We are exploiting the broad host range of the human opportunistic pathogen Pseudomonas aeruginosa strain PA14 to elucidate the molecular basis of bacterial virulence in plants, nematodes, insects and mice. In this report, we characterize the role that two PA14 gene products, MucD and AlgD, play in virulence. MucD is orthologous to the Escherichia coli periplasmic protease and chaperone DegP. DegP homologues are known virulence factors that play a protective role in stress responses in various species. AlgD is an enzyme involved in the biosynthesis of the exopolysaccharide alginate, which is hyperinduced in mucD mutants. A PA14 mucD mutant was significantly impaired in its ability to cause disease in Arabidopsis thaliana and mice and to kill the nematode Caenorhabditis elegans. Moreover, MucD was found to be required for the production of an extracellular toxin involved in C. elegans killing. In contrast, a PA14 algD mutant was not impaired in virulence in plants, nematodes or mice. A mucDalgD double mutant had the same phenotype as the mucD single mutant in the plant and nematode pathogenesis models. However, the mucDalgD double mutant was synergistically reduced in virulence in mice, suggesting that alginate can partially compensate for the loss of MucD function in mouse pathogenesis.
To explore conserved mechanisms of bacterial pathogenesis and host defence, we are using a multihost model system that involves the human opportunistic pathogen Pseudomonas aeruginosa. This model takes advantage of the fact that P. aeruginosa strain UCBPP-PA14 (PA14) causes disease in plants (Arabidopsis thaliana and lettuce; Rahme et al., 1995; 1997), insects (Galleria mellonella and Drosophila melanogaster;Jander et al., 2000) and mice (Rahme et al., 1995). PA14 also kills the nematode Caenorhabditis elegans (Mahajan-Miklos et al., 1999; Tan et al., 1999a). Previous work from our laboratory has shown that there is a remarkable degree of overlap between the virulence factors required for PA14 pathogenicity in its plant and animal hosts (Rahme et al., 1995; 1997; Mahajan-Miklos et al., 1999; Tan et al., 1999b). This permitted the relatively rapid identification of novel P. aeruginosa virulence factors required for maximum pathogenicity in mice, by screening PA14 mutant libraries for mutants that are less pathogenic in plants (Rahme et al., 1997) or that fail to kill C. elegans (Mahajan-Miklos et al., 1999; Tan et al., 1999b) or G. mellonella (S. Miyata, D. Lee, J. Villanueva, C. Sifri, S. Calderwood and F. Ausubel, unpublished data). Furthermore, the genetic tractability of the Arabidopsis, C. elegans and D. melanogaster hosts facilitates characterization of the interaction between P. aeruginosa virulence factors and host targets or defence mechanisms (Mahajan-Miklos et al., 1999). In this paper, we take advantage of the multihost model system to characterize the role in pathogenesis of a P. aeruginosa gene that shares significant homology with the Escherichia coli degP (htrA) gene.
The DegP protein was first identified in E. coli as a periplasmic serine protease that degrades improperly folded or damaged proteins and that is required for growth at high temperatures (Strauch and Beckwith, 1988; Lipinska et al., 1989; 1990;Strauch et al., 1989). Heat shock and other stresses in the cell envelope induce expression of degP via the alternate sigma factor RpoE, the RpoE regulators RseAB and the two-component sensor–activator pair CpxRA (Connolly et al., 1997; Raivio and Silhavy, 1999). Recently, E. coli DegP has also been shown to have chaperone activity that is capable of providing at least some protection from heat stress independently of its proteolytic activity (Spiess et al., 1999). In addition to degP, E. coli contains two degP homologues, degQ (hhoA) and degS (hhoB) (Bass et al., 1996; Waller and Sauer, 1996). Although neither of these genes is required for growth at high temperature, constitutive expression of degQ (but not degS) complements the inability of a degP mutant to grow at high temperature (Waller and Sauer, 1996). degP homologues have been identified in numerous Gram-negative bacteria as well as in Gram-positive bacteria, cyanobacteria, yeast, plants and humans (Pallen and Wren, 1997; Itzhaki et al., 1998).
In this report, we describe the characterization of a P. aeruginosa PA14 degP homologue that is tightly clustered with several other genes involved in regulation of the synthesis of the exopolysaccharide alginate (Govan and Deretic, 1996). The PA14 degP gene corresponds to the P. aeruginosa strain PAO1 mucD gene, mutations in which result in the formation of mucoid colonies as a result of the overproduction of alginate (Boucher et al., 1996). Because alginate has been implicated as a key virulence factor in chronic P. aeruginosa infections in the lungs of cystic fibrosis patients and because DegP homologues in other species are important virulence factors, it was of interest to determine whether P. aeruginosa PA14 MucD is a virulence factor in plant, nematode, insect or mouse pathogenesis and, if so, whether this is related to its role as a regulator of alginate biosynthesis. Our results, described in this paper and in a recent publication (Jander et al., 2000), indicate that MucD is an important virulence factor in plant, nematode, insect and mammalian hosts and that MucD is required for the production of an extracellular toxin involved in C. elegans killing. On the other hand, the P. aeruginosa PA14 AlgD protein and, by inference, its biosynthetic end-product alginate play a limited role, if any, in plant, nematode and mammalian hosts, except in a mucD mutant background in which they are necessary for pathogenesis in a mouse thermal burn injury model. In general, our results demonstrate the utility of the multihost pathogenesis model in characterizing the specific roles of P. aeruginosa virulence factors in pathogenesis.
