MrpJ Directly Regulates Proteus mirabilis Virulence Factors, Including Fimbriae and Type VI Secretion, during Urinary Tract Infection

Proteus mirabilis is a leading cause of catheter-associated urinary tract infections (CAUTIs) and urolithiasis. The transcriptional regulator MrpJ inversely modulates two critical aspects of P. mirabilis UTI progression: fimbria-mediated attachment and flagellum-mediated motility.

ABSTRACT

Proteus mirabilis is a leading cause of catheter-associated urinary tract infections (CAUTIs) and urolithiasis. The transcriptional regulator MrpJ inversely modulates two critical aspects of P. mirabilis UTI progression: fimbria-mediated attachment and flagellum-mediated motility. Transcriptome data indicated a network of virulence-associated genes under MrpJ's control. Here, we identify the direct gene regulon of MrpJ and its contribution to P. mirabilis pathogenesis, leading to the discovery of novel virulence targets. Chromatin immunoprecipitation followed by high-throughput sequencing (ChIP-seq) was used for the first time in a CAUTI pathogen to probe for in vivo direct targets of MrpJ. Selected MrpJ-regulated genes were mutated and assessed for their contribution to UTI using a mouse model. ChIP-seq revealed a palindromic MrpJ binding sequence and 78 MrpJ-bound regions, including binding sites upstream of genes involved in motility, fimbriae, and a type VI secretion system (T6SS). A combinatorial mutation approach established the contribution of three fimbriae (fim8A, fim14A, and pmpA) to UTI and a new pathogenic role for the T6SS in UTI progression. In conclusion, this study (i) establishes the direct gene regulon and an MrpJ consensus binding site and (ii) led to the discovery of new virulence genes in P. mirabilis UTI, which could be targeted for therapeutic intervention of CAUTI.

INTRODUCTION

Proteus mirabilis is a major cause of catheter-associated urinary tract infections (CAUTIs), a significant health care-associated public health burden worldwide. Acute P. mirabilis infections lead to cystitis, pyelonephritis, urolithiasis, and often bacteremia (1).

Two critical aspects of P. mirabilis pathogenesis are fimbria-mediated attachment to the urinary tract and flagella-mediated motility (2, 3). The P. mirabilis HI4320 genome contains 17 chaperone-usher fimbrial operons, the highest number found thus far in a single genome of any bacterial species (4). The majority of these fimbriae are conserved in clinical isolates (5), although the function of most has not yet been determined. The mannose-resistant Proteus-like (MR/P) fimbria is the best studied and potentiates bladder colonization, localization, urolithiasis, and cluster formation in a murine model of UTI (6, 7).

P. mirabilis modulates the inverse regulation of fimbrial adherence and flagellar motility using transcriptional regulators encoded within the majority of its fimbrial operons (2). MrpJ, a helix-turn-helix (HTH) domain-containing protein encoded at the end of the mrp operon, temporally regulates these two virulence processes during UTI. Transcribed at low levels in vitro, mrpJ is among the most upregulated genes during experimental UTI (8). MrpJ is encoded by the overwhelming majority of P. mirabilis isolates, including those with clinical origins (9). Homologs are encoded by many other Gram-negative bacteria and, outside Proteus spp., are most frequently found in the related Providencia and Morganella genera (10). However, MrpJ homologs have generally not been studied outside P. mirabilis (2).

Previously, we used a microarray-based approach to learn that MrpJ directly or indirectly modulates transcription of both known and putative virulence factors, including other fimbriae and a type VI secretion system (T6SS) (9). Using chromatin immunoprecipitation followed by high-throughput sequencing (ChIP-seq), a powerful tool to define the direct gene networks of transcriptional regulators by probing their binding sites on a genomic scale (11, 12), we have now identified direct gene targets of MrpJ and their contributions to P. mirabilis pathogenesis. By identifying an MrpJ consensus binding site, we found that MrpJ directly targets genes from multiple functional classes. We also uncovered pathogenic roles for the T6SS and MrpJ-regulated fimbriae. Dissecting the MrpJ regulatory network may therefore lead to new targets for therapy or prevention for CAUTI patients.

RESULTS

Establishment of physiologically relevant mrpJ expression parameters during in vitro culture.

Previous transcriptional profiling of MrpJ indicated a multifaceted role for this regulator (9). However, differences in transcription observed in that study could be due to either direct promoter binding by MrpJ or indirect effects by MrpJ modulation of other regulators. Here, we used ChIP-seq to identify genes directly regulated by MrpJ. To this end, duplicate experiments were done using either anti-His antibody or a no-antibody control for P. mirabilis HI4320 cells expressing His-tagged MrpJ at levels comparable to those produced during UTI (in vivo mimic) (Fig. 1A) (8). This tagged version of MrpJ was selected to overcome the limitation of low expression of mrpJ outside the host. A vector control was also used for immunoprecipitation with or without anti-His antibody. Because MrpJ regulates a switch between a flagellum-mediated motile state and a fimbria-mediated adherent state, we selected a fimbria-inducing early-stationary-phase culture (optical density at 600 nm [OD₆₀₀] ≈ 2.0) to screen the genome-wide MrpJ ChIP-interactome. Expression levels of MrpJ in the in vivo mimic were comparable between mid-logarithmic phase (used for the previous transcriptome analysis) and early-stationary-phase cultures (Fig. 1A). Immunoblot analysis confirmed MrpJ protein expression (see Fig. S1A in the supplemental material). We used quantitative reverse transcription-PCR (qRT-PCR) to verify transcriptional modulation of candidate target genes that are positively (mrpA and tssA) or negatively (fim8A and flhC) regulated (Fig. 1B). Likewise, MrpJ repressed swarm motility compared to a vector control, a hallmark of this regulator (see Fig. S1B in the supplemental material). The slight difference in motility reduction by His-tagged MrpJ compared to a non-epitope-tagged version is consistent with our previous observation (9).

FIG 1.

Expression and functionality of pMrpJ-His₆. (A) mrpJ expression level as determined by qRT-PCR. The graph shows the mrpJ expression fold change compared to a vector control (EV) and rpoA. The black bar represents the expression of mrpJ in mid-log stage, and the gray bar represents expression at the early stationary stage (OD₆₀₀ = 2.0), used for ChIP-seq analysis. (B) qRT-PCR data for candidate MrpJ target genes. Bars represent the fold change compared to EV and rpoA. The data represent the means of two independent experiments; error bars represent the standard errors of the mean.

Determination of MrpJ core gene regulon by ChIP-seq.

