Regulation of Escherichia coli fim gene transcription by GadE and other acid tolerance gene products

Abstract

Uropathogenic Escherichia coli (UPEC) cause millions of urinary tract infections each year in the United States. Type 1 pili are important for adherence of UPEC to uroepithelial cells in the human and murine urinary tracts where osmolality and pH vary. Previous work has shown that an acidic pH adversely affects the expression of type 1 pili. To determine if acid tolerance gene products may be regulating E. coli fim gene expression, a bank of K-12 strain acid tolerance gene mutants were screened using fimA-lux, fimB-lux, and fimE-lux fusions on single copy number plasmids. We have determined that a mutation in gadE increased transcription of all three fim genes, suggesting that GadE may be acting as a repressor in a low pH environment. Complementation of the gadE mutation restored fim gene transcription to wild-type levels. Moreover, mutations in gadX, gadW, crp, and cya also affected transcription of the three fim genes. To verify the role GadE plays in type 1 pilus expression, the NU149 gadE UPEC strain was tested. The gadE mutant had higher fimE gene transcript levels, a higher frequency of Phase-OFF positioning of fimS, and hemagglutination titres that were lower in strain NU149 gadE cultured in low pH medium as compared to the wild-type bacteria. The data demonstrate that UPEC fim genes are regulated directly or indirectly by the GadE protein and this could have some future bearing on the ability to prevent urinary tract infections by acidifying the urine and shutting off fim gene expression.

Introduction

Across the world, nearly 150 million urinary tract infections (UTIs) are recorded each year worldwide [1]. For women, UTIs are the most prevalent infections outside the intestinal tract around the world [2]. Approximately 10.5 million women each year have a urinary tract infection in the United States. It is estimated that 80 % of all urinary tract infections and over 100 000 hospitalizations are due to uropathogenic Escherichia coli (UPEC) [3–5]. Adherence to uroepithelial cells is a vital part of UPEC pathogenesis [6–8]. Human and murine urinary tracts are bathed in urine that is usually slightly acidic with osmolality that can vary from 50 to 1400 mOsm kg⁻¹ [9–11]. Thus, the pH within the urinary tract is one potential environmental factor in UPEC pathogenesis.

Escherichia coli has at least five acid resistance (AR) systems that allow the species to survive in an acidic environment [12–18]. The AR1 system is tied to cAMP receptor protein (CRP) and the RpoS alternative sigma factor [13, 19]. Systems two through four are induced under acidic conditions, usually by their respective, exogenous amino acids. These three AR systems make use of decarboxylases and their cognate antiporters to maintain less acidic pH within cells grown under extreme acidic conditions (pH 2). System two (AR2) is induced in both mid-exponential and stationary phase in a low pH environment and is composed of two glutamate decarboxylases and a glutamate: γ-aminobutyric acid antiporter [19–22]. The AR3 system is composed of an arginine decarboxylase (AdiA) and an arginine:agmatine antiporter (AdiC) induced by arginine under anaerobic/acidic conditions [23, 24]. Induction of the AR4 system occurs under anaerobic/acidic conditions in the presence of lysine. A fifth system, AR5, is induced under logarithmic phase growth but it is not well characterized.

Of the five AR systems, AR2 is the best characterized [12, 18, 19, 25]. The AR2 system requires the antiporter GadC and two inducible glutamate decarboxylases: GadA and GadB. The antiporter is responsible for transporting glutamate into the cell while transporting the product of glutamate decarboxylation, glutamate: γ-aminobutyric acid, out of the cell [19–21, 25–30]. GadE is part of the LuxR family of regulatory proteins [30]. Studies have shown that GadE is the central transcriptional activator of gadA/BC [31, 32]. A microarray study identified the yhiE gene (renamed gadE) that encodes the GadE transcriptional regulator protein in E. coli grown under acidic condition [30]. A 20 bp GadE binding sequence (GAD box: 5′-TTAGGATTTTGTTATTTAAA-3′) was shown to be located −63 bp from the transcriptional start site of both gadA and the gadBC operon [20, 33, 34].

Regulation of several genes occurs following E. coli growth in an acidic environment [29]. Previously, we demonstrated that acidic growth conditions regulated transcription of several fim genes and subsequent type 1 pilus expression in E. coli [35, 36]. Expression of type 1 pili are critical for UPEC colonization of the bladder in humans and mice [37–40]. Type 1 pili are encoded by nine fim genes (fimA-I) found as part of an operon [41]. The fimA gene encodes a 17 kDa protein that comprises the bulk of the type 1 pilus structure [42, 43]. A 314 bp invertible element named fimS found immediately upstream of the fimA gene houses the promoter for the fimA gene that can undergo site-specific recombination [41, 44–46]. This fimS element can switch from aligning the fimA promoter with the fimA gene (Phase-ON, piliated) to having the fimA promoter positioned in the oppositive direction (Phase-OFF, nonpiliated). Upstream of the fimS element are the fimB and fimE genes that encode the FimB and FimE site-specific recombinases that position the invertible fimS DNA element [41, 44–46]. FimB switches the fimS element in both directions, but favours a switch to the Phase-ON orientation, whereas FimE binding switches the fimS element from Phase-ON to Phase-OFF [45–51]. Thus, expression of type 1 pili is in part driven by the ratio of the FimB and FimE site-specific recombinases.

