Novel drug targets in cell wall biosynthesis exploited by gene disruption in Pseudomonas aeruginosa

For clinicians, Pseudomonas aeruginosa is a nightmare pathogen that is one of the top three causes of opportunistic human infections. Therapy of P. aeruginosa infections is complicated due to its natural high intrinsic resistance to antibiotics. Active efflux and decreased uptake of drugs due to cell wall/membrane permeability appear to be important issues in the acquired antibiotic tolerance mechanisms. Bacterial cell wall biosynthesis enzymes have been shown to be essential for pathogenicity of Gram-negative bacteria. However, the role of these targets in virulence has not been identified in P. aeruginosa. Here, we report knockout (k.o) mutants of six cell wall biosynthesis targets (murA, PA4450; murD, PA4414; murF, PA4416; ppiB, PA1793; rmlA, PA5163; waaA, PA4988) in P. aeruginosa PAO1, and characterized these in order to find out whether these genes and their products contribute to pathogenicity and virulence of P. aeruginosa. Except waaA k.o, deletion of cell wall biosynthesis targets significantly reduced growth rate in minimal medium compared to the parent strain. The k.o mutants showed exciting changes in cell morphology and colonial architectures. Remarkably, ΔmurF cells became grossly enlarged. Moreover, the mutants were also attenuated in vivo in a mouse infection model except ΔmurF and ΔwaaA and proved to be more sensitive to macrophage-mediated killing than the wild-type strain. Interestingly, the deletion of the murA gene resulted in loss of virulence activity in mice, and the virulence was restored in a plant model by unknown mechanism. This study demonstrates that cell wall targets contribute significantly to intracellular survival, in vivo growth, and pathogenesis of P. aeruginosa. In conclusion, these findings establish a link between cell wall targets and virulence of P. aeruginosa and thus may lead to development of novel drugs for the treatment of P. aeruginosa infection.

Introduction biochemical data have been identified and predicted to have an essential function in this pathogen [25,27].
The pathogenesis of P. aeruginosa infections is multifaceted, as illustrated by numerous virulence features. Previous studies have shown that the majority of bacterial outer leaflet murein, and LPS biosynthesis enzymes are crucial for the survival of most Gram-negative bacteria [11, 13-16, 18, 28]. However, no direct relationship between these genes and virulence of P. aeruginosa has been established. In order to fill the innovation gap and to search new targets for novel antibiotics development against P. aeruginosa infections, we used "evaluate and design" strategy. In this study, a sub-set of the PAO1 cell wall, murein and LPS biosynthesis targets were subjected to in-vivo gene deletion by constructing single-gene knockout strains of PAO1 followed by survival experiments of these mutants in vitro and in vivo. We report here the results obtained with a number of k.o mutants in murA, PA4450; murD, PA4414; murF, PA4416; ppiB, PA1793; rmlA, PA5163; waaA, PA4988. By using this genetic approach, we could assess directly the role of selected targets in pathogenesis and virulence in mice.

