (PA1821 and nudC) was subcloned from the complementing cosmid pCos5-9B into pUCPSK to create pCW32. This plasmid was subjected to in vitro transposon mutagenesis with EZ::TN<TET-1> and plasmids containing transposon insertions in fimL (pCW32-EZ::TN-7) or nudC (pCW32-EZ::TN-8) were isolated (Fig. 1A). pUCPSK, pCW32, pCW32-EZ::TN-7 and pCW32-EZ::TN-8 were transformed into wild-type PA103, mutant I and PA103ΔfimL and twitching motility was assayed (data not shown). Both pCW32 and pCW32-EZ::TN-8 (nudC insertion) but not pCW32-EZ::TN-7 (fimL insertion) restored twitching motility to the PA103 fimL mutants indicating that fimL is required for the ability to complement twitching motility to fimL mutants of PA103. Together, these results convincingly demonstrate that mutation of fimL (and not nudC) is responsible for the defects in twitching motility observed with P. aeruginosa fimL mutants.
Interestingly, when we tried complementation of twitching motility with pCW32 we found that this plasmid did not restore twitching motility to PAO1ΔfimL and completely abrogated twitching motility in wild-type PAO1, whereas inhibition of twitching motility by pCW32 was not observed with wild-type PA103 (see above). Repression of twitching motility in PAO1 did not occur with pCW32-EZ::TN-7 (fimL insertion) but did with pCW32-EZ::TN-8 (nudC insertion) suggesting that the presence of fimL on pCW32 was responsible for the loss of twitching motility in PAO1 (data not shown). To examine this further we subcloned fimL from pCW32-EZ::TN-8 (to utilize convenient cloning sites located in the EZ::TN) into the low-copy-number, IPTG-inducible expression vector pMMB208 in both orientations relative to the Ptac promoter (pJB23-1 and pJB23-2). Both orientations of fimL relative to Ptac could restore twitching motility in PAO1ΔfimL (data not shown), indicating that the defect in twitching motility in this mutant results from deletion of fimL. Titration of FimL expression from pJB23-1 (fimL with Ptac) with 0–10 mM IPTG shows that higher levels of expression of FimL causes a reduction in twitching motility zone size in PAO1 and PAO1ΔfimL (data not shown). The observed repression of twitching motility in PAO1 with the higher-copy-number pUCPSK clone pCW32 and with high IPTG concentrations when expressing FimL from pJB23-1 suggests that FimL levels within PAO1 need to be tightly regulated.
FimL is homologous to the N-terminal domain of ChpA
Twitching motility in P. aeruginosa is controlled by a complex chemosensory pathway comprised of the proteins PilG, PilH, PilI, PilJ, PilK, ChpA, ChpB and ChpC (D’Argenio et al., 2002; Wolfgang et al., 2003; Whitchurch et al., 2004). The central component of the pathway, ChpA, is a hybrid of homologues of the histidine kinase CheA and the response regulator CheY (Fig. 1C; Whitchurch et al., 2004). P. aeruginosa ChpA (2477 amino acids) is a highly complex protein possessing nine potential sites of phosphorylation: six histidine-containing phosphotransfer (HPt) domains, two novel serine- and threonine-containing phosphotransfer domains (SPt, TPt) and a CheY-like receiver domain at its C-terminus (Fig. 1C; Whitchurch et al., 2004). blastp analysis shows that P. aeruginosa FimL is homologous (33% identity, 51% similarity) to the N-proximal domains of P. aeruginosa ChpA and its homologues found in M. degredans, Pseudomonas fluorescens, Pseudomonas syringae, Pseudomonas putida, Xyella fastidiosa, Xanthomonas axonopodis, Xanthomonas campestris and Nitrosomonas europaea. FimL is also homologous to the N-terminal domain of the 991-amino-acid Xanthomonas oryazae PilL protein which has been shown to have a possible interaction with flagellar biosynthetic protein FlhF (Shen et al., 2001). X. oryazae PilL is significantly longer than FimL and is 90% identical to the first 992 amino acids of the 2422-amino-acid X. axonopodis ChpA homologue, and 82% identical to the first 984 amino acids of the 2345-amino-acid X. campestris ChpA homologue. As such it appears to be either a truncated version of the X. axonopodis and X. campestris PilL proteins (ChpA homologues) or (more likely) a sequencing error, as originally occurred in the case of P. aeruginosa ChpA which has a domain homologous to FimL (and which was originally referred to as PilL) (Whitchurch et al., 2004). Only P. aeruginosa and M. degredans encode both ChpA and FimL homologues.
The N-terminal 563 amino acids of P. aeruginosa ChpA are homologous to FimL across the entire length of FimL. Interestingly, this ‘FimL-like’ domain of ChpA contains two regions that are highly homologous to HPt and TPt domains, respectively, except that the predicted phosphorylatable residues of these domains have been replaced with glutamine (H–Q and T–Q substitutions respectively) (Fig. 1C and D). These glutamine substitutions would remove the ability of FimL to participate in phosphorelay reactions, but are otherwise structurally conservative substitutions. Therefore, although FimL possesses two domains that are homologous to potential phosphotransfer domains, we would predict that FimL is incapable of phosphorylation in these regions and that it may act as a competitive inhibitor of phosphotransfer interactions involving ChpA and its interacting partners. Similarly, the M. degredans FimL homologue is highly conserved across the entire FimL-like domain of M. degredans ChpA protein, including the HPt and TPt domains. The M. degredans FimL protein contains a histidine–arginine substitution in the HPt domain but retains the threonine residue in the TPt domain. There were no other identifiable functional domains in FimL according to smart or Pfam domain analysis.
Given the similarity of FimL to the N-proximal domain of the complex chemosensory protein ChpA, we considered the possibility that FimL participates in signal transduction control of twitching motility in P. aeruginosa by intersecting with this system. We expressed ChpA (pUCPChpA, pMMBChpA), the FimL-like N-proximal domain of ChpA (pJEN54, pJB24) and a clone containing all of pilG-pilK (pJEN53a) in fimL mutants of PA103 and PAO1 and assayed the ability of these clones to complement twitching motility to determine whether provision of any of these signalling components provided in trans could bypass the fimL mutation. None of the clones were able to restore twitching motility to fimL mutants (data not shown). Of note, expression of pUCPChpA in PAO1 suppresses twitching motility in the wild-type strain (Whitchurch et al., 2004).
fimL mutants have reduced surface-assembled tfp and twitching motility
The twitching motility phenotypes of the fimL mutants were examined via the subsurface stab assay. In this assay, normal twitching motility results in rapid colony expansion at the agar–Petri dish interface (Fig. 2A); non-twitching mutants such as pilA mutants produce no such zone of expansion (Fig. 2A). Whereas PAO1 elicits 2–3 cm twitching motility zones under these conditions, PA103 demonstrates very poor twitching motility at the agar–Petri dish interface when standard polystyrene Petri dishes are used, routinely producing zones of colony expansion < 1 cm in diameter after 24 h incubation at 37°C. We have found, however, that tissue culture-treated polystyrene dishes promote extensive twitching motility in PA103 (Fig. 2A) possibly due to enhanced binding of the tfp to the treated surface. Twitching motility stab assays using tissue culture-treated dishes show that fimL mutants of both PA103 and PAO1 are capable of some twitching motility. fimL mutants produce zones of colony expansion which are very reduced relative to wild type but which are significantly larger than that produced by the non-twitching pilA mutants (Fig. 2A).