We hypothesized that proteins that are part of one or more AR systems repress transcription of fim genes during UPEC growth in an acidic environment, such as within human kidneys. Repression of these fim genes by an AR system protein could in part explain the loss of type 1 pili on UPEC strains growing in the human kidney over time [39]. In this study, we have used fim-lux reporter fusions to screen AR systems mutants for their potential role in fim gene regulation. We demonstrate for the first time that GadE has the primary role in regulating fim gene transcription within E. coli grown under acidic conditions. Moreover, GadE appears to regulate transcription of several fim genes in E. coli , which in turn affects type 1 pilus expression.

Methods

Bacterial strains, plasmids, media, and growth conditions

The bacterial strains used in this study are shown in Table 1. E. coli strain NU149 was obtained from a patient with cystitis [39]. The λ red recombinase system with plasmids pKD4, pKD46, and pCP20 was used for construction of the Δcrp mutation [52]. Each fim-lux reporter plasmid [pHD-06 (fimE-lux), pWS145-38 (fimB-lux), and pHD-03 (fimA-lux) [53]] has the single copy number pPP2-6 plasmid as the backbone [36]. Luria broth (LB) and Luria agar (LA) were used for propagation of the bacteria and during the transformations with the addition of the following antibiotics: kanamycin, 40 μg ml⁻¹; ampicillin, 100 μg ml⁻¹, chloramphenicol 12.5 µg ml⁻¹ (Sigma Aldrich Chemical Company, St. Louis, MO). All bacteria were also grown in Luria broth (LB) buffered with sodium phosphate (0.1 M) set at pH 5.5, pH 5.5 with 400 mM NaCl (pH 5.5+), pH 7.0, or pH 7.0 with the addition of 400 mM NaCl (pH 7.0+) at 37 °C as previously described [36].

Table 1.

Bacterial strains and plasmids used in this study

Strain/Plasmid	Genotype	Source
Strain
EK227	K-12 (lambda-, F-)	[19]
EF333	gadC of EK227	[19]
EF337	adiA::MudI1734, gadC::Tn10 of EK227	[19]
EF362	rpoS::Tn10 of EK227	[19]
EF438	adiA:: pKnock of EK227	[19]
EF493	gadA::pRR10 of EK227	[19]
EF507	gadB::Km of EK227	[19]
EF530	cya::Km of EK227	[19]
EF548	gadA::pRR10, gadB::Km of EK227	[19]
EF828	gadX of EK227	[65]
EF861	gadW::Km of EK227	[65]
EF863	gadXW::Km of EK227	[65]
EF1007	gadE::Km of EK227	[63]
EF1079	Wild-type/ pPCRScript gadE of EK227	[63]
EF1083	gadE::Km/pPCRScript gadE of EK227	[63]
EF1093	gadE::Km/pPCRScript of EK227	[63]
EF1120	gadXW::Km, gadE::Tn10 of EK227	[63]
KL1	crp of EK227	This study
NU149	UPEC clinical isolate	[39]
NU149 gadE	gadE of NU149	[60]
Plasmid
pWS145-38	fimB-lux fusion on plasmid pPP2-6	[53]
pHD-03	fimA-lux fusion on plasmid pPP2-6	[53]
pHD-06	fimE-lux fusion on plasmid pPP2-6	[53]
pPP2-6	Single copy plasmid	[36]
pKD4	Flp recombinase sites	[52]
pKD46	l Red recombinase	[52]
pCP20	Flp recombinase	[52]
pPCRScript	Multicopy plasmid	[63]
pPCRScript gadE	gadE on pPCRScript Amp	[63]

Luminescence assays

Five hundred microlitres of each mid-logarithmic phase (optical density 600 nm between 0.4 – 0.55) or stationary phase (1.905 – 2.242) bacterial culture incubated at 37 °C with shaking (250 r.p.m.) was tested for bioluminescence using a FB 12 bioluminescence single tube luminometer (Zylux Corporation, Huntsville, AL). The luminescence results were divided by the OD₆₀₀ reading to standardize and were reported as relative luminescence units (RLU). Assays were performed at least three different times on different days for each particular assay and the data was expressed as the mean+standard deviation of these measurements.

Construction of the Δcrp mutant strain

The λ red recombinase system was used to construct the Δcrp mutation in the EK227 strain [52]. Briefly, the primer pair Crp1 (5- TGGCTCTGGAGAAAGCTTATAACAGAGGATAACCGCGCTTGTGTAGGCTGGAGCTGCTTCG −3′) and Crp2 (5′- CAAAATGGCGCGCTACCAGGTAACGCGCCACTCCGACGGACATATGAATATCCTCCTTAG −3′) were synthesized (Integrated DNA Technologies, Coralville, IA) and used in PCR amplifications with pKD4 plasmid DNA as a template under the following conditions: initial denaturation at 95 °C, 5 min; then 35 cycles of 95 °C, 1 min; 55 °C, 1 min; and 72 °C, 2 min. The resulting PCR product was concentrated and separated on a 0.8 % agarose gel, cut out, and the DNA extracted from the agarose gel. With this purified PCR product, an electroporation was performed into competent NU149/pKD46 cells, selecting for transformants on LA with kanamycin. Several transformants were screened for a loss of crp gene amplification using a PCR-based system with the Crp3 (5′- AACAGACCCGACTCTCGAAT −3′) and Crp4 (5′ - GTTTTGCCAGATGTTTTGCCAGATTCAGCAGAG - 3′) primer pair. One transformant was chosen for further analysis. To remove the kanamycin resistance gene, the pCP20 plasmid was introduced into the transformant by electroporation. The resulting UPEC strain was named KL1.

