Some microorganisms that endogenously uptake electrons from extracellular electron donors can also uptake electrons from cathode poised at an equivalent redox potential, coupling this electron uptake to reduction of an electron acceptor (Rowe et al., 2015). To determine if L. plantarum can uptake electrons from a cathode, we first tested whether it could consume current under different cathode potentials with an external physiological electron acceptor (nitrate) present (Brooijmans R. J. W. et al., 2009) in a bioelecrochemical reactor (Supplementary Figures S1A,B). To promote growth without usurping the role of the cathode as the electron donor, we supplied 5 mM glucose as a carbon source since under these conditions L. plantarum minimally utilizes the glucose to reduce nitrate (Brooijmans R. J. W. et al., 2009) (if the electron transport chain is reconstituted). We detected current consumption in a minimal medium (Aumiller et al., 2023) by L. plantarum under all the potentials tested (−0.15, −0.25, −0.40 and − 0.60 V), with the highest current at −0.60 V (all potentials are reported vs. Ag/AgCl_{KCl sat}) (Supplementarys Figure S1A,B). No current consumption was observed with heat killed cells under identical conditions (Supplementary Figure S1C), indicating that electron uptake required metabolically active cells. We also observed a higher and earlier current consumption when glucose was supplied (Supplementary Figure S1D), suggesting that cells could maintain a higher metabolic activity that resulted in a greater EET from the cathode. Because of all these observations, we chose to use a potential of −0.6 V and to provide 5 mM glucose to the medium for further experiments to maximize EET from the cathode.

To probe whether this electron uptake was coupled to the reduction of an electron acceptor, we next monitored current in the presence and absence of nitrate (Figure 1A). Approximately 18 h after inoculation of cells with nitrate, a maximum current consumption of −20 ± 5 μA/cm² at −0.60 V (Figure 1A) was observed. However, no significant current consumption occurred in the absence of nitrate (Figure 1A), indicating that the extracellular electron uptake required an external electron acceptor. The electron uptake also coincided with the reduction of nitrate to nitrite (Figure 1B). We compared these data to nitrate reduction by L. plantarum under open circuit (OC) conditions, in which the cathode is not electrically connected to the anode and thus electrons from electrical current cannot be supplied. Under OC conditions, L. plantarum reduced significantly less nitrate to nitrite (1.19 ± 1.16 mM vs. 8.98 ± 2.29 mM per day, p < 0.050) (Figure 1B). Using differential pulse voltammetry (Supplementary Figure S1E), we identified a redox active center at −0.52 V (pH = 6.3, vs. Ag/AgCl sat. KCl) which was not observed when nitrate or cells were not present in the medium. Thus, this result confirms the existence of a redox process associated with the bioelectrochemical reduction of nitrate or nitrite, similar to other observations (Pous et al., 2014). Further experiments indicated that nitrite could be biotically and abiotically reduced in the reactors at −0.6 V (Supplementary Figure S2A). The addition of nitrite in a bioelectrochemical reactor with heat killed cells showed a significant current consumption, which was slightly breater when viable cells were present (Supplementary Figure S2A) (ΔE = −0.6 V). Moreover, 1.3 mM of nitrite was consumed over 3 days in the presence of heat killed cells while 1.6 mM was consumed when viable cells were added (Supplementary Figure S2B). To understand the use of the cathode by L. plantarum to reduce nitrite, we minimized the abiotic electrochemical signal by reducing the cathode potential to −0.25 V and −0.4 V. Under these cathode potentials, the addition of nitrite led to a minimal current consumption, and adding viable cells did not vary current consumption levels (Supplementary Figure S2C), while significant nitrite was reduced (Supplementary Figure S2D). Thus, the nitrite biological reduction was mainly occurring without the mediation of the cathode (>91%) (see SI for electron balances). This indicated that L. plantarum predominately uses electrons from a cathode to drive the reduction of nitrate to nitrite, but a separate pool of electron donors is used by L. plantarum to reduce nitrite.