Cloning and DNA sequence analysis of a P. aeruginosa PA14 degP homologue
A P. aeruginosa PA14 degP homologue was identified by screening a cosmid library of PA14 chromosomal DNA with a probe derived from a P. syringae degP-like gene (see Experimental procedures for details). Restriction mapping, subcloning, DNA blot analysis and DNA sequence analysis of a region of the PA14 genome that hybridized to the P. syringae degP probe identified a PA14 degP homologue located immediately downstream of a group of genes involved in the regulation of alginate biosynthesis, algU (algT), mucA, mucB (algN) and mucC (algM) (Goldberg et al., 1993; Martin et al., 1993; Deretic et al., 1994; DeVries and Ohman, 1994; Martin et al., 1994; Hershberger et al., 1995; Schurr et al., 1996; Xie et al., 1996; Mathee et al., 1997). The same P. aeruginosa degP homologue was identified independently in P. aeruginosa strain PAO1 and designated mucD (Boucher et al., 1996). It appears likely that the degP/mucD regions of the PA14 and PAO1 genomes have the same organization. Polymerase chain reaction (PCR) primers corresponding to the PA14 degP/mucD sequence and to the published PAO1 mucB sequence amplified a predicted 580 bp product from PA14 chromosomal DNA, consistent with the same gene arrangement in PA14 and PAO1 (data not shown).
DNA sequence analysis of the PA14 degP open reading frame (ORF) revealed a predicted protein product that was identical in length (474 amino acids) but that contained four conservative amino acid differences compared with the PAO1 mucD product (position PA14:PAO1; 137V:I; 225 E:Q; 363 N:S; 441 I:V) (Boucher et al., 1996). Out of 1515 bp of PA14 DNA sequence comprising mucD and flanking regions, there were 19 nucleotide differences between the PAO1 and PA14 sequences. Accordingly, we have designated the PA14 degP homologue mucD. These sequence data have been submitted to the DDBJ/EMBL/GenBank databases under accession number AF343973.
Construction of a PA14 mucD mutant
To disrupt the PA14 mucD gene, a gene cassette encoding gentamicin resistance, aacC1, was cloned into a XhoI site located in the N-terminal end of the PA14 mucD ORF. This mucD::aacC1 construct was then used to replace the wild-type copy of mucD in the PA14 genome by marker exchange using the positive selection suicide vector pCVD442 (Donnenberg and Kaper, 1991). The structure of the mucD::aacC1 disruption mutation was confirmed by DNA blot analysis (data not shown). The logarithmic growth rate of the PA14 mucD mutant was the same as wild-type PA14 in both rich (LB or KB) and minimal (M9) media. However, the mucD mutant often exhibited a 1 or 2 h longer lag before growth commenced compared with wild-type PA14 when a saturated culture was subcultured into rich or minimal media. Whereas wild-type PA14 grew well on solid media at 42°C, the PA14 mucD mutant exhibited little or no growth, similar to E. coli degP mutants (Strauch et al., 1989; Lipinska et al., 1990) and the P. aeruginosa PAO1 mucD mutant (Boucher et al., 1996). The PA14 mucD mutant formed mucoid colonies after incubation on LB, KB or PIA agar media at 37°C similar to the PAO1 mucD mutant on PIA (Boucher et al., 1996).
The PA14 mucD gene complements PA14 mucD and E. coli degP mutants
A 2.0 kbp EcoRI–SalI fragment containing the PA14 mucD gene expressed from the E. coli lacZ promoter in plasmid pPY221 complemented both the mucoid and temperature-sensitive phenotypes of the PA14 mucD strain (Fig. 1A; data not shown). A separate complementation experiment was carried out with a 5.3 kbp fragment from PA14 that contained the entire algUmucABCD cluster. This fragment, carried on plasmid pPY222, was integrated into the chromosome of the PA14 mucD strain by homologous recombination upstream of the mucD::aacC1 insertion, leaving any possible polar effects of the mucD insertion mutation intact. This integrant had the same phenotype as wild-type PA14 with respect to temperature sensitivity and mucoidy (data not shown). These results show that the phenotypes associated with the insertion in mucD resulted from the loss of mucD expression rather than an unlinked mutation or a polar effect on a downstream gene. In addition, plasmid pPY221 containing the PA14 mucD gene under lacZ promoter control partially complemented the temperature-sensitive phenotype of the E. coli degP41 mutant (Fig. 1B; Strauch et al., 1989), indicating that P. aeruginosa MucD is most probably targeted to the periplasm like E. coli DegP and has at least some overlapping substrate specificity or function.
Construction of a PA14 algD internal in frame deletion
As described below, the PA14 mucD mutant is significantly less pathogenic in plant, nematode and mouse pathogenesis models. However, the mucoid phenotype of the mucD mutant raised the concern that the overproduction of alginate per se may be the cause of reduced virulence rather than the loss of the MucD protease or chaperone activity. To address this concern and to determine the role of alginate in virulence in our pathogenesis models, we constructed a PA14 algD mutant and a mucDalgD double mutant.
The algD gene, the first gene in a large operon containing genes required for alginate biosynthesis, encodes GDP-mannose dehydrogenase, the first committed step in alginate biosynthesis (Govan and Deretic, 1996). Given that it is unknown whether all the genes in this operon are dedicated solely to alginate biosynthesis, an internal in frame deletion in the PA14 algD gene was constructed to prevent any polar effects on downstream genes. As described in Experimental procedures, a PCR-based approach was used to generate a deletion in the PA14 algD gene that retains the first six and last 36 codons of the gene. The deleted algD gene was then used to replace the wild-type copy of algD in PA14 and in PA14 mucD. As expected, the mucDalgD double mutant was no longer mucoid on PIA plates, confirming that the mucoid phenotype of the mucD mutant resulted from the overproduction of alginate. In addition, because plasmid pPY226, which contains only a wild-type copy of algD, restored mucoidy in the mucDalgD mutant background, we concluded that the algD deletion does not have a polar effect on downstream alginate biosynthetic genes (data not shown).
