FumC is required for E. coli fitness in vivo

The fumarase, FumC, catalyzes the oxidative TCA cycle step by which fumarate is converted to malate (Fig 1). Previously, we have shown that deletion of fumC in E. coli CFT073 results in a fitness defect in the mouse bladder during urinary tract infection (UTI), supporting the hypothesis that the oxidative TCA cycle plays a role during infection [1, 5]. E. coli encodes an additional fumarase, fumA, adjacent to fumC in the CFT073 genome. Like FumC, FumA also catalyzes fumarate oxidation to malate during the oxidative TCA cycle, meaning that these enzymes can both perform the same reaction with equal substrate affinities [25] (Fig 1). A third fumarase, fumB, is encoded by E. coli; however, unlike FumA and FumC, FumB operates in the branched, reductive TCA pathway converting malate to fumarate (Fig 1). Despite operating in different TCA cycles, FumA and FumB have 90% identity to each other. To further investigate the in vivo roles of these additional fumarase enzymes, single mutant strains fumA and fumB, double mutant strains fumAB, fumAC, and fumBC, and a triple mutant strain fumABC were constructed in CFT073 and examined during experimental UTI. Based upon our previous findings, we hypothesized that the mutant strains with disruptions within the oxidative TCA cycle (fumA and/or fumC) would result in a fitness defect following infection while mutant strains with a disruption in the reductive TCA pathway (fumB) would be dispensable during UTI.

To directly test each individual fumarase enzyme, double mutant strains were constructed in which only one fumarase is present; double mutant strain fumAB only encodes fumC, double mutant strain fumAC only encodes fumB, and double mutant strain fumBC only encodes fumA. Interestingly, we observed that following co-challenge with wild-type CFT073, mutant strain fumAB displayed a fitness advantage in the bladder (P = 0.0011) (Fig 2A). In contrast, co-infection with wild-type CFT073 and double mutant strain fumAC or fumBC, both demonstrated a significant fitness defect in the bladder (P = 0.0002 and P = 0.0001, respectively) and kidneys (P = 0.0011 and P = 0.0067, respectively) (Fig 2B and 2C). Consistent with this, disruption of all three fumarases in mutant strain fumABC also resulted in a significant fitness defect in the bladder and kidneys (P = 0.0005 and P = 0.0012, respectively) (Fig 2D). These findings clearly indicate that any mutant strain with a disruption in the fumC gene (fumC, fumAC, fumBC, and fumABC) results in a fitness defect during infection suggesting that only fumC is required for UTI.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. FumC is required for in vivo fitness of CFT073.

In vivo Log₁₀ competitive indices (CI) were determined following 48 h co-challenge infections in female CBA/J mice with a 1:1 ratio of (A) double mutant fumAfumB and wild-type CFT073, (B) double mutant fumAfumC and wild-type CFT073, (C) double mutant fumBfumC and wild-type CFT073, and (D) triple mutant fumAfumBfumC and wild-type CFT073. (E) In vivo Log₁₀ CI was determined following infection with a 1:1:1 ratio of single mutants fumC, fumA and fumB. Each symbol represents bladder (closed circle) and kidneys (open square) from an individual mouse. Log₁₀ CI>1 indicates a fitness advantage and Log₁₀ CI<1 indicates a fitness defect of the mutant. Significant differences in colonization (**P<0.01, ***P<0.001) were determined with the Wilcoxon signed-rank test. Triple co-challenge data was analyzed using a one-way ANOVA with multiple comparisons (**P<0.01).

https://doi.org/10.1371/journal.ppat.1008382.g002

To further investigate the individual fumarase single mutants in vivo and rank their fitness in relationship to each other, an infection with all three single mutants fumA, fumB, and fumC in a 1:1:1 ratio was examined during experimental UTI. Only mutant strain fumC was statistically outcompeted by mutant strain fumA (P = 0.0041) and mutant strain fumB (P = 0.0022) in the mouse bladder suggesting loss of FumC is detrimental to UPEC UTI (Fig 2E). Indeed, a fitness advantage during experimental UTI was only observed for double mutant strain fumAB which is able to encode fumC. These findings suggested that only FumC of the oxidative TCA cycle is important for CFT073 colonization of the mouse urinary tract. Additionally, these results also suggest that eliminating non-essential fumarase enzymes, FumA and FumB, may enhance fitness for UPEC in the bladder microenvironment.

Loss of FRD ameliorates the fitness defect in the SDH mutant

Membrane-bound succinate dehydrogenase (SDH) consists of four subunits, encoded by sdhCDAB, which catalyzes succinate oxidation to fumarate during the oxidative TCA cycle and donates electrons to ubiquinone by being a component of the aerobic respiratory chain (Fig 1). In the branched, reductive TCA pathway fumarate reductase (FRD), encoded by frdABCD, catalyzes the reduction of fumarate to succinate and also can function in the terminal step of electron transfer to fumarate during anaerobic respiration when fumarate is available (Fig 1). Previously, disruption of SDH resulted in a significant fitness defect during experimental UTI [1, 17] and loss of FRD did not result in a fitness defect [5] suggesting that the oxidative TCA cycle is required for UTI while the reductive TCA pathway is dispensable [1]. Because only one of the two oxidative fumarase isozymes (FumC) was required for UPEC fitness, we decided to more thoroughly investigate the roles of oxidative and reductive respiration during experimental UTI.

In vivo co-challenge between single mutant strain sdhB and single mutant strain frdA resulted in a fitness defect of the sdhB mutant in both the bladder and kidneys (P = 0.0004 and P = 0.0413), providing additional evidence that the oxidative TCA cycle is crucial during UTI (Fig 3A). Interestingly, loss of FRD in the sdhB mutant background (sdhBfrdA double mutant strain) ameliorates the fitness defect observed for the mutant sdhB strain [1] and generates a fitness advantage in the bladder (P = 0.0105) (Fig 3B). These findings indicate that FRD, in the absence of SDH, creates a fitness defect in vivo. Indeed, when an in vivo co-challenge was performed between the sdhBfrdA double mutant and single mutant strain frdA no significant difference was observed between the strains (Fig 3C). To confirm that the fitness advantage of the sdhBfrdA double mutant strain was due to the loss of FRD, the frdABCD operon was expressed in the sdhBfrdA double mutant strain and the fitness advantage was lost and, moreover, complementation produced a fitness defect during UTI in the bladder and kidneys (P = 0.002 and P = 0.0117, respectively) (Fig 3D) reminiscent of the fitness defect observed in the sdhB single mutant. These experiments suggest that SDH may be important for UPEC UTI, but the presence of FRD has a more significant detrimental effect during infection. However, the fitness defect observed for the sdhB mutant strain during co-infection with frdA mutant strain, suggests that succinate oxidation is conditionally beneficial to infection when functional FRD is present. Only when the sdhBfrdA mutant strain is complemented with FRD do we observe that fumarate reduction is detrimental during infection in the absence of succinate oxidation. Here we propose that FRD is detrimental to UPEC during UTI but can only be observed in the absence of SDH.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Presence of FRD in the absence of SDH, creates a fitness defect in vivo.