Studies using time-lapse video microscopy to examine the dynamics of twitching motility at a cellular level have revealed that twitching motility is a complex social process. Wild-type bacteria move in a highly co-ordinated fashion, initially forming rafts of cells which move away from the colony edge, behind which an intricate lattice-like network of cells is formed (Semmler et al., 1999; see Fig. 2B). Non-twitching mutants (such as pilA mutants) demonstrate no differentiation of the colony edge (Semmler et al., 1999; see Fig. 2B). We used this slide microscopy technique to more closely examine the twitching motility of fimL mutants in both PA103 and PAO1 strain backgrounds. When examined after 2–4 h incubation at 37°C, fimL mutants of both strains behaved similarly and appear capable of the early stages of twitching motility forming large, exaggerated rafts of cells which move away from the colony edge but lack the development of the characteristic lattice network behind these rafts (Fig. 2B).
ELISA and Western blots using anti-PilA anti-serum were used to examine the degree of surface-assembled tfp and pilin production in fimL mutants. As we only have anti-sera specific for the PA103 PilA subunit, PAO1 strains were not included in these ELISA and Western analyses. We assayed tfp production and surface assembly from 8 h plate cultures to avoid problems with autolysis of PA103 strains (see below). ELISA of whole cells and Westerns of sheared tfp demonstrate that fimL mutants produce very small amounts of surface-assembled tfp (Fig. 2C and D), which are presumably sufficiently functional to facilitate the small amount of twitching motility observed in fimL mutants. Levels of cell-associated pilin subunit were assayed by Western analysis of lysed whole cells. These studies reveal that fimL mutants remain capable of producing pilin subunit (Fig. 2D). Thus, the reduction in levels of surface-assembled tfp seen in strains lacking fimL is not likely to result from a defect in pilin production per se, but possibly results from an inability to properly co-ordinate the biogenesis of the tfp structures either through defects in assembly or due to increased retraction.
Other tfp-related phenotypes of P. aeruginosa include sensitivity to certain bacteriophage and swarming motility. We assayed sensitivity to the tfp-specific bacteriophage PO4 (Bradley and Pitt, 1974) and found that the fimL mutants had wild-type levels of sensitivity to this bacteriophage (data not shown; Kang et al., 1997). Wild-type sensitivity of fimL mutants to phage PO4 is consistent with the ability of these strains to produce at least a small amount of functional surface-assembled tfp.
Pseudomonas aeruginosa is unique among Gram-negative bacteria that demonstrate flagella-mediated swarming motility in that P. aeruginosa swarming also requires tfp (Kohler et al., 2000). As strain PA103 is non-flagellated (Montie et al., 1982), we assayed swarming motility using PAO1 and its isogenic mutants on 0.8% nutrient broth (Oxoid) containing 0.5% glucose solidified with 0.5% agar as described previously (Deziel et al., 2001). Under these conditions PAO1ΔfimL demonstrated wild-type swarming motility (data not shown) whereas the isogenic pilA mutant demonstrated only a small zone of swarming motility (data not shown). We also examined flagella-mediated swimming motility through Luria broth (LB) set with 0.3% agar and found swimming to be normal for the isogenic PAO1 fimL and pilA mutants (data not shown).
fimL mutants form mature but irregularly shaped biofilms
The role of tfp and twitching motility in biofilm development by P. aeruginosa has recently been carefully examined using time-lapse confocal laser scanning microscopy (CLSM) of colour-coded bacteria (Klausen et al., 2003a,b). In these studies it was observed that in glucose media P. aeruginosa PAO1 biofilms develop by initially forming microcolonies by clonal growth of a sessile subpopulation (in which twitching motility has been downregulated) and thereafter migrating bacteria accumulate on top of the microcolonies (stalks) to form mushroom caps. This migration was found to be dependent on tfp (Klausen et al., 2003a). Given the markedly reduced twitching motility and levels of surface-assembled tfp seen in fimL mutants, we were interested in determining the influence that fimL mutation would have on biofilm development by P. aeruginosa. As PA103 forms poor biofilms (T. Tolker-Nielsen, unpubl. obs.), we assayed biofilm development in PAO1.
Initial examination of 4-day-old biofilms formed by green fluorescent protein (GFP)-tagged fimL mutants in glucose minimal media shows that these mutants form more irregular and asymmetric mushroom structures than wild-type PAO1 (Fig. 3A). To determine whether biofilm development by PAO1ΔfimL solely results from clonal growth (as is observed for pilA mutants) (Klausen et al., 2003a); we examined biofilm development by PAO1ΔfimL mutants more closely, by initiating biofilms with a 1:1 mixture of yellow fluorescent protein (YFP)- or cyan fluorescent protein (CFP)-tagged PAO1ΔfimL. Biofilm development was observed by CSLM over a 4 day period (Fig. 3B). Again, the 4-day-old biofilms formed more irregularly shaped structures than is usually observed for wild-type PAO1 (Fig. 3B). Surprisingly, however, these assays show that PAO1ΔfimL biofilm development occurs through a similar developmental procedure as observed for wild-type PAO1. The mixed colour caps atop the monocoloured stalks are indicative that the bacteria migrated to the top of stalks formed by a sessile subpopulation to form the mushroom caps (Fig. 3B; Klausen et al., 2003a). Time-lapse video microscopy of initial biofilm development confirms that fimL mutants migrate (twitch) on the surface in a manner indistinguishable from wild-type PAO1. It is possible that the magnitude of the tfp defect in the fimL mutant may depend on environmental conditions and may be less severe in the setting of biofilm growth than during growth on agar plates.
fimL mutants show increased autolysis
Many strains of P. aeruginosa, in particular PA103, demonstrate ‘plaque-like’ clearings associated with iridescent autolysis when grown on agar plates (Warner, 1950; Zierdt, 1971; D’Argenio et al., 2002). The mechanism of autolysis is unknown but is associated with Pseudomonas quinolone signal (PQS) production (D’Argenio et al., 2002), a secreted quinolone whose production is induced during stationary-phase growth and that interfaces with the Las and Rhl quorum-sensing systems (Diggle et al., 2003). Although plate-grown PAO1 does not exhibit significant autolysis, PQS-overproducing strains of PAO1 are highly autolytic. This phenotype can be suppressed by mutation in PQS biosynthetic genes (D’Argenio et al., 2002).
As plate-grown PA103 is visibly autolytic, we examined PA103ΔfimL for this property. As shown in Fig. 4A, PA103 fimL mutants show increased autolysis when cultured on LB agar plates. Given the association between PQS production and autolysis, we measured levels of cell-associated PQS in plate-grown cultures of wild-type PA103 and PA103ΔfimL at 8 and 16 h. PQS was not detectable in 8 h plate cultures for any of the PA103 strains (data not shown) but was measurable in PA103 after 16 h of plate growth (Fig. 4B). This may correspond to the onset of stationary-phase growth. No changes in cell-associated PQS levels were detected in the FimL mutant after 16 h of plate growth (Fig. 4B). It is possible that very small changes in PQS may be sufficient to trigger autolysis or that autolysis in fimL and vfr mutants is triggered via a mechanism independent of PQS production.
fimL is required for TTSS and for adhesion to epithelial cells
PA103 is a highly cytotoxic strain of P. aeruginosa that uses the TTSS to translocate effector proteins directly into epithelial cells. The transposon insertion mutant of fimL PA103 mutant I was originally isolated in a genetic screen to identify mutants that were non-cytotoxic towards epithelial cells (Kang et al., 1997). We confirmed the role of fimL in cytotoxicity using the in-frame deletion mutant PA103ΔfimL. Bacteria were added to the apical medium of HeLa cell monolayers and the amount of cell death after 5 h was assayed by LDH release (Fig. 5A). These studies found that PA103ΔfimL shows cytoxicity levels similar to the isogenic pilA mutant, to the TTSS mutant MutN (pscJ ) and to uninfected control [P = not significant (NS)].
