Extracellular electron transfer increases fermentation in lactic acid bacteria via a hybrid metabolism

  1. Sara Tejedor-Sanz
  2. Eric T Stevens
  3. Siiliang Li
  4. Peter Finnegan
  5. James Nelson
  6. Andre Knoesen
  7. Samuel H Light
  8. Caroline M Ajo-Franklin  Is a corresponding author
  9. Maria L Marco  Is a corresponding author
  1. Department of BioSciences, Rice University, United States
  2. Biological Nanostructures Facility, The Molecular Foundry, Lawrence Berkeley National Laboratory, United States
  3. Department of Food Science & Technology, University of California‐Davis, United States
  4. Department of Electrical and Computer Engineering, University of California‐Davis, United States
  5. Department of Microbiology, University of Chicago, United States

Abstract

Energy conservation in microorganisms is classically categorized into respiration and fermentation; however, recent work shows some species can use mixed or alternative bioenergetic strategies. We explored the use of extracellular electron transfer for energy conservation in diverse lactic acid bacteria (LAB), microorganisms that mainly rely on fermentative metabolism and are important in food fermentations. The LAB Lactiplantibacillus plantarum uses extracellular electron transfer to increase its NAD+/NADH ratio, generate more ATP through substrate-level phosphorylation, and accumulate biomass more rapidly. This novel, hybrid metabolism is dependent on a type-II NADH dehydrogenase (Ndh2) and conditionally requires a flavin-binding extracellular lipoprotein (PplA) under laboratory conditions. It confers increased fermentation product yield, metabolic flux, and environmental acidification in laboratory media and during kale juice fermentation. The discovery of a single pathway that simultaneously blends features of fermentation and respiration in a primarily fermentative microorganism expands our knowledge of energy conservation and provides immediate biotechnology applications.

Editor's evaluation

In this study, the authors describe unique metabolic strategies, including extracellular electron transfer, utilized by the lactic acid bacterium Lactiplantibacillus plantarum. The ability to shift and/or accelerate metabolism of lactic acid bacteria capable of extracellular electron transfer may have interesting biotechnological applications.

https://doi.org/10.7554/eLife.70684.sa0

eLife digest

Bacteria produce the energy they need to live through two processes, respiration and fermentation. While respiration is often more energetically efficient, many bacteria rely on fermentation as their sole means of energy production. Respiration normally depends on the presence of small soluble molecules, such as oxygen, that can diffuse inside the cell, but some bacteria can use metals or other insoluble compounds found outside the cell to perform ‘extracellular electron transfer’.

Lactic acid bacteria are a large group of bacteria that have several industrial uses and live in many natural environments. These bacteria survive using fermentation, but they also carry a group of genes needed for extracellular electron transfer. It is unclear whether they use these genes for respiration or if they have a different purpose.

Tejedor-Sanz, Stevens et al. used a lactic acid bacterium called Lactiplantibacillus plantarum to study whether and how this group of bacteria use extracellular electron transfer. Analysis of L. plantarum and its effect on its surroundings showed that these bacteria use a hybrid process to produce energy: the cells use aspects of extracellular respiration to increase the yield and efficiency of fermentation. Combining these two approaches may allow L. plantarum to adapt to different environments and grow faster, allowing it to compete against other species.

Tejedor-Sanz, Stevens et al. provide new information on a widespread group of bacteria that are often used in food production and industry. The next step will be to understand how the hybrid system is controlled and how it varies among species. Understanding this process could result in new biotechnologies and foods that are healthier, produce less waste, or have different tastes and textures.

Introduction

The ways in which microorganisms extract energy to maintain cellular functions are directly linked to their environment, including the availability of nutrients and cooperative or antagonistic interactions with other organisms (Haruta and Kanno, 2015). Microorganisms must also maintain redox homeostasis by responding to oxidative and reductive changes inside and outside the cell (Sporer et al., 2017). Ultimately, microorganisms that can effectively generate cellular energy while also managing redox requirements will maintain higher growth and survival rates, and therefore exhibit greater ecological fitness.

All organisms possess mechanisms to conserve energy, that is, to convert light or chemical energy into cellular energy in the form of ATP (Russell and Cook, 1995). During respiration, microorganisms rely on either oxygen (aerobic respiration) or other exogenous substrates (anaerobic respiration) as terminal electron acceptors. Some microorganisms, most notably Geobacter spp., can anaerobically respire using electron acceptors outside the cell, such as iron (III) oxides or an electrode (Renslow et al., 2013; Richter et al., 2012). This process is called extracellular electron transfer (EET). Regardless of the identity of the electron acceptor, ATP synthesis during respiration occurs via oxidative phosphorylation (Kim and Gadd, 2019). In oxidative phosphorylation, electrons from electron carriers are transported by an electron transport chain, which creates a proton motive force (PMF) for ATP generation. Under anaerobic conditions, some cells can also conserve energy using fermentation. In fermentation, microorganisms use internally supplied electron acceptors, and ATP is generated mainly through substrate-level phosphorylation (Kim and Gadd, 2019). In substrate-level phosphorylation, ATP is generated in the cytoplasm by transfer of phosphate from metabolic intermediates to ADP (Kim and Gadd, 2019).

Lactic acid bacteria (LAB) are a diverse group of aerotolerant, saccharolytic microorganisms in the Firmicutes phylum which mainly use fermentation for energy conservation. LAB are essential for many food fermentations, including fermented milk and meats, fruits and vegetables, and grains (Tamang et al., 2020). Strains of LAB are also used for industrial chemical production (Sauer et al., 2017) and as probiotics to benefit human and animal health (Vinderola et al., 2019). LAB are generally grouped by their differences in hexose metabolism (Salvetti et al., 2013). Some species perform homofermentation, reducing pyruvate to lactate as the sole metabolic end-product from glycolysis. Other LAB perform heterofermentation, producing lactate along with ethanol, acetate, and CO2 by the phosphoketolase pathway. However, for redox balancing, homofermentative LAB can also shift to a mixed acid fermentation and heterofermentative LAB use alternative electron acceptors, like fructose or citrate (Hansen, 2018). Although some LAB can respire in the presence of heme and menaquinone, those bacteria are unable to synthesize heme and many are also auxotrophic for menaquinone (Pedersen et al., 2012). Even those species capable of respiration still use fermentation metabolism as the primary mechanism to conserve energy (Pedersen et al., 2012). Therefore, LAB growth rates and cell yields are constrained by access to electron acceptors used to maintain intracellular redox balance during substrate-level phosphorylation.

The bioenergetics of anaerobic bacteria have been tightly linked to oxidative phosphorylation for anaerobic respiration and substrate-level phosphorylation for fermentation. However, experimental evidence shows a concurrent use of oxidative phosphorylation and substrate-level phosphorylation. For instance, some yeasts perform respiro-fermentation to enhance ATP production (Pfeiffer and Morley, 2014). Another example is the electron bifurcating mechanism used by some fermentative microorganisms such as Clostridium spp. (Herrmann et al., 2008; Li et al., 2008). Through that energy conservation strategy, cells can generate extra ATP through oxidative phosphorylation (Buckel and Thauer, 2013; Müller et al., 2018). Along with other examples that are not fully understood (Hunt et al., 2010Kracke et al., 2018), these observations suggest metabolisms that combine aspects of fermentation and respiration may exist.

We recently discovered that Listeria monocytogenes, a facultative anaerobic pathogen known to rely on respiratory metabolism, uses EET to reduce Fe3+ or an anode through a flavin-based extracellular electron transfer pathway (Light et al., 2018). Use of this pathway allowed L. monocytogenes to maintain intracellular redox balance via NADH oxidation. This capacity was associated with the presence of a gene locus, called a flavin-based EET (FLEET) locus, that was identified in many Gram-positive species in the Firmicutes phylum, including LAB. Studies in individual species of LAB such as Lactococcus lactis (Freguia et al., 2009; Masuda et al., 2010), Enterococcus faecalis (Hederstedt et al., 2020; Keogh et al., 2018), and Lactiplantibacillus pentosus (Vilas Boas et al., 2015) show that they can perform EET with an anode endogenously, that is without addition of molecules foreign to their native niches. These observations are quite surprising because endogenous EET has been mainly associated with respiratory organisms, even though some of these organisms also possess fermentative-type metabolism (Glasser et al., 2014). Those observations also raise the question of whether the FLEET locus is functional in LAB and what, if any, role it plays in energy conservation and metabolism.

Here, we explored EET across LAB and studied the implications of this trait at a metabolic and energetic level in Lactiplantibacillus plantarum, a homofermentative LAB capable of mixed acid fermentation and which can respire in the presence of exogenous heme and menaquinone. L. plantarum is of particular interest as it is a generalist LAB species found in insect, animal, and human digestive tracts and is essential for the production of many fermented foods (Behera et al., 2018; Duar et al., 2017). These findings have significance for the understanding of energy conservation strategies in primarily fermentative microorganisms and on lactic acid fermentations in food biotechnology.

Results

L. plantarum reduces extracellular electron acceptors

To determine whether L. plantarum can reduce extracellular electron acceptors, we first measured its ability to reduce insoluble ferrihydrite (iron (III) oxyhydroxide). Incubation of the model strain L. plantarum NCIMB8826 in the presence of ferrihydrite showed that this strain reduces Fe3+ to Fe2+ (Figure 1A and Figure 1—figure supplement 1A). Viable cells are required for iron reduction and this activity is dependent on the presence of exogenous quinone (DHNA, 1,4-dihydroxy-2-naphthoic acid) (Figure 1A and Figure 1—figure supplement 1A-B). The requirement for DHNA was hypothesized because DHNA is a precursor of demethylmenaquinone (DMK), a membrane electron shuttle utilized by L. monocytogenes for EET (Light et al., 2018), and L. plantarum lacks a complete DHNA biosynthetic pathway (Brooijmans et al., 2009a). For full activity, an electron donor (such as mannitol or glucose) was required to be present (Figure 1A and Figure 1—figure supplement 1A). Like L. monocytogenes (Light et al., 2018), the addition of riboflavin during the iron reduction assay also increased Fe3+ reduction in a dose-dependent manner (Figure 1—figure supplement 1C). Thus, L. plantarum reduces insoluble iron in a manner similar to L. monocytogenes.

Figure 1 with 3 supplements see all
L. plantarum can reduce both Fe3+ and an anode through EET.

(A) Reduction of Fe3+ (ferrihydrite) to Fe2+ by L. plantarum NCIMB8826 after growth in mMRS. The assays were performed in PBS supplemented with 20 µg/mL DHNA and/or 55 mM mannitol. Fe2+ was detected colorimetrically using 2 mM ferrozine. For L. plantarum inactivation, cells were incubated at 85℃ in PBS for 30 min prior to the assay. Significant differences were determined by one-way ANOVA with Tukey’s post-hoc test (n = 3), *** p < 0.001. (B) Two-chambered electrochemical cell setup for measuring current generated by L. plantarum. (C) Current density production over time by L. plantarum in CDM supplemented with 20 µg/mL DHNA and/or 110 mM mannitol. The anode was polarized at +0.2VAg/AgCl. The avg ± stdev of three biological replicates is shown. See also Figure 1—figure supplement 1 and Figure 1—figure supplement 2 and related data in Figure 1—source data 1.

Next, we investigated whether the ability of L. plantarum to reduce insoluble iron was altered by growth media. L. plantarum was able to reduce iron after growth in either complete (MRS) medium or chemically defined medium (CDM) (Figure 1—figure supplement 1B). Iron reduction was greater when mannitol, a sugar alcohol, rather than glucose, was provided as the sole carbon source in MRS (Figure 1—figure supplement 1B). However, reduction was highest when L. plantarum was incubated in mannitol-containing MRS (mMRS) with both DHNA and ferric ammonium citrate present (Figure 1—figure supplement 1D). The addition of riboflavin to the growth medium did not further increase iron reduction by L. plantarum (Figure 1—figure supplement 1E), potentially because riboflavin is already present in high quantities in MRS, a medium containing yeast extract (Tomé, 2021). Thus, L. plantarum was grown in mMRS supplemented with DHNA and iron before ferrihydrite reduction assays in all subsequent experiments.

L. plantarum EET activity was confirmed in a bioelectrochemical reactor by quantifying electron output as current (Figure 1B). L. plantarum reduced a carbon electrode (anode) polarized to +200 mVAg/AgCl in the presence of both DHNA and an electron donor (mannitol) (Figure 1C). No current was observed in the absence of L. plantarum (Figure 1—figure supplement 2A), indicating that current production stems from a biological process. L. plantarum produced a maximum current of 129 ± 19 µA/cm2 in mCDM (Figure 1C) and 225 ± 9 µA/cm2 in mMRS (Figure 1—figure supplement 2B). Under EET conditions in mCDM, the L. plantarum biomass was 2.7 mg (dry cell mass). Assuming 50% of the dry cell mass was protein, the specific electron transfer rate was 57 µmol electrons/mg- protein/hr and the current production was 1.5 mA/mg-protein. This value is lower than that reported for Geobacter sulfurrenducens (4–8 mA/mg-protein) (Marsili et al., 2010; Rose and Regan, 2015), the model species for direct EET, and higher than that of Shewanella oneidensis (0.67 mA/mg-protein) (Marsili et al., 2008), the model species for mediated EET. It should be noted that these species, unlike L. plantarum, can synthesize riboflavin and quinones and do not require the addition of either for EET activity. Similar to our iron reduction experiments, EET to an anode occurred with different electron donors and growth media (Figure 1—figure supplement 2B-C), and current increased after supplementation of riboflavin when it was omitted from the growth medium (Figure 1—figure supplement 2D). Because of these differences, CDM was amended with mannitol and riboflavin in subsequent experiments.

DHNA is found in concentrations of 0.089–0.44 μg/mL in commercial fermented beverages (Eom et al., 2012), and under laboratory conditions, microbes can synthesize and secrete DHNA leading to concentrations of 0.37–48 μg/mL (Isawa et al., 2002; Furuichi et al., 2006; Kang et al., 2015). To test whether EET in L. plantarum is relevant under these physiological concentrations, we probed whether L. plantarum can perform EET with a sub-physiological DHNA concentration of 0.01 μg/mL. Indeed, L. plantarum can reduce iron and produce significant current density (Figure 1—figure supplement 3A-B), although the magnitude of iron reduction and current was smaller than what was observed with 20 μg/mL. These results show that the concentrations of DHNA found in niches of L. plantarum can support EET and suggest the magnitude of EET will depend on the DHNA concentration.