In the sections that follow, the PA14 mucD, algD and mucDalgD double mutants were tested for a variety of phenotypes in comparison with wild-type PA14.
Loss of MucD function confers enhanced temperature and oxidative stress sensitivity, whereas loss of AlgD has no detectable impact
To extend our earlier observation that PA14 mucD exhibited temperature sensitivity, we assessed whether the algD mutation had any impact on this phenotype. Fresh overnight cultures of wild-type PA14 or the algD mutant showed no loss of viability after transfer to 45°C for 4 h, whereas the PA14 mucD mutant and the mucDalgD double mutant both exhibited an approximate 20-fold decrease in viability (data not shown).
Increased sensitivity to oxidative stress in vitro has also been reported for a variety of degP-like mutants, including PAO1 mucD (Boucher et al., 1996). In addition, alginate has been hypothesized to provide protection against oxidative stress and has been shown to scavenge hypochlorite and reactive oxygen species generated by macrophages and neutrophils (Learn et al., 1987; Simpson et al., 1989). Sensitivity to oxidative stress was tested as described in Experimental procedures using 5% H2O2-impregnated filter disks placed on top of agarose overlays containing suspensions of the mutant and wild-type strains. To avoid complications as a result of the temperature sensitivity of the mucD mutant, low-melting-point agarose was equilibrated at 37°C before the addition of the PA14 strains and pouring the overlays. Figure 2 shows that, under the conditions tested, the mucD and mucDalgD mutants both exhibited a comparable level of enhanced sensitivity to H2O2, whereas the algD mutant was similar to wild type.
The PA14 mucD and mucDalgD mutants exhibit decreased growth in Arabidopsis, whereas the algD mutant is unimpaired
As illustrated in Fig. 3, after infiltration of ≈ 104 cfu cm−2 leaf area into an Arabidopsis leaf, the growth of the PA14 mucD and mucDalgD mutants was significantly impaired compared with wild-type PA14 and the algD mutant. The growth of the PA14 algD and mucDalgD mutants, however, was not significantly different from that of the wild-type and mucD strains respectively.
The virulence of the mucDalgD mutant is synergistically reduced in a mouse thermal injury model
Table 1 shows that the PA14 mucD mutant was less virulent in a mouse thermal injury model compared with the wild type. Plasmid pPY222 (algUmucABCD) complemented the virulence defect of the mucD mutant when it was integrated upstream of the mucD::aacC1 insertion, demonstrating that the virulence defect in the mucD mutant is not attributable to an unlinked mutation or a polar effect of the insertion. In contrast to the mucD mutant, the PA14 algD mutant exhibited virulence similar to wild type. Interestingly, the PA14 mucDalgD double mutant was dramatically less virulent than the mucD single mutant, a result that was unexpected given the minimal or absent effect of the algD mutation alone. The phenotype of the double mutant demonstrates that the impaired virulence of the mucD mutant is not caused simply by the overproduction of alginate. In fact, these results suggest that alginate may be partially compensating for the loss of MucD function in vivo.
Table 1. The mucDalgD mutant is avirulent in a mouse thermal injury model.
No. of septic mice/total
Mice that developed sepsis (% mortality)
Six-week-old male AKR/J mice were subjected to a 5% total surface body area thermal injury on their abdominal skin as described in Experimental procedures. Immediately after the burn, mice were injected with 5 × 105 PA14 cells, and the number of animals that developed sepsis was monitored each day for 10 days. mucD + integrated algUmucABCD contains the plasmid pPY222 integrated into the chromosome upstream of the mucD::aacC1 insertion. The Fisher exact values (versus wild type): mucD = 0.015; mucD + algUmucABCD= 0.40; algD = 0.35; mucDalgD = 0.000000. With the Bonferroni correction for the comparison of four test groups with the wild-type control, a P-value equivalent to 0.05 (in a two-way comparison) is 0.0125. Using this method, the mucD mutant falls near to but slightly above the 0.0125 level of significance. The mucDalgD double mutant is clearly significantly different.
mucD + integrated algUmucABCD
The mucD gene product is essential for full virulence in C. elegans, whereas algD plays no role
PA14 kills C. elegans in a medium-dependent manner when it is fed to C. elegans as a sole source of food (Mahajan-Miklos et al., 1999; Tan et al., 1999a). On low-osmolarity (minimal) medium, approximately half the worms are killed by 50 h (slow killing), whereas on high osmolarity medium, at least 50% killing occurs by 4 h (fast killing). Interestingly, slow and fast killing appear to be mechanistically distinct. Slow killing requires direct contact between live bacteria and worms and is correlated with the accumulation of large numbers of intact PA14 in the lumen of the affected worms. Accumulation of intact bacteria in the lumen is not observed when C. elegans is fed E. coli. In contrast to slow killing, fast killing is mediated by extracellular small-molecular-weight toxins and does not require direct contact between worms and live bacteria.
