Bacterial culture media

All bacterial cultures were maintained in LB broth (Fisher BioReagents) or on agar plates with 1.5% (w/v) agar (Fisher BioReagents). Strains used in this study are listed in Table 1. Antibiotics were added to media for plasmid maintenance or marker selection at the following concentrations: kanamycin (kan, 50 μg/ml), chloramphenicol (cam, 30 μg/ml), ampicillin (amp, 100 μg/ml). Ten mM EGTA was used to stop P22 phage infection. Screening for pseudolysogens after P22 transductions entailed the use of Evans Blue-Uranine (EBU) plates consisting of (per L): 10 g tryptone, 5 g yeast extract, 5 g NaCl, 1.5 g glucose, 1.5% (w/v) agar. After autoclaving, the medium was supplemented with 40 mL of 12.5% (w/v) K₂HPO₄, 1.25 mL 1% (v/v) Evans Blue solution, and 2.5 mL 1% (w/v) Uranine [37].

Download:

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

Table 1. Strains and plasmids.

https://doi.org/10.1371/journal.pone.0339602.t001

Growth assays

For growth studies with minimal medium, we used M9 minimal medium supplemented with 19 mM ammonium chloride as previously described [30]. F-Asn was synthesized as previously described [29]. Growth of bacteria was measured in a 96-well clear, flat bottom plate with M9 minimal medium supplemented with the specified carbon sources at the concentrations stated and inoculated 1:100 with an overnight culture that had been washed twice in sterile water [30]. Growth was measured with hourly absorbance measurements (600 nm) at 37°C for 15 h in a SpectraMax i3X (Molecular Devices) microplate reader and the SoftMax Pro 6.1 software.

Minimum inhibitory concentration

The MIC for each strain was calculated from a growth assay after 15 h incubation at 37°C using three biological replicates with three technical replicates each (nine total data points). IC₅₀ and IC₉₀ values were calculated with Prism (GraphPad).

Mutant construction

Lambda Red mutagenesis was used to construct in-frame deletion mutants of targeted genes in Salmonella as previously described using primers listed in Table 2 [43]. Some mutations were constructed using pKD3 or PKD4 as PCR templates, while others were constructed using pCLF3 or pCLF4 as template [40,43]. Others were taken from the BEI resources collection [40] which were constructed similarly using pCLF3 or pCLF4 as templates, and then through phage P22 transduction moved into 14028, our laboratory’s wild-type strain (see below). Antibiotic cassettes were removed by electroporating with pCP20 which encodes FLP recombinase, as previously described [43].

Download:

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

Table 2. Oligonucleotides used in this study.

https://doi.org/10.1371/journal.pone.0339602.t002

P22 transductions

Alleles were transduced from one strain to another using phage P22HTint. To accomplish this goal, phage P22HTint was propagated on the donor strain and the resulting phage were used to infect the recipient strain. Transductants were selected on LB supplemented with the appropriate antibiotic and 10 mM EGTA. Colonies were streaked to isolation twice on this same medium. Transductants were then tested for pseudolysogeny and full length LPS by streaking against P22HTint on EBU plates [37].

fraD qRT-PCR

Bacterial strains were grown in LB broth using a roller drum at 37°C overnight. The cultures were then washed twice and resuspended in sterile water. The washed cells (500 μL) were used to inoculate 50 mL of M9 minimal medium supplemented with 5 mM glucose resulting in a 1:100 dilution of cells. These cultures were allowed to grow for 4 h, F-Asn was then added to 1 mM, and growth was continued for 30 min before withdrawing 15-mL aliquots for intracellular 6-P-F-Asp measurements and qRT-PCR. The cells were harvested by centrifuging at 5,000 x g for 10 min at 4°C. Pellets were stored at −20°C. From the qRT-PCR aliquots, RNA was extracted using the Purelink RNA Miniprep kit (Invitrogen). Briefly, pellets were thawed on ice and resuspended in freshly made lysis buffer and lysed cells run through the column per the manufacturer’s protocol. After the RNA was eluted from the column in 50 μL RNase-free water, residual DNA contaminants were removed by treatment with one-tenth volume of 10x DNase buffer and 2 U of DNase (Turbo DNA-free kit; Invitrogen) for 30 min at 37°C. At the end of this incubation, 5 μL of DNase inactivation reagent was added to the reaction, incubated for 5 min at room temperature (~23°C) with periodic mixing, and centrifuged at 19,000 x g for 1 min. The supernatant was removed and placed in an RNase-free microcentrifuge tube and stored at −20°C. For reverse transcription (RT), 100 ng of total RNA was used for cDNA synthesis using Superscript VILO (Invitrogen) per the manufacturer’s protocol. For quantitation of fraD expression, 1 μL of cDNA was used in a qPCR reaction using iTaq SYBR Green Supermix (BioRad) with either 300 μM of primers BA3405 and BA3406 (fraD) or BA2745 and BA2746 (rpoB) in the CFX96 Touch Real-time PCR machine (BioRad). Amplification efficiencies for each primer pair were calculated using a 10-fold dilution series standard curve of pooled cDNA from all samples from one of the biological replicates (fraD BA3405 and BA3506 was 91.48%; rpoB BA2745 and BA2746 was 90.42%; see Table 2). The C_t values were analyzed using the Pfaffl method where the rpoB control was used as the internal calibrator and the wild-type fraD expression is the reference condition and fraD expression in all mutant backgrounds is relative to the wild-type [45].

Measurement of 6-P-F-Asp concentration in Salmonella cells

To measure intracellular 6-P-F-Asp, 15-mL samples of cells were collected, pelleted (see qRT-PCR section above), and resuspended in 300 μL of LC grade water. Biological replicates were collected on three separate occasions. Each replicate was split into three aliquots of 100 μL. Two aliquots were used as analytical replicates for each analysis. 6-P-F-Asp was extracted from each 100 μL aliquot (equivalent to 5 mL of cells), by adding 250 μL of chilled LC grade water and 500 μL of chilled methanol (Fisher Optima LC/MS grade, Fisher Scientific) with 20 nmol [¹³C]-F-Asn added as internal standard. The cell suspension was vortexed for 2 min to facilitate pellet disruption and metabolite extraction. Samples were subjected to ten cycles of ultrasonication (30 s on, 30 s off cycles) using a Bioruptor® Pico (Diagenode). Following cell lysis, samples were centrifuged at 16,000 × g for 15 min at 4°C and the supernatants were transferred to new tubes and dried under vacuum (SpeedVac Concentrator, Thermo Scientific). Before mass spectrometry analysis, these dried pellets were resuspended in 50 μL water:acetonitrile (98:2) with 0.1% (v/v) formic acid (LC-MS grade, Thermo Scientific) and filtered using a 0.2-µm PTFE filter (Thermo Scientific). The supernatant of the flow-through fraction was then injected for liquid chromatographic-mass spectrometric analysis.