Iron reduction by LAB is associated with the presence of ndh2 and pplA

Because iron reduction by L. monocytogenes requires the genes in a 8.5 kb gene locus encoding a flavin-based EET (FLEET) pathway (Light et al., 2018), we looked for the presence of these genes in 1,788 LAB genomes deposited in NCBI. Homology searches identified the complete FLEET locus in 11 out of 38 genera including diverse LAB such as Enterococcus and Lacticaseibacillus (Figure 2A). The other LAB genera either lack multiple FLEET pathway genes or, as was observed for all 68 strains of Lactococcus, contain all genes except for pplA, which is predicted to encode an extracellular flavin-binding reductase. Among the lactobacilli, genomes of 19 out of 94 species contain the entire FLEET system (Figure 2—figure supplement 1). The lactobacilli species with the entire FLEET locus are homofermentative and are distributed between different phylogenetic groups (Zheng et al., 2020). These data show that the FLEET locus is conserved across LAB genera besides L. plantarum, including other homofermentative LAB species known to colonize host and food environments.

Figure 2 with 2 supplements see all
The FLEET genes ndh2 and pplA are associated with iron reduction by LAB.

(A) Heatmap showing the genera in the Lactobacillales order containing FLEET genes. Homology searches were conducted using tBLASTx for 1788 complete LAB genomes in NCBI (downloaded 02/25/2021) against the L. plantarum NCIMB8826 FLEET locus. A match was considered positive with a Bit-score >50 and an E-value of <10–3. Arrows designate genera tested for iron reduction activity; green = EET-active with Fe3+, red = EET-inactive with Fe3+. (B) Reduction of ferrihydrite in PBS with 20 μg/mL DHNA and 55 mM mannitol after growth in mMRS supplemented with 20 μg/mL DHNA and 1.25 mM ferric ammonium citrate. The avg ± stdev of three biological replicates per strain is shown. (C) Relative expression of NCIMB8826 FLEET locus genes in mMRS with 20 μg/mL DHNA and 1.25 mM ferric ammonium citrate compared to growth in mMRS. Significant differences in expression were determined by the Wald test (n = 3) with a Log2 (fold change) > 0.5 and an FDR-adjusted p-value of <0.05. See also Figure 2—figure supplement 1 and Figure 2—figure supplement 2 and related data in Figure 2—source data 1.

Figure 2—source data 1

EET locus homology matches at genus level via tBLASTx from complete Lactobacillales genome dataset for Figure 2A, raw data for Figure 2B and relative expression data of Figure 2C.

https://cdn.elifesciences.org/articles/70684/elife-70684-fig2-data1-v1.xlsx

To determine whether LAB FLEET gene presence was associated with EET activity, a diverse collection of LAB strains were examined for their capacity to reduce ferrihydrite. The assay showed that isolates of L. plantarum, Lactiplantibacillus pentosus, Lacticaseibacillus rhamnosus, Lacticaseibacillus casei, Enterococcus faecium, and Enterococcus faecalis are capable of Fe3+ reduction (Figure 2B). The genomes of those species also contain a complete FLEET locus (Figure 2A and Figure 2—figure supplement 1). Conversely, strains of Lactococcus lactis, Ligilactobacillus murinus, Levilactobacillus brevis, Pediococcus pentosaceus, and Streptococcus agalactiae showed little to no iron reduction activity (Figure 2B). The presence of FLEET-associated genes varied between those species, but only strains of species found to contain both ndh2, a predicted membrane-bound, type-II NADH dehydrogenase, and pplA were able to reduce iron under the conditions tested.

L. plantarum NCIMB8826 exhibited the highest EET activity resulting in at least 2.5-fold greater Fe3+ reduction than the other L. plantarum strains (Figure 2B). Remarkably, however, the L. plantarum NCIMB8826 genome and the genomes of 138 other L. plantarum strains queried all harbored a complete FLEET locus including ndh2 and pplA (Figure 2—figure supplement 1 and Figure 2—figure supplement 2A). Among those strains tested for the capacity to reduce Fe3+, L. plantarum NCIMB700965 and 8.1 could not reduce Fe3+ but possessed all genes in the FLEET pathway. Closer examination of both strains by aligning their FLEET loci with NCIMB8826 revealed unique IS30-family transposons in the intergenic promoter regions spanning ndh2 and pplA (Figure 2—figure supplement 2A). These genes were minimally expressed in L. plantarum NCIMB700965 and 8.1 in comparison to NCIMB8826 (Figure 2—figure supplement 2B). ndh2 and pplA were also the only two genes in the FLEET gene locus that were induced when L. plantarum NCIMB8826 was incubated in mMRS supplemented with DHNA and iron (Figure 2C and Figure 1—figure supplement 1D). Both ndh2 and pplA were induced (~1.6 fold, p < 0.05) in MRS containing mannitol, DHNA, and ferric ammonium citrate (Figure 2C), but were not upregulated when either DHNA or ferric ammonium citrate were omitted from the culture medium (Figure 2—figure supplement 2C). Taken together, these data show that widespread iron reduction in LAB is tightly associated with the presence and upregulation of ndh2 and pplA, suggesting they are required for EET.

Ndh2 is required and PplA is conditionally required for L. plantarum EET

In order to confirm the necessity of ndh2 and pplA for EET in L. plantarum, we constructed ndh2 and ppA deletion mutants of L. plantarum NCIMB8826. Both mutants were significantly impaired in their capacities to reduce ferrihydrite compared with the wild-type strain (Figure 3A). The ndh2 and pplA deletion mutants also had different effects on the oxidation-reduction potential (ORP) of mMRS. ORP is defined as the ratio of all oxidative to reductive components in an environment (Killeen et al., 2018) and is an important environmental condition which influences the outcome of LAB fermentations such as flavor development in cheese (Morandi et al., 2016) and the growth of spoilage microorganisms (Olsen and Pérez-Díaz, 2009). Expectedly for the L. plantarum EET phenotype, significant reductions in mMRS ORP only occurred during L. plantarum growth when DHNA was included in the culture medium (Figure 3—figure supplement 1A). Although ORP declined for all three strains in a manner consistent with other ORP-reducing enzymatic activities (for example the reduction of oxygen by NADH oxidase) (Tachon et al., 2010), wild-type L. plantarum resulted in greater reductions in ORP compared to either mutant in mMRS, and these differences were significant at most time points measured over a 12 hr period (Figure 3B). The effects on ORP occurred in the absence of changes in growth rates, cell yields, and medium pH (Figure 3—figure supplement 1A-D). The ΔmVmax was reached during mid-exponential phase (approximately 5 hr) (Figure 3—figure supplement 1B), and at that time, wild-type L. plantarum cells but not the Δndh2 or ΔpplA strains were active in the ferrihydrite reduction assay (Figure 3—figure supplement 1E). This difference in ferrihydrite reduction activity similarly persisted in stationary phase cells (12 hr) (Figure 3—figure supplement 1F). These observations show that ndh2 and pplA contribute to the capacity of L. plantarum to reduce iron and have relevance to the ORP-dependent activities occurring during food fermentations (van Dijk et al., 2000).

Figure 3 with 2 supplements see all
L. plantarum requires ndh2 and conditionally pplA for EET.

(A) Reduction of Fe3+ (ferrihydrite) to Fe2+ with wild-type L. plantarum or EET deletion mutants in the presence of 20 μg/mL DHNA and 55 mM mannitol after growth in mMRS supplemented with 20 μg/mL DHNA and 1.25 mM ferric ammonium citrate. Significant differences determined by one-way ANOVA with Dunnett’s post-hoc test, **** p < 0.0001. (B) Redox potential of mMRS supplemented with 20 μg/mL DHNA and 1.25 mM ferric ammonium citrate after inoculation with wild-type L. plantarum or EET deletion mutants. Significant ORP differences between the wild-type and mutant strains determined by two-way RM ANOVA with Tukey’s post-hoc test, * p < 0.05 (WT vs. Δndh2); # p < 0.05 (WT vs. ΔpplA). (C) Current density generated by wild-type L. plantarum and deletion mutants in mCDM supplemented with 20 μg/mL DHNA. The avg ± stdev is shown. (D) Current density generated by L. plantarum and two L. lactis strains lacking pplA in mCDM. For L. plantarum, the mCDM was supplemented with 20 μg/mL DHNA. The data correspond to the average of two (D) or three (A to C) biological replicates per strain. See also Figure 3—figure supplement 1 and Figure 3—figure supplement 2 and related data in Figure 3—source data 1.

Use of an anode as an external electron acceptor instead of ferrihydrite showed a similar, but not identical genetic dependency. L. plantarum Δndh2 produced a significantly lower current density (Figure 3C) and a lower peak current (Figure 3—figure supplement 2A). Surprisingly, L. plantarum ΔpplA was able to produce the same amount of current as the wild-type strain, suggesting that the lipoprotein PplA is not essential and might not be involved in anode reduction through EET. This observation led us to investigate the anodic-EET ability of other LAB species lacking pplA like Lactococcus lactis (Figure 3D). DHNA was not provided to these strains because they can synthesize demethylmenaquinone and other quinones (Rezaïki et al., 2008). Both L. lactis strain IL1403 and strain KF147 were capable of current generation, confirming that PplA is not essential for LAB to produce current. This is consistent with the finding that other extracellular reductases besides PplA are responsible for EET activity in Gram-positive bacteria (Light et al., 2019). Taken together these results show that EET activity is dependent upon the presence of the putative FLEET locus, and specifically ndh2 and conditionally pplA.

L. plantarum increases energy conservation and balances intracellular redox state when performing EET

Building from studies in E. faecalis (Keogh et al., 2018), it has been suggested that EET improves growth by either enabling iron to be acquired as a macronutrient or by enhancing respiration (Jeuken et al., 2020). It is worth noting that several studies have shown that L. plantarum does not require iron to grow (Elli et al., 2000; Weinberg, 1997). To test whether EET allowed increased iron acquisition by L. plantarum, we measured intracellular iron by Inductively Coupled Plasma-Mass Spectrometry (ICP-MS). There was no significant difference in intracellular iron concentrations between L. plantarum growth in mMRS supplemented with DHNA and iron compared to growth in mMRS alone (Figure 4—figure supplement 1). Moreover, deletion of ndh2 did not significantly change the amount of intracellular iron (Figure 4—figure supplement 1). ICP-MS showed that other redox-active metals used for EET, such as manganese and copper (Kouzuma et al., 2012; Fan et al., 2018) were also not affected (Figure 4—figure supplements 1 and 2). In contrast to studies in E. faecalis in which iron supplementation leads to intracellular accumulation of this metal (Keogh et al., 2018), these data show that L. plantarum does not use EET to increase its acquisition of iron or other redox-active metals, suggesting EET may instead play a role in energy conservation.

We next sought to understand if EET impacts energy conservation in L. plantarum by comparing its growth and ATP levels in the presence of a polarized anode. The highest current density (i.e. greatest EET activity) produced by L. plantarum in mannitol CDM typically occurred within 24 hr after inoculation into the bioreactor (Figure 1C). At this point, there was an approximately 4-fold higher dry cell weight and 2-fold higher numbers of viable cells compared to L. plantarum incubated in open circuit (OC) conditions (Figure 4A–B) Current density declined from its maximum value when L. plantarum cells performing EET were in exponential growth (Figures 1C and 4C). By comparison, growth was not observed until two days later under OC conditions (Figure 4C). During peak current production, intracellular ATP levels were significantly higher (4.5-fold) under EET compared to OC conditions (Figure 4D and Table 1). These results strongly suggest faster ATP accumulation under EET conditions allowed L. plantarum to exit lag phase more rapidly. ATP levels were also greater in L. plantarum when in the presence of both mannitol and DHNA, compared to either mannitol or DHNA separately (Figure 4D). Thus, EET allows L. plantarum to initiate growth and accumulate ATP more rapidly, indicating that EET significantly increases energy conservation in L. plantarum.

Figure 4 with 3 supplements see all
Growth, ATP, and redox balance of L. plantarum changes when an anode is provided as an extracellular electron acceptor.

These measurements and the current density plot shown in Figure 1C are from the same experiment. (A) Viable cells and (B) dry weight at the point of maximum current density under current circulating conditions (EET) and at open circuit conditions (OC) at the same time point. (C) Change in cell numbers measured by OD600 over time in the bioreactors under EET (continuous line) and OC conditions (dotted lines). (D) ATP production per OD600 unit and (E) NAD+/NADH ratios at the point of maximum current density. The bioreactors were shaken vigorously to dislodge cells before sampling. The avg ± stdev of three biological replicates is shown. Significant differences were determined by one-way ANOVA with (A and B) Dunn-Sidak post-hoc test (n = 3) and (D and E) Dunnett’s post-hoc test (n = 3), * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001. See also Figure 1 panel C and Figure 4—figure supplement 3 and related data in Figure 4—source data 1.

Table 1
Bioenergetic balances suggest energy conservation under EET conditions occurs via substrate-level phosphorylation.

The reactors contained 20 µg/mL of DHNA and mannitol as the electron donor. Balances were calculated with data obtained by day four from Figure 5. See also Supplementary file 1. SLP stands for substrate-level phosphorylation. Yfermentation refers to the total fermentation products obtained (see Supplementary file 1) per mol of sugar consumed. Ymannitol is the ATP produced from the total fermentation products per mol of sugar consumed, and YATP is the dry weight measured per mol of ATP produced from fermentation products.

NADH consumed*Calculated ATP(from metabolites)Biomass yieldYfermentationYmannitolYATP
UnitsmMmMg-dw/mol-mannitolmmol product/mmol-mannitolmol ATP/mol mannitolg dw/mol ATP
EET6.44 ± 0.48 via anode16.69 ± 2.72 via SLP16.6 ± 1.54.85 ± 0.331.53 ± 0.131.59 ± 0.133.09 ± 0.36
OC5.51 ± 0.97 via SLP5.7 ± 0.67.21 ± 1.410.87 ± 0.090.89 ± 0.098.06 ± 0.86
  1. *

    Calculated based on production of 3 mol of NADH produced per mol of mannitol, 1 mol of NAD+ per lactate, 2 mol of NAD+ per ethanol, 2 mol of NAD+ mol per succinate produced and 0.5 mmol of NAD+ per mol of electrons harvested on the anode.

  2. Calculated based on production of 1 mol of ATP per lactate, 2 mol per acetate, 1 mol per ethanol, and 3 mol per succinate produced.

