|Home | About | Journals | Submit | Contact Us | Français|
Overproduction of the exopolysaccharide alginate and conversion to a mucoid phenotype in Pseudomonas aeruginosa are markers for the onset of chronic lung infection in cystic fibrosis (CF). Alginate production is regulated by the extracytoplasmic function (ECF) σ factor AlgU/T and the cognate anti-σ factor MucA. Many clinical mucoid isolates carry loss-of-function mutations in mucA. These mutations, including the most common mucA22 allele, cause C-terminal truncations in MucA, indicating that an inability to regulate AlgU activity by MucA is associated with conversion to the mucoid phenotype. Here we report that a mutation in a stable mucoid strain derived from the parental strain PAO1, designated PAO581, that does not contain the mucA22 allele, was due to a single-base deletion in mucA (ΔT180), generating another type of C-terminal truncation. A global mariner transposon screen in PAO581 for non-mucoid isolates led to the identification of three regulators of alginate production, clpP (PA1801), clpX (PA1802), and a clpP paralogue (PA3326, designated clpP2). The PAO581 null mutants of clpP, clpX and clpP2 showed decreased AlgU transcriptional activity and an accumulation of haemagglutinin (HA)-tagged N-terminal MucA protein with an apparent molecular mass of 15 kDa. The clpP and clpX mutants of a CF mucoid isolate revert to the non-mucoid phenotype. The ClpXP and ClpP2 proteins appear to be part of a proteolytic network that degrades the cytoplasmic portion of truncated MucA proteins to release the sequestered AlgU, which drives alginate biosynthesis.
Acquisition of chronic pulmonary infection with Pseudomonas aeruginosa is a major cause of morbidity and mortality in patients with cystic fibrosis (CF). The emergence of clinical mucoid isolates of P. aeruginosa in a CF patient, due to overproduction of a capsular polysaccharide, alginate, is an indicator of decreased life expectancy (Govan & Deretic, 1996; Henry et al., 1992). The accumulation of mucus plugs in the airways, composed in part of bacterial alginate, reduces airway conductance, accelerating the rate of decline in pulmonary function. Copious amounts of alginate also enable the pathogen to grow as compact microcolonies (Worlitzsch et al., 2002), referred to as biofilms by some investigators, which increases the organism's resistance to host phagocytosis and antibiotics. Extensive research efforts over the past three decades have led to the basic understanding of the biosynthesis and regulation of alginate production in P. aeruginosa. Increased transcription of genes that encode alginate biosynthetic enzymes involves multiple regulatory and signalling molecules (Govan & Deretic, 1996; Lyczak et al., 2002; Ramsey & Wozniak, 2005). The key to mucoid conversion is the expression of the algD operon, whose transcription is increased when the extracytoplasmic function (ECF) σ factor AlgU (also called AlgT) binds to the promoter of this operon. Additional regulation is engendered by the two-component signal transduction systems encompassing both FimS–AlgR and KinB–AlgB, and the transcriptional factor AlgZ (Baynham & Wozniak, 1996), which has recently been renamed AmrZ (Baynham et al., 2006).
The activity of AlgU is regulated by the algUmucABCD cluster, which encodes a cognate anti-σ factor, MucA, as well as the negative regulators MucB and MucD. Disruption of the production of these three proteins leads to mucoid conversion. However, disruption of mucC does not result in mucoid conversion. Coincidently there is no mucC orthologue in the genome of a related alginate-producing organism, Pseudomonas syringae (Keith & Bender, 2001). The regulation of AlgU activity follows a paradigm similar to that of Escherichia coli σE (Qiu et al., 2007). According to the model, the N terminus of RseA (the MucA orthologue) is bound by ClpP (Flynn et al., 2003), and is degraded by the ClpXP protease complex (Chaba et al., 2007). The degradation of RseA is initiated by the inner membrane-associated proteases DegS and YaeL (RseP) (Alba et al., 2002; Alba & Gross, 2004; Grigorova et al., 2004; Kanehara et al., 2003; Walsh et al., 2003), followed by ClpXP-mediated degradation of the cytoplasmic N terminus of RseA to release the sequestered σE into the cytoplasm (Chaba et al., 2007). This model also suggests that MucB binds to the periplasmic domain of MucA stabilizing this anti-σ factor. MucD is an orthologue of E. coli DegP, an HtrA family serine protease involved in the degradation of misfolded proteins, and is crucial for protein quality control in the periplasm. MucD negatively regulates mucoidy (Boucher et al., 1996; Yorgey et al., 2001). We have previously provided evidence that MucD suppresses alginate overproduction by targeting misfolded and unfolded polypeptides that otherwise lead to activation of AlgW, which subsequently initiates the proteolysis of MucA and release of active AlgU (Qiu et al., 2007). Of note, AlgW production was initially thought to suppress alginate overproduction (Boucher et al., 1996). However, the homologue of AlgW in E. coli, DegS, is a protease that acts on the periplasmic domain of a MucA homologue in E. coli, RseA, to activate σE (Alba & Gross, 2004). Using mariner-based transposition mutagenesis in two non-mucoid strains of P. aeruginosa, PAO1 and PA14, we have previously identified a small envelope protein, MucE, whose production activates AlgW, leading to increased alginate production (Qiu et al., 2007). Furthermore, through screening of inducers of alginate overproduction, cycloserine has been identified as an inducer of AlgW, leading to alginate overproduction (Wood et al., 2006). The regulatory protease AlgW most likely degrades the portion of MucA found in the periplasm; however, MucA has an extensive cytoplasmic domain which is likely where the protein binds and sequesters the AlgU transcriptional activator. To date, no cytoplasmic proteases have been identified that affect the steady-state levels of the cytoplasmic domain of MucA.