To construct a standard curve, different amounts of standard 6-P-F-Asp (0, 20, 80, 160, 200, 400 or 600 nmol) were spiked into the extraction solvent and were prepared as described above. Ten μL of each sample (equivalent to 1 mL of cells) was introduced into an Ultimate 3000 Ultra Performance LC (Thermo Scientific) system with an HSS T3, C18 column (Waters, 2.1 μm × 100 mm, 1.8 µm) and coupled into a triple quadrupole mass spectrometer (Thermo Quantiva TQ-S). Mobile phases were A [0.1% (v/v) formic acid (LC-MS grade, Thermo Scientific) in water] and B [0.1% (v/v) formic acid in acetonitrile]. The separation started with 2% B for 3 min at a flow rate of 100 μL/min and was then followed by a gradient: 3–4 min, 2–10% B; 4–7 min, 10–98% B; 7–7.5 min, 98% B; 7.5–7.6 min, 98–2% B; 7.6–10 min, 2% B. The mass spectrometer was operated in positive ion electrospray ionization mode (ESI+) with a capillary voltage at 4 kV, source temperature 100°C, desolvation temperature 350°C, sheath gas flow 12 L/min and auxiliary gas flow 13 L/min. The gas flow rate for the collision cell was 0.15 mL/min. A multiple reaction monitoring (MRM) mode was used. While the m/z 376 → 242 transition of 6-P-F-Asp with collision energy 15eV was selected for quantitation, m/z 301 → 283 of [¹³C]-F-Asn with collision energy 13eV was used for normalization. Skyline (v20.2, MacCoss Lab, Department of Genome Sciences, University of Washington, Seattle, WA, USA) was used for calculating the peak area of transition.

Measurement of F-Asn in mouse fecal samples

For sample preparation, 1 mL of the extraction solvent (water/acetonitrile, 80/20) with [¹³C]-F-Asn was added to the fecal sample for F-Asn extraction. The fecal slurry was lysed by the bath sonication with Bioruptor® (Diagenode, Belgium) followed by centrifugation for protein precipitation. The supernatant was dried and reconstituted in water-acetonitrile-formic acid (98/2/0.1, v/v) before LC-MS/MS analysis. For the standard-curve analysis, different amounts of F-Asn standard compound (5.15, 12.9, 25.8, 51.5, 128.8, 257.7, 515.5, 1288.6 nmol) were spiked into the extraction solvent and were prepared as described above. F-Asn and [¹³C]-F-Asn were separated using Dionex UltimateTM 3000 RSLC liquid chromatography system (Thermo Scientific, USA) with a C18 column (Kinetex® 1.7 µm C18 100 Å, LC Column 100 x 2.1 mm, Phenomenex Inc, USA). Mobile phases were A [0.1% (v/v) formic acid (LC-MS grade, Thermo Scientific) in water] and B [0.1% (v/v) formic acid in acetonitrile]. The separation started with 2% B for 3 min at a flow rate of 100 μL/min and was then followed by a gradient: 3–7 min, 2% to 98% B; 7–7.5 min, 98% B; 7.5–10 min, 2% B. Thermo Quantiva triple quadrupole mass spectrometer (Thermo Scientific, USA) was used in a positive-ion mode. Transition conditions, including retention time, precursor ion, product ion, and collision energy, are described in Table 3. Skyline (v20.2, MacCoss Lab, Department of Genome Sciences, University of Washington, Seattle, WA, USA) was used for calculating the peak area of transition.

Download:

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

Table 3. Transition conditions for F-ASN and ¹³C₆-F-ASN for LC-MS analysis.

https://doi.org/10.1371/journal.pone.0339602.t003

Measurement of F-Asn in mouse chow using mass spectrometry

For sample preparation, the mouse chow was lyophilized and approximately 10 mg was aliquoted into a 1.5 mL microcentrifuge tube and weighed. Five hundred μL of methanol and 500 uL of water was added as an extraction solvent and the sample was spiked with 1 μL of 2.58 mM [13C]-F-Asn. The sample was vortexed and centrifuged at 14,800 x g for 1 hour. The supernatant was transferred to a clean 1.5 mL tube and allowed to dry. For the standard curve analysis, different amounts of F-Asn standard (ranging from 0.000001 to 1.00 nmol) were spiked into the extraction solvent. After drying, the samples were reconstituted in 50 μL of 80/20 acetonitrile/water with 0.1% (v/v) formic acid (LC − MS grade, Thermo Scientific) and injected for mass spectrometry analysis. F-Asn and [13C]-F-Asn were separated using a Dionex UltimateTM 3000 RSLC liquid chromatography system (Thermo Scientific, USA) with an Acquity UPLC BEH Amide column (1.7 µm, 130 Å, 2.1 mm x 150 mm, Waters Corporation, USA). Mobile phases were A [0.1% (v/v) formic acid (LC-MS grade, Thermo Scientific) in 90/10 water/acetonitrile (v/v)] and B [0.1% (v/v) formic acid in acetonitrile]. The sample injection volume was 1 μL. Gradient elution was applied at a flow rate of 300 μL/min. The gradient began with 80% B for 6 minutes followed by 6 − 20 min, 80 − 50% B; 20 − 26 min, 50% B; 26 − 28 min, 50 − 80% B; 28 − 35 min, 80% B. A Thermo Quantiva triple quadrupole mass spectrometer (Thermo Scientific, USA) was used in a positive-ion mode. Transition conditions, including retention time, precursor ion, product ion, and collision energy, are described in Table 3. Skyline (v 22 MacCoss Lab, Department of Genome Sciences, University of Washington, Seattle, WA, United States) was used for calculating the peak area of transitions. The detection limit of F-Asn was determined to be 50 nM.