In vivo Log₁₀ competitive indices (CI) were determined following 48 h co-challenge infections in female CBA/J mice with a 1:1 ratio of (A) single mutant strain sdhB and single mutant strain frdA, (B) double mutant sdhBfrdA and wild-type CFT073, (C) double mutant strain sdhBfrdA and single mutant strain frdA and (D) double mutant strain sdhBfrdA complemented with pGEN-frdABCD and wild-type CFT073 containing pGEN-MCS. Each symbol represents bladder (closed circle) and kidneys (open square) from an individual mouse. Log₁₀ CI>1 indicates a fitness advantage and Log₁₀ CI<1 indicates a fitness defect of the mutant. Significant differences in colonization (*P<0.05, **P<0.01, ***P<0.001) were determined with the Wilcoxon signed-rank test.

https://doi.org/10.1371/journal.ppat.1008382.g003

The presence of at least one fumarase enzyme is required for in vitro E. coli growth

To determine if the fumarase enzymes are required for growth in vitro, growth of single mutant strains fumA and fumB, double mutant strains fumAB, fumAC, and fumBC, and triple mutant strain fumABC were measured in LB medium, defined medium containing either 0.2% glycerol or 0.2% glucose as the sole carbon source, and human urine which mimics the milieu of the host urinary tract. Only triple mutant strain fumABC displayed a minor growth defect in LB medium (Fig 4A), a slightly more drastic growth defect in defined medium containing 0.2% glycerol (Fig 4B), a modest growth defect in defined medium containing 0.2% glucose (Fig 4C), as well as an observable growth defect in human urine (Fig 4D) compared to wild-type CFT073. These observations suggested that the presence of at least one fumarase enzyme is required for growth in multiple aerobic media conditions. The in vitro growth defect of triple mutant strain fumABC was restored to the wild-type phenotype when complemented with pGEN-fumC and cultured in LB medium as well as defined medium containing 0.2% glycerol or 0.2% glucose (S1A, S1B and S1C Fig respectively). The poor in vitro growth observed for the triple mutant strain fumABC may contribute to the in vivo fitness defect observed during infection (Fig 2D); however, the single and double mutant strains with disruption in fumC which also resulted in in vivo fitness defects (Fig 2B, 2C and 2E) displayed no deficiency in any of the growth mediums tested. Therefore, generalized growth defects do not demonstrate a potential mechanism for loss of in vivo fitness for mutants lacking fumC.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Optimal E. coli growth in vitro requires the presence of at least one fumarase.

Growth of wild-type CFT073, fumarase single mutant strains fumA, fumB, and fumC, double mutant strains fumAB, fumAC, fumBC and triple mutant strain fumABC in (A) LB medium, (B) defined medium containing 0.2% glycerol as the sole carbon source (C) defined medium containing 0.2% glucose as the sole carbon source, and (D) human urine. OD₆₀₀ values were recorded each hour and the mean of three independent trails is plotted.

https://doi.org/10.1371/journal.ppat.1008382.g004

SDH is required for in vitro growth of E. coli

To determine the roles of SDH and FRD in vitro, the single mutant strains, sdhB and frdA, and double mutant strain, sdhBfrdA, were cultured in LB medium, defined medium containing either 0.2% glycerol or 0.2% glucose as the sole carbon source, and human urine. Strains that were unable to express SDH; single mutant strain sdhB and double mutant strain sdhBfrdA, had a very slight growth defect in LB medium compared to wild-type CFT073 (Fig 5A). Single mutant strain sdhB was also observed to have a slight growth defect in defined medium containing 0.2% glycerol while the double mutant strain sdhBfrdA had the most pronounced defect (Fig 5B). Both growth defects in LB medium and defined medium containing 0.2% glycerol were complemented with expression of sdhB in the single mutant sdhB and double mutant sdhBfrdA (S2A and S2B Fig, respectively). All mutant strains had a similar growth phenotype as observed for the wild-type strain in defined medium containing 0.2% glucose (Fig 5C). The single mutant strain sdhB and the double mutant strain sdhBfrdA both had a growth defect in human urine (Fig 5D) similar to the defect observed in LB medium (Fig 5A). These results are not surprising since LB medium and human urine both consist of peptides and amino acids. While the in vivo fitness defect of the single mutant strain sdhB (Fig 3A) can be partially explained by the growth defects observed in vitro, the loss of both sdhB and frdA in the double mutant strain was found to have no fitness defect in vivo (Fig 3B) and cannot explain its poor ability to grow in vitro. These findings suggest that under the conditions tested, which were aerobic, SDH is required for optimal growth of E. coli in vitro.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. In vitro growth for E. coli SDH and FRD mutant strains.

Growth of wild-type CFT073 and succinate dehydrogenase mutant strain sdhB, fumarate reductase mutant strain frdA, and double mutant strain sdhBfrdA in (A) LB medium, (B) defined medium containing 0.2% glycerol as the sole carbon source, (C) defined medium containing 0.2% glucose as the sole carbon source, and (D) human urine. Mean OD₆₀₀ values determined at each hour time point are plotted against time.

https://doi.org/10.1371/journal.ppat.1008382.g005

Loss of FumC results in delayed growth during iron-limitation

Because the urinary tract is an iron-limited environment and iron is essential for UPEC pathogenesis, we sought to determine the effect of disruptions within the oxidative and reductive TCA pathways during iron-limitation in vitro. Previously, iron availability in E. coli has been shown to be important for expression of respiratory genes [26]. During iron-limitation, expression of aerobic respiratory genes were found to be upregulated while expression of anaerobic respiratory genes including frdABCD was downregulated [26]. Iron control has not been observed for SDH [27]. Classified as a class II enzyme, FumC, lacks an iron-sulfur cluster and does not require iron for its activity [28] while class I enzymes, FumA and FumB, are iron-dependent hydrolases [29]. During iron-restricted conditions fumC was observed to have increased expression as fumA maintained constant levels of expression suggesting a functional FumA requires iron for assembly of the 4Fe-4S iron-sulfur center [30]. Because FumA and FumC both carry out the same reaction within the oxidative TCA cycle, it has been proposed that FumC acts as an alternative enzyme substituting for FumA during iron-limitation and oxidative stress [30].

