A flavin-based extracellular electron transfer mechanism in diverse Gram-positive bacteria.

Extracellular electron transfer (EET) describes microbial bioelectrochemical processes in which electrons are transferred from the cytosol to the exterior of the cell1. Mineral-respiring bacteria use elaborate haem-based electron transfer mechanisms2-4 but the existence and mechanistic basis of other EETs remain largely unknown. Here we show that the food-borne pathogen Listeria monocytogenes uses a distinctive flavin-based EET mechanism to deliver electrons to iron or an electrode. By performing a forward genetic screen to identify L. monocytogenes mutants with diminished extracellular ferric iron reductase activity, we identified an eight-gene locus that is responsible for EET. This locus encodes a specialized NADH dehydrogenase that segregates EET from aerobic respiration by channelling electrons to a discrete membrane-localized quinone pool. Other proteins facilitate the assembly of an abundant extracellular flavoprotein that, in conjunction with free-molecule flavin shuttles, mediates electron transfer to extracellular acceptors. This system thus establishes a simple electron conduit that is compatible with the single-membrane structure of the Gram-positive cell. Activation of EET supports growth on non-fermentable carbon sources, and an EET mutant exhibited a competitive defect within the mouse gastrointestinal tract. Orthologues of the genes responsible for EET are present in hundreds of species across the Firmicutes phylum, including multiple pathogens and commensal members of the intestinal microbiota, and correlate with EET activity in assayed strains. These findings suggest a greater prevalence of EET-based growth capabilities and establish a previously underappreciated relevance for electrogenic bacteria across diverse environments, including host-associated microbial communities and infectious disease.

Listeria monocytogenes is a fermentative gram-positive bacterium that is frequently associated with decaying plant matter in the environment, but which transforms into an intracellular pathogen upon encountering a mammalian host.[5] Despite lacking a lifecycle or genes conventionally associated with EET, a 25-year-old observation of extracellular ferric iron reductase activity[6] led us to question whether L. monocytogenes possessed a novel EET mechanism. Since electrons transferred out of the cell can be captured by an electrode, electrochemical measurements provide a useful tool for assaying EET.[7] Chronoamperometry experiments showing that L. monocytogenes produces a robust electric current in the presence of growth substrate thus provided evidence of EET (Fig. 1a, Extended Data Fig. 1a). Moreover, cyclic voltammetry experiments, which monitor electric current while the electrochemical potential is systematically varied, revealed a distinctive catalytic wave reminiscent of other electrochemically active bacteria (Extended Data Fig. 1b).[8,9]

Fig. 1.

An uncharacterized genetic locus associated with EET activity.

(a) Chronoamperometry results from L. monocytogenes (L.m.) or Shewanella oneidensis (S.o.)-inoculated electrochemical reactors. For L. monocytogenes experiments, the +/− signify whether electron donor (glucose) was included in the medium; lactate was used as an electron donor for S. oneidensis. Results are representative of three independent experiments (n = 3). (b) A representative of the thirty-six independent mutants identified from the ferric iron reduction screen minutes after ferrozine agar overlay (arrow). (c) Location of transposon insertions (triangles) in mutants identified as having decreased ferric iron reductase activity in the genetic screen. Arrows represent genes, with the darker gray signifying inclusion in a multi-gene operon. Previously uncharacterized genes on the locus have been assigned names based on putative functions. (d) Ferric iron reductase activity of transposon mutants identified from the screen. Results are expressed as means and standard errors from three independent experiments (n = 3). (e) Maximum electric current achieved from chronoamperometry experiments with representative EET mutants. Strains that statistically differ from wildtype (***, P < 0.001 [ANOVA with Dunnett’s posttest]) are indicated.

Extended Data Figure 1.

Electrochemical analyses of L. monocytogenes.

(a) The double chamber cell used for electrochemical experiments. Abbreviations stand for: working electrode (WE), reference electrode (RE), counter electrode (CE), cation exchange membrane (CEM). Inlets and outlets for N2 gas are labeled. (b) Cyclic voltammograms of wildtype and ndh2::tn L. monocytogenes strains. “Abiotic” refers to an uninoculated control. Arrows highlight the initiation of the catalytic wave. Results are representative of three independent experiments (n = 3).

To address the genetic basis of EET activity, ~50,000 colonies of a pooled L. monocytogenes himar1 transposon library were grown on Fe3+-containing agar plates. Mutants with decreased colorimetric change following an Fe2+-indicator overlay were visually identified and the location of their transposon insertion was mapped to the genome (Fig. 1b). From this screen, thirty-four independent transposon insertions that localized to a largely uncharacterized 8.5-kilobase locus were identified – with at least one insertion disrupting each of the 8 genes in this region (Fig. 1c). Based on the putative function of protein products, genes in the EET locus were assigned names (dmk-, eet-, and fmn- prefixes) that are used hereafter. The only transposon insertions outside the identified locus disrupt ribU, the substrate-binding subunit of a riboflavin transporter (Fig. 1c).[10] After confirming that the mutants had diminished ferric iron reductase (Fig. 1d) and electrochemical activity (Fig. 1e, Extended Data Fig. 1b), we turned to the molecular basis of EET.