We also tested the ability of the cloned fimL gene to restore cytotoxic activity to PA103ΔfimL. As expected, we found that addition of PA103ΔfimL transformed with the vector control (pUCPSK) to HeLa monolayers resulted in cell death similar to HeLa cells exposed to media alone or to MutN containing pUCPSK (Fig. 5B). The fimL clone pCW32 and pCW32-EZ::TN8, which carries a Tet insertion in nudC (Fig. 1A), restored the ability of PA103ΔfimL to kill HeLa cells whereas pCW32-EZ::TN7, which has a Tet insertion in fimL (Fig. 1A), did not restore cytotoxicity (Fig. 3B). These results confirm that fimL and not nudC or some other secondary mutation elsewhere on the genome is mediating the defect in cytotoxicity observed with PA103 fimL mutants.
As tfp are the major adhesin used by P. aeruginosa for attachment to epithelial cells, one role of tfp in cytotoxocity is to mediate the initial stages of attachment of bacteria to epithelial cells. Indeed, we have previously demonstrated that functional tfp are required for cytotoxicity towards epithelial cells (Kang et al., 1997; Comolli et al., 1999a). Given the reduced levels of surface-assembled tfp in fimL mutants (Fig. 2C and D), we predicted that fimL mutants would be severely defective in their ability to attach to epithelial cells. As shown in Fig. 5C, PA103ΔfimL shows very poor adhesion to HeLa cells similar to that observed with the non-piliated pilA mutant (P = NS). Therefore, the loss of cytotoxicity in fimL mutants at least in part results from defective adhesion.
fimL regulates Vfr
vfr mutants have been reported to be deficient in twitching motility and TTS (Beatson et al., 2002b; Wolfgang et al., 2003). In particular, the twitching motility and tfp phenotypes of vfr mutants are very reminiscent of what we have observed for fimL mutants, suggesting that fimL and vfr mutants may function in a common pathway. Interestingly, it was also reported that PAO1 vfr mutants demonstrate mild autolysis although the role of this cAMP-binding transcriptional regulator in autolysis is unknown (D’Argenio et al., 2002). To more rigorously compare the similarities between fimL and vfr mutants, we generated isogenic in-frame deletion mutants of vfr in PAO1 and PA103 and assessed the twitching motility, tfp production, autolysis and PQS production in these mutants in the appropriate strain background. All of the mutants had similar growth rates when grown in LB with vigorous aeration (data not shown). We found that both fimL and vfr mutants of PAO1 and PA103 show similar defects in twitching motility by stab assay (Fig. 2A), and PA103Δvfr assembles small amounts of surface tfp (Fig. 2C and D), although PA103ΔfimL has a more severe reduction in surface-assembled tfp than PA103Δvfr (Fig. 2C and D; recall that the anti-pilin antibody does not recognize PAO1 pilin, which precludes direct quantification of PAO1 pilin). Similar to PA103ΔfimL, plate-grown PA103Δvfr exhibits increased autolysis and wild-type level PQS production relative to wild type (Fig. 4A and B). It should be noted that in the culture conditions under which we performed these assays, autolysis of PA103 is minimized and autolysis is not observed in any of the PAO1 strains (including PAO1Δvfr) even after 10 days of incubation at 30°C (data not shown). The PA103 fimL and vfr mutants also show similar defects in cytotoxicity of epithelial cells in vitro (P = NS; Fig. 5A and C) and are both defective in adhesion to epithelial cells relative to wild type (P < 0.05; Fig. 5C). Interestingly, PA103Δvfr shows increased adherence to epithelial cells relative to PA103ΔfimL (P < 0.05). This is consistent with our observations that PA103Δvfr produces more surface-assembled tfp than PA103ΔfimL.
Given the phenotypic similarities between vfr and fimL mutants, we tested whether cloned vfr could complement fimL mutants. The clones pUCPvfrF and pUCPvfrA contain vfr subcloned into pUCPSK in both orientations relative to the Plac promoter and both can complement vfr mutants of PAK and PAO1 (Beatson et al., 2002a). We transformed both pUCPvfrF (vfr with Plac) and pUCPvfrA (vfr against Plac) into wild type and fimL mutants of PA103 and PAO1 and examined twitching motility via stab assay. Neither plasmid altered the twitching motility phenotypes of wild-type strains (data not shown). pUCPvfrF but not pUCPvfrA was able to fully restore twitching motility to fimL mutants of both PA103 and PAO1 (Fig. 6A, data not shown). Furthermore, pUCPvfrF but not pUCPvfrA significantly reduced the autolysis phenotype of PA103ΔfimL (data not shown). Thus, it appears that fimL mutant phenotypes can be complemented by vfr provided in trans. Furthermore, our observations suggest that either a high level of vfr expression is required for complementation of fimL mutants and/or that expression of vfr from the endogenous vfr promoter in pUCPvfrA is poor in fimL mutants.
Vfr expression and production is reduced in fimL mutants when cultured on agar
As the twitching motility defect in fimL mutants could be complemented by extrachromosomal copies of vfr, we examined vfr transcription in fimL mutants to determine whether FimL might regulate vfr expression. A Pvfr–lacZ transcriptional fusion reporter was constructed using a mini-CTX-lacZ promoter reporter system (Becher and Schweizer, 2000; Hoang et al., 2000). The mini-CTX system is based on the lysogenic phage ΦCTX and facilitates the integration of genetic elements in single copy at a defined location (phage attachment site attB) on the P. aeruginosa chromosome. The mini-CTX-Pvfr–lacZ reporter was introduced into PA103, PAO1, PA103ΔfimL and PAO1ΔfimL. A promoterless mini-CTX-lacZ was also introduced into the wild-type strains as a negative control.β-Galactosidase activity was assayed from 8 h and 16 h LB plate cultures. PAO1ΔfimL shows a reproducible and significant reduction in Pvfr activity relative to wild type (66–72% of wild type; n = 9–12; P < 0.05; Fig. 6B). These findings were confirmed by measuring steady-state Vfr protein levels of plate-grown PAO1 by Western blot analysis. At both 8 h (data not shown) and 16 h (Fig. 6C), we observed decreased levels of Vfr protein in PAO1ΔfimL.
While our results thus far suggest that FimL regulates vfr transcription, we considered the possibility that FimL could also be modulating Vfr activity by affecting the levels of cAMP, a cofactor necessary for at least some functions regulated by Vfr (West et al., 1994). As shown in Fig. 6D, cAMP levels were not significantly different from wild type in PAO1ΔfimL and PAO1Δvfr.
Together, our data show that FimL modulates multiple virulence factors, including tfp biogenesis and function, TTS, and autolysis, at least in part, via modulation of vfr expression and production. Along with the fact that fimL mutants can be complemented with extrachromosomal copies of vfr, these results indicate that the reduction in vfr production in plate-grown fimL mutants in plate culture is contributing to the observed fimL-related phenotypes. Thus, it seems that FimL is required for normal vfr transcription under conditions in which functional tfp are required for twitching motility, such as for growth on agar surfaces.