PCR for fimS orientation determination

To determine the orientation of the fimS invertible element, previously described PCR techniques were used and products visualized with FOTO/Analyst PC Image Software (Fotodyne, Hartland, WI) [54, 55]. To quantify the percentage of Phase-ON or Phase-OFF bacteria, a standard curve was prepared as described by Teng et al. [56] using locked-ON (DH5a/pAON-1) [34] and locked-OFF bacteria (NU149 cells passaged five times on agar shown to be 100 % Phase-OFF) [54] as PCR templates and the ImageQuant 5.2 software (GE Healthcare, Chicago, IL) to measure pixilation for each band. The Phase-ON and Phase-OFF bands were standardized to the ftsZ bands for each condition.

Extraction of total RNAs and conversion to cDNAs

Total RNA was extracted from E. coli NU149 cells grown with shaking to mid-exponential phase (OD₆₀₀ 0.5–0.8) at 37 °C in pH 5.5, pH 5.5+, pH 7.0, or pH 7.0+ buffered LB medium using TRIzol Reagent (Invitrogen, Carlsbad, CA) with 50 µl lysozyme (10 mg ml⁻¹, New England Biolabs, Ipswich, MA). All RNA samples were digested with DNase I (New England Biolabs) to remove contaminating DNA. For each RNA sample, 2 µg RNA was converted to cDNA using a SuperScript First-Strand Synthesis kit (Invitrogen) following the protocol recommended by the manufacturer.

Quantitative reverse-transcribed PCR (qRT-PCR)

The qRT-PCRs were performed using a BioRad CFX96 system (BioRad, Hercules, CA). One microlitre of a 1/10 dilution of cDNA was used in each qRT-PCR. A Power SYBR Green mastermix (Applied Biosystems, Foster City, CA) was used and products were detected with an FAM setting. The ftsZ gene was used to standardize between the samples using the EcFtsZ1/EcFtsZ2 primer pair [36, 55]. A fimB product was amplified using the primer pair FimB5/FimB6 [35]. Amplification of the fimE product was done with the FimE1/FimE2 primer pair [36]. The qRT-PCR parameters were as follows: initial denaturation for 10 min at 95 °C followed by 38 cycles of 15 s at 95 °C and 1 min at 58 °C. Data were analysed by using the 2^-ΔΔC method [57] by standardizing the fimB and fimE transcript levels to ftsZ transcript levels. A minimum of three separate analyses were done for each gene.

Hemagglutination assays

Hemagglutination assays were performed with 1 % guinea pig erythrocytes (Hardy Diagnostics, Santa Maria, CA) with and without the addition of 2 % (wt/vol) mannose (Sigma Aldrich Chemical Company) on the EK227, EF1007, EF1007/pPCRScript gadE, NU149, NU149 gadE, and NU149 gadE/pPCRScript gadE strains grown statically in pH 5.5, pH 5.5+, pH 7.0, and pH 7.0+ buffered LB at 37 °C as previously described [58]. The titres represent the average of three separate runs.

Statistics

A Student’s t test was used for statistical analyses. P values of <0.05 were considered significant.

Results

Transcriptional changes in fim gene expression in mid-logarithmically cultures grown in buffered LB

Previously, we showed that E. coli growth in an acidic environment caused transcriptional changes in several fim genes [36]. Our hypothesis was one or more AR system proteins regulated fim gene transcription in E. coli growing in an acidic environment. As described above, there are several AR systems important for E. coli survival in an acidic environment [12, 16, 18, 59]. To assess which AR system(s) might be involved in the transcriptional regulation of fim genes, we transformed single copy number plasmids containing fimA-, fimB-, or fimE-lux fusions into E. coli strains with mutations in genes encoding proteins tied to the assorted AR 1–3 systems.

Most of the AR system mutations had no significant effect on fimB-lux transcription when the E. coli strains were grown to mid-logarithmic phase in pH 5.5, pH 5.5+, pH 7.0, and pH 7.0+ buffered LB (Table S1). However, we found that a few AR system mutants had effects on fimB transcription when the bacteria were grown under acidic conditions. A mutation in the gadE gene caused a 1.3-fold increase in transcription of fimB when the E. coli was grown in pH 5.5 buffered LB (P<0.001) (Fig. 1a, Table S1). Complementation of the gadE mutation brought fimB transcript levels down to slightly less than the wild-type level, whereas the gadE mutant with a blank vector showed transcription similar to the gadE mutant.

Fig. 1.

Assessing a gadE mutation and complementation on fim-lux transcription in mid-logarithmically grown cultures of E. coli in in buffered pH 5.5 and 7.0 LB media with different osmolarities. The fim genes tested were a) fimB, b) fimE, and c) fimA. All of the bacterial cultures were grown to mid-logarithmic phase in pH 5.5, pH 5.5+, pH 7.0, and pH 7.0+ buffered LB. Luminescence was standardized to the wild-type strain grown in pH 7.0 buffered LB that was set to 1. The strains that were compared included EK227 (wild-type, open white bar), EF1007 (gadE mutant, solid black bar), EF1083 (gadE mutant with pPCRScript gadE plasmid, right striped bar), and EF1093 (gadE mutant with pPCRScript, left striped bar). Data represents the mean±standard deviation from at least three separate runs. * Denotes P<0.05 and ** denotes P<0.001.

