Deacidification by FhlA-dependent hydrogenase is involved in urease activity and urinary stone formation in uropathogenic Proteus mirabilis

Proteus mirabilis is an important uropathogen, featured with urinary stone formation. Formate hydrogenlyase (FHL), consisting of formate dehydrogenase H and hydrogenase for converting proton to hydrogen, has been implicated in virulence. In this study, we investigated the role of P. mirabilis FHL hydrogenase and the FHL activator, FhlA. fhlA and hyfG (encoding hydrogenase large subunit) displayed a defect in acid resistance. fhlA and hyfG mutants displayed a delay in medium deacidification compared to wild-type and ureC mutant failed to deacidify the medium. In addition, loss of fhlA or hyfG decreased urease activity in the pH range of 5–8. The reduction of urease activities in fhlA and hyfG mutants subsided gradually over the pH range and disappeared at pH 9. Furthermore, mutation of fhlA or hyfG resulted in a decrease in urinary stone formation in synthetic urine. These indicate fhlA- and hyf-mediated deacidification affected urease activity and stone formation. Finally, fhlA and hyfG mutants exhibited attenuated colonization in mice. Altogether, we found expression of fhlA and hyf confers medium deacidification via facilitating urease activity, thereby urinary stone formation and mouse colonization. The link of acid resistance to urease activity provides a potential strategy for counteracting urinary tract infections by P. mirabilis.


Scientific Reports
| (2020) 10:19546 | https://doi.org/10.1038/s41598-020-76561-w www.nature.com/scientificreports/ medium deacidification during mix acid fermentation 17 and acid resistance in the anaerobic environment 18 . In addition, hydrogenases of FHL might play a role in virulence by neutralizing hydroxyl free radicals (OH·), providing energy and maintaining acid-base homeostasis 19 . In this regard, fhlA and FHL-2 genes have been shown to be induced in UPEC during urinary tract infections 20 . Recently, fhlA and the genes encoding the components of FHL-2, were found to be likely to contribute to mouse colonization in uropathogenic Proteus mirabilis by the genome-wide study 21 . In view that enzymes and the maturation machinery required for production of FHL hydrogenases are completely different from human proteins 22 , these pathways would be promising targets for the development of new antimicrobial strategies. Proteus mirabilis is an important pathogen of the urinary tract, especially in patients with indwelling urinary catheters 5 . UTIs and CAUTIs involving P. mirabilis are typically complicated by the formation of bladder and kidney stones due to alkalinization of urine from urease-catalyzed urea hydrolysis, leading to catheter blockage and permanent renal damage 3 . In addition, urinary stones deposit on the catheter surface, facilitating the formation of crystalline biofilms 3 . On the basis of the genome-wide study by Armbruster et al. 21 , we investigated if FhlA and FHL hydrogenase are associated with virulence factor expression of P. mirabilis. Here, we show that FhlAregulated hydrogenase gene expression accelerated medium deacidification, whereby facilitating urease activity and urinary stone formation. Moreover, we confirmed the fitness defect of fhlA and hydrogenase gene (hyfG) mutants in UTI mouse model of P. mirabilis. This is the first report revealing P. mirabilis fhlA and hydrogenase genes participate in urease activity, urinary stone formation and virulence. This finding provides a perspective for development of new therapeutic approaches to counteract UTIs caused by P. mirabilis.