Construction of suicide vectors for replacement of target genes
For gene replacement the sacB-based strategy based on the pEX18Ap suicide vector [29,30] was used. The suicide vectors for deletion of target genes were constructed by amplification of approximately 400 bp flanking regions using primers with integrated restriction sites (Table 1) allowing directed insertion into mobilisable vector pEX18Ap. PCR-products of upstream regions were cleaved with EcoRI and BglII and downstream regions with HindIII and BglII (only in case of PA1793 (ppiB) BamHI instead of BglII). The gentamycin-GFP cassette of pPS858 was excised using BamHI and vector pEX18Ap was digested with EcoRI and HindIII. Cleaved fragments and vector were then combined in ligation mixture (Fig 2). The constructs were confirmed by sequencing and then transformed into the E. coli donor strain ST18 which was used for conjugational transfer of the plasmids into P. aeruginosa PAO1 wild-type (WT) as described previously [29]. Briefly, E. coli ST18 containing the plasmid were streaked and transferred to LB plate containing aminolevulinic acid, carbenicillin 100 μg/mL and gentamicin 10 μg/mL. After controls and main cultures with and without antibiotics, in 1.5 mL reaction tube 100 μL PAO1 and 1 mL ST18 culture were mixed. The mixture was centrifuged at 11.000 g for 1 min, the pellet was suspended in 100 μL LB medium and plated on LB agar plate overnight at 37˚C. Using an inoculation needle, the colonies from the conjugation plate transferred to LB-antibiotic plate and incubated overnight at 37˚C. Single colonies obtained on the conjugation plate restreaked on LB-antibiotic agar plates. After counter-selection on LB-agar plates containing 5% sucrose and 80 μg/ml gentamycin, the obtained clones were tested for carbenicillin sensitivity by replica plating. In order to confirm the loss of plasmid borne DNA due to recombination events, genomic DNA was isolated from the potential k.o mutants and used in PCR together with primers specific for internal sequences of backbone of the plasmid. The correct replacement of the target gene by the gentamycin cassette was confirmed by site specific PCR and sequencing. Primers located in approximately 800 and 1400 base pairs distance from the ends of the cassette and directed outwards were used in combination with primers located minimum 500 base pairs up-and downstream from the target gene. The resulting PCR products were isolated and sequenced.
Agarose bead preparation with P. aeruginosa PAO1 WT and P. aeruginosa PAO1 mutants Agarose beads were prepared as described previously [31] with some modifications. P. aeruginosa PAO1 (ATCC 47085) wild-type (WT) was cultured overnight at 37˚C in LB and cell wall targets mutants were cultivated in LB supplemented with gentamycin (80 μg/ml). After centrifugation at 1500 g for 10 min sedimented bacteria were resuspended in 1 ml sterile PBS (pH 7.4) and added to 5 ml 2% Agarose prewarmed to 50˚C. The bacteria-agarose mixture was transferred rapidly to 5 ml 50˚C prewarmed heavy white mineral oil. After intensive vortexing the mixture was cooled on ice for 5 minutes followed by centrifugation at 1500 g for 10 min. The resulting agarose beads were washed three times in sterile PBS (pH 7.4). The load of bacteria in the agarose beads was quantified by plating 10-fold serial dilutions on Columbia blood agar plates. The inoculums for infection in mice were prepared by diluting the bead suspension with PBS (pH 7.4) to 2x10 8 CFU/ml.

Lung infection of mice and quantification of bacteria in the lung
For the infection studies we used female NMRI outbred mice at the age of min. 6 weeks. The animals were anesthetized using a Ketamin-Xylazin mixture applied by subcutaneous  injection. Mice were intratracheally infected with 50 μl of the bead suspension at a concentration of 2x10 8 CFU/ml resulting in 1x10 7 CFU. After 72 hours of infection, the mice were sacrificed by CO 2 inhalation. Lungs were isolated and homogenized through cell strainer. Mouse lung homogenates were diluted 1:1, 1:10 and 1:100 with sterile PBS (pH 7.4). The resulted dilutions were plated on Columbia blood agar plates in case of P. aeruginosa PAO1 wild-type, and with addition of gentamycin (80 μg/ml) for Knockout mutants. The colony counts were determined for each dilution after overnight incubation at 37˚C.

Macrophage-mediated bactericidal assay
Macrophage-mediated bactericidal assays were carried out as described previously with few modifications [32]. The murine macrophage cell line J774A.1 (Leibniz-Institut DSMZ-Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH) was used to examine the survival in vitro. Macrophages were grown in DMEM supplemented with 10% fetal bovine serum (FBS) in a 5% CO 2 atmosphere at 37˚C. Mid-log-phase P. aeruginosa PAO1 wild-type grown in LB broth at 37˚C and for mutants with addition of gentamycin were collected by centrifugation at 6.000 g and suspended to an OD 600 of 0.4 in DMEM without FBS. Macrophages (5 × 10 7 ) were incubated in 1 ml DMEM with the wild-type and mutants of P. aeruginosa (5 ×10 7 CFU) for 30 min at 37˚C. In order to eliminate extracellular P. aeruginosa, three washes by centrifugation at 163 g for 5 min at room temperature in PBS for wild-type and with PBS containing 400 μg/ml gentamycin were carried out. After the final wash, macrophages were allowed to adhere to tissue culture flasks in DMEM medium supplemented with gentamicin (400 μg/ml). After two hours of incubation, macrophages with internalized bacteria were harvested and lysed with 0.25% SDS and 0.025% SDS and the live intracellular bacteria were counted by plating serial dilutions of the lysates on LB plates for wild-type and LB containing gentamycin (80 μg/ml) for the mutants. The bacterial counts were determined with respect to the CFU/ml.