Among different mucoid P. aeruginosa isolates from CF patients as well as in mucoid laboratory isolates, missense and nonsense mutations have been identified throughout the mucA gene (Anthony et al., 2002; Boucher et al., 1997). For instance, both CF patient isolate FRD1 and the laboratory-induced mucoid variant of PAO1, PAO578, harbour the same mutation in mucA (designated the mucA22 allele), and these strains have been utilized extensively to study the regulation of alginate synthesis (Govan & Deretic, 1996; Ramsey & Wozniak, 2005). The mucA mutant alleles have been identified in about 44 % of P. aeruginosa isolates from an Australian adult CF centre (Anthony et al., 2002), but in as many as 84% of mucoid P. aeruginosa isolates from other populations of CF patients (Boucher et al., 1997). Additional sequencing studies of the mucA gene from CF sputum isolates have identified some early colonizing strains that are free of the mucA mutations (Qiu et al., 2007). Thus, while mucA allelic variants such as mucA22 can commonly lead to the mucoid phenotype, other variants or controlling factors are also involved in producing this phenotype.
Prior work has identified two mucoid derivatives of strain PAO1, PAO579 and PAO581, harbouring uncharacterized genetic changes, designated muc-23 (PAO579) and muc-25 (PAO581) variants (Fyfe & Govan, 1983). We chose the latter strain to search for additional components that regulate alginate production in P. aeruginosa, and to identify the genetic change of the mucA25 variant in PAO581. We were able to determine via physical mapping of the chromosome of PAO581 that no large-scale DNA rearrangements or deletions had occurred in comparison with the sequenced parental strain PAO1 (Stover et al., 2000). The wild-type MucA protein is a trans-inner-membrane protein with 194 aa (Mathee et al., 1997). According to the NCBI Conserved Domain Database (www.ncbi.nlm.nih.gov/Structure/cdd/cdd.shtml), MucA, like E. coli RseA, has two functional domains, one located in the cytoplasm (MucA7–57, equivalent to the N-terminal region of RseA) and the other in the periplasm (MucA113–170, equivalent to the C-terminal region of RseA). In addition, the MucA84–104 region has been shown elsewhere to contain a transmembrane domain (Mathee et al., 1997; Rowen & Deretic, 2000). In this study, we found that the mucoid phenotype of PAO581 was due to a frameshift mutation resulting in the production of a truncated MucA protein, MucA25, which only retained the predicted cytoplasmic domain. Additional analysis of non-mucoid variants generated in PAO581 using a mariner-based saturation transposon mutagenesis approach identified proteases encoded within the clpP–clpX cluster (PA1801 and PA1802) and the clpP2 (PA3326) gene as new regulators of alginate biosynthesis. We demonstrate, for the MucA25 variant, that stabilization of the mucoid phenotype is due to the susceptibility of these truncated forms of MucA to proteolysis by the ClpXP and ClpP2 proteases, leading to the degradation of N-terminal MucA and release of the sequestered AlgU to activate alginate biosynthesis.
E. coli DH5α was used as the host for molecular cloning. P. aeruginosa strains PAO1 and PAO581 were grown at 37 °C (unless specified) in Luria broth (LB), or on Luria (L) agar or Pseudomonas Isolation Agar (PIA; Difco) plates, supplemented with gentamicin or carbenicillin (300 μg ml–1) when required. The strains of P. aeruginosa used for conjugative transfer of plasmids were grown at 42 °C in LB. The E. coli strains were grown in LB, or on L agar supplemented with carbenicillin (100 μg ml–1), gentamicin (13 μg ml–1) or kanamycin (40 μg ml–1), when required. Other strains as well as plasmids used are listed in Table 1.