Through its fermentative pathways, L. plantarum produces organic molecules, e.g., acetate, that can also serve as electron donors. To establish to what degree the cathode vs. endogenous molecules served as the electron donor for L. plantarum to reduce nitrate, we quantified the electron uptake needed for nitrate reduction. We first calculated the electrons consumed from the cathode from the current consumption data and compared this value to the number of electrons theoretically needed to reduce the consumed nitrate to nitrite in the bioreactor medium (Figure 1B). Significantly more electrons were used to reduce nitrate than electrons supplied by the cathode: only 6.9 ± 1.6% of the electrons needed for nitrate reduction were supplied by the cathode (i.e., coulombic efficiency). This difference indicates that additional electron donors, such as an organic fermentation intermediates, may be utilized by source for L. plantarum to reduce nitrate.

When L. plantarum donates electrons to an extracellular anode, it conserves energy more rapidly and increases its fermentative flux (Tejedor-Sanz et al., 2022). To probe the impact of the electron uptake on energy conservation and metabolism, we measured metabolites of L. plantarum under OC and cathodic EET conditions from glucose fermentation, employing this time 10 mM of sugar to facilitate the analyses of metabolites. Under EET and OC conditions, L. plantarum predominantly produced the end products of fermentation lactate, ethanol, acetate and succinate (Figures 2A,B). Formate (<0.02 mM) and pyruvate (<0.06 mM) were detected in all bioreactors at trace levels, and no acetoin or 2,3-butanediol were detected at any time. By day 5, these fermentation products contained >98% of the initial carbon supplied as glucose, indicating these measurements provide a complete view of L. plantarum carbon metabolism under these conditions.

Interestingly, the fermentation process was affected by electron uptake in L. plantarum. Over the first 1.7 days, fermentation products were similar under EET and OC conditions (Figure 2A). Glucose was fully consumed in all reactors after 24 h (Supplementary Figure S3A), lactate was the main fermentation product (22.5 ± 0.1 mM), and the nitrate concentration was unchanged (Supplementary Figure S3B). After day 1.75, however, there was a gradual increase in current consumption accompanied by nitrate consumption (238 μmol day ⁻¹ vs. 5.5 μmol day ⁻¹ under OC conditions) (Figure 2A; Supplementary Figure S3B). Moreover, we observed a significant increase in production of acetate and succinate and an increased consumption of lactate under EET conditions. For instance, lactate consumption rates were of 35.8 μmol day ⁻¹ under EET conditions and of 9.8 μmol day ⁻¹ under OC conditions. Thus, electron uptake and nitrate reduction coincided with a shift of L. plantarum metabolism toward a more heterogeneous distribution of end products (Figure 2B, 46% lactate, 38% acetate, 15% ethanol and 2.3% succinate) than under OC conditions, in which lactate remained the dominant product. This strongly suggests that, after exhausting glucose, L. plantarum utilizes the cathode as an electron donor and nitrate as an electron acceptor, allowing lactate to be sequentially oxidized to pyruvate then acetate (Figure 2D). The absence of nitrate or the polarized cathode did not induce the metabolic shift, probing that this phenomenon is associated with electron uptake to reduce nitrate (Figure 2B). We also observed this shift in the spectrum of end-metabolites when we repeated the test using the lower concentration of glucose (5 mM) we employed in previous assays (Supplementary Figures S3C,D, S5). In fact, even higher relative levels of acetate (41%) were detected and lower levels of lactate (29%), while we observed an enhanced current consumption in the bioelectrochemical reactors. This suggests that the higher the electron uptake, the more pronounced the metabolic shift is.

