S. enterica ridA mutants have a defect in competitive fitness

In minimal glucose medium, a ridA mutant has a slight, often immeasurable, growth defect in monoculture (Figure 2, Figure 3). Despite the lack of a significant growth phenotype, activity of branched chain aminotransferase (IlvE, EC:2.6.1.42) is decreased by ~40% in a ridA mutant under these conditions. This demonstrable biochemical defect indicates 2AA has accumulated 30, 31, 32. Serine can be added to the medium to increase 2AA stress. The presence of serine prevents growth of a ridA mutant in S. enterica by partially inactivating serine hydroxymethyltransferase (GlyA, EC:2.1.2.1) and, to a lesser extent, cysteine desulfurase (IscS, EC:2.8.1.7) 5, 13, 18. In competition experiments, S. enterica ridA mutants and wild type cells were co-inoculated into nutrient medium and minimal glucose medium. The competitive index (CI) for strains was calculated as the ratio of strains after growth in coculture normalized to the ratio present in the inoculum. The CI is a measure of competitive fitness that has typically been used to study the requirement for gene products when cells are grown in competition for a single limiting resource 22, 23. In contrast, experiments herein were designed to define competition under conditions where no single resource was limiting. In other words, CI values presented here define differences in general fitness (i.e., growth rate) of a strain that may not be apparent in monoculture experiments. No difference in fitness was detected between wild type and a strain lacking RidA grown in nutrient medium. In contrast, the ridA mutant had a significant fitness defect when the two strains were cocultured in minimal glucose medium. In minimal medium, the ridA mutant had a CI of 0.037 (ln(CI) = -3.3), which reflects a ~27-fold decrease in fitness compared to wild type (Figure 2). The significant fitness difference between wild type S. enterica and the ridA mutant amplified a growth difference that was minimal, or undetectable, in monoculture and underlined the sensitivity of this approach. Results from this competition assay aligned with the 2AA-dependent decrease in IlvE activity of a ridA mutant grown in minimal medium, validating that 2AA stress is present in S. enterica during growth in minimal medium 30, 31.

–

FIGURE 2: S. enterica ridA mutants have a fitness defect in minimal medium. (A) Growth of S. enterica wild type (black) and a ridA mutant (red) in nutrient broth (NB, circles) and minimal glucose (11 mM) medium (triangles). (B) Competitive fitness of a S. enterica ridA mutant (ridA3::MudJ) grown in competition with wild type S. enterica in nutrient or minimal medium. The underlined genotype indicates the strain for which the competitive indices are shown in the corresponding graph. An asterisk (*) indicates a significant difference (p ≤ 0.05 as determined by Sidak’s multiple comparisons test) in CI values compared to control experiments in which the ridA mutant was grown in competition with a second ridA mutant (ridA4::Cm) in the relevant medium. Data shown are the mean and standard deviation of three independent biological replicates.

Isoleucine restores fitness of a S. enterica ridA mutant in coculture

Isoleucine in the growth medium restored full competitive fitness to the ridA mutant (Figure 3). This result was expected, as isoleucine is known to allosterically inhibit serine/threonine dehydratase IlvA and prevent the formation of 2AA 30, 31. These data supported the idea that the fitness defect of a ridA mutant was due to accumulated 2AA.

–

FIGURE 3: Glycine and thiamine alleviate fitness defects of S. enterica ridA mutants. (A) Growth of S. enterica wild type (black) and a ridA mutant (red) in minimal glucose (11 mM) medium supplement with nothing (circles), 1 mM L-isoleucine (triangles) or 1 mM glycine and 100 nM thiamine (diamonds). (B) Competitive fitness of a S. enterica ridA mutant (ridA3::MudJ) grown in competition with wild type S. enterica in minimal glucose medium with indicated supplements. An asterisk (*) indicates a significant difference (p ≤ 0.05 as determined by Sidak’s multiple comparisons test) in CI values compared to control experiments in which the ridA mutant was grown in competition with a second ridA mutant (ridA4::Cm) in the indicated medium. Data shown are the mean and standard deviation of three independent biological replicates.