Mouse experiments

For competition experiments, a 1:1 mixture of two different bacterial strains were inoculated intragastrically (i.g.) and then recovered from ceca to determine the competitive index (CI) of the two strains. Streptomycin-treated mice were administered 20 mg of streptomycin 24 h prior to inoculation with Salmonella strains [46]. Streptomycin-treated mice were inoculated with 10⁷ total CFU of Salmonella while mice that were not treated with streptomycin were inoculated with 10⁹ total CFU. CI was determined by plating on media selective for WT (LB cam) and mutant (LB kan). For the competitions with the fraB E214A mutant, which is not marked, CI was determined by screening 52 individual colonies recovered from XLD plates for antibiotic resistance by patching on LB plates containing the appropriate antibiotic, thus the limit of detection on these experiments is less than 100-fold.

For the single-infection Swiss Webster experiments, streptomycin-treated Swiss-Webster mice were inoculated with a single bacterial strain at 10⁷ CFU. Where indicated, 100 μL of 1 M F-Asn was administered intragastrically (i.g.) daily and/or a probiotic strain was administered in a 1:1 mixture with the tested Salmonella strain. Ceca and feces were harvested on day four post-infection. CFU enumeration was determined by plating on media selective for WT (LB cam) and the mutant (LB kan). For the fraB E214A mutant, which is not marked, CFU enumeration was determined by plating on XLD.

For experiments involving the CBA/J mice on a high-fat chow, CBA/J mice were switched to this diet (Bioserv S3282) three days prior to inoculation with a single Salmonella strain, either JLD1214 or ASD4355, at 10⁹ CFU (day 0). Where indicated, the mice were also inoculated with either 10⁹ CFU of ASD203 or E. coli Nissle on day 0, and 100 μL of 1 M F-Asn was administered intragastrically on days 0–3. Salmonella strains JLD1214 and ASD4355 were enumerated from feces or cecum by plating on LB cam while ASD203 was enumerated by plating on LB kan. Lipocalin was measured from feces as previously described [6].

Limiting dilution series

Cecal contents from a 1:1 competition of WT and fraB80::kan mutant were used to calculate the percentage of fraB mutants that had become resistant to F-Asn intoxication in vivo. The cecum of each infected mouse was homogenized and serially diluted. One dilution series was plated on LB kan plates to determine the total number of fraB80::kan mutants present. The other dilution series were performed in 96-well plates in medium selective for fraB80::kan mutants that have become resistant to F-Asn (M9 medium supplemented with 5 mM glucose, 5 mM F-Asn, and 50 μg/ml kanamycin). Each plate was incubated at 37°C overnight. After growth overnight, the lids were removed, and the absorbance was read at 600 nm in the Spectra Max M2 (Molecular Devices). The frequency at which a particular dilution yielded growth was used to determine the percentage of resistant mutants present: percentage mutant = -ln(frequency that growth was not obtained)/total bacteria per well. For example, if 3 of 10 wells had no growth at a dilution where there were 240 bacteria per well, we determined that 0.5% of the fraB80::kan mutants were F-Asn resistant.

Mutation fluctuation tests- To calculate the mutation rate to F-Asn resistance, 94 colonies of fraB80::kan were inoculated into the wells of a 96-well plate containing 1X M9 medium supplemented with 5 mM glucose and 5 mM F-Asn. One well was inoculated with wild-type as a positive growth control and one well was inoculated with wild-type but the growth medium was supplemented with chloramphenicol as a negative control for growth. Three of the wells were immediately titered by serial dilution and drip plated on LB plates to enumerate the starting number of CFU in each well. The plate was lidded and incubated at 37ºC for 16 h. After incubation, three of the wells were titered by serial dilution and drip plated on LB plates to enumerate the total CFU in each well after incubation. Then, the lid was removed and the absorbance was read at 600 nm in a Spectra Max M2 plate reader (Molecular Devices). The mutation rate in events per culture (m) = -ln(# of wells with no growth/total number of wells). This rate was then multiplied by the number of cell divisions per culture to obtain the mutation rate per cell division using the method of Luria and Delbrück [47].

Isolation of transposon mutants resistant to F-Asn

A mutant library was created by transducing a T-POP transposon [Tn10dTet[del25] [48]] into the fraB80::kan mutant background, HMB206. To prepare the lysate, Salmonella enterica serovar Typhimurium strain TH3923 was grown overnight in LB at 30°C. The cells were washed and subcultured to OD₅₅₀ of ~0.4. One mL of 0.5 mg/mL mitomycin C was added to induce phage production, and the cells were allowed to lyse with shaking at 37°C for 4 h. The solution was filter sterilized, transferred to a glass tube, and 100 μL of chloroform was added. The transduction was performed immediately after lysate preparation. A 5-mL culture of HMB206 carrying pNK2880, a plasmid that encodes a Tn10 transposase with relaxed target specificity, was grown in LB with ampicillin at 37°C overnight with shaking. In a sterile flask, 5 mL of the TH3923 lysate was added to 5 mL of the HMB206 + pNK2880 culture, and incubated for 25 min at 37°C. The cells were washed twice in LB with EGTA and resuspended in 50 mL of LB with EGTA for a 60-min outgrowth. At this point, the number of mutants in the culture was determined by titering on LB with tetracycline. Tetracycline was then added to the remainder of the culture and the culture allowed to grow at 37°C overnight after which glycerol stocks were created and stored at −80°C. The library contained 25,000 mutants.

A selection was applied to the library to obtain mutants that could grow in the presence of F-Asn. Ninety-six mutants were obtained using a 96-well format in which 2 μL of library was inoculated into 200 μL of medium and one mutant was isolated from each well after overnight growth at 37 °C. Growth of all mutants in M9 + glucose + F-Asn was confirmed. Inverse PCR was used to amplify the transposon insertion site. Genomic DNA of the mutant was purified using the GenElute bacterial genomic DNA kit (Sigma). The genomic DNA was digested with the restriction enzyme NlaIII (NEB). The reaction was incubated at 37°C for 3 h, and then the enzyme was heat inactivated at 65°C for 20 min. The digested DNA was self-ligated to form circular DNA using T4 DNA ligase at 15°C overnight (NEB). The ligation was purified with a QIAprep® Spin Miniprep Kit (Qiagen). The DNA was then digested with DraI for 3 hours at 37°C and then the enzyme heat inactivated for 20 min at 65°C (NEB). PCR was performed with primers IPCRF and IPCRR [49]. The PCR product was purified with QIAquick® PCR Purification Kit (Qiagen) and submitted for Sanger sequencing to the Plant-Microbe Genomics Facility at the Ohio State University.