Type II NADH dehydrogenase, or Ndh1 in L. monocytogenes, catalyzes electron exchange from cytosolic NADH to a lipid-soluble quinone derivative, the first step in the respiratory electron transport chain.[11] One of the genes in the EET locus, ndh2, encodes a protein with an N-terminal type II NADH dehydrogenase domain and a unique transmembrane C-terminal domain that is absent from functionally characterized enzymes (Fig. 2a). Consistent with ndh2 encoding a novel NADH dehydrogenase, we observed that EET activation correlated with cellular NAD+ levels (Extended Data Fig. 2). Furthermore, proteins encoded by two other genes in the EET locus, DmkA and DmkB, are homologous to enzymes MenA and HepT, which catalyze terminal steps in the production of the quinone demethylmenaquinone (Fig. 2b). In E. coli, the different quinones demethylmenaquinone, menaquinone, and ubiquinone are employed to selectively channel electrons to different electron acceptors.[12] By analogy, we reasoned that a distinct quinone derivative and NADH dehydrogenase might functionally segregate electron fluxes for EET and aerobic respiration.

Fig. 2.

A parallel electron transfer pathway segregates EET from aerobic respiration.

(a) Domain layout of L. monocytogenes proteins Ndh1 and Ndh2. Abbreviations stand for type II NADH dehydrogenase domain (NDH II) and C-terminal domain (CTD). Gray regions represent predicted transmembrane helices. (b) Predicted reactions catalyzed by DmkA/DmkB and paralogous L. monocytogenes proteins MenA/HepT, as well as MenG (highlighted by blue arrow). Abbreviations stand for demethylmenaquinone (DMK), menaquinone (MK), isopentenyl pyrophosphate (IPP), and 1,4-dihydroxy-2-napthoyl-CoA (DHNA). The ‘x’ refers to an unknown number of isoprene repeats, which may differ between the two quinones. (c) Colony-forming units (CFU) after 24 hours in “aerobic respiration medium.” Results (n = 3) are expressed as means and standard errors. The ΔcydAB/ΔqoxA mutant lacks terminal cytochrome oxidases and thus provides an aerobic respiration-deficient control. (d) Probable electron transfer pathways inferred from mutants with EET (red) or aerobic respiration (blue) phenotypes. Dashed arrows highlight the path of electron flow and solid lines track quinone synthesis.

Extended Data Figure 2.

EET activity maintains cellular redox homeostasis.

NAD+/NADH ratio in wildtype and ndh2::tn strains supplemented with ferric ammonium citrate under aerobic or microaerophilic conditions. Results from three independent experiments (n = 3) are expressed as means and standard errors. A statistically significant difference (*, P = 0.0015 [unpaired two-sided t test]) between microaerophilic cells incubated with or without iron is indicated.

To clarify the relationship between EET and aerobic respiration, we formulated an “aerobic respiration medium” that contained non-fermentable glycerol as the sole carbon source. Despite exhibiting wildtype levels of ferric iron reductase activity (Extended Data Fig. 3a), a positive control that lacked terminal cytochrome oxidases (ΔcydAB/ΔqoxA), ΔmenA, hepT::tn, and Δndh1 strains failed to grow on aerobic respiration medium (Fig. 2c). By contrast, EET mutants grew similarly to wildtype under these conditions (Fig. 2c). Moreover, menG, which encodes the enzyme that converts demethylmenaquinone to menaquinone, is contained on an operon with hepT and is essential for growth on aerobic respiration medium, but not ferric iron reductase activity (Fig. 2c, Extended Data Fig. 3). Collectively, these results support the conclusion that a demethylmenaquinone derivative used by Ndh2 and a menaquinone derivative used by Ndh1 are selective for downstream enzymes that function in EET and aerobic respiration, respectively (Fig. 2d).

Extended Data Figure 3.

Evidence that a distinct menaquinone derivative functions in aerobic respiration.

(a) Ferric iron reductase activity of mutants described in Fig. 2 demonstrates that genes essential for growth on aerobic respiration media are dispensable for EET. Results from three independent experiments (n = 3) are expressed as means and standard errors. (b) The L. monocytogenes hep/men operon. Notably, the demethylmenaquinone transferase, menG, which encodes the enzyme that converts demethylmenaquinone to menaquione (Fig. 2b), neighbors the hepT and hepS genes, which function in quinone biosynthesis and are essential for aerobic respiration (Fig. 2c).

We next sought to address downstream steps responsible for electron transfer from the quinone pool to extracellular electron acceptors. FmnB is a predicted lipoprotein that is annotated as possessing flavin mononucleotide (FMN) transferase activity. Homologous FMN transferases catalyze a posttranslational modification in which an FMN moiety is covalently linked to a threonine sidechain of substrate proteins (Fig. 3a).[13,14] To identify protein substrates of FmnB, wildtype and fmnB::tn cells were subjected to a comparative mass spectrometric analysis. Only two L. monocytogenes peptides met the criteria of selective FMNylation in the wildtype sample and both of these mapped to distinct regions in the protein product of the neighboring gene in the EET locus, PplA (Supplementary Table 1).

Fig. 3.

A surface-associated flavoprotein establishes the extracellular component of EET apparatus.