Our results indicate a complex electron flow during extracellular electron uptake: the cathode and an endogenous fermentation product (lactate) serve as electron donors, while exogenous nitrate acts as an electron acceptor. To understand the impact of this electron flow on the intracellular redox state of L. plantarum, we analyzed how the reducing equivalents generated from glycolysis, i.e., NADH, and the electrons from the cathode were used. For this analysis, we estimated the NADH regenerated via fermentation and via the reduction of nitrate using the measured concentrations of end-fermentation products and nitrate (see Methods section for methodology). Under OC conditions, we estimated that a 81%, a 18% and a 0.7% of all the NADH was regenerated via lactate ethanol and succinate production, respectively (Figure 2C), in agreement with prior observations of homofermentation (Ferain et al., 1996). However, when cells consumed electrons from the cathode (EET + nitrate), NADH regeneration via lactate was reduced to 45, 30% via ethanol, 21% of the via the reduction of nitrate to nitrite and a 4% via succinate (Figure 2C). Thus, electron uptake by L. plantarum changes the strategy that it uses to balance its intracellular redox state.

Extracellular electron transfer can affect catabolic processes and how they are coupled to anabolic processes, including cell growth. To examine how electron uptake impacted L. plantarum anabolism, we measured cell viability and biomass under EET and OC conditions. Even though we did not observe visible cell attachment, for quantifying growth we employed a graphite rod, an electrode with a smoother surface than graphite felt, to reduce cell attachment. When L. plantarum cultures produced the maximal current density (day 2), they contained a higher number of viable cells compared to OC conditions (Figure 3A). However, maximal current consumption (Supplementary Figure S4A) coincided with the entry of L. plantarum into the stationary. This observation suggests that when L. plantarum anabolism slows due to substrate depletion (glucose), electron uptake increases. Further supporting this observation, there were no significant differences in the cell density over time (Supplementary Figure S4A) or final biomass (Figure 3B) between OC and EET conditions, both in the presence and absence of nitrate (Supplementary Figure S4B). Thus, electron uptake coupled to nitrate reduction increases cell viability, but does not increase final biomass. We hypothesize that metabolic shift observed toward acetate production when electron uptake occurs may generate extra ATP, providing the cells with the additional energy needed for survival in the stationary phase and under glucose depravation. However, it may not be enough to support significant, if any, additional growth.

We next explored the molecular mechanisms supporting electron uptake. We studied whether cathodic EET requires molecules that activate aerobic and anaerobic respiration, heme and 1,4-dihydroxy-2-naphthoic acid (DHNA), respectively (Brooijmans R. et al., 2009), by adding these molecules exogenously and examining current consumption and nitrate reduction (Figure 4A). Heme addition did not affect the magnitude of the current consumed, but delayed the start of electron uptake by cells (Figure 4B). Heme supplementation did not affect nitrate reduction either (Figure 4C). Similarly, the addition of DHNA (Figures 4D,E) and omission of riboflavin (Figures 4F,G) did not impact current consumption nor nitrate reduction. Differential pulse voltammetry on the cathode of the riboflavin-free bioreactors also probed that residual riboflavin was not acting as a freely-diffusing electron shuttle since the characteristic redox peak of flavins at −0.4 V was absent from the cathodic chamber (inset of Figure 4E). This result is in accordance with the inability of L. plantarum to synthesize riboflavin (Brooijmans R. J. W. et al., 2009). Thus, the extracellular electron uptake machinery does not require any of the external cofactors, i.e., heme, quinone-precursors, or free riboflavin, necessary to activate the other known electron transfer chains in L. plantarum, other LAB and EET in other species (Wu et al., 2014; Mevers et al., 2019).

Those results indicate that L. plantarum uses electrons from a cathode directly, or via a mediated approach by either using other molecules present in the culturing medium acting as electron shuttles or by self-synthesizing them. We found that cysteine, a redox active compound that can behave as an electron shuttle between other electroactive species and minerals (Doong and Schink, 2002; Liu et al., 2014) was present at 0.8 mM in the medium used. Removing cysteine from the medium resulted in similar current consumption levels by L. plantarum than when cysteine was added to the medium (Supplementary Figures S5A,B). Thus, cysteine is not a mediator for cathodic electron uptake. Since no other compounds present in the CDM are known to act as electron shuttles, we posit that cells were performing direct EET with the cathode or were using a yet unidentified self-synthesized shuttle molecule or detached protein.

