Nitrate Reduction Stimulates and Is Stimulated by Phenazine-1-Carboxylic Acid Oxidation by Citrobacter portucalensis MBL

We did not observe PCA (PCA_red/PCA_ox $E_{{}_{\frac{1}{2}}}^{o\text{′}}=-116$ mV versus NHE [9]) oxidation in the absence of a terminal electron acceptor in either the abiotic and biotic regimes (Fig. 1A, gray curves; Fig. 1B, PCA-only column). When 10 mM nitrate (NO₃⁻/NO₂⁻ $E_{{}_{\frac{1}{2}}}^{o\text{′}}=+433$ mV versus NHE [15]) was added (right panels), C. portucalensis MBL readily oxidized PCA_red at an initial rate of −25.23 ± 0.84 μM/h (Fig. 1A, top right panel, yellow curve, and Fig. 1B). Nitrate did not oxidize PCA abiotically (Fig. 1A, top left panel, yellow curve). During both dissimilatory and assimilatory nitrate reduction, nitrate is first reduced to nitrite (21). Nitrite (NO₂⁻/NO $E_{{}_{\frac{1}{2}}}^{o\text{′}}=+350$ mV versus NHE [15]) drove slow abiotic oxidation of PCA_red at a steady rate of −1.48 ± 0.29 μM/h after an initial short period of reduction (Fig. 1A, top left panel, orange curve, and Fig. 1B). This abiotic rate of oxidation is insufficient to be responsible for the rate of PCA_red oxidation by cells in the presence of nitrate.

When C. portucalensis MBL is incubated with PCA_red and nitrite, there is less PCA_red oxidation than in the abiotic case (Fig. 1A, top right panel, orange curve). We interpret this to mean that C. portucalensis catalyzes the oxidation of PCA_red when an appropriate terminal electron acceptor is available but reduces PCA_ox when such an electron acceptor is absent. We observed PCA_ox reduction when the cells started with 200 μM PCA_ox, but the rate of reduction decreased according to the provided terminal electron acceptor (Fig. S1). Adding nitrate to the PCA_ox condition caused reduction to be nondetectable, and the presence of nitrite, which abiotically oxidizes PCA_red, caused a slight dampening in the reduction rate by 1.25 ± 0.09 μM/h (Fig. S1).

We assayed whether other gammaproteobacteria can also oxidize PCA in a terminal electron acceptor-dependent manner (Fig. 1). Using the same assay as described above, Escherichia coli MG1655, P. aeruginosa UCBPP-PA14 Δphz*, Pseudomonas chlororaphis phzB::TnluxAB, and Pseudomonas aureofaciens phzB::lacZ (all of which cannot synthesize PCA either naturally or due to the specified mutations) were incubated with PCA and 10 mM nitrate or nitrite, or no terminal electron acceptor. With the exception of P. chlororaphis and P. aureofaciens, all strains exhibited PCA oxidation with nitrate, with C. portucalensis MBL being the fastest (Fig. 1B). The assayed P. aureofaciens and P. chlororaphis oxidized PCA_red with nitrite faster than the abiotic control: −2.75 ± 0.80 and −3.10 ± 0.58 μM/h, respectively, versus −1.48 ± 0.29 μM/h abiotically (Fig. 1B). The reduction assay for the other strains reflected our observations with C. portucalensis MBL: when a terminal electron acceptor that stimulated PCA_red oxidation was present, no reduction was observed (Fig. S1). None of the strains exhibited significant growth in these assays (Fig. S2 and S3).