(a) Post-translational modification catalyzed by the FMN transferase family of enzymes, of which FmnB is a member.[13,14] (b) Domain architecture of PplA. Abbreviations stand for: unstructured (US), FMNylated domain 1 (FMN1), and FMNylated domain 2 (FMN2). The lipidated cysteine on the N-terminus after signal peptidase processing is colored red and FMNylated threonines yellow. (c) Analysis of FmnB substrate specificity. SDS-PAGE of recombinant PplA after incubation under specified conditions. UV illumination of the gel (bottom) allows for visualization of protein with covalently bound flavin. Results are representative of three independent experiments (n = 3). See Supplementary Figure 1 for uncropped gel. (d) Model of the molecular basis of EET. DmkA and DmkB synthesize a demethylmenaquinone (DMK) derivative (lower inset). RibU and FmnA secrete FAD that is used by FmnB to post-translationally modify PplA (upper inset). EET is achieved by a series of electron transfers. Ndh2 transfers electrons from NAD to DMK. Electrons are transferred from DMK to FMN groups on PplA or free flavin shuttles – possibly with involvement from uncharacterized membrane proteins in the EET locus, EetA and EetB – and ultimately to terminal electron acceptor.

Like FmnB, PplA is a predicted lipoprotein and a trypsin-shaving experimental approach, in which extracellular surface-associated proteins liberated through a partial digestion of the cell wall are identified by mass spectrometry, confirmed that PplA is associated with the surface of the cell (Supplementary Table 2). The N-terminal lipidation site on PplA is followed by ~30 amino acids that are predicted to be unstructured. N-terminal unstructured regions are a common feature of bacterial lipoproteins and are thought to provide a loose tether that allows the active portion of the protein to diffuse further from the membrane and to partially or fully penetrate the cell wall.[15] Thus, this property coupled with the covalently bound redox-active FMNs is consistent with PplA representing the extracellular component of the EET machinery that facilitates electron transfer, via its FMNs, to extracellular electron acceptors.

Following its unstructured N-terminal region, PplA has sequential domains that share 59% sequence identity with each other. From the proteomic analysis, it is evident that the FMNylated threonines on PplA assume equivalent positions on each of these related domains (Fig. 3b). To further clarify the mechanism of FMNylation, FmnB substrate specificity was tested using recombinant FmnB and PplA. These assays confirm that FmnB catalyzes FMNylation of PplA and demonstrate that the enzyme specifically uses flavin adenine dinucleotide (FAD) as substrate (Fig. 3c, Extended Data Fig. 4).

Extended Data Figure 4.

Recombinant FmnB FMNylates PplA at two discrete sites.

Deconvoluted mass spectra from a single experiment (n = 1) of (a) recombinant PplA and (b) recombinant PplA incubated with FAD + FmnB. The observed molecular weight change (877 Da) is consistent with two posttranslational FMNylations (2 × 438.3 Da) on PplA.

Considering that both FmnB and PplA are membrane-anchored lipoproteins, FmnB must require a mechanism of acquiring FAD substrate in order to modify PplA. The only transposon insertions identified outside the EET locus disrupt ribU, which has previously been shown to encode the substrate-binding subunit of an ECF (energy-coupling factor) transporter that functions in riboflavin uptake.[10] In addition to a substrate-binding subunit, ECF transporters contain a transmembrane subunit and two distinct ATPase subunits, which drive transport of substrate across the membrane (Extended Data Fig. 5a).[10] FmnA in the EET locus shares 50% sequence identity with EcfT, the transmembrane subunit of the RibU-ECF riboflavin transporter, and this led us to hypothesize that it might interact with RibU to promote FAD secretion (Extended Data Fig. 5b). Consistent with this interpretation, proteomic analysis of ribU::tn and fmnA::tn strains revealed a dramatic decrease in PplA FMNylation (Supplementary Table 1). Furthermore, addition of FAD to the growth medium specifically restored ferric iron reductase activity to the ribU::tn and fmnA::tn strains (Extended Data Fig. 5c). Based on these findings, we propose that RibU and FmnA establish a transporter that secretes the FAD required for FmnB-catalyzed FMNylation of PplA.

Extended Data Figure 5.

Proposed role of RibU and FmnA in FAD secretion.

(a) Simplified adaptation of a previously proposed model of L. monocytogenes riboflavin uptake through the RibU, EcfT, EcfA, and EcfA’ transporter.[10] According to this model, EcfT, EcfA, and EcfA’ couple ATP hydrolysis with conformational changes that result in substrate bound to RibU being released into the cytosol. (b) Based on protein homology (FmnA shares 50% sequence identity with EcfT) and the expectation that extracellular FAD is required for FmnB to catalyze FMNylation of PplA, we propose the FmnA interacts with RibU to promote FAD secretion. (c) Ferric iron reductase activity of strains incubated with 0.5 mM FAD for 1 hour. The ability of exogenous FAD to specifically rescue ferric iron reductase activity to the fmnA::tn and ribU::tn strains is consistent with FmnA and RibU functioning in FAD secretion. Results from three independent experiments (n = 3) are expressed as means and standard errors. Statistically significant differences (*, P = 0.038 and **, P < 0.0001 [unpaired two-sided t test]) between untreated and FAD-treated cells are indicated.