Histopathology

Samples of the proximal colon were removed from mice, and a portion was immersion fixed in 10% (v/v) neutral buffered formalin, processed by routine methods, and embedded in paraffin wax by the Comparative Pathology and Mouse Phenotyping Shared Resource (CPMPSR) at the Ohio State University. Sections (4 μm) were stained with hematoxylin and eosin (H&E) as previously described [39] and scored in a blinded fashion by a veterinary pathologist, board certified by the American College of Veterinary Pathologists (ACVP).

Animal assurance

All animal work was performed using protocols approved by our Institutional Animal Care and Use Committee (IACUC; OSU 2009A0035) and in accordance with the relevant guidelines set forth in the PHS “Guide for the Care and Use of Laboratory Animals”.

Alleles of fraB have different phenotypes with regard to F-Asn sensitivity

We previously reported that the growth of a fraB80::kan mutant of Salmonella is inhibited by the presence of F-Asn with an IC₅₀ of 19 μM [30]. Here, we report that the growth of a fraB1::cam mutant of Salmonella is unaffected by the presence of F-Asn (Fig 2). We hypothesize that the fraB1::cam mutation is polar on fraD expression, while the fraB80::kan mutation is not. Since fraD expression is required for synthesis of the toxic metabolite, 6-P-F-Asp, expression of fraD would be required for F-Asn to inhibit the growth of a fraB mutant. Both mutations replace nearly all of fraB, except for ten codons each at the 5’ and 3’ termini. The kan cassette is oriented in the same orientation as fraB, while the cam cassette is oriented in the opposite direction (Fig 2). Although neither cassette has a Rho-independent terminator of transcription, the kan (but not cam) promoter may increase fraD expression above wild-type levels. This possibility prompts the question of how much growth inhibition would be observed if expression of the operon was not altered. This question is relevant to drug discovery efforts, as it is essential to determine if a putative small molecule inhibitor of FraB would inhibit the growth of wild-type Salmonella in the presence of F-Asn [50–52]. To address this key question, we utilized multiple strategies.

Download:

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

Fig 2. Different alleles of fraB exhibit different growth phenotypes, altered fraD expression levels, and variable 6-P-F-Asp build-up.

Growth of different fraB mutants in M9 minimal medium containing (A) 5 mM glucose, (B) 5 mM F-Asn, or (C) 5 mM glucose and 1 mM F-Asn. (D) A diagram of the different fraB alleles indicating the antibiotic cassette, and its orientation, used to replace fraB, or the deletion of the antibiotic cassette, or the location of the point mutation E214A (not to scale). The IC₅₀ and IC₉₀ values are listed for all fraB mutants, along with the respective 95% confidence interval. The expression of fraD relative to wild-type, and the concentration of 6-P-F-Asp for each strain are also shown. The fraD expression and 6-P-F-Asp measurements were collected from the same cultures that were grown for 4 h in M9 minimal medium supplemented with 5 mM glucose, at which point F-Asn was added to 1 mM and incubation continued for 30 min. qRT-PCR data were obtained from two biological replicates, each with two technical replicates. 6-P-F-Asp concentrations (expressed as nanomoles per volume cells * the optical density of the culture, or nmol/OD ml) were determined from three biological replicates, each with two technical replicates. “None” indicates that inhibition of growth was not observed at concentrations of F-Asn up to 5 mM. ^a A 95% confidence interval could not be calculated for this strain, so a 50% confidence interval was used instead. Color gradient from white to red indicates level of growth inhibition observed. E) Graphical representation of the relative fraD expression and 6-P-F-Asp concentration for each strain (data from panel D). Dotted line is the limit of detection line (LOD) at 42 nmol per OD*mL of cells which was measured based on a calibration curve and area under the curve analysis for standard samples. Color gradient from white to red indicates level of growth inhibition observed.

https://doi.org/10.1371/journal.pone.0339602.g002

First, we utilized a fraB E214A mutation in the native fra locus [27]. This mutation should have no polar effects on fraD expression. Additionally, the FraB E214A mutant is catalytically inactive, making it a good proxy for a wild-type cell in the presence of a presumptive drug targeting FraB [27,50–52]. Second, we constructed a strain in which the cam cassette was inserted in the same orientation as fraB (fraB53::cam). Finally, we also made strains in which the cam and kan cassettes were removed by Flp-mediated recombination (ΔfraB1, ΔfraB53, and ΔfraB80). For each strain, we then measured fraD expression, the accumulation of 6-P-F-Asp, and growth in the presence of F-Asn (Fig 2).

The fraD gene is most highly expressed, 6-P-F-Asp is at its highest concentrations, and growth is the most severely inhibited, in the fraB80::kan and fraB53::cam mutants (Fig 2). In contrast, the lowest expression of fraD, the lowest concentrations of 6-P-F-Asp, and the least growth inhibition are observed with the fraB1::cam, ΔfraB1, and ΔfraB53 mutants. The fraB E214A and the ΔfraB80 mutants had fraD expression similar to wild-type. 6-P-F-Asp accumulation and growth inhibition were also intermediate in these mutants compared to the other strains. Although the IC₅₀ of F-Asn for the E214A mutant is similar to that of the fraB80::kan mutant, the magnitude of the inhibitory effect is lower based on growth in vitro. The expression of fraD in the fraB E214A or ΔfraB mutants is driven by the native fraB promoter, while fraD expression is increased in the fraB80::kan mutant, presumably by the kan promoter. Despite the seemingly strong correlations, not all the results are easily interpreted.

First, similar to the fraB80::kan strain, the fraB53::cam mutant is severely intoxicated by F-Asn and consistently shows high 6-P-F-Asp accumulation but fraD expression is not higher than the wild-type. It is also not clear why the ΔfraB1 and ΔfraB53 mutants fail to accumulate as much 6-P-F-Asp as the ΔfraB80 mutant. It is likely that there are aspects of FraD (post-transcriptional) regulation that remain to be elucidated. Second, the fraB53::cam but not the fraB1::cam mutant is highly intoxicated by F-Asn (Fig 2). The difference in these mutations is the orientation of the cam cassette. There is intoxication by F-Asn only when the cassette is oriented with fraB and not opposite to fraB. When oriented opposite to fra, there is a stop codon created in the fraB gene that is likely polar on fraD (i.e., premature termination of transcription due to uncoupling of translation and transcription); indeed, fraD is not expressed in fraB1::cam (Fig 2). Without FraD, there can be no synthesis of the toxic intermediate, 6-P-F-Asp, explaining the absence of intoxication in fraB1::cam. In the fraB80::kan mutant, the kan cassette is oriented with fraB, and fraD expression is three-fold higher than wild-type, providing an explanation for the enhanced intoxication of this strain by F-Asn (Fig 2).