FimL is required for expression of TTSS components
FimL appears to influence vfr transcription under conditions in which fimL mutants present vfr-related phenotypes such as reduced tfp assembly and twitching motility and increased autolysis on agar plates. PA103 vfr and fimL mutants are also both defective in killing epithelial cells in vitro (Fig. 5). Given that Vfr has been recently reported to control expression of the TTSS apparatus and effectors (Wolfgang et al., 2003), we wished to determine whether fimL might also be affecting vfr transcription under the conditions encountered in the cytotoxicity assays. We assayed β-galactosidase activity from the PA103 mini-CTX-Pvfr–lacZ reporter strains after incubation with epithelial cells. These assays showed that vfr promoter activity was reduced to about 53% of wild type in PA103ΔfimL mutants under these conditions (n = 6; P < 0.05; Fig. 5D). We also tested the ability of pUCPvfr F to complement the cytotoxicity defect of PA103 fimL mutants. Expression of Vfr in trans from pUCPvfr F restored cytotoxicity to PA103ΔfimL (Fig. 5B).
We next determined whether the defect in vfr transcription observed in fimL mutants translates into defects in secretion and/or production of TTSS system components. The proteins ExoU and ExoT are TTSS effectors which are translocated by PA103 into epithelial cells and mediate cytotoxicity and cell-rounding respectively. ExoT production and secretion was assayed from the same samples used for assaying Pvfr activity in the presence of HeLa cells. Western analyses showed that PA103ΔfimL was defective in the secretion of ExoT into the media (Fig. 5E). Loss of secretion could result from a block in function of the TTSS apparatus and/or from loss of production. Western analysis of protein production in the bacterial cell showed that PA103ΔfimL produced little to no detectable ExoT under these conditions (Fig. 5E), indicating that FimL is required for normal production of ExoT in the presence of epithelial cells. This finding is consistent with a role for FimL in controlling gene expression via Vfr. Taken together, our observations suggest that FimL affects tfp function, cytotoxicity and TTS, at least partly, through modulation of vfr expression.
fimL mutants spontaneously revert to wild-type phenotypes
During this study we were surprised to observe that fimL mutants of both PA103 and PAO1 often reverted to an apparent wild-type twitching phenotype. In such cases, reversion was characterized by flares of twitching cells erupting from the normally smooth margins of the fimL mutant colonies when grown on the agar surface (Fig. 7). Reversion was also evident with stab assays in 1% agar where flares of twitching motility caused by reversion were evident at the agar/plastic interface (Fig. 7). Cultures taken from flares of twitching cells surrounding previously non-twitching colonies invariably yielded a variant with a stable wild-type twitching motility phenotype. Given these observations, it is important to note that throughout all of our assays with fimL mutants we were careful to take special precautions to ensure that we were not inadvertently characterizing revertant strains, particularly when fimL mutants showed little to no change from wild-type phenotypes. For instance, we routinely confirmed by microscopic examination that the fimL colonies used for initiating assays, as well as colonies obtained after completion of the assay, demonstrated no evidence of reversion.
To determine whether fimL mutants or fimL revertants were hypermutators, we compared their mutation frequency with wild-type strains by measuring the frequency of spontaneous streptomycin or rifampicin resistance development after 24–48 h (data not shown). We found no difference in the general mutation rate between wild type and fimL mutants or revertants.
We selected specific revertants for further examination in various phenotypic assays. Twitching motility stab assays and slide microscopy show that PAO1 and PA103 fimL revertants (PAO1ΔfimL-Rev and PA103ΔfimL-Rev) demonstrate wild-type twitching motility under the conditions of these assays (Fig. 2A and B). ELISA and Western assays with anti-PilA anti-serum show that PA103ΔfimL-Rev is hyperfimbriate but has normal levels of cell- associated pilin (Fig. 2C and D). PA103ΔfimL-Rev shows decreased autolysis and PQS levels when grown on agar plates (Fig. 4A and B), suggesting that high levels of Vfr may repress PQS production. Cytotoxicity and adhesion assays show that these phenotypes are also restored in PA103ΔfimL-Rev (Fig. 5A and C). Secretion of ExoT into the media is fully restored in the PA103ΔfimL-Rev (Fig. 5E). Production of ExoT within the bacterial cell is also restored in PA103ΔfimL-Rev, although this is reduced relative to wild-type ExoT production (Fig. 5E). β-Galactosidase assays with fimL revertants isolated from the PAO1 and PA103 mini-CTX-Pvfr–lacZ reporter strains show that fimL revertants have Pvfr activity restored to wild-type levels when assayed from 8 h plate cultures (Fig. 6B for the PAO1 data; data not shown for PA103) or when PA103ΔfimL-Rev was cultured in the presence of epithelial cells (Fig. 5D). Interestingly, PAO1ΔfimL-Rev, but not PA103 ΔfimL-Rev, shows a significant increase in Pvfr activity relative to wild type when assayed from plates after 16 h incubation at 37°C (n = 9; P < 0.05; Fig. 5D for the PAO1 data; data not shown for PA103). The increase in vfr transcription observed with plate-grown PAO1ΔfimL-Rev correlated well with increased Vfr protein levels (Fig. 6C). Strikingly, cAMP levels were also greatly increased in PAO1ΔfimL-Rev, approximately 4- to 10-fold at 8 h (data not shown) and 10- to 40-fold at 16 h (Fig. 6D). Our observations suggest that the suppressor mutation(s) in the fimL revertants causes increased levels of vfr transcription and production, increased levels of cAMP and increased PQS. These findings are consistent with the ability of pUCPvfr F to complement fimL mutants (see above).
As Pvfr activity is increased in fimL revertants, we tested whether the vfr promoter or the gene itself might be a site of the extragenic supressor mutation in the fimL revertants. We sequenced a 1.3 kb region that encompassed the vfr coding region, as well as 110 bp downstream and 559 bp upstream of vfr from wild-type PAO1 and PA103, the isogenic fimL mutants of each and five independently isolated revertants of each of PAO1ΔfimL and PA103ΔfimL. Neither vfr nor its promoter contains the site of suppressor mutation in the revertant fimL strains (data not shown). As cAMP levels were also elevated in PAO1ΔfimL-Rev, we sequenced the coding regions of the two adenylate cyclase genes, CyaA or CyaB, as well as the upstream intergenic regions of 149 and 217 bp, respectively, which should include their respective promoters. We found no mutations in PAO1ΔfimL-Rev compared with the wild-type sequence in the database. Given the similarity of FimL to the N-terminal domain of ChpA, it is possible that FimL functions to modulate ChpA activity through interaction with ChpA or its interacting partners. We considered the possibility therefore that the FimL-like domain of ChpA might be a candidate for acquiring the suppressor mutation and sequenced this domain from wild-type and fimL revertant strains. Sequence analysis shows that this domain of ChpA is not mutated in the revertants (data not shown). It is likely that the site of secondary suppressor mutation in the revertants is occurring in a gene that encodes an as yet unidentified regulator of vfr and adenylate cyclase transcription.
Tfp assembly and function, as well as the production and secretion of virulence factors, are controlled by complex regulatory circuits in P. aeruginosa (Mattick, 2002). These include sensor/regulator modules that regulate tfp, including AlgR and FimS, PilR and PilS, the Chp system, transcriptional activators such as Vfr and its cofactor cAMP, and three interrelated quorum-sensing systems. We now identify a novel gene product, FimL, that has homology to the N-proximal domain of ChpA and which affects tfp, TTS and autolysis. Furthermore, we show that FimL regulates these diverse processes at least in part through modulation of Vfr production. Finally, we have discovered an unusual property of this gene, unique among genes in the pilus biogenesis pathway. Upon inactivation, the fimL mutants acquire secondary suppressor mutations at a significant rate. Our studies indicate that Vfr and FimL are components of a common pathway that co-ordinately controls virulence factors that are critical to the pathogenesis of human infections.