We next sought to further understand the mechanism involved by exploring the role of different proteins in EET in L. plantarum. Lactic acid bacteria, including L. plantarum, possess a flavin-based extracellular electron transfer (FLEET) locus (Supplementary Figure S6A) that is associated with an ability to perform EET with insoluble electron acceptors (Light et al., 2018; Tejedor-Sanz et al., 2022; Tolar et al., 2023). Because certain proteins in other bacterial species can be used for both interacting with insoluble donors and acceptors (Xie et al., 2021), we interrogated the role of certain genetic elements within the FLEET locus in the cathodic EET chain. Since DHNA is not required for electron uptake, we reasoned that DmkB (heptaprenyl diphosphate synthase complex II), an enzyme which catalyzes terminal steps in the production of menaquinone-7 from DHNA, is unlikely to be required for it. Instead, we targeted Ndh2 (the type II NADH-dehydrogenase), PplA, EetA, and EetB proteins, since they have a role in reducing insoluble electron acceptors in this species (Tejedor-Sanz et al., 2022; Tolar et al., 2023). Electron uptake, quantified as total charge consumed from the cathode (Supplementary Figure S6B), and nitrate reduction by Δndh2, ΔpplA, and ΔeetA/B strains (Supplementary Figures S6C,D) were similar to wild-type. These results indicate PplA, Ndh2 and the proteins EetA and EetB are not components of the cathodic EET chain. Because of this finding and our previous result showing that riboflavin is not required for electron uptake, we posit that FmnA and FmnB, proteins involved in the flavinilyzation of PplA (Light et al., 2018) and with their respective coding-genes within the FLEET locus, are unlikely to be essential elements in the cathodic EET chain. Overall, this analysis indicates that FLEET proteins are not redox enzymes mediating the electron uptake capacity in L. plantarum, confirming that the cathodic and anodic EET mechanisms are genetically different.

Since in our bioreactors EET from a cathode is nitrate-dependent, we next studied the role of the nitrate reductase A, the only enzyme known to catalyze nitrate reduction in L. plantarum. To this end, we tested the electron uptake capacity of a strain that lacks the operon ΔnarGHJI encoding the sole nitrate reductase. We observed that current consumption and nitrate reduction were completely inhibited in the ΔnarGHJI strain (Figure 5A; Supplementary Figure S7A). Furthermore, the catalytic redox center with a midpoint potential of −0.52 V that we previously observed in wild type strain (Supplementary Figure S1D) was absent from the voltammograms obtained using ΔnarGHJI strain (Supplementary Figure S7B). Complementation of narGHJI via plasmid expression in the deletion background (narGHJI⁺ strain) restored the current consumption and nitrate reduction levels capacity to wild type levels (Figure 5A; Supplementary Figure S7C). Thus, nitrate reductase is essential for the nitrate reduction-dependent cathodic EET in L. plantarum.

To understand how electrons were transported across this enzyme during EET from a cathode, we analyzed the requirement of each subunit of nitrate reductase A in the process. This protein is membrane bounded and contains 3 subunits: NarI (subunit γ), NarH (subunit β), and NarG (subunit α) (Sparacino-Watkins et al., 2014; Figure 5D). When L. plantarum employs a soluble electron donor to perform anaerobic respiration on nitrate, both an external quinone and heme source are needed to restore NarI activity to receive electrons from a NADH-dehydrogenese (Brooijmans R. J. W. et al., 2009). Electrons are then transferred to NarH to the NarG subunit, the catalytic site of the enzyme (Figure 5D). Since L. plantarum is heme auxotroph and NarI is a bd-cytochrome described to accept electrons from membrane-embedded menaquinones (Guigliarelli et al., 1996), the transport of electrons cannot occur in the absence of exogenous heme and quinones. Thus, electrons from the cathode cannot be transported through NarI subunit. We then studied the electron transport across NarH and NarG during electron uptake from a cathode. These subunits harbor Fe-S clusters, described to be electron transport centers in nitrate reductase A in L. plantarum and other species (Guigliarelli et al., 1996; Sparacino-Watkins et al., 2014). We disrupted electron transport across narH by mutating specific cysteine residues within NarH that bind three 4Fe-4S to alanine, a mutation that results in the loss of several Fe-S clusters and full enzymatic activity (Guigliarelli et al., 1996). To build this mutant we constructed a plasmid to overexpress the narGH*JI genes in a ΔnarGHJI background. Eliminating electron transport (Figure 5B) of NarH subunit did not affect current consumption capacity in L. plantarum nor nitrate reduction (Supplementary Figure S7C). This suggests that electrons do not flow through the Fe-S clusters of the NarH subunit when provided by a cathode. Thus, these results imply that electrons flow only via NarG subunit within the nitrate reductase when L. plantarum consumes electrons from a cathode.