To mimic the iron-limited environment of the urinary tract, all respiration mutant strains were cultured in iron-free minimal medium with 0.2% glucose as the carbon source, with and without 36 μM FeCl₃. When single mutant strains fumA, fumB, and fumC were grown in iron-limited minimal medium, as expected, only the fumC mutant strain possessed a lag in growth (Fig 6A). With addition of iron in the medium the growth of each strain was improved compared to growth in iron-free conditions; however, the fumC mutant still possessed a lag in growth during iron conditions yet was able to reach a final OD saturation similar to wild-type (Fig 6A). Both single mutant strains fumA and fumB had no defects in either minimal media condition (Fig 6A). Therefore, iron can partially restore the growth defect of the fumC mutant strain to wild-type levels suggesting that iron-limitation is the basis for the required presence of FumC for optimal growth. Consistent with this notion, the double mutant strain fumAB, which has a functional FumC that does not require iron, grew similarly to wild-type in both conditions (Fig 6B). Double mutants lacking FumC (fumAC and fumBC) resulted a slight lag during exponential growth in iron-free minimal medium and when iron was provided (Fig 6B). The triple mutant fumABC was completely crippled in growth during all tested conditions, even with iron present (Fig 6B). In medium without iron, single mutant sdhB and double mutant sdhBfrdA grew slightly better than the wild-type; however, when iron was added to the medium, these mutants displayed a slight lag in growth rate compared to wild-type CFT073 and single mutant frdA (Fig 6C). These data may be explained by previous reports referenced above, which demonstrated the down-regulation of anaerobic respiratory genes and the up-regulation of aerobic respiratory genes during iron-limitation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. The growth defect of fumC can be rescued with addition of exogenous iron.

Wild-type CFT073 and (A) single fumarase mutant strains fumA, fumB, and fumC, (B) double and triple fumarase mutant strains fumAB, fumAC, fumBC, fumABC and (C) SDH and FRD mutant strains were examined in iron-free defined medium containing 0.2% glucose with and without 36 μM FeCl₃. OD₆₀₀ was recorded every hour and symbols indicate the mean values of three trials. Black symbols indicate the absence of iron and white symbols denote medium containing iron.

https://doi.org/10.1371/journal.ppat.1008382.g006

To confirm that FumA is functioning as the preferred fumarase enzyme, qPCR analysis of wild-type CFT073 and the single fumarase mutants were examined during iron-replete and iron-deplete conditions under aerobic growth. As expected, in wild-type CFT073 fumA was upregulated while fumC was downregulated during iron-replete compared to iron-deplete conditions (S3 Fig) indicating that FumA is the main enzyme in growth conditions with iron present. In fact, fumA expression was trending up in every mutant strain and likewise, fumC was downregulated (S3 Fig). These findings confirm that in aerobic growth conditions with iron available, FumA is the preferred enzyme for converting fumarate to malate in the oxidative TCA cycle, while FumC is preferred under iron-limitation.

Disruption of SDH and the loss of FumC results in the excretion of acidic end products

To determine if the oxidative and reductive TCA pathway mutant strains behave as expected during aerobic and anaerobic conditions, phenol red in LB-1.5% agar was used as a pH indicator to test for the excretion of acidic fermentation end products. In this phenotypic assessment a change in pH, which is dependent on proton concentration, is indicated by a color change of the phenol red. Neutral pH on phenol red agar is orange, alkaline pH on phenol red agar is pink, and acidic pH on phenol red agar is yellow. Under anaerobic and aerobic conditions, the oxidative and reductive TCA pathway mutant strains were examined on plain phenol red agar, phenol red agar containing 0.2% glucose or 0.2% glycerol to mimic fermentative conditions (during anaerobic conditions), and phenol red agar containing 0.2% glucose or 0.2% glycerol with the addition of 0.2% fumarate to serve as a final electron acceptor for anaerobic respiration (during anaerobic conditions).

Following incubation under anaerobic conditions at 37°C on phenol red agar alone, wild-type CFT073 and all of the fumarase mutants turned orange indicating a neutral pH (Fig 7A). During anaerobic conditions on phenol red agar containing glucose or glycerol as the sole carbon source, wild-type CFT073 and the fumarase mutants were observed to be yellow indicating an acidic pH, suggesting that these strains are capable of carrying out fermentation. Exogenous fumarate in the phenol red agar containing either glucose or glycerol also resulted in a yellow color for wild-type and all of the mutant strains (Fig 7A). Interestingly, transitioning the phenol red agar plates to aerobic conditions at 25°C, resulted in only the fumarase mutants with disruptions in fumC (mutant strains fumC, fumAC, fumBC, fumABC) to remain yellow and acidic on the phenol red agar containing glucose alone or glucose with the addition of exogenous fumarate (Fig 7A). Wild-type CFT073, single fumarase mutant strains, fumA and fumB, as well as double mutant strain fumAB became alkaline presenting as a pink color on the phenol red agar for all conditions(Fig 7A). Surprisingly, only the triple mutant strain fumABC turned yellow following aerobic incubation on phenol red agar containing glycerol alone or phenol red agar containing glycerol and exogenous fumarate (Fig 7A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Loss of FumC results in acidic end products.

Wild-type CFT073, single, double, and triple fumarase mutants were examined on LB agar containing phenol red (plain), LB agar containing phenol red and 0.2% glucose, LB agar containing phenol red and 0.2% glycerol, LB agar containing phenol red, 0.2% glucose, and 0.2% fumarate, and LB agar containing phenol red, 0.2% glycerol, and 0.2% fumarate. Un-inoculated phenol red agar plates with orientation of strains are located on the right and were treated under the same conditions. Alkaline pH is pink, neutral pH is orange, and acidic pH is yellow. (A) Plates were spotted with bacterial culture, incubated under anaerobic conditions at 37°C for 24 h, imaged, and then transitioned to aerobic conditions at 25°C and imaged at 6 h and 24 h. (B) Plates were spotted with bacterial culture, incubated under aerobic conditions at 37°C, imaged at 6 and 24 h, transitioned to 25°C and imaged at 48 h.

https://doi.org/10.1371/journal.ppat.1008382.g007

Interestingly, when the strains were examined on plain phenol red agar under aerobic conditions at 37°C following a 6 h incubation, only the triple mutant strain fumABC was orange compared to wild-type CFT073 and the other single and double fumarase mutant strains which were pink (Fig 7B). Under aerobic conditions over the span of 6–48 h we noted that all of the phenol red agar conditions tested (Fig 7B) had similar results to the phenol red agar plates examined during anaerobic conditions and transitioned to aerobic conditions (Fig 7A). Fumarase mutant stains deficient in fumC maintained acidity on phenol red agar containing glucose and glucose with the addition of fumarate (Fig 7A and 7B). These conditions with glucose also resulted in the fumarase mutant strains deficient in fumC having reduced growth compared to wild-type CFT073 (Fig 7B). On phenol red agar containing glycerol or glycerol with the addition of fumarate, only the triple mutant strain fumABC remained acidic (Fig 7A and 7B).

Following anaerobic incubation at 37°C, wild-type CFT073 and the SDH and FRD mutants displayed an orange hue on phenol red agar alone, while the addition of glucose or glycerol to the phenol red agar caused the strains to turn yellow (Fig 8A). Addition of fumarate as a final electron acceptor to the phenol red agar containing glucose or glycerol also resulted in a yellow color to be observed (glucose may result in a more intense yellow color than glycerol) (Fig 8A). After transitioning the phenol red agar plates to aerobic conditions for 24 h at 25°C, all of the strains turned pink on phenol red agar alone as expected. Interestingly, single mutant strain sdhB and double mutant strain sdhBfrdA turned yellow on phenol red agar containing either carbon source with or without the addition of fumarate while wild-type CFT073 and the frdA mutant strain turned orange on phenol red agar containing carbon source alone to slight pink on phenol red agar with addition of a carbon source and fumarate (Fig 8A). These observations are not surprising since the sdhB and sdhBfrdA mutant strains are unable to operate an oxidative TCA cycle to carry out oxidative respiration, thus use of the exogenous fumarate as an electron acceptor during anaerobic conditions would result in a yellow color as was observed.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Disruption of SDH results in acidic end products.