Using MrpJ-DNA complexes generated during this culture condition, we identified a total of 78 enriched peaks by ChIP-seq (Table 1; predicted functions of these genes are listed in Table S1 in the supplemental material). MEME software (13) identified an enriched MrpJ binding motif within 68 of these sequences. Table 1 details the gene annotation and genomic coordinates for these motif centers relative to the HI4320 genome, including the “Fold Above Threshold” (FAT) scores, which indicate relative enrichment of regions in the ChIP-seq data, and the chromosomal strand designation (sense or antisense) for each motif. We identified a strongly enriched MrpJ binding motif within the bound regions (Fig. 2A; MEME E value = 1.4e⁻⁵⁸). This motif harbors end palindromes (ACnC), known to be a common characteristic of HTH-family transcriptional regulators (14). Figure 2B shows the nonrandom relative distribution of MrpJ-binding motifs relative to the ChIP-enriched peak center; as expected for a genuine MrpJ binding site, the motifs were more likely to be at the center of the query sequence (P = 1.6e⁻⁷). A total of 50 ChIP-enriched peaks were found to be upstream of genes (within 1 kb of the translational start site) and therefore likely promoters (Table 1). An additional site, 1,945 nucleotides (nt) upstream of flhD, is noted in Table 1. Even though this site is relatively distant from the flhD coding sequence, it is within a very large (2,758 nt in HI4320) intergenic region that is typically found upstream of flhD in P. mirabilis genomes.

TABLE 1.

ChIP-enriched MrpJ gene targets

Locus	FATa	Motif strand	Binding motif	Upstream of gene 1 or gene 2 (distance)b:		Inside gene
				Gene 1	Gene 2
41623	1	+	ACCCGCACACTGAGTG	NAc	NA	PMI0023
146908	1	–	ACCCGCTTCACGCGTA	NA	NA	PMI0117; lon
160359	3	–	ACCCATATTGAGAGTA	PMI0128; tesB (543)	NA	NA
178198	1	+	ACTCACTTAATGAGTG	PMI0143 (17)	PMI0144 (246)	NA
304748	4	+	CGTCATTTTATGAGTA	NA	NA	PMI0255; mrpC′
368496	1	+	ACAATTATTAGGAGAG	NA	NA	PMI0311
420952	1	–	ACCCGCTAAATGGGTG	PMI0371; aroL (23)	NA	NA
446530	2	+	ATTTCATTTACGAGTA	PMI0390; pheA (355)	NA	PMI0391
446662	2	+	ATCCTTAGTGAGAATA	PMI0390; pheA (487)	NA	PMI0391
545191	2	+	AAGCCCAGCATGGGTG	PMI0493 (96)	NA	NA
556934	2	NA	NA	NA	NA	PMI0511
748440	1	+	ACCCGCTTTCAGTGTG	PMI0689; clpS (398)	NA	PMI0688; cspD
791400	6	–	ATGCGATTAGTGGGTA	PMI0720 (410)	NA	NA
805494	1	+	ACCCCCATTATGAGTA	NA	NA	PMI0730
830177	1	+	ACTCTAATAATGAGTA	PMI0749; tssA, or vipA (364)	NA	NA
830376	1	–	ACAAATATAATGAGTA	PMI0749; tssA, or vipA (563)	NA	NA
831185	2	–	ACACTCAATATGCGTA	PMI0750; hcp, or idrA (525)	NA	NA
831489	6	+	ACACTTTTTATGACTG	PMI0750; hcp, or idrA (221)	NA	NA
874317	1	NA	NA	NA	NA	PMI0785; ompA
893397	1	NA	NA	PMI0807; cspB (149)	NA	NA
915441	1	–	ACTCAAATTAAGACTA	PMI0832 (204)	PMI0833 (143)	NA
1063935	1	–	ATACAAAGTATGACTA	NA	NA	PMI0999
1097492	1	+	ACCTGCAGGAGGAGTA	NA	NA	PMI1033
1120654	1	–	ACGCTATTAATGGGTA	NA	NA	PMI1055
1143561	2	+	ACTCATTAAATGAGTG	PMI1075 (336)	NA	NA
1169104	3	NA	NA	PMI1102 (388)	NA	NA
1169220	1	–	AGACCTAAAATGAGTA	PMI1102 (272)	NA	NA
1169425	1	–	ACTCGTTAAATGATTA	PMI1102 (67)	NA	NA
1185445	2	–	ACACTCAATATGCGTA	PMI1117; hcp3 (534)	NA	NA
1185756	6	+	ACACTTTTTATGACTG	PMI1117; hcp3 (223)	NA	NA
1220355	1	+	GCGCGATGAAAGAATA	PMI1158; mdtJ (277)	NA	NA
1288340	1	NA	NA	NA	NA	PMI1219
1319274	1	–	ACAACCTACGCGTGTA	PMI1247; metQ (210)	NA	NA
1361366	1	+	ACGCATTTTAGGAGTA	PMI1286; osmB (83)	NA	NA
1430072	3	–	ACTTATTACAGGAATG	PMI1350; ompW (44)	PMI1351 (400)	NA
1469620	1	+	ACAATACTGATGAGTA	PMI1391 (87)	PMI1392; gloA (94)	NA
1488618	1	+	TGTCATATAAAGAGTA	PMI1409; lpp (122)	NA	NA
1496694	5	–	ACTCACAATAAGAGTG	NA	NA	PMI1417
1554041	3	+	ACACTAAATAAGAGTA	PMI1470; fim8J (122)	NA	NA
1743057	1	+	ACACGCTTACCGAGTA	PMI1629; fliE (61)	PMI1630; fliF (195)	NA
1787097	1	+	ACTCAAATAAAGAGTA	PMI1672; flhD (1945)	NA	NA
1790075	3	NA	NA	PMI1676; cspC (23)	PMI1677 (433)	NA
1790175	3	–	ACCTGATTTATGATTG	PMI1676; cspC (123)	PMI1677 (333)	NA
1812688	1	–	ACTCAAAATATGAGTA	PMI1695 (256)	PMI1696; rppB (728)	NA
1847188	2	–	TATCGTATTCTGAGTA	PMI1730; rcsB (527)	NA	PMI1729; rcsD, or rsbA
1847325	1	–	ATTCTCATTACGAATA	PMI1730; rcsB (390)	NA	PMI1729; rcsD, or rsbA
2024237	5	+	ACTCACAGAATGAGTA	PMI1877; pmfA (591)	NA	NA
2024574	4	+	ACCCGTATTAGGAGTA	PMI1877; pmfA (254)	NA	NA
2044431	1	+	ACCTCCAATGTGGGCA	PMI1898; grcA (21)	PMI1899; ung (314)	NA
2048671	1	+	ACACAAGTTATGTGTG	NA	NA	PMI1902; recN
2250234	4	NA	NA	PMI2084; leuA (162)	NA	NA
2283778	1	–	ACTCTCATTAGGGGTA	NA	NA	PMI2112; hcp
2413314	1	–	ACACGCAGTGCGTGTA	PMI2224; pmpJ (280)	NA	NA
2417633	2	NA	NA	PMI2227 (81)	PMI2228 (342)	NA
2486762	1	–	ACTCACACGGCGAGTA	NA	NA	PMI2287; glnD
2575961	2	+	ACTCCCTTTATGACTG	NA	NA	PMI2357
2900642	1	–	ACACAGGCTACGGGTA	PMI2651; fixC (516)	NA	PMI2652; fixB
2951826	1	NA	NA	PMI2699; mxiH (40)	NA	NA
2978366	2	+	ACTCCCAATACGAGTA	PMI2727 (143)	PMI2728; atfA (727)	NA
3278698	6	+	ACACTTTTTATGATTA	PMI2990; idsA (232)	NA	NA
3303639	2	NA	NA	NA	NA	PMI3009; uspA
3305922	2	+	ACGCATAGCACGAGTA	NA	NA	PMI3011; pitA
3398605	1	–	ACTCCTTTTAAGAGTA	PMI3095 (292)	NA	NA
3460037	2	+	ACCCTTTAAATGAGTA	NA	NA	PMI3142
3477985	3	+	ACCCGCAGCACGAGTG	NA	NA	PMI3165; coaD
3514609	3	+	AACCGAATGAGGAGTG	PMI3199; cpxR (157)	PMI3200; cpxP (27)	NA
3514803	2	+	GGCCATATGATGGGTA	NA	NA	PMI3200; cpxP
3582302	1	–	TCCCCAATCGGGAGTA	PMI3254; rpsJ (86)	NA	NA
3636105	1	+	ACGCGCGTGGCGAGTC	NA	NA	PMI3316; wecA
3677592	2	–	ACTCAATAAATGAGTA	PMI3353 (730)	NA	NA
3685797	1	–	ACTCGCTTTAAGTGTA	PMI3359 (353)	NA	NA
3730038	1	–	ACCCAGATAACGAGTG	NA	NA	PMI3402
3735675	1	+	GCCCTCATTAAGGCTG	PMI3408; dacB (307)	PMI3409; greA (107)	NA
3754708	1	+	ACTCGCATTATGAATA	NA	NA	PMI3424; deaD
3780491	1	+	AGACTAATTGTGGATA	PMI3451; nrdD (46)	NA	NA
3796948	5	–	ACACGTAGAGTGAGTA	PMI3467; ampH (584)	NA	NA
3977658	1	–	ACCTGTTAAACGAGTA	NA	NA	PMI3637
3994945	2	+	ACACACCTGAAGAGTG	NA	NA	PMI3660