The term “extracellular electron shuttle” refers to redox-active small molecules that are cyclically reduced by cells and oxidized by extracellular electron acceptors.[16,17] The relevance of shuttles for EET is exemplified by Shewanella species, which use an efflux-type transporter to secrete flavins that can shuttle electrons to acceptors that are not directly contacting the cell.[18-20] In contrast to Shewanella, L. monocytogenes is a flavin auxotroph and thus, by definition, environmental flavins must be present in its replicative niche. Indeed, micromolar flavin concentrations are typical of nutrient-rich environments, like the plant/animal biomass and mammalian host where L. monocytogenes proliferates.[21,22] To determine whether flavins could be used as electron shuttles, we tested the effect of exogenous riboflavin, FMN, and FAD on EET activity. Injection of FMN into an L. monocytogenes-inoculated electrochemical chamber resulted in a pronounced increase in electric current (Extended Data Fig. 6a). Moreover, while flavins caused a marked concentration-dependent enhancement in the reduction of insoluble ferric (hydr)oxide, cells immersed in soluble ferric iron exhibited a high baseline level of activity that was unresponsive to flavins (Extended Data Fig. 6b). These data thus support the conclusion that L. monocytogenes can use environmental flavins to shuttle electrons to outlying acceptors.

Extended Data Figure 6.

Flavin shuttles promote EET activity.

(a) Chronoamperometry results from L. monocytogenes-inoculated electrochemical reactors with 1 μM FMN injections at the indicated time points. Results are representative of three independent experiments (n = 3). (b) The effect of flavins on L. monocytogenes (Lm) ferric iron reductase activity with insoluble ferric (hydr)oxide (top) and soluble ferric ammonium citrate (bottom). With insoluble substrate the local iron concentration for most cells is low, whereas with soluble substrate the concentration of iron in the direct vicinity of cells is high (insets). Results from three independent experiments (n = 3) are expressed as means and standard errors.

Integrating insight into the role of the components of the EET apparatus, we arrive at a molecular model of electron travel from intracellular NADH, to membrane-confined quinone, to extracellular flavoprotein/shuttles, and ultimately to a kinetically favorable terminal electron acceptor (Fig. 3d). Next, to determine whether EET established a bona fide growth-supporting activity, we screened a library of common microbial growth substrates and found that the inclusion of ferric iron or an electrode was required for anaerobic growth on the sugar alcohols xylitol and D-arabitol (Fig. 4a, Extended Data Fig. 7). Moreover, while genes for aerobic respiration, but not EET, were essential for aerobic growth on xylitol, this pattern was reversed under anaerobic conditions, with the EET genes being essential and aerobic respiration genes dispensable (Fig. 4a, Extended Data Fig. 7). These data thus demonstrate that the distinct electron transport chains that segregate aerobic respiration and EET promote aerobic and anaerobic growth, respectively.

Fig. 4.

EET supports anaerobic growth, confers a competitive advantage in the intestinal lumen, and is active in multiple Firmicutes.

(a) L. monocytogenes CFU (left) and electric current (right) from chronoamperometry experiments conducted with xylitol growth medium. The (−) signifies a control condition without an electrode. Results from three independent experiments (n = 3) are expressed as means and standard errors. (b) Mice (n = 5) were fed bread inoculated with a 1:1 mixture of Δhly L. monocytogenes and Δhly/ndh2::tn strains. The competitive index at three post-infection time points is indicated. Median values and statistically significant differences (*, P = 0.01 [unpaired two-sided t test]) between the Δhly/ndh2::tn mutant and a control that competed two Δhly strains are indicated. Results are representative of three independent experiments (n = 3). (c) Iron reductase activity in a panel of Firmicutes species, expressed as a percentage of wildtype L. monocytogenes activity. Results from at least three independent experiments (n = 7 for ndh2::tn, L. garviae, E. durans; n = 6 for L. innocua E. faecalis S. mutans; n = 5 for C. maltaromaticum, E. casseliflavus, S. gallolyticus, B. subtillis; n = 4 for L. lactis, E. faecium, E saccharolyticus, B. circulans, L. plantarum, E. raffinosus; n = 3 for L. casei, E. coli K12) are expressed in arbitrary units as means and standard errors. Strains that statistically differ from ndh2::tn are indicated (*, P < 0.05 [ANOVA with Dunnett’s posttest]). Some Lactobacillales lack the ability to synthesize 1,4-dihydroxy-2-napthoyl-CoA (DHNA), the precursor for demethylmenaquinone biosynthesis, and require an exogenous source for quinone-dependent processes.[36] Organisms with EET genes and menC (which catalyzes an essential step in DHNA biosynthesis) are colored gray. L. casei, L. plantarum and E. raffinosus contain genes for EET, but not menC. The remaining species lack EET genes.

Extended Data Figure 7.

EET supports anaerobic growth on ferric iron.

(a) Growth following incubation of L. monocytogenes strains on xylitol media without (left) or with (right) ferric iron under aerobic (top) or anaerobic (bottom) conditions. Results are representative of three independent experiments (n = 3). Strain labels are colored based on attributed deficiencies (Fig. 2d) in aerobic respiration (blue) or EET (red). Ndh1 and Ndh2 are likely functionally redundant under aerobic conditions, as a growth phenotype is only observed in the double mutant. Note that visual evidence of ferrous iron production in the agar adjoining anaerobically growing cells. (b) CFU of L. monocytogenes strains anaerobically incubated in xylitol media without (−) or with (+) ferric supplementation. Results for soluble ferric ammonium citrate (top) and insoluble ferric (hydr)oxide (bottom) are shown. Dashed lines denote the number of cells at the start of the of the experiment. Results from three independent experiments (n = 3) are expressed as means and standard errors. Statistically significant differences (***, P < 0.0001 [unpaired two-sided t test]) in the ferric iron-supplemented condition are noted.

