Identification of virulence determinants in uropathogenic Proteus mirabilis using signature-tagged mutagenesis

The application of STM has been successful in identifying a number of virulence genes in many bacterial organisms (Chiang & Mekalanos, 1998; Mei et al., 1997; Sanschagrin et al., 2008; Schwan et al., 1998; Shea et al., 1996). Using this method, mutants in a mixed population can readily be identified by their unique DNA signature tag located at the end of the transposon. This attribute is particularly desirable following infection within a host in which the population of transposon mutants not recovered from an experimental infection (i.e. attenuated mutants) can be identified from among the input pool of mutants. Thus mutants absent from the recovered population, following infection and host colonization, potentially contain transposon insertions within virulence genes. For a recent and thorough review of this technique, see Mazurkiewicz et al. (2006).

Previously, we reported the findings of a non-saturating STM screen in which genes affecting motility, iron acquisition, transcriptional regulation, phosphate transport, urease activity, cell-surface structures and key metabolic pathways were found to be required for P. mirabilis UTIs (Burall et al., 2004). However, we estimated that the theoretical coverage of the genome in that study was only 46.5 %. To increase coverage to at least 70 %, we carried out a second STM screening of 1880 P. mirabilis strain HI4320 transposon mutants. Performing co-challenge competition experiments using the CBA strain mouse model of ascending UTI, we identified 37 transposon mutants that were significantly outcompeted by wild-type HI4320. The results of this unbiased study when combined with a previous STM study (Burall et al., 2004) and numerous traditional virulence studies (Allison et al., 1992a, b, 1994; Bahrani et al., 1991, 1994; Belas & Suvanasuthi, 2005; Belas et al., 1998; Bondarenko et al., 1993; Chippendale et al., 1994; Dattelbaum et al., 2003; Jacobsen et al., 2008; Johnson et al., 1993; Jones & Mobley, 1988; Jones et al., 1990; Li et al., 1999, 2002; Liaw et al., 2004; Loomes et al., 1993; Massad et al., 1994b; Mobley & Chippendale, 1990; Mobley et al., 1996; Peerbooms et al., 1983; Senior et al., 1987; Swihart & Welch, 1990; Tolson et al., 1997; Walker et al., 1999; Wang et al., 2008; Wassif et al., 1995; Welch, 1987; Wray et al., 1986; Zunino et al., 2001, 2003, 2007) provide a fairly complete picture of urovirulence properties for this agent of complicated UTIs, P. mirabilis. Newly identified genes include fimbrial, flagellar, metabolic and transport genes.

Culture media and phenotypic testing. P. mirabilis HI4320 was routinely cultured in Luria broth (LB) (l^–1: 0.5 g NaCl, 5.0 g yeast extract, 10.0 g tryptone). Both non-swarming agar (l^–1: 0.4 g NaCl, 5.0 g yeast extract, 10.0 g tryptone, 15.0 g agar, 5.0 ml glycerol) and Luria agar (l^–1: 0.5 g NaCl, 5.0 g yeast extract, 10.0 g tryptone, 15.0 g agar) were used to maintain the transposon mutants and prevent swarming (Belas et al., 1991). Swarming phenotypes of P. mirabilis transposon mutants were compared with the wild-type HI4320 swarming phenotype on swarming agar (Burall et al., 2004). Modified urea segregation agar was used to identify the presence of urease activity in the P. mirabilis transposon mutants as described previously (Burall et al., 2004; Hu et al., 1990). Minimal A medium [l^–1: 10.5 g K₂HPO₄, 4.5 g KH₂PO₄, 0.47 g sodium citrate and 1.0 g (NH₄)₂SO₄] was prepared and autoclaved. After cooling, 1 ml 1 M MgSO₄, 10 ml 20 % glycerol and 1 ml 1 % nicotinic acid were added per litre and used to screen for auxotrophic P. mirabilis transposon mutants (Belas et al., 1991). Minimal A medium agar was made by the addition of 2.5 % Bacto Agar to minimal A medium. Antibiotics were added as necessary at the following concentrations: kanamycin, 25 µg ml^–1; tetracycline, 15 µg ml^–1; and ampicillin, 100 µg ml^–1.