Mutations in both gadE transcriptional regulator genes (gadW and gadX) significantly diminished fimB transcription compared to wild-type following growth in pH 5.5 and pH 5.5+ LB (Table S1). The gadXW double mutant showed significantly less fimB transcription when grown in pH 5.5 and pH 5.5+ media (Table S1). A triple mutant where gadXW were deleted combined with a transposon insertion mutation of gadE showed fimB transcription results that were similar to the gadW single mutant strain when compared with the wild-type strain. Thus, several AR2 system proteins appear to regulate fimB genes when E. coli is growing in an acidic environment, supporting our hypothesis of AR system regulation.

Besides the AR2 system involvement in regulating fim gene transcription, we also found AR1 system involvement. Deletion of the crp gene also caused significant downregulation of fimB transcription in the E. coli populations grown in pH 5.5 and pH 5.5+ buffered LB compared to the wild-type bacteria grown in the same media to mid-logarithmic phase (Table S1).

An assessment of fimE-lux transcription was then done on mid-logarithmic cultures. Again, most of the AR system mutant strains exhibited no significant change in fimE transcription levels (Table S2). However, a cya mutant displayed significantly increased fimE transcript levels when grown in pH 5.5 LB when grown in pH 5.5, pH 5.5+, pH 7.0, and pH 7.0+ buffered LB versus wild-type. A crp mutant showed a significant increase in fimE transcription, but only in pH 7.0 and pH 7.0+ buffered LB (Table S2).

Mutations of the AR2 system were also examined for their effect on fimE-lux transcription during mid-logarithmic phase. A mutation in the gadE gene caused a pH deptendent effect on fimE transcription. Luminescence activity significantly increased in the gadE mutant strain grown in pH 5.5 and pH 5.5+ LB compared to wild-type (Fig. 1b, Table S2). An examination of the transcriptional regulators of gadE showed a gadX mutation caused significantly elevated fimE transcription when the strain was grown in pH 5.5+ and pH 7.0+ LB versus the wild-type strain (Table S2). The triple gadXW gadE mutant strain displayed fimE transcription levels that were significantly lower in pH 5.5 and pH 5.5+ LB compared to the wild-type strain grown in the same media. From this analysis, it was apparent that several AR2 system proteins regulated fimE transcription as well.

Lastly, fimA-lux transcription was assessed in the AR system mutants grown to mid-logarithmic phase. As was observed for the fimB- and fimE-lux reporter fusions, most of the AR system mutations had no significant effect on fimA transcription (Table S3). However, the crp mutant had a 1.4-fold decrease in fimA transcription compared to the wild-type strain when grown in pH 5.5 LB (P<0.002).

Mutations in several AR2 system genes also caused significant changes in fimA transcription. A gadE mutation led to 1.5-fold (P<0.001) and 1.7-fold (P<0.002) increases in fimA transcription compared to the wild-type when the strains were grown in pH 5.5 and pH 5.5+ LB media, respectively (Fig. 1c, Table S3). Complementation of the gadE mutation dropped fimA transcription below the wild-type level when the strains were grown in any of the buffered LB media. Again, the gadE mutant with the empty vector had transcription levels similar to the gadE mutant.

A gadX mutation led to significant declines in fimA transcription following growth in pH 5.5 and pH 5.5+ LB, respectively (Table S3). Similarly, gadW and gadXW mutants displayed significantly lower fimA transcription versus wild-type when the strains were grown in pH 5.5 and pH 5.5+ LB, respectively. A gadXW gadE triple mutant displayed significantly lower levels of fimA transcription compared to wild-type when grown in all four types of media. Thus, the AR2 system mutants had the biggest effect on fimA transcription in mid-logarithmic cultures.

Transcriptional changes in fim gene expression in stationary phase cultures grown in buffered LB

The data above demonstrated transcriptional changes in three fim genes when the E. coli was grown to mid-logarithmic phase. RpoS and the AR2 system are important during stationary phase growth of E. coli in an acidic environment [19–22, 60]. Next, we tested the same strains in the same growth media after stationary phase growth. First, fimB-lux transcription was examined. Again, most of the AR mutants tested had no significant effect on fimB transcription (Table S4). The crp mutation had significantly higher fimB transcription compared to wild-type when grown in pH 5.5 LB, and the cya mutant displayed a significant increase in fimB transcription in pH 7.0 LB. The rpoS mutant showed significantly higher fimB transcription when grown in pH 7.0 and pH 7.0+ LB versus wild-type.

Mutations in several AR2 system genes also had effects on fimB transcription. The gadE mutant had 1.2-fold (P<0.001) and 1.6-fold (P<0.001) increases in fimB transcription compared to wild-type when grown in pH 5.5 and pH 5.5+ LB media, respectively (Table S4). Complementation of the gadE mutation brought fimB transcription back to or lower than wild-type levels, whereas the gadE mutant with the empty vector pPCRScript plasmid exhibited transcription similar to the gadE mutant. Both the gadX and gadW mutants showed significantly lower fimB transcription after growth in pH 5.5 and pH 5.5+ LB media, respectively. The gadW mutant also displayed significantly lower fimB transcription versus wild-type following growth in pH 7.0 and pH 7.0+ LB. An analysis of the gadXW double mutant showed significantly lower fimB transcription in all of the growth media tested when compared to the wild-type strain. Lastly, a gadXW gadE triple mutant was tested to determine if the added gadE mutation had an overall effect on fimB transcription. The triple mutant also exhibited a significant decrease in fimB transcription in all of the growth media used versus wild-type (Table S4).