Several lines of evidence suggest that FimL is involved in the regulation of tfp function and twitching motility in P. aeruginosa. fimL mutants showed decreased twitching motility, and this phenotype was restored by introduction of the cloned fimL gene. Surface-assembled tfp are decreased although intracellular levels of PilA are normal in fimL mutants. As twitching motility has been shown to involve concerted extension and retraction of multiple tfp (Merz et al., 2000; Skerker and Berg, 2001), our findings suggest that FimL affects the regulation of tfp assembly and function rather than production.
Under certain conditions, tfp are required for mature biofilm development. Interestingly, PAO1ΔfimL forms mature biofilms under the conditions used in this study. Time-lapse video microscopy shows PAO1ΔfimL migrates across the glass surface similarly to wild-type PAO1. However, the mature biofilms appear somewhat irregular and resemble those formed in a ChpA mutant (M. Klausen, C. B. Whitchurch, T. Tolker-Nielsen and J. Engel, unpublished studies), consistent with our hypothesis that FimL may intersect with the Chp signalling system. We were careful to ensure that the biofilm assays were not initiated with fimL revertants, and we also determined that reversion had not occurred during the course of the assay.
We present multiple lines of evidence that FimL modulates vfr expression and that the defects in tfp biogenesis, twitching motility, TTS and cytotoxicity observed in fimL mutants may be accounted for (at least in part) by a decrease in vfr expression. First, similar to FimL, vfr regulates twitching motility and expression of the TTSS apparatus and effectors (Beatson et al., 2002a; Wolfgang et al., 2003). Second, provision of vfr in trans complements fimL mutant phenotypes. Third, vfr transcription is depressed in fimL mutants when grown on plates or in the presence of epithelial cells. Finally, vfr expression is upregulated in a FimL revertant, and Vfr-regulated phenotypes, including twitching motililty and TTS, are restored.
These observations also suggest that FimL functions to differentially control vfr expression under specific conditions in which tfp are functional, as is the case for twitching motility or for TTS-dependent killing and disruption of host signalling pathways in epithelial cells. The reduction in vfr promoter activity in fimL mutants under these conditions is relatively mild (only 50–70% of wild type) but is accompanied by a decrease in Vfr protein levels. Interestingly, vfr expression in fimL mutants is unaffected in broth-grown cultures (C.B. Whitchurch and J.N. Engel, unpubl. obs.) and we are currently pursuing the significance of this observation.
We also discovered that fimL and vfr mutants of PA103 show increased autolysis when grown for prolonged time periods on LB plates, while ΔfimL revertants exhibit wild-type autolysis and an exaggerated increase in vfr and cAMP levels. The mechanism of autolysis is unknown, but has been shown to correlate with PQS production (D’Argenio et al., 2002). PQS biosynthesis is a complex process that is regulated by MvfR (also known as PqsR) (Diggle et al., 2003; Deziel et al., 2004). Recently, 4-hydroxy-2-heptylquinoline (HHQ) has been identified as the immediate precursor to PQS and to have intercellular signalling properties of its own (Deziel et al., 2004). Conversion of HHQ to PQS is regulated by LasR but not by Mvfr. However, PQS production is still synthesized, although with delayed kinetics, in a lasR mutant (Diggle et al., 2003). Interestingly, a lasR mutant of PA14 shows increased autolysis, which correlates with increased HHQ levels. This observation may explain the high baseline autolysis observed in PA103, a strain that is naturally deficient in LasR. Our experiments suggest that PQS is not increased in PA103 fimL or vfr mutants. We propose, instead, that the observed increase in autolysis in the fimL and vfr mutants may arise from increased HHQ (or HHQ precursor) levels, which are exaggerated by the absence of LasR in PA103. Autolysis and PQS levels are decreased in the FimL revertent, possibly because of upregulation of an enzyme that degrades HHQ and PQS or that shunts one or more of its precursors into another pathway.
The interrelationship of the PQS/HHQ cell communication systems is intricate and incompletely understood, but is probably important in the pathogenesis of P. aeruginosa infections (Collier et al., 2002; Guina et al., 2003). We could only detect PQS production in PA103 after prolonged growth on plates, at which time it has probably entered stationary-phase growth. Similarly, PQS production in PAO1 and PA14 has been shown to be induced during late logarithmic and stationary phase (Diggle et al., 2002; Deziel et al., 2004). Together with published data of others demonstrating that Vfr regulates the Las and Rhl quorum-sensing systems (Albus et al., 1997), our results suggest that Vfr may regulate all three known quorum-sensing systems of P. aeruginosa. Our data are consistent with a model in which FimL regulates Vfr, which functions under the conditions of our assays to negatively regulate PQS production and autolysis. Our findings also support the notion that cAMP, presumably by binding to and modulating Vfr activity, may also regulate PQS production and autolysis.
These observations also suggest that FimL regulates vfr expression under specific conditions in which tfp are functional, such as twitching motility or TTS-dependent killing and disruption of host signalling pathways in epithelial cells. The reduction in vfr promoter activity in fimL mutants is relatively mild (only 50–70% of wild type). Our findings suggest that even a minor reduction in vfr transcript levels within the cell has a significant impact on Vfr-dependent phenotypes and/or that FimL is also functioning via other effectors. Indeed, while both fimL and vfr mutants show reduced levels of surface-assembled tfp compared with wild type, we found that the defect is more severe in fimL mutants, an observation which supports the notion that FimL might also be controlling additional gene products necessary for functional tfp. This prediction is supported by initial microarray analysis that reveals that FimL and Vfr transcriptional targets include both genes in common and non-overlapping sets of genes (J. Sargent, J. West and J. Engel, unpubl. obs.).
The homology between FimL and the N-terminus of ChpA, a complex CheA-like histidine kinase that also regulates tfp biogenesis and twitching motility (Whitchurch et al., 2004), is intriguing and prompts the hypothesis that FimL modulates this signal transduction system. However, the relationship between FimL, Vfr and the Chp chemosensory regulon may be complex. While our data places FimL upstream of Vfr, it has been reported that the Chp system components, along with most of the known tfp genes, are themselves targets of Vfr (Wolfgang et al., 2003). Overexpression of Chp system components did not restore twitching motility to fimL mutants, suggesting that FimL may regulate additional factors required for twitching motility, as suggested by preliminary microarray studies. Nonetheless, FimL may still have a modulatory role on the activity of ChpA or its partners in phosphotransfer relays.
Our studies have uncovered a role for FimL in regulating TTS. fimL mutants show loss of adherence to epithelial cells, presumably caused by reduced levels of functional surface-assembled tfp. The decrease in ExoU-mediated cytotoxicity can be explained in part by the reduction in adhesion. Remarkably, we found that fimL mutants are also defective in the production, secretion and translocation of TTSS effectors. This finding can be explained by decreased production of Vfr, which has recently been shown to regulate transcription of the TTS apparatus and effectors (Wolfgang et al., 2003), as well as by the loss of surface pili, which have recently been shown to be required for production and translocation of TTS effectors (T. Jakobsen and J. Engel, unpubl. data).