A caveat to this simple interpretation is that RidA also deaminates the threonine-derived enamine product of IlvA, 2AC. 2AC is an obligatory intermediate in the biosynthesis of isoleucine that is eliminated when isoleucine inhibits IlvA. Thus, it was possible that RidA improved fitness of S. enterica by increasing biosynthetic flux to isoleucine. Distinguishing these two possibilities was complicated by the fact that isoleucine would eliminate the need for RidA in both scenarios. Coculture experiments to separate the possible contributions of RidA were based on the response of S. enterica to higher levels of 2AA stress generated when exogenous serine was added. The assumption was that the effect of 2AA stress generated by endogenous serine would be similar in consequence, if not magnitude, to that generated by addition of exogenous serine. Specifically, S. enterica ridA mutants grow well in the presence of 2AA stress if glycine and thiamine are provided. Together, these nutrients bypass the two key enzymes damaged by 2AA, GlyA and IscS and allow growth in the presence of 2AA stress 5, 13. Notably, the presence of glycine and thiamine in the medium improved fitness of a ridA mutant in coculture with wild type, making them near equal (Figure 3). Importantly, a minor, but statistically significant difference in fitness remained. These data supported, but were not conclusive, that the competitive defect of a ridA mutant was caused by 2AA-dependent damage to the target enzymes GlyA and IscS. The slight fitness defect that remained left the possibility that either i) decreased flux to 2-ketobutyrate (from 2AC) in the isoleucine biosynthetic pathway contributes to the fitness defect of a ridA mutant, or ii) 2AA-dependent damage to PLP-dependent target enzymes beyond GlyA and IscS contribute to the fitness defect of a ridA mutant in S. enterica.

RidA proteins are required for full fitness of E. coli

Past studies have identified multiple differences between the metabolic networks of S. enterica and E. coli 21, 33, 34. E. coli encodes two RidA proteins, RidA and TdcF, which have been assessed for a role in 2AA stress 21. The ridA gene is monocistronic, while tdcF is in the tdc operon that encodes enzymes involved in a putative threonine catabolic pathway 35, 36 and are thought to be expressed only under anaerobic conditions 37, 38. Evidence of 2AA stress was observed in E. coli only with significant manipulation of the genome and growth medium. Loss of TdcF and/or RidA had no observable effect on growth nor activity of known 2AA target enzymes, unless 2AA stress was artificially intensified by expression of a feedback-insensitive variant of IlvA (encoded by the ilvA219 allele) in the presence of exogenous serine 21.

Unlike S. enterica, the absence of RidA proteins in E. coli (RidA and TdcF) had no effect on the growth, or competitive fitness of strains grown in minimal medium. In medium supplemented with 5 mM serine, the ridA tdcF double mutant strain had a small, but statistically significant, difference in growth rate (Figure 4) reminiscent of S. enterica ridA mutants grown in minimal glucose medium. Despite the minimal growth difference in monoculture, the ridA tdcF mutant was notably less fit than wild type in minimal medium supplemented with serine (Figure 4). Unlike the situation in S. enterica, there was no significant defect in IlvE activity of the E. coli tdcB ridA mutant when grown under these conditions that produced a clear fitness defect (5 mM serine) (Table 1).

–

FIGURE 4: RidA proteins are required for full fitness of E. coli. (A and B) Shown are growth profiles of wild type E. coli (black), a ridA mutant (red, ridA790::Km), tdcF mutant (blue, tdcF12::Km) and a ridA tdcF double mutant (grey, ∆ridA890tdcF13::Cm). Strains were grown in (A) minimal glucose (11 mM) medium with no supplementation or (B) supplemented with 5 mM L-serine. (C) Shown is competitive fitness of an E. coli ridA tdcF double mutant (∆ridA890 tdcF13::Cm) grown in coculture with wild type E. coli in minimal glucose (11 mM) medium with or without L-serine (5 mM). Data shown are mean and standard deviation of three independent biological replicates. The underlined genotype indicates the strain for which the CIs are shown in the corresponding graph. An asterisk (*) indicates a significant difference (p ≤ 0.05 as determined by Sidak’s multiple comparisons test) in CI values compared to control experiments in which ∆ridA890 tdcF13::Cm was competed with an different strain lacking ridA and tdcF (∆ridA890 ∆tdcF14 ) in the indicated medium.