Either electroporation or a chemical transformation method was used for the transformation of E. coli (Qiu et al., 2007). Triparental conjugations, using pRK2013 as the helper plasmid (Figurski & Helinski, 1979), were used for genetic manipulation of P. aeruginosa PAO581 and PAO1 strains.
Biparental conjugations were carried out for transposon mutagenesis, using E. coli SM10 λpir carrying plasmid pFAC as the donor strain (Wong & Mekalanos, 2000) and PAO581 as the recipient strain. After 4–6 h of mating of P. aeruginosa and E. coli cells on L agar plates at 37 °C, bacteria were streaked onto PIA plates supplemented with 300 μg gentamicin ml–1. Non-mucoid colonies were identified and subjected to further genetic analyses. The chromosomal DNA of non-mucoid mutants was isolated using the QIAamp genomic DNA kit (Qiagen). About 2 μg DNA was digested with SalI overnight at 37 °C followed by purification and self-ligation using Fast-Link DNA ligase (Epicentre). The circular closed DNA was used for inverse PCR using GM3OUT and GM5OUT primers (Table 1), which were designed based on the sequence of the Gmr gene, and the PCR products were purified and sequenced.
PFGE coupled with Southern hybridization analyses were carried out as described previously (Head & Yu, 2004) for comparison of the genome organization between P. aeruginosa strains PAO581 and PAO1. Southern blot hybridization was also used to monitor the copy number of transposon insertions using the Gmr gene as the probe template.
Bacterial cells were harvested from LB (supplemented with 150 μg carbenicillin ml–1 ). OD600 was measured and equal amounts of bacterial cells for each sample to be used were pelleted. The bacterial cells were washed with cold PBS buffer (pH 7.4) once and then pelleted for protein extraction. The total cellular protein extracts were prepared using the ReadyPreps kit (Epicentre). Protein samples were heated at 95 °C for 5 min, electrophoresed on an SDS-PAGE gel (15% acrylamide), and transferred onto a nitrocellulose membrane (pore size 0.1 μm, Schleicher & Schuell). Immunoblots were developed by using rat anti-haemagglutinin (HA) monoclonal antibody (Roche) or rabbit anti-RNAP (RNA polymerase α-subunit) polyclonal antibody (courtesy of M. J. Chamberlin, University of California, Berkeley) as the primary antibody, and horseradish peroxidase-labelled goat anti-rat (Roche) or anti-rabbit IgG (Kirkegaard & Perry laboratories) as the secondary antibody. Enhanced chemiluminescence (ECL; Amersham Biosciences) was used for detection.
The P. aeruginosa strains carrying the P1algU promoter–lacZ fusion were grown on three different PIA plates at 37 °C, and cells were harvested and resuspended in cold PBS buffer. OD600 was recorded, and β-galactosidase activity was measured after SDS/chloroform permeabilization for each sample and the Miller units were recorded. One Miller unit is equivalent to 1000 × (A420/–1.75 × A550/OD600 ml–1 min–1). The reported values represent the average in triplicate of at least three experiments.
The alginate assay was based on a method published previously (Knutson & Jeanes, 1968) with the following modifications. P. aeruginosa and mutants were grown at 37 °C on 50 ml PIA plates in triplicate for 24 h. Bacterial growth was removed from plates and suspended in 40 ml PBS. The OD600 was recorded. The alginate standard curve was made using d-mannuronic acid lactone (Sigma) in the range 0–100 μg ml–1.
Analysis of β-galactosidase was done with one-way ANOVA followed by a post-hoc analysis of all relevant pair-wise comparisons. Analysis of alginate assays was done using the t test. All analyses were performed using SigmaStat software (v. 3.5, Systat Software).
The stable mucoid variant, PAO581 was isolated in vitro following the incubation of the non-mucoid P. aeruginosa PAO with phage E79 (Fyfe & Govan, 1980). Since this strain carried undefined muc mutation(s) (designated the muc-25 variant; Fyfe & Govan, 1983), we first tested whether there were large genomic alterations in this strain. Using macrorestriction digestions of the genome coupled with separation by PFGE, no large-scale recombination, inversion, deletion, or amplification was noted in PAO581 as compared with PAO1 (see Supplementary Fig. S1). This is consistent with earlier data (Fyfe & Govan, 1983) which indicate that the mucoid phenotypes associated with muc mutation(s) designated muc-22, muc-23 and muc-25 are not associated with major changes in DNA organization.