One of the most intriguing aspects of this study has been the observation that the fimL mutant phenotype is readily overcome presumably by an extragenic spontaneous suppressor mutation. Importantly, fimL mutants and revertants are not hypermutators. The revertents appear to overproduce Vfr and cAMP, which may synergistically act to increase transcription of many downstream gene targets. The high selective pressure for developing twitching motility has been amply demonstrated by the reversion of P. aeruginosa PAK lasI and rhlI null mutants. After having undergone mutations in vfr and algR (respectively) to achieve a defective twitching phenotype, these mutants develop further compensatory mutations at high frequency when stabbed in solid media (Beatson et al., 2002b). As the site of the compensatory mutation in the FimL revertants is not in Vfr, nor in the genes encoding adenylate cyclase (cyaA or cyaB), or the N-terminal domain of ChpA, we predict that it is occurring in a gene that encodes a regulator of vfr and adenylate cyclase transcription. Given the lack of homology with transcriptional regulators, it is highly likely that FimL controls vfr transcription indirectly via one or more intermediary proteins. Identifying the site(s) of suppressor mutation in fimL revertants is an important avenue of future research to expand our understanding of how FimL modulates vfr expression in P. aeruginosa. Further analysis of FimL, Vfr and ChpA will help elucidate the mechanism via which these proteins intersect to control tfp biogenesis and function, TTS, PQS and HHQ production, and other aspects of P. aeruginosa virulence.
Bacterial strains, plasmids and media
Bacterial strains and plasmids used in this study are listed in Table 1. E. coli were routinely cultured in LB broth or on LB agar containing the appropriate antibiotic at 37°C. P. aeruginosa was routinely cultured in LB or on LB agar (1–1.6%) or Vogel-Bonner (VBM) agar (Vogel and Bonner, 1956) plus the appropriate antibiotic at 37°C unless otherwise indicated. For light microscopy, the cells were grown on media that contained 4 g l−1 tryptone, 2 g l−1 yeast extract, 2 g l−1 NaCl, 1 g l−1 MgSO4·7H2O and 8 g l−1 GelGro (ICN) as a solidifying agent. The following antibiotic concentrations were used for the selection of E. coli: tetracycline 5 µg ml−1 for plasmid selection and 40 µg ml−1 for cosmid selection; ampicillin 100 µg ml−1, chloramphenicol 25 µg ml−1 and kanamycin 50 µg ml−1. The concentration of antibiotics for the selection of P. aeruginosa were carbenicillin 250 µg ml−1, chloramphenicol 250 µg ml−1, rifampicin 50 µg ml−1, streptomycin 500 µg ml−1, tellurite 150 µg ml−1 and tetracycline 200 µg ml−1.
Table 1. Bacterial strains and plasmids used in this study.
Strain or plasmid
Source or reference
Cm, chloramphenicol; Tc, tetracycline; Gm, gentamicin; Km, kanamycin; Ap ampicillin; Tel, tellurite.
Identification of PA103 mutant I transposon insertion site
A library of PA103 genomic fragments cloned into cosmid pLAFR5SK1 (Hauser et al., 1998) was screened by mating E. coli strain S17-1 containing the PA103 cosmid library to PA103 mutant I. These strains were mated on LB agar overnight at 37°C and exconjugants were selected on VBM agar containing tetracycline. Pools of eight cosmid clones were initially screened, followed by screening for individual cosmids from pools that restored twitching motility to PA103 mutant I.
Identification of the PAO1293 fimL mutant
Pseudomonas aeruginosa strain PAO1293, a chloramphenicol-resistant derivative of the completely sequenced PAO1 strain, was mutagenized with a 2.2 kb mTn5-Tc transposable element (de Lorenzo et al., 1990) and a library of over 12 000 mutants were screened by twitching stab assay for defects in twitching motility (Beatson et al., 2002a). One such twitching-defective mutant PAO1293-31E6 was selected for further study. To identify the point of transposon insertion marker rescue cloning of the chromosomal fragment containing the transposon (and associated tetracycline resistance marker) was performed by ligating a PstI digest of chromosomal DNA from 31E6 into pBluescript II SK (to produce pSB62.4). The DNA sequence flanking the point of insertion was determined by automated sequencing using a primer designed to match one end of the transposon (primer Tn5.I: 5′-GCGGCCAG ATCTGATCAAGAG-3′).
Bacterial protein expression studies
T7 expression was accomplished following the procedure of Studier and Moffatt (1986) using E. coli strain BL21(DE3). fimL was cloned as a 3.3 kb KpnI fragment from pCos5-9B into pGEM-7 either in the same direction as the T7 promoter (pJC207+) or in the opposite direction of the T7 promoter (pJC207–). One millilitre of a bacterial culture at an A600 of 0.1 was incubated in M9 minimal medium supplemented with 0.5% methionine assay medium (Difco, Detroit, MI) for 60 min at 37°C and then induced with 1 mM IPTG for another 60 min at 37°C. Cultures were subsequently incubated with 100 µg ml−1 rifampicin for 60 min at 37°C before labelling with 10 µCi µl−1 Tran35S-Label Metabolic Labeling Reagent (ICN Pharmaceuticals, Costa Mesa, CA). Samples were centrifuged, resuspended in 50 µl of SDS-PAGE sample buffer (Laemmli, 1970), boiled for 5 min and then loaded onto 10% SDS-PAGE gels (Laemmli, 1970) followed by autoradiography.
Construction of isogenic mutants
In-frame deletions of fimL were constructed as follows. An approximately 1 kb region 5′ of fimL and including the first 96 bp of fimL was amplified with primers fimL1 (5′-GCGG GATCCGTTTAGCGCAATCATCGAAG-3′) and fimL2 (5′-GCGAAGCTTGGCAATGAACTGCTCAAGAC-3′). An approximately 1 kb region 3′ of fimL and encompassing the final 326 bp of fimL was amplified with primers fimL3 (5′-GCGAAGCTTATCACCGCGTACCTGGAATC-3′) and fimL4 (5′-GCGGGATCCTCCTCGATCTCGTCTTCCTG). The PCR products were cloned into pGEM-T and sequence was confirmed. Construction of the fimL deletion suicide clone for allelic exchange is described in Table 1.
In-frame deletions of vfr was constructed as follows. An approximately 1 kb region 5′ of vfr and including the first 32 bp of vfr was amplified with primers vfr1 (5′-GAATTCGTC GATGTACTGCACGTAGG-3′) and vfr2 (5′-AAGCTTTTTGAG TTTGGGTGTGTGGG-3′). An approximately 1 kb region 3′ of vfr and encompassing the final 11 bp of vfr was amplified with primers vfr5 (5′-AAGCTTGGCACCCGCTGAACAGCACC-3′) and vfr6 (5′-GGATCCATTCAACTGGCCCACGATGC). The PCR products were cloned into pGEM-T and sequence was confirmed. The cloned PCR products were excised and the fragments concatamerized and cloned into pOK12. The vfr in-frame deletion construct was then shuttled into the suicide vector pJEN34.
Other suicide clones for allelic exchange used in this study were constructed as described in Table 1. Allelic exchange mutants were constructed using the sucrose selection system described previously (Schweizer, 1992; Alm and Mattick, 1996). All suicide clones contain the genes sacB/sacR which promote sensitivity to sucrose, and oriT which enables conjugal transfer. The constructs were then transformed into the E. coli donor strain S17-1 in preparation for mating into P. aeruginosa. After conjugation, transconjugates were selected on 5% sucrose media. This allows selection of colonies in which the plasmid sequences have been excised while leaving the homologously recombined mutated gene in the chromosome. All mutants were genotypically confirmed by Southern blotting.