Generation of an STM library using bacterial conjugation. P. mirabilis transposon mutants were constructed and screened using an STM technique adapted by Hensel et al. (1995). Transformation of E. coli S17 λpir with each of the 48 pUT/mini-Tn5Km2 plasmids containing a unique DNA signature tag (provided by C. Tang, Imperial College, London, UK) was performed as previously described (Zhao et al., 1999). Unique DNA signature tags were previously screened and selected based on non-cross-reactivity with one another (Burall et al., 2004). Bacterial conjugation reactions between recipient strain P. mirabilis HI4320 and 48 E. coli S17 λpir (pUT/mini-Tn5Km2) (Amp^R Kan^R) donor strains were performed as described previously using a modified protocol (Burall et al., 2004). Overnight cultures of P. mirabilis HI4320 (Tet^R) and each of the 48 E. coli donor strains (Kan^R Amp^R) were mixed at a ratio of 2 : 1 or 5 : 1. Following centrifugation and removal of supernatant, LB medium without antibiotic was added to the pellet and held at room temperature for 5 min. After resuspension of the pellet, 50 or 100 µl of the suspension was pipetted onto membranes (0.45 µm pore size) and aseptically positioned on non-swarming agar plates without antibiotic. After drying, plates were incubated at 37 °C for 60–90 min to minimize replication of siblings. Following incubation, membranes were aseptically overlaid with LB medium without antibiotic onto non-swarming agar plates containing kanamycin and tetracycline and incubated at 37 °C overnight. Transconjugants were screened for ampicillin sensitivity to ensure loss of the pUT vector and incorporation of the mini-Tn5 transposon into the genome of P. mirabilis, as described previously (Burall et al., 2004). Simultaneously, screening for resistance to kanamycin and tetracycline was performed to select only P. mirabilis transposon mutants containing the kanamycin transposon. In the library, 59/1920 transconjugants selected (3 % conjugation rate failure) had to be reselected to create pools in which each one of the 48 unique DNA signature tags was represented only once per pool. Within each pool, one of the 48 mutants served as a negative control. The signature tag of the negative control failed to hybridize and did not cross-hybridize with other tags. For this study, 40 pools (1880 tagged transposon mutants not including negative controls) were created, each with 48 P. mirabilis transposon mutants containing a uniquely tagged transposon (Kan^R Tet^R). P. mirabilis mutants (Kan^R Tet^R Amp^S) were stored in 96-well plates at –80 °C in 80 % glycerol.

CBA mouse model of ascending UTI. To identify attenuated signature-tagged mutants, all 1880 P. mirabilis transposon mutants were tested in the CBA/J murine model of ascending UTIs as described previously (Burall et al., 2004; Hagberg et al., 1983; Johnson et al., 1993). For primary screens, each of the 40 pools was transurethrally infected into a total of five mice for a 4-day infection. Subsequent secondary screens of 23 smaller pools containing 20–25 mutants (each DNA signature tag represented only once per pool) were also transurethrally infected into a total of five mice for a 4-day infection. Co-challenge competition experiments between P. mirabilis HI4320 and putatively attenuated P. mirabilis transposon mutants were performed with a modified protocol (Burall et al., 2004). For each of the 93 competition experiments, a total of five mice were infected with either 1x10⁷ or 5x10⁷ c.f.u. of a bacterial suspension containing a 1 : 1 ratio of P. mirabilis HI4320 and a P. mirabilis transposon mutant. It was not uncommon for mice to die midway through the 7-day infection period; this appeared to be caused primarily by obstruction of urine flow by urease-mediated stone formation. In these instances, experiments were repeated with additional mice to achieve data from a minimum of five mice. Viable counts from plates were expressed in c.f.u. (ml urine)^–1 or c.f.u. (g tissue)^–1, as described previously (Burall et al., 2004).

In vitro screening using LB and minimal A media. Prior to co-challenge competition experiments, P. mirabilis transposon mutants putatively attenuated in both in vivo primary and secondary screens were tested in vitro to disqualify mutants that may have been attenuated in the mouse model due only to minor growth defects. In vitro primary and secondary screening using LB and minimal A media was performed as described previously (Burall et al., 2004). Mutants were placed in pools of 15 for primary screening and then regrouped into pools of five to ten for secondary screening.