As shown in Fig. 4A and B, the PA14 mucD mutant was impaired in killing C. elegans in both the fast killing and the slow killing assays. In contrast, the algD and mucDalgD mutants were indistinguishable from the wild type and the mucD mutant respectively (Fig. 4A and B), suggesting that algD has no effect on virulence in either the fast or the slow killing assays. The extended time required for killing in the slow killing assays (Fig. 4B) presents two experimental problems. First, the rate at which eggs hatch internally in hermaphrodite worms is enhanced when worms are feeding on PA14 compared with E. coli. Because internal hatching of eggs kills the parent, it can be difficult to determine accurately the amount of PA14-mediated killing. Secondly, because PA14 does not kill the early larval stages, rapidly developing offspring of the original L4 (hermaphrodite) worms are difficult to distinguish from their parents, once again making it difficult to determine an accurate rate of killing. To overcome these experimental difficulties associated with the production of progeny, a male sterile mutant (fer-1;Achanzar and Ward, 1997) was substituted for wild-type N2 worms. As shown in Fig. 4C, the PA14 mucD mutant exhibited a similar defect in its ability to kill fer-1 worms as in its ability to kill wild-type worms. Figure 4C also shows that the integrated plasmid pPY222 (carrying algUmucABCD) complements the mucD mutant to full virulence in the slow killing assay, demonstrating that the reduced virulence phenotype of the mucD mutant is not caused by an unlinked mutation or a polar effect of the mucD insertion. Plasmid pPY222 also fully complemented the mucD mutation in the fast killing assay (data not shown).
MucD is required for the production of an extracellular virulence factor that is not pyocyanin
Because our laboratory had shown previously that fast killing is mediated by low-molecular-weight secreted toxins, the fact that the PA14 mucD mutant is dramatically reduced in its ability to mediate fast killing of C. elegans suggests that it is impaired in the production or export of an extracellular virulence factor or toxin. Thus, in addition to its roles in protecting against environmental stress, MucD may play a role in the production of virulence-related toxins.
PA14-mediated fast killing of C. elegans can occur without contact with live bacteria by growing a lawn of PA14 on a 0.45 µm filter placed on PGS agar and then removing the filter from the agar surface (Mahajan-Miklos et al., 1999). When L4-stage C. elegans are placed on this ‘conditioned’ agar medium, the worms die with similar kinetics to C. elegans that are in direct contact with a lawn of PA14. Consistent with the results obtained with the fast killing assay described above, the PA14 mucD strain exhibited dramatically reduced killing compared with wild-type PA14 using this filter assay (S. Mahajan-Miklos, personal communication).
One class of P. aeruginosa extracellular factors required for fast killing are phenazines, tricyclic secondary metabolites produced by a variety of pseudomonads (Turner and Messenger, 1986). Although phenazines are required, they are not sufficient for fast killing of C. elegans. PA14 mutants impaired in the C. elegans fast killing assay fall into at least two classes: mutants that produce reduced levels of the phenazine pyocyanin (class I) and mutants that synthesize wild-type levels of pyocyanin (class II). The mucD, algD and mucDalgD mutants showed no reduction in pyocyanin production in broth culture compared with wild-type PA14, exhibiting the following relative levels of production: wild type, 100% (SD ± 17); mucD, 136% (± 20); algD, 109% (± 42); mucDalgD, 221% (± 19). Surprisingly, the mucDalgD double mutant reproducibly produced elevated levels of pyocyanin. It is unlikely that the mucD mutant synthesized reduced levels of pyocyanin when grown on PGS agar in the C. elegans fast killing assay. Pyocyanin stains the agar a characteristic blue-green or brown colour depending on the composition of the medium, and no change in coloration was observed with the mucD mutant in comparison with the wild type.
Mahajan-Miklos et al. (1999) also characterized the mechanism by which phenazines (pyocyanin) kill C. elegans by demonstrating that the C. elegans P-glycoprotein (PGP) mutant [NL130 pgp-1(pk17); pgp-3(pk18)] exhibited enhanced susceptibility to PA14-mediated fast killing compared with wild-type nematodes. In general, PGPs are membrane proteins that have been implicated as components of active efflux pumps of a diverse range of exogenous toxic compounds (Higgins, 1992; Schinkel et al., 1994; Broeks et al., 1995; Higgins, 1995). Mahajan-Miklos et al. (1999) found that the enhanced susceptibility of the pgp mutant worms was correlated with pyocyanin production. That is, on high osmolarity media, the pgp mutant nematodes suppressed the phenotype of class II PA14 fast killing mutants that still synthesized normal levels of pyocyanin, but did not suppress the phenotype of class I PA14 fast killing mutants that synthesized reduced levels of pyocyanin. This suggested that pyocyanin (or other related phenazines) was a substrate of the PGP-1 and/or PGP-3 efflux pumps.
Because the PA14 mucD mutant produced at least wild-type levels of pyocyanin in broth cultures, we reasoned that pgp-1 pgp-3 mutant nematodes should suppress the virulence defect of the mucD mutant. Indeed, Fig. 5 shows that, on high-osmolarity media, the pgp-1 pgp-3 nematodes were almost as susceptible to killing by the mucD mutant as they were to wild-type PA14, consistent with the conclusion that mucD is not impaired in pyocyanin production on PGS agar medium. Under these high-osmolarity conditions, the pgp-1 pgp-3 mutant worms die at a rate comparable with wild-type N2 worms when exposed to wild-type PA14. In the case of the negative control, as expected, neither the wild-type nor the pgp-1 pgp-3 worms were susceptible to the pyocyanin-deficient PA14 mutant 3E8.
Combined, the observations that the mucD mutant exhibits a defect in the C. elegans fast killing assay, that the mucD mutant produces wild-type levels of pyocyanin and that pgp-1 pgp-3 worms are more susceptible to mucD than a pyocyanin-deficient PA14 mutant are consistent with the conclusion that MucD is involved in the production of an extracellular toxin that is not pyocyanin.