Finally, since no redox-molecule was found to mediate electron uptake pin our previous tests, we next probed the possibility of L. plantarum employing an EET-mediated strategy to interact with the cathode by releasing detached proteins, like nitrate reductase, as previously observed for hydrogeneses and formate dehydrogenes in other species (Deutzmann et al., 2015). To do so, we performed a media swap assay to test current consumption capacity of cell-free medium (0.2 μm filter employed) from a current consuming bioreactor with cells and in the presence of nitrate (Figure 5C). This filtered medium could not support the current consumption levels observed when L. plantarum cells were present, indicating that free enzymes were not or minimally responsible for the electron uptake. We associate the background current with that one associated with abiotic nitrite reduction and a possible presence of oxygen traces. All together, these observations indicate that EET from a cathode involves the presence of whole cells with the nitrate reductase A, and NarG is the key enzyme for electron uptake. We therefore propose a model for this EET chain in L. plantarum in which a fraction of the electrons used to reduce nitrate flow from the cathode to the nitrate reductase A and, specifically, to NarG subunit through a yet unknown intermediate step(s) (Figure 5D).

Lactiplantibacillus plantarum NCIMB8826 strains were grown overnight (aprox. 18 h) at 37°C from glycerol stocks in MRS commercial media (BD, Franklin Lakes, NJ, USA) under static (non-aerated) conditions. Cells were harvested by centrifugation (5,250 g, 12 min, 4°C) and washed twice in 1xPBS. When L. plantarum wild type EET activity versus the mutants was compared, cells were grown as described and the number of cells was normalized across the strains prior to its inoculation in the electrochemical cells. For growing the strains in the bioelectrochemical reactors we used a Chemically Defined Media (CDM) (Aumiller et al., 2023) (see composition in SI) containing 5 mM of glucose and 10 mM of nitrate. The media was adjusted to pH 6.5 and then filter sterilized. For the tests for studying growth of L. plantarum, the strain containing the empty vector pSIP403, and the CDM also contained erythromycin (10 μg/mL). The complementation strains of narGHJI and the narGH*JI mutated strains were grown with 10 μg/mL erythromycin.

The strains, plasmids, primers and DNA fragments used in this study are listed in Supplementary Tables S2–S4, respectively. L. plantarum NCIMB8826 wild type, pplA, ndh2 and narGHJI strains were kindly provided to us by Maria Marco group at the UC Davis. L. plantarum NCIMB8826 eetA/B deletion mutant was constructed by CRISPR-Cas9 toolbox according to Huang et al. (2019). Briefly, the upstream and downstream homologous arms were amplified from genomic DNA of L. plantarum (see Supplementary Table S4 for primers). The crRNA targeting eetA/B was designed by using CRISPOR.¹ The sgRNA was synthesized as an oligo to insert into the pHSP02 backbone. All the fragments were assembled in ApaI-XbaI digested pHSP02 editing plasmid to create pSTS02 by Gibson assembly (Gibson et al., 2009). For CRISPR editing, L. plantarum strain harboring helper plasmid pLH01 was induced with 100 ng/mL Sakain P peptide (GenScript, Piscataway, NJ) for RecE/T expression and was subsequently prepared as competent cells. The pSTS02 was then delivered into L. plantarum NCIMB8826 by electroporation. The deletion mutants were selected on MRS plates containing 10 μg/mL erythromycin and 10 μg/mL chloramphenicol. Colony PCR was performed to confirm the deletion of the target gene (see Supplementary Table S4 for primers).