^aFAT, fold above threshold score, a measure of relative enrichment.

^bThe distance, in nucleotides, from the predicted translational start codon is indicated in parentheses.

^cNA, not applicable.

FIG 2.

Binding motif analysis and functional categories of genes. (A) MrpJ binding motif generated using MEME. The relative frequency of each base at different positions in the putative consensus motif is represented by the height of each letter, and the maximum score of 2 bits corresponds to 100% nucleotide conservation at that position. (B) Distribution of motifs relative to the ChIP-seq peak center. The y axis indicates the fraction of the 68 motifs identified by MEME. The x axis indicates the distance of the motif center from the corresponding ChIP-seq peak center. (C) Functional categories of genes associated with MrpJ binding. Some genes were categorized in more than one class, and therefore the total number adds up to more than the number of enriched MrpJ-binding sequences (n = 78). KEGG annotation was used to classify genes (15).

MrpJ directly regulates a diverse set of genes.

A number of different cellular processes were found to be directly regulated in the MrpJ ChIP-interactome (Fig. 2C). ChIP-enriched MrpJ binding sequences (Table 1) were functionally categorized according to their annotation in KEGG (15). The diversity of functional classes in the MrpJ direct regulon mirrors the previously reported MrpJ-regulated genes identified by microarray analysis (9). Of the 78 MrpJ-binding targets, 27 can be unambiguously assigned as upstream of a single gene (Table 1). Of the 27 genes, 13 are >1.5-fold differentially expressed during MrpJ overexpression (compared to all genes, P = 0.0029 [Fisher exact test, one tailed]), strongly supporting the premise that genes with upstream MrpJ binding sites are often direct regulatory targets of MrpJ. Genes identified by ChIP-seq that had the topmost regulatory changes in the prior transcriptome analysis were the flagellar genes PMI1629 fliE (−261.8-fold), PMI1630 fliF (−286.3-fold), and PMI1672 flhD (−5.16-fold); the T6SS genes PMI0749 tssA (also called vipA) (12.8-fold), PMI0750 hcp (named idrA in P. mirabilis BB2000; 2.23-fold), and PMI2990 idsA (2.15-fold); and the fimbrial genes PMI1470 fim8J (−37.36-fold), PMI1877 pmfA (−3.1-fold), and PMI2728 atfA (8.3-fold). Also noteworthy is PMI0833, which encodes a putative exported protein and is repressed 95-fold by MrpJ.

MrpJ was originally characterized as a repressor of flagellum-mediated motility (2, 16), and as expected, we detected interactions with the flagellar master regulatory operon flhDC. However, the MrpJ binding site identified by ChIP-seq was considerably further upstream of the site that was previously identified by gel shift and ChIP-PCR analyses (2, 9). In addition, MrpJ bound the promoter region of the divergently transcribed class 2 flagellar operons fliE and fliFGHIJK (17). Furthermore, MrpJ bound the promoters or coding sequences of other genes known to regulate swarming differentiation (e.g., rppB, rsbA, rcsB, and lon) (18–21). Thus, MrpJ has multiple points of control over motility and swarming.

We found that multiple genes encoding the T6SS are direct targets of MrpJ (tssA, idsA, hcp [also called idrA], and hcp3) (Table 1), consistent with our previous finding that T6SS genes are transcriptionally induced by MrpJ (9). Notably, we found four enriched sequences containing the consensus MrpJ-binding motif in between the divergently transcribed T6SS genes PMI0749 (tssA) and PMI0750 (hcp). We also detected MrpJ targeting upstream of T6SS gene PMI1117 (hcp3), which is highly similar to PMI0750 and has MrpJ binding sites identical to those found upstream of hcp. Thus, MrpJ directly regulates the T6SS structural operon (tss, also called vip) as well as three of the five predicted T6SS effector operons in P. mirabilis HI4320 (i.e., idsA, hcp, and hcp3) (22).

A number of fimbriae emerge as direct targets of MrpJ: P. mirabilis fimbria (PMF), P. mirabilis P-like pili (PMP), ambient-temperature fimbria (ATF), and fimbria 8 (Fim8). MrpJ consensus binding sites were found preceding the starting genes of these operons, which for some operons is the major structural component gene (atfA and pmfA) and for others is a regulatory mrpJ paralog gene (fim8J and pmpJ). We found one consensus binding site on the atfA, fim8J, and pmpJ promoters, while the pmfA promoter harbors two strong MrpJ binding motifs (Table 1). This finding provides a mechanistic explanation for the previous observation of an altered fimbrial profile on the P. mirabilis cell surface when mrpJ was expressed in trans (2). Likewise, we previously observed MrpJ-mediated transcriptional activation (e.g., atfA and mrpA) and repression (e.g., fim8A and pmfA) of fimbrial genes (9). Thus, MrpJ directly regulates multiple fimbrial targets and acts as both a transcriptional activator and a repressor. Notably, the atfA site is located within a 100-nt region that was previously shown to interact with AtfJ (23), suggesting that MrpJ paralogs bind similar consensus sequences.