Wild-type CFT073, single mutant strains, sdhB and frdA, and double mutant strain sdhBfrdA were examined on LB agar containing phenol red (plain), LB agar containing phenol red and 0.2% glucose, LB agar containing phenol red and 0.2% glycerol, LB agar containing phenol red, 0.2% glucose, and 0.2% fumarate, and LB agar containing phenol red, 0.2% glycerol, and 0.2% fumarate. Un-inoculated phenol red agar plates with orientation of strains are located on the right and treated under the same conditions. Alkaline pH is pink, neutral pH is orange, and acidic pH is yellow. (A) Plates were spotted with bacterial culture, incubated under anaerobic conditions at 37°C for 24 h, imaged, and then transitioned to aerobic conditions at 25°C and imaged at 6 h and 24 h. (B) Plates were spotted with bacterial culture, incubated under aerobic conditions at 37°C, imaged at 6 h and 24 h, transitioned to 25°C and imaged at 48 h.

https://doi.org/10.1371/journal.ppat.1008382.g008

Following incubation of inoculated phenol red agar plates under aerobic conditions over the span of 6–48 h (Fig 8B) we noted that the strains looked similar to those inoculated on the phenol red agar plates examined during anaerobic conditions transitioned to aerobic conditions (Fig 8A). We also observed that the sdhB and sdhBfrdA mutant strains did not grow as well as wild-type CFT073 and the frdA mutant strain (Fig 8B). SDH mutant strains maintained acidity on phenol red agar containing carbon sources, glucose or glycerol, and also on phenol red agar containing a carbon source with the addition of fumarate (Fig 8A and 8B). These findings suggest that disruptions within enzymes, FumABC and SdhBFrdA that function in both the oxidative and reductive TCA pathways cause acidic end products to be excreted under anaerobic as well as aerobic conditions. Loss of oxidative TCA cycle enzyme SdhB alone also results in acidic end products following anaerobic and aerobic conditions.

Disruption of both the oxidative and reductive TCA pathway results in sensitivity to oxidative and acid stress

During respiration the reduced electron carriers NADH and FADH₂ formed during the oxidative reactions of the TCA cycle ultimately transfer electrons to oxygen as a final electron acceptor providing a proton gradient for the cell to produce energy by oxidative phosphorylation. Despite this aerobic respiratory chain being highly efficient in suppling energy and generating proton motive force, damaging reactive oxygen species, such as superoxide (O₂^-), hydrogen peroxide (H₂O₂), and hydroxyl radicals, can also be produced resulting in oxidative damage to the cell. These reactive oxygen species oxidize and inactivate [4Fe-4S] cluster-containing dehydratases such as FumA and FumB [28, 31–33]. As a countermeasure against oxidative stress, cellular defense mechanisms including superoxide dismutases (SODs) help convert the hydroxyl radicals to O₂ and H₂O₂ which in turn induces the oxidative stress regulator, OxyR, to activate expression of a cohort of scavenging and repair enzymes. FumC is a known member of the superoxide response (soxRS) regulon [34] and lacks an iron-sulfur cluster making it oxygen-stable and unaffected by oxidative stress compared to its labile FumA and FumB counterparts.

We reasoned that the viability of the oxidative and reductive TCA pathway mutant strains would be decreased during oxidative stress due to their presumed partial defects in cellular respiration. To determine a suitable bactericidal concentration of H₂O₂ with at least three logs of killing of wild-type CFT073, various concentrations of H₂O₂ were examined over a 60-minute time period. For this pilot experiment wild-type CFT073 was inoculated into LB medium containing 0.0%, 0.03%, 0.3% and 3% H₂O₂ and sample was removed every 10 minutes for 1 hour, diluted, and drip-plated on plain agar to determine colony forming units (CFU). We observed optimal killing of wild-type in LB medium containing 0.3% H₂O₂ (S4A Fig), defined by a 3-log decrease in CFU, and examined the bactericidal effects of H₂O₂ at this concentration for all of the oxidative and reductive TCA pathway mutants. To control for any potential growth of the mutants during this assay, we incubated all TCA pathway mutants in LB medium containing 0% H₂O₂ which resulted in no statistically significant difference in CFU between wild-type CFT073 during a 30-minute incubation (S4B Fig). Triple mutant strain fumABC was observed to have the greatest susceptibility to 0.3% H₂O₂ at 20, 30, and 40 minutes post-incubation with a statistically significant decrease in CFU compared to wild-type CFT073 (P< 0.0001) (Fig 9A). This trend of susceptibility to 0.3% H₂O₂ by the fumABC triple mutant was observed until 40 minutes post-incubation at which time detectible viability was eliminated (Fig 9A). Here we see that complete loss of both the oxidative and reductive TCA pathways in which the reversible reaction between fumarate and malate (fumABC) was disrupted results in an increased sensitivity to in vitro-induced exogenous oxidative stress. At 40 minutes post-incubation, the sdhB mutant was also found to have a significantly decreased CFU compared to CFT073 (P = 0.0003) (Fig 9A). All mutant strains tested lost viability 10 minutes earlier than the wild-type, indicating that intact cellular respiration capabilities improve overall survival to exogenous sources of oxidative stress. Our inability to observe minor differences of killing between the mutant strains may be due to the lack of this assay’s sensitivity to distinguish the susceptibility of the mutants to H₂O₂ during this assay.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Disruption of both the oxidative and reductive TCA pathway results in sensitivity to oxidative stress.

(A) Wild-type CFT073 and the oxidative and reductive TCA pathway mutant strains were examined in LB medium containing 0.3% H₂O₂. Samples were collected every 10 minutes, diluted, and plated on LB agar to determine CFU. Colored symbols indicate the mean of three independent biological replicates. Two-way ANOVA was use to assess statistical differences at each time point (***P<0.001, ****P<0.0001). (B) Wild-type CFT073 and the oxidative and reductive TCA pathway mutant strains were incubated in 100 mM MES buffered LB, pH 2.5, for 60 minutes, diluted, and plated on LB agar to determine CFU. The dashed line represents the intended input CFU/ml at time zero. The mean is indicated by the bar height and the error bars display SEM of three biological replicates.

https://doi.org/10.1371/journal.ppat.1008382.g009

During acid stress, the bacterial cell undergoes physiological, metabolic, as well as proton-consuming acid resistance mechanisms that reduces influx of protons into the cell. Under normal growth conditions, proton motive force is generated by the coupling of metabolic redox reactions with the export of protons. However, during aerobic growth under mild acid stress (pH 5.0–7.0) electron transport chain components including cytochrome oxidase (cyo) genes, NADH dehydrogenase II (ndh) genes, succinate dehydrogenase (sdh), and NADH dehydrogenase I (nuo) genes have been found to be upregulated [35, 36]. This enables the cell to counteract intracellular increases in pH by exporting protons. To test the viability of TCA cycle mutant strains in acidic conditions, we cultured bacteria in LB buffered to pH 7, pH 5, and pH 2.5 for one hour before enumerating CFU. At pH 7 and pH 5, none of the mutants displayed significantly altered CFU compared to wild-type CFT073 (S5A and S5B Fig, respectively). However, when grown at pH 2.5, the triple fumABC mutant had significantly reduced CFU (P<0.0001), which indicates an increased susceptibility to acid-induced stress (Fig 9B). The sdhB mutant also had reduced CFU, although this was not statistically significant (Fig 9B). Collectively, these in vitro experiments suggest that the fumABC and sdhB mutants are more susceptible to exogenous radicals and acid stress.