Labelling and hybridization screening of STM mutants. For all in vivo and in vitro screens, DNA signature tags of bacteria in the input inoculum and from recovered bacteria (output sample) were PCR amplified, labelled with digoxigenin–dUTP (Roche) and hybridized to donor DNA dot-blot membranes containing a representative of each DNA signature tag. For in vivo screens, recovered bacteria were isolated from the bladder and kidneys. Visual comparison of the hybridized blots between input and output samples was performed to identify putatively attenuated transposon mutants. For in vivo primary screens, transposon mutants that lacked a hybridization signal in either or both output samples, had a reduced hybridization signal compared with the input sample or had an intensity equal to or less than that of the negative control were scored anywhere from being slightly attenuated to unrecoverable. Primary P. mirabilis transposon mutants attenuated from the primary screen were regrouped into secondary pools of 20–25 mutants each and subjected to a secondary screen. Mutants of the secondary screen in the output sample lacking a hybridization signal compared with the input sample or having an intensity equal to or less than that of the negative control were considered reproducibly unrecoverable. For in vitro primary and secondary screens, putatively attenuated mutants that were able to survive passage in LB and minimal A media, and thus were recovered as seen by hybridization on donor DNA dot blots, were kept as putatively attenuated mutants to be tested in co-challenge competition experiments. The labelling and hybridization procedure was performed for all input and output samples of primary and secondary in vivo and in vitro screens as described previously with minor modifications (Burall et al., 2004). A two-step PCR was used to amplify the DNA signature tags. A 1 : 5 dilution of sample was used as template for PCR I. Following PCR I, the product was loaded on a 1.6 % (w/v) SeaPlaque agarose gel (Cambrex) and electrophoresed. A 3.5 µl sample of the 80 bp gel slice was used for PCR II. A sample of 10–25 µl of the PCR II digoxigenin-labelled fragments was used for hybridization to donor DNA blots prepared with 10 µl plasmid DNA. The primers used in PCRs I and II are listed in Table 1.

Table 1. Oligonucleotide primers used in this study

Sequence analysis. Arbitrary PCR and molecular cloning were carried out on P. mirabilis transposon mutants that were statistically outcompeted by wild-type HI4320 to identify the gene in which the transposon was inserted. Arbitrary PCR as described by Bahrani-Mougeot et al. (2002) was carried out as two sequential amplification steps with primers complementary to the transposon end reading outward and arbitrarily designed primers reading inward from the unknown sequence. Arbitrary PCR 1 primer (Arb 1) and primer 1955 were prepared and used under PCR conditions described previously (Burall et al., 2004). Arbitrary PCR 2 was performed with arbitrary PCR 2 primer (Arb 2) and primer 1954 using 1 µl product from arbitrary PCR 1 as described previously (Burall et al., 2004). Primers used in arbitrary PCRs 1 and 2 are listed in Table 1. Gel-isolated arbitrary PCR product was either directly sequenced or cloned into pCR2.1-TOPO (Invitrogen) and sequenced at the Biopolymer Core Facility at the University of Maryland, MD, USA, or at the Sequencing Core Facility at the University of Michigan, MI, USA. The sequences of primers 1955 and 1954 are both located within the transposon. For all sequencing reactions, primer 1954 was used to determine the interrupted gene sequence.

Sequence analysis and translation were performed using online search engines at http://expasy.org. Homology searches (BLAST) of nucleotide and protein sequences (Altschul et al., 1990) were completed using the public database GenBank (). All nucleotide sequences of identified transposon mutants were examined using the Sanger Institute's database of the sequenced and annotated P. mirabilis HI4320 genome at (Pearson et al., 2008). The nucleotide sequence was analysed for presence of the transposon and placed in the Sanger Institute's P. mirabilis BLAST server to query exact sequence matches. By examining matches of unknown nucleotide sequence at the end of the transposon to known P. mirabilis sequence, a transposon insertion could be determined at the exact nucleotide within the genome. The correlating P. mirabilis nucleotide number where the transposon inserted and that of the Sanger Institute's annotation of P. mirabilis was used to identify the gene.

SDS-PAGE and Western blot analysis. Attenuated transposon mutants of P. mirabilis confirmed by co-challenge competition experiments were analysed by Western blotting for the presence of flagellin (Mobley et al., 1996) and the MrpA fimbrial subunit (Burall et al., 2004; Li et al., 1999). Bacterial suspensions were standardized at OD₆₀₀=0.8 and protein was precipitated using TCA. Samples were denatured for 5 min by boiling in sample buffer, separated by 15 % SDS-PAGE gel and transferred to a PVDF membrane (Millipore). In Fig. 1(a), samples were standardized using a 2-D Quant kit (Amersham) and 11 µg protein was loaded onto the gel. After Western blotting, membranes were incubated with rabbit polyclonal antiserum against either flagellin or MrpA followed by a secondary goat anti-rabbit IgG conjugated to horseradish peroxidase. Membranes were developed using ECL Plus Western blotting detection reagents (GE Healthcare).