Plant virulence assays
The plant virulence assay was performed in lettuce leafs as described previously [33]. 10 μl of stationary-phase cultures were diluted in LB medium (10 7 CFU/ml; OD 600 = 0.01), washed twice and resuspended in 10 mM MgSO 4 . Bacterial suspensions were inoculated into the midribs of Romaine lettuce leaves. Alu-dishes containing Whatman filter papers soaked with 10 mM MgSO 4 and inoculated leaves were kept at room temperature for seven days. Symptoms were monitored daily for seven days.
Culture conditions and growth curve P. aeruginosa PAO1 was grown in LB medium or minimal M9 medium, if not otherwise stated. M9 medium contained 2% glucose as a carbon source in all cases. For growth of all PAO1 mutants both medium types were supplemented with gentamycin. For growth characteristics comparison, glycerol (2%), instead of glucose as a carbon source, was used.

Construction of k.o mutants in P. aeruginosa PAO1
For validation of cell wall biosynthesis genes as potential new drug targets in P. aeruginosa, we created single gene knockout strains of P. aeruginosa PAO1 using sacB-based strategy on the pEX18Ap suicide vector (Fig 2). After the putative k.o mutants were identified by their carbenicillin-sensitive, gentamycin-resistant, and sucrose-tolerant phenotype, correct insertion of Gm-gfp cassette at the target site and excision of pEX18Ap vector backbone was confirmed. The latter event was examined by PCR using total DNA isolated from k.o-candidates together with primers specific to pEX18Ap vector backbone and Gm-gfp cassette (S1 Fig). In case where no fragment could be amplified, clones were further tested for replacement of target gene by Gm-gfp cassette. For this, primer combinations located minimum 500 bp from the flanking regions and those located in the cassette directed outwards were used. This means that the four primers combined in four pairs of which two were expected to result in a fragment in case of correct insertion. Amplified fragments were purified and sequenced for final confirmation.
Growth characteristics of the k.o mutants: ΔrmlA failed to grow Prior to studying the effect of the k.o mutations on the mice model, we first examined the growth characteristics of the wild-type and the mutant strains in both minimal and complex media. In liquid minimal medium M9 with glucose as the sole carbon source at 37˚C, ΔmurA, ΔmurD, ΔmurF and ΔppiB mutant strains showed significantly reduced growth rate compared to that of the wild-type (Fig 3A). In minimal M9 medium with glycerol as carbon source, ΔmurA, ΔmurD, ΔmurF and ΔppiB mutant strains also showed significantly reduced growth rate compared to the parent strain ( Fig 3B). Unexpectedly, ΔrmlA mutant strain failed totally to grow in M9 neither when supplemented with glucose nor with glycerol. The doubling time of all mutants except ΔwaaA grown in the M9 medium either with glucose or glycerol was significantly longer compared to the wild-type strain. Unexpectedly the wild-type culture displayed short stationary-phase between 5 and 6 hours in M9 plus glucose and began to grow exponentially again after 6 hours. In glycerol supplemented media this short delayed growth appeared also between 6 and 7 hours. In addition, cultures of the mutant strains did not achieve the same final optical density as the parent strain in the M9 medium. In LB medium all mutants exhibited similar growth rate to wild-type except delayed growth in initial mid-log of ΔrmlA cultures (data not shown).