Previously, the muc-25 locus of PAO581 had been roughly mapped to the region between pruAB (67.5 min) and hisI (69 min), very close to the algUmucABCD cluster (68 min) (Fyfe & Govan, 1983). Therefore, we sequenced the algUmucABCD genes of PAO581. A single base deletion at T180 of mucA was identified, leading to creation of a premature stop codon (TGA) at position 285. The resulting frameshift encoded a truncated polypeptide of 94 aa containing the N-terminal 59 aa of MucA. This protein was probably located in the cytoplasm due to the absence of the transmembrane (MucA84–104) and periplasmic (MucA113–170) domains. No mutation was found in the algU, mucB, mucC or mucD genes of PAO581. To determine whether this mutation could be identified in naturally occurring mucoid P. aeruginosa isolates, we sequenced the algU–mucA genes of a series of variant mucoid isolates obtained from the lungs of transgenic CF mice that emerged after 6–12 months of chronic lung infection (Coleman et al., 2003). Interestingly, the same mucA mutation as that of PAO581 was identified in one of the mucoid isolates (Strain 1003, Table 1). We next replaced the wild-type mucA gene in PAO1 with the mucAΔT180 allele from PAO581, and the resultant strain PAO1DR1 became mucoid, as expected (data not shown). The mucoid phenotype of both PAO581 and PAO1DR1 could be suppressed by the introduction of wild-type mucA in trans on a plasmid (pUCP20T-mucA) (lac promoter, data not shown). Thus, the previously uncharacterized muc-25 mutation was due to deletion of a single base at T180 in the mucA gene, and has been designated the mucA25 allele.
To determine whether other factors might be important for the expression of the mucoid phenotype in PAO581, we used mariner transposon mutagenesis, which gives rise to high-density insertion of a TA-flanked gentamicin-resistance marker into the chromosome of target bacteria, including P. aeruginosa (Qiu et al., 2007; Rubin et al., 1999; Wong & Mekalanos, 2000). In total, 86 non-mucoid mutants were isolated out of about 100 000 GmR mutants of PAO581 screened, and 60 clones were found with single insertions (see Supplementary Table S1). Insertion sites were identified by inverse PCR and sequencing. The majority of the non-mucoid mutants harboured a single insertion, as confirmed by Southern hybridization using the Gmr cassette as a probe (data not shown). As expected, most insertions were mapped to the known positive regulatory genes, including algU, algR (Deretic et al., 1989), algB (Goldberg & Dahnke, 1992) and amrZ (algZ) (Baynham & Wozniak, 1996; Tart et al., 2006). These results, while confirming the previous findings, also validated the reliability and efficiency of the mariner-based transposon mutagenesis. There were about two to three independent transposon insertions in the first seven genes of the 18 kb alginate biosynthetic operon, algD (PA3540), alg8 (PA3541), alg44 (PA3542), algK (PA3543), algE (PA3544), algG (PA3545) and algX (PA3546), and in the unlinked algC (PA5322) gene. However, no insertions were found in the other five genes, algL (PA3547), algI (PA3548), algJ (PA3549), algF (PA3450) and algA (PA3551). These results were consistent with previous findings, since transposon inactivation of algL is only possible if the transcription of the entire algD operon is suppressed (Jain & Ohman, 2005), and inactivation of algI, algJ and algF would not lead to a complete loss of mucoidy, as these genes encode proteins involved in alginate acetylation, but not biosynthesis (Franklin et al., 2004).
Several other metabolic genes were also disrupted by transposon insertions, including carA (PA4758, encoding the small subunit of carbamoylphosphate synthase), carB (PA4756, encoding the large subunit of carbamoylphosphate synthetase), purF (PA5426, encoding the glutamine phosphoribosylpyrophosphate amidotransferase), purM (PA0945, encoding the phosphoribosylformyglycinamidine cycloligase), and pyrD (PA3050, encoding dihydroorotate dehydrogenase). However, the role of these metabolic genes in the mucoid phenotype was not investigated further.