Growth in a variety of media, including in human urine, was not sufficient to explain the negative impact of fumC mutations during UTI. Similarly, it was not possible to invoke growth defects to explain the conditional and detrimental impact of FRD on UPEC fitness. Therefore, we sought to determine if the increased susceptibility to radicals and acidic pH in TCA cycle mutant strains might provide explanation for their observed fitness defects in vivo. Because these in vitro experiments partially mimic bactericidal mechanisms deployed by neutrophils, we conducted pharmacological neutrophil depletion experiments in mice prior to performing co-challenge. If this is indeed the underling mechanism contributing to fitness defects, we predicted that neutrophil depletion using cyclophosphamide would rescue the in vivo phenotypes observed for the mutant strains fumC and sdhB. While blood tests indicated that the pharmacological depletion of neutrophils was successful and we observed a 3–4 log increase in total CFUs comparing cyclophosphamide- to PBS-treated mice (S6A and S6C Fig), the observed fitness defects in fumC and sdhB were not reversed by neutrophil depletion (S6B and S6D Fig). Howeverr, the analysis of blood samples did indicate that in addition to neutropenia, the mice were also anemic. It remains possible that the anemia is contributing to the increased fitness defect (P = 0.0101) observed for the fumC mutant in the bladders of cyclophosphamide-treated mice (S6B Fig). These results further support the notion that iron availability in vivo could be a substantial contributing factor to the preferential utilization of FumC in the urinary tract.

Absence of FumC facilitates resistance to bactericidal antibiotics

Currently, the treatment for UTI is a course of antibiotics, most commonly trimethoprim or ciprofloxacin for more serious cases. These two antibiotics have different methods for inhibiting infection, by deploying bacteriostatic or bactericidal mechanisms. Bacteriostatic antibiotics prevent bacterial growth without affecting viability, and in the clinical setting, would allow the host immune system to clear the infection. In contrast, bactericidal antibiotics effectively eliminate the viability of bacteria and have even been proposed to produce these lethal effects in part by hyperactivation of the respiratory chain and concomitant production of reactive species [37]. To better understand the role for the TCA cycle in E. coli susceptibility to clinically relevant antibiotics, all mutants and wild-type CFT073 were cultured in LB medium containing a single antibiotic. Three bacteriostatic antibiotics were chosen for these experiments: chloramphenicol, trimethoprim, and tetracycline. Strains were inoculated at ~10⁷ CFU and incubated in 10 μg/ml chloramphenicol for 180 minutes. The growth of all strains was similarly stunted by the presence of chloramphenicol (Fig 10A). In 5 μg/ml of trimethoprim, sdhBfrdA was the only strain differentially affected; however, the decrease in CFU was minimal and not significant (Fig 10B). CFU of the triple fumABC mutant strain was decreased by 2.5-logs over the 180-minute incubation with 10 μg/ml tetracycline (Fig 10C). Although not statistically significant, this trend may indicate that absence of all fumarase activity is an important compensation mechanism when protein translation is inhibited.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. Absence of FumC leads to decreased susceptibility to bactericidal antibiotics.

Wild-type CFT073 and the oxidative and reductive TCA pathway mutant strains were examined in LB medium containing (A) chloramphenicol 10 μg/ml (B) trimethoprim 5 μg/ml or (C) tetracycline 10 μg/ml. Samples were collected over the course of 180 minutes, diluted and plated on LB agar to determine CFU. Colored dots signify the mean of three biological replicates. The Minimum Bacterial Concentration (MBC) of (D) ciprofloxacin 0.002–32 μg/ml (E) ampicillin 0.016–256 μg/ml or (F) streptomycin 0.064–24 μg/ml observed for wild-type CFT073 and the mutant strains were recorded following overnight incubation on Mueller-Hinton agar with antibiotic test strips. The values shown are averages from two independent trials.

https://doi.org/10.1371/journal.ppat.1008382.g010

Bactericidal antibiotics are effective at killing bacteria by inhibiting essential cellular mechanisms such as cell wall synthesis, protein translation, and DNA replication. We performed two in vitro methods; a liquid LB medium assay containing antibiotic and a Mueller-Hinton agar-based assay using antibiotic test strips to examine the bactericidal effects of ciprofloxacin, ampicillin, and streptomycin on the respiration mutants. Bacteria cultured with 250 ng/ml ciprofloxacin in LB medium for 3 hours displayed up to 4-logs loss of CFU (S7A Fig). Both the frdA and fumAC mutant strains were more resistant to ciprofloxacin than wild-type CFT073, displaying no decrease in CFU over time (S7A Fig). The sdhB and sdhBfrdA mutant strains both had a 2-log decrease in CFU within 45 minutes, indicating increased susceptibility to ciprofloxacin. The minimum bactericidal concentration (MBC) was determined to be 0.047 μg/ml for the triple mutant strain fumABC, compared to 0.010 μg/ml for wild-type CFT073 (Fig 10D). This suggests that loss of fumarase activity increases tolerance to bactericidal antibiotics targeting DNA gyrase. When incubated in LB medium with 5 μg/ml ampicillin, every mutant strain displayed a higher final CFU compared to CFT073 (S7B Fig). The CFU of mutant strains fumABC and fumBC increased in the presence of ampicillin, demonstrating no effect of the antibiotic during 3 hours of incubation, unlike wild-type CFT073 which decreased by 2.5-logs (S7B Fig). Similarly, results from agar plates with an antibiotic test strip indicated that the fumAC, fumBC, and fumAB mutant strains were the least susceptible to ampicillin (Fig 10E). However, unlike the results obtained from LB medium containing ampicillin, mutant strains sdhB and sdhBfrdA had a lower MBC than wild-type CFT073 (Figs 10E and S7B). The most striking bactericidal effect was with the use of streptomycin, which inhibits protein synthesis. In liquid culture, the sdhB and sdhBfrdA mutant strains had a 5.5-log decrease in CFU, wild-type had a 4-log decrease in CFU, and the triple mutant fumABC displayed an increase in CFU over time (S7C Fig). Interestingly, the four most resistant strains to 5 μg/ml streptomycin in the liquid assay were those lacking FumC. These results were consistent with the findings that mutants lacking fumC examined on the Mueller-Hinton agar with a streptomycin test strip demonstrated decreased susceptibility (Fig 10F). These findings are consistent with the notion that certain bactericidal antibiotics depend upon either a robust electrochemical gradient across the cell membrane, generated during maximal re-oxidation of reduced NADH to NAD⁺, or even that bactericidal antibiotics cause hyper-activation of membrane respiration and formation of reactive oxygen species within the cell [37] since many of the tested TCA cycle mutants are more resistant to these bactericidal antibiotics than the parental strain.