We next asked whether EET played a role in host colonization. Consistent with EET being dispensable for aerobic growth, EET-deficient mutants resembled wildtype L. monocytogenes in an intracellular macrophage growth assay and an intravenous infection model (Extended Data Fig. 8). Since anaerobic growth mechanisms are important for microbial proliferation within the intestinal lumen,[23,24] we hypothesized that the foodborne pathogen might utilize EET in this context. Consistent with the hypothesis, the fecal burden of the ndh2::tn strain was decreased ~6-fold in a streptomycin-pretreated model of L. monocytogenes intestinal colonization (Fig 4b). These results thus suggest a role for EET within the dysbiotic gut and raise the possibility that EET establishes a generally significant metabolic activity within the mammalian gastrointestinal tract.

Extended Data Figure 8.

EET genes are dispensable for L. monocytogenes intracellular growth.

(a) Murine bone-marrow-derived macrophages were infected with L. monocytogenes and CFU were enumerated at the indicated times. Results from three independent experiments (n = 3) are expressed as means and standard errors. (b) L. monocytogenes burdens in mouse organs 48 hours after intravenous infection. Representative results from two independent experiments (n = 2) are expressed as medians and standard errors.

We next turned to the phylogenetic distribution of the identified EET genes. BLAST searches revealed that homologs of the genes are widespread in hundreds of species that span the Firmicutes phylum (Extended Data Fig. 9a, Supplementary Table 3). Many of these genes likely encode functional EET systems, as the identified locus is typically conserved, though noteworthy distinctions are evident in some genomes (Extended Data Fig. 9b). Microbes that possess a locus with EET genes adopt a wide range of different lifestyles, including within thermophilic (Caldanaerobius, Thermoanaerobacter, etc.) and halophilic (Halolactibacilli, Halothermothrix, etc.) habitats. Orthologs of the identified EET genes are found in a number of human pathogens (Clostridium perfringens, Enterococcus faecalis, Streptococcus dysgalactiae, etc.), members of the human microbiota (Clostridia, Enterococci, Streptococci, etc.), and lactic acid bacteria that have commercial applications in food fermentation or probiotics (Lactococci, Lactobacilli, Oenococci, Tetragenococci, etc.) (Supplementary Table 3). Functionality of identified loci could explain previous reports of EET-like activity in a number of species[25-35] and assays of ferric iron reductase activity of a panel of Firmicutes provided additional evidence that the presence of necessary genetic components correlates with EET activity (Fig. 4c).

Extended Data Figure 9.

Identified EET loci are widespread in the Firmicutes phylum.

(a) Phylogenetic tree constructed from select Ndh2 homolog sequences. A more comprehensive list of organisms that possess an EET locus is provided in Supplementary Table 3. The percentage of replicate trees that gave the depicted branch topology in a bootstrap test of 1,000 replicates is labeled. (b) Distinct EET loci from select genomes are shown. While the arrangement of genes varies, a locus with EET genes is present in many genomes. Some loci contain ECF transporter ATPase subunits (homologous to those depicted in Extended Data Fig. 5a) that likely function with RibU and FmnA subunits in flavin transport. The dmkA-like gene found in Caldanaerobius fijiensis (and other genomes) lacks homology to dmkA, but is annotated as catalyzing the same reaction. The pplA variant in some genomes contains a single FMNylated domain (rather than two) and this property is indicated by a shorter arrow. A few bacteria (including the Lactococci) lack a recognizable locus and distribute EET genes throughout the genome.

In conclusion, the studies presented here establish a novel electron transport chain that supports growth on extracellular electron acceptors. This mechanism lacks an elaborate multi-heme apparatus and, partly by taking advantage of the single-membrane gram-positive cell architecture, is characterized by significantly fewer electron transfer steps than comparable systems in mineral-respiring gram-negative bacteria.[1] Interestingly, the identified EET genes are present in a wide-ranging group of microorganisms that occupy a diverse array of ecological niches. Defying conventional views of EET, this distinctive system is abundant in bacteria that prioritize fermentative metabolic strategies and reside in nutrient-rich environments, including the lactic acid bacteria. Within this context, environmental flavins seems to represent a feature of the ecological landscape that can be exploited to promote EET activity. These observations suggest that, rather than a specialized process confined to mineral-respiring bacteria, utilization of extracellular electron acceptors represents a fundamental facet of microbial metabolism relevant across diverse environments. In addition to obvious bioenergetic applications, characterization of flavin-based EET mechanism thus establishes new avenues for the study of electrochemical activities throughout the microbial world.

Methods

L. monocytogenes strains and growth conditions.

All L. monocytogenes strains used in this study were derived from wildtype 10403S (Supplementary Table 4). Transduction methods were used to introduce transposons into distinct genetic backgrounds, as previously described.[37,38] L. monocytogenes cells were grown at 37 °C and spectrophotometrically measured by optical density at a wavelength of 600 nm (OD600). Anaerobic conditions were achieved with the BD GasPak™ EZ pouch system or an anaerobic chamber (Coy Laboratory Products) with an environment of 2% H2 balanced in N2.