Morphological and colony architectural changes
Using PAO1 k.o mutant cultures grown in minimal and complex media, we investigated colony morphology of PAO1 wild-type, ΔmurA, ΔmurD, ΔmurF, ΔppiB, ΔrmlA and ΔwaaA grown in vitro. For colony morphology studies, single colonies were inoculated onto fresh medium plates (LB or M9 supplemented with glucose). All strains used in this study formed smooth colonies when cultured on LB. We noted that the ΔmurA and ΔmurF colonies in LB had different color with slightly shining and smoother textured surfaces (Fig 4A). On the other hand, dramatic changes in colonies morphology were observed when the wild-type and mutant strains were grown on M9 agar. All mutants colonies were smaller compared to wildtype except ΔrmlA strain which lost the ability to grow in minimal media (Fig 4A). Such effect in rmlA mutant (Fig 3) has also been reported for E. coli [18]. ΔmurA strain showed similar undulated colonies like wild-type but were smaller in size. ΔmurD formed flat smooth colonies with less undulations while ΔwaaA formed lobed and smooth colonies. Moreover, ΔmurF and ΔppiB colonies had irregular and highly textured surfaces with marked wrinkles (Fig 4A). Taken together, the deletion of cell wall biosynthesis genes influenced the morphological appearance especially in minimal conditions. Scanning electron microscopy showed that whereas PAO1 wild-type grown in LB exhibited normal rod shape morphology with a cell length of 1.2-2.5 μm (Fig 4B), ΔmurA, ΔmurD, ΔppiB, ΔrmlA and ΔwaaA mutants grown in  LB exhibited no alteration in cell morphology. Markedly ΔmurF cells were highly elongated ( Fig 4B) up to 6.8 μm compared to the wild-type and other mutants. This abnormal cell elongation is likely caused by defective cell division because most of elongated cells didn't show any cell dividing pattern which may indicate that the formation of septa is suppressed in ΔmurF mutants [34]. Thus, ΔmurF may have an important role in septal formation and late cell division events [34,35]. In M9 medium, length measurements demonstrated clearly that ΔmurD and ΔmurF cells are shorter when compared to wild-type and other mutants. For ΔrmlA strain in M9 very few cell found to be documented, because ΔrmlA strain lost growth ability in M9 medium and most probably these few cells were from the initial inoculum. These prominent and different changes in cell size, different colony morphologies and surface structure demonstrated that each Knockout gene affects the cellular phenotype differently.
Extracellular DNA in P. aeruginosa cultures P. aeruginosa PAO1 as well as clinical P. aeruginosa isolates are known to release extracellular DNA [4,36]. Accumulated evidence confirms that the extracellular DNA functions as a cellto-cell interconnecting matrix component in biofilms [4,[36][37][38]. Moreover, extracellular DNA induces antibiotic resistance in biofilms [4]. The extracellular DNA in the medium of growing P. aeruginosa PAO1 strains was assessed by stained agarose gel electrophoresis. The LB medium was used because different growth rates were observed in minimal medium. Software-aided DNA density volume measurements, showed that the P. aeruginosa cultures except culture of ΔmurF, contained a low basal level of extracellular DNA in the initial-log phase of growth (2 hours), and that a large amount of DNA was released in the late-log phase of growth (Fig 5). After 4 hours, sudden increase of extracellular DNA in wild-type, ΔmurD and ΔmurF cultures was observed, while the DNA amount remained low in the other mutants. This increase was followed by substantial increases of extracellular DNA after 8 hours in all except in ΔrmlA cultures. Obviously ΔrmlA cultures contained the lowest extracellular DNA among the whole time course. Fig 5 shows that a large amount of extracellular DNA was released specifically in the late-log phase of ΔmurD growth. The low level of DNA release in ΔrmlA cultures correlated well with the delayed initial mid-log in LB and suggested that quorum sensing and biofilms might be affected seriously in this mutant [36][37][38].