To better understand these results, we performed identical MBC determinations under anaerobic conditions. We observed a general trend toward reduced susceptibility to bactericidal antibiotics across all of the strains tested (S8 Fig). The results for ampicillin and ciprofloxacin under anaerobiosis were similar to what was observed under aerobic conditions, where double and triple mutants containing a disruption in fumC were 2- or 3-fold more resistant than the parental strain (S8A and S8B Fig). As expected, we observed a dramatic increase in resistance to the aminoglycoside streptomycin under anaerobic conditions where the parental CFT073 MBC shifted over 10-fold from 4 μg/ml to 48 μg/ml (Figs 10F and S8C). This indicates that the decreased PMF that occurs during anaerobic growth likely results in decreased uptake of the aminoglycoside and reduced susceptibility, which was expected because it is known that aminoglycosides are ineffective toward anaerobic microbes. However, similar to what was observed under aerobic conditions, any strain lacking FumC was more resistant to streptomycin than the parental strain; fumC 79 μg/ml, fumAC 128 μg/ml, fumBC 112 μg/ml, and fumABC 128 μg/ml (S8C Fig). While far from conclusive, the collective results from the MBC experiments demonstrate some unappreciated linkage between the TCA cycle and the mechanism underlying susceptibility to these bactericidal antibiotics.

E. coli motility is dependent on FumC

Bacterial swimming motility is an easily quantified phenotype that can be used to indirectly assess the electrochemical gradient across the cell membrane. Flagella rotation is dependent on proton motive force (PMF), or specifically the H⁺ gradient created by membrane respiration during the oxidation of the reduced electron carrier NADH produced by the oxidative TCA cycle. During this process electrons are passed through an electron transport chain and protons are deposited across the cytoplasmic membrane. The movement of H⁺ down the chemiosmotic gradient, through the flagellum stator drives the rotation of flagella. Because flagella-mediated motility is required for colonization of the urinary tract and contributes to UPEC virulence [38] we sought to determine if the fitness defects of the oxidative and reductive TCA pathway mutant strains during UTI was due to a deficiency in motility. To address this, mutant strains were spotted onto 0.25% soft agar and their swimming phenotypes were compared to that of wild-type CFT073. Only double mutant strain sdhBfrdA had a hypermotile swimming phenotype with a diameter of 58.75 mm compared to the swimming diameter, 36.42 mm, of wild-type CFT073 (P = 0.0001) (Fig 11A and 11B). Interestingly, only mutants with a disruption within one of the fumarase encoding genes, fumC, had a hypomotile swimming phenotype compared to the swimming phenotype of wild-type CFT073. Single mutant strain fumC, double mutant strains fumAC and fumBC, and triple mutant strain fumABC were significantly decreased with swimming diameters of 7.53 mm, 6.59 mm, 8.25 mm, and 7.61 mm, respectively (P<0.0001) (Fig 11A and 11B). These findings suggest that the in vivo fitness defects observed for the oxidative and reductive TCA pathway mutant strains with disruptions in FumC may be due to impaired motility influencing poor UPEC colonization within the urinary tract.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 11. FumC is required for motility in CFT073.

(A) Wild-type CFT073 and each oxidative and reductive TCA pathway mutant strain was stabbed into semi-soft motility agar and incubated overnight at 30°C for 16 h. Images of representative motility agar plates following incubation with inoculated strains. Outer swim edge is indicated by a white line. (B) Bars represent the average swimming diameters (in mm) of biological triplicates. Error bars represent the standard errors of the means (SEM). Statistical significance was determined by one-way ANOVA (***P <0.001, ****P <0.0001). (C) The invertible element (IE) assay was performed on strains grown statically in LB at 37°C for 18 h and standardized to OD₆₀₀ = 0.5. Bars depict percent population of bacteria with the fim promoter within the IE positioned in the on orientation (black bar) or off orientation (white bar) as calculated with pixel density values for bands of PCR products in (Fig 9A).

https://doi.org/10.1371/journal.ppat.1008382.g011

It has been previously reported that flagellar expression and motility is directly affected by the overexpression of type 1 fimbriae (fim) whose promoter is located within an invertible element (IE) that is capable of switching from an on to an off orientation and vice versa [39, 40]. While decreased motility and flagellar expression can result from constitutive expression of the fim operon, increased motility is not indicative of decreased fim expression [39]. We wanted to better understand if this reciprocal relationship could explain the motility findings for strains lacking fumC or for the sdhBfrdA double mutant. We found that the fim promoter was in the OFF-orientation for 92%-100% of the bacterial population for the majority of the respiration mutant strains; single mutant strains sdhB and fumC, double mutant strains sdhBfrdA, fumAB, fumAC, fumBC, and triple mutant strain fumABC (Figs 11C and S9A). These findings suggest that the decreased motility of the single, double, and triple mutant strains with a disruption in fumC is not due to overexpression of fim since the IE of 92%-100% of the bacterial population was in the OFF-orientation. Likewise, the hypermotile swimming phenotype of double mutant strain sdhBfrdA cannot be explained by the orientation of the IE since 97% of the bacterial population was in the OFF-orientation. There is a moderate relationship (R² = 0.87) between motility and the percent of bacteria population with the fim promoter in the ON-orientation within the fumarase mutant strains; however, this correlation is the inverse of expected results (S9B Fig). Interestingly, there is also a moderate relationship (R² = 0.70) between motility and the percent of bacteria population with the fim promoter in the ON-orientation within the SDH and FDR mutant strains (S9C Fig).

To determine if transcription is responsible for the motility phenotypes of the oxidative and reductive TCA pathway mutant strains, quantitative PCR was performed to quantify the transcription of flagellar and fimbrial genes (S10 Fig). Expression of the RNA polymerase sigma factor for the flagellar operon, fliA, and flagellin, fliC, was downregulated greater than 2-fold in the single mutant strain fumC, double mutant strains fumAB, fumAC, fumBC, and the triple mutant strain fumABC compared to wild-type CFT073 (S10A and S10B Fig). Gene expression for the pyelonephritis associated pili subunits, papA and papA2 and type one fimbriae major subunit, fimA, was also downregulated greater than 2-fold, although not statistically significant, in the single mutant fumC, double mutant strains fumAB, fumAC, fumBC, and the triple mutant strain fumABC compared to wild-type CFT073 (S10C, S10D and S10E Fig). As expected, expression of fimA for the CFT073 fim L-ON strain, in which the type 1 fim promoter is mutated in the on-orientation, and the CFT073 fim L-OFF strain, in which the type 1 fim promoter is mutated in the off-orientation was upregulated and downregulated, respectively, greater than 5-fold compared to wild-type CFT073 (S10E Fig). Up-regulation or down-regulation of flagellar and fimbrial gene expression in the SDH and FRD mutant strains was not observed (S10 Fig). These transcriptional findings suggest that it is likely a physiological mechanism, for example disruptions in the TCA affecting the proton gradient across the membrane, playing a role in the reduction of motility in the absence of fumC and the increase in motility in the absence of SDH and FRD.