Results
Counterparts of the FHL hydrogenase hyc locus, fhlA and fdhF genes in P. mirabilis N2. To investigate the role of FhlA and FHL in uropathogenic P. mirabilis, we searched the genome of P. mirabilis HI4320 for the homologous genes of FhlA, FHL hydrogenase and formate dehydrogenase H (FDH-H encoded by fdhF). Only a putative FHL hydrogenase 4 operon (hyf) is present in P. mirabilis HI4320. No similar FHL hydrogenase 3 operon (hyc) was found in P. mirabilis HI4320. We amplified the entire DNA fragment (11,956 bp) containing hyfABCDEFGHIJhycI and the upstream region of hyfA from the genomic DNA of P. mirabilis N2 using the KOD DNA polymerase of high fidelity and efficiency by primers designed from P. mirabilis HI4320 (Table 1) and the sequence of the product was determined. hyfA and hyfH may encode electron carrier proteins containing 4Fe-4S domain; hyfB, hyfD and hyfF may encode proton-conducting membrane transporters; hyfC and hyfE may encode membrane-anchored subunits; hyfG and hyfI may encode [NiFe] hydrogenase large and small subunits respectively; hyfJ and hycI may encode proteins for maturation of HyfG. As shown in Fig. 1a, the P. mirabilis N2 hyf locus consists of hyfABCDEFGHIJhycI. The putative proteins encoded by hyf locus of P. mirabilis N2 shared 59-86% similarities with their respective orthologues in E. coli MG1655 and CFT073 respectively, with highest similarity between hydrogenase large subunit (HyfG vs HycE) (Fig. 1a). The nucleotide sequences of fhlA (2572 bp) and fdhF (2810 bp) containing the upstream promoter region were also acquired by amplification with primers designed from P. mirabilis HI4320 (Table 1) and sequencing the product. The BLAST search revealed FhlA and FDH-H orthologues with 71% and 85% similarities to that of uropathogenic E. coli CFT073 and nonpathogenic E. coli MG1655 respectively (Fig. 1a). The nucleotide sequences of fhlA, hyf locus and fdhF were assigned accession number MT492462, MT492463 and MT492464 respectively by the GenBank database.  18 , we first investigated whether FhlA and Hyf hydrogenase affect acid resistance in P. mirabilis N2. We performed homologous recombination to construct kanamycin-resistant fhlA and hyfG mutants, followed by Southern blotting to verify the mutant clone. In view that the large subunit of hydrogenase is critical for hydrogenase activity 18 , hyfG deletion stands for functional loss of Hyf activity in the study. There was no difference between the growth of the wild-type bacteria and respective mutants (Fig. 6c). We then tested the ability of the wild-type strain, fhlA and hyfG mutants and respective complemented strains to survive the acid exposure. We found that the survival rate of both fhlA and hyfG mutants was significantly lower than that of the wild-type strain and the respective complemented strains (Fig. 1b). The results indicate that FhlA and Hyf are involved in resistance to extreme acid exposure in P. mirabilis. FhlA was reported as an enhancer binding protein which could hydrolyze ATP to activate RpoN-dependent transcriptions through the RpoN-interacting motif, GAFTGA 23 . Thus, we constructed the site-directed mutant, fhlAcSD strain (the fhlA mutant containing an altered FhlA with the conserved RpoN-interacting motif GAFTGA changing into GAISGA in the pGEM-T Easy vector) which exhibited no growth defect compared to the wild-type bacteria. Then we assessed the survival in acid using the fhlAcSD strain, rpoN mutant and the rpoN-complemented strain. The survival of fhlAcSD strain and rpoN mutant was reduced relative to the wild-type and the rpoN-complemented strain. We concluded that FhlA (requiring the GAFTGA motif for RpoN-interaction), RpoN and HyfG contribute to the acid resistance (Fig. 1b).
hyf and fdhF are regulated by FhlA and induced by formate and anaerobiosis. In E. coli, the transcription of both fdhF and hyf operons is regulated by FhlA 8,10 . To identify conserved promoter elements for FhlA-dependent transcriptions, we submitted position weight matrices of FhlA and RpoN from characterized FhlA-and RpoN-dependent promoters ( Table 2) to the Regulatory Sequence Analysis Tools (RSAT) sever 24 . TGG CAC GNNNNTTGCA/T and the palindromic sequence TGA/TC-A/GAA/TA/GAT-GA/TCA were shown to be the conserved binding sites for RpoN and FhlA respectively. We found two putative FhlA and one RpoN conserved regulatory sequences present in the promoter regions of all the determined fhlA, hyf and fdhF sequences (Fig. 2a). We therefore performed the reporter assay to examine if fhlA, hyf and fdhF are regulated by FhlA and RpoN. We found hyf and fdhF promoter activities were reduced in the fhlA and rpoN mutants relative to the wild-type and respective complemented strains at 3, 5, 7, and 24 h after incubation (Fig. 2b,c). In addition, the RpoN-interacting domain of FhlA is responsible for regulation of hyf and fdhF promoter activities (Fig. 2b,c). According to the RpoN conserved regulatory sequence not in the fhlA gene direction, the reporter assay showed loss of rpoN or RpoN-dependent fhlA had no effect on fhlA promoter activity during the time period tested (Fig. 2d). It is known fdhF, hyf and fhlA could be induced by either formate or anaerobiosis in E. coli 8,10,25 . We tested the effect of 30 mM formate or anaerobic condition on the promoter activities of fhlA, fdhF and hyf after incubation for 5 h. We found P. mirabilis fdhF and hyf were induced significantly by 30 mM formate and anaerobiosis respectively (Fig. 2f,g) while formate and anaerobiosis seemed not have a significant effect on fhlA promoter activity (Fig. 2e). The results of fhlA promoter activity contrasted with previous reports that showed formate and anaerobiosis could trigger expression of FhlA-regulated hypABCDE-fhlA in E. coli 10,26 . Figure 2b-d showed the wild-type P. mirabilis displayed hyf, fdhF and fhlA promoter activities under the aerobic condition at 3, 5, 7 and 24 h after incubation. In summary, promoter activities of fdhF and hyf were under the control of FhlA in response to formate and anaerobiosis.