Competition experiments to determine the fitness of a fraB E214A mutant in mouse models

The intermediate level of growth inhibition observed with the fraB E214A mutant in the presence of F-Asn suggested that this mutant may not be as attenuated in mice as the fraB80::kan mutant. This notion was confirmed in several mouse models using competition experiments in which the wild-type was inoculated with a mutant in a 1:1 ratio (Fig 3). If the mutant is recovered in less than a 1:1 ratio, it is considered less fit than the wild-type. Initial studies were performed in a streptomycin-treated C57BL/6 mouse model. In this model, the fraB E214A mutant has a 4-fold decrease in fitness compared to wild-type, while the fraB80::kan mutant has a 524-fold decrease in fitness (Fig 3A). Similarly, in the streptomycin-treated Swiss Webster mouse model, the fraB E214A mutant is significantly less attenuated than the fraB80::kan mutant (9-fold vs > 10,000-fold, respectively) (Fig 3B).

Download:

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

Fig 3. The Salmonella fraB E214A mutant is attenuated in different murine models.

In each experiment, a 1:1 mixture of two different bacterial strains is inoculated intragastrically (i.g.) and then recovered later from ceca to determine the competitive index (CI) of the two strains. The pairs of strains inoculated are listed on the X-axis. The use of strain JLD1214 (wild-type 14028 marked with a chloramphenicol-resistance cassette in a neutral location [24,30,39,53,54]) is indicated as “WT”. The use of strain HMB206 (fraB80::kan) is indicated as “fraB”. The use of strain ASD1312 (fraB E214A) is indicated as “fraB E214A”. A) Streptomycin-treated C57BL/6 mice inoculated with 10⁷ total CFU and harvested after four days. B) Streptomycin-treated Swiss-Webster mice inoculated with 10⁷ total CFU and harvested after four days. C) CBA/J mice inoculated with 10⁹ total CFU and harvested after either one or ten days. D) Streptomycin-treated CBA/J mice inoculated with 10⁷ total CFU and harvested after four days. Black circles indicate that the CI was determined by plating on media selective for WT (LB cam) or the mutant (LB kan). Open circles indicate that the CI was determined by screening 52 individual colonies recovered from XLD plates for antibiotic resistance by patching on LB plates containing the appropriate antibiotic, thus the limit of detection on these experiments is less than 100-fold. The data were transformed and the mean with standard deviation is shown. Statistical significance of logCI being different than zero was determined using a one sample t-test (two-tailed). Statistical significance of two competitions being different from each other was determined using a student’s t-test (unpaired, two-tailed). * = P value <0.05, ** = P value <0.005, *** = P value <0.0005, **** = P value <0.00005.

https://doi.org/10.1371/journal.pone.0339602.g003

We next chose to test the fraB mutant phenotypes in a CBA/J model. In this model, the gastrointestinal tract slowly becomes inflamed in the presence of Salmonella (8–12 days post-infection) without the need for perturbation of the normal microbiota with antibiotics [34,35]. At 10 days post-infection, the fraB80::kan mutant was significantly attenuated (25-fold) while the fraB E214A mutant was not (Fig 3C). However, the presence of only the wild-type or the mutant in 50% of the mice suggested that the bacterial population was undergoing a population bottleneck. To circumvent this issue, we repeated the experiment but pre-treated the mice with streptomycin. In streptomycin-treated CBA/J mice, both the fraB80::kan and the fraB E214A mutants are equally attenuated one day post-infection (10- and 8-fold, respectively) (Fig 3D). After 10 days post-infection, the fraB80::kan mutant was no longer recovered. Comparison to the number of wild-type recovered indicates that the fraB80::kan mutant is at least 10,000-fold attenuated. The fraB E214A mutant was not recovered either, but since the fraB E214A mutant is not marked with an antibiotic resistance gene, Salmonella was first recovered on XLD plates and then 52 individual colonies from each mouse were patched on LB cam, with growth indicating the strain was wild-type and no growth indicating the strain was a fraB E214A mutant. The limit of detection in this experiment is low, but the fraB E214A mutant was nonetheless not recovered.

The roles of SPI1 SPI2, and ttrA

Salmonella benefits from SPI1 and SPI2, each encoding a type III secretion system, as they provoke an inflammatory response that disrupts the microbiota [10]. The disruption of the microbiota provides access to nutrients including 1,2-propanediol, ethanolamine, glucarate, galactarate, L-lactate, and F-Asn [7,8,31,55–57]. Unique access to nutrients is beneficial, but gaining increased energy over the competition (e.g., through respiration of metabolites) is even more advantageous from a pathogen expansion perspective. The oxidative burst that accompanies inflammation leads to the accumulation of tetrathionate, nitrate, and even oxygen, all of which are used as terminal electron acceptors by Salmonella for respiration [56,58–60]. The ttrA gene is required for Salmonella to utilize one of these electron acceptors, tetrathionate [59].

In 2014, we tested and believed that we had confirmed the requirement of SPI1 and SPI2 in order for Salmonella to gain access to F-Asn [24]. However, the studies performed here reveal an error in that paper [24]. In our earlier study, we demonstrated that two independently constructed fraB::kan mutants were attenuated in a variety of mouse models that are susceptible to inflammation, and not attenuated in mouse models that are not susceptible to inflammation. Both mutants had fitness restored by complementation with a plasmid encoding the fra island. Those results are correct. However, because SPI1 and SPI2 are required for Salmonella to induce inflammation of the intestine, we then tested the hypothesis that in the absence of SPI1 and SPI2, a fraB mutant would no longer gain access to F-Asn and be intoxicated. Therefore, a competition experiment was performed between a SPI1 SPI2 double mutant and a SPI1 SPI2 fraB triple mutant in germ-free or strep-treated C57BL/6 mice. Both strains were recovered in equivalent numbers indicating that the fraB mutation does not cause attenuation in the SPI1 SPI2 mutant background [24]. Unfortunately, we used a fraB1::cam mutation to construct the SPI1 SPI2 fraB triple mutant because the SPI2 locus had already been disrupted with a kan cassette [24]. Since we now know that the fraB1::cam mutation is not intoxicated by F-Asn, there was no toxicity for the SPI1 and SPI2 mutations to alleviate in the SPI1 SPI2 fraB triple mutant. (Incidentally, the figure in that paper [24] is mislabeled and indicates that it was a fraB::kan mutation in the SPI1 SPI2 strains, although the figure legend correctly states that it was a fraB1::cam mutation.)

