|Home | About | Journals | Submit | Contact Us | Français|
The formation of biofilms by the opportunistic pathogen Pseudomonas aeruginosa is a developmental process governed by a novel signal transduction system composed of three two-component regulatory systems (TCSs), BfiSR, BfmSR, and MifSR. Here, we show that BfiSR-dependent arrest of biofilm formation coincided with reduced expression of genes involved in virulence, posttranslational/transcriptional modification, and Rhl quorum sensing but increased expression of rhlAB and the small regulatory RNAs rsmYZ. Overexpression of rsmZ, but not rsmY, coincided with impaired biofilm development similar to inactivation of bfiS and retS. We furthermore show that BfiR binds to the 5′ untranslated region of cafA encoding RNase G. Lack of cafA expression coincided with impaired biofilm development and increased rsmYZ levels during biofilm growth compared to the wild type. Overexpression of cafA restored ΔbfiS biofilm formation to wild-type levels and reduced rsmZ abundance. Moreover, inactivation of bfiS resulted in reduced virulence, as revealed by two plant models of infection. This work describes the regulation of a committed biofilm developmental step following attachment by the novel TCS BfiSR through the suppression of sRNA rsmZ via the direct regulation of RNase G in a biofilm-specific manner, thus underscoring the importance of posttranscriptional mechanisms in controlling biofilm development and virulence.
Pseudomonas aeruginosa is an important human pathogen associated with chronic colonization of a wide array of human tissues and medical devices. Biofilm formation by P. aeruginosa is considered one of the primary causes of persistent infections in immunocompromised hosts and in the lungs of individuals with cystic fibrosis (8, 9, 13, 18, 21). Secretion of numerous toxic compounds and degradative enzymes contributes to the pathogenesis of P. aeruginosa infections (42, 44), many of which are coordinated via the two N-acylhomoserine lactone (AHL)-mediated quorum-sensing (QS) systems, termed Las and Rhl (12, 42, 43, 56, 72). Recent studies have further suggested that P. aeruginosa biofilm formation, virulence, and polysaccharide gene expression are coordinated via a sensitive and complex signaling network composed of multiple two-component regulatory systems (TCSs).
TCS signaling pathways represent a major regulatory mechanism in bacteria and archaea, and are also found in simple eukaryota and higher plants (74). They translate external and internal stimuli into adaptive responses by a variety of mechanisms, including control of gene expression and methylation of target proteins. Classical TCS pathways share a conserved core architecture: a homodimerizing histidine kinase protein sensory domain and a cognate receiver or response regulator domain, coupled mechanistically through a histidine-aspartic acid phosphorelay (63).
In P. aeruginosa, a multicomponent system composed of several orphan TCS sensor kinases reciprocally regulates genes involved in acute and chronic infections. The hybrid sensor kinase RetS is required for type III secretion system (TTSS) activation, concomitant repression of exopolysaccharide production and biofilm formation, and colonization/dissemination in murine acute infection models (19, 31, 75). Genes under RetS control are inversely regulated by two other sensor kinases: GacS and LadS. Inactivation of RetS was shown to result in hyperattachment with elevated Psl exopolysaccharide locus expression and suppression of TTSS virulence (19). This phenotype was abolished by a secondary mutation in gacS (19). In contrast, LadS inactivation was found to result in decreased attachment, elevated TTSS expression, and reduced exopolysaccharide production, suggesting that LadS may function to counteract RetS activity (70).
The two hybrid regulators appear to inversely control the phosphorylation state of GacS/GacA (20), which in turn activate the expression of the small RNAs (sRNAs) rsmZ and rsmY that serve as antagonists of the translational regulator RsmA. The binding of RsmA to specific mRNA targets differentially affects their stability, turnover, and translation rates. Therefore, the multicomponent system consisting of RetS, LadS, and GacS/GacA controls the expression of a significant number of P. aeruginosa virulence and attachment genes at the level of mRNA translation and/or stability (19). Thus, it is not surprising that in P. aeruginosa PAO1, the Gac/Rsm system positively regulates the production of the quorum-sensing signal N-butanoyl-homoserine lactone (C4-HSL) and of extracellular virulence factors, such as hydrogen cyanide (46, 73).
Transitions to later stages of biofilm formation by P. aeruginosa, which coincide with distinct phenotypes compared to planktonic and initially attached bacterial cells (54, 60, 64), have recently been shown to be regulated by a multistep phosphorelay system composed of three previously undescribed TCSs, BfiSR, BfmSR, and MifSR, which were found to be sequentially phosphorylated during biofilm formation (47). Inactivation of bfiS, bfmR, and mifR arrested biofilm formation at the transitions to the irreversible-attachment, maturation-1, and maturation-2 stages, respectively, without affecting growth, motility, polysaccharide production, or initial attachment (47). These findings suggested that the novel signal transduction system represents a genetic program required for P. aeruginosa biofilm development. However, little is known of the downstream targets of these novel TCSs.