In Fig. 1A, it is clear that C. portucalensis MBL completes its oxidation of ∼200 μM PCA_red within 10 h when nitrate is available. We repeated this experiment in anaerobic culture tubes inoculated with C. portucalensis MBL, 10 mM NO₃⁻, and either 200 μM PCA_red, 200 μM PCA_ox, or no PCA and measured nitrate and nitrite concentrations over time via ion chromatography. We observed that PCA_red oxidation significantly increased the rate of nitrate reduction to nitrite (Fig. 2A). Nitrate consumption was stoichiometrically matched by nitrite production. We did not observe the production of any other nitrogen oxides or ammonium (data not shown). The no-PCA control did not show any nitrate reduction over the first 8 h (3 ± 55 μM/h; 95% confidence interval reported for all rate measurements, calculated from a linear regression of the data from the first two time points for each curve in Fig. 2A). During this time, 131 ± 49 μM/h nitrate was reduced in the PCA_red condition. In contrast, the PCA_ox control exhibited only a putative increase in nitrate reduction (35 ± 35 μM/h). To verify that nitrate was reduced to nitrite, we tracked nitrite production. We observed that over the first 8 h the no-PCA control produced nitrite at the rate of 22 ± 3 μM/h, the PCA_ox control at 58 ± 2 μM/h, and the PCA_red condition at 147 ± 44 μM/h of nitrite over the first 8 h. Thus, we estimate the effect of PCA_red versus PCA_ox to be 96 ± 60 μM/h of increased nitrate reduction or 89 ± 44 μM/h of increased nitrite production. The increase in nitrate reduction due to PCA_red was greater than the absolute number of electrons the PCA_red could provide: PCA redox and nitrate reduction are both two-electron processes (9, 21), and a process without a redox cycle would predict that oxidizing PCA_red at a rate of 25 μM/h (Fig. 1B) would reduce at most 25 μM/h nitrate to nitrite. However, the lowest range of the confidence intervals suggests that at least an extra 36 μM/h nitrate was reduced by the cells when PCA_red was provided, implying that PCA is stimulating nitrate reduction by some other means than a one-to-one electron donation; our data showing PCA reduction capability (Fig. S1) suggests that the cells may be cycling PCA under these conditions, though this conclusion awaits definitive demonstration.

The effect of PCA oxidation by C. portucalensis MBL on its rate of nitrate reduction was outsized (Fig. 2A). This is consistent with two explanations. (i) A prior report argues that neutral red (a synthetic phenazine) oxidation affects electrosynthesis during anaerobic respiration primarily by changing gene regulation via menaquinone reduction (16), and so it is plausible that PCA oxidation may increase transcription of a rate-limiting factor in the electron transport chain to nitrate. (ii) We observed that a PCA redox cycle by C. portucalensis MBL is possible: the cells naturally reduce PCA_ox, but this is not detectable while nitrate is present, potentially meaning that the PCA is being reoxidized as soon as it is reduced. If this is the case, it is possible that PCA redox cycling may also shunt more electrons to nitrate reduction. An alternative explanation for why PCA_ox reduction is not observed in the presence of nitrate is that the cells are preferentially reducing nitrate instead of PCA. While we cannot directly test this hypothesis given our current experimental set up, this will be possible once we develop methods to genetically modify C. portucalensis MBL. We have not been able to determine whether reduced PCA (PCA_red) serves as an effective electron donor to the cell’s metabolism from our above observations. Even if there is no direct physiological benefit from PCA oxidation, that cells can catalyze this reaction and stimulate nitrate reduction has environmental ramifications far beyond the cell.

Despite all being capable of nitrate reduction, not all the gammaproteobacteria that we tested were capable of oxidizing PCA_red with nitrate (Fig. 1), suggesting certain enzymes are responsible for this process. We are developing genetic tools in C. portucalensis MBL to test this directly. Nitrate-dependent PCA_red oxidation is likely to be common in anoxic environments, such as soils, as it is not species specific and occurs readily. Our observed rate of PCA_red oxidation (25 μM/h) is significant relative to the measured concentration and half-life of PCA in the rhizosphere (∼2.5 μM and ∼3.5 days, respectively) (22). Allowing that the conditions of our experiments may have resulted in an overestimate of the natural PCA_red oxidation rate in the environment, even if the laboratory PCA_red oxidation rate is orders of magnitude faster than in nature, this process would still be environmentally relevant. Furthermore, if the outsized effect of PCA_red oxidation on nitrate reduction that we observed generalizes to other species, the production and reduction of phenazines by organisms like pseudomonads likely affect the rate of nitrate consumption in their environs, adding another function to the phenazine arsenal (18). We propose that cells may catalyze a PCA redox cycle whenever they have internal stores of reducing equivalents and a usable terminal electron acceptor (Fig. 2B). While both intracellular (e.g., sulfur) and extracellular (e.g., humic) bacterial redox cycles have been described (23, 24), to our knowledge this has not been appreciated for secreted redox active metabolites, such as phenazines. Our observation underscores the possibility that these molecules may act as “electron buffers,” enabling cells to reduce and oxidize them according to whether they are lacking a terminal electron acceptor or an electron donor, respectively, and in so doing significantly impact other biogeochemical cycles, such as the nitrogen cycle.