The MucD catalytic serine is required for fast killing
The recent demonstration that the E. coli DegP protein, the apparent orthologue of MucD, has chaperone activity led us to examine the respective contributions of the proteolytic and chaperone activities to MucD function with respect to three phenotypes, mucoidy, temperature sensitivity and C. elegans fast killing. To address this question, mucDS217A was constructed, in which the codon encoding the serine at the protease catalytic site (determined by alignment with E. coli degP) was mutated to encode alanine (see Experimental procedures for details). MucDS217A, carried on plasmid pPY227-3, provided no detectable complementation of the temperature sensitivity of a mucD mutant in stationary phase broth cultures at 45°C and, in fact, slightly exacerbated the temperature sensitivity, whereas the wild-type mucD clone restored temperature resistance fully (data not shown). On the other hand, as shown in Fig. 6, the mucDS217A clone did delay and partially suppress the onset of the mucoid phenotype in the mucD strain. Thus, although the in vivo stability of the altered MucD protein has not been tested, the effect of the mucDS217A clone on mucoidy indicates that MucDS217A is expressed in vivo. Moreover, the corresponding amino acid change in E. coli DegP did not affect the amount of MucD protein or make any detectable change in secondary structure (Skorko-Glonek et al., 1995; Spiess et al., 1999). In any case, the partial suppression of mucoidy by mucDS217A is consistent with the hypothesis that MucD has chaperone activity that contributes to its function, similar to the E. coli DegP protein. Finally, mucDS217A did not restore C. elegans fast killing activity to a mucD mutant, whereas wild-type mucD (pPY221) complemented fully (data not shown). Overall, these results suggest that the MucD catalytic serine is required for the production of an extracellular virulence factor(s) necessary for C. elegans fast killing.
We have characterized a variety of phenotypes of a P. aeruginosa strain PA14 mucD mutant. The predicted PA14 MucD protein is 37% identical to the E. coli DegP protein (by Lipman–Pearson protein alignment), and MucD and DegP appear to be orthologues. The PA14 mucD mutant forms mucoid colonies and exhibits increased sensitivity to temperature and oxidative stress. The simplest explanation for the mucoid phenotype is that the loss of MucD function leads to the accumulation of improperly folded or damaged proteins (or other stress signals) in the periplasm, which in turn leads indirectly to alginate overproduction.
In addition to mucoidy, the PA14 mucD mutant exhibits reduced virulence in three dramatically different pathogen–host model systems, Arabidopsis, C. elegans and mice. A reduction in virulence of the mucD mutant was also seen in a P. aeruginosa–wax moth (G. mellonella) pathogen–host model, in which wild-type PA14 yielded an LD50 of 1, whereas the PA14 mucD mutant exhibited an LD50 of 8 (Jander et al., 2000). Interestingly, the mucD mutant appears to be impaired in the production or secretion of an extracellular low-molecular-weight virulence factor required for killing C. elegans. MucD lacking the protease catalytic serine (MucDS217A) failed to complement the mucD mutation in the C. elegans fast killing assay, suggesting that the production of the fast killing toxin(s) may depend on the protease activity and not the hypothesized chaperone activity of MucD. However, MucDS217A did partially suppress the mucoid phenotype of the mucD mutant.
The mucoid phenotype of the mucD mutant led us to explore the relevance of alginate production on the pathogenicity-related phenotypes of the mucD mutant. Consequently, we constructed a deletion of a key alginate biosynthetic gene, algD, in both wild-type and mucD backgrounds. As expected, the algD deletion in the mucD background abolished the mucoid phenotype on agar plates. The ability to synthesize alginate had no effect on the temperature sensitivity or sensitivity to H2O2. With respect to pathogenicity, the algD mutation alone had minimal or no impact on virulence in an Arabidopsis leaf infiltration assay, two different C. elegans killing assays and a mouse thermal burn injury model. In contrast to these results, however, the algD mutant exhibited reduced virulence in the wax moth pathogenesis model (wild-type PA14 yielded an LD50 of 1, whereas PA14 algD mutants exhibited an LD50 of 10; Jander et al., 2000). The absence of a major role for AlgD in Arabidopsis and C. elegans was also reflected in the results obtained with the mucDalgD double mutants, which exhibited similar levels of virulence to the mucD mutant in these two models. Interestingly, however, the mucDalgD mutant was dramatically reduced in virulence in a mouse thermal injury model compared with wild type or either mutant alone. In the wax moth model, the mucDalgD double mutant also showed an additive negative impact on virulence, exhibiting an LD50 of 80 (Jander et al., 2000).
In summary, the two most significant findings of this work were the observed synergistic effect of the mucD and algD mutations on PA14 virulence in mice and the demonstration that the catalytic serine of MucD is essential for the production of an extracellular virulence factor(s) required for C. elegans fast killing.
Are E. coli degP and P. aeruginosa mucD orthologues?
Many Gram-negative bacteria examined to date contain multiple degP homologues, and mutations in these homologues often yield varied phenotypes (Pallen and Wren, 1997). P. aeruginosa PAO1 contains two degP homologues, mucD and algW, whereas E. coli contains degS and degQ. A search of the sequenced PAO1 genome (http://www.pseudomonas.com/index.html) (Stover et al., 2000) revealed no additional highly homologous family members. Several observations suggest that mucD and degP are orthologues. First, both PA14 mucD and E. coli degP mutants are temperature sensitive (Lipinska et al., 1989; Strauch et al., 1989). Secondly, mucD can partially complement the temperature sensitivity phenotype of an E. coli degP mutant. Thirdly, both mucD and degP are apparently transcriptionally regulated by orthologous stress-activated sigma factors (AlgU and RpoE respectively; Yu et al., 1995), whereas there is no evidence that algW, degQ or degS (Bass et al., 1996; Boucher et al., 1996; Waller and Sauer, 1996) is regulated in a similar fashion. In P. aeruginosa, mucD is situated downstream of algU in what appears to be an algUmucABCD operon. Although the E. coli degP gene is not linked to rpoE, the E. coli rpoE gene is in an operon with the mucABC homologues, rseABC (De Las Penas et al., 1997; Missiakas et al., 1997).