Urease activity of P. mirabilis is pH-dependent and loss of hyf or fhlA gene affects urease activity and urinary stone formation. Urease can catalyze the hydrolysis of urea into ammonia, rising pH value and promoting urinary stone formation 27,28 . The urease activity changes with different pH value in Helicobacter pylori, being activated by acidic pH down to pH 2.5 and 3 29 . To understand whether urease activity of P. mirabilis is affected by pH value, we monitored the urease activity in LB broth at different pH values and found that urease activity was very low at pH 4, increased gradually in the pH range from 5 to 9 and dropped suddenly at pH 10 ( Fig. 4a). Knowing fhlA and hyfG contributed to acid resistance, we assessed the urease activity in wild-type strain, hyfG and fhlA mutants and respective complemented strains in the pH range from 5 to 9. The data showed that urease activity of hyfG and fhlA mutants were lower than the wild-type and respective complemented strains in the pH range from 5 to 8 (Fig. 4b). Likewise, fhlAcSD strain and rpoN mutant had reduced urease activity relative to the wild-type bacteria in this pH range (Fig. 4b).
The rpoN mutant could restore the activity to a certain level from pH 5 to 9 after rpoN-complementation (Fig. 4b). Interestingly, we found the urease activity of hyfG, fhlA and fhlAcSD mutants was not different from that of the wild-type strain at pH 9 while urease activity of rpoN mutant still decreased relative to the wild-type bacteria (Fig. 4b). The results show that both urease activity and the involvement of fhlA and hyfG in the urease activity is pH-dependent in P. mirabilis. Subsequently, we assessed the urinary stone formation using the same synthetic urine at pH 5.8. We found that the ability of stone formation in fhlA and hyfG mutants was significantly lower than that of wild-type strain and respective complemented strains (Fig. 4c). The ability to form urinary stones of fhlAcSD mutant significantly decreased relative to the wild-type, in contrast to that of rpoN mutant being comparable to the wild-type strain.
To differentiate between direct proton consumption and modulation of urease activity for the role of hyf and fhlA in medium deacidification, we included the ureC (encoding urease subunit) mutant in deacidification assay in synthetic urine. No medium deacidification but slight acidification by the ureC mutant was observed (Fig. 3a). In addition, using synthetic urine medium in which urea has been replace by an equivalent amount of NH 4 Cl, the medium deacidification did not occur in wild-type and mutants of hyfG, fhlA and ureC (Fig. 3b). Slight acidification was observed for wild-type and the mutants. The data suggest that urea hydrolyzed by urease is an essential process to deacidify synthetic urine medium for P. mirabilis and the role of hyfG and fhlA in medium deacidification is mainly dependent on modulation of urease activity. These results suggest that fhlA-regulated hyf expression could help P. mirabilis to deacidify the environment via facilitating urease activity and subsequent urinary stone formation.

fhlA-regulated hyf expression assists bacterial colonization in the mouse urinary tract.