Of greater interest were the transposon insertions in the non-mucoid mutants mapped to the tig-clpP-clpX (PA1800–1802) polycistronic loci and a monocistronic locus, PA3326, a gene paralogous to clpP (designated clpP2) (Fig. 1, Table 1). The clpP2 gene encodes a polypeptide of 201 aa, homologous to ClpP (PA1801, 213 aa). Eleven out of the 51 transposon insertions into coding sequences (21.6%) were mapped to the tig (two insertions), clpP (three insertions), clpX (four insertions) and clpP2 (two insertions) loci. Southern blot analyses confirmed that only a single copy of the transposon was inserted into the chromosome in these null mutants (data not shown). The relevance of these genes to mucoid conversion was confirmed by complementation tests. The mucoid phenotype and alginate synthesis could be restored to the tig (PAO581DR23), clpP (PAO581DR58) and clpX (PAO581DR45) null mutants by the pUCP20T-borne tig-clpP-clpX (pUCP20T-tig-clpXP) genes in trans (Table 2, Fig. 2). In addition, introduction of pUCP20-clpP2 could restore mucoidy to the clpP2 null mutants of PAO581DR3 (Table 2) and PAO581DR9 (not shown), suggesting that the ClpP2 protease is also required for the mucoid phenotype. However, the non-mucoid phenotype of the tig null mutants, PAO581DR23 and PAO581DR6, seemed to result from polar effects of the transposon insertion on the downstream clpP and clpX, since pUCP20T-tig did not restore alginate synthesis (Table 2, Fig. 2, PAO581DR23). Both pUCP20T-tig-clpXP and pUCP20T-clpXP induced a high level of alginate production in the tig mutant PAO581DR23 (Table 2). To confirm the function of ClpXP and to verify that the transposon insertion did not cause other unknown mutations, we cloned the GmR interrupted allele of PAO581DR58 (clpP::GmR) into the suicide vector pEX100T and used it to disrupt the clpP gene of PAO581. The resultant clpP mutant of PAO581 was non-mucoid, and the mucoid phenotype could be complemented by plasmid-borne clpP in trans.
Since the tig-clpP-clpX gene cluster has been identified to be involved in the regulation of alginate synthesis, the tig-clpP-clpX cluster of PAO581 was subjected to sequence analyses. Compared with the tig-clpP-clpX cluster of the parental nonmucoid strain PAO1, some missense mutations were found, suggesting that these polymorphisms could be required for the mucoid phenotype of PAO581. Therefore, the tig-clpXP and clpXP genes of PAO581 were cloned into pUCP20T to test for mucoid induction in PAO1. However, overexpression of these PAO581-derived genes, tig-clpXP (PAO581), clpXP (PAO581) or the clpP2 gene from PAO581 (pUCP20T-clpP2), was not sufficient to cause mucoid conversion in PAO1 (Table 2). Furthermore, plasmid-borne alleles from PAO581 were able to restore the mucoid phenotype of the non-mucoid mutants PAO581DR58 (clpP::GmR), PA581DR23 (tig::GmR), PAO581DR45 (clpX::GmR) and PAO581DR3 (clpP2::GmR) (Table 2). The pUCP20T-tig-clpXP genes from PAO1 could also complement in trans the null mutations of tig, clpX and clpP in PAO581 (results were identical to those recorded for PAO581-derived alleles in Table 2). These results indicate that the polymorphisms of the tig-clpXP loci were not responsible for the mucoid conversion in PAO581, and represented functional allelic variants that likely arose when PAO1 was mutagenized to produce the mucoid variant of PAO581 or during the laboratory passages of PAO581.
AlgU activity was assayed in various strains, including PAO1 and PAO581, and the isogenic null mutants of tig, clpX, clpP and clpP2, as well as two other non-mucoid strains with mutations in algU (PAO581DR7) or algR (PAO581DR51) (Table 1). The P1 promoter of algU has been reported to be an AlgU-dependent autoregulatory promoter (Firoved & Deretic, 2003; Schurr et al., 1995). The P1algU promoter–lacZ fusion was integrated into the CTX phage attachment site (attB) on the bacterial chromosome of PAO581 and the isogenic mutants (Hoang et al., 2000). AlgU expression was 2.2-fold higher in PAO581 than that in the wild-type non-mucoid strain PAO1 (P<0.001 ANOVA, P<0.05 Dunnett's test) thus validating that the mucoid phenotype of PAO581 was caused by the increased level of functional AlgU (Fig. 3). Furthermore, AlgU activities of the isogenic, non-mucoid tig, clpX, clpP and clpP2 null mutants of PAO581 were comparable to that of PAO1, and were two to threefold lower than that of PAO581 (P<0.001 ANOVA, P<0.05 Dunnett's test for pair-wise comparisons). A very low level of β-galactosidase activity was detected in the algU null mutant PAO581DR7 (negative control), which was 2.8-and sixfold lower than that of PAO1 and PAO581, respectively (Fig. 3). The AlgU activity was decreased slightly in the algR null mutant PAO581DR51 (positive control), but was 1.8-fold higher than that of PAO1 (Fig. 3, P<0.001 ANOVA, P<0.05 Dunnett's test). AlgR is an AlgU-dependent downstream regulatory gene for alginate synthesis (Deretic et al., 1989), and its disruption could not affect the activity of the upstream regulator AlgU. These results suggest that ClpX, ClpP and ClpP2 are involved in the increase in AlgU activity, which is associated with increased alginate synthesis and the subsequent development of the mucoid phenotype.