An examination of fimE-lacZ transcription was also undertaken after stationary growth in the same four growth media. Most of the mutants had no effect on fimE transcription (Table S5). Testing of the cya mutant resulted in a significant increase in fimE transcription compared to wild-type following growth in all of the media, whereas the crp mutant displayed higher fimE transcription versus wild-type when grown in pH 7.0 LB (1.8-fold, P<0.002).

Involvement of the AR2 system in regulating fimE transcription during stationary phase growth was also observed. The gadE mutant showed significantly more fimE transcription compared to wild-type when grown in pH 5.5 and pH 5.5+ LB (Table S5). Complementation of the gadE mutation brought fimE transcription back to or lower than wild-type levels. Other AR2 system mutants also significantly affected fimE transcription. The gadX mutant demonstrated significantly lower fimE transcription versus wild-type in pH 7.0 and pH 7.0+ LB. A gadW mutant had uniformly diminished fimE transcription in all four types of media that were tested versus wild-type. When a gadXW double mutant and gadXW gadE triple mutant were examined, fimE transcription results were similar to those observed for the gadW mutant.

Lastly, fimA-lacZ transcription was examined in the various AR system mutants after stationary phase growth. The majority of the mutants displayed no significant difference in fimA transcription when compared to wild-type (Table S6). Growth in all four media caused fimA transcription to be significantly elevated in the rpoS mutant compared to wild-type. The cya mutant had a little less than a three-fold lower fimA transcription versus wild-type when grown in pH 7.0 LB and pH 7.0+ LB that was significant compared to wild-type.

Transcription of fimA was also affected by several AR2 system mutations. A higher level of fimA transcription compared to wild-type was reported when the gadE mutant was grown in pH 5.5 LB (1.3-fold, P<0.001) and pH 5.5+ LB (1.5-fold, P<0.001), respectively. The gadX mutant displayed significantly lower fimA transcription when measured against wild-type when grown in all four types of buffered LB. A gadXW double mutant as well as a gadXW gadE triple mutant also showed significantly lower fimA transcription versus wild-type when grown in the four buffered LB conditions. Thus, AR2 system proteins appear to regulate fimA transcription in stationary phase cultures.

Transcription of fim genes also affected by a UPEC strain grown in an acidic environment

Our luminescence assay results above showed a mutation in the gadE gene had significant effects on fimB and fimE transcription when the E. coli strains were grown in an acidic environment. To demonstrate that GadE may regulate fim gene transcription in a UPEC strain, a qRT-PCR analysis was performed with a gadE mutant strain that had been previously described [61] compared to wild-type and complemented gadE mutant strains grown in the same in vitro environments noted above.

An examination of the wild-type strain showed fimB transcript abundance was the highest when the strain was grown in pH 7.0 LB (Fig. 2a). Transcription of fimB in the wild-type strain was lower after growth in the following LB media compared to growth in pH 7.0 medium: 2.5-fold in pH 7.0+ LB (P<0.04), 5.7-fold in pH 5.5 LB (P<0.006), and 53-fold when grown in pH 5.5+ LB (P<0.003). The fimE transcript abundance was not significantly different in RNA samples collected from the wild-type strain grown in pH 7.0 LB compared to growth in pH 7.0+ LB or pH 5.5 LB (Fig. 2b). However, fimE transcript levels were 1.9-fold lower in strain NU149 grown in pH 5.5+ LB versus when grown in pH 7.0 LB (P<0.05).

Fig. 2.

Quantitative determination of the effect of the gadE mutation on (a) fimB and (b) fimE transcript expression by qRT-PCR. UPEC NU149 (open white bar), NU149 gadE (solid black bar), and NU149 gadE/pPCRScript gadE (striped bar) cells were grown to mid-logarithmic phase in pH 7, pH 7+, pH 5.5, and pH 5.5+ buffered LB media. The fold change in fimB and fimE transcript levels that were corrected using ftsZ are shown as the mean±SD from three separate runs. Both the fimB and fimE transcript levels were set to 1.0 at pH 7 for the wild-type strain. For significant differences, * denotes P<0.05 and ** denotes P<0.01.

Neither the gadE mutant nor the complemented gadE mutant had significant differences in fimB or fimE transcript abundance when the UPEC strain was grown in pH 7.0 LB or pH 7.0+ LB as compared with the wild-type strain grown under the same conditions. However, the NU149 gadE strain showed significantly higher transcription of fimB versus wild-type when the mutant was grown in pH 5.5 LB (1.5-fold, P<0.05) and pH 5.5+ LB (1.6-fold, P<0.05). Furthermore, fimE transcription was higher in the gadE mutant compared to wild-type when both strains were grown in pH 5.5 LB (2.3-fold, P<0.02) and pH 5.5+ LB (2.6-fold, P<0.02). Complementation of the gadE mutation restored fimB and fimE transcript abundance back to wild-type levels. These results demonstrate that GadE has an effect on both fimB and fimE transcription when a UPEC strain is grown in an acidic environment, but the regulatory effect is more pronounced in regard to fimE transcription.

Positioning of the fimS invertible element affected in UPEC strain grown in an acidic environment

In our previous studies, we have demonstrated that the position of the fimS invertible element was changed when E. coli was grown in pH 5.5 media versus pH 7.0 media [33, 34, 45, 46]. From the qRT-PCR results described above, we showed that fimE transcript levels rose the most when the UPEC gadE strain was grown in pH 5.5 buffered LB with or without added NaCl and compared to the wild-type UPEC strain. If the ratio of FimB and FimE skewed to favour FimE, then positioning of the fimS invertible element would likely shift to a Phase-OFF orientation [41, 44–46].