Filter-sterilized brain-heart infusion medium (Difco) or variants of chemically defined Listeria synthetic medium (LSM)[39] were used in all studies. “Aerobic respiration medium” replaced the glucose in LSM with 50 mM glycerol. The requirement of an electron acceptor to support L. monocytogenes growth on xylitol was identified by comparing aerobic versus anaerobic (absent an alternative electron acceptor) growth on carbon sources, using PM1 and PM2A plates of the Phenotype MicroArray (Biolog). “Xylitol medium” replaced the glucose in LSM with 50 mM xylitol.

Gene name assignment.

The identified EET locus is widely conserved in L. monocytogenes isolates and encompasses the genes lmrg_02179-lmrg_02186 in L. monocytogenes 10403S (which correspond to lmo2634-lmo2641 in L. monocytogenes EGD-e). Identified EET genes were assigned dmk or fmn prefixes based on putative roles in demethylmenaquinone biosynthesis or PplA FMNylation, respectively. The eet prefix was assigned to the remaining genes, which at present lack high-confidence functional assignments. The only previously named gene, pplA, was so called based on the role of its cleaved signal peptide as a signaling pheromone (a function that seems unrelated to the mature protein).[40]

Bioelectrochemical characterization and measurements.

Chronoamperometry and cyclic voltammetry were carried out using a Bio-Logic Science Instruments potentiostat model VSP-300. All measurements were performed using double chamber electrochemical cells (Extended Data Fig. 1a) and consisted of an Ag/AgCl reference electrode (CH Instruments), a Pt wire counter electrode (Alfa Aesar), and a 6.35 mm-thick graphite felt working electrode with a 16-mm radius (Alfa Aesar).

Electrochemical cells were prepared with 120 mL of modified LSM (containing 0.8 μM FMN as the sole flavin) and an open circuit potential was performed in the absence of bacteria. Once the current stabilized, the electrochemical cell was inoculated to a final OD600 of ~0.1. The medium in the electrochemical chamber was mixed with a magnetic stir bar for the course of the experiment. For current acquisition, the applied potential was set at +0.4 V vs Ag/AgCl. To maintain anaerobic conditions, electrochemical cells were continuously purged with N2 gas. Cyclic voltammetry measurements in the potential region of −0.8 to +0.4 V vs Ag/AgCl and a scan rate of 10 mV s−1 were conducted immediately prior to inoculation and 3 hours later. Electric currents are reported as a function of the geometric surface area of the electrode. To test the effect of flavins on electrochemical activity, FMN was injected into the L. monocytogenes-inoculated electrochemical chamber to a final concentration of 1 μM.

For S. oneidensis experiments, the glucose in LSM was replaced with sodium lactate and S. oneidensis was inoculated to an OD600 of 0.1. Growth-supporting L. monocytogenes experiments on xylitol medium were conducted in a similar fashion, but the electrochemical cell was inoculated to an OD600 of ~0.002 and the medium from the electrochemical chamber was sampled at regular intervals for the enumeration of CFU.

Screen of mutants with diminished ferric iron reductase activity.

A previously described method was adapted to screen for L. monocytogenes mutants with diminished ferric iron reductase activity.[6] Approximately 250 colony-forming units/plate of a pooled himar1 transposon library, generated as previously described,[38] were grown on brain-heart infusion agar supplemented with 0.1 mg/mL ferric ammonium citrate. After 24 hours at 37 °C, plates were removed from the incubator and a 10-mL overlay (0.8% agarose and 2 mM ferrozine) was applied. Colorimetric change resulting from ferrozine binding to Fe2+ was visually tracked for ~10 minutes. Colonies with diminished colorimetric change were selected and the location of the transposon insertion identified by Sanger sequencing, as previously described.[41]

Ferrozine assay of ferric iron reductase activity.

L. monocytogenes cells grown to mid-log phase were washed twice, normalized to an OD600 of 0.5, and resuspended in fresh medium supplemented with 4 mM ferrozine. Experiments were initiated by adding 100 μL of cells to an equivalent volume of 50 mM ferric ammonium citrate or ferric (hydr)oxide and were conducted in triplicate at 37 °C in 96-well format using a plate reader. OD562 measurements were made every 30 seconds for up to an hour. Maximal rates (typically over 2 minutes) calculated from a Fe2+ standard curve are reported. Assays were generally performed in LSM, with glucose serving as the electron donor. However, because some of the respiratory mutants grew poorly in these conditions, these strains were assayed in brain-heart infusion medium (with glucose remaining as the electron donor). For FAD complementation studies, prior to washing steps, strains grown to mid-log were split and, after adding 0.5 mM FAD to one aliquot, incubated for 1 hour at 37 °C. To test the effect of flavins, riboflavin, FMN, or FAD was titrated into cells resuspended in a LSM base that lacked flavins.

To prepare other species (detailed in Supplementary Table 4) for the ferric iron reductase assay, cells were grown anaerobically in brain-heart infusion medium for 36 hours. Sub-cultures in brain-heart infusion medium supplemented with 25 mM ferric ammonium citrate were then grown to mid-log phase. Cells were washed twice, resuspended in fresh brain-heart infusion medium, and cell densities were normalized to wildtype L. monocytogenes. Next, ferrozine was added to a final concentration of 2 mM and 100 μL of cells were dispensed in a 96-well plate. The experiment was initiated by adding 100 μL of brain-heart infusion medium supplemented with 10 mM ferric ammonium citrate and OD562 measurements were made as described for the L. monocytogenes ferric iron reductase assay.