Because fermentation, anaerobic respiration, and aerobic respiration are each associated with a different NAD+/NADH ratio, energy conservation is linked to intracellular redox homeostasis (Holm et al., 2010). Therefore, we probed redox homeostasis in L. plantarum under EET conditions by measuring intracellular NAD+/NADH at the point of maximum current density (Figure 4E). L. plantarum showed an 8-fold higher NAD+/NADH ratio under EET conditions compared to OC (Figure 4E). This result was not limited to the presence of a polarized anode as L. plantarum also contained a significantly higher NAD+/NADH ratio when Fe3+ was available as a terminal electron acceptor (Figure 4—figure supplement 3). These NAD+/NADH ratios are more similar to those found for in E. coli performing aerobic respiration (de Graef et al., 1999) or G. sulfurreducens performing anaerobic respiration than in LAB performing fermentation (Guo et al., 2017). Taken together, our data indicate that EET is involved in energy conservation, and the intracellular redox balance during EET mimics a respiratory rather than a fermentative process.

EET increases fermentative metabolism through substrate-level phosphorylation and reduction in extracellular pH

Metal-reducing bacteria use EET in anaerobic respiration (Richter et al., 2012; Shi et al., 2007). Ndh2 is considered an anaerobic respiratory protein, and L. plantarum can perform anaerobic respiration with exogenous menaquinone and heme using an electron transport chain (Brooijmans et al., 2009b). This led us to hypothesize that those electron transport proteins could also be involved for EET to conserve energy as part of anaerobic respiration. To test this hypothesis, we examined whether any of the known electron transfer proteins needed for PMF generation in aerobic and anaerobic respiration are required for L. plantarum EET. Neither the addition of heme to restore bd-type cytochrome (cydABCD) used in aerobic respiration, nor deletion of the respiratory nitrate reductase (ΔnarGHJI) significantly altered current production (Figure 5—figure supplement 1A-B). Because Ndh2 is a type-II NADH dehydrogenase which does not contribute to a proton gradient (Lin et al., 2008; Nakatani et al., 2020), these observations show that while EET does involve a respiratory protein, it does not involve any of the known PMF-generating electron transfer proteins in L. plantarum.

Respiration is also associated with the tricarboxylic acid (TCA) cycle. L. plantarum, like other LAB, does not possess an oxidative branch of the TCA cycle and only contains a reductive branch (Tsuji et al., 2013). To probe whether the reductive branch was active during EET, we also examined production of succinate, the terminal end-product of the reductive branch. EET did not increase the succinate concentration (Figure 5—figure supplement 2). Moreover, we did not detect any intermediates of the reductive branch of the TCA cycle, that is oxalacetate, malate, or fumarate. This indicates that EET did not cause additional metabolic flux through its TCA cycle. Thus, none of the known metabolic pathways or electron transport proteins associated with anaerobic respiration, besides Ndh2, are required for EET. These results suggest increased energy conservation during EET in L. plantarum is not through canonical anaerobic respiration.

An alternative hypothesis is that increased energy conservation under EET conditions is driven by changes in fermentation. L. plantarum uses glycolysis to convert mannitol to two molecules of pyruvate which are then converted mainly to lactate or ethanol via NADH-consuming steps, or acetate via an ATP-generating reaction using substrate-level phosphorylation (Dirar and Collins, 1972). Thus, shifting toward production of acetate from to lactate or ethanol production can increase ATP yield during fermentation. Additionally, NADH can be re-generated by oxidizing pyruvate to yield 2,3-butanediol, using acetoin as an intermediate. Fermentation in L. plantarum also decreases the pH of the surrounding media.

To probe changes in fermentation, we measured the concentrations of mannitol, acetate, lactate, ethanol, acetoin, 2,3-butanediol, formate, and pyruvate and the pH in L. plantarum cultures during OC and EET conditions. After four days, we accounted for ~80% and ~ 55% of the total carbon under EET and OC conditions (for all metabolite concentrations see Supplementary file 1), giving us a quantitative view of metabolism under EET conditions. Surprisingly, under EET conditions, the distribution of major end-fermentation products (acetate, lactate, and ethanol) did not change, but their yield per cell was 2.6-fold higher compared to OC conditions (Figure 5A). While we did not detect acetoin or 2,3-butanediol, formate was found at trace levels, and pyruvate was found at similar, low levels under EET and OC conditions (Figure 5—figure supplement 2). After accounting for mannitol consumption, we observed that EET allowed cells to produce ~1.75 x more fermentation products per each mol of mannitol utilized (Yfermentation, Table 1). The culture medium pH was also significantly lower than under OC (Figure 5B), a result which may indicate that EET conferred higher levels of acid stress on L. plantarum, and therefore, reductions in cell viability, despite EET leading to higher cell numbers overall (as measured by dry cell weight) (Figure 4A–B). A similar acidification of the medium was observed for ΔpplA, but not for Δndh2, when an anode was present as electron acceptor, indicating that ndh2-dependent EET is needed to decrease the pH (Figure 3—figure supplement 2B). When much lower, sub-physiological levels of DHNA were supplied (0.01 μg/mL), a smaller but significant decrease in the pH of the medium was also observed (Figure 1—figure supplement 3C). Overall, these results show that EET allows L. plantarum to ferment to ~1.75 x greater extent and to acidify the medium to a greater extent as well.

Figure 5 with 2 supplements see all
Fermentation fluxes are increased when an anode is provided as an extracellular electron acceptor.

Results are from the same set of experiments as the current density plot shown in Figure 3C. (A) Metabolic yields of L. plantarum end-fermentation products under open circuit conditions (OC) and current circulating conditions (EET) in mCDM supplemented with 20 μg/mL DHNA. (B) pH measurements and (C) mannitol, (D) lactate, (E) acetate, and (F) ethanol concentrations over time under OC and EET conditions. (G) Schematic of proposed model for NADH regeneration during fermentation of mannitol in the presence of an anode as electron sink for L. plantarum. The avg ± stdev of three biological replicates is shown. Significant differences were determined by one-way ANOVA with Dunn-Sidak post-hoc (n = 3), ** p < 0.01. See also Figure 5—figure supplement 1 and Figure 5—figure supplement 2 and related data in Figure 5—source data 1.

We also observed that EET led to higher cellular metabolic fluxes, that is, higher changes in metabolites per cell per unit time. Although the final OD600nm and dry cell weight were not significantly different (Supplementary file 1), L. plantarum utilized mannitol and produced acetate and lactate more rapidly under EET than OC conditions (Figure 5C–F). Cells performing EET were ~2 fold faster at consuming mannitol (Figure 5C) between days 1 and 3. Mannitol consumption increased between day 1 and day 2, approximately when the cells transitioned to higher current density (Figure 3C), suggesting that increased EET drove that increased consumption. The overall rates of acetate and lactate production also increased 3.4 and 3.6 times (Figure 5D–E), respectively. Measurements of metabolites produced by ΔpplA and Δndh2 strains confirmed that, like for current production to an anode, the EET-associated increased metabolic flux in L. plantarum requires the presence of Ndh2, but not PplA (Figure 3—figure supplement 2C). Overall, these data indicate that ndh2-dependent EET increases both the flux and final yield of fermentation in L. plantarum.

Because the production of acetate yields ATP, these results also suggested that the increase in ATP generation under EET conditions may be due to substrate-level phosphorylation. To probe whether EET-associated increase in fermentative flux could account for the changes in ATP generation, we calculated fermentation balances (Table 1). Our measurements account for 80% of the carbon under EET conditions (see Supplementary file 1), leaving a maximum of ~20% systematic uncertainty in these calculations. The concentrations of fermentation products detected (Supplementary file 1) were used to estimate the total ATP in the presence and absence of EET. The estimated ATP was 3-fold higher under EET conditions than OC conditions (Table 1), a result that is consistent with the ~2.5 fold higher accumulation of ATP measured at maximum current density (Figure 4D). Overall, this quantitative analysis shows that the vast majority of the increased energy conservation under EET conditions can be accounted for by an increase in fermentation yield and substrate-level phosphorylation.

EET shifts how L. plantarum uses electron acceptors and converts ATP into biomass

Thus far, our results provided an unusual picture of the energy metabolism of L. plantarum under EET conditions; while EET significantly shifted the intracellular redox state to a more respiratory-like balance, its increased ATP yield was mainly accounted for by an increased fermentative yield. Another major difference in fermentation and anaerobic respiration is the use of the endogenous versus exogenous electron acceptors. To more deeply understand how L. plantarum uses organic molecules and the anode as electron acceptors when performing EET, the electron balances under EET and OC conditions were calculated (Table 1 and Supplementary file 1). We estimated the NADH produced using two different methods (see Supplementary file 1 for methodology) and the NADH re-oxidized through the reduction of the anode (measured as current) and via substrate-level phosphorylation. This allowed us to obtain a global balance of the NAD+/NADH ratio. Under OC conditions between 35% and 66% of the NADH produced from the oxidation of mannitol to pyruvate (a range is given using the two methods used) was re-oxidized to NAD+ (5.5 mM NADH consumed, Table 1), qualitatively agreeing with the low NAD+/NADH ratios measured (Figure 4E). In contrast, electron balance calculations showed that between 77% and 96% of the NADH produced under EET conditions was re-oxidized (17 mM NADH consumed, Table 1), a result that is consistent with the significantly higher NAD+/NADH ratios measured (Figure 4E). Interestingly, these calculations estimate that 55–69% of the total NADH generated was oxidized through fermentation, while 21–28% of the NADH was oxidized using the electrode as a terminal electron acceptor (Table 1). Thus, L. plantarum growing under EET conditions achieves a more oxidized intracellular redox balance by more completely fermenting mannitol to lactate and ethanol and by using the electrode as a terminal electron acceptor (Figure 5G). These observations reinforce that the energy metabolism of L. plantarum under EET conditions utilizes elements of both fermentation and anaerobic respiration.

In rapidly dividing cells, energy conservation, a catabolic process, is associated with growth, an anabolic process. However, catabolism need not be coupled with anabolism (Russell and Cook, 1995). To determine how catabolic and anabolism are connected under EET conditions, the ATP requirements to grow biomass (YATP) were estimated using the calculated ATP and the measured dry weight. Under OC conditions, the YATP obtained (8.06 ± 0.8 g dw/mol ATP) for L. plantarum was similar to that observed previously (10.9 g-dw/mol ATP) (Dirar and Collins, 1972). Hence, without EET, the ATP generated from fermentation was converted into biomass nearly at the expected efficiency. In contrast, a significantly lower YATP was reached for L. plantarum performing EET (3.07 ± 0.35 g dw/mol ATP) (Table 1). This observation indicates that under EET conditions, either more ATP is required to produce biomass or more ATP is utilized by other functions such as for PMF-generation and intracellular pH regulation (Russell and Cook, 1995). EET conditions also resulted in 79% more ATP per mol of fermented mannitol (Ymannitol). Consequently, molar biomass yields (g-dw/mol-mannitol) under EET conditions were significantly lower (Table 1), in agreement with previous observations in respiratory electroactive species (Esteve-Núñez et al., 2005). These calculations show that when L. plantarum performs EET, anabolism and catabolism processes are differently coupled than under OC conditions. ATP is produced more efficiently, but this it is less efficiently utilized to make biomass. Overall, these results show an intriguing pattern of coupling between anabolism and catabolism, indicative of a novel energy metabolism in L. plantarum during EET.

EET is active in vegetable fermentations

Our results inspired us to explore whether EET could occur in a physiological niche of LAB such as fermented foods. LAB are necessary for the making of many fermented fruit and vegetable foods and the properties of those foods depend on the metabolic diversity of the LAB strains present (Gänzle, 2015). Plant tissues also contain a much wider variety of carbon substrates and potential electron acceptors than the CDM used in our prior experiments. To study the physiological and biotechnological relevance of EET in food fermentations, kale juice was fermented using L. plantarum as a starter culture (Figure 6A). The fermentation of kale juice was measured under EET conditions (a polarized anode with or without DHNA), and an OC control (a nonpolarized anode with DHNA) was used to separate the role of DHNA and electron flow to the anode on the fermentation process. An additional bioreactor without cells, but with DHNA, was operated to identify any possible electrochemical-driven conversion of substrates. When L. plantarum was added to the prepared kale juice, approximately 10-fold more current was generated during EET conditions with DHNA (EET+ DHNA) as compared to abiotic and biological non-EET promoting conditions (no DHNA) (Figure 6B). This current was comparable to the current generated in laboratory medium (Figure 3C), indicating that robust EET by LAB is possible in the complex physiological conditions of a food fermentation.

Figure 6 with 1 supplement see all
EET in a kale juice increases the production of fermentation end products.

(A) Preparation of kale juice medium used for fermentation in bioelectrochemical reactors. (B) Current density production measured from kale juice medium over time in the presence of L. plantarum and 20 μg/mL DHNA, no DHNA, or under abiotic conditions with addition of 20 μg/mL DHNA. The anode polarization was maintained at 0.2 VAg/AgCl. (C) Normalized total quantities of the metabolites detected per cell (CFUmax used for calculations). (D) pH measurements over time under different conditions tested on a second set of kale juice fermentations performed under the same conditions. (E) Production rate per viable cell, r, of lactate, acetate, and succinate. The avg ± stdev of three biological replicates is shown. See also Figure 6—figure supplement 1 and related data in Figure 6—source data 1.

We next investigated the impact of EET on L. plantarum growth and metabolism in the kale juice fermentation. Significant changes in the pH and fermentation products were detected under EET conditions (Figure 6C–D). These differences occurred in the absence of significant changes in viable cell numbers (Figure 6—figure supplement 1A) at the time points measured. As previously observed using laboratory culture media, an approximately 2-fold greater accumulation of total end-fermentation products per cell was obtained when cells interacted with an anode in the presence of DHNA (Figure 6C). In the kale juice fermentation, EET+ DHNA conditions enhanced both lactate and acetate production per cell without changing the distribution of metabolites (Figure 6E and Figure 6—figure supplement 1B). Thus, when DHNA was provided, EET enhanced the overall yield of fermentation end-products and their production rates per cell, mimicking our observations in laboratory medium (Figure 5C–D). EET also led to a significantly higher acidification of the kale juice compared to OC conditions, and the presence of DHNA dramatically enhanced this pH drop (Figure 6D). In general, when no DHNA was supplied but an anode was present as an electron acceptor, the fermentation process was very similar to OC conditions. This means in kale juice, a source of quinones is essential to support L. plantarum EET activity. Overall, these results show that EET under physiological conditions impacts cellular metabolism in L. plantarum by increasing metabolic flux which ultimately can affect the flavor profile of fermented foods (Chen et al., 2017).