To confirm the finding that FHL and FhlA were responsible for mouse colonization in P. mirabilis 21 , we investigated if fhlA and hyfG are associated with UTIs caused by P. mirabilis. The colonization ability was assessed in wildtype strain and mutants of hyfG and fhlA using the UTI mouse model 30 . In the bladder, the colonization ability www.nature.com/scientificreports/ of fhlA and hyfG mutant were significantly lower than wild-type strain (Fig. 5a). Both fhlA and hyfG mutants exhibited a significantly low ability of colonization in the kidney relative to the wild-type strain (Fig. 5b).
Loss of fhlA or hyf does not affect proton motive force-related motility, drug susceptibility and growth under carbonyl cyanide m-chlorophenyl hydrazone (CCCP). Previous studies showed that Hyf complex is similar to respiratory NADH dehydrogenase complex I 11,13,14 and the complex I was shown to provide the proton motive force (PMF) 31 , which could affect acid resistance, susceptibilities of polymyxin B and aminoglycosides, motility and growth under treatment of the PMF uncoupler [32][33][34][35][36][37] . We monitored the PMF- The urease activity of wild-type bacteria regrown in LB for 5 h was determined by phenol-hypochlorite assay at pH 5-9. The relative urease activity is calculated with the following formula: OD 625 of test/OD 625 of wt at pH 7. Significant differences from the result at pH 7 were determined by using the Student's t test (*P < 0.05). (b) Urease activities of the wild-type, mutants (fhlA, hyfG and rpoN), complemented strains (fhlAc, hyfGc and rpoNc) and fhlAcSD strain regrown in LB for 5 h were measured at different pH values. The relative urease activity is calculated with the following formula: OD 625 of test/OD 625 of wt at pH 7. Significant differences from the wild-type bacteria at each pH were determined by using two-way ANOVA with Tukey's multiple-comparison test (*P < 0.05; **P < 0.01). (c) Stone formation of the wild-type, mutants (fhlA, hyfG and rpoN), complemented strains (fhlAc, hyfGc and rpoNc) and fhlAcSD strain. Bacteria were regrown and adjusted to 2 × 10 8 CFU/ml with synthetic urine and incubated for 1 h at 37 ℃. The optical density at 630 nm of suspensions was measured as the intensity of stone formation. The relative urinary stone formation is calculated with the following formula: OD 630 of test/ OD 630 of wt. Significant differences from the wild-type bacteria were determined by using one-way ANOVA with Tukey's multiple-comparison test (*P < 0.05; **P < 0.01 www.nature.com/scientificreports/ related phenotypes including drug susceptibilities and motility in the wild-type bacteria and fhlA and hyfG mutants. Neither the MICs of the drugs (polymyxin B, streptomycin and gentamicin) used nor the swarming and swimming motility was different between wild-type strain and respective mutants (Table 3; Fig. 6a,b). We also examined the effect of CCCP, a PMF uncoupler, on growth of the wild-type, hyfG mutant and fhlA mutant. The growth curves of all strains showed no difference in the presence or the absence of CCCP (Fig. 6c). These results indicate that mutation of fhlA or hyfG did not affect the PMF-related phenotypes examined except acid resistance. It implies that the consumption of protons may be the main reason for FhlA-regulated hyf expression in acid resistance.