In this study, we demonstrate that BfiSR, which is required for transitioning to the irreversible attachment stage, regulates biofilm development via the transcription of RNase G (CafA). Inactivation of cafA results in increased rsmZ levels and arrested biofilm formation, a phenotype that is consistent with bfiS or bfiR inactivation, while overexpression in the ΔbfiS background results in restoration of biofilm formation and reduction of rsmZ levels comparable to P. aeruginosa wild-type biofilm formation. We further demonstrate that reduced rsmZ levels are essential for biofilm formation to occur and that inactivation of bfiS renders P. aeruginosa avirulent. We propose a novel mechanism for the posttranscriptional control of rsmZ that involves the BfiSR-dependent turnover of rsmZ RNA levels via RNase G under biofilm growth conditions. rsmZ turnover is required for biofilm development to occur.
All bacterial strains and plasmids used in this study are listed in Table Table1.1. P. aeruginosa strain PAO1 and strain PA14 were used as parental strains. With the exception of the proteomic and virulence studies, all experiments were carried out in both parental backgrounds to ensure that the ΔbfiS and ΔbfiR mutant phenotypes were identical in both parental strains. This was also necessary as the Arabidopsis virulence studies (see below) can be carried out only using P. aeruginosa PA14 strains. All planktonic cultures were grown in minimal medium containing glutamate as a sole carbon source (53) or in Luria-Bertani (LB) broth in shake flasks at 220 rpm. Biofilms were grown as described below at 22°C in 1/20 diluted LB broth or minimal medium containing glutamate as a carbon source (53).
Biofilms were grown using a once-through continuous flow tube reactor system for biofilm sample collection and in flow cells (BioSurface Technologies) for the analysis of biofilm architecture, as previously described (47, 54, 55). Quantitative analysis of confocal scanning laser microscopy (CSLM) images of flow cell-grown biofilms was performed using COMSTAT (24).
Preparation of crude protein extracts and protein determination were carried out as previously described (60). The resulting proteins were subsequently separated by 2D-PAGE and analyzed using 2D ImageMaster software (GE Healthcare, Piscataway, NJ) (53, 54). Differentially produced proteins were identified essentially as previously described using a QStarXL mass spectrometer (Applied Biosystems) (47).
Acyl-homoserine lactones (AHLs) were extracted from filter-sterilized supernatants of planktonic or biofilm cells with acidified (0.1% acetic acid) ethyl acetate (52, 53). The resulting organic fractions were evaporated to dryness and subsequently resuspended in 1.5 ml of sterile water. The concentration of N-(3-oxododecanoyl)homoserine lactone (3O-C12-HSL) was determined using an Agrobacterium tumefaciens A136-based bioassay while C4-HSL was determined using a P. aeruginosa rhlA-lacZ C4-HSL reporter bioassay (17, 42). Quantitation was based on β-galactosidase activity (39). Synthetic N-butanoyl-dl-homoserine lactone (Sigma-Aldrich) was used as a standard for absolute C4-HSL concentration determinations.
Isolation of mRNA and cDNA synthesis were carried out as previously described (1, 2, 47, 60). Quantitative reverse transcriptase PCR (qRT-PCR) was performed using an Eppendorf Mastercycler ep realplex instrument (Eppendorf AG, Hamburng, Germany) and a KAPA SYBR FAST qPCR Kit (Kapabiosystems, Woburn, MA), with oligonucleotides listed in Table S1. mreB was used as a control. The stability of mreB levels was verified by 16S RNA abundance using primers HDA1/HDA2 (37). Relative transcript quantitation was accomplished using the ep realplex software (Eppendorf AG) by first normalizing transcript abundance (based on the threshold cycle [CT] value) to mreB, followed by determining transcript abundance ratios. Melting curve analyses were employed to verify specific single-product amplification.
β-Galactosidase activity of strains harboring the rsmY or rsmZ promoter reporter construct (3, 4) was determined using the Miller assay (39) with the following modification: instead of using total cells, specific β-galactosidase activity was determined using protein extracts, obtained as previously described (60). An extinction coefficient for o-nitrophenyl-β-galactoside cleavage at 420 nm of 4,500 nl/nmol/cm was used. Furthermore, β-galactosidase activity in planktonic and attached cells was determined by fluorescent microscopy. Microscopic analyses of rsmY and rsmZ expression during surface-attached growth were accomplished by allowing the respective reporter strains to attach to microscope slides under flowing conditions in flow cells, as previously described (54), or under static conditions in medium containing 0.02 g/liter methlyumbelliferyl β-d-galactopyranoside dissolved in N,N-dimethylformamide. β-Galactosidase activity was assessed via microscopy by examination under long-wave UV excitation (10, 54). All samples were analyzed using an exposure time of 3 s, with cells illuminated only during the period of data collection.