Discussion

Increases in fermentation and energy conservation from EET have important bioenergetic implications for the mainly fermentative LAB. We showed that L. plantarum and other diverse LAB species perform EET if riboflavin and quinones are present. L. plantarum EET activity requires an NADH dehydrogenase (Ndh2) and conditionally requires an extracellular, flavin-binding reductase (PplA). EET in L. plantarum generates a high NAD+/NADH ratio, increases fermentation yield and flux, shortens lag phase, and increases ATP production. Thus, EET in L. plantarum is a hybrid energy metabolism that contains metabolic features of fermentation, redox features of anaerobic respiration, and predominately uses substrate-level phosphorylation to conserve energy. This pathway is active in L. plantarum with physiologically relevant DHNA concentrations and in a food fermentation and results in an increased metabolic flux and acidification rate.

The combined EET fermentation hybrid metabolism is distinct from anaerobic respiration, fermentation, and other energy conservation strategies

When performing EET, the metabolism of L. plantarum, a primarily fermentative bacterial species, is fundamentally different from EET-driven, anaerobic respiration of metal-reducing bacteria. Although aspects of EET in L. plantarum and metal-reducing Geobacter spp. are similar, such as the upregulation of NADH dehydrogenase, the reduction rate of extracellular electron acceptors, and the high NAD+/NADH ratio, other aspects of energy metabolism during EET in these two organisms are starkly different (see comparison in Supplementary file 2). Geobacter spp. direct their metabolic flux through the TCA cycle, rely almost exclusively on extracellular electron acceptors to regenerate NADH, and produce ATP exclusively through oxidative phosphorylation. In contrast, L. plantarum regenerates a substantial fraction of its NADH by directing metabolic flux through fermentative pathways. Additionally, oxidative phosphorylation is not a major mechanism of energy conservation in L. plantarum during EET, as supported by three lines of evidence: the marginal metabolic flux through the reduced branch of TCA cycle, no involvement of known PMF-generating proteins, and that increased ATP levels can be accounted for by increased substrate-level phosphorylation. While additional data are required to eliminate the possibility that oxidative phosphorylation is occurring in L. plantarum during EET, we can qualitatively state that substrate-level phosphorylation is the major mechanism for ATP generation.

Comparing EET and respiratory metabolism in LAB also reveals substantial differences in these metabolisms (Supplementary file 2). While both metabolisms require quinones, respiration also requires exogenous heme. Our findings and similar findings in E. faecalis (Pankratova et al., 2018) confirm that heme is not required for EET. Moreover, EET also differs from respiration in LAB because it occurs at the start of or prior to exponential phase growth, does not change the final cell density, and increases fermentation with no effect on the resultant proportions of lactate, acetate, and ethanol (Duwat et al., 2001). Thus, EET in LAB diverges from respiration in metal-reducing bacteria or LAB in its metabolic pattern and energetic consequences. In addition, while EET provokes a shift in fermentative metabolism in other bacteria upon the addition of artificial mediators (Vassilev et al., 2021; Emde and Schink, 1990), L. plantarum EET is active upon the presence of a mediator present in a complex food system.

This EET mechanism is also a novel energy conservation strategy compared to known fermentative metabolisms in LAB (comparison in Supplementary file 2). L. plantarum and other LAB, reduce alternative intracellular electron acceptors like citrate, fructose, and phenolic acids, resulting in increased intracellular NAD+/NADH ratios (Hansen, 2018). This metabolic activity is especially important for heterofermentative LAB in order to synthesize additional ATP through acetate kinase (Gänzle, 2015). Unlike these examples, however, the reduction of extracellular Fe3+ or an anode by EET requires a respiratory protein (Ndh2) and the shuttling of electrons outside of the cell. In addition, the reduction of the oxygen and organic compounds for cofactor regeneration by LAB leads to a metabolic shift toward acetate production and altered metabolic end-product ratios (Gänzle, 2015), which does not occur during EET. These differences show how the hybrid metabolism under EET conditions is distinct from other pathways that alleviate reduced intracellular conditions in LAB.

Previous studies have reported a simultaneous use of fermentation and electron transport elements, such as in respiro-fermentation in Saccharomyces cerevisiae (Blom et al., 2000). However, respiro-fermentation produces ATP and maintains intracellular redox balance through substrate-level phosphorylation and/or oxidative phosphorylation using separate pathways. Our data strongly suggests a single pathway is responsible for both ATP generation and intracellular redox balance. This hybrid fermentation mode is also different from the electron bifurcating mechanism, in which the extra ATP generation is driven by the creation of a H+ or Na+ potential from the oxidation of a ferredoxin (Buckel and Thauer, 2013). Unlike in this example, EET in LAB does not involve PMF creating elements and EET drives ATP generation through substrate-level phosphorylation. Another poorly understood example of the use of substrate-level phosphorylation and electron transport chains to balance intracellular redox state is found in the non-fermentative bacterium S. oneidensis. Although this species is a respiratory bacterium, it relies predominately on substrate-level phosphorylation to grow anaerobically with the exogenous electron acceptor fumarate (Hunt et al., 2010). In this scenario, it is unclear if there are changes in intracellular redox state or metabolism in this species.

In contrast to and expanding upon these studies, our work elucidates a qualitatively and quantitatively different blending of fermentation and respiration. L. plantarum EET-associated metabolism contains features of both fermentation (e.g. substrate-level phosphorylation, high fermentation product yields) and respiratory metabolisms (e.g. NAD+/NADH ratios, NADH dehydrogenase required) (Supplementary file 2). Quantitatively, this hybrid metabolism leads to an overall ~1.75x-more efficient and ~1.75x-faster energy conservation (increased Ymannitol, mannitol flux), but an overall ~1.5 fold weaker coupling between anabolism and catabolism (lower YATP). Additionally, the increased NAD+/NADH ratio arises from using an ~2:1 ratio of endogenous to extracellular electron acceptors. Thus, to our knowledge, the hybrid strategy that L. plantarum uses to generate ATP performing EET constitutes a novel mode of energy conservation in a primarily fermentative microorganism.

Different mechanisms of EET appear to be widespread in LAB

Based on our observations and others, we propose that EET is widespread in LAB and occurs by different mechanisms. Besides L. plantarum, we showed that L. lactis is able to generate current despite lacking pplA. Current generation by L. lactis was observed previously, found to be riboflavin dependent, and resulted in a small metabolic shift (yet to be defined) in which the flux through NADH-oxidizing pathways was reduced and ATP generating pathways were increased (Freguia et al., 2009; Masuda et al., 2010). L. lactis can also perform EET by reduction of tetrazolium violet and this activity depends on the presence of both quinones and an NADH dehydrogenase (NoxAB) (Tachon et al., 2009). E. faecalis is another LAB that performs EET, and similar to L. plantarum, it requires quinones (Pankratova et al., 2018) and a type-II NADH dehydrogenase (Ndh3) (Hederstedt et al., 2020) for Fe3+ reduction. In contrast to this mechanistic similarity, E. faecalis performs EET using matrix-associated iron resulting in both increased final cell biomass and intracellular iron (Keogh et al., 2018). Moreover, unlike L. plantarum and L. lactis, the presence of PplA is not necessary for anode reduction or Fe3+ reduction (Hederstedt et al., 2020). The conditional need for PplA in EET may be explained by the different prior growth conditions used and/or related to the existence of different mechanisms and proteins depending on the redox potential of the extracellular electron acceptor. Other flavin-binding, extracellular reductases amongst Gram-positive organisms, such as FrdA (acting on fumarate) have been identified in L. monocytogenes and UrdA (acting on urocanate) in Enterococcus rivorum (Light et al., 2019). Thus, there may exist a yet unidentified extracellular reductase in L. plantarum and L. lactis required for anode reduction. Thus, our findings elucidate a new pattern of metabolic changes associated with EET. It seems likely that these many mechanisms reflect the ability of EET to alleviate constraints of intracellular redox balance in fermentative metabolism across LAB.

EET has important implications for ecology and biotechnology of LAB

Conservation of the FLEET locus among different LAB species supports the premise that this hybrid fermentation with EET provides an important metabolic strategy for these bacteria in their natural habitats. LAB with a complete FLEET locus are homofermentative, thus underscoring the distinct ways homofermentative and heterofermentative LAB have evolved for energy conservation (Salvetti et al., 2013). L. plantarum and other LAB with FLEET systems such as L. casei are genetically and metabolically diverse and grow in a variety of nutrient rich environments including dairy and plant foods and the digestive tract (Cai et al., 2009; Martino et al., 2016; Siezen et al., 2010a). Those environments also are rich sources of sources of quinones, flavins, and extracellular electron acceptors such as iron (Cataldi et al., 2003; Fenn et al., 2017; Kim, 2017; Roughead and McCormick, 1990; Walther et al., 2013). Increased organic acid production and environmental acidification by LAB with this hybrid metabolism would provide an effective mechanism to inhibit competing microorganisms and confer a competitive advantage for growth. The increased ATP relative to biomass generation observed during growth on mannitol might also give sufficient readiness for using this energy later on to outcompete neighboring organisms (Russell and Cook, 1995). These effects of EET may be particularly important on plant tissues and intestinal environments, wherein LAB tend to be present in low numbers. Besides our observation that L. plantarum performs EET in kale juice, the FLEET pathway is important for intestinal colonization by both L. monocytogenes (Light et al., 2018) and E. faecalis (Lam et al., 2019), and L. plantarum FLEET genes including ndh2 and pplA were highly induced in the small intestine of rhesus macaques (Golomb et al., 2016).

The hybrid fermentation metabolism of LAB also has technological relevance. For many LAB food fermentations, acidification of the food matrix is required to prevent the growth of undesired microorganisms and result in a more consistent and reproducible product (Marco et al., 2021). Starter cultures are frequently selected based on their capacity for rapid growth and acid production (Bintsis, 2018). In the presence of an anode, exposure of L. plantarum to EET conditions during kale juice fermentation increased the acidification rate. Thus, this shows that EET metabolism is active in complex nutritive environments such as kale leaf tissues that contain other potential electron acceptors besides the anode and diverse electron donors (glucose, fructose, sucrose) (Thavarajah et al., 2016). This example also shows how electro-fermentation, the technological process by which fermentation is modulated using electrodes, can be used to control food fermentations (Moscoviz et al., 2016; Schievano et al., 2016; Vassilev et al., 2021). Because L. plantarum also increased fermentation flux when the electrode was available as an electron sink, higher quantities of organic acid flavor compounds were formed. Therefore, by the manipulation of extracellular redox potential, food electro-fermentations may be used to control microbial growth. This would allow the creation of new or altered sensory profiles in fermented foods, such as through altered organic acid production and metabolism or synthesis of other compounds that alter food flavors, aromas, and textures.

Final perspective

We expect that our study will improve the current understanding of energy conservation in primarily fermentative microorganisms and contribute to establishing the ecological relevance of EET in lactic acid bacteria. This work will ultimately allow the use of EET to electronically modulate the flavor and textural profiles of fermented foods and expand the use of lactic acid bacteria in bioelectronics, biomedicine, and bioenergy (Moscoviz et al., 2016). The identification of the precise components and full bioenergetics involved in L. plantarum EET will be key to unravel physiological and ecological questions and to develop other biotechnological applications.

Materials and methods

Key resources table
Reagent type (species) or
resource
DesignationSource or referenceIdentifiersAdditional information
Strain, strain background (Lactiplantibacillus plantarum)NCIMB8826Dandekar, 2019Strain information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus plantarum)NCIMB8826-RYin et al., 2018Rifampicin-resistant mutant of NCIMB8826Strain information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus plantarum)MLES100This studyDeletion mutant of NCIMB8826 lacking ndh2Plasmid information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus plantarum)MLES101This studyDeletion mutant of NCIMB8826 lacking pplAPlasmid information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus plantarum)MLEY100This studyDeletion mutant of NCIMB8826 lacking narGHIJPlasmid information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus plantarum)B1.3Yin et al., 2018Strain information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus plantarum)AJJ11Yu et al., 2021Strain information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus plantarum)8.1Yu et al., 2021Strain information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus plantarum)ATCC 202195Wright et al., 2020Strain information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus plantarum)NCIMB700965Heeney and Marco, 2019Strain information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus pentosus)BGM48Golomb et al., 2013Strain information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus casei)BL23Mazé et al., 2010Strain information listed in Supplementary file 3
Strain, strain background (Levilactobacillus brevis)ATCC 367Makarova et al., 2006Strain information listed in Supplementary file 3
Strain, strain background (Lactococcus lactis)KF147Siezen et al., 2010bStrain information listed in Supplementary file 3
Strain, strain background (Lactococcus lactis)IL1403Bolotin et al., 2001Strain information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus Rhamnosus)GGKankainen et al., 2009Strain information listed in Supplementary file 3
Strain, strain background (Lactiplantibacillus murinus)ASF361Wannemuehler et al., 2014Strain information listed in Supplementary file 3
Strain, strain background (Enterococcus faecalis)ATCC 29212Minogue et al., 2014Strain information listed in Supplementary file 3
Strain, strain background (Enterococcus faecalis)ATCC 8459Kopit et al., 2014Strain information listed in Supplementary file 3
Strain, strain background (Pediococcus pentosaceus)ATCC 25745Makarova et al., 2006Strain information listed in Supplementary file 3
Strain, strain background (Streptococcus agalactiae)ATCC 27956McDonald and McDonald, 1976Strain information listed in Supplementary file 3
Strain, strain background(Escherichia coli)DH5αTaylor et al., 1993fhuA2 lac(del)U169 phoA glnV44 Φ80’ lacZ(del) M15 gyrA96 recA1 relA1 endA1 thi−one hsdR17, amplification of cloning vector

Strains and culture conditions

Request a detailed protocol

All strains and plasmids used in this study are listed in Supplementary file 3. Standard laboratory culture medium was used for routine growth of bacteria as follows: Lactiplantibacillus spp., Lacticaseibacillus spp., Levilactobacillus brevis, Ligilactobacillus murinus, and Pediococcus pentosaceus, MRS (BD, Franklin Lakes, NJ, USA); Lactococcus lactis and Streptococcus agalactiae, M17 (BD) with 2% w/v glucose; Enterococcus faecalis, and Enterococcus faecium, BHI (BD); and Escherichia coli, LB (Teknova, Hollister, CA, USA). Bacterial strains were incubated without shaking except for E. coli (250 RPM) and at either 30 or 37 °C. Where indicated, strains were grown in filter-sterilized MRS (De MAN et al., 1960) lacking beef extract with either 110 mM glucose [gMRS] or 110 mM mannitol [mMRS], or a chemically defined minimal medium (Supplementary file 4) with 125 mM glucose [gCDM] or 125 mM mannitol [mCDM] for 18 hr (Aumiller et al., 2021). Riboflavin (1 mg/L) was routinely added to the CDM. When indicated, culture medium was supplemented with 20 μg/mL of the quinone 1,4-dihydroxy-2-naphthoic acid (DHNA) (Alfa Aesar, Haverhill, MA, USA), 1.25 mM ferric ammonium citrate (C6H8FeNO7) (1.25 mM) (VWR, Radnor, PA, USA), riboflavin (Sigma-Aldrich, St. Louis, MO, USA), or 5 μg/mL erythromycin (VWR).