In agreement with a prior microarray study (9), other direct targets of MrpJ included genes involved in metabolism, signaling and transport, membrane proteins, and lipoproteins (Table 1 and Fig. 2C). MrpJ binding enrichment was found in promoters or coding sequences of genes encoding major outer membrane proteins such as ompA, ompW, and lpp. Another target was cold shock proteins, with MrpJ binding detected in the promoters or coding sequences of cspB, cspC, cspD, and deaD and the putative cold shock protein PMI0720. Although CspA-family proteins were originally identified because they were induced at low temperature, these RNA-binding proteins are now known to function in a variety of conditions, including nutrient stress, osmotic stress, oxidative stress, cell division, and virulence (24–26). Even though metabolic genes comprised 14 possible MrpJ targets (Fig. 2C), no single metabolic pathway stood out. Three of these genes—hcp hydroxylamine reductase, glnD (protein-PII) uridylyltransferase, and PMI3637, a putative carbon-nitrogen hydrolase—are part of nitrogen sensing or metabolism. Two genes, aroL and pheA, contribute to aromatic amino acid biosynthesis; two others, tesB and PMI0999 triacylglycerol lipase, participate in lipid metabolism. Annotations for all of the genes listed in Table 1 are listed in Table S1 in the supplemental material.

Confirmation of MrpJ binding on candidate promoters.

We used ChIP-qPCR to confirm MrpJ binding on selected promoters (Fig. 3A to C). This method of independent ChIP experiments, followed by targeted qPCR, allows analysis of interactions with the chromosome in its native conformation. Consistent with the ChIP-seq data, we found MrpJ-His₆ ChIP enrichment relative to the vector (EV) and no-Ab control on T6SS (Fig. 3A) and fimbrial (Fig. 3B) targets of MrpJ, but not for non-MrpJ-interacting gene rpoA. Individual experimental data are included in the supplemental material (see Fig. S2A and B).

FIG 3.

Confirmation of MrpJ binding on candidate genes. (A to C) PCR on candidate gene promoters was used to measure MrpJ-His₆ ChIP enrichment relative to the vector (EV) and no-antibody control. (A) T6SS gene promoters. (B) Fimbrial gene promoters. (C) Motility and mrp fimbrial gene promoters. The graph represents the mean of two independent experiments; error bars represent standard deviations. rpoA is a non-MrpJ-interacting negative-control gene. ■, anti-His IP of EV; ▧, anti-His IP of MrpJ-His₆. (D) In vitro MrpJ-His₆ binding of MrpJ consensus site. A 29-mer double-stranded DNA probe containing the MrpJ consensus binding site from the tssA promoter was mixed with MrpJ-His₆, and DNA-protein interactions were detected by slower migration in a native polyacrylamide gel. The shift was outcompeted using 200-fold excess unlabeled DNA (competitor). A solid arrow indicates unshifted DNA, and the bracket indicates shifted DNA-protein complexes.

We also tested qPCR enrichment of the mrpA fimbrial gene and fliE and fliF flagellar gene promoters (Fig. 3C) to confirm MrpJ binding. Surprisingly, a previously identified autoregulatory and ChIP-enriched region on the mrpA promoter (major structural gene of MR/P fimbriae) (9) was not identified by ChIP-seq; however, the mrpA promoter was again identified here as an MrpJ target by ChIP-qPCR using the same input DNA used for ChIP-seq. A possible explanation for the apparent lack of mrpA in the ChIP-seq data set is noise in the ChIP-seq data, which may obscure weakly bound sites; indeed, the ChIP-qPCR enrichment for the mrpA promoter was lower (9-fold) than ChIP-qPCR enrichment of MrpJ-binding targets detected by ChIP-seq (range, 13- to 61-fold; Fig. 3). Another possibility is that differences between ChIP-seq and the standard ChIP methods led to detection of overlapping but distinct targets (9, 27).

To biochemically investigate interaction of purified MrpJ-His₆ with a synthesized MrpJ consensus binding sequence, we used an electrophoretic mobility gel shift assay (EMSA). Complementary 29-mer oligonucleotides, designed to encompass the MrpJ consensus site centered on nt 830177 in the tssA promoter, were hybridized and labeled to facilitate detection. Mixture of this DNA sequence with increasing amounts of MrpJ-His₆ led to a shift in apparent size, and this shift was abrogated by addition of unlabeled probe, providing further evidence that ChIP-enriched sequences represent genuine MrpJ binding targets (Fig. 3D).

T6SS contributes to P. mirabilis pathogenic fitness.

MrpJ plays a critical role in pathogenesis since its deficiency significantly attenuates UTI progression in a murine model in both independent and cochallenge experiments (9, 16). Several MrpJ-regulated genes have previously been shown to contribute to virulence, including those encoding flagella, MR/P and PMF fimbriae, and Pta toxin (28–32). However, mutation of these individual MrpJ targets results in a relatively modest deficiency in mice compared to mutation of MrpJ itself, suggesting that MrpJ regulates additional virulence factors. Therefore, we investigated the role of novel MrpJ direct targets in the pathogenic fitness of P. mirabilis.

T6SS secretion apparatus genes are encoded in an operon (PMI0749-0734). Our ChIP-seq data revealed that MrpJ has binding sites 370 and 578 bp upstream of the predicted translational start site of tssA (PMI0749), at the beginning of this operon. We first investigated conservation of this gene, which encodes the sheath protein of the T6SS needle complex (22), among a collection of 48 P. mirabilis clinical isolates (5) using PCR. Among these isolates, 94% (45/48) were tssA positive. We found that tssA was conserved among isolates from both urine (93%, 37/40) and diverse anatomical sites (100%, 8/8). We also analyzed tssA conservation in 54 sequenced genomes of P. mirabilis using BLAST (see Table S2 in the supplemental material). Although these sequenced isolates of P. mirabilis were obtained from a variety of sources, their tssA gene was found to be ≥99% identical with HI4320.

Previous studies reported a role for P. mirabilis T6SS in strain recognition during swarming (22, 33). Because the P. mirabilis T6SS is active during swarming (when the mrp genes are not expressed) (22, 34), yet MrpJ induces transcription of T6SS genes (9), we hypothesized that T6SS also plays a swarm-independent role in P. mirabilis pathogenesis. Therefore, we tested whether a deficiency in tssA affects UTI progression using a mouse bladder cochallenge model. Pathogenic fitness of the tssA mutant was significantly compromised in the urine, kidneys, and spleen at 7 days postinfection (median outcompetition, 2.5- to 5-fold), indicating that the T6SS aids bacterial fitness during pyelonephritis and possibly systemic dissemination (Fig. 4B and C). The T6SS mutant was not attenuated in the bladder. The compromise in UTI fitness did not originate from a growth defect, since no notable difference during in vitro culture or coculture of wild-type and mutant strains was detected (Fig. 4A). Thus, our data suggest that the T6SS contributes to P. mirabilis UTI.