Bacteria and growth conditions

E. coli CFT073 was isolated from the blood and urine of a patient with acute pyelonephritis [53]. Bacteria was routinely cultured in lysogeny broth (LB) medium (per liter; 0.5 g NaCl, 10 g tryptone, 5 g yeast extract). For in vitro growth experiments, wild-type and mutants were cultured in MOPS defined medium [54] containing either 0.2% (w/v) glucose or 0.2% (w/v) glycerol as the sole carbon source. Wild-type CFT073 and mutants were examined for growth in pooled human urine collected from at least 4 healthy females and filter sterilized. LB medium and human urine cultures were inoculated 1:100 and defined medium cultures were inoculated 1:50 from overnight LB medium bacterial cultures and incubated with aeration at 37°C. For in vitro iron-limitation experiments, bacteria grown in MOPS defined medium containing iron and 0.2% glucose were washed prior to being sub-cultured 1:50 into MOPS defined medium containing 0.2% glucose without iron. In vitro growth was examined in iron deplete and replete MOPS defined medium containing 36 μM FeCl₃. Growth curves were performed in triplicate and OD₆₀₀ was recorded every hour.

Construction of mutants

E. coli CFT073 deletion mutants (Table 1) were constructed using the lambda red recombinase system [55]. Primers homologous to sequence within the 5’ and 3’ ends of the gene to be targeted were designed to replace the gene with a nonpolar kanamycin-or chloramphenicol-resistant cassette amplified from template plasmid pKD4 or pKD3, respectively [55]. E. coli CFT073 mutants were confirmed by PCR amplification using primers flanking the target gene sequence and comparing gene product size to wild-type CFT073 PCR product size. To decipher negligible product sizes PCR reactions were digested with a restriction enzyme (New England Biolabs). The antibiotic resistance cassette of the fumC mutant was removed with plasmid pCP20 to unmark the strain for deletion of additional genes; fumA and/or fumB. Oligonucleotides used to construct and confirm the fumA, fumB, fumAB, fumAC, fumBC, fumABC, and sdhBfrdA mutant strains are listed in Table S1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Bacterial strains used in this study.

https://doi.org/10.1371/journal.ppat.1008382.t001

Complementation of mutants

To perform in vitro and in vivo complementation experiments, complementation plasmids were designed by PCR amplifying frdABCD, sdhB, and fumC from wild-type CFT073 genomic DNA using Easy-A high fidelity polymerase (Agilent), digesting with either HindIII and SacI or SphI and NotI restriction enzymes (New England Biolabs), and cloning into pGEN-MCS [56]. Oligonucleotides used for complementation construct design are listed in Table S1. The sequence of pGEN-frdABCD, pGEN-sdhB, and pGEN-fumC was confirmed by sequence analysis prior to transformation into the sdhBfrdA double mutant, sdhB single mutant, and fumABC triple mutant, respectively.

Experimental UTI

The fitness contribution of the CFT073 oxidative and reductive TCA pathway mutant strains during co-challenge competition was assessed in the CBA mouse model of ascending UTI [57, 58]. Female CBA/J mice (6–8 week old; 20 to 22 g; Jackson Laboratories) were anesthetized with ketamine/xylazine and transurethrally inoculated with a 50 μl bacterial suspension of 2 x 10⁸ CFU per mouse using a sterile polyethylene catheter (I.D. 0.28 mm x O.D. 0.61 mm) connected to an infusion pump (Harvard Apparatus). In vivo co-challenges were performed with a bacterial suspension containing a 1:1 ratio of wild-type E. coli CFT073 and CFT073 antibiotic-resistant mutant in PBS or a 1:1 ratio of two CFT073 antibiotic-resistant mutants in PBS or a 1:1:1 ratio for three CFT073 antibiotic-resistant mutants in PBS. For neutrophil depletion experiments, mice were injected via the intraperitoneal route with 100 μl of sterile PBS alone or sterile PBS containing 150 mg/kg of cyclophosphamide at 96 h and 100 mg/kg at 24 h prior to infection. Blood samples were analyzed to confirm absence of neutrophils in cyclophosphamide treated mice. Input CFU/ml was determined by plating serial dilutions (Spiral Biotech) of the inoculum onto LB agar with and without antibiotic. Urine of infected mice was collected 48 h post infection by abdominal massage and plated for CFU/ml. Following subsequent euthanization, bladder and kidneys were aseptically removed, weighed, and homogenized (OMNI International) in 3 ml PBS, and appropriate dilutions were spiral plated on LB agar with and without antibiotic to determine the output CFU/g of tissue. Viable counts were enumerated using QCount software (Spiral Biotech) and CFU from antibiotic-containing medium (mutant CFU) were subtracted from the total CFU from plates lacking antibiotic to determine the number of wild-type bacteria. For co-challenge experiments, competitive indices (CI) were calculated by dividing the ratio of the CFU of mutant to wild-type recovered from each mouse following infection by the ratio of the CFU of mutant to the CFU of wild-type present in the input. CI data were log-transformed and analyzed by the Wilcoxon signed-rank test to determine statistically significant differences in colonization (P-value <0.05). A series of Mann Whitney tests were used to analyze CI data of PBS control vs cyclophosphamide mice in each respective organ site. A CI˃1 indicates that the mutant out-competes the wild-type strain and a CI<1 indicates that the mutant is out-competed by the wild-type strain.

Anaerobic testing on phenol red

Overnight LB medium cultures of wild-type CFT073 and mutants were spotted (5 μl) onto LB agar containing 0.04 g/liter phenol red alone, phenol red and 0.2% (wt/vol) glycerol, phenol red and 0.2% (wt/vol) fumarate-0.2% (wt/vol) glycerol, phenol red and 0.2% (wt/vol) glucose, and phenol red or 0.2% (wt/vol) fumarate-0.2% (wt/vol) glucose. Plates were incubated under anaerobic (BD GasPak EZ Anaerobe) conditions at 37°C for 24 h and subsequently imaged. These plates were transitioned to aerobic conditions at 25°C and imaged at 6 h and 24 h. Strains were also spotted on phenol red agar plates for examination under aerobic conditions at 37°C and imaged at 6 h and 24 h. These plates were then transitioned to 25°C and imaged at 48 h post-inoculation.