In order to identify the in vivo DNA binding sites of the response regulator BfiR, 24-h-old biofilms of P. aeruginosa PAO1/pJN4196His, bearing the His6/V5-tagged BfiR, were subjected to chromatin immunoprecipitation (ChIP) analysis. In vivo DNA-protein cross-linking using 1% formaldehyde for 10 min at 37°C and immunoprecipitation using anti-V5 antibodies (Invitrogen Corp.) were done essentially as previously described (7, 33, 57). Following immunoprecipitation, DNA was liberated by reversing the cross-linking via incubation with 0.5 M NaCl in Tris-EDTA (TE) buffer at 65°C for 4 h. The DNA was subsequently amplified using arbitrary P. aeruginosa primers (35), subcloned into pCR2.1 (Invitrogen Corp.) to create a ChIP clone library, sequenced, and identified by BLAST analysis (http://ncbi.nlm.nih.gov/BLAST). ChIP assays were done in triplicate. Reaction mixtures containing no antibody or no His6/V5-tagged BfiR were used as controls.
BfiR binding to the putative cafA promoter was confirmed using a LightShift chemiluminescent electrophoretic mobility shift assay (EMSA) kit (Thermo Scientific). Briefly, biotinylated target DNA fragments PcafA1 (−75 to +100 relative to translational start site, representing sequence present in all identified cafA ChIP clones) and PcafA2 (−75 to 0 relative to translational start site) were amplified using primer pairs PcafA_for/PcafA1 and PcafA_for/PcafA2 (see Table S1 in the supplemental material). A total of 1 pmol of target DNA was incubated for 30 min at room temperature with 3 pmol of purified His6/V5-tagged BfiR in 25 mM Tris-Cl, pH 8, 5 mM MgCl2, 0.5 mM dithiothreitol, 1 mM EDTA, and 50 ng/μl poly(dI-dC) as nonspecific competitor DNA. For specific competition, nonbiotinylated target DNA (25 pmol) was used. Samples were separated on a 5% native polyacrylamide-Tris-borate-EDTA (TBE) gel and blotted onto a Hybond nylon membrane, and bands were visualized according to the manufacturer's instructions (Thermo Scientific).
The role of BfiSR in virulence was assessed using two plant infection models. The lettuce infection model allows for high-throughput qualitative analysis of virulence, whereas the Arabidopsis thaliana infection model provides a quantitative approach and permits the tracking of bacterial cell proliferation in planta (61). The lettuce infection model was used for the analysis of virulence of PAO1 strains and was carried out essentially as described by Filiatrault et al. (15) using store-bought romaine lettuce leaves, which were each infected with ~106 cells. The A. thaliana infection model was used to test the virulence of PA14 strains. Following 2 weeks of growth in 1/2 MS salts (2.2 g/liter Murashige and Skoog basal medium), plants were infected with P. aeruginosa PA14 and its isogenic mutant strains at a final optical density at 600 nm (OD600) of 1.0 in twofold-diluted MS salts and incubated for a period of 12 days at a 25/22°C, with a 16-h light/8-h dark cycle. All plants were inspected daily for signs of infection and/or death as evidenced by wilting, discoloration, and necrosis.
A Student's t test was performed for pairwise comparisons of groups, and multivariate analyses were performed using a one-way analysis of variance (ANOVA) followed by a postpriori test using Sigma Stat software.
To elucidate the mechanism by which this novel TCS regulates the transition of P. aeruginosa biofilms to the irreversible-attachment stage, proteins controlled in a BfiSR-dependent manner were identified using a proteomic approach. Proteins obtained from PAO1 and the isogenic bfiS mutant following 24, 72, and 144 h of biofilm growth were analyzed by 2D-PAGE, and differentially produced proteins were subsequently identified by liquid chromatography-tandem mass spectrometry (LC-MS-MS). ΔbfiS biofilms produced multiple proteins that were detectable in PAO1 biofilms only at the early stages of development (Table (Table2).2). Furthermore, multiple proteins present in 72- and 144-h-old wild-type biofilms were absent in the ΔbfiS mutant biofilms following 144 h of growth. Together, these data support our findings that ΔbfiS biofilm formation is arrested at the transition to the irreversible-attachment stage.
Proteins found to be absent in ΔbfiS biofilms are involved in transport and secretion, metabolism, and posttranscriptional/translational modifications (Table (Table2).2). Of these, three proteases (Tig, ClpP, and Lon), which are involved in global posttranslational modification, chaperone functions, and protein turnover and which are known to be essential for the regulation of a wide range of processes including multicellular behavior in a variety of bacterial species (5, 11, 29, 36, 41, 48, 66), are encoded in an operon (PA1800-PA1803). Furthermore, in contrast to wild-type PAO1, ΔbfiS biofilms exhibited the production of P. aeruginosa translation initiation factor-2 InfB, which plays an essential role in the regulation of protein production by allowing for the unusual process of formylation-independent initiation of translation (62). ΔbfiS biofilms also lacked several virulence factors and quorum-sensing (QS)-regulated proteins (Table (Table2).2). Overall, a total of 10 proteins previously identified to be QS modulated (56, 72) were differentially produced in ΔbfiS biofilms.