Discussion
To our knowledge, this study is the first to demonstrate that RpoN-dependent FhlA and hydrogenase Hyf participate in acid tolerance by medium deacidification and facilitate urease activity, urinary stone formation and mouse colonization. A typical characteristic of P. mirabilis UTIs is the development of urinary stones. P. mirabilis would rapidly invade bladder epithelium cells and start to form extracellular clusters within the bladder lumen adjacent to the urothelium 38,39 . The concentrated urease activity within extracellular clusters would cause mineral deposition that serves as an early step in urinary stone formation 38 . Urease decomposes urea to produce Horizontal bars indicate the average for each group, and the limit of detection was 100 CFU/g organ. Significant differences were determined using the Wilcoxon rank sum test (*P < 0.05; **P < 0.01). wt, wild-type; hyfG, hyfG mutant; fhlA, fhlA mutant.  40 and facilitating the formation of infectious urinary stones, carbonate apatite and struvite 38 . P. mirabilis extracellular clusters draw a massive neutrophil response and provide the environment to form biofilms 39 . The formation of extracellular clusters is critical to prevent clearance of bacteria by neutrophils and would influence the severity of infection 38,39 . In addition, medium deacidification has been shown to be involved in acid resistance of Salmonella enterica serovar Typhimurium 41 . We found FhlA and HyfG not only participated in medium deacidification and acid resistance (Figs. 1b, 3) but also affected urease activity and urinary stone formation in P. mirabilis (Fig. 4b). We identified that urease activity was pH-dependent, increasing in the pH range from 5 to 9 (Fig. 4a) and was dependent on FhlA and Hyf in the pH range from 5 to 8 (Fig. 4b). We showed that FhlA and Hyf-mediated deacidification or urine alkalization assisted in urease activity, stone formation and mouse colonization (Figs. 3a, 4b, 21 . The published Proteus genomes including P. mirabilis HI4320, P. mirabilis BB2000, P. hauseri, P. penneri and P. vulgaris contain only a hydrogenase 4-like operon but not the hydrogenase 3-like operon. It is likely that hydrogenase 4-like operon is the functional counterpart of hydrogenase 3-like operon for encoding the protein complex joining with FDH-H to form FHL in Proteus species. The facts that PCR produced a product of 4228 bp but not the predicted 524 bp using P. mirabilis N2 genomic DNA as the template with primers annealing to the conserved region of hycD and hycE (data not shown) and P. mirabilis hyfG mutant displayed reduced survival against acid exposure (Fig. 1b) as the hyc mutant 18 reinforce the notion. Hyf belongs to group 4a hydrogenase and is highly conserved in gammaproteobacteria 13 . The key role of Hyf is to convert proton to hydrogen, namely deacidification 11,13 . The structure of Hyf is similar to respiratory complex I, indicating Hyf is probably involved in the formation of proton gradient and energy production 13,19 . We found hyfG and fhlA are required for the full extent of acid tolerance in P. mirabilis (Fig. 1b). The dye DiOC 2 commonly used to detect membrane proton gradient does not work in P. mirabilis N2 as is the case for P. mirabilis HI4320 33 . Instead, we determined other PMF-related phenotypes, motility and drug susceptibility, to know whether Hyf is involved in conservation of www.nature.com/scientificreports/ the proton gradient and PMF. No difference of swarming motility, swimming and drug susceptibility between wild-type strain and hyfG or fhlA mutant may indicate that the formation of PMF by FhlA-Hyf is negligible or compensable in these conditions. Although proton consumption should be the main reason for fhlA and hyfinvolved acid tolerance. Possibly, it cannot be ruled out that the extent of PMF provided by FhlA-regulated Hyf may contribute to acid resistance of P. mirabilis. Recent studies showed that FHL-2 (FDH+Hyf) might couple with ATPase to generate PMF 42,43 and formate affects ATPase activity and changes the number of thiol groups 43 . Hyf activity is ATP-dependent, while the formate can increase the F 0 F 1 -ATPase activity 42 . It is worth noting that the effect of formate on ATPase activity disappeared in hyf mutant or under the respiratory condition 44 . In addition, the formate dehydrogenase-H activity was dependent on Hyf and F 0 F 1 -ATPase 16 . These studies indicate FHL-2 may contribute to regulation of formate-associated ATPase under fermentative condition. In this regard, our finding that P. mirabilis Hyf did not play an important role in generating PMF under aerobic condition may attribute to coupling of ATPase with FHL-2 only under fermentative or formate-rich conditions. Thus, it could not be ruled out that Hyf is involved in PMF formation under oxygen-limited condition during infection and contribute to virulence. Unlike fhlA and hyfG mutants, the urease activity of rpoN mutant still significantly decreased compared to the wild-type strain at pH 9 (Fig. 4b). Apparently, RpoN would have other regulatory pathways to affect urease activity at pH 9. In P. mirabilis, urease subunits and accessory proteins encoded by ureDABCEFG operon has been shown to be regulated positively by UreR and negatively by H-NS in the transcription level 45 . Previous study showed that RpoN and its cognate enhancer binding protein NtrC regulated expression of urease gene operon indirectly through control nac, encoding a transcriptional factor Nac which could activate urease gene operon in K. pneumonia 46 . We will investigate if other enhancer binding proteins such as NtrC, QseF and PspF are involved in the pathways of RpoN-regulated urease expression. In addition, we will examine whether UreR and H-NS are subject to the control of RpoN.