Cell wall mutants are sensitive to macrophage-mediated killing
On the basis of the association of outer leaflet murein, and LPS biosynthesis genes with virulence of most Gram-negative bacteria, we hypothesized that deletion of the these genes would lead to attenuation of intracellular macrophage growth [11, 13-16, 18, 28]. Encouraged by the growth rate decrease of most of the mutants in defined media, we measured the sensitivity of bacterial cells to macrophage-mediated killing. We performed macrophage cell culture invasion assay for both the wild-type and mutants using the murine J774 macrophage cell line in vitro. Macrophage cultures were infected with bacterial suspensions and intracellular bacteria were recovered in order to determine the number of surviving bacterial cells. As shown in Fig  6, all cell wall mutants were much more sensitive to macrophage-mediated killing than wildtype PAO1. These results suggested that cell wall biosynthesis genes are necessary for survival inside macrophages.

Intratracheal lung infection of NMRI mice
For further validation of outer cell wall biosynthesis genes as potential targets for novel drugs against P. aeruginosa infections, NMRI mice were intratracheally infected with agarose beads loaded with P. aeruginosa PAO1 and mutants as described in the method section. In our preliminary studies, incubation duration of 72 hours was found to yield the most significant results. Mice were separated in different groups of 10 animals each. 72 hours post infection the lungs were dissected and their homogenates were plated in different dilutions for quantification of the bacterial infectivity. In this study, P. aeruginosa PAO1 ΔmurF and ΔwaaA did not show any decrease in total bacterial counts compared to the wild-type (WT) strain (Fig 7). In contrast, ΔmurA, ΔmurD, ΔppiB and ΔrmlA show reduced bacterial burden in mice. These results demonstrated that the murA, murD, ppiB and rmlA genes are required for infectivity and growth of P. aeruginosa in an in vivo model of infection while murF and waaA genes not.

The k.o mutations in cell wall targets compromise P. aeruginosa in a plant virulence model
Apparently, the deletion of cell wall biosynthesis genes substantially decreased the in vitro intracellular survival in the murine macrophages and the in vivo survival in NMRI mice model of infection. For further assessment of the missing genes on the virulence of P. aeruginosa, we employed the lettuce leaf model of infection. Earlier studies used the plant as an in vivo model for identification of unknown P. aeruginosa virulence factors (genes) in mammalian pathogenesis [33,[39][40][41]. All these studies clearly demonstrated that the P. aeruginosa virulence mechanisms and factors are conserved between plant and animal models [40,41]. Interestingly, the plant model disclosed a significant difference between the wild-type and mutants in infection symptoms. We found that lettuce was not susceptible to infection by ΔmurD, ΔmurF and ΔrmlA (Fig 8). These mutants strain did not cause any infection symptoms to the leaves even after a prolonged incubation period. In contrast, the wild-type, caused necrotic lesions just after three days post-infection. ΔwaaA and ΔppiB elicited delayed weak rotting symptoms on lettuce stems. Surprisingly, ΔmurA proliferated in lettuce leaves and elicited disease symptoms similar to those elicited by wild-type (Fig 8). At day 7 of infection, ΔmurA P. aeruginosa strain invaded the entire midrib of a lettuce leaf resulting in severe maceration (S2 Fig); the infection severity is even more than what was observed for wild-type.