DNA sequence analysis

Request a detailed protocol

The FLEET gene locus of L. plantarum NCIMB8826 was identified using NCBI basic local alignment search tool (BLAST) (McGinnis and Madden, 2004) using the L. monocytogenes 10403S FLEET genes (lmo2634 to lmo2641) as a reference. L. plantarum genes were annotated based on predicted functions within the FLEET pathwa (Light et al., 2018). FLEET locus genes were identified in other LAB by examining 1,788 complete Lactobacillales genomes available at NCBI (downloaded 02/25/2021). A local BLAST (ver 2.10.1) database containing these genomes was queried using tBLASTx with NCIMB8826 FLEET genes a reference. A gene was considered to be present in the Lactobacillales strain genome if the Bit-score was >50 and the E-value was <10–3 (Pearson, 2013). Heatmaps showing the percentage of strains in Lactobacillales genera and the Lactobacillus-genus complex (Zheng et al., 2020) identified to contain individual FLEET genes were visualized using the R-studio package ggplot2 (Wickham, 2011) with clustering done through UPGMA. The FLEET loci of L. plantarum strain 8.1 and NCIMB700965 were aligned to the NCIMB8826 genome in MegAlign Pro (DNAstar Inc, Madison, WI, USA).

Insoluble iron reduction assays

Request a detailed protocol

Cells were collected by centrifugation at 10,000 g for 3 min, washed twice in phosphate-buffered saline (PBS), pH 7.2 (http://cshprotocols.cshlp.org), and adjusted to an optical density (OD) at 600 nm (OD600nm) of 2 in the presence of 2.2 mM ferrihydrite (Schwertmann and Fischer, 1973; Stookey, 2002) and 2 mM ferrozine (Sigma-Aldrich). Where indicated, 55 mM glucose or mannitol, 20 μg/mL DHNA, and riboflavin were added. After 3 hr incubation at 30 °C, the cells were collected by centrifugation at 10,000 g for 5 min and the absorbance of the supernatant was measured at 562 nm with a Synergy 2 spectrophotometer (BioTek, Winooski, VT, USA). Quantities of ferrihydrite reduced were determined using a standard curve containing a 2-fold range of FeSO4 (Sigma-Aldrich) (0.25 mM to 0.016 mM) and 2 mM ferrozine. The FeSO4 was dissolved in 10 mM cysteine-HCl (RPI, Mount Prospect, IL, USA) to prevent environmental re-oxidation of Fe2+ to Fe3+ in the standard curve. For testing iron reduction activity of cells with a DHNA concentration of 0.01 μg/mL in the medium, iron(III) oxide nanoparticles < 50 nm (Sigma-Aldrich) were used as insoluble iron form (Figure 1—figure supplement 3).

L. plantarum mutant construction

Request a detailed protocol

L. plantarum NCIMB8826 ndh2, pplA, and narGHIJ deletion mutants were constructed by double-crossover homologous recombination with the suicide plasmid pRV300 (Leloup et al., 1997). For mutant construction, upstream and downstream flanking regions of these genes were amplified using the A/B and C/D primers, respectively, listed in Supplementary file 5. Splicing-by-overlap extension (SOEing) PCR was used to combine PCR products as previously described (Heckman and Pease, 2007). PCR products were digested with restriction enzymes EcoRI, SacI, SacII, or SalI (New England Biolabs, Ipswich, MA, USA) for plasmid ligation and transformation into E. coli DH5α. The resulting plasmids were then introduced to L. plantarum NCIMB8826 by electroporation. Erythromycin-resistant mutants were selected and confirmed for plasmid integration by PCR (see Supplementary file 5 for primer sequences). Subsequently, deletion mutants were identified by a loss of resistance to erythromycin, PCR (see Supplementary file 5 for primer sequences) confirmation, and DNA sequencing (http://dnaseq.ucdavis.edu).

Bioelectrochemical reactors (BES) construction, operation, and electrochemical techniques

Request a detailed protocol

L. plantarum NCIMB8826 strains were grown overnight (approximately 16–18 hr) from glycerol stocks in MRS. Cells were harvested by centrifugation (5200 g, 12 min, 4 °C) and washed twice in PBS. When L. plantarum wild-type EET activity versus the Δndh2 (MLES100) and ΔpplA (MLES101) deletion mutants was compared, cells were grown as described and the number of cells was normalized across the three strains prior to inoculation in the BES. The bioreactors consisted of double-chamber electrochemical cells (Adams & Chittenden, Berkeley, CA) (Figure 1B) with a cation exchange membrane (CMI-7000, Membranes International, Ringwood, NJ) that separated them. A three-electrode configuration was used consisting of an Ag/AgCl sat KCl reference electrode (BASI, IN, USA), a titanium wire counter electrode, and a 6.35-mm-thick graphite felt working electrode (anode) of 4 × 4 cm (Alfa Aesar, MA, USA) with a piece of Ti wire threaded from bottom to top as a current collector and connection to the potentiostat. We used a Bio-Logic Science Instruments (TN, USA) potentiostat model VSP-300 for performing the electrochemical measurements (chronoamperometry). The bioreactors were sterilized by filling them with ddH2O and autoclaving at 121 °C for 30 min. The water was then removed and replaced with 150 mL of filter sterilized mMRS or mCDM media for the working electrode chamber, and 150 mL of M9 medium (6.78 g/L Na₂HPO₄, 3 g/L KH2PO4, 0.5 g/L NaCl, 1 g/L NH4Cl) (BD) for the counter electrode chamber. Both media of the working electrode chamber were supplemented with 20 μg/mL DHNA or 0.01 μg/mL diluted 1:1 in DMSO:ddH2O where appropriate. To test the role of bd-cytochrome, heme was added in a final concentration of 10 μg/mL (diluted 1:1 in DMSO: ddH2O). The medium in the working electrode chamber was continuously mixed with a magnetic stir bar and N2 gas was purged to maintain anaerobic conditions for the course of the experiment. The applied potential to the working electrode was of +0.2 V versus Ag/AgCl (sat. KCl) (BASI, IN, USA). Reactors run under OC conditions were similarly assembled but kept at open circuit and used as control for non-current circulating conditions. Once the current stabilized, the electrochemical cells were inoculated to a final OD600 of 0.12–0.15 with the cell suspensions prepared in PBS. Current densities are reported as a function of the geometric surface area of the electrode (16 cm2). The bioreactors were sampled by taking samples under sterilized conditions at different time points for subsequent analysis. The samples for organic acids analyses were centrifuged (15,228 g, 7 min) and the supernatant was separated for High-Performance Liquid Chromatography (HPLC) assessments. Samples for ATP and NAD+/NADH analyses were flash frozen in a dry ice/ethanol bath.

Metabolite analysis

Request a detailed protocol

Organic acids, ethanol, and sugar concentrations were measured by HPLC (Agilent, 1260 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 1260 Infinity II RID. A five-point calibration curve based on peak area was generated and used to calculate concentrations in the unknown samples. The following standards were included in the HPLC measurements: acetate, formate, pyruvate, malate, lactate, succinate, oxalacetate, fumarate, ethanol, acetoin, butanediol, mannitol, and glucose. No gaseous products were measured.

BES biomass growth determination

Request a detailed protocol

Bioreactors were shaken to remove the cells attached to the working electrode and afterwards sampled to measure viable cells (colony forming units [CFUs]) and total biomass (dry weight). 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 (1:1000 to 1:1000000) in sterile PBS and plated on MRS for CFUs enumeration after overnight incubation at 30 °C. Dry weight was determined using a 25 mL sample collected at approximately maximum current density. Cells were harvested by centrifugation (5250 g, 12 min, 4 °C) and washed twice in 50 mL ddH2O. Afterwards cells were resuspended in 1 mL of ddH2O and transferred to microfuge tubes (previously weighted). Cells were harvested by centrifugation (5250 g, 12 min, 4 °C), and the tubes were then transferred to an evaporator to remove humidity. 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.

RNA-seq library construction and transcriptome analysis

Request a detailed protocol

L. plantarum NCIMB8826 was grown in triplicate to exponential phase (OD600 1.0) at 37 °C in mMRS with or without the supplementation of 20 μg/mL DHNA and 1.25 mM ferric ammonium citrate. Cells were collected by centrifugation at 10,000 g for 3 min at 4 °C, flash frozen in liquid N2 and stored at –80 °C prior to RNA extraction as previously described (Golomb et al., 2016). Briefly, frozen cell pellets were resuspended in cold acidic phenol:chloroform:isoamyl alcohol (pH 4.5) [125:24:1] (Invitrogen, Carlsbad, CA, USA) before transferring to 2 mL screw cap tubes containing buffer (200 mM NaCl, 20 mM EDTA), 20% SDS, and 300 mg 0.1 mm zirconia/silica beads. RNA was extracted by mechanical lysis with an MP Fastprep bead beater (MP Biomedicals, Santa Ana, CA, USA) at 6.5 m/s for 1 min. The tubes were centrifuged at 20,000 g at 4 °C for 3 min and the upper aqueous phase was transferred to a new tube. The aqueous phase was extracted twice with chloroform:isoamyl alcohol [24:1] (Fisher Scientific, Waltham, MA, USA), The aqueous phase was then transferred to a new tube for RNA ethanol precipitation (Green and Sambrook, 2020). RNA was then quantified on a Nanodrop 2000c (ThermoFisher), followed by double DNAse digestion with the Turbo DNA-free Kit (Invitrogen) according to the manufacturer’s protocols. The quality of the remaining RNA was checked using a Bioanalyzer RNA 6000 Nano Kit (Agilent Technologies, Santa Clara, CA, USA) (all RIN values > 9) and then quantified with the Qubit 2.0 RNA HS Assay (Life Technologies, Carlsbad, CA, USA). For reverse-transcription PCR (RT-PCR), 800 ng RNA was converted to cDNA with the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturer’s protocols. Quantitative RT-PCR was performed on a 7,500 Fast Real-Time PCR System (Applied Biosystems) using the PowerUp SYBR Green Master Mix (ThermoFisher) and RT-PCR primers listed in Supplementary file 5. The 2-ΔΔCt method was used for relative transcript quantification using rpoB as a control (Livak and Schmittgen, 2001).

For sequencing, ribosomal-RNA (rRNA) was depleted from 4 μg RNA using the RiboMinus Eukaryote Kit v2 with specific probes for prokaryotic rRNA (ThermoFisher) following the manufacturer’s instructions. RNA was then fragmented to approximately 200 bp, converted to cDNA, and barcoded using the NEBnext Ultra-directional RNA Library Kit for Illumina (New England Biolabs, Ipswitch, MA, USA) with NEBnext Multiplex Oligos for Illumina (Primer Set 1) (New England Biolabs) following the manufacturer’s protocols. cDNA libraries containing pooled barcoded samples was run across two lanes of a HiSeq400 (Illumina, San Diego, CA, USA) on two separate runs for 150 bp paired-end reads (http://dnatech.genomecenter.ucdavis.edu/). An average of 36,468,428 raw paired-end reads per sample was collected (Supplementary file 6). The DNA sequences were quality filtered for each of the 12 samples by first visualizing with FastQC (ver. 0.11.8) (Andrews, 2010) to check for appropriate trimming lengths, followed by quality filtering with Trimmomatic (ver. 0.39) (Bolger et al., 2014). Remaining reads then were aligned to the NCIMB8826 chromosome and plasmids using Bowtie2 (ver. 2.3.5) in the [-sensitive] mode (Langmead and Salzberg, 2012). The resulting ‘.sam’ files containing aligned reads from Bowtie2 were converted to ‘.bam’ files with Samtools (ver 1.9) (Li et al., 2009) before counting aligned reads with FeatureCounts in the [-- stranded = reverse] mode (ver. 1.6.4) (Liao et al., 2014). Reads aligning to noncoding sequences (e.g. rRNA, tRNA, trRNA, etc.) were excluded for subsequent analyses. Differential gene expression based on culture condition was determined with DESeq2 (Love et al., 2014) using the Wald test in the R-studio shiny app DEBrowser (ver 1.14.2) (Kucukural et al., 2019). Differential expression was considered significant with a False-discovery-rate (FDR)-adjusted p-value < 0.05 and a Log2 (fold-change) >0.5. Clusters of Orthologous Groups (COGs) were assigned to genes based on matches from the eggNOG (ver. 5.0) database (Huerta-Cepas et al., 2019).

Redox probe assays

Request a detailed protocol

Hamilton oxidation-reduction potential (ORP) probes (Hamilton Company, Reno, NV, USA) were inserted into air-tight Pyrex (Corning Inc, Corning, NY, USA) bottles containing mMRS supplemented with 20 μg/mL DHNA and/or 1.25 mM ferric ammonium citrate and incubated in a water bath at 37 °C. A custom cap for the Pyrex bottles was 3D printed with polylactic acid filament (2.85 mm diameter) such that the ORP probe threads into the cap and an o-ring seal can be used to provide an air-tight seal between the probe and the cap. The ORP was allowed to equilibrate over 40 min before L. plantarum NCIMB8826, Δndh2 (MLES100), or ΔpplA (MLES101) were inoculated at an OD600 of 0.10. Two uninoculated controls were used to measure baseline ORP over time. The ORP data was collected via Modbus TCP/IP protocol (Stride Modbus Gateway, AutomationDirect, Cumming, GA, USA) into a database (OSIsoft, San Leandro, CA, USA) and analyzed in MATLAB (Mathworks, Nantick, MA, USA). pH was measured using a Mettler Toledo SevenEasy pH meter (Mettler Toledo, Columbus, OH, USA). Cells were collected at either 24 hr or at the greatest ORP difference between the wild-type and mutant strains (ΔmVmax) by centrifugation at 10,000 g for 3 min and used for ferrihydrite reduction analyses.