The above results suggest that the tig and clp mutants could not degrade the anti-σ factor MucA protein to release AlgU, thus producing the non-mucoid phenotype in P. aeruginosa. If this is the case, overexpression of AlgU in these mutants should bypass the requirement for tig, clpP, clpX and clpP2. Therefore, the pUCP20T-PBAD-algU construct was introduced into these null mutants. All of these conjugants converted to the mucoid phenotype when inoculated onto PIA plates supplemented with 0.5–2.5% arabinose, whereas in the absence of arabinose the strains remained nonmucoid on PIA plates (data not shown). These results suggest that the endogenous AlgU is sequestered by the truncated MucA25 protein in the tig, clpX, clpP and clpP2 null mutants of PAO581, as these cofactors are apparently needed for release of AlgU from the anti-σ factor MucA.
To determine whether the conversion of P. aeruginosa to the mucoid phenotype is dependent on both a truncated MucA protein and the Clp proteases, which could degrade the cytoplasmic portion of this sequestering anti-σ factor, we created the HA-tagged mucA25 and HA-tagged wild-type mucA fusions in the pBAD/pUCP20T vector, and transferred these constructs into PAO1, PAO581, PAO581DR23 (tig::GmR), PAO581DR45 (clpX::GmR), PAO51DR58 (clpP::GmR) and PAO581DR3 (clpP2::GmR) (Table 1). As expected, there was no phenotypic change in wild-type PAO1 expressing HA-tagged mucA25 (Fig. 4). However, overexpression of the mucA25 allele suppressed the mucoid phenotype in PAO581, while the vector control had no effects. Similarly, all PAO581 strains carrying the wild-type mucA allele in trans remained mucoid when arabinose was absent, but converted to non-mucoid when the level of arabinose was >0.01% (data not shown). Similarly, the IPTG-inducible pVDtac24-based construct pDR206, carrying the HA tag–mucA fusion, and pDR207, carrying an HA tag–mucA22 fusion, could also suppress the mucoid phenotype of PAO581 under IPTG induction (data not shown). However, higher levels of inducers (>0.05% arabinose or 5 mM IPTG) were required for the mucA25- or mucA22-carrying constructs to suppress the mucoid phenotype of PAO581 than for mucoid suppression by the wild-type mucA-carrying constructs. These results indicate that the mucoid phenotype of PAO581 is due to the truncation of anti-σ factor MucA as a consequence of mucA25 mutation, and that ClpXP might be responsible for degrading the N terminus of MucA.
The in vivo stability of MucA protein was monitored by the Western blot analysis of the HA-tagged wild-type MucA protein and HA-tagged MucA25 proteins in PAO1 and PAO581. Similar levels of wild-type HA–MucA fusion protein (~25 kDa) could be detected in PAO1, PAO581 and the tig-clp mutants of PAO581 (data not shown), suggesting that the full-length HA-tagged MucA protein is stable in all of these genetic backgrounds. However, the levels of the truncated MucA25 (~15 kDa) proteins were 4.3-fold lower in PAO1 than in PAO581 after arabinose induction (Fig. 4). The levels of HA-tagged MucA25 proteins were also increased in the clp– mutants of PAO581 compared with PAO1 (Fig. 4). The level of HA-tagged MucA25 was over threefold higher in PAO581DR3 (clpP2::GmR) and PAO581DR58 (clpP::GmR) than in PAO581 (Fig. 4). There was an accumulation of the MucA25 variant protein under the condition of arabinose induction in the null mutants of tig, clpX, clpP and clpP2 (data not shown). Therefore, in the absence of the ClpXP protease, MucA25 was not degraded, suggesting that this protease complex may be responsible for the degradation of the N terminus of MucA25 to release the sequestered AlgU into the cytoplasm for activating the transcription of alginate biosynthetic genes.
In E. coli, the mature form of DegS is a periplasmic protease that regulates the activity of σE, as does the P. aeruginosa AlgW protein, the orthologue of DegS, which initiates MucA proteolysis. AlgW-mediated proteolysis can also be activated by the accumulation of misfolded periplasmic proteins (Qiu et al., 2007). To determine whether AlgW is needed for the mucoid phenotype in mucA25 mutant PAO581, we inactivated algW (PA4446) by inserting a TetR marker (algW::TetR) into the chromosomal copy of algW and also created an in-frame deletion of algW in PAO581. Both mutants still retained the mucoid phenotype, indicating that algW is not required for mucoidy in PAO581. We next disrupted the mucP (PA3649) gene of PAO581, which encodes another inner membrane-bound protease orthologous to E. coli YaeL (RseP). MucP is also involved in the cleavage of MucA. The mucP (mucP::TetR) mutant remained highly mucoid, indicating that this protease was not required for MucA cleavage and the release of AlgU in PAO581. It appears as if the truncation of MucA bypasses requirements for AlgW- and MucP-mediated cleavage of the periplasmic portion of MucA, because the truncated MucA proteins are susceptible to degradation by the active ClpXP complexes acting on the cytoplasmic portion.