We found urinary stone formation ability of rpoN mutant was not impaired while fhlA mutant (lacking the enhancer binding protein, FhlA) was. In this regard, it was established that amino acids have effects on stone formation. For example, serine and asparagine could serve as the catalysts of this process 47 . Addition of asparagine did increase stone formation of wild-type P. mirabilis (data not shown). In addition, expression of asparagine synthetase gene (asnA) was downregulated by RpoN in wild-type P. mirabilis (data not shown). It is likely that loss of rpoN increases the asparagine level, compensating for the effect of rpoN loss.
This study showed P. mirabilis fdhF and hyf but not fhlA were induced significantly by formate and anaerobiosis respectively (Fig. 2e-g). Previous reports showed formate and anaerobiosis could trigger transcription of FhlA-regulated hypABCDE-fhlA operon in E. coli 10,26 . Two reasons could account for the discrepancy of fhlA expression in response to formate. First, In E. coli, hypABCDE-fhlA constitute an operon subject to the control of FhlA, namely self-regulation of FhlA 10 . This is not the case for P. mirabilis, whose fhlA promoter is not dependent on RpoN and FhlA (Fig. 2a). Second, formate serves as a ligand to activate FhlA, thus leading to transcription of FhlA-dependent genes including hypABCDE-fhlA operon in E. coli 48 but not fhlA in P. mirabilis. As for anaerobic induction of fhlA, there are an FNR binding site in the promoter region and an OxyS-interacting site in the 5′UTR of fhlA in E. coli 26,49 . FNR is a global regulatory protein that regulates gene expression in response to oxygen deprivation in E. coli and it was shown that FNR acts as an anaerobic activator of hypABCDE-fhlA operon expression to control the expression of the hydrogenase maturation genes (Hyp) 26 and FhlA. OxyS, a small regulatory RNA (sRNA) which is induced in response to the oxygen and oxidative stress in E. coli, inhibits fhlA translation by pairing with a short sequence overlapping the Shine-Dalgarno sequence, thereby blocking ribosome binding and translation 49 . Neither corresponding OxyS was found after searching for P. mirabilis sRNAs in the Bacterial Small Regulatory RNA Database (BSRD, https ://kwanl ab.bio.cuhk.edu.hk/BSRD/), nor FNR binding site was present in the P. mirabilis fhlA promoter region analyzed by RSAT. Therefore, unlike E. coli fhlA which is regulated by OxyS sRNA and transcription factor FNR sensing oxygen availability, the P. mirabilis promoter of fhlA was unresponsive to anaerobiosis (Fig. 2e).
This study implies that the proton-consuming acid resistance mechanism such as glutamate-and argininedependent acid resistance 50 may also contribute to urease activity in the presence of glutamate or arginine. Our preliminary data showed urease activity of wild-type P. mirabilis was induced by arginine and glutamate. What is the biological significance of the proton-consuming acid resistance of FHL? The different metabolites or signals would trigger different acid resistance system. Under stationary phase or the presence of extracellular glutamate, glutamate-dependent acid resistance would be important to bacterial acid response 51 . Likewise, arginine-dependent acid resistance is triggered by extracellular arginine 50 . However, FHL is the amino acid-independent acid resistance induced by formate or anaerobiosis 50 . In this study, we found that fdhF, hyfG and fhlA of FHL genes of wild-type P. mirabilis were expressed when cultured in the aerobic condition (Fig. 2b-d) and the expression of fdhF and hyf could also be induced by formate or anaerobiosis (Fig. 2f,g).
In this study, we showed FhlA-dependent hyf expression accelerates deacidification to facilitate urease activity and infectious urinary stone formation of P. mirabilis (Fig. 7) and subsequently mouse colonization. This is the first report investigating how P. mirabilis FHL-associated genes are involved in virulence and linking the ability of acid resistance to urease activity of P. mirabilis. The formate-induced FHL pathway provides a perspective for development of new approaches to counteract P. mirabilis UTIs. The prohibition against food producing formate could be a preventive measure to combat P. mirabilis UTIs. For example, uptake of aspartame increases urine formate level 52 , which could trigger FHL system to facilitate urease activity and subsequently urinary stone formation and in vivo colonization. In addition, our findings support the notion that manipulation of the urine pHn (the pH above which calcium and magnesium phosphates come out of solution in urine) could form the basis of a strategy to prevent catheter encrustation in those with urinary tract colonization by urease-positive bacteria 53 10.0. pH was adjusted to 5.8 and urine was sterilized by passing through a 0.2 μm pore-size filter.