Additional competition experiments showed that RidA was the homolog critical for full fitness of E. coli grown in minimal medium with 5 mM exogenous serine (Figure 5). There was no difference between the fitness of the ridA tdcF double mutant and a ridA mutant. In contrast, the ridA tdcF double mutant was less fit than a tdcF mutant when grown in coculture. Isoleucine restored full fitness to the E. coli strains, eliminating the effect of the ridA mutation as found in S. enterica. These data were consistent with a scenario that 2AA stress negatively impacted fitness of E. coli.

–

FIGURE 5: RidA contributes to the competitive fitness of E. coli. Shown are competitive fitness values for an E. coli ridA tdcF double mutant (∆ridA890 tdcF13::Cm) grown in coculture with (A) an E. coli ridA mutant (ridA790::Km) or (B) a tdcF mutant (tdcF12::Km) in minimal glucose (11 mM) medium with the indicated supplements. L-serine was provided at a final concentration of 5 mM and L-isoleucine each provided at 1 mM. Data shown are mean and standard deviation of three independent biological replicates. The underlined genotype indicates the strain for which the CIs are shown in the corresponding graph. An asterisk (*) indicates a significant difference (p ≤ 0.05 as determined by Sidak’s multiple comparisons test) in CI values compared to control experiments in which ∆ridA890tdcF13::Cm was competed with an isogenic strain also lacking ridA and tdcF (∆ridA890 ∆ tdcF14) under each condition.

2AA stress is induced in E. coli with increased serine concentration

The above results could be explained in a scenario where 5 mM serine generated enough 2AA to compromise fitness, but not enough to cause detectable damage to target enzyme IlvE. If this was the case, increased concentrations of serine might increase 2AA production to a level that could generate a growth defect in monoculture. Consistently, both the ridA and ridA tdcF mutants had a growth defect with 20 mM serine, while the growth of wild type was not affected (Figure 6). The serine-induced growth defects of the ridA and ridA tdcF mutants were eliminated with the addition of isoleucine or glycine, consistent with the pattern in S. enterica with 2AA-dependent growth defects. As expected based on the growth defects in monoculture, E. coli ridA and ridA tdcF mutants were outcompeted by wild type in minimal medium containing 20 mM serine, and the ridA tdcF mutant having a more severe defect than the ridA single mutant (Figure 8). The addition of isoleucine restored full fitness to each stain in the presence of 20 mM serine. The observation that TdcF had an impact on fitness in the presence of elevated serine concentration was not entirely unexpected. E. coli strains with synthetically increased 2AA levels exhibited a growth defect only when both TdcF and RidA were absent, consistent with their redundant activities 21. In total, the results with E. coli suggested that either i) 2AA levels are generally lower in E. coli than S. enterica when strains lack RidA proteins, and/or ii) 2AA is less detrimental to the metabolic network of E. coli than it is to that of S. enterica. Neither of these directly addressed the disconnect between the lack of detectable 2AA damage (i.e., IlvE activity) and competitive fitness defect of an E. coli ridA mutant in minimal medium with 5 mM serine.

–

FIGURE 6: Glycine does not restore competitive fitness to E. coli ridA tdcF double mutants under 2AA stress. Shown is competitive fitness of an E. coli ridA tdcF double mutant (∆ridA890 tdcF13::Cm) grown in coculture with wild type E. coli in minimal glucose (11 mM) medium with the indicated supplements – L-serine (5 mM), glycine (1 mM), L-isoleucine (1 mM). Data shown are mean and standard deviation of three independent biological replicates. The underlined genotype indicates the strain for which the competitive indices are shown in the corresponding graph. An asterisk (*) indicates a significant difference (p ≤ 0.05 as determined by Sidak’s multiple comparisons test) in CI values compared to control experiments in which ∆ridA890tdcF13::Cm was competed with an isogenic strain lacking ridA and tdcF (∆ridA890 ∆ tdcF14) in each medium.