Elimination of auxotrophic mutants or mutants with growth defects
The 217 transposon mutants identified as attenuated in both primary and secondary in vivo screens were tested in vitro to identify those with minor growth defects that may decrease generation time in rich LB medium and to identify mutants with auxotrophic mutations by lack of growth in minimal A medium. Following passage of the 217 transposon mutants in vitro, 141 mutants were not outcompeted in either LB or minimal A medium. The remaining 76 transposon mutants were placed in secondary pools and each was retested in both LB and minimal A media. Of these transposon mutants, 21 were outcompeted after performing morning and afternoon passages for 4 days in LB and minimal A media. Thus a total of 196 transposon mutants were not outcompeted in LB or minimal A medium and displayed no growth defects after primary and secondary in vitro screens.

Co-challenge and identification of attenuated mutants
CBA mice were co-challenged with wild-type P. mirabilis HI4320 and each of the 93 transposon mutants that were considered most attenuated out of the 196 transposon mutants based on hybridization intensity results from the dot blots. Of the 93 transposon mutants tested, 37 were statistically outcompeted by wild-type HI4320. Attenuation by at least 100-fold in urine, bladder or kidney samples was observed in 15 of the 37 statistically significant transposon mutants outcompeted in the co-challenge experiment (Table 2).

Table 2. Co-challenge with P. mirabilis HI4320 and attenuated STM mutants in the CBA mouse model of ascending UTI Five female CBA/J mice were inoculated transurethrally with 107 c.f.u. (approx. equal numbers of wild-type and mutant). After 7 days, mice were euthanized and organs were harvested to determine the number of c.f.u. g–1 or c.f.u. ml–1 for wild-type and mutant strains. NA, Not applicable (ratios were not statistically significantly different from 1); NS, not statistically significant.

Table 3. Identification of disrupted genes in attenuated P. mirabilis transposon mutants

Transposon mutants 11B2 and 20C4, also containing unique tags, had identical disruptions within cpsF, which encodes a glycosyltransferase required for lipopolysaccharide core modification (Klena et al., 1992; MacLachlan et al., 1991; Parker et al., 1992). Previously, this mutant has also been found to compromise urinary tract virulence (Allison et al., 1994). The aberrant swarming phenotype of mutant 20C4 was similar to that of a P. mirabilis U6450 TnphoA insertion mutant of cpsF unable to assemble capsular polysaccharide (Gygi et al., 1995). Upon further analysis, mutant 11B2 produced a swarming phenotype like that of HI4320 and not of 20C4. It is possible that there was a compensatory mutation in transposon mutant 11B2 restoring the swarming phenotype to that of HI4320 or that the difference in the tag sequence itself somehow governed this phenotype.

Transport
The disrupted gene of transposon mutant 7A1, exbD, encodes a biopolymer transport protein. ExbD works in conjunction with TonB and ExbB to form an energy transducing system that uses the proton motive force of the cytoplasmic membrane to transport micronutrients across the outer membrane into the cell. The TonB–ExbB–ExbD system is essential for internalization of iron–siderophore complexes across outer-membrane receptors (Braun et al., 1996; Fischer et al., 1989). Thus these mutants would be outcompeted in the iron-limited urinary tract (Alteri & Mobley, 2007).

Gene regulation
Transposon mutant 4B6 was identified as disrupting the transcriptional activator protein NhaR, a DNA-binding protein of the LysR family of transcriptional regulators, and was attenuated in the urine (P=0.001) and kidneys (P=0.004) and overall (P<0.0005) in vivo. NhaR expression has been shown to be induced in response to alkaline conditions and activates transcription of the pgaABCD operon required for poly-β-1,6-N-acetyl-D-glucosamine production (Goller et al., 2006); the latter promotes E. coli fitness by aiding cell-to-cell adhesion, attachment to surfaces and stabilization of biofilm structures (Itoh et al., 2005). Unlike uropathogenic E. coli, P. mirabilis produces urease, which creates an alkaline microenvironment (Jones et al., 1990). When P. mirabilis infects the urinary tract, urease-mediated urea hydrolysis increases environmental pH, which may cause activation of NhaR; this could lead to increased expression of genes that may ultimately contribute to the persistence of infection. Presently, gene targets of the P. mirabilis NhaR regulator have yet to be identified. The finding of nhaR in our STM screening provides evidence that the urinary tract is alkaline in the presence of P. mirabilis and supports the premise that P. mirabilis HI4320 responds to alkaline conditions by producing virulence factors to aid in UTIs.