Role of mucD and algD in Arabidopsis, mouse, C. elegans and wax moth pathogenesis
The results reported here and by Jander et al. (2000) show that MucD is essential for full virulence in Arabidopsis, mice, C. elegans and wax moths. This is consistent with previous reports of reduced virulence of a PAO1 mucD mutant in a mouse septicaemia model (Yu et al., 1996). Mutations in presumptive degP orthologues in S. typhimurium (htrA) and Y. enterocolitica (htrA/gsrA) also result in attenuated virulence in mouse models (Johnson et al., 1991; Baumler et al., 1994; Li et al., 1996; Yamamoto et al., 1996). In contrast, PAO1 mutants lacking algW, which does not appear to be an orthologue of degP, showed no reduction in virulence compared with wild type in a neutropenic mouse model (Boucher et al., 1996).
In contrast to mucD, which appears to be generally required for pathogenesis in a variety of hosts and infection models, the relevance of algD to virulence is much more limited. AlgD does not appear to be a significant PA14 virulence factor for Arabidopsis or C. elegans pathogenesis. Moreover, although alginate is implicated as a virulence factor in P. aeruginosa infections in the lungs of cystic fibrosis patients and mouse lung infection clearance models (Govan and Deretic, 1996; Yu et al., 1998), our results indicate that alginate is not an important virulence factor in a mouse thermal injury model. This is consistent with a previous report that a P. aeruginosa PAO1 algD mutant was not impaired in virulence in a burned mouse model (Goldberg et al., 1995). These results suggest that, in mammalian pathogenesis, alginate may only be an important virulence factor in chronic infections that involve the formation of biofilms such as in the lungs of cystic fibrosis patients. Although alginate was not required for pathogenesis in the plant, nematode and mouse models, alginate does appear to be important for acute infection in wax moths, in which the algD mutant exhibited a 10-fold reduction in virulence, showing an LD50 of 10 versus an LD50 of 1 for wild-type PA14 (Jander et al., 2000).
In contrast to our finding that alginate does not appear to be important for PA14 pathogenesis in Arabidopsis, Yu et al. (1999) observed a significant drop in virulence in an alginate mutant of the plant pathogen P. syringae pv. syringae 3525 in its natural host bean. In addition, other exopolysaccharides have also been reported to play important roles in the virulence of a variety of plant pathogens (Leigh and Coplin, 1992; Newman et al., 1994; Condemine et al., 1999).
Do AlgD and MucD provide protection from oxidative stress?
The observation that PA14 and PAO1 mucD mutants exhibit increased sensitivity to oxidative stress in vitro (Boucher et al., 1996) is consistent with the hypothesized role of DegP orthologues in mammalian pathogens as providing protection from the deleterious effects of the macrophage oxidative burst (Johnson et al., 1991). However, in plants, recent evidence suggests that H2O2 produced in the plant defence-associated oxidative burst in tobacco leaves and potato tubers is not directly deleterious to invading microbes, including an Erwinia chrysanthemi oxyR mutant that is hypersusceptible to H2O2 (Miguel et al., 2000). If this result proves to be generally true in plant–pathogen interactions, the role of MucD in plant pathogenesis would function by some mechanism other than protecting against an oxidative burst.
Alginate has also been reported to provide protection against oxidative stress, including scavenging hypochlorite and reactive oxygen species generated by macrophages or neutrophils (Learn et al., 1987; Simpson et al., 1989). Although our in vitro assay suggests that alginate does not protect from H2O2 stress, it remains possible that it could provide protection in vivo, as the timing and levels of alginate expression and the types of reactive oxygen species encountered would probably be different. In the burned mouse infection model, alginate apparently plays a limited (if any) role in virulence in wild-type P. aeruginosa, but the fact that the mucDalgD double mutant was dramatically less virulent that the mucD single mutant suggests that the overproduction of alginate in the mucD mutant may partially compensate for the loss of MucD function. One way in which alginate overproduction could compensate for MucD function is by protecting periplasmic proteins from oxidative damage, thereby reducing the need for the repair and salvage functions of MucD. A similar mechanism could be operating in the wax moth model, given the fact that the mucDalgD mutant also exhibits dramatically lower virulence than either the mucD or the algD mutants alone.
The MucD catalytic serine is required for the synthesis or export of a low-molecular-weight toxin
There are at least two classes of toxins required for fast killing of C. elegans. One class includes phenazines, whereas the identity of the second class is unknown. The PA14 mucD mutant is impaired in fast killing, yet produces the phenazine pyocyanin at wild-type or higher levels. In addition, pgp-1 pgp-3 mutant worms exhibit enhanced sensitivity to the mucD strain. These data indicate that the mucD mutant belongs to the class of PA14 mutants impaired in C. elegans fast killing that are defective in a toxin distinct from pyocyanin. Furthermore, we have provided evidence that the production of this toxin may be dependent on the proteolytic activity of MucD, with the caveat that the stability of the presumptive proteolytically inactive mutant MucD protein has not been determined. For example, MucD could be involved in the generation of low-molecular-weight toxic peptides cleaved from a larger precursor protein, similar to the function of a degP homologue in Lactococcus lactis that is required for the proteolytic activation of an exported autolysin (Poquet et al., 2000). Alternatively, the loss of MucD proteolytic activity may result in increased cell envelope stress, which could have pleiotropic effects on the expression of a variety of genes, including some essential for fast killing. We are currently pursuing experiments to identify the toxin(s) that are deficient in PA14 mucD mutants. It will be of interest to determine whether these putative toxins play a significant role in virulence in other hosts.