FIG 4.

The T6SS gene tssA (also called vipA) is required for pathogenic fitness of P. mirabilis. (A) The tssA mutant and wild-type parent have nearly identical growth kinetics in LB medium in vitro. (B) A tssA mutant has a competitive defect during experimental UTI. Mice were cochallenged (1:1) with tssA mutant and wild-type strains by transurethral catheterization. Bacteria were enumerated at 7 days postinfection. The bar represents the median CFU for each sample. The limit of detection is 200 CFU. (C) The competitive index (CI) was calculated for each cochallenge experiment. The bar represents the median CI for each sample. Statistical significance was determined using the Wilcoxon matched-pair signed-rank test against a hypothetical value of zero (indicated by a dashed line). Two independent experiments, with 10 mice in each, were conducted.

Novel MrpJ fimbrial targets contribute to virulence.

A number of P. mirabilis fimbriae were discovered as direct targets of MrpJ in this study. MrpJ bound sites upstream of pmfA, atfA, fim8J and pmpJ, the first genes of their respective operons (Table 1). Although not detected by ChIP-seq, fim14A is also regulated by MrpJ (9). While PMF fimbriae have been shown to contribute to UTI (30, 35), ATF fimbriae were not found to significantly contribute (36). The other MrpJ-regulated fimbriae have not been directly tested for their roles in UTI; however, both fim8A and fim14A transcripts have been detected in urine voided by infected mice (8). Also, a gene from the fim14 operon was identified in a signature-tagged mutagenesis study as a virulence candidate (37). Although pmpA is highly conserved in clinical strains of P. mirabilis (5) and PmpA has been detected in a canine urinary isolate (38), a mutant has not been assessed in a UTI model. Therefore, we mutated the major structural genes of these MrpJ-regulated fimbriae (fim8A, fim14A, and pmpA) to evaluate their contribution to P. mirabilis uropathogenesis.

We constructed seven P. mirabilis fimbrial mutants to test the roles of these three fimbriae: a pmpA fim8A fim14A triple mutant; pmpA fim8A, pmpA fim14A, and fim8A fim14A double mutants; and pmpA, fim8A, and fim14A single mutants (Fig. 5). To our knowledge, this is the first report of construction of a triple mutant in this uropathogen.

FIG 5.

P. mirabilis HI4320 and fimbrial mutant cochallenge in a mouse model of ascending UTI. Mice were cochallenged (1:1) with mutants of MrpJ fimbrial targets pmpA, fim8A, and fim14A (triple, double, and single) and wild type. At 7 days postinfection, CFU were measured and the CI was calculated for each cochallenge experiment. The bar represents the median CI for each sample. Statistical significance was determined for CI using the Wilcoxon matched-pair signed-rank test against a hypothetical value of zero (indicated by a dashed line). Two independent experiments, with 10 mice in each, were conducted for panels A to E; one experiment of 10 mice per group was conducted for panels F and G.

Mice were coinfected (1:1 ratio of inoculum) with the mutants and wild-type P. mirabilis, and the bacterial burden was evaluated at 7 days postinfection (see Fig. S3 in the supplemental material). The ratio of wild type to mutant was then calculated for each experiment (see Materials and Methods) (Fig. 5). The pmpA fim8A fim14A triple mutant showed a significant defect in all samples, indicating that one or more of these fimbriae contribute to UTI (median outcompetition, 10- to 34-fold; Fig. 5A). To tease apart the contribution of each fimbria, we tested each double-mutant combination (Fig. 5B to D). Both combinations lacking pmpA (pmpA fim8A and pmpA fim14A) were significantly impaired in both bladder and kidney colonization, implicating a requirement for PMP fimbriae during UTI. Indeed, the pmpA single mutant was attenuated in both organs, as well as the urine (Fig. 5E). However, the magnitude of the defect in the bladder for these pmpA mutants was less than that observed for the triple mutant (outcompetition range, 1.8- to 15-fold).

In mice cochallenged with wild-type and triple fimbrial mutant bacteria, the mutant was frequently unrecoverable from one or more organs, indicating that it had been completely outcompeted by wild type (Table 2 and see Fig. S3 in the supplemental material). In comparison, fewer mice challenged with single or double mutants cleared the mutant bacteria from any organ, particularly the urine or bladder. In a pooled statistical model comparing outcomes for each mutant, all single or double fimbrial mutants were more likely to have recoverable mutant CFU than the triple mutant (P < 0.01, logistic regression). Subsequently, the likelihood of clearance from each individual site (urine, bladder, kidneys, and spleen) was examined. The probability of complete clearance of the triple mutant from the bladder compared to any of the double mutants was significant (P < 0.05 for the pmpA fim8A mutant and P < 0.01 for the pmpA fim14A or fim8A fim14A mutant). Thus, even though the fim8A fim14A double mutant, as well as fim8A and fim14A single mutants, did not have an apparent fitness defect in this model (Fig. 5E, F, and G), the data suggest overlapping or redundant roles of Fim8, Fim14, and PMP fimbriae, especially in the bladder. In vitro analysis of fimbrial mutants in monoculture demonstrated no significant difference compared with wild type, suggesting that the colonization defects during UTI are not due to growth retardation (see Fig. S4A in the supplemental material). Furthermore, coculture competition and serial passaging over 72 h revealed no difference for in vitro fitness (see Fig. S4B to F in the supplemental material). Together, our data demonstrate that novel MrpJ fimbrial targets contribute to P. mirabilis pathogenesis.

TABLE 2.

Mice with unrecoverable mutant bacteria^a

Mutant	No. of mice with no mutant bacteria/total no. of mice examined (%)b
	Urine	Bladder	Kidneys	Spleen
pmpA fim8A fim14A	5/13 (38)	9/16 (56)	5/16 (31)	6/13 (46)
pmpA fim8A	1/11 (9)	2/14 (14)*	1/14 (7)	6/14 (43)
pmpA fim14A	3/15 (20)	2/18 (11)**	5/19 (26)	0/10 (0)
fim8A fim14A	0/15 (0)	1/17 (6)**	0/18 (0)	3/10 (30)
pmpA	0/19 (0)	2/19 (11)**	4/20 (20)	2/7 (29)
fim8A	0/4 (0)	0/6 (0)	0/6 (0)	2/6 (33)
fim14A	1/7 (14)	0/8 (0)	2/8 (25)	0/8 (0)

^aMice were not included if the mouse died before the experimental endpoint, both mutant and wild type had been cleared (no recoverable CFU), or, for urine counts, if no urine could be collected prior to sacrifice.