We have repeated the experiments here, placing a fraB80::kan mutation into a SPI1 SPI2 background. In competition experiments using CBA/J mice at ten days post-infection, the fraB80::kan mutant was attenuated 25-fold compared to wild-type, while a fraB80::kan SPI1 SPI2 triple mutant was not attenuated compared to a SPI1 SPI2 double mutant (Fig 3C). While this finding would appear to confirm that a fraB mutation does not cause attenuation when present in a SPI1 SPI2 background, it is not an unambiguous result as the SPI1 SPI2 strains appeared to undergo a population bottleneck in these mice, with roughly half of the mice containing only fraB mutant bacteria and half of the mice containing only fraB⁺ bacteria at the end of the experiment. This bottleneck is likely due to both strains in the competition lacking SPI1 and SPI2 and failing to induce inflammation. When the CBA/J mice were pre-treated with streptomycin, the bottleneck effects were ameliorated and the fraB80::kan mutant was greatly attenuated compared to wild-type (>30,000-fold). Moreover, the fraB80::kan SPI1 SPI2 triple mutant was attenuated, but not as much as the SPI1 SPI2 double mutant (251-fold) (Fig 3D). Thus, it appears that the SPI1 and SPI2 mutations at least in CBA/J mice partially abrogate the fraB80::kan phenotype. Since inflammation eliminates competitors for F-Asn [7], we hypothesize that the SPI1 SPI2 mutants fail to provoke inflammation and leave intact competitors for F-Asn. Therefore, there is less F-Asn is available to intoxicate the fraB mutant. Interestingly, in strep-treated C57BL/6 mice and strep-treated Swiss-Webster mice, the SPI1 and SPI2 mutations did not abrogate the fraB80::kan phenotype at all (Fig 3A, 3B). This outcome is likely attributable to the fact that streptomycin treatment eliminates competitors for F-Asn, thereby allowing the fraB mutant to acquire F-Asn and to be intoxicated by 6-P-F-Asp. Thus, SPI1 and SPI2 are essential for fraB phenotypes in some scenarios but not others presumably because access to F-Asn is being regulated by other factors. These results also emphasize how different mouse models could result in different conclusions as to the essentiality of specific genes for Salmonella pathogenesis.

We also confirmed and extended our previous studies on the dependence of the fraB phenotype on ttrA, a gene required for the utilization of tetrathionate as an electron acceptor (Fig 3) [8,24,61,62]. Confirming our previously published result, the fraB phenotype is dependent on ttrA in a streptomycin-treated C57B/6 mouse model (Fig 3A) [24]. Additionally, we determined that the ttrA1 mutation partially abrogates the attenuation of a fraB80::kan mutant in streptomycin-treated Swiss webster mice (Fig 3B). Interestingly, in a streptomycin-treated CBA/J model, tetrathionate utilization is required for a fraB80::kan phenotype at one day post-infection but not at 10 days post-infection when a fraB80::kan ttrA1 double mutant is just as attenuated as the fraB80 single mutant (Fig 3D). The mechanism by which ttrA affects the intoxication of a fraB mutant by F-Asn is not clear.

Acquisition of resistance of a Salmonella fraB mutant to F-Asn

An important aspect of any therapeutic strategy is to understand the mechanism(s) by which cells can become resistant. To determine the mechanisms by which a Salmonella fraB80::kan mutant may become resistant to F-Asn, we performed a selection for resistant mutants. A library of 25,000 T-POP mutations was created in the Salmonella fraB80::kan strain HMB206. The T-POP transposon is a mTn10 derivative that lacks a transcription terminator at the end of the tetA gene encoding the tetracycline efflux pump [48]. This design allows transcription to proceed into neighboring genes, albeit under the control of tetracycline. Thus, in addition to the disruption of genes, artificial activation can be achieved. Selection was applied to the library by growing the cells in M9 minimal medium containing 5 mM glucose, 5 mM F-Asn, kanamycin and tetracycline. To avoid the characterization of siblings, the library was grown in the wells of a 96-well plate and one mutant was obtained from each well. The insertion site was determined for 96 mutants resistant to F-Asn. Of 96 sequences, nine mapped to fraD, 86 mapped to fraA and one mapped to narU. When the narU mutation was transduced into HMB206 (a clean background), it no longer conferred resistance to F-Asn. This finding indicates narU is not responsible for the F-Asn-resistance phenotype and that there was a second mutation in the original strain (which we did not map). Thus, in this genetic screen, only fraD and fraA conferred resistance to F-Asn. Interestingly, fraE mutations were not isolated, which is consistent with our findings that the ansB gene encodes an asparaginase that can partially compensate for the lack of fraE [26].

We determined the spontaneous mutation rate to F-Asn resistance (no transposons, just natural mutation rate) using a fluctuation test in which 94 separate cultures of a fraB80::kan mutant were grown in the presence of 5 mM F-Asn. The number of cultures that yielded no resistant mutants was used to calculate the mutation rate using the method of Luria and Delbrück [47]. The rate of spontaneous mutation to F-Asn resistance was determined to be 6.5 x 10⁻⁸ mutants per cell division. This rate is similar to the rate that Luria and Delbrück obtained when selecting for resistance to T1 phage and is expected as simple knock-outs confer resistance in each system (i.e., tonB or fhuA for T1^r, and fraA or fraD for F-Asn^r).

When a wild-type (JLD1214) and fraB80::kan mutant (HMB206) are inoculated into CBA/J mice in a 1:1 ratio (10⁹ CFU), the wild-type is recovered in 1000-fold higher numbers than the mutant after 16 days (Fig 4). We were curious as to why fraB80::kan mutants were recovered given the evidence that F-Asn is bactericidal to the fraB80::kan mutant during infection of mice [30]. We hypothesized that the fraB80::kan mutants may be acquiring secondary mutations (presumably in fraA or fraD) that make them resistant to F-Asn. To test this hypothesis, we performed a limiting-dilution experiment on recovered homogenized ceca from ten infected mice to determine the percentage of fraB80::kan mutants that had become resistant to F-Asn. In four of the mice, all of the fraB80::kan mutants recovered had become resistant to F-Asn. In two mice, none of the fraB80::kan mutants recovered had become resistant to F-Asn. Three mice yielded intermediate values in which 0.4%, 0.4% and 50% of the fraB80::kan mutants recovered were resistant to F-Asn. No fraB80::kan mutants were recovered at all from the tenth mouse so a percentage could not be obtained. In conclusion, fraB80::kan mutants often become resistant to F-Asn in this mouse model.