Discussion
Drug development against P. aeruginosa infections requires detailed information on the putative drug targets which are validated. Construction and characterisation of specific gene deletions of targets in the host strain is an excellent way for such validations. We constructed knockout mutants of six cell wall targets to investigate the essentiality of candidate genes. We employed a verified knockout system using pEX18Ap suicide vector as our successful tool for knockout [29,30]. We were able to inactivate the cell wall targets at the bacterial chromosomal locus. In support of a role for cell wall biosynthesis and involvement in environmental versatility of P. aeruginosa [11-14, 16, 18-21, 24], the construction of the P. aeruginosa murA, murD, murF, ppiB, rmlA and waaA mutants resulted in dramatic changes in the growth properties in minimal media compared with the wild-type (Figs 3 and 4). Major changes in the size and morphology of the colonies were also observed when cultures were grown on solid media. Previously, cell elongation in P. aeruginosa was reported to be caused by nutrient deprivation and anaerobic respiration [34,42]. It was postulated that bacteria elongate to increase their nutrient uptake and to respond to nitric oxide, a spontaneous byproduct of the anaerobic respiration.
Remarkably, here we observed that ΔmurF cells were highly elongated (Fig 4B) compared to the wild-type and other mutants. Unlike the previous reports, our results suggest that cell elongation could be a consequence of gene function loss and has significant effects on the physiological process, cell division and on formation of septal peptidoglycan [34,35]. The issues underlying the short cell size exhibited by wild-type and mutants cells and their subsequent adaptation to growth on minimal medium have not been further clarified. In this study we have clearly demonstrated that knockout of cell wall targets strongly remodels the extracellular DNA release in LB media. Significantly, ΔrmlA released very low DNA amount which suggests that rmlA contributes to interbacterial signaling and multi-cellular development processes in Pseudomonas biofilms [4,[36][37][38].
Furthermore, our results also demonstrated the impaired ability of these mutants to replicate in murine macrophages. In this study, we tested the infectivity of PAO1 mutants in the lung infection model. Using this method we are able also to examine directly whether individual gene knockout in PAO1 has lethality effect in the PAO1 strain in the lung. With two major exceptions, the growth of these k.o mutants was severely diminished in vivo in mice compared to that of wild-type (Fig 7). This indicates that P. aeruginosa PAO1 ΔmurF and ΔwaaA do not constitute a regulator controlling P. aeruginosa growth-associated virulence, but the other targets do so. Interestingly, the P. aeruginosa PAO1 ΔmurF and ΔwaaA mutants were attenuated for intracellular macrophage survival.
Discernibly, several studies have shown that the P. aeruginosa virulence determinants are conserved within animal and plant models [33,40,41]. The mutant of murA proliferated in lettuce leaves and severely caused disease symptoms even more than the wild-type especially at late infection phase (S2 Fig). It is interesting to speculate that murA mutation was spontaneously complemented in the plant model. The peptidoglycan building stone, UDP-N-MurAc, forms from UDP-N-acetyl-D-glucosamine (UNAG) and phosphoenolpyruvic acid via murA reactions. murA catalyzes unusual reaction where the transfer of the enolpyruvyl moiety of phosphoenolpyruvic acid to the 3 0 -hydroxyl group of UNAG mediated through cleavage of C-O bond of PEP and not via addition-elimination process utilizing the high energy P-O bond [43,44]. The shikimate pathway in plants contains AroA (5-enolpyruvylshikimate-3-phosphate synthase), the only enzyme known to catalyze the identical unusual reaction as murA [44,45]. This finding indicates that the AroA enzyme from the plant might replace the function of murA. Other possible explanation is the abolition of the synthesis of UDP-N-MurAc, could be replaced by another similar sugar from the plant.
Understanding the details of aftereffect of the functional loss of targeted gene would lay the framework for a focused strategy of drug design. Future work could now be directed towards whole-genome effects of these mutants. Microarray-based comparative techniques could be applied to learn more about the interplay between these mutations and ecological versatility of this pathogen. The current report provides initial insights into the relationship between the targets studied and the cell and colony characteristics of P. aeruginosa. This study also provides definitive evidence that cell wall targets activity contributes to intramacrophage/mice survival and pathogenesis. Thus, these results identify peptidoglycan/LPS assembly in the P. aeruginosa cell wall as essential targets which are expected to significantly increase the probability of finding novel anti-pseudomonas therapeutic drugs.

S1 Fig. Genetic analysis by specific PCR to pEX18Ap vector backbone and Gm-gfp cassette.
Knockout of the native genes in the PAO1 strain was verified by PCR analysis. Chromosomal DNA of mutant strains was used as templates in PCR reactions. Upper gel part): using internal specific primers for Gm-gfp cassette, the expected 425 bp. Which confirm the insertion of the Gm-gfp cassette at the chromosomal locus. Lower gel part): For these mutants the absence of pEX18Ap vector backbone was confirmed by using backbone specific primers (538 bp PCR Product). The PCR reactions were analyzed by gel-electrophoresis on 1% agarose gel. The first lane contains molecular size markers (GeneRuler 1 kb DNA Ladder, Thermo Scientific). (PDF)