Comparison of 2D-PAGE protein patterns of ΔbfiS biofilms to ΔlasI and ΔrhlI mutants defective in las and rhl QS signal production revealed that the ΔbfiS biofilm proteome most closely resembled that of ΔrhlI biofilms since a large number of the QS-regulated proteins undetected in ΔbfiS biofilms were also absent in ΔrhlI but present in ΔlasI biofilms (Table (Table3;3; see Fig. S1A in the supplemental material). Among these were chitinase and phenazine biosynthetic proteins, which have been previously demonstrated to be dependent on Rhl QS signaling (56, 72). Moreover, ΔbfiS biofilms, but not planktonic cells, were characterized by 4-fold reduced levels of N-butanoyl-homoserine lactone (C4-HSL) compared to wild-type biofilms (not shown). No difference in N-(3-oxododecanoyl)homoserine lactone (3O-C12-HSL) levels were observed (data not shown).
The proteomic data were validated using qRT-PCR, demonstrating that, on average, transcript levels of selected QS-regulated genes were 5-fold lower in ΔbfiS biofilms than in wild-type biofilms (Fig. (Fig.1).1). In contrast, analysis of rhlA transcript levels (RhlA was not detected by 2D-PAGE) revealed that rhlA was 13.7-fold ± 6.2-fold increased (Fig. (Fig.1).1). Furthermore, differences in virulence were assessed by determining the expression of the type III secretion systems (TTSS) represented by pscL (PA1725; exsD-pscBCDEFGHIJKL), pcrV (PA1706; pcrGVH-popBD), and pscU (PA1690; PA1967 pscOPQRSTU). Deletion of bfiS resulted in a significantly increased expression of pscU, pcrV, and pscL (3.8-, 10-, and 20-fold, respectively) compared to wild-type biofilms following 144 h of growth (Fig. (Fig.11).
The overall changes in transcript levels following bfiS inactivation were unlike those detected in a ΔrhlR mutant (Table (Table4).4). This is further supported by the finding that restoration of C4-HSL levels in ΔbfiS did not restore biofilm formation to wild-type levels (see Fig. S1B in the supplemental material). However, TTSS and Rhl-regulated genes were expressed in ΔbfiS in a manner similar to expression in ΔgacA and ΔrsmYZ (23, 28, 44, 73) and opposite to expression in mutant inactivated in rsmA (Table (Table4).4). We therefore explored a possible link between the novel TCS BfiSR and the LadS/RetS/GacAS/RsmA/RsmYZ multicomponent system. This was done by analyzing the expression of genes central to this regulatory network, in particular rsmA, rsmY, and rsmZ.
Under planktonic growth conditions, no difference in rsmAYZ gene expression was detected when transcript levels of PAO1/PA14 and ΔbfiS were compared (Table (Table5).5). However, while rsmA, rsmY, and rsmZ abundance decreased up to 11-fold during wild-type biofilm formation, deletion of bfiS coincided with unchanged rsmA and elevated rsmYZ levels in biofilm cells compared to the free-swimming counterparts (Fig. (Fig.2),2), indicating that BfiS-dependent modulation of rsmYZ is surface contact dependent. Interestingly, decreased transcript levels of rsmA, rsmY, and rsmZ were detectable as soon as 24 h after initial attachment by P. aeruginosa wild type (Table (Table55).
The findings suggested that once cells are attached, P. aeruginosa wild-type biofilm formation coincides with an overall decrease in the expression of rsmA, rsmY, and rsmZ, a response absent in ΔbfiS biofilms.
We next sought to confirm the unexpected finding of decreasing rsmY and rsmZ levels during surface-associated growth and to determine the timing of this reduction. Consequently, chromosomal transcriptional fusion constructs were employed, and β-galactosidase activity of planktonic cells attaching to a glass surface under flowing conditions over a period of 24 h was monitored by bright-field and fluorescent microscopy. The percentage of cells with detectable β-galactosidase activity dropped significantly within 10 h after initial attachment (Fig. (Fig.33 A and C; see Fig. S2 in the supplemental material). While more than 90% of all planktonic cells used as the inoculum displayed β-galactosidase activity, less than 40% of cells harboring the rsmY-lacZ reporter displayed β-galactosidase activity, and less than 20% demonstrated rsmZ-lacZ activity (Fig. (Fig.3).3). As cells can attach reversibly (attachment via their poles ) or irreversibly (longitudinal attachment ) to surfaces during the attachment stage, β-galactosidase activity was further evaluated depending on how cells were attached to the glass surface (Fig. (Fig.3B).3B). Of all the cells demonstrating polar attachment, approximately 60% of cells harboring the rsmY-lacZ reporter and 26% of cells harboring the rsmY-lacZ reporter displayed β-galactosidase activity (Fig. (Fig.3B).3B). In contrast, less than 2% of all irreversibly attached cells displayed β-galactosidase activity (Fig. 3B and C), indicating that once P. aeruginosa cells have committed to surface-attached growth, expression of rsmY and rsmZ ceases. Thus, under flowing conditions, no evidence of rsmY or rsmZ reporter activity was detectable at ≥15 h after initial attachment (Fig. (Fig.3C;3C; see Fig. S2). It is of interest that under static conditions, rsmY and rsmZ transcription ceased within less than 4 h (data not shown). Moreover, rsmYZ transcription was detectable in nonadherent single cells collected in the biofilm effluent (data not shown).