Download:

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

Fig 4. A fraD mutant and a mutant lacking the fra island are slightly attenuated in a CBA/J mouse model.

In each experiment, 6 to 10-week-old CBA/J mice were inoculated intragastrically (i.g.) with a 1:1 mixture of two different bacterial strains (10⁹ total CFU) and then recovered from ceca 15 days later to determine the competitive index (CI) of the mutant. The pairs of bacterial strains inoculated are listed on the X-axis. The wild-type strain is JLD1214 which is 14028 marked with a neutral chloramphenicol resistance cassette. The fraD4::kan strain is HMB184, the fraA4::kan strain is HMB247, the fraRBDAE80::kan is HMB205, and the fraB80::kan strain is HMB206. B) The competitive index data from panel A of fraD4::kan, fraA4::kan and fraRBDAE80::kan combined. The CI was determined by plating on media selective for WT (LB cam) or the mutant (LB kan). The data were transformed and the mean with standard deviation is shown. Statistical significance of a logCI being different than zero was determined using a one sample t-test (two-tailed). Statistical significance of two competitions being different from each other was determined using a student’s t-test (unpaired, two-tailed). * = P value <0.05, ** = P value <0.005.

https://doi.org/10.1371/journal.pone.0339602.g004

Since suppressor mutations that arise in vivo render a fraB80::kan mutant resistant to F-Asn, we tested the hypothesis that these mutants might still be attenuated due to a lack of F-Asn utilization. We have previously tested this idea using a fraD mutant and a fra island mutant (lacking all of fraR-fraBDAE) and found that neither were attenuated in a strep-treated Swiss Webster mouse model [30]. Here, we re-examined this possibility using larger numbers of CBA/J mice. We tested a fraD mutant, fra island mutant, and a fraA mutant. We found that in competition with wild-type (JLD1214), the fraA mutant was 2.6-fold attenuated, the fraD mutant was 14-fold attenuated, and the fra island mutant was 9-fold attenuated (Fig 4). Thus, any suppressor mutant that can no longer utilize F-Asn should be at least slightly attenuated in CBA/J mice over time.

Use of single infections to determine the amount of inflammation caused by a Salmonella fraB mutant

We have typically measured the fitness of fraB mutants during competitive infections with the wild-type given their higher precision compared to single and separate infections of mice with these strains. However, we recognize that the presence of wild-type Salmonella during competitive infections would contribute additively to inflammation and thereby help the fraB mutant gain access to F-Asn [7]. If a fraB mutant were to be used alone, SPI1 and SPI2 in the fraB mutant should induce inflammation thus engendering F-Asn availability but the F-Asn would soon after intoxicate the fraB mutant. If inflammation subsides upon elimination of the fraB mutant, the availability of F-Asn may also decline and promote a revival of the remaining fraB mutant cells. This oscillation triggered by the undulating inflammation and F-Asn prompts the question of whether the need for inflammation (a pre-requisite for F-Asn) be offset by adding exogenous F-Asn. To answer this question, we infected streptomycin-treated Swiss Webster mice with wild-type alone or the fraB E214A mutant alone. Four days later, the ceca were harvested for enumeration of bacteria, the proximal colon was collected for histopathology, and fecal samples were collected for determination of F-Asn concentration. The fraB E214A mutant was recovered 48-fold lower than the wild-type, however, the inflammation induced by the fraB mutant was similar to that induced by the wild-type (Fig 5). Surprisingly, providing exogenous F-Asn did not substantially change the results.

Download:

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

Fig 5. A fraB E214A mutant is attenuated in a single infection streptomycin-treated mouse model.

Streptomycin-treated Swiss Webster mice were infected with the indicated strains. On day 4 post-infection, CFU were enumerated (A, D), F-Asn in feces was measured by LC-MS/MS (B, E), and histopathology scores of proximal colon were determined (C, F). Mock samples are mice that were not inoculated with Salmonella and “strep-only” mice are mice that received only streptomycin-treatment. Some groups were administered 29.4 mg of F-Asn intragastrically (i.g.) daily. For panels A-C the wild-type is JLD1214 which is 14028 marked with a chloramphenicol resistance cassette in a neutral location in the chromosome, and the fraB E214A mutant is ASD1312. For panels D-F the wild-type is JLD1214 and the fraB E214A mutant is ASD4355 which is marked with a chloramphenicol resistance cassette in a neutral location in the chromosome. Where indicated, the probiotic strain ASD203 was inoculated i.g. on day 0 (10⁹ CFU). Panel A shows the results from two separate experiments (5 mice per group in each experiment, 10 mice per group total). The inoculum was 1.3 x 10⁸ and 8.4 x 10⁷ for the wild-type and 9.8 x 10⁷ and 7.2 x 10⁷ for the fraB E214A mutant. The CFU recovered of the wild-type strain was determined by plating on LB cam, while the number of fraB mutant recovered was determined by plating on XLD. Panel D shows the results from two separate experiments (5 mice per group in each experiment, 10 mice per group total). The inoculum was 1.2 x 10⁸ and 5 x 10⁷ for the wild-type and 1.5 x 10⁸ and 2.8 x 10⁷ for the fraB E214A mutant. When probiotic ASD203 is present, CFU of each strain was determined by plating on media selective for WT (LB cam), the mutant (LB cam), or the probiotic (LB kan). Probiotic counts are indicated by the blue squares. The data were transformed and the mean with standard deviation is shown. Statistical significance of two groups being different from each other was determined using an ordinary one-way ANOVA. * = P value <0.05, ** = P value <0.005, *** = P value <0.0005, **** = P value <0.00005.

https://doi.org/10.1371/journal.pone.0339602.g005

Probiotics enhance the attenuation of fraB E214A in streptomycin-treated Swiss Webster mice

E. cloacae can enhance the attenuation of a fraB mutant in germ-free mice [24]. Since E. cloacae cannot utilize F-asn [32], we hypothesize that this enhancement is due to E. cloacae competing with Salmonella for a large proportion of the nutrients available, driving Salmonella to consume F-Asn as a higher proportion of its nutrient pool. We decided to re-visit this idea, using what we refer to as our “Salmonella probiotic” strain, ASD203, which lacks SPI1 and SPI2 and therefore cannot induce inflammation [10,39]. Additionally, this strain cannot consume F-Asn due to a deletion of the entire fra island (fraR-fraBDAE). The ansB gene is also deleted as this enzyme is capable of converting F-Asn to F-Asp [26]. The addition of this “probiotic” should provide a competitor to Salmonella for all nutrients except F-Asn, effectively forcing the use of F-Asn as a nutrient. The probiotic strain was added in a 1:1 ratio with either wild-type or fraB E214A mutant Salmonella, in the presence or absence of exogenous F-Asn. The number of fraB mutant bacteria recovered, and the resulting inflammation, were dramatically reduced by the probiotic, regardless of whether additional F-Asn was provided (Fig 5).