Cell-surface structures and motility
In our previous STM study (Burall et al., 2004), only one motility-related gene, cheW, was identified. Here, we report findings of three additional motility genes important in P. mirabilis HI4320 colonization. Two genes of the flagellum regulon, fliF and flgE, were identified in our STM screen. fliF encodes protein subunits that make up the MS ring of the flagellum, which is the first synthesized substructure and is crucial for flagellar assembly (Grunenfelder et al., 2003; Kubori et al., 1992; Wu & Newton, 1997). The other flagellar transposon mutant had a disruption in the hook protein encoded by flgE. This mutant can form the PL ring complex and distal rod, but cannot assemble a complete flagellum (Komoriya et al., 1999). Both Western blot and swarming results of the flgE and fliF transposon mutants were identical; no flagellin protein was detected and a swarming-deficient phenotype was observed, probably due simply to lack of flagella synthesis. In addition, mutant 8A6, in which the transposon disrupts fliF, appeared to be overproducing mrpA (Fig. 1a).

The five known P. mirabilis fimbriae are mannose-resistant/Proteus-like fimbriae (MR/P) (Sareneva et al., 1990), P. mirabilis fimbriae (Massad & Mobley, 1994), ambient-temperature fimbriae (Massad et al., 1994a), non-agglutinating fimbriae (Tolson et al., 1995; Wray et al., 1986) and P. mirabilis P-like pili (Bijlsma et al., 1995). The completed annotation of the P. mirabilis HI4320 genome has revealed an additional 12 putative fimbrial operons (Pearson et al., 2008). Mutant 4A3 contained a disruption of the gene that encodes PMI3001, a putative fimbrial protein, which is located within one of these newly identified fimbrial operons. PMI3001 is located in fimbrial operon 14 (Pearson et al., 2008). Mutant 4A3 was attenuated and outcompeted by wild-type HI4320 in the murine model of ascending UTI, suggesting that this newly identified fimbrial operon may play a role in the pathogenesis of UTIs, as has been shown for the mrp operon (Bahrani et al., 1994).

Transposon mutant 36D2 was outcompeted in vivo and contained a disruption in PMI2575, which encodes a putative adhesin-type autotransporter. Autotransporters have been associated with virulence functions such as adhesion, aggregation, invasion, biofilm formation and toxicity (Alamuri & Mobley, 2008; Wells et al., 2007). By Western blot analysis, it appeared that mutant 36D2 was negative for synthesis of the MrpA fimbrial subunit (Fig. 1). There is no clear explanation, however, as to why a mutation in this adhesin would result in lack of MrpA expression.

Metabolism
A total of seven metabolic genes were identified in our STM screen including those involved in nucleotide, carbohydrate and amino acid metabolism. Mutation of these genes did not interfere with growth of P. mirabilis in vitro in rich medium, although disruption of guaB and aceE resulted in an increased lag phase compared with wild-type HI4320 in minimal A medium, probably due to defects in nucleotide metabolism and central metabolism, respectively. Transposon mutants 5F1 and 2E2 had disruptions in genes involved in purine metabolism. Mutant 5F1 disrupted guaB encoding inosine 5'-monophosphate dehydrogenase, which catalyses the first step of GTP synthesis. The cyaA gene disrupted in transposon mutant 2E2 encodes adenylate cyclase, which catalyses the breakdown of ATP to 3',5'-cyclic adenosine monophosphate (cAMP). Interestingly, this mutant was unable to produce either FlaA or MrpA. Four genes were involved in carbohydrate metabolism. Mutant 11G3 was disrupted in cbbC, a probable molybdopterin-containing oxidoreductase; this is similar to formate dehydrogenase, which is involved in glyoxylate, dicarboxylate and methane metabolism. Transposon mutant 11D4 disrupted edd, encoding 6-phosphogluconate dehydratase, which is involved in the Entner–Doudoroff pathway. Mutant 10E4 disrupted sdhC, which encodes the succinate dehydrogenase cytochrome b₅₅₆ subunit, which is part of the citric acid or TCA cycle. This mutant was also found by Western blot analysis not to produce MrpA, the major fimbrial subunit of MR/P fimbria. Mutant 8B5 had a transposon disruption in aceE encoding pyruvate dehydrogenase, which is involved in glycolysis. Mutant 15C3 had a disruption in sdaA encoding L-serine deaminase 1, which catalyses L-serine to pyruvate and ammonia and is involved in amino acid metabolism of glycine, serine, threonine and cysteine.