The complementation of narGHJI was achieved by using pSIP403 plasmid (Sørvig et al., 2003). The narGHJI gene fragments were amplified from genomic DNA of L. plantarum NCIMB8826 (see Supplementary Table S4 for primers) and was cloned into NcoI-EcoRI digested pSIP403 backbone by Gibson assembly (Gibson et al., 2009) to generate pSL02. The narH [4Fe-4S] Cys > Ala mutation was constructed based on pSL02. Gene fragment containing the Cys > Ala mutations was synthesized as a gBlock (Twist Bioscience, San Francisco, CA) and was inserted into pSL02 by Gibson assembly to create pSL06. All plasmids were verified by Sanger sequencing and were delivered in L. plantarum NCIMB8826 by electroporation.

The bioreactors consisted of double chamber electrochemical cells (Adams & Chittenden, Berkeley, CA) with a cation exchange membrane (CMI-7000, Membranes International, Ringwood, NJ) that separated them. We used a 3-electrode configuration consisting of an Ag/AgCl sat KCl reference electrode (BASI), a titanium wire counter electrode, and a working electrode of either 6.35-mm-thick graphite felt working electrode of 4×4 cm (Alfa Aesar) or a graphite rod of 1.2 cm of diameter and 3 cm length (used for cell viability tests), both with a piece of Ti wire threaded as a current collector and connection to the potentiostat. We used a Bio-Logic Science Instruments potentiostat model VSP-300 for performing the electrochemical measurements. The bioreactors were sterilized by filling them with ddH₂O and autoclaving at 121°C for 30 min. After this, each chamber media was replaced with 150 mL of filter sterilized CDM media (for the working electrode chamber), and 150 mL of M9 media (BD) (for the counter electrode chamber). When indicated, nitrate (aprox. 10 mM) was supplemented as electron acceptor and glucose as carbon source (5 or 10 mM). When appropriate, the CDM was supplemented (before filter-sterilized) with 20 ug/mL of DHNA diluted 1:1 in DMSO: ddH₂O where appropriate or with heme at a final concentration of 10 ug/mL (diluted 1:1 in DMSO: ddH₂O). Riboflavin was omitted from the mineral stock solution of the CDM in the test in which the effect of the presence of flavins in the medium was studied.

The media in the working electrode chamber was mixed with a magnetic stir bar for the course of the experiment and N₂ gas was continuously purged in the working electrode chamber to maintain anaerobic conditions. The applied potential to the working electrode was of −0.6 V versus Ag/AgCl (sat. KCl) (BASI). All the experiments were tested under 30°C. Reactors at open circuit (OC) were similarly assembled and used as control for non-current circulating conditions. Once the current stabilized, the electrochemical cells were inoculated to a final OD_600nm of 0.1–0.15 with the cell suspensions prepared in M9 medium. Current densities are reported as a function of the geometric surface area of the electrode (16 cm² for carbon felt). Differential Pulse Analyses tests were performed from −0.75 V to 0.4 V, 50 mV pulses height, 500 ms pulses width, 1 mV step height and 1,000 ms step time. In the experiment where metabolites production was tested, we inoculated with an initial OD_600NM = 0.05 to slow the consumption of glucose and examine metabolites production within time.

The bioreactors were sampled under sterilized conditions at different time points for subsequent analysis. The samples for organic acids analyses were centrifuged (15,228 g for 7 min) and the supernatant was separated for HPLC assessments.

Bioreactors were prepared using the same methodology as before, with the CDM medium containing 5 mM glucose and 10 mM of nitrate. We used an erytromycin resistant strain (10 ug/mL) as electron uptaking strain and ΔnarGHJI as control. We cultured the strains in a BES with the cathode polarized at ΔE_cathode = −600 mV_{Ag/AgCl, sat. KCl, and after aprox.} 42 h, we collected and filtered the medium (0.22 μm) from each BES, and used add it all into a new abiotic BES to measure current consumption in the absence of cells.