As described previously, the mucE gene (PA4033) encodes a small periplasmic protein whose overexpression causes mucoid conversion in PAO1 and other isolates of P. aeruginosa with the wild-type mucA gene. Therefore, overexpression of mucE provides a strong inducing signal that leads to AlgU activation and alginate synthesis (Qiu et al., 2007), which provides a convenient way to test for the requirement of a specific gene in mucoid conversion. To determine whether ClpP and ClpX are required for mucoid conversion in wild-type PAO1, or ClpXP is a specific case for PAO581 alone, we replaced the clpP and clpX genes of PAO1 with the GmR-disrupted alleles of PAO581DR58 (clpP::GmR) and PAO581DR45 (clpX::GmR), respectively. These mutants were introduced with the mucE-overexpressing plasmid. The clpP and clpX mutants of PAO1 were unable to convert to the mucoid phenotype when mucE was induced. This suggests that ClpP is required for mucoid conversion in PAO1. To determine whether ClpXP is required for the maintenance of the mucoid phenotype in clinical isolates, we disrupted those genes in CF24, a stable mucoid strain isolated from a CF patient (Boucher et al., 1997). The resultant clpP and clpX mutants became non-mucoid and the mucoid phenotype could be restored by the plasmid-borne clpXP, suggesting that clpP and clpX are also necessary for the mucoid phenotype that arises in clinical isolates (data not shown).
We have presented evidence that ClpXP and ClpP2 proteins are needed for the degradation of the N-terminal portion of MucA resident in the cytoplasm of P. aeruginosa, and that the resultant increase in AlgU activity leads to the increased biosynthesis of alginate. By analogy to E. coli, it appears that a similar process occurs in P. aeruginosa and underlies the emergence of the mucoid phenotype associated with progression of lung disease in CF patients, as described in our proposed model for the role of Clp proteases in the MucA25-mediated mucoid phenotype (Fig. 5). However, no periplasmic protease genes were targeted in any of the non-mucoid transposon insertion mutants of PAO581. The truncation of MucA resulting from the single-base deletion leads to the loss of the periplasmic C terminus. Therefore, cleavage of truncated MucA by the E. coli orthologous proteases AlgW (DegS) and MucP (RseP/YaeL) is not necessary for the liberation of AlgU and production of the mucoid phenotype in PAO581 with the mucA25 allele.
ClpXP is an ATP-dependent serine protease complex responsible for the regulated degradation of proteins and SsrA-mediated protein quality control (Gottesman et al., 1997; Gottesman, 1999). The ring hexamers of the ClpX ATPase bind, denature and then translocate protein substrates into the degradation chamber of the double-heptameric-ring peptidase ClpP (Joshi et al., 2004). In Bacillus subtilis, ClpXP is involved in a series of cellular processes, including stationary-phase gene expression, competence development and sporulation (Msadek et al., 1998), and it is involved in virulence in some pathogenic bacteria (Butler et al., 2006). In addition, the ClpXP protease complexes regulate flagellum synthesis via the master regulator FlhD/FlhC in Salmonella enterica serovar Typhimurium (Tomoyasu et al., 2002). ClpXP also controls the expression of type III protein secretion systems through regulation of RpoS and GlrR levels in E. coli (Iyoda & Watanabe, 2005). The results presented here show that in P. aeruginosa ClpXP regulates alginate synthesis via degradation of the cytoplasmic portion of the truncated anti-σ factor MucA25. Whether or not P. aeruginosa ClpXP regulates additional cellular processes in P. aeruginosa, as it does in E. coli and Salmonella, is not yet known.
Notably, the overexpression of the tig-clpXP cluster did not induce the mucoid phenotype in wild-type PAO1, indicating that ClpXP activity alone cannot degrade the full-length MucA. The periplasmic domain of the full-length protein likely must first be cleaved by the periplasmic proteases AlgW and MucP, as has been described in the orthologous system in E. coli (Alba et al., 2002; Alba & Gross, 2004). It is also notable that unlike most eubacteria, such as E. coli, that harbour only a single copy of the clpP gene (Porankiewicz et al., 1999), P. aeruginosa harbours three copies of clpP genes, including clpP, encoding the orthologue of the E. coli ClpP protein, and the paralogous clpP2 (PA3326) and clpP3 (PA2189) genes, encoding hypothetical ClpP proteins. In this study we showed that both clpP and clpP2 are involved in the increase in alginate synthesis in strains with truncated MucA proteins. The clpP2 gene in the P. aeruginosa chromosome lacks the adjacent ORF encoding an ATPase that is found in other ClpP homologues. We also found some sequence heterogeneity in the N termini of the clpP and clpP2 genes, although the significance of this is unknown.