In all assays the strains to be tested were inoculated from fresh cultures grown overnight on LB agar (1.6%). Twitching motility stab assays in LB solidified with 1% agar (BBL) were performed as described previously (Alm and Mattick, 1995) except that tissue culture-treated polystyrene dishes (Corning) were used. Microscopic analysis of twitching motility on GelGro (ICN) slides was performed as described previously (Semmler et al., 1999).
Swarming motility was assayed on plates composed of 0.8% nutrient broth (Oxoid) supplemented with 0.5% glucose and solidified with 0.5% agar (BBL) (Rashid and Kornberg, 2000). Plates were dried overnight at room temperature and the strains to be tested were point inoculated to the surface of the agar and incubated at 37°C for 6 h.
Swimming motility was assayed by stab inoculation of the strain to be tested into swim agar plates (LB solidified with 0.3% agar) followed by incubation for 6 h at 37°C. Motility was assessed qualitatively by examining the circular turbid zone formed by the bacterial cells migrating away from the point of inoculation.
Phage sensitivity assays
Phage sensitivity was assayed using the tfp-specific P. aeruginosa bacteriophage PO4. The phage stock was titred by adding serial dilutions of the phage to 100 µl of PA103 grown to stationary phase, and the mixture was combined with 0.7% LB top agarose and then poured onto LB agar. Phage sensitivity was assayed in strains PA103, PA103 mutant I (fimL) and PA103pilA::GmR by spotting 5 µl of serially diluted phage onto top agarose seeded with 100 µl of bacteria which has been prepared by diluting a stationary-phase culture to an A600 of 0.6 in LB broth.
Overnight LB cultures of the strains to be tested were diluted 1:100 into fresh LB, grown with vigorous aeration at 37°C to mid-log phase and 100 µl plated onto LB agar plates. After incubation at 37°C for 16–24 h, the plates were examined for the characteristic ‘plaque-like’ clearing associated with autolysis.
Overnight cultures were diluted 1:100 in 5 ml LB and grown with vigorous aeration at 37°C to mid-log phase. 100 µl of each culture was plated on LB agar and grown at 37°C in a humidified bag for 8 or 16 h. The entire bacterial lawn was resuspended in 5 ml of phosphate-buffered saline (PBS) and PQS was extracted as previously described (Gallagher et al., 2002). Briefly, 300 µl of culture resuspension was extracted twice with 900 µl of acidified ethyl acetate by vigorously vortexing for at least 30 s, followed by centrifugation at 16 000 g for 5 min. An aliquot (800 µl) of the upper organic phase was transferred to a microcentrifuge tube and allowed to dry overnight at room temperature. The dried extracts were resuspended in 20 µl of a 1:1 acetonitrile:methanol mixture by vortexing and pulse spinning multiple times. The entire volume of each extract was loaded onto a Silca Gel 60 F254 plate (20 × 20 cm; EM Science) which had been activated by soaking in 5% KH2PO4 and baking at 100°C for 1 h. Thin layer chromatography (TLC) was performed with a solvent mixture containing 17:2:1 methylene chloride:acetonitrile:1,4-dioxane. When the solvent front neared the top of the plate, the plates were removed, air dried and photographed under long-wave UV illumination using a GelDoc photoimager (Bio-Rad). Quantification was performed using the Quantity One software (Bio-Rad).
PAO1 (ATCC 15692) and the isogenic PAO1ΔfimL were fluorescently tagged at an intergenic neutral chromosomal locus with gfp, cfp or yfp in mini-Tn7 constructs as described previously (Klausen et al., 2003b). Biofilm cultivation, image acquisition and analyses were performed as described previously (Klausen et al., 2003a,b). In these studies modified FAB medium (Heydorn et al., 2000) was supplemented with 30 mM glucose for batch overnight cultures and with 0.3 mM glucose for biofilm cultivation.
Three single colonies for each strain were used to inoculate 5 ml of starter cultures that were grown in LB overnight shaking at 37°C, diluted 1:100 into 5 ml of LB and were grown with vigorous aeration at 37°C until mid-log growth. One hundred microlitres were plated onto a 1.6% LB plate and incubated at 37°C in a humidified tray for 8 or 16 h. PBS (5 ml) was added to the plate to remove the bacteria and vortexed vigorously. The A600 was measured and 10–20 µl were used to measure cAMP levels according to manufacturer's instructions (cAMP Biotrak EIA System, Amersham Biosciences). The cells were diluted into PBS (500 ml), centrifuged at 8000 r.p.m. (eppendorf microfuge) and resuspended in 500 µl of Lysis Buffer 1B. The pellet was frozen at −80°C, and thawed at room temperature, sonicated (Branson sonifier 45) three times for 10 s on ice and 100 µl was used to measure cAMP using an ELISA non-acetylation assay.
PilA immunoblotting and ELISA
Rabbit polyclonal anti-sera to PA103 PilA was generated using the C-terminal 18 amino acids of PilA (TCTSTQEEMF IPKGCNEP) (Johnson et al., 1986) conjugated to keyhole limpet haemocyanin (Animal Pharm Services). Detection of PilA in whole cell and in sheared surface-assembled tfp samples were performed essentially as described elsewhere (Whitchurch et al., 2004) except for the following differences. In this study, overnight LB cultures were subcultured 1:100, incubated to mid-log phase and 100 µl plated onto LB agar plates. After incubation at 37°C for 8 h the bacteria were resuspended in PBS and ELISA and preparation of whole cell and surface tfp samples performed as described previously (Whitchurch et al., 2004). ELISAs were performed in Immulon 4HBX Microtiter® Immunoassay plates (DYNEX Technologies) and A405 of wells was measured using a microplate spectrophotometer (SPECTRAmax® 340PC384, Molecular Devices) Samples for PilA immunoblots were displayed with NuPAGE 12% Bis-Tris Gel using 1× NuPAGE MES SDS Running Buffer (Invitrogen) and electroblotted onto Immobilon-P transfer membrane (Millipore) in the Tris-glycine system described by Towbin et al. (1979). Membranes were blocked with 5% milk, probed with a 1:40 000 dilution of primary anti-PilA antibody in 1% skim milk powder in PBS. Membranes were then incubated with a 1:10 000 dilution of goat anti-rabbit immunoglobulin G conjugated with horseradish peroxidase (HRP) (Jackson Immunoresearch) in 1% skim milk powder in PBS followed by detection by enhanced chemiluminescence (ECL) using the ECL Western Blotting Detection Reagents from Amersham Biosciences.
Plate-grown bacteria (8 and 16 h, prepared as described for the cAMP assays) were resuspended in PBS (A600≈ 1), pelleted and resuspended in 50 µl of Laemmli buffer, passed three times through a 27 g needle to shear bacterial DNA, and 10 µl were electrophoresed through one 10% Nu-PAGE gel (Invitrogen). Immunoblotting was performed as described above, using a rabbit polyclonal antibody to Vfr (a kind gift of Dr Susan West, University of Wisconsin; 1:5000 dilution) followed by incubation with a goat anti-rabbit HRP (1:10 000) dilution.