L. monocytogenes growth on xylitol and ferric iron.

To test electron acceptor usage capabilities, xylitol medium was inoculated with L. monocytogenes and incubated at 25° C in an anaerobic chamber. Conditions testing putative electron acceptors contained 50 mM ferric ammonium citrate or ferric (hydr)oxide, prepared as previously described.[42] For the ferric ammonium citrate experiments, 50 mM sodium citrate was included in the ferric ammonium citrate-lacking control condition and CFU were enumerated following overnight incubation in a 96-well plate (Greiner Bio-One). Ferric (hydr)oxide experiments were conducted in 6-well plate (Costar) and CFU were enumerated 6 days after inoculation.

NAD+/NADH measurements.

L. monocytogenes cells grown overnight in LSM were washed and resuspended in 500 μL of medium. Cells were then split and 50 mM ferric ammonium citrate was added to one aliquot. To test aerobic conditions, 14-mL tubes were placed in a shaking (200 RPM) incubator. To achieve microaerophilic conditions, the headspace in the tube was purged with argon gas and the tightly capped tube was placed in a stationary incubator. After 1.5 hours at 37°C, bacteria were harvested by centrifugation, resuspended in PBS, and lysed by vortexing with 0.1 mm-diameter zirconia-silica beads. NAD+/NADH measurements were performed using the NAD/NADH-Glo Assay (Promega).

Assay of FmnB FMN transferase activity

Constructs of fmnB and pplA that truncated the signal peptide were subcloned into the pMCSG58 vector. Protein overexpression and purification followed previously described protocols.[43] Purified PplA and FmnB were incubated overnight at a 10:1 molar ratio in assay buffer (0.5 M NaCl and 10 mM Tris, pH 8.3) with putative flavins substrates. Since homologous FMN transferases require a magnesium cofactor,[13] the effect of the chelator ethylenediaminetetraacetic acid (EDTA) on activity was tested. Samples were analyzed by SDS-PAGE and protein bands with covalent flavin modifications were visualized by UV illumination.

To identify the basis of posttranslational modifications, intact protein mass measurements of PplA were made using a Synapt G2-Si mass spectrometer that was equipped with an electrospray ionization (ESI) source and a C4 protein ionKey (inner diameter: 150 μm, length: 50 mm, particle size: 1.7 μm), and connected in-line with an Acquity M-class ultra-performance liquid chromatography system (UPLC; Waters, Milford, MA). Acetonitrile, formic acid (Fisher Optima grade, 99.9%), and water purified to a resistivity of 18.2 MΩ·cm (at 25 °C) using a Milli-Q Gradient ultrapure water purification system (Millipore, Billerica, MA) were used to prepare mobile phase solvents. Solvent A was 99.9% water/0.1% formic acid and solvent B was 99.9% acetonitrile/0.1% formic acid (v/v). The elution program consisted of a linear gradient from 1% to 10% B (v/v) over 1 min, a linear gradient from 10% to 90% B over 4 min, isocratic flow at 90% B for 5 min, a linear gradient from 90% to 1% B over 2 min, and isocratic flow at 1% B for 18 min, at a flow rate of 2 μL/min. The ionKey column and the autosampler compartment were maintained at 40 °C and 6 °C, respectively. Mass spectra were acquired in the positive ion mode and continuum format, operating the time-of-flight (TOF) analyzer in resolution mode, with a scan time of 0.5 s, over the range m/z = 400 to 5000. Mass spectral deconvolution was performed using ProMass software (version 2.5 SR-1, Novatia, Monmouth Junction, NJ).

L. monocytogenes protein trypsinization.

One milliliter of L. monocytogenes cells grown in LSM to mid-log phase was washed, resuspended in 100 μL of 100 mM NH4HCO3 (pH 7.5), and incubated at 100 °C for 10 minutes. Cells were lysed by bead beating for 15 minutes at 4 °C. RapiGest SF (Waters) was added to lysed cells at a final concentration of 0.1% and the sample was incubated at 100 °C for 5 minutes. After adding 5 μL of 100 mM dithiothreitol, samples were incubated at 58 °C for 30 minutes. Next, 15 μL of 100 mM iodoacetamide was added and sample were incubated for an additional 30 minutes. Samples were then digested overnight with 10 μL Trypsin Gold (Promega). The following morning, 10 μL of 5% trifluoroacetic acid was added and samples were incubated at 37 °C for 90 minutes. Samples were centrifuged for 30 minutes, to remove hydrolyzed RapiGest, and supernatant was collected.

L. monocytogenes intracellular growth assays.

Bone marrow-derived macrophages prepared from 6- to 8-week-old female mice were plated overnight on coverslips and infected with L. monocytogenes strains at a multiplicity of infection of 0.1. Macrophage monolayers were washed with PBS and fresh medium was added thirty minutes after infection. At 1 hour post-infection, 50 μg/mL gentamicin was added to kill extracellular bacteria. To enumerate L. monocytogenes CFU, macrophages were lysed by transferring coverslips to 10 mL of water, as previously described.[44]

L. monocytogenes intravenous infections.

Eight-week-old female C57BL/6 mice (The Jackson Laboratory) were infected with 1 × 105 CFU in 200 μL of PBS by tail vein injection. Forty-eight hours post-infection, spleens and livers were harvested, homogenized, and plated for the enumeration of CFU.