^bStatistical significance values were calculated using logistic regression compared to the triple mutant (pmpA fim8A fim14A). *, P < 0.05; **, P < 0.01.

DISCUSSION

The key to bacterial pathogenesis is optimal utilization of cellular and environmental resources toward establishing an infectious niche while surviving the host's immune response. Rapid changes in gene expression help achieve this goal by enabling quick adaptation, and this is where transcriptional regulators play the role of a master conductor. Transcriptional modulation could be local (specific to one set of genes or processes) or global (regulation of a variety of genes and processes simultaneously). Our recent transcriptome study revealed MrpJ as a global multifaceted regulator, making it the first fimbria-associated regulator of its kind to orchestrate a diverse virulence-associated gene network in a uropathogen (9). Here, we determined the core in vivo regulon of MrpJ and established the role of selected targets in P. mirabilis pathogenesis.

MrpJ and its regulon are targets for therapeutic intervention.

MrpJ orchestrates the regulation of multiple gene classes (Fig. 2C). We chose to focus on the major themes of fimbriae and T6SS because MrpJ directly binds multiple promoters of these genes, and this correlated well with our previous microarray study. However, there were additional MrpJ targets that may yield useful information about P. mirabilis adaptation to the urinary tract, motility, or swarming differentiation. For example, an MrpJ motif was identified within the gene encoding Lon protease, which has been previously linked to both motility and virulence gene expression (20) and contributes to infection in a CAUTI model (39). Another site was within a putative intimin/invasin gene (PMI0023). It is not yet clear how these intergenic MrpJ binding sites are significant. MrpJ binding was also found in the promoters for clpS and cspC. ClpS is an adapter protein that modulates protein degradation and has been tied to both virulence and swarming in Pseudomonas aeruginosa (40). Likewise, CspC has a role in motility and virulence in Salmonella enterica (41) and contributes to both bladder and kidney colonization in a CAUTI model (39).

Direct regulation of fimbrial genes is particularly important because fimbriae (42) allow the pathogen to adhere both to the urinary tract and to catheters (3, 7, 42–44). This makes fimbriae an attractive target for therapeutic interventions, such as vaccines and small molecule inhibitors (45, 46). Indeed, several moderately successful single target P. mirabilis vaccine studies have been conducted in mice using fimbrial proteins MrpA, UcaA, and PmfA (47–50).

P. mirabilis encodes 17 chaperone-usher fimbriae that are conserved among a wide variety of clinical isolates, and multiple types can be expressed simultaneously (4, 5). Fimbrial coexpression is a unique attribute of P. mirabilis in comparison with uropathogenic Escherichia coli and Salmonella enterica, two other pathogens also known to harbor multiple fimbriae (51–53). Unlike P. mirabilis, fimbrial regulation seems to be hierarchical in these two pathogens, where one fimbria predominates at any time. Thus, fimbrial biology in P. mirabilis is complex and involves cross talk that we have only just begun to appreciate (23).

Therefore, we focused on the role of novel MrpJ fimbrial targets in UTI pathogenesis. Here, we established the virulence role of PMP fimbriae in the urine, bladder, and kidneys (Fig. 5). We also found subtle contributions for Fim8 and Fim14 that are only apparent in a triple fimbrial mutant, establishing that these fimbriae may have overlapping binding targets or redundant contributions to infection (Fig. 5A). Similarly, mutation of atfA was previously found to have no effect on virulence (36), but ATF fimbriae are specifically detected in mice infected with mrpA mutant bacteria (54), suggesting that ATF fimbriae should be reinvestigated in the context of other fimbriae. Subtle or overlapping contributions to UTI have previously been noted for other fimbriae produced by P. mirabilis, specifically, unique contributions of MR/P and PMF fimbriae (55).

T6SS contributes to UTI pathogenesis.

The initial finding that MrpJ induces T6SS transcription, as well as the subsequent finding that MrpJ directly regulates T6SS genes, was surprising because the P. mirabilis T6SS had previously been exclusively associated with bacterial strain-strain competition during swarming (22, 56), a condition where the mrp operon is strongly repressed (34). MrpJ-mediated direct regulation of both structural components of the needle apparatus and secreted effector genes therefore points toward a swarm-independent role of this T6SS, and we found a fitness defect of a T6SS needle mutant during experimental UTI (Fig. 4C and D). T6SSs potentiate virulence in other pathogenic organisms, including Vibrio cholerae and Pseudomonas aeruginosa (22, 57). Recently, the P. mirabilis T6SS was found to contribute to infection in a transposon insertion-site sequencing (Tn-seq) screen using a polymicrobial CAUTI model (39). In that model, which investigated UTI progression in mice that had a catheter segment pushed into and remaining in place in the bladder for the duration of the experiment, the T6SS was found to strongly contribute to P. mirabilis fitness when Providencia stuartii was also present. Interestingly, T6SS was not identified as a contributor to single-species CAUTI using a Tn-seq approach. Further work is warranted to tease out T6SS contributions using both the traditional and the CAUTI models.

Currently, we do not know which, if any, of the multiple predicted T6SS-secreted effectors mediate virulence and whether these factors act directly on the host or instead attack competing bacterial invaders. We speculate that differences in MrpJ binding sites in different isolates could also translate to distinct phenotypes. For example, HI4320, the strain we used for ChIP-seq, and BB2000, a strain often used for motility studies, both encode a T6SS secretion apparatus operon and a divergently transcribed effector operon (22, 33). BB2000 retains all four MrpJ binding motifs found in HI4320; however, the two sites closer to the effector operon are located further away from the coding sequence (562 and 285 nt, compared to 518 and 240 nt in HI4320). Notably, the effector operons encoded by these two strains (pef in HI4320 and idr in BB2000) are completely different after the first two, conserved T6SS genes (50). Future efforts are warranted to determine the full list of T6SS effectors and their individual contributions to P. mirabilis virulence.

Noncanonical and intragenic MrpJ binding sites add complexity to its regulatory mechanism.

Advancements in genomics and bioinformatics in recent years have led to an increasing demonstration and deepening appreciation of intragenic or coding region binding by transcriptional regulators in bacteria. In many instances, these intragenic binding sites regulate an internal promoter, unidentified open reading frames, or noncoding regulatory RNAs, and they have been suggested to influence chromosomal conformation toward optimal regulation of transcription (11, 58–60).

Here, we identified 32 noncanonical enrichments within intragenic or annotated coding regions, suggesting (i) the influence of chromosomal conformation in MrpJ-mediated transcriptional regulation, (ii) alternative promoters within operons, or (iii) unidentified transcripts. Notably, MrpJ was found to bind in the coding region of rcsD (also called rsbA) (PMI1729). This motif center is situated 361 bp upstream of gene rcsB, which encodes a two-component signal transduction system response regulator known to play an important role in P. mirabilis swarmer cell differentiation and recently reported to regulate mrpA transcription (19). Although rcsDB is considered to be an operon, our data suggest that MrpJ specifically regulates rcsB via an mRNA that initiates within the upstream gene. This is consistent with our transcriptomic study, where we detected MrpJ-dependent regulation of rcsB but not rcsD transcripts (9).