Bacteria were grown at 37°C for 18 h with shaking. The culture was diluted to an A600 of 0.08 and grown shaking to an A600 of ≈0.5. The exact inoculum was enumerated by serial dilution onto LB plates. HeLa cells (105 per well) were grown for 16–18 h on 12-well transwell tissue culture plates (COSTAR Corning, 12 mm diameter, 0.4 µm pore size), washed twice with PBS, and the medium was replaced with 1 ml of MEM-Lite (MEM; Sigma Chemical), 20 mM Hepes buffer (pH 8.0), 3.5% sodium bicarbonate in the upper compartment and 1.5 ml of MEM-Lite in the lower compartment. Cells were infected with a multiplicity of infection (moi) of 250 (OD600 1 = 109 bacteria per millilitre) for 1 h at 37°C, the filters were excised with a scalpel and washed four times in PBS. Cells were lysed by incubation for 30 min at 25°C as previously described (Comolli et al., 1999b). Bacterial counts were enumerated by serial dilution onto LB plates and normalized to the input inoculum.
HeLa cells (2 × 105) were seeded in 24-well tissue culture plate(s) 16, 17, 18 h before infection and grown overnight at 37°C in the presence of 5% CO2. Cells were washed three times with PBS and overlayed with 1 ml of MEM-Lite. Bacteria were grown as described for the adhesion assay and HeLa cells were infected with an moi of 50. After 5 h of co-cultivation, the amount of lactate dehydrogenase released into the medium (150 µl aliquot) was determined according to the manufacturer's instructions (Promega CytoTox 96 Non Radioactive cytotoxicity assay G1780).
HeLa cells (2 × 106) were seeded in 10 cm tissue culture plate(s) 16, 17, 18 h before infection and grown overnight at 37°C in the presence of 5% CO2. Cells were washed three times with PBS and overlayed with 8 ml of MEM-Lite. Bacteria were grown as described for the adhesion assay and HeLa cells were infected with an moi of 150 and incubated for 1.5 h. To quantify ExoT production and secretion, 6 ml of culture medium was subjected to centrifugation (7000 g for 20 min at 4°C) to pellet bacteria. Before centrifugation, viable counts were enumerated by serial dilution onto LB plates.Supernatant (5 ml) was filtered through low protein binding PES filters (Pall) and proteins were precipitated with 55% ammonium sulphate. After incubation on ice for 18 h, precipitated proteins were concentrated by centrifugation at 13 000 g at 4°C for 20 min. Pellets were resuspended in 250 µl of PBS. Equal amounts of sample were loaded on a 10% NuPAGE Bis-Tris gel (Invitrogen) using 1× NuPAGE MOPS buffer. To determine bacterial production of ExoT, the bacterial pellet was resuspended in 100 µl of BugBuster protein extraction reagent (Novagen) and 1 µl of Benzonase (Novagen) was added. After 30 min incubation at room temperature, extracts were centrifuged at 20 000 g for 15 min at 4°C. Supernatants were collected and protein concentrations were determined with the BCA protein assay kit (Pierce) according to the manufacturer's instructions. Equal amounts of total proteins were electrophoresed on a 10% NuPAGE Bis-Tris gel (Invitrogen) using 1× NuPAGE MOPS buffer. Proteins were visualized by Western blot analysis essentially as described for PilA immunoblotting except that membranes were blocked with 5% BSA, and a 1:6000 dilution of anti-ExoT anti-sera (Hauser et al., 1998) and a 1:2500 dilution of the HRP-conjugated goat anti-rabbit anti-sera were used.
β-Galactosidase reporter assays
The promoter region of vfr was amplified by PCR using the primers Pvfr-forward (5′-AAGCTTAGGAAGGCTTCGC AGCTCTC-3′) and Pvfr-reverse (5′-GGATCCGTCTAGGTG TTTGAGTTTGG-3′) and cloned into the mini-CTX-lacZ integration vector (Becher and Schweizer, 2000; Hoang et al., 2000). This clone (pCW70) encompasses the region −487 to +46 relative to the first nucleotide of the start codon of vfr cloned upstream of the promoterless lacZ gene. pCW70 and the promoterless mini-CTX-lacZ integration vector were transformed into E. coli S17-1 and integrated into the P. aeruginosa chromosome as previously described (Hoang et al., 2000). Unwanted plasmid sequences were removed by utilizing the Flp-FRT recombination procedure (Hoang et al., 1998).
Assays of Pvfr promoter activity from LB broth and plate cultures were performed as follows. P. aeruginosa strains containing mini-CTX-Pvfr–lacZ or the promoterless mini-CTX-lacZ were grown overnight from single colonies in LB broth. Overnight cultures were diluted 1:100 into LB broth and incubated with vigorous aeration at 37°C for 8 h or 16 h. To assay β-galactosidase activity from plate grown cultures, overnight LB broth cultures were diluted 1:100, grown with vigorous aeration at 37°C to mid-log phase and 100 µl plated onto LB agar plates. After incubation at 37°C for 8 h, the bacteria were resuspended in 5 ml of PBS by vortexing. Pvfr activity in the presence of HeLa cells was performed as follows. PA103 strains containing mini-CTX-Pvfr–lacZ or the promoterless mini-CTX-lacZ were co-cultivated with HeLa cells for 1.5 h as described in the procedure for immonoblotting of ExoT secretion and production in the presence of HeLa cells. Bacteria were centrifuged and the bacterial pellet was resuspended in 250 µl of PBS and 200 µl used to assay for β-galactosidase activity.
β-Galactosidase assays were performed as previously described (Miller, 1972) with modifications to adapt it to 96-well plate. Aliquots (200 µl) of each sample were transferred to the first row of an 8 × 12-well microtitre plate. Samples were serially diluted in PBS twofold down the eight rows of the plate, the OD600 of each well was recorded using a microplate spectrophotometer (SPECTRAmax® 340PC384, Molecular Devices) and the plate was stored at −80°C. The plate was thawed at room temperature, 100 µl of B-PER lysis buffer (Promega) added to each well and the plate incubated at 37°C for 1 h. Fifty microlitres of o-nitrophenyl-β- d-galactopyranoside (ONPG) solution at 4 mg ml−1 in PBS were added to each well and the OD420 was measured for 30 min at 1 min intervals using a microplate spectrophotometer (SPECTRAmax® 340PC384, Molecular Devices) and analysed using SoftMaxPro® 4.3.1 (Molecular Devices). Maximum rates of β-galactosidase activity in undiluted culture and OD600 of undiluted culture were calculated from suitable dilutions. Specific activity was calculated by dividing the maximum rate with the calculated OD600 of undiluted cultures.
Mutation rate assays
Minimal inhibitory concentrations of rifampicin and streptomycin were determined for both PAO1 and PA103 as described elsewhere (Andrews, 2001). Mutation rate assays were performed as described previously (Oliver et al., 2000) except for the following modification. Single colonies were inoculated into 5 ml of LB broth and incubated for 20 h at 37°C. Overnight cultures were diluted 1:100 into 25 ml of LB broth and incubated 37°C overnight. Selective LB agar plates contained either rifampicin (50 µg ml−1) or streptomycin (500 µg ml−1).
Results are presented as means ± SD. Student's two-tailed t-test, one-way anova and Student–Newman–Keuls tests were used to determine differences between means. Values of P < 0.05 were considered to be significant.
We thank L. Turnbull for assistance with statistical analyses. The work done in the Mattick laboratory was supported by the National Health and Medical Research Council of Australia. James C. Comolli was supported by the Bank of America-Gianinni Foundation. Jacob J. Bertrand was supported by the Ford Foundation. Joanne N. Engel was supported by grants from the University Wide AIDS Research Program, the NIH (R01 AI42806), the Cystic Fibrosis Foundation and the American Lung Association. During a portion of this work, Joanne N. Engel was an established investigator of the American Lung Association.
Note added in proof
While this paper was in revision, Shan et al. (2004) reported that PA1822 is involved in twitching motility (Microbiology150: 2653–2661).