To determine whether differences in rsmYZ levels observed during wild-type and ΔbfiS biofilm formation (Fig. (Fig.2A)2A) are due to reduced transcription of rsmYZ, we made use of rsmY and rsmZ chromosomal transcriptional fusions and determined β-galactosidase activity in planktonic and 24- and 144-h-old biofilm cells. Transcriptional fusion data showed no difference in rsmY and rsmZ expression levels between ΔbfiS and its isogenic parental strain under planktonic (data not shown) or biofilm growth conditions (Fig. (Fig.2B).2B). Overall, rsmY-lacZ and rsmZ-lacZ reporter activity increased by 19- and 139-fold, respectively, in ΔbfiS between 24 and 144 h of biofilm growth (Fig. (Fig.2C).2C). Changes in RNA levels over the same period of time were comparable (Fig. (Fig.2C).2C). For wild-type biofilms, transcription of rsmY and rsmZ changed 2.1- and 125.4-fold, respectively, while RNA levels increased by only 4.5- and 3.3-fold (Fig. (Fig.2C).2C). Thus, while rsmYZ transcription levels are similar in the ΔbfiS and wild-type strains regardless of growth conditions, rsmYZ abundance, in particular that of rsmZ, differs significantly between ΔbfiS and the wild type under biofilm growth conditions.
We next asked whether impaired biofilm formation due to bfiS inactivation is caused by increased rsmYZ levels by assessing biofilm development by ΔretS and ΔgacS, strains previously shown to exhibit altered rsmYZ levels (Table (Table4).4). Both ΔbfiS and ΔretS are characterized by increased rsmYZ levels compared to the wild type (Fig. (Fig.2A)2A) (19); however, they differ with respect to rsmYZ transcription (Fig. (Fig.2B).2B). Moreover, in contrast to ΔbfiS, a ΔretS mutant hyperattaches to polystyrene and overexpresses genes required for Psl and Pel polysaccharide biosynthesis (19, 47). The biofilm architecture of ΔretS was indistinguishable from a ΔbfiS (or ΔbfiR) mutant (Fig. (Fig.44 and Table Table6),6), indicating that rsmYZ levels play an essential role in biofilm development. It is of interest that a ΔgacS mutant, characterized by decreased rsmYZ levels and decreased rsmYZ transcription under planktonic (Table (Table4)4) and biofilm (Fig. (Fig.2B)2B) conditions, formed biofilms comparable to the wild type (see Fig. S3 in the supplemental material).
To determine which regulatory small RNA, rsmZ or rsmY, is responsible for biofilm development being arrested in a manner similar to inactivation of bfiS (or bfiR), strains overexpressing the sRNAs were tested. When placed in a PA14 background (PA14/pJN-rsmY), overexpression of rsmY resulted in biofilms that were overall similar in architecture to wild-type biofilms, being composed of large microcolonies exceeding 100 μm in diameter (Fig. (Fig.4).4). However, quantitative analysis of the biofilm architecture by COMSTAT revealed a 50% reduction in the overall biomass of PA14/pJN-rsmY biofilms compared to that of PA14 biofilms (Table (Table6).6). In contrast, biofilms overexpressing rsmZ (PA14/pJN-rsmZ) lacked large microcolonies and, instead, were composed of a thin layer of cells at the substratum with an average height of 4.5 μm and the presence of smaller cellular aggregates which were less than 30 μm in diameter (Fig. (Fig.44 and Table Table6).6). Qualitative and quantitative analyses confirmed biofilms formed by the rsmZ-overexpressing strain (PA14/pJN-rsmZ) to be comparable to ΔbfiS and ΔbfiR biofilms, as well as to ΔretS biofilms (Fig. (Fig.44 and Table Table6).6). Similar results were obtained when overexpression studies of rsmY and rsmZ were carried out in PAO1 (not shown).
These findings indicate that the arrest in biofilm development due to bfiS inactivation is based mainly on increased levels of rsmZ. Furthermore, the data indicate that rsmY and rsmZ do not appear to have redundant functions in biofilms.
We next addressed the question of how BfiSR regulates the degradation of small RNAs. We hypothesized that this is accomplished indirectly through the regulation of genes/proteins involved in posttranslational/transcriptional modifications (Table (Table2).2). To determine the direct regulatory targets of BfiSR, chromatin immunoprecipitation (ChIP) analysis was consequently employed to identify the DNA binding sites of BfiR, the cognate response regulator of BfiS.
Under biofilm growth conditions, we identified a total of six DNA binding sites from triplicate ChIP analyses. These DNA sites were not identified when ChIP was repeated using a blank vector or when it was carried out under planktonic growth conditions. Among the identified DNA binding sites were three located upstream of hypothetical genes (PA0937, PA1393, and PA2224), as well as three upstream of genes encoding proteins involved in posttranslational/transcriptional modification. These were located in the 5′ region of PA3157 encoding a probable acyltransferase; in PA3083 encoding an aminopeptidase involved in translation, posttranslational modification, and degradation of proteins, peptides, and glycopeptides; and in sequences located directly upstream of PA4477, a gene encoding the RNase G CafA (Fig. (Fig.55 A), which has been implicated in Escherichia coli in 16S RNA maturation and rapid decay of mRNA (26, 68, 69). Since more than 50% of all sequences identified by the ChIP approach were located upstream of cafA, we further investigated cafA and a possible link to BfiSR.