MrpJ-mediated regulation of flhDC and mrp operons is complex.

MrpJ-dependent regulation of flagella appears to be accomplished through both direct and indirect mechanisms. In addition to binding the flagellar class 1 flhDC and class 2 fliE and fliF promoters, MrpJ directly regulates rcsB (−2.6-fold) and rppB (−1.9-fold), both of which encode parts of two-component systems that regulate swarming motility (18, 21, 61–63). Regulation of flagella and coordination of adherence, swimming, and swarming motility involve multiple interwoven inputs (6, 64). Work from several groups has demonstrated that motility regulators have overlapping contributions and frequently modulate each other in addition to directly controlling motility. Particularly notable to this study is that Rcs and Rpp, both regulated by MrpJ, also modulate flhDC. Likewise, at least one MrpJ regulatory target, Rcs, in turn transcriptionally represses mrpJ (19).

Other MrpJ-regulated motility regulators (e.g., umoA and umoB) (9, 65, 66) were not identified as direct MrpJ targets by ChIP-seq. This may be because of indirect regulation via, for example, FlhD₄C₂, but direct MrpJ binding of umo promoters cannot be ruled out without further investigation. While the focus of this study was MrpJ's contribution to UTI, a time when the mrp operon is highly expressed and motility appears to be repressed (8), other motility modulators likely dominate over MrpJ in suitable environments.

Two known MrpJ targets were not found in our ChIP-seq data set. First, MrpJ binds a 550-nt fragment of the flhDC promoter both in vitro (gel shift) and in vivo (ChIP-PCR) (2, 9). Yet, the ChIP-seq MrpJ-binding motif in the flhDC promoter is 1.7 kb from the transcriptional start. Second, MrpJ positively autoregulates the mrp fimbrial operon: expression of MrpJ in trans leads to increased mrpA transcript (9) and MrpA protein (16). We believe MrpJ directly regulates this operon because we have detected MrpJ-mrp promoter interactions using ChIP-PCR (9) (Fig. 3C). Furthermore, the MrpJ-interacting region is within an invertible element known to be crucial for mrp expression (9, 67, 68). Based on our previous work with AtfJ (23), it appears that MrpJ and its paralogs can act by binding sites far (>500 nt) from gene starts. Perhaps these regulators achieve control, at least in some instances, by looping together far-away sequences. Indeed, we found an MrpJ binding site in mrpC′, which is encoded in an operon adjacent to mrp (4).

An explanation of why we failed to detect expected binding events by ChIP-seq is that the previously identified binding targets might be relatively weak MrpJ targets. This is consistent with ChIP-qPCR measurement of MrpJ-mrp promoter interaction, where this promoter was enriched relative to nontarget DNA, but at a lesser fold increase than MrpJ targets identified by ChIP-seq (Fig. 3). Previously reported MrpJ-ChIP experiments used endpoint PCR to detect interactions with mrp and flhD promoters, and therefore, while useful for identifying MrpJ-binding DNA sequences, that method is not quantitative. During ChIP-seq analysis, variation in the control sample led to noise that would have obscured less intense signal peaks, such as any that may have been present for mrp. A corollary is that there may be additional MrpJ binding sites that we did not identify by ChIP-seq because they were lost in signal noise.

MrpJ paralogs add regulatory intricacy.

The P. mirabilis HI4320 genome carries 14 additional mrpJ paralogs besides mrpJ itself (2), and direct MrpJ regulation of several fimbriae is consistent with a proposed regulatory cascade of these paralogs (23). Because each of the MrpJ-targeted fimbrial operons carries an mrpJ paralog (i.e., pmpJ, atfJ, fim8J, and fim14J), there are multiple opportunities for cross talk and adaptation to specific environments. MrpJ has been postulated to be the dominant regulator of this fimbrial cascade during UTI (9).

It is conceivable that MrpJ paralogs function as heterodimers, adding further fine transcriptional control. One such candidate is AtfJ, whose atf fimbrial operon is under direct control of MrpJ-mediated positive transcriptional regulation, detected both by our ChIP-seq and by previous transcriptomic analyses (9). AtfJ has an autoregulatory effect and binds a distal region of its promoter in vitro (23). This AtfJ-binding region overlaps with the MrpJ consensus motif, suggesting that both of these paralogs bind the same DNA sequence. Potential synergism between MrpJ and AtfJ is yet to be explored.

Conclusion.

We present here the first report of ChIP-seq in a CAUTI pathogen coupled with the discovery of novel virulence factors in P. mirabilis UTI. We defined the direct gene regulon of MrpJ and established new roles for its fimbrial and T6SS targets in uropathogenesis. We found overlapping contributions of fimbriae in P. mirabilis UTI and established MrpJ as a dominant player of fimbrial cross talk in this pathogen. Future work remains to examine the mechanism of MrpJ transcriptional control, the pathogenic contribution of other MrpJ targets, and the therapeutic potential of novel virulence targets presented here.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

Bacterial strains and plasmids from this study are in Table 3. HI4320, a clinical isolate from a catheterized patient in a nursing home (69), was used as the wild-type strain for ChIP-seq and mutagenesis studies. All strains were cultured at 37°C in low-salt Luria broth (LB; per liter: 10 g tryptone, 5 g yeast extract, 0.5 g NaCl) or on LB medium solidified with 1.5% agar unless otherwise noted.

TABLE 3.

Bacterial strains and plasmids used in this study

Strain or plasmid	Descriptiona	Source or reference
Strains
HI4320	P. mirabilis strain collected from an elderly, long-term catheterized woman's urine	69
Mutants
tssA	T6SS mutant (PMI0749); Kanr	This study
pmpA fim8A fim14A	Triple fimbrial mutant; Kanr	This study
fim8A fim14A	Double fimbrial mutant; Kanr	This study
pmpA fim14A	Double fimbrial mutant; Kanr	This study
pmpA fim8A	Double fimbrial mutant; Kanr	This study
pmpA	Single fimbrial mutant; Kanr	This study
fim8A	Single fimbrial mutant; Kanr	This study
fim14A	Single fimbrial mutant; Kanr	This study
Plasmids
pLX3607	Ampr; IPTG-inducible cloning vector, empty vector (EV)	16
pLX3805	pLX3607 derivative containing mrpJ, pMrpJ	16
pNB010	pLX3805 derivative containing His-tagged mrpJ	9
pACD4K-CloxP	TargeTron vector with loxP sites flanking kanamycin cassette	Sigma
pACD4K-C	TargeTron vector without loxP sites	Sigma
pAR1219	pBR322-based T7 helper plasmid	Sigma
pQL123	Encodes IPTG-inducible cre recombinase	73
pJS044	pACD4K-CloxP with fim14A TargeTron fragment	This study
pJS045	pACD4K-CloxP with pmpA TargeTron fragment	This study
pJS048	pACD4K-CloxP with fim8A TargeTron fragment	This study
pCZ02	pACD4K-C with fim8A TargeTron fragment	This study
pMP225	pACD4K-CloxP with tssA TargeTron fragment	This study

^aKan^r, kanamycin resistance; Amp^r, ampicillin resistance.