The anaerobiosis was established and maintained in the anaerobic chamber (Whitley A35 anaerobic workstation, 5% H 2 -5% CO 2 -90% N 2 ).  www.nature.com/scientificreports/ Acid resistance assay. Acid resistance assay was performed as described by Wu et al. 55 with some modifications. Overnight bacterial cultures were diluted 500-fold with fresh LB broth in centrifuge tubes and incubated at 37 °C with shaking at 220 rpm for 4 h. Cells were incubated in LB medium at pH 3.0 (± 0.1) for 2 h. Acid-treated cells and the untreated control (cells before the acid treatment) then were washed with phosphatebuffered saline (PBS), serially diluted in PBS and plated on LSW − agar plates to determine the CFU. Acid survival rate (expressed as percent) was calculated with the following formula: 100 × (CFU after acid treatment/CFU before acid treatment).
Swarming assay. The swarming assay was performed on LB agar (tryptone, 10 g/l; yeast extract, 5 g/l; NaCl, 0.5 g/l; agar, 1.5%, w/v) plates as described previously 30 . The swarming migration distance was monitored by following swarm fronts of the bacterial cells and recording progress every hour.
Swimming assay. The swimming assay was performed on LB agar (agar, 0.3%, w/v) plates as described previously 30 . The swimming migration distance was recorded at 18 h after inoculation.
MIC assay. MICs of antibiotics for wild-type and mutant strains were determined by the broth microdilution method according to the guidelines proposed by the Clinical and Laboratory Standards Institute.
Growth curve analysis. Bacteria were grown overnight at 37 ℃, the cultures were diluted to an initial optical density at 600 nm (OD 600 ) of 0.01 in LB with or without 7.5 μM CCCP and then the OD 600 values were measured at 1-h intervals up to 8 h with a spectrophotometer.
Urease activity (phenol-hypochlorite) assay. Phenol-hypochlorite assay is based on the detection of ammonia released during urea hydrolysis. Ammonia can react with phenol-hypochlorite at high pH to form blue indophenol 56 . Bacteria grown to log phase (5 h after subculture) were adjusted to 2 × 10 9 CFU using LB and incubated statically in LB containing urea at 0.5 M for 2 h at 37 ℃. Supernatants were collected (70 μl per well of 96-well microplates) for urease activity measurement after centrifugation at 12,000 rpm for 5 min. Urease activity was determined by measuring ammonia production after 2-h incubation in LB containing urea at 0.5 M.
Briefly, 70 μl of phenol reagent (2%(w/v) phenol in 75% ethanol) and 70 μl of alkali reagent (0.28 M sodium hydroxide and 100 mM sodium hypochlorite) were added to each well. After one hour, the absorbance at 625 nm was measured using a microplate reader (SpectraMax M2, USA).
Urinary stone formation assay. Urinary stone formation assay was performed as described previously 27,28 with some modifications. Wild-type and mutant strains were grown overnight in LB broth and diluted 100-fold in the same medium. After incubation for 5 h, bacteria were adjusted to 2 × 10 8 CFU/ml with synthetic urine and incubated for another 1 h in the static condition at 37 ℃. The optical density at 630 nm of well-suspended suspensions was measured as the intensity of stone formation.
UTI mouse model. The mouse model of UTIs was used as described previously 57 . Six-to eight-week old female ICR mice were injected transurethrally with overnight cultures of bacteria at a dose of 1.5 × 10 7 CFU per mouse. On day 3 after injection, mice were sacrificed and bladder and kidney samples were collected to determine the viable bacterial count. All animal experiments were performed in strict accordance to the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Laboratory Animal Center (Taiwan), and the protocol was approved by the Institutional Animal Care and Use Committee of National Taiwan University College of Medicine.