RidA has a role beyond 2AA stress control in E. coli

The data gathered for ridA mutants of E. coli appeared consistent with a model in which, i) the fitness defect on 5 mM serine is due to 2AA stress, despite the stress being too low to impair IlvE activity, ii) high levels of exogenous serine (20 mM) generate enough 2AA to cause a growth defect, and iii) glycine and isoleucine restore growth via the mechanism(s) characterized in S. enterica ridA mutants. Based on this scenario, glycine was expected to suppress the competitive fitness defect of the ridA tdcF mutant. Unexpectedly, glycine failed to improve fitness and the ridA tdcF mutant remained less fit than wild type in medium with serine (5 mM) and glycine (Figure 7). Furthermore, the addition of glycine did not restore competitive fitness defects of ridA or ridA tdcF mutants in the presence of 20 mM serine, despite its ability to restore growth in monoculture (Figure 8). A possible explanation for these results is that, in E. coli RidA has a role in optimizing flux through isoleucine biosynthesis. In this scenario, isoleucine contributes to restoring fitness by satisfying a need for isoleucine created by sluggish de novo synthesis, in addition to its role in decreasing 2AA generation. The finding that glycine eliminates the 2AA-generated growth defect in monoculture, while failing to restore the fitness defect of the ridA tdcF mutant, supports the conclusion that decreased fitness of E. coli on minimal medium with serine is not solely due to 2AA stress.

–

FIGURE 7: Glycine does not restore competitive fitness to E. coli ridA tdcF double mutants under 2AA stress. Shown is competitive fitness of an E. coli ridA tdcF double mutant (∆ridA890 tdcF13::Cm) grown in coculture with wild type E. coli in minimal glucose (11 mM) medium with the indicated supplements – L-serine (5 mM), glycine (1 mM), L-isoleucine (1 mM). Data shown are mean and standard deviation of three independent biological replicates. The underlined genotype indicates the strain for which the competitive indices are shown in the corresponding graph. An asterisk (*) indicates a significant difference (p ≤ 0.05 as determined by Sidak’s multiple comparisons test) in CI values compared to control experiments in which ∆ridA890 tdcF13::Cm was competed with an isogenic strain lacking ridA and tdcF (∆ridA890 ∆ tdcF14) in each medium.

The kinetics of _ECIlvA model 2AA and 2AC production

The above data suggested that fitness defects of E. coli ridA and ridA tdcF mutants in coculture were not solely due to 2AA-mediated damage to PLP-dependent enzymes. The deaminase activities of other members of the Rid protein family improve flux through their respective pathway(s) and thus increase fitness. Based on this precedent, it was considered possible that fitness defects of E. coli mutant strains in coculture were due to impaired flux to isoleucine in the absence of RidA. For this to be the case, the data indicated flux was reduced only when exogenous serine was present. In principle, impaired flux could be due to i) decreased 2AC formation by IlvA, or ii) decreased deamination of 2AC to 2KB in the absence of RidA. Only the former possibility appeared to be amenable to an effect by serine, since competition between serine and threonine for IlvA could potentially alter the 2AC:2AA ratio. In S. enterica, kinetic models indicated that the production of 2AC from threonine was not affected by the level of serine present in vivo 39. The catalytic parameters (K_M and k_cat) of the E. coli protein (_ECIlvA) were determined with serine and threonine as substrates (Figure S1). Using these parameters of _ECIlvA to model endogenous 2AC and 2AA production in E. coli indicated that 2AC production was not impacted by the increased production of 2AA (Table 2). The ratios of 2AC:2AA predicted by modeling were consistent between S. enterica and E. coli. In total, these data show that while 2AC genertion is not predicted to differ, the amount of 2AA relative to 2AC increases dramatically in E. coli strains grown with serine. In total the data are consistent with a scenario in which 2AA stress is not the sole cause of fitness defects in strains lacking RidA homologs. Fitness defects may be exacerbated by impaired flux through isoleucine biosynthesis that results from lower 2AC deamination in the absence of RidA activity, the direct effect of serine on this flux remains unclear.

–

FIGURE 8: Glycine does not restore competitive fitness to ridA or ridA tdcF mutants in the presence of elevated serine concentrations. Shown are competitive indices of (A) an E. coli ridA mutant (ridA790::kan) and (B) a ridA tdcF double mutant (∆ ridA890 tdcF13::Cm) grown in coculture with wild type E. coli in minimal glucose (11 mM) medium with the indicated supplements – L-serine (20 mM), glycine (1 mM), L-isoleucine (1 mM). Data shown are mean and standard deviation of three independent biological replicates. The underlined genotype indicates the strain for which the CIs are shown in the corresponding graph. An asterisk (*) indicates a significant difference (p ≤ 0.05 as determined by Sidak’s multiple comparisons test) in CI values compared to control experiments in which each strain was competed with an isogenic strain lacking the relevant RidA homolog(s) in each medium.