Mucoid strains of P. aeruginosa recovered from CF patients usually produce truncated MucA proteins (Anthony et al., 2002; Boucher et al., 1997). It appears as if these truncated proteins are more stable in mucoid strains than in wild-type strains with full-length mucA (Figs 4 and and5).5). We observed that the levels of the truncated MucA25 protein were higher in mucoid PAO581 than in wild-type PAO1 when expression was induced from the PBAD promoter using different concentrations of arabinose (Fig. 4). However, the increased expression of the MucA25 protein converted PAO581 from a mucoid to non-mucoid phenotype, indicating that the increased sequestration of AlgU prevented high-level synthesis of alginate. The stability of the truncated MucA proteins in mucoid strains may be needed to impose some level of regulation on the biosynthesis of alginate when the organism acquires the mucoid phenotype. Thus, additional control over alginate synthesis in mucoid strains might come via the ClpP proteases, which may in turn be influenced by the ClpX ATPase. However, before concluding that this hypothesis is valid, further experimental results are needed.
The involvement of at least two related but unlinked Clp proteases, ClpP and ClpP2, in alginate synthesis in stable mucoid P. aeruginosa strains would be consistent with the concept that alginate production must be regulated even when the polysaccharide is produced in copious quantities. In P. aeruginosa PAO581, both ClpP and ClpP2 were required for degradation of truncated MucA and production of the mucoid phenotype. Thus, differential expression of these two proteases could allow a mucoid cell to control the level of alginate synthesized. However, the two Clp proteases differed in their functionality when used to complement chromosomal mutations. Plasmid-borne tig-clpXP, clpXP and clpP were all able to complement the insertionally inactivated clpP and clpP2 genes. This indicates that ClpP can provide the functionality needed for conversion to the mucoid phenotype in the absence of ClpP2, if sufficient levels of ClpP are made. However, the converse was not true, as pUCP20-clpP2 could not complement the clpP null mutant PAO581DR58. This outcome suggests that there are differences in the protease activity of ClpP and ClpP2 on the MucA25/AlgU complex, consistent with the finding that the AlgU activity of PAO581DR3 (clpP2::GmR) is significantly higher than that of PAO581DR58 (clpP::GmR) (Fig. 4).
The regulation of alginate synthesis in P. aeruginosa is complicated, with direct control from proteins encoded by the algUmucABCD operon and the algR, algW, mucP and amrZ loci. However, it appears there are additional regulatory circuits, as shown here, indicating that the clpXP and clpP2 genes encode proteins likely to be involved in regulating alginate synthesis via their action on MucA/AlgU cytoplasmic complexes. The AlgsU protein also affects the production of other P. aeruginosa virulence factors, including inhibition of flagellum formation by regulating expression of fleQ (Tart et al., 2005, 2006), and expression of the RpoH heat shock-response σ factor. Another alginate regulatory factor, AlgR, inhibits the activity of the P. aeruginosa type III secretion system (Wu et al., 2004). Overall, the interactions of all of these factors and the use of master control circuits are likely to be critical for maintaining a stable and viable phenotype for the mucoid variants of P. aeruginosa.
In summary, by using mariner-based transposition and complementation assays, we have identified the tig-clpP-clpX and clpP2 loci as encoding regulatory factors that control alginate synthesis in mucoid P. aeruginosa. We also identified the muc-25 mutation as a single-base deletion (T180) in mucA. These findings add to the already extensive knowledge about regulation of alginate synthesis in P. aeruginosa, and identify additional factors that regulate how this organism can convert to the mucoid phenotype needed for maximal pathogenesis in the CF lung, yet also show that our knowledge of regulation of alginate synthesis may still not be complete.
This work was supported by a research grant (NNA04CC74G) from the National Aeronautics and Space Administration (NASA), research grants from the NASA West Virginia Space Grant Consortium and by NIH grant AI 48917-04. N. E. H. was supported by a training grant (NNG04GN56H) from the NASA Graduate Student Researchers Program (GSRP). We thank Kristy D. Dillon for the technical assistance with the alginate assay. We thank Dr John Mekalanos of Harvard Medical School for mariner transposon pFAC and Dr Daniel Wozniak of Wake Forest University School of Medicine for review and comments on this manuscript.
The GenBank/EMBL/DDBJ accession number for the algUmucABCD sequence of Pseudomonas aeruginosa PAO581 is EF635219.
A supplementary figure showing genomic profiles of PAO1 and the isogenic strain PAO581 (PAO1 mucA25), and a supplementary table listing mariner transposon insertions leading to the non-mucoid phenotype in the stable mucoid strain, are available with the online version of this paper.