Three additional mutants identified as PMI1184, PMI1193 and PMI1000 encoded putative exported proteins and a putative phage-related protein, respectively (Table 3). A total of seven mutants (11A4, 15A6, 15B4, 17F5, 22F6, 33B5 and 37D5) were unable to be identified due to limited or poor DNA sequencing results despite persistent attempts. An additional mutant, 17A4, had a transposon insertion in the intragenic region between PMI1420 and PMI1421.

Phenotypic characterization
All 37 attenuated transposon mutants were examined for synthesis of known P. mirabilis virulence factors including flagella and fimbriae. Production of flagella and MR/P fimbriae was analysed using rabbit serum raised against flagellin and the MrpA subunit, respectively, and analysed by Western blotting. Of the 37 attenuated mutants, a total of eight had a defect in producing either of these proteins: four were unable to produce FlaA and five were unable to produce MrpA (Table 3; Fig. 1). This suggests the possibility that pleiotropic effects resulted from certain sites of transposon insertion. Modified urea segregation agar was used to identify urease-negative mutants in the library of P. mirabilis transposon mutants. Of 1880 transposon mutants tested, however, none of the STM mutants were deficient for urease enzymic activity. Screening on swarming agar was performed to identify mutants with aberrant or deficient swarming phenotypes compared with the swarming phenotype of wild-type HI4320. Visual comparison of the swarming differentiation patterns such as the number and length of swarming rafts was made between HI4320 and each transposon mutant. Of 1880 transposon mutants tested, eight had swarming patterns distinct from that of the wild-type HI4320. Swarming patterns varied from no swarming rafts, multiple short swarming rafts and a small star-shaped raft to elongated swarming rafts (Table 3; Fig. 1). The growth phenotypes of the 37 attenuated transposon mutants were verified on minimal A medium agar and also by independent growth curves in LB and minimal A media. After 24 h, all mutants grew as well as wild-type HI4320 on minimal A medium and in independent LB growth curves (data not shown). Two transposon mutants with disruptions in aceE and guaB demonstrated an increased lag phase (12 and 23 h, respectively) relative to wild-type HI4320 (2.5 h) but displayed identical exponential growth rates to the wild-type in independent liquid cultures of minimal A medium.

Conclusions
For this STM study, a total of 1880 P. mirabilis HI4320 transposon mutants were screened for virulence following transurethral challenge of approximately 500 CBA mice. A total of 40 pools, each containing 48 mutants, were tested in a primary in vivo challenge. A total of 23 groups of approximately 20–25 mutants were tested in a secondary murine screen. All attenuated mutants were tested for growth defects in LB and minimal A media. Mutants outcompeted by the wild-type strain in vitro were eliminated from further consideration. In total, 37/1880 STM transposon mutants were identified as being attenuated in vivo.

In comparison with our previous STM study (Burall et al., 2004), many mutants had transposon disruptions in genes belonging to similar classifications, although no single gene was identified in both studies. Further analysis revealed that genes identified in both studies did not belong to similar putative operons. The total number of transposon mutants created in both STM studies (3968) predicts a theoretical coverage of 69.6 % of the P. mirabilis HI4320 genome. The incomplete saturation of the genome may explain why the same gene was not identified in both STM studies. Together with the numerous virulence studies in which targeted mutants have been assessed for virulence (see Introduction), we are approaching a complete cataloguing of urovirulence genes for P. mirabilis, a prominent cause of urolithiasis, catheter-associated UTIs and other complicated UTIs.

P. mirabilis HI4320 expresses virulence factors that facilitate infection of the bladder and ascension of the ureters to the kidneys to cause pyelonephritis. Using STM, we identified 37 P. mirabilis HI4320 mutants that were significantly attenuated or exhibited reduced fitness in the murine model of ascending UTIs. Virulence factors identified by STM included genes associated with cellular processes, transport, regulation, cell-surface structures and motility, and metabolism, which contribute to the uropathogenesis of P. mirabilis. To satisfy molecular Koch's (Falkow's) postulates for these putative virulence genes, it will be necessary to construct specific mutants in the discovered genes, assess their attenuation and complement the mutants with the cloned determinants.