Strains, media, and chemicals

Strains used in this study are derivatives of Salmonella enterica serovar Typhimurium LT2 or Escherichia coli K-12 and are listed in Table 3. All strains were grown at 37°C unless otherwise stated. S. enterica strains were routinely grown in Nutrient Broth (NB; Difco; 8 g/liter, 5 g/liter NaCl). E. coli strains were grown in Lysogeny Broth (LB; 10 g/liter tryptone, 5 g/liter NaCl, 5 g/liter yeast extract) as a rich medium. No-carbon E (NCE) medium with 11 mM glucose, supplemented with 0.1 mM MgSO₄ and trace elements was used as a minimal medium 41. Other supplements were added as noted to the following concentrations: L-isoleucine (1 mM), L-glycine (1 mM), thiamine (100 nM), L-aspartate (1 mM), and L-serine (5 mM). Antibiotics were used as indicated: kanamycin (Km; 50 μg/ml for rich media or 12.5 μg/ml for minimal media) and chloramphenicol (Cm; 20 μg/ml for rich media or 5 μg/ml for minimal media). Chemicals were purchased from MilliporeSigma (St. Louis, MO).

Growth analyses

Cultures were started from a single colony and grown overnight in rich medium with shaking at 37°C (2 ml). Cells were pelleted and resuspended in an equal volume of sterile saline. Cell suspensions (5 μl) were used to inoculate the indicated medium (195 μl). Growth was monitored in a 96-well plate as the change in optical density at 650 nm (OD₆₅₀) over time using a BioTek ELx808 plate reader (BioTek Instruments, Winooski, VT) with a slow shaking speed (210 rpm). Data were plotted using Graphpad Prism version 8.4. Growth experiments were performed with three independent biological replicates unless otherwise stated.

Competition assays

Competition assays were performed generally as described 22, 44, 45. Briefly, overnight cultures of each strain were grown in rich medium (2 ml) at 37°C with shaking. Cells were pelleted and resuspended in 2 ml sterile saline and optical densities at 650 nm (OD₆₅₀) were normalized to 0.1. 5 ml of the indicated medium was inoculated with 50 μl of cell suspension from each of the two competing strains. Control experiments were performed for each condition in which strains of the same relevant genotype, with different antibiotic resistance cassettes, were competed against each other. These controls accounted for differences in competitive fitness caused by the different antibiotic resistance cassettes in each strain. Colony forming units (CFUs) of cocultures were determined by spreading serial dilutions at T=0 (i.e., before incubation of the cocultures) onto media to differentiate between each respective strain. For S. enterica, media were as follows: (i) minimal medium supplemented with L-serine, selective for wild type (DM9404); (ii) minimal medium supplemented with chloramphenicol, selective for a chloramphenicol-resistant ridA mutant (DM17845), and (iii) minimal medium supplemented with kanamycin, selective for a kanamycin-resistant ridA mutant (DM3480). For E. coli, media were as follows: (i) LB supplemented with chloramphenicol, selective for chloramphenicol-resistant ridA and ridA tdcF mutants (DM14820, DM14682), and (ii) LB supplemented with kanamycin, selective for kanamycin-resistant ridA and tdcF mutants (DM14597, DM14669). Some E. coli strains (DM14520, DM15051) did not contain antibiotic resistance cassettes and could not be isolated from cocultures by spreading onto selective media. CFUs of these strains were determined by counting CFUs on LB media and subtracting the CFUs of the competing strain, which could be selected for with antibiotics. Cocultures were grown to stationary phase at 37°C with shaking. CFUs were enumerated for cultures after co-incubation as described above. The competitive index (CI) was defined and calculated as follows:

–

Each competition assay was performed at least three times with similar results. GraphPad Prism version 8.4 was used for data plotting and statistical analysis.