Hydrogen peroxide and acid resistance assays

For hydrogen peroxide assays, approximately 10⁸ CFU/ml of wild-type CFT073 or mutant was inoculated into LB medium containing either 0% H₂O₂, 0.03% H₂O₂, 0.3% H₂O_2, or 3% H₂O₂. For acid resistance assays, overnight cultures were washed in phosphate buffered saline and 10⁷ CFU was inoculated into 100 mM MES buffered LB (LB-pH 2.5, LB-pH 5.0, and LB-pH 7.0). Input and output samples were collected from each assay to perform 1-fold serial dilutions in a microwell plate and a total of 10 μl from each dilution was drip plated onto LB agar. All hydrogen peroxide assays were carried out under static conditions at room temperature and samples were collected at 10, 20, 30, 40, 50 and 60 min post-inoculation and acid resistance assays were carried out at 37°C and samples were collected at 1h post-inoculation. Following incubation of LB agar plates at 37°C for 18 h viable colonies were enumerated at countable dilutions. Assays were performed in triplicate and averaged. A 2-way ANOVA was run to determine statistical difference between CFT073 and mutant strains at each time point.

Antibiotic susceptibility assay

Wild-type CFT073 and the oxidative and reductive TCA pathway mutant strains were examined in LB medium containing chloramphenicol 10 μg/ml, trimethoprim 5 μg/ml, tetracycline 10 μg/ml, streptomycin 5 μg/ml, ampicillin 5 μg/ml, ciprofloxacin 250 ng/ml. Samples were collected over the course of 180 minutes, diluted and plated on LB agar to determine CFU. Assays were performed in triplicate and averaged. For antibiotic susceptibility testing using antibiotic test strips, overnight LB medium cultures of wild-type CFT073 and all oxidative and reductive TCA pathway mutant strains were diluted 1:100 in Mueller-Hinton broth, except the fumABC triple mutant which was diluted 1:50, and grown shaking at 37°C for approximately 2–3 h to an OD₆₀₀ = 0.5. Cultures were adjusted to lowest OD with phosphate buffered saline and a total of 150 μl was spread plate onto Mueller-Hinton agar. Once dry, a Minimum Inhibitory Concentration (MIC) Test Strip (Liofilchem MTS) was placed face up in the center of the agar plate using sterile forceps. The following antibiotics were tested: streptomycin 0.064–24 μg/ml, ampicillin 0.016–256 μg/ml, and ciprofloxacin 0.002–32 μg/ml. Input CFU was quantitated by performing 1-fold serial dilutions in a microwell plate and a total of 10 μl from each dilution was drip plated onto LB agar. Plates were incubated at 37°C for 18 h aerobically or for 24h in an anaerobic chamber and the MIC on the strip where inhibited growth was observed was recorded. Assays were performed in duplicate and averaged.

qPCR

For fimbrial and flagellar gene expression studies, overnight LB medium cultures of wild-type CFT073, oxidative and reductive TCA pathway mutant strains, type 1 fimbriae mutants fim L-ON and fim L-OFF [59] were back diluted (1:100) and grown shaking in LB medium at 37°C for approximately 2–3 h to an OD₆₀₀ = 0.5–0.6. For RNA stabilization a 2:1 ratio (v:v) of RNA protect (Qiagen) was added to 250 μl of bacterial sample and spun down (Thermo Scientific, Legend Micro 21, RT, 5 min, 9,600 x g). For gene expression of fumarase genes during iron replete and deplete conditions, samples were collected from MOPS defined medium cultures with and without 36 μM FeCl₃ grown at 37°C for approximately 4 h to an OD₆₀₀ = 0.5–0.6. For RNA stabilization a 2:1 ratio (v:v) of RNA protect (Qiagen) was added to 100 μl of bacterial sample and spun down (Thermo Scientific, Legend Micro 21, RT, 5 min, 9,600 x g). Subsequent RNA isolation was performed with the RNeasy Mini kit (Qiagen) followed by DNAse-treatment of RNA with the TURBO DNA-free kit (Invitrogen). PCR with primers to amplify housekeeping gene, glyceraldehyde-3-phosphate dehydrogenase (gapA), was performed to check for contaminating DNA. cDNA synthesis was performed on DNA-free RNA samples with the iScript kit (Bio-Rad). 5 ng of total cDNA was added to each reaction, qPCR was performed with SyberGreen PowerUp reagent (Invitrogen) with primers to amplify the following genes, fliA; RNA polymerase sigma factor for flagellar operon, fliC; flagellin assembly, fimA; type 1 fimbrial protein, papA and papA2; major pillin subunit A, and fumA, fumB, fumC; fumarase. gapA expression was used as the internal control to calculate relative expression values of experimental genes. The comparative threshold cycle (C_T) method was used to determine the relative fold-change compared to wild-type CFT073. Oligonucleotides used in the qPCR experiments are listed in Table S1.

Motility assays

The swimming motility of wild-type E. coli CFT073 and mutants was observed on soft agar plates (10 g/L tryptone, 5 g/L NaCl, and 2.5 g/L Bacto agar). Overnight bacterial cultures grown shaking at 37°C were normalized to an OD₆₀₀ = 1.0 in tryptone broth (10 g/L) and using a sterile inoculating needle stabbed into the center of the motility agar and incubated at 30°C for 16 h. The diameter of swimming was measured for each strain. Experiments were performed in duplicate and repeated for a total of three biological replicates. Statistical significance was determined by one-way ANOVA with Dunnett’s multiple comparisons test.

Invertible element PCR assay

Pyelonephritis isolate CFT073, previously constructed type 1 fimbriae mutants with the promoter situated in the on orientation (fim L-ON) or situated in the off orientation (fim L-OFF) within the invertible element (IE) [59], and all oxidative and reductive TCA pathway mutant strains were incubated statically in LB medium at 37°C for 18 h. Cultures were standardized to an OD₆₀₀ = 0.5. The IE PCR assay was conducted as described previously [60]. In brief, IE_ F (5’-AGTAATGCTGCTCGTTTTGC-3’) and IE_R (5’-GACAGAGCCGACAGAACAAC-3’) were used to amplify the 601-bp invertible element. PCR products were subsequently digested with SnaBI and electrophoresed on a 3% agarose-1× TAE (Tris-acetate-EDTA buffer) gel to determine the orientation of the fim promoter within the IE. Digestion of the IE PCR product with the fim promoter positioned in the on orientation results in a 403 bp and a 198 bp band while digestion of the IE PCR product with the fim promoter positioned in the off orientation results in a 440 bp and a 161 bp band. To determine the proportion of IE DNA with the fim promoter situated in the on and off orientation pixel intensities of the bands on the gel image calculated by Image Lab 6.0 Software (Bio-Rad) were examined. In brief, pixel intensities of both PCR product bands of the fim promoter positioned in the on or off orientation were added together to find the total bacterial population with the fim promoter positioned in the on or off orientation, respectively. Each total bacterial population value (total fim on or total fim off) was then divided by the total sum of pixel intensity (total fim on and total fim off) to calculate the percent bacterial population with the IE containing the fim promoter positioned in the on or off orientation.

Ethics statement

All animal experiments were performed in accordance to the protocol (08999–3) approved by the University Committee on Use and Care of Animals at the University of Michigan. This protocol is in complete compliance with the guidelines for humane use and care of laboratory animals mandated by the National Institutes of Health.