Our analysis showed that gadE mutant cells grown at pH 7 had the invertible element positioned approximately equal (ratio ON/OFF=79 : 21%) to the wild-type cells (ratio ON/OFF=80 : 20%) and the complemented mutant (ratio ON/OFF=77 : 23%, Fig. 3). Under pH 7+ growth conditions, all three strains shifted significantly towards a Phase-OFF orientation [wild-type (ratio ON/OFF=50 : 50%), gadE mutant (ratio ON/OFF=49 : 51%), and complemented mutant (ratio ON/OFF=51 : 49%)]. In pH 5.5 LB, the gadE mutant had more cells in the Phase-OFF orientation (ratio ON/OFF=21 : 79%) than the wild-type (ratio ON/OFF=52 : 48%), consistent with the observed qRT-PCR effects on fimB and fimE transcription noted above. More Phase-OFF oriented cells were also present in the gadE mutant grown in pH 5.5+ LB (ratio ON/OFF=51 : 49%) versus wild-type cells (ratio ON/OFF=20 : 80%). Together, these results indicated that the derepression of fimB transcription in the gadE mutant led to more Phase-ON oriented cells when grown in a high osmolality environment. In a low osmolality/low pH environment, less fimB was transcribed in the gadE mutant strain compared with wild-type, suggesting another regulatory system such as an acid tolerance system regulator may be involved in regulating fimB transcription in the absence of GadE.

Fig. 3.

Determination of the 314 bp fimS invertible element orientation in E. coli NU149, NU149 gadE, and NU149 gadE/pPCRScript gadE grown in pH 5.5, pH 5.5+, pH 7.0, and pH 7.0+ buffered LB by multiplex PCR. Analysis was performed on chromosomal DNA isolated from NU149 (WT), a NU149 gadE mutant (e), and a complemented mutant NU149 gadE/pPCRScript gadE (E+). Multiplex PCRs were set up with INV and FIMA primers to amplify ‘Phase-ON’ DNA (450 bp), FIME and INV primers to amplify ‘Phase-OFF’ DNA (750 bp), and EcFtsZ 1 and 2 primers to amplify the ftsZ gene (302 bp). Each multiplex was run at least twice. The lanes were loaded as follows: lane 1, WT (pH 7.0); lane 2, gadE (pH 7.0); lane 3, gadE/pPCRScript gadE (pH 7.0); lane 4, WT (pH 7.0+); lane 5, gadE (pH 7.0+); lane 6, gadE/pPCRScript gadE (pH 7.0+); lane 7, WT (pH 5.5); lane 8 (pH 5.5); lane 9, gadE/pPCRScript gadE (pH 5.5); lane 10, WT (pH 5.5+); lane 11, gadE (pH 5.5+); and lane 12, gadE/pPCRScript gadE (pH 5.5+).

Type 1 pili expression lower in K-12 and UPEC gadE mutant strains grown in an acidic environment

As noted above, transcription of fimE was higher than transcription of fimB when the UPEC gadE mutant strain was grown under acidic growth conditions, suggesting that the ratio of FimE to FimB changed to favour FimE. Positioning of the fimS invertible element changed to favour a Phase-OFF orientation in the gadE mutant strain compared to wild-type when both strains were grown in pH 5.5, pH 5.5+, pH 7.0, and pH 7.0+ LB. To determine if the level of type 1 pili concurrently changed in the gadE mutant strain grown under the same growth conditions, HA assays were done using guinea pig erythrocytes.

First, the K-12 strain was examined for the effect of a gadE mutation on type 1 pili expression. The HA titres did not vary between the wild-type, gadE mutant or complemented gadE mutant following growth in pH 7.0 or pH 7.0+ buffered LB (Table 2). However, the gadE mutant had a four-fold lower HA titre than either the wild-type or complemented gadE mutant when the strains were grown in pH 5.5 LB. In pH 5.5+ LB, the HA titre was two-fold lower for the gadE mutant compared to both the wild-type and complemented gadE mutant strains. The addition of 2 % mannose abolished the HA titres, confirming that the HA titres were mannose-sensitive. Thus, the gadE mutation affected type 1 pili expression in pH 5.5 and pH 5.5+ LB.

Table 2.

Effect of a gadE mutation and complementation of the mutation on type 1 pilus production in E. coli measured by the hemagglutination (HA) titre in a K-12 strain and a uropathogenic clinical isolate

		HA Titre
Strain	pH 5.5	pH5.5+*	pH 7.0	pH 7.0+*
EK227	64	32	256	128
EF1007 (gadE)	16	16	256	128
EF1083 (gadE/pPCRScript gadE)	64	32	256	128
NU149	256†	64	1024	256
NU149 gadE	128	32	1024	256
NU149 gadE/pPCRScript gadE	256	64	1024	256

*+=addition of 400 mM NaCl.

†Mean value from at least three separate runs.

The results demonstrated that the HA titres did not vary in the NU149, NU149 gadE or NU149 gadE/pPCRScript gadE strains when they were grown in pH 7.0 LB or pH 7.0+ LB (Table 2). However, when the gadE mutant was grown in pH 5.5 LB, the HA titre was two-fold lower compared to the wild-type strain, which was complemented by the pPCRScript gadE plasmid. Furthermore, growth of all of the strains in pH 5.5+ LB had the lowest HA titres, although the gadE mutant displayed an HA titre that was two-fold lower than either the wild-type or complemented strains. The addition of 2 % mannose abolished the HA titres, confirming that the HA titres were mannose-sensitive. These results demonstrate that GadE has an effect on type 1 pili expression under acidic growth conditions in two strains of E. coli .