The binding of BfiR to the sequence upstream of cafA was verified by electrophoretic mobility shift assays (Fig. (Fig.5B),5B), with BfiR binding to the sequence present in all identified cafA ChIP clones (−75 to +100 relative to the translational start site, PcafA1), as well as to 75 bp directly upstream of the translational start site (PcafA2). In addition, cafA expression was found to be BfiS dependent, with deletion of bfiS resulting in a (5 ± 0.52)-fold reduction of cafA transcript abundance (not shown).
While ΔcafA displayed normal attachment to polystyrene using a microtiter plate assay (not shown), ΔcafA failed to develop biofilm architecture observed in mature wild-type biofilms following 144 h of growth and accumulated 10-fold less biomass over the same period of time (Fig. (Fig.5C5C and Table Table6).6). However, ΔcafA biofilms were similar to those of the ΔbfiS strain in architecture, in that both strains produced flat unstructured biofilms devoid of large microcolonies and complex differentiated structures (Fig. (Fig.44 and and5C;5C; Table Table6).6). Further support for the role of CafA was obtained by the finding of increased rsmA, rsmY, and rsmZ levels in a cafA mutant biofilm (Fig. (Fig.5D).5D). The changes in expression were comparable to those observed for ΔbfiS biofilms compared to its isogenic parental strain PA14 (Fig. (Fig.5D5D).
To confirm that the ΔbfiS biofilm phenotype was due to reduced cafA expression, we overexpressed cafA in a ΔbfiS mutant and analyzed the biofilm architecture. Overexpression of cafA in a ΔbfiS mutant restored the biofilm architecture to wild-type levels (Fig. (Fig.5C5C and Table Table6).6). Importantly, cafA overexpression also resulted in 10-fold reduced rsmZ levels in ΔbfiS biofilms (data not shown). These findings confirmed BfiSR to be involved in coordinating biofilm formation through CafA.
We next asked whether CafA is responsible for the sRNA turnover observed in P. aeruginosa wild-type biofilms. To do so, rsmA and rsmYZ levels were determined in biofilms overexpressing cafA (PA14/pJN-cafA). P. aeruginosa harboring a blank vector (PA14/pJN105) was used as a control. No differences in rsmA and rsmY levels were detected (Table (Table5).5). However, rsmZ levels were found to be (3.4 ± 0.44)-fold reduced in PA14/pJN-cafA compared to PA14/pJN105 (Table (Table5).5). The finding indicates that CafA, which was implicated in E. coli in wide-scale mRNA maturation and turnover processing (68, 69), specifically targets rsmZ under biofilm growth conditions in P. aeruginosa.
The above findings demonstrated an influence of BfiSR on biofilm development, posttranslational/transcriptional modifications, protease synthesis, and the components of the multicomponent switch RetS/LadS/ GacAS/RsmA, in particular rsmY and rsmZ, which have been previously implicated to be involved in P. aeruginosa virulence (19, 70). This prompted us to test the effect of bfiS mutation on P. aeruginosa virulence using two distinct in vivo plant infection models. These alternative nonvertebrate host models were chosen as they have been previously demonstrated to result in the identification of bacterial virulence factors and to correlate with virulence outcomes obtained using the burned-mouse pathogenicity model (50, 51). Using the lettuce leaf infection model, which allows for the determination of infection dissemination along the leaf midrib, ΔbfiS infected only a small portion of the midrib within 4 to 5 days while at the same time P. aeruginosa PAO1 disseminated along the entire midrib (see Fig. S4 in the supplemental material). Similar results were obtained using P. aeruginosa PA14 and the isogenic PA14 ΔbfiS mutant (data not shown).
Additionally, P. aeruginosa ΔbfiS was avirulent, as determined using the A. thaliana infection model (plant death was used as an indicator of virulence). While more than 34% of Arabidopsis plants were killed by P. aeruginosa PA14 (Fig. (Fig.6)6) within 5 days postinfection, with the percentage of dead plants rising to 62% and 100% following 7 and 12 days of infection, respectively, the isogenic ΔbfiS mutant was unable to establish infection within 5 days. Only 4% ± 7% of all plants were killed at 7 days postinfection (Fig. (Fig.6).6). Thus, despite the increase in expression of TTSS and rhlA (Fig. (Fig.1),1), the ΔbfiS strain was unable to mediate the characteristic cytotoxic response observed by the parental strain. To better compare ΔbfiS virulence, we made use of a gacS mutant which was previously shown to elicit weak symptoms using nonmammalian virulence models and no or low mortality in an acute mouse model (49, 51, 67). Consistent with previous findings, infection by ΔgacS elicited only a weak response 5 days postinfection (25% death) but increased to more than 40% following 7 days (Fig. (Fig.6).6). While overall similar results were obtained for ΔcafA, fatality/mortality caused by ΔcafA was more variable compared to that of the PA14, ΔbfiS, or ΔgacS strain since plant death varied between 0 and 58% (average, 33.3%) 5 days postinfection and between 12 and 66% (average, 45.8%) following 7 days of infection (Fig. (Fig.6).6). Taken together, these results suggest a contribution of BfiSR signaling to virulence of P. aeruginosa in nonmammalian virulence models.