Strains, plasmids and growth conditions
Pseudomonas aeruginosa strain UCBPP-PA14 (referred to as PA14 for simplicity) is a human clinical isolate capable of causing disease in animals and plants (Rahme et al., 1995; Mahajan-Miklos et al., 1999; Tan et al., 1999a). PA14 3E8 contains a TnphoA insertion mutation in the phenazine biosynthetic gene phzB (Mahajan-Miklos et al., 1999;). All cloning steps were carried out in E. coli DH5α (λpir) (Kolter et al., 1978). Wild-type E. coli strain KS272 and the isogenic mutant degP41 (strain KS474) have been described previously (Strauch et al., 1989). The plasmids used and constructed in the course of this work are listed in Table 2. Bacterial strains were grown at 37°C in Luria broth (LB; Miller, 1972), King's B or King's A media (KB, KA; King et al., 1954), M9 minimal media (Miller, 1972) or Bacto Pseudomonas isolation agar (PIA; Difco) as noted. Antibiotic concentrations used for PA14 were: rifampicin, 100 µg ml−1; gentamicin, 30 µg ml−1; carbenicillin, 300 µg ml−1; kanamycin, 200 µg ml−1.
Source of aacC1 gentamicin cassette (aacC1 was cloned from pPC110 into pUC7)
Cosmid clone carrying PA14 mucD
PA14 algUmucABCD in pUC19
mucC and 5′ end of mucD in pUC19
mucD′::aacC1 in pUC19
mucD′::aacC1 in pCVD442
Internal in frame deletion of algD in pUC19
Internal in frame deletion of algD in pCVD442
PA14 mucD in pUC19
PA14 mucD in pUCP18
PA14 algUmucABCD in pBR322
PA14 algD in pUCP19
PA14 mucD Ser-217Ala in pUCP18
Cloning and DNA sequence analysis of a P. aeruginosa PA14 degP homologue
A degP DNA probe was generated by IPCR from a degP::TnphoA mutant strain (LS46) of P. syringae pv. maculicola ES4326 (Stevens, 1998). Chromosomal DNA from this strain was digested with HincII, the resulting fragments circularized by ligation, then used as templates for IPCR with primers specific to the 5′ end of TnphoA. The TnphoA primers used were: 5′-AATATCGCCCTGAGCA-3′ (P1); 5′-ACTCACTATGCGCTGAATAA-3′ (P2). A 415 bp PCR product was obtained and sequenced. It included 245 bp of degP flanked on either side by 90 bp and 80 bp of TnphoA sequence. This PCR product was labelled and used to probe a cosmid library of PA14 chromosomal DNA (Rahme et al., 1995). One cosmid clone, pH126, that hybridized to the probe was analysed further and found to contain a 5.3 kbp SalI fragment that hybridized to the P. syringae degP probe. The 5.3 kbp SalI fragment was subcloned into pUC19 to yield pPY201. A 1.3 kbp XhoI–SalI subclone of the SalI fragment that also hybridized to the P. syringae degP probe was sequenced and found to contain a portion of a gene homologous to E. coli degP and to other known degP homologues. When the adjacent EcoRI–XhoI fragment was sequenced, the EcoRI–XhoI and XhoI–SalI fragments were found to contain the entire degP homologue ORF. To test whether the PA14 degP homologue lay immediately downstream of the algUmucABC cluster, two primers, DEG7 and MUCB1, were used to amplify a predicted 580 bp fragment from pPY201. DEG7 (5′-ACGTACGTACGTACGTACGT) corresponds to the 5′ end of PA14 mucD, and MUCB1 (5′-AGATGGTGACCGTCGTCG) corresponds to the PAO1 mucB sequence.
Construction of a PA14 mucD mutation by insertion with the gentamicin resistance cassette
A 1.5 kbp PstI fragment containing the 5′ end of the PA14 degP homologue plus upstream sequences was subcloned from pPY201 into pUC19 to yield pPY204. A SalI fragment containing the gentamicin resistance gene aacC1 from pUC7G was subcloned into the XhoI site early in the degP gene to yield pPY205. Flanking restriction sites, SalI and SphI, in the pUC19 polylinker were used to subclone this construct from pPY205 into corresponding sites in the positive selection suicide vector pCVD442 (Donnenberg and Kaper, 1991) to generate pPY206. pPY206 was then used to replace the wild-type copy of degP by marker exchange as described previously (Donnenberg and Kaper, 1991).
Construction of mucD complementing clones
The 2.0 kbp mucD-containing EcoRI–SalI fragment from pPY201 was subcloned into the corresponding sites in pUC19 to yield pPY220.
The 2.0 kbp mucD-containing EcoRI–SalI fragment from pPY201 was subcloned into the corresponding sites in pUCP18 to yield pPY221 with mucD expression driven by the lacZ promoter.
The 5.3 kbp SalI fragment from pPY201 containing the entire algUmucABCD cluster was subcloned into pBR322 using the HindIII and BamHI flanking sites in pPY201 and the corresponding sites in the tetracycline resistance gene of pBR322, generating pPY222.
The codon encoding the catalytic serine of MucD (Ser-217, numbered from full-length or unprocessed MucD) was mutated to encode alanine in pPY220 using the Stratagene QuikChange site-directed mutagenesis kit. The 0.7 kbp XhoI–PstI fragment containing this mutation was sequenced to ensure that there were no additional changes and then used to replace the corresponding fragment in pPY221.
Plasmids for complementation were electroporated into PA14, PA14 mucD, PA14 mucDalgD, E. coli (KS272) and E. coli degP41 (KS474) (Strauch et al., 1989).