ATP and NAD+/NADH quantification

Request a detailed protocol

Frozen cell pellets were suspended in PBS and lysed by mechanical agitation in a FastPrep 24 (MP Biomedicals) at 6.5 m/s for 1 min. The cell lysates were then centrifuged at 20,000 g for 3 min at 4 °C. ATP and NAD+ and NADH in the supernatants were then quantified with the Molecular Probes ATP Quantification Kit (ThermoFisher) and the Promega NAD/NADH-Glo Kit (Promega, Madison, WI, USA), respectively according to the manufacturers’ instructions.

Inductively coupled plasma-mass spectrometry (ICP-MS)

Request a detailed protocol

L. plantarum was inoculated in mMRS with or without 20 μg/mL DHNA and 1.25 mM ferric ammonium citrate at an OD600 of 0.10 for 3.5 hr. Cells were then collected by centrifugation at 10,000 x g for 3 min and washed twice in PBS to remove cell-surface-associated metals. Viable cell numbers were enumerated by plating serial dilutions on MRS laboratory culture medium and the resulting cell materials were digested by incubating at 95 °C for 45 min in a 60% concentrated trace metal grade HNO3, allowed to cool, then diluted with MilliQ water to a final concentration of 6% HNO3. The contents were quantified with internal standards with an Agilent 7,500Ce ICP-MS (Agilent Technologies, Palo Alto, CA) for simultaneous determination of select metals (Na, Mg, Al, K, Ca, Cu, Zn, Ba, Mn, Fe) at the UC Davis Interdisciplinary Center for Plasma Mass Spectrometry (http://icpms.ucdavis.edu/).

Kale juice fermentation assay

Request a detailed protocol

Green organic kale purchased from a market (Whole Foods) was washed with tap water and air dried for 1 hr as previously recommended (Kim, 2017). A total of 385 g of the leaves and stems were shredded with an electric food processor in 1 L ddH20. The kale juice was then diluted with 0.35 L ddH2O and autoclaved (121 °C, 15 min). The juice was then centrifuged under sterile conditions at 8000 rpm for 20 min and the supernatant was collected. A rifampicin-resistant variant of L. plantarum NCIMB8826-R (Tachon et al., 2014) (grown for 19 hr in MRS medium at 37 °C, 50 μg Rif/mL) was inoculated to an estimated final OD of approximately 0.05, and DHNA (20 μg/mL) was added where appropriate. Cells were collected and washed as previously described for the bioelectrochemical assays in mCDM. The anodic chambers of bioreactors assembled as previously described (anode of 4.3*6 cm) were filled with 125 mL of the inoculated kale juice and incubated at 30 °C purged with N2. After 1 hr, the anodes were polarized to 0.2 V versus Ag/AgCl (sat. KCl) (EET conditions) or kept at open circuit (OC, no EET). Viable cells were measured by plating 10-fold serial dilutions in MRS agar plates with 50 μg/mL of Rif.

Calculations

Request a detailed protocol

The total electrons harvested on the anode were estimated by integrating the area (charge) under the chronoamperometric curve (current response (A) over time (s)), which was corrected by subtracting the current baseline obtained before L. plantarum was added to the system. This obtained charge was then converted to mol of electrons using the Faraday constant (96,485.3 A*s/mol electrons).

Data accession numbers

Request a detailed protocol

L. plantarum RNA-seq data are available in the NCBI Sequence Read Archive (SRA) under BioProject accession no. PRJNA717240. A list of the completed Lactobacillales genomes used in the DNA sequence analysis is available in the Harvard Dataverse repository at https://doi.org/107910/DVN/IHKI0C.

Data availability

L. plantarum RNA-seq data are available in the NCBI Sequence Read Archive (SRA) under BioProject accession no. PRJNA717240. A list of the completed Lactobacillales genomes used in the DNA sequence analysis is available in the Harvard Dataverse repository at https://doi.org/10.7910/DVN/IHKI0C All other data generated or analysed during this study are included in the manuscript and supporting files.

The following data sets were generated
    1. Stevens E
    (2021) NCBI BioProject
    ID PRJNA717240. Lactiplantibacillus plantarum transcriptome under extracellular electron transfer (EET)-conducive conditions.

References

    1. McDonald TJ
    2. McDonald JS
    (1976)
    Streptococci isolated from bovine intramammary infections
    American Journal of Veterinary Research 37:377–381.
    1. Russell JB
    2. Cook GM.
    (1995)
    Energetics of bacterial growth: balance of anabolic and catabolic reactions
    Microbiol Rev 59:48–62.
  1. Book
    1. Vinderola G
    2. Ouwehand A
    3. Salminen S
    4. von Wright A
    (2019)
    Lactic Acid Bacteria: Microbiological and Functional Aspects
    CRC Press.
    1. Wickham H
    (2011) ggplot2
    Wiley Interdisciplinary Reviews 3:180–185.
    https://doi.org/10.1002/wics.147

Decision letter

  1. Gisela Storz
    Senior and Reviewing Editor; National Institute of Child Health and Human Development, United States

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Extracellular electron transfer increases fermentation in lactic acid bacteria via a hybrid metabolism" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Gisela Storz as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) The methods and results of metabolite analyses are not sufficiently described. Further explanation is required if conclusions are to be made about yield, fermentation, and physiology.

2) Some claims are overstated and need to be revised or, better yet, backed up with more experimental data.

Reviewer #1 (Recommendations for the authors):

1) The only weakness: The methods and results of metabolite analyses are not sufficiently described. This requires explanation if conclusions are to be made about yield, fermentation, and physiology.

The paper provides strong evidence that electron diversion to external acceptors is an electron sink. However, while quantitative conclusions are made about fermentation, only ~30% of the mannitol can be accounted for in the EET condition that is the focus of this paper. Put another way, the paper is about where electrons go, but the majority of the electrons are missing.

Clarify if this strain can excrete products such as sugar polymers, sugars, pyruvate, formate, glycerol, acetoin, or butanediol. State if standards for these were included in the product analysis (an RI detector can pick up nearly all of these). Explain why formate isn't present (was it not measured?), as this would allow determination of flux through PFL vs. a pyruvate excretion, a carbon sink of acetoin, or butanediol. Could missing carbon be due to excretion of a partially oxidized sugar like fructose (which would also show up on an RI detector), or accumulation of storage polysaccharide?

A) Address these questions in the section around line 299; where fermentation products are mentioned as being present or absent. What was looked for, but not found (lactate yes, succinate no), vs which were not looked for, (pyruvate? Formate? Acetoin, etc). This makes it clear that carbon is not missing, just that the methods were unable to detect it.

B) Table 1 and all other tables making calculations need significant qualifiers-- since NADH flux is guessed from mannitol disappearance (the assumption of 3 NADH, which would not be true if sugars, glycerol, or pyruvate were made), and NADH fates + ATP yields are based on a small subset of fermentation products detected. Statements like line 347 "EET allowed cells to produce more fermentation products per mol mannitol" can't be made. Nor can one conclusively say what % of electrons went to electrodes, nor what the YATP is. If anything, the fact that the mix of lactate, acetate and ethanol do not shift dramatically to acetate production (which is typically the result of any system with an NADH sink, such as interspecies hydrogen transfer), is surprising and indicates something unexplained is going on.

Again, the lack of a carbon balance takes nothing away from qualitative conclusion that mannitol electrons are diverted to EET, or the strength of the EET evidence. The electrons at the electrode or in the iron are real and countable, this just limits the quantitative statements that can be made.

2) L136; Reported Geobacter current densities easily reach 1000 µA/cm2 for flat surfaces, not 100-200.

3) Figure 3B: Many questions about this experiment. It is the only time it appears, yet is used as a quantitative way to compare mutants.

a) State if the iron(III) in the medium was reduced by mutants during this assay-for redox potential to drop equally suggests that it was (which contradicts 3A).

b) Provide the final pH during these experiments-- pH can affect E{degree sign}'.

c) Lack of DHNA is always used as a reliable negative control in the other experiments, provide any data that shows if the addition of DHNA is essential for the mV drop in this assay, so we know it is also an assay of EET.

Lacking better evidence/controls, this is one of the weaker bits of data in the main body, especially next to the more robust Fe/current data. Appropriate for supplements, but not conclusive as is.

4) An idea regarding the ∆pplA phenotype-the two conditions require EET, but iron oxides should be lower potential acceptors than the anode which is poised at a high driving force of >+400 mV. The anode is plenty high to oxidize something like DHNA (E{degree sign}' ~70 mV) or chelated metals. PplA could be more necessary near the low potential of flavins. The supplied data can't be used to compare the effect of specific riboflavin concentrations on iron reduction rates (only extents) vs current production rates, but it might be available.

5) Line 311, More data would be needed to use succinate production as evidence for or against respiration. A mol of pyruvate could be excreted as succinate due to TCA cycle oxidation (indicating respiration), manifesting as excess electrons relative to the amount of succinate produced (as additional pyruvate would have to be used to make OAA). Conversely, succinate could be due to electron disposal via a reductive pathway (indicating fermentation). Lacking good carbon or electron recovery, neither can be ruled out, though the non-respiratory reductive pathway is most logical.

6) Table 1/line 403-: The low YATP in cells using EET is a possible 'different energy conversion strategy'. (Side note- an energy spilling/Russell mechanism would be an 'ATP-dissipating', not -accumulating). I'd suggest uncoupling of acid production from growth by low pH seems likely. In most strains, simple glycolytic pathway functions at low external pH, but complex protein synthesis and replication processes become inhibited, ATP from substrate turnover is needed to maintain pmf and internal pH, and high internal organic acids become toxic. Citable yield/pH data can be hard to locate as it comes from the 80's in the age of calling every lactic acid bacterium a Streptococcus, but since the non-EET cultures at neutral pH got close to the Bauchop and Elsden YATP, and EET cultures at pH 4 were a fraction of this, these data look like common behaviour of many lactic acid bacteria. In future comparisons, pH control could help estimate energy yields from EET. The first paper I found in a quick search to illustrate pH effects was: Russell and Dombrowski, DOI: 10.1128/aem.39.3.604-610.1980.

7) The value for Geobacter ATP/substrate yield in Table S2 seemed very high. Mahadevan says Geobacter produces 0.5 mol ATP/mol acetate when using extracellular acceptors (3x higher when using internal acceptors like fumarate). If I understand the adjustment for 6-carbon compounds, would this make the value for Geobacter using EET to be 1.5 mol ATP/mol 6-C substrate?

Reviewer #2 (Recommendations for the authors):

Line 21 – acronyms like this make our field more challenging to understand and are unnecessary. Does the work presented actually support 'energy conservation'? Growth seems to occur after EET activity takes place rather than with it as discussed below.

Line 27 – 'fermentation yield' referring to biomass yield or fermentation product yield?

Line 29, Line 45, Line 75, Line 86, Line 479 – blending fermentation and respiration has been known for over a decade in Shewanella (see Hunt et al., 2010, JBact). The description on Line 517 regarding the major finding from this reference is incorrect. PMF is not used by S. oneidensis anaerobically to generate ATP as demonstrated by the robust anaerobic growth of an ATP synthase mutant.

Line 61 – implied here is that LAB are auxotrophic for heme and menaquinone, but should be explicitly stated.

Page 8 (top) – the comparison to other organisms is not particularly helpful unless current per unit cell or unit biomass is quantified, nor does it seem particularly relevant to the overall theme of this manuscript.

Line 152 – predicted to encode

Line 182 – predicted to encode

Line 183 – under the condition tested.

Line 200 – isn't this already known from the Listeria work?

Line 206 – unclear what 'extracellular redox potential' is, unclear to me if any differences are significant in Figure 3B. I guess this is a measurement of Fe(II) / Fe(III) in the medium?

Figure 3D – did the authors include a strain that lacked the cluster?

Line 252 – unclear to me how ICP-MS can distinguish between iron inside a cell and associated with the cell surface.

Line 256 – have other metals been tested? Frequently manganese substitutes for iron in enzymes from non-iron dependent bacteria. Manganese oxide is also a known substrate for EET in Geobacter and Shewanella and could be nutritionally significant to LAB.

Line 262 – does this mean a large amount of the cells that grew are now dead?

Figure 4 (and Figure 1) – I don't understand why maximum current output is completely decoupled from when biomass spikes. At 24h (maximum current), basically no growth has occurred. At 2 days and 5 days, when biomass is the greatest (in the WT culture), very little current is produced. This seems like direct evidence that EET is not coupled with metabolism in these bacteria. The ATP measurements were made at 24h, where OD of the mutant is decreasing (thus less ATP would be expected). Same issue with NAD/NADH measurement.

Figure 4C – Lines are undefined in the figure legend. Also, growth should be plotted on a semi-log scale, see: https://schaechter.asmblog.org/schaechter/2018/07/why-you-must-plot-your-growth-data-on-semi-log-graph-paper.html

Figure 5 – based on the growth curve shown in Figure 4, the OC culture lags behind and does not reach maximum OD until day 5. What do these measurements look like at day 5 (or 6)? It seems likely that the authors are drawing conclusions from cells in two different states (stationary phase vs exponential phase) rather than focusing on the finding that having an external electron acceptor appears to accelerate the metabolism of cells.

Line 465 – Description of the experiment setup in 2B does not include addition of flavins for the iron reduction assay, so it is unclear why the authors claim that both flavins and quinones are necessary for EET in diverse LAB species.

Line 473 – Iron and electrode reduction is dependent on addition of DHNA. Are these cells likely to ever encounter 20 ug/mL DHNA in the environment? Is the EET activity observed relevant to normal physiology for these organisms, or a lab artifact caused by providing a non-native electron carrier? The argument beginning on line 491 underscores the artificial nature to me.

Line 559 – none of the references here provide any quantitation quinone concentration in the digestive tract. (Relates to the comment above from line 473).

Line 637 – is ferrozine toxic to these cells?

Line 646 – Were these mutants complemented?

Line 653 – the suicide vector should be identified and referenced here.

General comments:

Unclear why Figure 1C shows twice as much current twice as fast as Figure 3C in what appears to be identical conditions?

Reviewer #3 (Recommendations for the authors):

Figure 1A – though the tubes are aligned with the plot below, it may be helpful for some readers to have labels on this image to clarify the conditions of each.

Figure 1C – for the top line, the "+DHNA+mannitol" label should be on one line

Figure 4A – labeling should be more consistent ("anodic" vs. "EET")

Figure 4A – "log" should be removed from the y-axis label

Figure 4C – points on this plot should be connected with straight lines

Figure 5D – axis label is cut off

Figure 5G – FLEET should not be mentioned in the model at all. Instead, Ndh2 should be labeled

Figure 6A – Label differences between the two kale juice bottles

Authors should also provide more information for the definitions of YATP and the values in Table 1

Line 137: Where the authors compare current density capacity for L. plantarum to that of S. oneidensis--are the quoted values for S. oneidensis using (an) endogenously produced shuttle(s), or are they for S. oneidensis + (an) added shuttle(s)? This should be clarified, because L. plantarum has no current density capacity without added DHNA.