Biofilm development by the opportunistic pathogen P. aeruginosa is a complex process driven by environmental signals and coordinated regulatory events. Here, we have initiated the characterization of the first TCS, BfiSR, which is required for the development of P. aeruginosa biofilms following attachment (47), by investigating BfiSR-dependent protein production and gene expression profiles, as well as its direct genomic targets. The proteomic data suggested that bfiS inactivation resulted in reduced/deficient Rhl QS signaling, which was corroborated by a biofilm-specific reduction in C4-HSL levels. Despite the Rhl defect, ΔbfiS was not impaired in rhlAB expression or swarming, a trend suggestive of a link to the RetS/LadS/GacAS/Rsm multicomponent signaling system. Moreover, Rhl-regulated genes were expressed similarly in ΔbfiS, ΔgacA, and ΔrsmYZ strains (23, 28) (Table (Table4).4). However, analysis of rsmAYZ levels in ΔbfiS biofilms indicated that BfiSR-mediated effects are distinct from the regulatory trends associated with the multicomponent system. For instance, the ΔgacA strain differed from ΔbfiS with respect to rsmYZ gene expression (4, 19, 20, 30, 70) (Table (Table4).4). Furthermore, while the ΔbfiS phenotype detected here was opposite to that of an rsmA mutant (inactivation of rsmA coincides with increased C4-HSL levels, increased expression of Rhl-regulated genes, but reduced rhlAB, rsmYZ, and TTSS expression compared to its parental strain) (3, 4, 23, 28, 30, 44, 70) (Table (Table4),4), no difference in rsmA levels was detected in ΔbfiS grown planktonically or as a biofilm (Fig. (Fig.22 and Table Table55).
Instead, inactivation of bfiS coincided with increased rsmYZ levels under biofilm, but not planktonic, growth conditions. In contrast, biofilm formation by P. aeruginosa wild type coincided with decreased rsmYZ levels. These findings are in contrast to previous reports demonstrating that reduced rsmY and rsmZ levels coincide with acute infection and P. aeruginosa remaining in the planktonic mode of growth (19, 70). Instead, here we demonstrate that the progression of biofilm development appears to correlate with the reduction of sRNA abundance (but not transcription levels) compared to planktonic cells (Table (Table55 and Fig. Fig.2).2). Reduction of rsmYZ transcription was noted as soon as cells attached to a surface. The biochemical and microscopic reporter assays suggest that initiation of surface-associated growth, in particular the transition from reversible to irreversible attachment, correlates with a downregulation of the sRNAs at the transcriptional level (Fig. (Fig.33 and Table Table5).5). As downregulation of sRNA transcription (and RNA levels) was noted for both the P. aeruingosa wild type and ΔbfiS mutant during the transition from planktonic to surface-attached growth (Fig. (Fig.33 and Table Table5;5; see Fig. S2 in the supplemental material), the observed reduction in sRNA is independent of BfiSR. However, it is likely that reduced transcription and, thus, RNA levels are dependent on reduced cell density experienced by attaching cells, considering that cell density dependence of rsmYZ expression has previously been reported for P. aeruginosa (28).
The importance of reduced rsmYZ levels for biofilm formation was supported by the finding that a ΔretS mutant overexpressing rsmYZ was comparable in architecture to ΔbfiS biofilms (Fig. (Fig.44 and Table Table6;6; see Fig. S3 in the supplemental material), while a ΔgacS mutant formed wild-type-like biofilms [see Fig. S3, although it exhibited accelerated biofilm formation (47)]. Overexpression studies using rsmY and rsmZ confirmed that increased levels of rsmZ were sufficient to arrest biofilm formation (Fig. (Fig.4),4), thus underscoring a requirement for reduced rsmZ levels for normal biofilm development to occur.