L. monocytogenes oral infections.

Previously described models of L. monocytogenes oral infection were adapted to address the role of EET in the intestinal lumen.[45,46] Prior to infection, 5 mg/mL of streptomycin sulfate was added to the drinking water of 8-week-old female C57BL/6 mice (The Jackson Laboratory). After 24 hours, mice were transferred to fresh cages and chow was removed to initiate an overnight fast. Forty-eight hours after streptomycin addition to the water, mice were isolated, fed a small piece of bread with 3 μL of butter and an inoculum with 108 CFU of L. monocytogenes, and returned to cages containing standard drinking water and chow. To confine L. monocytogenes to the intestinal lumen, a Δhly parental strain (which have greatly reduced intracellular growth and spread) was used in these experiments. Inoculums were prepared with a 1:1 ratio of Δhly and an erythromycin-resistant Δhly strain (Δhly/erm, derived as previously described[47]) or Δhly and Δhly/ndh2::tn. Following infection, stools were collected, homogenized, and dilutions were plated. Because total parental strain CFU did not statistically differ between conditions, results are simply reported as a competitive index (i.e., the ratio of streptomycin to erythromycin-resistant CFU). This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All protocols were reviewed and approved by the Animal Care and Use Committee at the University of California, Berkeley (AUP-2016–05-8811).

Identification of protein substrates of FmnB.

Wildtype and fmnB::tn strains grown in LSM were prepared for proteomic analysis as described in the protein trypsinization section. Peptides with >50% FMNylated peptide relative ion abundance in the wildtype sample and <5% in the fmnB::tn sample were identified using Progenesis QI for Proteomics software (version 4.0, Waters) and validated by manual inspection of the data. To address the FMNylation status of PplA, ribU::tn and fmnA::tn mutants were prepared in the same manner.

Trypsin-shaving analysis of surface-associated proteins.

Trypsin-shaving experiments adapted a previously described method.[4] Cells grown in brain-heart infusion medium were washed twice and resuspended in a shaving buffer (1 M sucrose + 1 mM HEPES, pH 7). Lysozyme from chicken egg white (Sigma) was added to a concentration of 0.1 mg/mL. Cells were incubated at 37 °C for 60 minutes and released surface-associated components were separated from the cell by centrifugation. The supernatant (surface-associated protein fraction) was dialyzed overnight in digestion buffer (100 mM NH4HCO3, pH 7.5) and the pellet (total protein fraction) was resuspended in digestion buffer. Samples were prepared for proteomic experiments as described in the protein trypsinization section. A label-free relative quantification approach[48,49] implemented in Progenesis QI for Proteomics software (version 4.0, Waters) identified proteins disproportionately abundant in the surface-associated fraction.

Liquid chromatography-mass spectrometry analysis of trypsin-digested proteins.

Samples of trypsin-digested proteins were analyzed in triplicate using the Acquity M-class UPLC and Synapt G2-Si mass spectrometer, as follows. The mass spectrometer was equipped with a nanoelectrospray ionization (nanoESI) source that was connected in-line with the UPLC. The UPLC was equipped with trapping (Symmetry C18, inner diameter: 180 μm, length: 20 mm, particle size: 5 μm) and analytical (HSS T3, inner diameter: 75 μm, length: 250 mm, particle size: 1.8 μm, Waters) columns. Solvent A was 99.9% water/0.1% formic acid and solvent B was 99.9% acetonitrile/0.1% formic acid (v/v). The elution program consisted of a linear gradient from 1% to 10% B (v/v) over 2 min, a linear gradient from 10% to 35% B over 90 min, a linear gradient from 35% to 90% B over 1 min, isocratic flow at 90% B for 6 min, a linear gradient from 90% to 1% B over 1 min, and isocratic flow at 1% B for 20 min, at a flow rate of 300 nL/min. The column and autosampler compartments were maintained at 35 °C and 6 °C, respectively. Ion mobility-enabled HD-MSE data[50,51] were acquired in the positive ion mode and continuum format, operating the TOF analyzer in resolution mode, with a scan time of 0.5 s, over the range m/z = 50 to 2000. An optimized wave velocity of 850 m/s was used for the traveling wave ion mobility cell. Collision-induced dissociation was performed in the ion transfer cell with a collision energy ramp from 30 to 78 V. Data acquisition was controlled using MassLynx software (version 4.1), and tryptic peptides were identified using Progenesis QI for Proteomics software (version 4.0, Waters).

Bioinformatics analysis of identified EET genes.

Ndh2 homologs were identified by searching the sequence of the unique C-terminal domain of Ndh2 on the PSI-BLAST server.[52] To perform a phylogenetic analysis, representative homologs were selected and aligned by ClustalW.[53] The maximum likelihood method was used to infer the evolutionary history of identified sequences in Mega 7.0.26 and confidence limits of branch points were estimated by 1,000 bootstrap replications.[54,55] The information about EET genetic loci summarized in Supplementary Table 3 was acquired by analyzing genomic context of identified genes in the PATRIC 3.5.1 (https://www.patricbrc.org) database.

Statistics and reproducibility.

No statistical methods were used to pre-determine sample size. The investigators were not blinded to allocation during experiments and outcome assessment. Statistical analyses were performed in Prism 5 for Mac OS X (GraphPad Software) and Progenesis QI for Proteomics version 4.0.

Data availability statement.

The datasets generated during the current study are available from the corresponding author on reasonable request.