A high fat diet eliminates the need for streptomycin pretreatment in a Salmonella gastroenteritis model

Mice are quite resistant to Salmonella-mediated inflammation of the gastrointestinal tract unless the microbiota is first disrupted with streptomycin (or other antibiotics) [63–68]. These streptomycin-treated mouse models have been extremely useful for studying Salmonella-mediated inflammation. However, the use of streptomycin is not ideal when the research objective is to study the interaction of Salmonella with an intact or “normal” microbiota. The CBA/J mouse model solves this problem in that Salmonella can persist in these mice and given enough time (8–12 days) begin to cause inflammation of the gastrointestinal tract and thrive in this inflamed environment. A disadvantage of the CBA/J model is that only about one third of the mice become “high responders” with a high Salmonella burden and inflammation. The remaining mice tend to have a low Salmonella burden and no inflammation. This heterogeneity leads to high variability. Recent reports [69,70] that a high-fat diet can increase susceptibility of mice to Salmonella-mediated inflammation prompted us to test the hypothesis that the percentage of CBA/J mice that become high responders could be increased by placing them on a high-fat diet. This idea was confirmed when we compared the recovery of Salmonella from orally inoculated CBA/J on normal chow versus high-fat diet. The CFU recovered from the mice on a high-fat diet were much more consistent and all mice were high responders (Fig 6).

Download:

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

Fig 6. Optimization of a diet-enhanced colitis mouse model.

Three different mouse models were compared for recovery of wild-type Salmonella (JLD1214), after oral inoculation, from feces or cecum, as indicated. Swiss-Webster mice treated with streptomycin were orally inoculated with 1 x 10⁷ CFU; CBA/J mice on normal chow were inoculated with 1 x 10⁹ CFU; and CBA/J mice that had been on high-fat diet for ten days were inoculated with 1 x 10⁹ CFU. The data was transformed and the mean with standard deviation is shown. Statistical significance of two groups being different from each other was determined using an ordinary one-way ANOVA. * = P value <0.05, ** = P value <0.005, *** = P value <0.0005, **** = P value <0.00005.

https://doi.org/10.1371/journal.pone.0339602.g006

We next performed an experiment in which mice were placed on a high-fat diet and were then infected with either wild-type Salmonella or the fraB E214A mutant (Fig 7A) F-Asn was gavaged on days zero through three post-infection which caused greater than a 100-fold reduction of the fraB mutant. Interestingly, after daily F-Asn gavaging was stopped, the fraB mutant CFU recovered to wild-type levels on day 5. Unlike normal mouse chow, the high-fat chow has no F-Asn (detection limit of 50 nM, see Materials and Methods) accounting for the need for F-Asn supplementation to inhibit the fraB mutant. When combined with F-Asn, the administration of the Salmonella probiotic strain ASD203 had no effect on the wild-type infection but greatly reduced the fraB mutant recovered (greater than 10,000-fold). The effect of ASD203 on the fraB mutant persisted through day six when the experiment was ended, and indeed, enumeration of ASD203 showed that this strain persisted throughout the experiment (Fig 7C). Most importantly, the combination of F-Asn and the probiotic ASD203 greatly reduced the inflammation caused by the fraB mutant (greater than 1000-fold on day 6) (Fig 7B).

Download:

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

Fig 7. A combination of F-Asn and a probiotic can reduce the bacterial burden and inflammation caused by a Salmonella fraB E214A mutant during infection of CBA/J mice on a high fat diet.

Groups of five female CBA/J mice were placed on a high-fat diet, ad libitum, 3 days prior to oral inoculation with either 1 x 10⁹ CFU of wild-type (JLD1214) or a fraB E214A mutant (ASD4355). The inoculation of these Salmonella strains marks day 0 and is indicated on the X-axis. Also on day 0, where indicated, the mice were orally inoculated with 1 x 10⁹ CFU of a probiotic strain, either ASD203 or E. coli Nissle, as indicated below the X-axis. On days 0-3, all mice were orally gavaged with 100 μl of 1 M F-Asn. A) On days 1, 3, and 5 Salmonella were enumerated in feces. On day 6, the mice were sacrificed and Salmonella were enumerated in the cecum. B) From the mice in panel A, the lipocalin concentration in feces was measured on days 0, 2, and 4, and in cecum on day 6. C) From the mice in panel A, the probiotic Salmonella strain ASD203 was enumerated from feces on days 1, 3, and 5, and from the cecum on day 6. The data were transformed and the mean with standard deviation is shown. Statistical significance of two groups being different from each other was determined using an ordinary one-way ANOVA. * = P value <0.05, ** = P value <0.005, *** = P value <0.0005, **** = P value <0.00005.

https://doi.org/10.1371/journal.pone.0339602.g007

Since a genetically modified Salmonella strain is less appealing as a probiotic, we tested E. coli Nissle, a commercially available probiotic. Like ASD203, E. coli Nissle cannot cause inflammation, and is unable to utilize F-Asn, and thus should act as a competitor for nearly all nutrients that Salmonella can utilize. E. coli Nissle should therefore drive Salmonella to utilize a higher proportion of F-Asn. This expectation was confirmed when E. coli Nissle, combined with F-Asn, dramatically reduced the number of Salmonella fraB E214A mutant bacteria recovered from mice on day 1, but then had progressively less effect on days 3, 5, and 6 (Fig 7A). It appears that E. coli Nissle did not persist in this experiment, although we did not measure the levels of this probiotic strain since it has no easily selectable antibiotic-resistance marker.