Authors should offer a mechanism for the conditional pplA phenotype

Switching between the terms "electron donor" and "carbon source" may be confusing for some readers; one term should be used consistently.

Figure 3-—figure supplement 1F is not discussed in the text.

Line 204: typo in "pplA"

The ndh2 mutant should be included in the kale juice fermentation experiments to test the contribution of Ndh2 in this context.

Readers may not be familiar with the term "Crabtree-positive"

"SLP" should be defined on page 3

The introduction does not provide enough detail for a broad audience to understand the metabolic pathways being discussed. It would help if some of the detail about L. plantarum metabolism (such as lines 318-321, 553-555) was moved from the results/discussion to the introduction. If L. plantarum is able to grow via respiration (as suggested in lines 304-308), this should also be mentioned in the introduction. The introduction should also be revised to ensure that terms are used unambiguously and to remove inaccurate generalizations. Some examples are given in the comments below.

It is not totally accurate to say that energy per se is generated or created during metabolism. The authors should revise these statements to say that energy is obtained or conserved or that ATP is generated.

This paper discusses detailed metabolic pathways and the terms used to describe these pathways should be used carefully and precisely to avoid confusion for the reader. For instance, the authors should precisely define "substrate-level phosphorylation" and "fermentation" and should not use these terms interchangeably. "Substrate-level phosphorylation" refers to a reaction that generates ATP directly, without the use of the proton motive force and the membrane-associated ATP synthase enzyme. "Fermentation" refers to the pathways that restore the NAD+/NADH ratio and allow growth with substrate-level phosphorylation as the sole mechanism of ATP generation.

Lines 50-51 and 373-374: Some organic compounds can also be respired (for example fumarate).

Lines 51-52 should say "Fermentative bacteria produce ATP solely via substrate-level phosphorylation" because substrate-level phosphorylation also occurs when bacteria are growing by respiration (i.e., in glycolysis).

Lines 71-73: It is unclear how the property of being "exergonic" or "endergonic" is relevant in the context here.

Lines 85-86: "…EET is associated with non-fermentative respiratory organisms…" Does this refer specifically to endogenous EET (i.e., without an added mediator?). This should be clarified because there is a long history of research on fermentation+EET in the presence of added mediators (see next comment).

Related to the above point: In the introduction or discussion, the authors should mention or acknowledge other work examining how growth in the presence of an electrode affects fermentative metabolisms. Some examples are:

Glasser NR, Kern SE, Newman DK. Phenazine redox cycling enhances anaerobic survival in Pseudomonas aeruginosa by facilitating generation of ATP and a proton-motive force. Mol Microbiol. 2014 Apr;92(2):399-412. doi: 10.1111/mmi.12566.

Emde R, Schink B. Enhanced Propionate Formation by Propionibacterium freudenreichii subsp. freudenreichii in a Three-Electrode Amperometric Culture System. Appl Environ Microbiol. 1990 Sep;56(9):2771-6. doi: 10.1128/aem.56.9.2771-2776.1990.

And other references discussed in reviews such as:

Moscoviz R, Toledo-Alarcón J, Trably E, Bernet N. Electro-Fermentation: How To Drive Fermentation Using Electrochemical Systems. Trends Biotechnol. 2016 Nov;34(11):856-865. doi: 10.1016/j.tibtech.2016.04.009.

Vassilev I, Averesch NJH, Ledezma P, Kokko M. Anodic electro-fermentation: Empowering anaerobic production processes via anodic respiration. Biotechnol Adv. 2021 May-Jun;48:107728. doi: 10.1016/j.biotechadv.2021.107728.

Comparing Figure 1A to Figure 3A, the extent of Fe3+ reduction seems quite different. Were the conditions the same for these two experiments? If so, an explanation should be added to the text.

Lines 104-106: Can DHNA itself be used as an electron shuttle? This should be addressed.

Line 137: Where the authors compare current density capacity for L. plantarum to that of S. oneidensis--are the quoted values for S. oneidensis using (an) endogenously produced shuttle(s), or are they for S. oneidensis + (an) added shuttle(s)? This should be clarified, because L. plantarum has no current density capacity without added DHNA.

Line 139: "current production was not dependent on the carbon source or growth medium, and current increased after supplementation of riboflavin in the growth medium". Both of these claims are hard to evaluate because the plots in figure 1 and figure1-—figure supplement 2 are shown for different lengths of time, different carbon sources (electron donors), and different media. To make it easier for the reader to compare, it would help to have the name of the medium as a title for each current density plot. To judge whether riboflavin truly increases current, we should compare figure 1c to figure 1-—figure supplement 2D. In figure 1c, with no riboflavin added, current increases to ~120 uA/cm2 at around 1 day. In the supplemental figure, this increase is not seen until after riboflavin is added. Without a control, it is hard to know whether this increase was dependent on the riboflavin addition.

Line 154: Please define homofermentative.

Line 160: "the percentage of" should be removed

figure 2-—figure supplement 2C: I am confused about how this experiment controls for the possibility that reduction of soluble iron (i.e., the ferric ammonium citrate ), which wouldn't necessarily require EET, contributes to the higher levels of ferrous iron observed when DHNA and ferric ammonium citrate are both added. Maybe a control with no ferrihydrite added would tell us whether ferric ammonium citrate is being reduced.

Line 194: more background should be provided for the FLEET-inducing conditions

Lines 304-309: More background information is needed to understand the motivation behind these experiments. Were oxygen and nitrate available as electron acceptors?

Line 305: "…needed for PMF generation in aerobic…"

Line 414: "previous results"; does this refer to results in this paper or a prior publication? If a prior publication, it should be cited here.

https://doi.org/10.7554/eLife.70684.sa1

Author response

Essential revisions:

1) The methods and results of metabolite analyses are not sufficiently described. Further explanation is required if conclusions are to be made about yield, fermentation, and physiology.

We thank the Editor and Reviewers for raising these important concerns. We revised the Materials and methods and Results sections to more comprehensively describe the methods and results of the metabolite analyses. Within the Results, we included new data (Figure 5 and Supplement file 1) showing the levels of formate and pyruvate under extracellular electron transfer and open circuit conditions. We also clarify that we included standards for a number of metabolites, but that our measurements show that these compounds were absent or sufficiently low to not be within the range of detection.

Second, we also now provide a more complete picture which allows us to make quantitative conclusions about yield, fermentation, and physiology during EET in L. plantarum. Our new data on formate and pyruvate levels allowed us to track carbon and electrons at ~80% and ~85%, respectively. With this information, we show that EET increases the yield of ATP and fermentation products per mole of mannitol (Ymannitol, Yfermentation in Table 1) by ~1.75x. This result indicates that EET enables more efficient energy conservation and greater mannitol fermentation in L. plantarum. The better accounting of electrons reveals that L. plantarum uses an ~2:1 ratio of endogenous to exogeneous electron acceptors during EET. Interestingly, the yield of biomass per ATP and per mannitol (YATP, Ymannitol) decreases by ~1.5x, showing that while EET triggers faster, more efficient energy conservation, these catabolic processes are also less coupled to anabolic processes. Thus, the revised manuscript provides an improved, multifaceted view of this hybrid metabolism.

2) Some claims are overstated and need to be revised or, better yet, backed up with more experimental data.

We revised claims to more precisely reflect our findings and, for other claims, provided additional experimental data to further support them as follows:

(1) We provide new data on formate and pyruvate and the implications of those data on yield of ATP and fermentation products (discussed in our response to point 1 above).

(2) We provide new evidence that L. plantarum performs EET with physiologically-relevant (low) concentrations of DHNA (concern raised by Reviewers #2 and #3). DHNA is present at a range of concentrations (0.09-0.5 ug/mL) in foods and other environments inhabited by L. plantarum. Our new data (Figure 1—figure supplement 3) shows that L. plantarum supplied with 0.010 ug/mL DHNA, which is on the lower end of this range, still produces significant current and iron reduction. Additionally, we now show that this DHNA-dependent EET causes a significant decrease in pH (Figure 1—figure supplement 3), indicating EET affects the physiology of L. plantarum.

Lastly, while these data indicate that physiologically relevant levels of DHNA have physiological effects on L. plantarum in laboratory culture media, we also qualify our conclusions to explain that the extent and physiological impact of EET will depend on the environmental and nutritional conditions of the niche L. plantarum is inhabiting (lines 153-162).

(3) More background information is provided on the rationale and need for measuring oxidation-reduction potential (ORP) (concern raised by Reviewers #1 and 2) (lines 213-217). Appropriate statistical tests are also now applied to the ORP data. Lastly, new data are included (Figure 3B and Figure 3—figure supplement 1) showing that DHNA supplementation is required for robust ORP reduction by wild-type L. plantarum, further linking EET activity to its ability to modulate extracellular ORP during fermentation.

(4) Reviewer #3 suggested that the claim that EET proceeds independently of respiration may be overstated. We now explain our current understanding and the limits on that understanding. Our results show EET proceeds via a respiratory protein, but ATP generation during EET is mainly produced by substrate-level phosphorylation, not oxidative phosphorylation. We also acknowledge that EET could involve other as-yet unidentified respiratory components (lines 478-485 and 504-5) and that we cannot exclude ATP generation by oxidative phosphorylation (lines 482-485).

Reviewer #1 (Recommendations for the authors):

1) The only weakness: The methods and results of metabolite analyses are not sufficiently described. This requires explanation if conclusions are to be made about yield, fermentation, and physiology.

The paper provides strong evidence that electron diversion to external acceptors is an electron sink. However, while quantitative conclusions are made about fermentation, only ~30% of the mannitol can be accounted for in the EET condition that is the focus of this paper. Put another way, the paper is about where electrons go, but the majority of the electrons are missing.

Clarify if this strain can excrete products such as sugar polymers, sugars, pyruvate, formate, glycerol, acetoin, or butanediol. State if standards for these were included in the product analysis (an RI detector can pick up nearly all of these). Explain why formate isn't present (was it not measured?), as this would allow determination of flux through PFL vs. a pyruvate excretion, a carbon sink of acetoin, or butanediol. Could missing carbon be due to excretion of a partially oxidized sugar like fructose (which would also show up on an RI detector), or accumulation of storage polysaccharide?

We thank the Reviewer for the Reviewer’s strong endorsement of our manuscript and clear recommendations for improvements.

In the revised manuscript, we provide additional data showing where ~80% of the carbon goes under EET conditions (Supplement File 1) and where the majority of the carbon goes under open circuit conditions. With this more complete information, our balances account for between 77-96% of the electrons (see Supplement file 1). These new data more completely support our conclusion that EET increases fermentation product yield, metabolic flux, and environmental acidification.

A) Address these questions in the section around line 299; where fermentation products are mentioned as being present or absent. What was looked for, but not found (lactate yes, succinate no), vs which were not looked for, (pyruvate? Formate? Acetoin, etc). This makes it clear that carbon is not missing, just that the methods were unable to detect it.

We agree with the Reviewer on this recommendation to provide more clarity on the methods and results of the metabolite analyses. As a result, we have included additional information in the Materials and Methods (lines 712-720) and in the Results (lines 323-325 and 328-335). As described in detail below, we can account for ~80% of the carbon and electrons under EET conditions, which captures a majority of both the carbon and electron flow.

With our HPLC methods, we were able to detect the following compounds using standards: acetate, formate, pyruvate, malate, lactate, succinate, oxalacetate, fumarate, ethanol, acetoin, butanediol, mannitol, glucose and fructose (now indicated on lines 712-720 in Materials and Methods). Despite being extensively studied, L. plantarum have not been reported to make storage polysaccharides, and thus it is very unlikely that storage polysaccharides could account for the remaining missing carbon.

We detected large and significant changes in the concentration of main fermentation end-products, acetate, lactate and ethanol (Figure 5). We detected significant, but small changes, in pyruvate and formate production (now shown in Figure 5—figure supplement 2). We did not detect fructose, malate, glucose, oxaloacetate, fumarate, acetoin or 2,3-butanediol in the samples. We have included this new information in the revised Results section (lines 333-334). We also now include carbon in biomass as part of the carbon recovery in an updated Table 1 and updated Table S1. (We also revised other portions of Table S1 to correct a prior error.)

Taken together, these updated measurements increase the carbon recovery observed under both EET and OC conditions. Under EET conditions, ~80% of the carbon is recovered in fermentation products, while under OC conditions, we can account for ~55% of the carbon. We did not measure CO2 production because our bioelectrochemical reactors are not gas tight. Thus, we suggest the missing carbon is likely due to CO2 production, but we cannot rule out production of other unidentified metabolites.

B) Table 1 and all other tables making calculations need significant qualifiers-- since NADH flux is guessed from mannitol disappearance (the assumption of 3 NADH, which would not be true if sugars, glycerol, or pyruvate were made), and NADH fates + ATP yields are based on a small subset of fermentation products detected. Statements like line 347 "EET allowed cells to produce more fermentation products per mol mannitol" can't be made. Nor can one conclusively say what % of electrons went to electrodes, nor what the YATP is. If anything, the fact that the mix of lactate, acetate and ethanol do not shift dramatically to acetate production (which is typically the result of any system with an NADH sink, such as interspecies hydrogen transfer), is surprising and indicates something unexplained is going on.

We agree with the Reviewer that the qualitative observations are quite surprising and that our quantitative calculations need to carefully state what uncertainties remain. We also believe that our calculations need to be considered in the context of the prior, extensive literature on L. plantarum fermentation (McFeeters and Chen, 1996 – Food Microb; Teusink et al., 2006 – J Biol Chem):, (Teusink et al., 2009 – Plos Comp Biol).

Additional context to the changes made in the manuscript: our improved data (Table 1) now aligns with existing literature on L. plantarum, allowing us to make quantitative assessments of changes in ATP and NADH. Specifically, our findings now have carbon recoveries (~80%) within 10% of previous studies of L. plantarum metabolism (~90-105%, see references below). Additionally, our studies probe the same spectrum of metabolites found in these prior studies. with the sole exception of CO2 (CO2 cannot be measured using our existing bioelectrochemical reactors, since they are not gas-tight). Because CO2 production does not regenerate NADH or produce ATP, the lack of CO2 accounting will not affect ATP or NADH balances. Thus, the conclusions we draw about quantifying ATP generated by substrate-level phosphorylation aligns with other quantitative conclusions from prior work on L. plantarum metabolism (Dirar and Collins, 1972 – J Gen Microbiol).