Citrobacter portucalensis MBL was isolated in our previous work (20). In the comparative PCA oxidation and reduction experiments, we used strains of gammaproteobacteria that cannot synthesize phenazines, either natively or due to mutations. We used E. coli MG1655, P. aeruginosa UCBPP-PA14 Δphz* (25), P. chlororaphis phzB::TnLuxAB (strain PCL1119 obtained from G. Bloemberg, Leiden University [26]), and P. aureofaciens phzB::lacZ (strain 30-84Z obtained from L. Pierson, University of Arizona [27]). The wild-type pseudomonads can synthesize PCA, but the mutants used cannot. All strains were grown and incubated under the same conditions. The basal medium for the experiments contained 20 mM potassium phosphate buffer (final pH 7), 1 mM sodium sulfate, 10 mM ammonium chloride, 1× SL-10 trace elements, 1× freshwater salt solution (17.1 mM sodium chloride, 1.97 mM magnesium chloride, 0.68 mM calcium chloride, and 6.71 mM potassium chloride), and 1× 13-vitamin solution (10 μM morpholinepropanesulfonic acid [MOPS] [pH 7.2], 0.1 μg/ml riboflavin, 0.03 μg/ml biotin, 0.1 μg/ml thiamine HCl, 0.1 μg/ml l-ascorbic acid, 0.1 μg/ml d-Ca-pantothenate, 0.1 μg/ml folic acid, 0.1 μg/ml nicotinic acid, 0.1 μg/ml 4-aminobenzoic acid, 0.1 μg/ml pyridoxine HCl, 0.1 μg/ml lipoic acid, 0.1 μg/ml NAD, 0.1 μg/ml thiamine pyrophosphate, and 0.01 μg/ml cyanocobalamin). Depending on the experimental condition, as indicated in the figure legends, a terminal electron acceptor would be added (10 mM nitrate or nitrite) or omitted. For oxidized PCA, a 10 mM stock in 20 mM NaOH was prepared. For reduced PCA, an 800 μM stock in the basal medium was reduced by electrolysis. Both stocks were diluted into wells in 96-well plates (BRAND Cat./No 781671) to a final target concentration of 200 μM.

All cell incubations and experiments were performed at 30°C. Cells were preserved in 35% glycerol stocks at −80°C. Two days prior to the experiments, frozen cells for each strain assayed were struck out on lysogeny broth (LB) agar plates and incubated overnight. The evening prior to the experiment, a patch from the streaks was inoculated into liquid LB in a respective culture tube and incubated slanted, shaking at 250 rpm, overnight. The morning of the experiment, 1 ml of each cell culture was washed three times into the basal medium by spinning for 2 min at 6,000 × g, aspirating the supernatant, and gently resuspending with a pipette. The optical density at 600 nm (OD₆₀₀) of each washed culture was measured. The cultures were brought into a Coy glove box, where they were washed three times into the same basal medium that had been made anoxic, following the same procedure as described above. After being left to stand for 1 to 2 h, the cells were inoculated into the different experimental conditions at a target starting OD₆₀₀ of 0.1.

All PCA redox measurements were performed in a Coy chamber (5% hydrogen/95% nitrogen headspace) using a BioTek Synergy 5 plate reader. Reduced PCA concentration was measured by fluorescence (excitation 360 nm and emission 528 nm) (28). Plates were incubated shaking on the “medium” setting. For nitrogen oxide concentration measurements, C. portucalensis MBL cells were prepared as described above but incubated in culture tubes in the Coy chamber to allow for sampling. In these cultures, no acetate was provided. The tubes were kept at 30°C, but not shaking. Samples were filtered through a 0.2-μm cellulose-acetate spin filter and stored at −80°C prior to analysis. Nitrate and nitrite concentrations were measured by ion chromatography using a Dionex ICS-2000 instrument.

Initial redox rates under the conditions with supplied nitrite did not include the first 1.5 h of the assay because that period showed a systematic artifact due to fluorescence quenching (Fig. 1A, lower left panel). The linear fits were calculated over the first 5 h after the first detectable PCA_red measurement, which was determined to be appropriate based on scanning for R² values over increasing time frames (Fig. S4 and S5). Ninety-five percent confidence intervals are calculated as the estimated value ± 1.96 × standard error. When comparing rates, the reported error is the geometric mean of the intervals for the two measurements. All plots were generated using Bokeh, and the legends and titles were adjusted using Inkscape. All the raw data and the Jupyter notebook used for their analysis are available at https://github.com/ltsypin/Cportucalensis_observation.