Discussion

Uropathogenic E. coli survive in pH stressed environments [10, 14, 16, 50], such as within the human or murine urinary tract. Previously, we showed that acidic pH affects UPEC type 1 pilus expression [33, 34]. Our working hypothesis was that when E. coli is grown in an acidic environment (e.g. within the human or mouse urinary tract) one or more AR system proteins regulate fim gene transcription, and in turn, the expression of type 1 pili. Using AR system mutants, fim-lux reporter fusions, PCR, and hemagglutination titering; we have demonstrated that several AR system mutants affected fim gene transcription. In particular, a gadE mutant displayed higher fimE transcription compared to the wild-type strain when grown in an acidic environment, leading to the fimS element switching to a more Phase-OFF orientation and less type 1 pili production.

Five different AR systems maintain homeostasis in E. coli growing in an acidic environment. We tested mutants in an E. coli K-12 background comprising the AR1-3 systems. No effect on fim gene transcription was observed when using the AR3 system mutants. The AR1 system mutants that we tested had some effect on fim gene transcription. Our analysis of the crp mutant showed lower fimB transcription and higher fimE transcription when the strain was grown in pH 7.0 media in comparison to the wild-type strain. In addition, the cya mutant had higher fimE transcription in all of the media tested during the mid-logarithmic and stationary growth phases. Previous studies have explored some AR1 system mutations to determine their effect on type 1 pilus expression. Muller et al. [62] used a Locked-ON mutant to show that a crp mutation caused a two-fold decline in fimA transcription versus the wild-type strain when grown in unbuffered LB, suggesting Crp activated transcription of fimA. We observed fimE transcription changes in our cya mutant compared to wild-type, but another study demonstrated no effect on fimbriation when using a cya mutant grown in nutrient broth with and without glucose [63]. Divergent strains and divergent growth conditions could have an effect on fim gene transcription and type 1 pilus expression.

Our analysis of the rpoS mutant demonstrated no significant difference in any of the fim gene transcription levels following mid-logarithmic growth versus wild-type. However, stationary phase grown rpoS mutant cells had significantly higher fimB transcription and no change in fimE transcription when the cultures were grown in pH 7.0 and 7.0+ media, switching the ratio of FimB and FimE to favour FimB. These results are in line with a previous study that showed RpoS strongly repressed fimB transcription in stationary phase grown cultures [64].

Although we saw some fim gene transcriptional changes using AR1 system mutants, the most marked changes in fim gene transcription were observed using AR2 system mutants. Acid response system two is the glutamate decarboxylase system [20, 21, 27, 28], which is positively regulated by the GadE protein [30, 34, 65, 66]. Transcription of gadE is in turn activated by GadX and GadW [22, 65, 67–74]. Even though the AR2 system mutants had the most obvious effects on fim gene transcription, the differences were all less than three-fold. Previous work has shown a gadE mutation caused a 60- to 80-fold change in gadA-gadC transcript abundance compared to wild-type [75], which is a much higher fold difference in transcription compared to the fim transcription changes we observed in this study. The glutamate and glutamate decarboxylases are integral to the survival of E. coli in an acidic environment, so the high transcriptional changes in the gadA-C genes are warranted. On the other hand, regulation of type 1 pilus expression in E. coli growing in an acidic environment does not have the same survival urgency because the bacterial cells will still survive with or without type 1 pilus expression.

The most pronounced effect on fim gene transcription was the result of a mutation in the gadE gene. All three fim genes (fimA, fimB, and fimE) had elevated transcription in the E. coli K-12 gadE mutant strain grown to mid-logarithmic and stationary phase in pH 5.5 and pH 5.5+ media, although the most significant effect was on fimE transcription. Complementation of the gadE mutation led to fim gene transcription returning to wild-type levels. Moreover, similar changes in fimB and fimE transcription were observed in a UPEC gadE mutant strain, confirming that the transcriptional changes were not strain specific. Our previous study with E. coli grown in buffered LB showed lower transcription of fimA, fimB, and fimE in a pH 5.5 versus a pH 7 environment [36]. No GAD box was observed in the promoter sequences of the three fim genes [20, 33, 34], so the GadE regulatory effect observed in this study is likely to be through an intermediary. GadE can activate lrp transcription and Lrp affects positioning of the fimS DNA element [30]. The results from the current study suggest that GadE may be indirectly repressing transcription of these three fim genes when the E. coli cells are grown in an acidic niche, such as the human or murine urinary tract.

Since gadE transcription is activated by the GadX and GadW proteins, we also examined the effects on fim gene transcription using gadX and gadW single mutants as well as a gadXW double mutant. Both gadX and gadW single mutations led to lower fimB transcription in E. coli grown to stationary phase under pH 5.5 conditions, but only the gadW mutant affected fimB transcription in pH 5.5 mid-logarithmic culture. The gadXW double mutant also displayed lower fimB transcription compared to the wild-type strain grown in pH 5.5 broth medium. Our results using a gadXW gadE triple mutant demonstrated that GadXW appear to have a dominant effect over GadE in regard to the regulation of fimB transcription in both mid-logarithmic and stationary phase cultures. For fimE transcription, the gadXW gadE triple mutant was the only one that showed a significant effect in pH 5.5 cultures grown to mid-logarithmic phase; whereas gadW, gadXW, and gadXW gadE mutants all showed lower transcription levels in pH 5.5 stationary phase versus the wild-type strain. Lastly, fimA transcription was significantly lower in the gadX, gadXW, and gadXW gadE mutants compared to wild-type in all four growth conditions during mid-logarithmic and stationary phase growth. Again, GadX and GadW appeared to have dominant effects over GadE.