ChIP-seq.

We performed duplicate ChIP-seq experiments as described previously (9, 70). This yielded 78 sequences that were analyzed for enriched sequence motifs using MEME-ChIP (version 4.12.0; default parameters except the HI4320 genome sequence was used as a background model) (11). Analysis of the identified motif position relative to ChIP-seq peak position using CentriMo (71) indicated a strong relationship between motif position and ChIP-seq peak position (P = 1.6e⁻⁷). Additional details are in the supplemental material.

Quantitative ChIP-PCR and qRT-PCR.

Independent ChIP experiments to validate ChIP-seq data were performed as detailed earlier (9) with one modification: 1% formaldehyde cross-linking was done with freshly pelleted and washed P. mirabilis cells. All qPCR experiments were conducted using a CFX Connect real-time system (Bio-Rad). We used 3 μl of ChIP-enriched DNA or 10% input DNA (starting DNA material) and 30 μM forward/reverse primers per amplification reaction. “No Ab” and “anti-His” IP analyses were performed with both vector control (EV) and pMrpJ-expressing cell lysates, where “no Ab” IP was used as a nonspecific control. A negative, bead-only control was included. rpoA was used as a non-MrpJ target gene control. Fold enrichments were calculated by 2^−ΔΔCT, where ΔΔC_T corresponds to the control adjusted threshold cycle (ΔC_T[His IP] − ΔC_T[no Ab IP]), and ΔC_T corresponds to the 10% input normalized C_T (C_T[IP] − C_T[Input × dilution factor]). C_T is the raw qPCR amplification cycle value, and the dilution factor for 10% input was 3.3219 (log₂ of 10).

RNA from mid-logarithmic-phase (optical density at 600 nm [OD₆₀₀] = 0.70 to 0.85) or early-stationary-phase (OD₆₀₀ = 1.9 to 2.1) aerated broth cultures was isolated and analyzed as described previously (2), except the 2× SYBR green qPCR master mix was obtained from Biotool.

For both ChIP-qPCR and qRT-PCR, the cycle was as follows: 95°C for 3.0 min; 40 cycles of [95°C for 10 s, 56°C for 30 s (+ plate read), 72°C for 20 s]; and melt curve analysis (56 to 95°C with increments of 0.5°C per 5 s [+ plate read]). The primers are listed in Table S3 in the supplemental material.

EMSA.

Complementary oligonucleotides were synthesized to encompass the MrpJ consensus site in the tssA promoter centered at nt 830177, plus five extra nucleotides on either end (total length, 29 nt; sequences are provided in Table S3 in the supplemental material). Equimolar concentrations of both oligomers were mixed, heated to 95°C, and slowly cooled to room temperature before end labeling with digoxigenin (DIG) using a DIG gel shift kit (2nd Generation; Roche) according to the manufacturer's directions. MrpJ-His₆ was purified using Ni-NTA beads (Qiagen) from a lysate of E. coli M15 (pLX2501) as previously described (16). Probes were incubated with or without MrpJ-His₆ in a reaction mixture containing poly[d(A-T)], poly-l-lysine, and reaction buffer. Protein-DNA complexes were separated on a 10% native, continuous polyacrylamide gel in ice-cold 0.5× Tris-borate-EDTA buffer (30 mA, 15 min) and transferred to nylon membrane by electroblotting (400 mA, 30 min). Detection of gel shifts was accomplished using the DIG gel shift kit and a ChemiDoc Touch imaging system (Bio-Rad).

Generation and analysis of P. mirabilis mutants.

Briefly, the TargeTron system (Sigma) was used to disrupt genes by the insertion of an intron that specifically targets the gene of interest using mutagenic PCR aided by three primers (IBS, EBS1d, and EBS2) (72). This mutated intron was ligated into vector pACD4K-CloxP (or, for the fim8A single mutant, pACD4K-C), followed by sequencing to confirm retargeting before transforming into HI4320 containing the T7 helper plasmid pAR1219. The resulting transformants were screened by kanamycin selection and genotyped by PCR. The kanamycin selection cassette, flanked by loxP sites, was removed using a cre recombinase encoded on plasmid pQL123 (73) before making double or triple mutants.

The independent growth of each mutant was analyzed by in vitro monoculture or coculture growth curves as described previously (74). Briefly, an overnight 37°C aerated culture of each strain was standardized to an OD₆₀₀ of 0.8. A 1:100 dilution of this culture was used for growth curve experiments by measuring the OD₆₀₀ at specified time points.

In vitro coculture of HI4320 and each mutant was performed as described earlier by passaging and sampling for up to 72 h (72, 74), and CFU were determined by serial dilution and plating on LB and LB supplemented with kanamycin to estimate the proportion of the wild-type versus the mutant bacterial titer.

Mouse model of UTI.

Animal experiments were approved by the New York University Langone Medical Center and University of Michigan Medical School Institutional Animal Care and Use Committees (IACUC) and were performed as described previously (9) with modifications. Briefly, overnight cultures of wild-type or mutant P. mirabilis were adjusted to an estimated density of 2 × 10⁸ CFU/ml (OD₆₀₀ of 0.2) and then mixed 1:1. Ten 5- to 6-week-old female CBA/J mice (Jackson Laboratory) were infected via transurethral catheterization while under anesthesia. Mice were cochallenged with 50 μl of the wild-type/mutant 1:1 mixture (10⁷ CFU inoculum/mouse). After 7 days, bacterial titers were determined by duplicate plating of serial dilutions of organ homogenates and urine samples on both LB agar and LB agar supplemented with kanamycin (25 μg/ml). For all samples in which the bacterial burden was above the limit of detection (200 CFU/g tissue), a competitive index (CI) was calculated as follows: CI = (mutant output/wild-type output)/(mutant input/wild-type input). All cochallenge experiments, with the exception of the fim8A and fim14A single mutants, were repeated twice (total n = 20). Mice that died before the experimental endpoint of 7 days were excluded from analysis.

Statistical analysis.

Graphing and analyses of data were done using GraphPad Prism software. A Wilcoxon matched-pair signed-rank test was used to analyze statistical significance for animal experiments. CI values were log₁₀ transformed and compared to a hypothetical value of 0 (that is, neither mutant nor wild type had an advantage). Logistic regression was computed using R software at the University of Michigan Consulting for Statistics, Computing, and Analytics Research core (CSCAR) and was used to compare complete clearance of mutant bacteria from mice cochallenged with fimbrial mutants.

Accession number(s).

Raw Illumina sequences and processed ChIP-seq data are available under GEO accession number GSE96042.