Protein purification

E. coli K12 ilvA was cloned into BspQI sites of pTEV18 as described previously. The resulting plasmid, pTEV18-ilvA (K12), was electroporated into BL21-AI. The resulting strain (DM18490) was grown overnight in 40 ml of Terrific Broth (TB; 12 g/liter tryptone, 24 g/liter yeast extract, 0.4% glycerol (v/v)) supplemented with 150 mg/ml ampicillin. 15 ml of the overnight culture was used to inoculate each of two 2.8-liter baffle-bottomed Fernbach flasks containing 1.5 liters of TB with 150 mg/ml ampicillin. The resulting cultures were grown at 30°C with shaking at 135 rpm to an OD₆₅₀ of 0.5 before the addition of arabinose (0.2% (w/v)) and IPTG (200 mM). Cultures were then grown at 23°C with shaking at 135 rpm for 18 h before being harvested by centrifugation at 6000 x g for 20 min. Cell pellets were resuspended in 2 ml/g cells of lysis buffer (50 mM potassium phosphate, 150 mM NaCl, 100 mM PLP, 2 mg/ml lysozyme, 0.2 mg/ml DNase, pH 7.5) and incubated on ice for 45 min. Cells were mechanically lysed using a Constant Systems OneShot Cell Disruptor at 20 kpsi. The lysate was then clarified by centrifugation at 45000 x g for 30 min before being filtered using a 0.45 mm syringe-driven filter. Filtered lysate was then loaded onto a 5 ml HisTrap HP column (GE Healthcare). The column was washed with 3 column volumes (CV) of 100% Binding Buffer (50 mM potassium phosphate, 150 mM NaCl, 20 mM imidazole, pH 7.5), followed by 3 CV of 96% Binding Buffer and 4% Elution Buffer (50 mM potassium phosphate, 150 mM NaCl, 500 mM imidazole, pH 7.5). IlvA was eluted with a linear gradient from 4-100% Elution Buffer over the course of 10 CV. A flow rate of 2 ml/min was used to each step in protein purification. Fractions containing eluted protein were assessed for purity before being concentrated using an Amicon Ultra-15 centrifugal filter (MilliporeSigma) with a molecular weight cutoff of 30 kDa. Buffer exchange of the protein into Storage Buffer (50 mM potassium phosphate, 150 mM NaCl, 10% glycerol (v/v) pH 7.5) was performed using a PD-10 column (Cytiva). Protein concentration was determined using the Peirce BCA Protein Assay Kit (Thermo Scientific). Purified IlvA was drop frozen in liquid nitrogen and stored at -80°C until use. Purity of the protein was determined by densitometer tracing to be 94% (Figure S2).

Enzyme assays

All enzyme assays were performed at room temperature.

Serine/Threonine Dehydratase assay -purified protein

The rates of 2-ketobutyrate and pyruvate formation were determined by measuring the absorbance at 230 nm39, 3, 47 every 10 s for 2 min. Reactions were carried out in a quartz 96-well plate and measurements were acquired using a SpectraMax Plus spectrophotometer (Molecular Devices). Reaction mixtures (100 ml) contained 50 mM potassium phosphate buffer (pH 7.5), 200 nM of purified IlvA. Reactions were initiated with the addition of the indicated concentrations of L-threonine or L-serine. The absorbance at 230 nm of known concentrations of either 2-ketobutyrate or pyruvate were used to generate a standard curve to convert the change in absorbance (DA₂₃₀/min) to the rate of product formation (mmol product/min) for each reaction. Data visualization and statistical analyses were performed using GraphPad Prism version 8.4.

 

The rates of 2AA and 2AC synthesis were modeled using the equation below 39. V_prod refers to the rate of product formation (2AA or 2AC) in µM/s, k_cat(Sub) refers to the catalytic constant for _ECIlvA using the relevant substrate (s^-1), [E_t] is the active site concentration of IlvA in E. coli, [Sub] is the concentration of substrate for _ECIlvA in vivo in µM, [Comp Sub] is the concentration of competing substrate in µM, K_M(Sub) is the Michaelis-Menten constant for _ECIlvA using the relevant substrate in mM, and K_{M(Comp Sub)} is the Michaelis-Menten constant for _ECIlvA using the relevant competing substrate in mM. When modeling the rate of 2AC formation, “Sub” refers to L-threonine and “Comp Sub” refers to L-serine. When modeling the rate of 2AA formation, “Sub” refers to L-serine and “Comp Sub” refers to L-threonine. 10 µM was used as the value for active site concentrations for _ECIlvA in vivo 48 and physiological concentrations of L-serine (68 µM) and L-threonine (180 µM) in E. coli 49, as well as elevated concentrations of L-serine (5000 µM and 20000 µM) were used when calculating the product formation by _ECIlvA.

–