Construction of PA14 algDΔ4
Flanking sequences of the algD gene were amplified by PCR from PA14 chromosomal DNA using primers corresponding to published P. aeruginosa strain PAO1 algD sequences. The following primers were used: 5′-TGCTAGTCTAGAGTTCGTCTGCAAGTCATTCGG (ALG5); 5′-CGGGATCCAAAGATGCTGATTCGC ATCG (ALG9); 5′-ATGCGAGCTCGCAGAATTCATTGGTGGTGAGG (ALG13); 5′-CCGGATCCGAGCTGTTCGTCGACCTGGTG (ALG14). To facilitate subsequent cloning, each primer was designed to contain a restriction site at its 5′ end: ALG5, XbaI; ALG9 and ALG14, BamHI; and ALG13, SacI. ALG5–ALG9 (5′ end of algD) and ALG14–ALG13 (3′ end of algD) primer pairs were used to amplify 750 and 900 bp fragments, respectively, flanking the algD gene. These PCR-amplified fragments were cloned into the XbaI and SacI sites in pUC19 as follows to generate pALGΔ4PUC: XbaI–ALG5–ALG9–BamHI and BamHI–ALG14–ALG13–SacI. This construct contains a 1.8 kbp contiguous clone in which all but the first six and last 36 codons of algD are deleted. The BamHI site linking the upstream and downstream fragments adds two additional codons and maintains the algD reading frame. Finally, the algDΔ4 construct was subcloned into the XbaI and SacI sites in pCVD442 to generate pALGΔ4CVD, which was then used to replace the wild-type copy of algD in PA14 and PA14 mucD as described previously (Donnenberg and Kaper, 1991). Both constructs were confirmed by PCR amplification of chromosomal DNA.
Construction of an algD complementing clone
The algD gene was amplified (Expand High Fidelity PCR system; Boehringer Mannheim) from PA14 chromosomal DNA using the ALG5 and ALG13 primers (see above). The amplified product was cloned into the pUCP19 XbaI–SacI sites such that expression of the algD gene was driven by the plasmid lacZ promoter.
Temperature sensitivity growth assays
Overnight cultures (5 ml) were grown in KB medium at 37°C. Aliquots (1 ml) of these cultures were then incubated on a tube roller at 45°C. Initial cfus and cfus after 4 h were determined by serial dilution and plating on KB supplemented with the appropriate antibiotics.
Hydrogen peroxide sensitivity assays
Each strain was grown overnight in LB plus rifampicin at 37°C. The following morning, 500 µl of each culture was subcultured into 5 ml of LB and grown for 5–6 h to early stationary phase. Aliquots (3 ml) of each strain were harvested by centrifugation and resuspended in 10 mM MgSO4 at an OD600 of 0.2. This suspension was added (5 µl ml−1 1.5% low melt agarose pre-equilibrated at 37°C) and used to pour 5 ml overlays on LB plates. These overlays were allowed to solidify for ≈ 2 h at room temperature. A sterile filter paper disk (Schleicher and Schuell no. 740-E, 1/4 inch) was then placed in the middle of each plate, and 10 µl of 5% H2O2 was pipetted onto the disk. Plates were incubated overnight at 37°C, and zones of inhibition were measured on the following day.
Plant pathogenicity assays
The virulence of PA14 strains was tested in Arabidopsis ecotype LL grown in a greenhouse and subsequently in a plant growth chamber as described previously (Rahme et al., 1995). Saturated cultures of PA14 were grown in LB and prepared as described previously (Rahme et al., 1995). Bacterial suspensions at an OD600 of 0.002 were infiltrated into individual leaves of 5- to 6-week-old plants as described previously for the Arabidopsis pathogen P. syringae (Dong et al., 1991). Plants were then incubated at 29°C and 90–100% humidity in a plant growth chamber. Bacterial proliferation in leaves was determined by taking two leaf punches per infiltrated leaf followed by grinding each pair of punches in 200 µl of 10 mM MgSO4 in microfuge tubes and diluting and plating for cfus on appropriate selective media.
Mouse full-thickness skin thermal burn model
Six-week-old male AKR/J mice (Jackson Laboratories) weighing between 25 and 30 g were given a 5% total surface area burn on their abdominal skin as described previously (Stevens et al., 1994). In mortality studies, immediately after the burn, mice were injected with 5 × 105 PA14 cells, and the number of animals that developed sepsis was monitored each day for 10 days. Animal study protocols were reviewed and approved by the Subcommittee on Animal Studies of the Massachusetts General Hospital. Fisher exact test with the Bonferroni correction was used for statistical analysis (http://home.clara.net/sisa/fishrhlp.htm).
Overnight cultures of PA14 and derivatives were grown in King's A + 10 µM FeCl3 with appropriate antibiotics. Overnight cultures (3 µl) were subcultured into 3 ml of fresh King's A + 10 µM FeCl3 without antibiotics (rifampicin interferes with the colorimetric assay) and grown for 16–24 h on a tube roller at 37°C. For each culture, 1 ml was centrifuged, and pyocyanin assays were conducted on supernatants as described previously (Essar et al., 1990). For each strain tested, quadruplicate cultures were grown and assayed. Pyocyanin production was normalized to cfu ml−1, the mean (and SD) calculated and represented as a percentage of wild-type production.
We thank Shalina Mahajan-Miklos for helpful discussions, and Gerald Pier and Julie Stone for critical reading of the manuscript. P.Y. was supported by an American Cancer Society Fellowship and NIH Training Grant T32 GM07035. This work was supported in part by a grant from Aventis SA to Massachusetts General Hospital.