Besides GadE acting as a regulator of these fim genes, it is possible that GadX and GadW could not only be affecting gadE transcription, but could also be directly affecting transcription of the fim genes. Seo et al. demonstrated that gadX and gadW mutants have wide ranging effects on transcription of a variety of genes in E. coli cultured in pH 5.5 medium when compared to wild-type [75]. One of the genes that showed less transcript abundance in the gadW mutant versus wild-type grown in pH 5.5 medium was fimI that sits in the middle of the fim operon [76]. The effects we observed are more subtle (generally less than a two-fold change) that may have been missed in the prior study, but subtle changes in the ratio of FimB and FimE would have dramatic effects on positioning of the fimS element.

Certainly, GadX and GadW could be acting indirectly through one or more intermediaries to regulate fim gene transcription. Several studies have shown that GadX regulates a number of genes in enteropathogenic E. coli , including the per gene that encodes the Per regulatory protein [77–79]. GadW can complex with GadE to regulate enterohemorrhagic E. coli virulence factor genes [80]. Since GadX and GadW regulate other virulence factor genes in other E. coli pathotypes, it is a reasonable assumption that either protein may be regulating UPEC fim genes as well.

Other global regulators are tied to regulating the transcription of E. coli fim genes based on the growth environment as depicted in Fig. 4. Two examples are RcsB and OmpR that we have shown affect type 1 pilus expression. Besides its role in osmotic stress, OmpR is also involved in acid resistance [59, 81]. Although ompR transcript levels do not increase in E. coli grown in a low pH environment [35, 81], OmpR protein levels do increase [35]. Escherichia coli grown in a pH 7 niche exhibit fimB transcriptional activation due to RcsB [46]. In addition, RcsB can form heterodimers with GadE and other proteins to provide acid tolerance to E. coli growing in stationary and logarithmic phase cultures [34, 82–84].

Fig. 4.

Model for how proteins regulate fim gene transcription (modified from Schwan [41]). The inverted repeat left and right (IRL and IRR) are shown as open boxes. Binding sites for integration host factor (IHF I and II) and leucine-responsive protein (Lrp1, 2, and 3) are also represented as open boxes. Genes are displayed as black boxes and the promoters are shown as bent black arrows. The dark grey arrows correspond to FimB and the light grey arrows are for FimE. Black arrows signify an effect on the fimS element. Solid green arrows indicate confirmed binding associated with stimulatory effects, whereas dashed green arrows indicate presumed stimulatory effects. Solid red arrows indicate confirmed binding associated with repressing effects, whereas dashed red arrows indicate presumed repressing effects.

What is the relevance of GadE and other acid tolerance proteins affecting fim gene transcription in UPEC? Escherichia coli is found as a commensal species in the large intestines of many animals. In this environment, the pH is closer to 7 [85] and GadE expression could be reduced, allowing FimB to be expressed at a greater level than FimE, causing the fimS invertible element to be aligned to favour transcription of the fim operon [45–51]. Type 1 piliated E. coli cells bind to the intestinal epithelium [86]. For some women, the intestinal E. coli can then colonize the vaginal surface, which is also close to a pH of 7. Again, the environment would favour fimA transcription and type 1 piliated E. coli could then attach to vaginal epithelial cells [87]. Some E. coli serotypes can now gain entry to the urethra and ascend to the bladder. Now, the E. coli transition from being in a near pH 7 environment to an acidic environment where gadX and gadW transcription would be activated followed by activation of gadE transcription [12]. As observed from this study, more GadE expression would cause repression of fimA, fimB, and fimE. More GadX could also induce hns and rpoS transcription, which could cause other regulatory changes in fim gene transcription. The fimS element would switch to a Phase-OFF position and fimA transcription would be lower. Less FimA would be produced and fewer type 1 pili would be present on the E. coli cells in this acidic environment. However, UPEC cell attachment to mannose receptors via the type 1 pili would in part negate the loss of type 1 pili expression. Further ascension into the kidneys would change the environment to an even more acidic niche [10], which could lead to even more repression of fimA transcription and subsequent loss of more type 1 pili on the surface of the E. coli .

Our study has several limitations. One limitation is the E. coli cells were grown in buffered LB with and without added NaCl. Growth in a mouse or human urinary tract bathed in urine is a more complex environment than can be achieved using LB. Another limitation is we do not know how the AR2 system proteins may be regulating expression of type 1 pili. Electrophoretic mobility shift assays with purified GadE, GadX, or GadW protein could help to answer the question as to whether any of these proteins directly regulate fim gene transcription or is regulation an indirect effect? The absence of a GAD box upstream of the fimA, fimB, and fimE genes suggests the regulation is indirect, possibly through H-NS, RpoS or Lrp [30]. Lastly, we have examined two E. coli strains and acid response regulation could vary in other E. coli strains. However, several UPEC strains exhibit a loss of type 1 pili expression over time when infecting mouse kidneys [39] and UPEC strain UTI89 displays a lower HA titre when grown in acidic LB compared to pH 7 LB, which suggests that AR2 system regulation may occur in a broader range of strains.