The finding that rsmYZ transcription and rsmYZ RNA levels deviated in P. aeruginosa wild-type and ΔbfiS biofilms indicated the presence of a BfiSR-dependent posttranscriptional regulatory mechanism via RNase G (CafA) in modulating rsmYZ levels. RNase G is a homologue of the essential E. coli RNase, RNase E. Whereas RNase E plays a key role in the degradation of mRNA and the processing of tRNA and rRNA in E. coli, the biological functions of RNase G appear more limited. While both ribonucleases share the propensity to cleave RNA within single-stranded segments that are AU rich and prefer RNA substrates that bear a single phosphate group at the 5′ end (22, 25, 68), RNase G has been confirmed to be required in E. coli only for adhE and eno mRNA (encoding alcohol dehydrogenase and enolase, respectively) decay and 16S rRNA processing (26, 32, 69, 71). However, the findings that cafA overexpression restores viability of an RNase E deletion mutant and reduces the accumulation of approximately 100 transcripts (31) suggest a broader role for RNase G in RNA processing. Here, we demonstrate that inactivation of cafA arrested biofilm development and resulted in elevated biofilm rsmYZ levels, comparable to those of ΔbfiS, with cafA overexpression restoring biofilm formation by ΔbfiS to wild-type levels (Fig. (Fig.55 and Table Table6).6). Restoration of biofilm formation coincided with reduced rsmZ levels without affecting rsmA or rsmY (data not shown), indicating that RNase G specifically targets rsmZ, but not rsmY, for degradation. Thus, our findings add rsmZ as a potential novel RNA substrate of ribonuclease G. Furthermore, as RNase G appears to target only rsmZ, BfiSR has to modulate rsmY and rsmZ via distinct mechanisms. rsmYZ decay studies will be conducted in the future to determine CafA specificity for rsmZ.
The differential effects of bfiS inactivation on rsmY and rsmZ are not surprising, given that the production and turnover of the CsrA/RsmA-antagonizing sRNAs are known to be modulated by a number of distinct mechanisms. For instance, inactivation of various parts of the E. coli degradosome had various effects on the decay rates of csrB and csrC, suggesting that the turnover of the two RNAs is differentially regulated (65). Similarly, in P. aeruginosa, the Sm-like RNA-binding protein Hfq interacts with and stabilizes rsmY, but not rsmZ (58, 59). The finding that BfiSR modulates rsmY and rsmZ at the posttranscriptional level via distinct mechanisms is underscored by the ChIP data suggesting that cafA is only one of many genes regulated by BfiSR. Furthermore, RNase G (CafA) is not the only posttranscriptional modulator affected in ΔbfiS. For instance, 2D-PAGE analysis revealed differential production of exoribonuclease RNase R in a BfiS-dependent manner (Table (Table2).2). RNase R has been shown to play an important role in mRNA decay, in particular, of rRNA and transcripts with substantial secondary structure (processing at colder temperatures) (6). Inactivation of rnr has been found to result in elevated levels and/or half-lives of over 100 transcripts in Pseudomonas putida (16). It is thus possible that rsmY degradation may be dependent on exoribonuclease RNase R under biofilm growth conditions.
Based on our findings, we propose a novel mechanism for the posttranscriptional control of rsmZ (Fig. (Fig.7).7). This mechanism is based on the BfiSR-dependent decay of rsmZ RNA levels via RNase G under biofilm growth conditions (Fig. (Fig.7).7). BfiSR-dependent modulation of rsmZ appears to be independent of functions ascribed to the multicomponent system with respect to rsmYZ transcription, as BfiSR modulates neither transcription of rsmYZ (Fig. (Fig.2)2) nor attachment, motility, or polysaccharide production as do other members of the LadS/RetS/GacAS/Rsm signaling system (Table (Table4)4) (47). Thus, while the LadS/RetS/GacAS/Rsm signaling system is essential for the reciprocal modulation of rsmYZ at the transcriptional level, BfiSR acts on rsmZ at the posttranscriptional level, via RNase G, in a biofilm-specific manner to reduce rsmZ levels to those required for the progression of biofilm development (Fig. (Fig.7).7). To our knowledge, this is the first demonstration of the posttranscriptional modulation of the small RNA rsmZ, a component of the RetS/LadS/GacAS/Rsm signaling system, in biofilms.
In addition to biofilm development, BfiSR signaling in P. aeruginosa plays an important role in the regulation of a number of functions that contribute to virulence. Under the conditions tested, both ΔgacS and ΔcafA displayed reduced virulence while ΔbfiS was avirulent (Fig. (Fig.6).6). As rsmYZ levels differ significantly in these strains (Fig. (Fig.2;2; Tables Tables44 and and5)5) (4, 70), it is unlikely that rsmYZ abundance can account for the reduced virulence of ΔbfiS observed here. It is likely that disruption in Rhl QS signaling (Tables (Tables22 and and3;3; Fig. Fig.1),1), responsible for regulating the expression of multiple virulence determinants and secondary metabolites, can account for the observed phenotype.
In conclusion, our data suggest that BfiSR plays an essential role in virulence and biofilm development by P. aeruginosa by triggering global changes in the production of genes/proteins associated with a wide array of functions, including virulence, transport and secretion, QS, and posttranslational/transcriptional modification. According to the results of this study, most, if not all, major effects of BfiSR can be ascribed to the combined action of rsmY and rsmZ. There are multiple levels of regulation to modulate sRNAs. The observation that BfiSR modulates rsmYZ RNA abundance, with RNase G affecting rsmZ, adds an additional level of signaling potential and regulation to an already complex sRNA-modulating system. A more thorough understanding of how bacteria activate BfiSR to regulate sRNAs may be key for developing novel strategies to target biofilms.
This work was supported by a grant from the Army Research Office (W911NF0710604).
Published ahead of print on 23 July 2010.
†Supplemental material for this article may be found at http://jb.asm.org/.