Organic acids, ethanol and sugar concentrations were measured by HPLC (Agilent, 1,260 Infinity), using a standard analytical system (Shimadzu, Kyoto, Japan) equipped with an Aminex Organic Acid Analysis column (Bio-Rad, HPX-87H 300 × 7.8 mm) heated at 60°C. The eluent was 5 mM of sulfuric acid, used at a flow rate of 0.6 mL min⁻¹. We used a refractive index detector 1,260 Infinity II RID. A five-point calibration curve based on peak area was generated and used to calculate concentrations in the unknown samples. We included the following standards in the HPLC measurements: acetate, formate, pyruvate, malate, lactate, succinate, oxalacetate, fumarate, ethanol, acetoin, butanediol, mannitol and glucose. No gaseous products were measured.

Bioreactors were shaken to remove the cells attached to the working electrode and afterwards sampled to measure total biomass (dry weight). Dry weight was determined after taking 75 mL of sample of the bioreactor supernatant after the current consumption was significantly reduced. Pellets were harvested by centrifugation (5,250 g, 12 min, 4°C) and washed twice in ddH2O (50 mL/sample/wash) to remove the salts. Afterwards the suspended pellets were transferred to 1.5 mL Eppendorf tubes, centrifuged again and let dry at 30°C using an evaporator. The microfuge tubes were then cooled in a desiccator for 30 min and the weight of each tube was measured to determine cell weight. The difference between the weight of each tube with the pellet and before containing it allowed us to determine the dry weight mL⁻¹. Samples for CFU enumeration were collected under sterile conditions at the time of inoculation and at the time of approximately maximum current density. Samples were serially diluted (1:1,000 to 1:10,00,000) in sterile PBS, and plated on MRS for CFUs enumeration after overnight incubation at 30°C.

Nitrate and nitrite were measured by either ion chromatography and/or with an enzymatic kit. The chromatograph used was a Dionex ICS-2100 system (Thermo Scientific, USA). The enzymatic kit used was a Cell Biolabs’ OxiSelect™ Nitric Oxide (Nitrite/Nitrate) Assay Kit (Cell Biolabs, USA). Nitrate in the samples was converted to nitrite by nitrate reductase enzyme and then total nitrite was detected with Griess reagents as a dye product.

The coulombic efficiency or electron recovery on the cathode was calculated as the fraction of the electrons consumed on the cathode over the total theoretical electrons needed to reduce nitrate to nitrite. The electrons consumed on the cathode were estimated by integrating the area (charge) under the chronoamperometric curve [current response (A) over time (s)]. This obtained charge was then converted to mol of electrons using the Faraday constant (96,485.3 A*s mol⁻¹ electrons) to obtain the experimental electron consumption. We estimated the total theoretical electrons used to reduce all the nitrate measured to nitrite considering that 2 mol of electrons are needed to reduce 1 mol of nitrate.

Total NADH produced was calculated based on the known stoichiometry of the fermentation pathways to assume 1 mol of NADH is produced for each mole of lactate, acetate, ethanol, pyruvate, succinate and formate detected from glucose fermentation. For EET pathways, the concentration of additional NADH generated from the electron supplies by the cathode was calculated by determining the moles of electrons supplied by the cathode, as previously described (Tejedor-Sanz et al., 2020). Under OC conditions, all NADH was generated from fermentation, while under EET conditions, NADH was generated from fermentation and from the cathode electron supply. L. plantarum can consume NADH in three ways: by reducing pyruvate to lactate (1 mol NADH/mol lactate produced) or to ethanol (2 mol NADH/mol ethanol produced), by reducing pyruvate or phosphoenolpyruvate to succinate via the reduced branch of the tricarboxylic acid cycle (2 mol NADH/mol succinate produced), and by reducing nitrate to nitrite (1 mol of NADH/mol nitrate consumed).