In response to the Reviewer’s comment, we now use two different sets of assumptions to NADH generation and present the resulting values as a range of NADH regeneration. As described in Supplementary file 1, Method 1 uses the concentration of metabolites to calculate NADH production, essentially assuming that all NADH production stems from production of metabolites. Because it does not include NADH produced from mannitol consumption directed towards biosynthesis or other unidentified fermentation products, Method 1 likely underestimates total NADH produced, leading to higher values of NADH regeneration. Conversely, Method 2 estimates the NADH produced from mannitol consumption, essentially assuming that NADH is produced whenever mannitol is consumed. Since mannitol consumption may lead to biosynthesis or unidentified fermentation products without NADH generation, Method 2 likely overestimates the total NADH produced, leading to lower NADH regeneration values. In light of this uncertainty, we now use both Methods and present a range for NADH regeneration.

Again, the lack of a carbon balance takes nothing away from qualitative conclusion that mannitol electrons are diverted to EET, or the strength of the EET evidence. The electrons at the electrode or in the iron are real and countable, this just limits the quantitative statements that can be made.

We recognize that our measurements do not account for all the carbon, however, in the revised manuscript, these quantitative uncertainties are now limited to ~20% because of the much improved carbon and electron balance. To address the Reviewer’s concern, we have now indicated the limited extent of these uncertainties in the manuscript (lines 363-365).

2) L136; Reported Geobacter current densities easily reach 1000 µA/cm2 for flat surfaces, not 100-200.

Following Reviewer #2’s suggestion, we now compare the current production of different species normalized per g of protein biomass (line 140-145). This normalized value is a better metric to compare EET at a cellular level among different species.

3) Figure 3B: Many questions about this experiment. It is the only time it appears, yet is used as a quantitative way to compare mutants.

a) State if the iron(III) in the medium was reduced by mutants during this assay-for redox potential to drop equally suggests that it was (which contradicts 3A).

b) Provide the final pH during these experiments-- pH can affect E{degree sign}'.

c) Lack of DHNA is always used as a reliable negative control in the other experiments, provide any data that shows if the addition of DHNA is essential for the mV drop in this assay, so we know it is also an assay of EET.

Lacking better evidence/controls, this is one of the weaker bits of data in the main body, especially next to the more robust Fe/current data. Appropriate for supplements, but not conclusive as is.

We now provide background and justification for our use of ORP experiments (lines 213-217). Extracellular redox potential is also referred to as oxidation-reduction potential (ORP), which is the ratio of all oxidative to reductive components in the environment (Killeen et al., 2018 – Am. J. Enol. Vitic.). Redox reactions are prevalent in food fermentations (see Hansen, 2018 – Curr. Op. Food. Sci. for a comprehensive list) and ORP directly influences the outcome of fermentations with lactic acid bacteria (Brasca et al., 2007 – J. App. Micro., Olsen and Pérez-Díaz 2009 – J. Food Sci.).

A) Already after one hour incubation in mMRS there was a significant difference in ORP between the wild-type and ndh2 deletion mutant. This change was also observable for the pplA deletion mutant after three hours incubation. ORP differences between the wild-type and the two mutant strains persisted for the duration of the experiment. Additionally, we note that other factors besides EET will affect ORP. Although our new data show that DHNA, and therefore EET capacity, is a major driver of ORP reduction, L. plantarum and related bacteria (e.g. L. lactis) possess NADH oxidase and therefore are able to reduce O2, thereby also contributing to ORP reduction (Tachon et al., 2010 – App. and Env. Micro). Hence, although we expect that iron was reduced by the wild-type L. plantarum strain, it was not the only redox active compound. We revised the text (lines 219-228) and Figure 3B to clarify these distinctions.

B) pH values along with the experimental data collected in response to comment C (below), have been added to a new Figure 3 – supplement figure 1. The culture medium pH was identical between the wild-type and mutant strains. Therefore, it is very unlikely that pH alone accounts for the differences in the observed ORP values.

C) We appreciate the Reviewer’s suggestion of including a control group lacking DHNA. We performed that experiment and the results are included as Figure 3 – supplement figure 1. As expected, the absence of DHNA during L. plantarum growth led to significantly higher ORP values.

4) An idea regarding the ∆pplA phenotype-the two conditions require EET, but iron oxides should be lower potential acceptors than the anode which is poised at a high driving force of >+400 mV. The anode is plenty high to oxidize something like DHNA (E{degree sign}' ~70 mV) or chelated metals. PplA could be more necessary near the low potential of flavins. The supplied data can't be used to compare the effect of specific riboflavin concentrations on iron reduction rates (only extents) vs current production rates, but it might be available.

We agree with the Reviewer that L. plantarum likely uses different mechanisms and proteins for EET depending on the redox potential of the extracellular electron acceptor. Indeed, PplA uses FMN as its redox co-factor (E0 ~-150 mV and -350 mV), and so it may only reduce electron acceptors with lower potentials (for ferrihydrite E0 ~ -100 mV -+100 mV). We also believe that the mechanisms and proteins for EET depend on the prior growth conditions for L. plantarum before addition of the extracellular electron acceptor. In the revised manuscript, we provide these hypotheses and indicate that testing them is the subject of future work in our laboratories (lines 551-553).

5) Line 311, More data would be needed to use succinate production as evidence for or against respiration. A mol of pyruvate could be excreted as succinate due to TCA cycle oxidation (indicating respiration), manifesting as excess electrons relative to the amount of succinate produced (as additional pyruvate would have to be used to make OAA). Conversely, succinate could be due to electron disposal via a reductive pathway (indicating fermentation). Lacking good carbon or electron recovery, neither can be ruled out, though the non-respiratory reductive pathway is most logical.

While the Reviewer’s reasoning is sound, L. plantarum, like other lactic acid bacteria, only possesses the reductive branch of the TCA cycle; it does not possess an oxidative branch (Tsuji et al., 2013 – Enzyme Microb Technol). Thus, we can rule out the possibility of TCA cycle oxidation, leaving the reductive branch as the only way to produce succinate. In agreement with the prior literature, we detected only trace amounts of succinate amongst the many TCA cycle metabolites we looked for: succinate, fumarate, oxaloacetate, citrate, malate. Thus, we are confident that the metabolic flux through the reductive branch of the TCA cycle was marginal, if present. In the revised manuscript, we now explain that L. plantarum does not possess an oxidative branch of the TCA cycle and that our measurements were sensitive to, but did not detect, these other TCA cycle intermediates (lines 308-310 and 312-314).

6) Table 1/line 403-: The low YATP in cells using EET is a possible 'different energy conversion strategy'. (Side note- an energy spilling/Russell mechanism would be an 'ATP-dissipating', not -accumulating). I'd suggest uncoupling of acid production from growth by low pH seems likely. In most strains, simple glycolytic pathway functions at low external pH, but complex protein synthesis and replication processes become inhibited, ATP from substrate turnover is needed to maintain pmf and internal pH, and high internal organic acids become toxic. Citable yield/pH data can be hard to locate as it comes from the 80's in the age of calling every lactic acid bacterium a Streptococcus, but since the non-EET cultures at neutral pH got close to the Bauchop and Elsden YATP, and EET cultures at pH 4 were a fraction of this, these data look like common behaviour of many lactic acid bacteria. In future comparisons, pH control could help estimate energy yields from EET. The first paper I found in a quick search to illustrate pH effects was: Russell and Dombrowski, DOI: 10.1128/aem.39.3.604-610.1980.

We thank the Reviewer for these insights. We removed the energy-spilling reference from the manuscript. Since our data shows that final biomass is not affected by EET whereas energy conservation is, it seems likely that growth is not strongly coupled to energy conservation under EET conditions. We strongly agree with the Reviewer’s suggestion that the low pH under EET conditions may uncouple ATP generation and growth. We also hypothesize that this ATP produced from substrate-level phosphorylation might be used to create proton motive force to maintain intracellular pH and biosynthetic functions. We have included this hypothesis in the main manuscript (lines 406-409).

7) The value for Geobacter ATP/substrate yield in Table S2 seemed very high. Mahadevan says Geobacter produces 0.5 mol ATP/mol acetate when using extracellular acceptors (3x higher when using internal acceptors like fumarate). If I understand the adjustment for 6-carbon compounds, would this make the value for Geobacter using EET to be 1.5 mol ATP/mol 6-C substrate?

We thank the Reviewer for this correction. We incorrectly provided the thermodynamic boundary for the ATP/substrate yield from the referenced study. We now provide the value of 1.5 mol ATP/mol 6 carbon substrate (0.5 mol ATP/mol acetate) (Mahadevan et al., 2006 – Appl Environ Microbiol) in Supplement File 2.

Reviewer #2 (Recommendations for the authors):

Line 21 – acronyms like this make our field more challenging to understand and are unnecessary. Does the work presented actually support 'energy conservation'? Growth seems to occur after EET activity takes place rather than with it as discussed below.

We agree with the Reviewer that a broad scientific audience will find the FLEET acronym confusing and have eliminated it from the manuscript.

The Reviewer also questions whether our findings indicate that EET supports energy conservation in L. plantarum. In biochemistry, energy conservation is defined as the conversion of light or chemical energy into cellular energy in the form of ATP by an organism (Russell and Cook 1995 – Microbiol Rev). In rapidly dividing cells, energy conservation, a catabolic process, is associated with growth, an anabolic process. However, catabolism need not be coupled with anabolism, and energy conservation can still occur in cells that are not rapidly dividing, e.g. in resting cell suspensions (Russell and Cook 1995 – Microbiol Rev).

In L. plantarum, EET increases the rate at which chemical energy (in mannitol) is converted to cellular energy (in ATP) per cell by ~1.75-fold (Figure 5C), indicating that EET increases the rate of energy conversation per cell. EET increases the amount of cellular energy obtained per chemical energy (Ymannitol) (Table 1) by ~1.75-fold, indicating that EET is also a more efficient energy conservation strategy in L. plantarum.

To understand how the anabolic process of energy conservation is coupled to the catabolic process of growth, we can compare changes in ATP to changes in biomass. At the start of EET, increased ATP levels are accompanied by a significant shortening of lag phase (Figures 1C, 4C). Since lag phase is the period microorganisms use to accumulate ATP for exponential growth, this shortened period suggests catabolic and anabolic processes are well coupled at this stage. However, by stationary phase, the biomass yield (Table 1) decreased under EET conditions, suggesting a much poorer coupling of catabolic and anabolic processes.

Line 27 – 'fermentation yield' referring to biomass yield or fermentation product yield?

The text was revised to clarity that ‘fermentation product yield’ is being discussed here.

Line 29, Line 45, Line 75, Line 86, Line 479 – blending fermentation and respiration has been known for over a decade in Shewanella (see Hunt et al., 2010, JBact). The description on Line 517 regarding the major finding from this reference is incorrect. PMF is not used by S. oneidensis anaerobically to generate ATP as demonstrated by the robust anaerobic growth of an ATP synthase mutant.

First, our work was very much inspired by the observations in S. oneidensis and other work, so we apologize for the inadvertent omission of the Hunt et al. J. Bacteriology 2010 reference in the introduction and mis-stating of the ATP synthesis result. In the revised manuscript, we removed the reference to S. oneidensis as a well understood respiratory microorganism (line 45), have replaced reference 14 with the Hunt et al. J. Bacteriology 2010 (our original intention) (line 80), removed the word recent when describing these findings (line 85), and revised the discussion to correctly describe that fumarate reduction supports the creation of ATP via substrate level phosphorylation, not PMF in S. oneidensis (line 523).

Second, we do not claim that our work is the first observation of a hybrid metabolism that blends fermentation and respiration. Rather, this work quantitatively elucidates a different blending of fermentation and respiration that both contrasts with and expands upon previous work. As a counterpoint to the prior observation that substrate-level phosphorylation is the major energy conservation strategy during anaerobic fumarate reduction in a non-fermentative organism (S. oneidensis), our work demonstrates that a blending of fermentation and respiration occurs during reduction of solid electron acceptors in a primarily fermentative organism (L. plantarum) (lines 28-30, 525-527). Going beyond prior qualitative observations, we show that this hybrid metabolism leads to ~1.75x-more efficient and ~1.75x-faster energy conservation (Ymannitol, mannitol flux), but an overall ~1.5-fold weaker coupling between anabolism and catabolism (lower biomass yield). Additionally, we discover that this hybrid metabolism substantially affects the intracellular redox state and show redox change arises from a ~2:1 use of endogenous to extracellular electron acceptors. This fuller discussion is included in the manuscript (lines 525-535).

Line 61 – implied here is that LAB are auxotrophic for heme and menaquinone, but should be explicitly stated.

The text was revised to incorporate the Reviewer’s suggestion (lines 66-68).

Page 8 (top) – the comparison to other organisms is not particularly helpful unless current per unit cell or unit biomass is quantified, nor does it seem particularly relevant to the overall theme of this manuscript.

We agree with the Reviewer that current should be normalized per unit biomass, and we believe that this normalized value provides context for how significant EET is as a metabolic activity. In the revised manuscript, we normalize the current from L. plantarum per unit protein and compare this value to reported values for Geobacter sulfurreducens and Shewanella oneidensis MR-1 (lines 140-147).

Line 152 – predicted to encode

The text was revised to state the predicted protein encoded by pplA.

Line 182 – predicted to encode

The text was revised to state the predicted protein encoded by ndh2.

Line 183 – under the condition tested.

The text was revised to incorporate the Reviewer’s suggestion.

Line 200 – isn't this already known from the Listeria work?

Prior work in ListeriaListeria monocytogenes (Light et al., 2018) showed that iron reductase activity was associated with the presence of PplA and Ndh2. However, as a respiratory species, L. monocytogenes has a metabolic strategy that is distinct from lactic acid bacteria. Thus the finding that iron reduction is associated with PplA and Ndh2 in lactic acid bacteria is new. Additionally, our work shows for the first time in any bacterium that ndh2 and pplA are upregulated during growth in the presence of DHNA and iron.

Line 206 – unclear what 'extracellular redox potential' is, unclear to me if any differences are significant in Figure 3B. I guess this is a measurement of Fe(II) / Fe(III) in the medium?

Extracellular redox potential is also referred to as oxidation-reduction potential (ORP), which is the ratio of all oxidative to reductive components in the environment (Killeen et al., 2018 – Am. J. Enol. Vitic.). The text has been revised to provide a definition of ORP and this additional background to highlight its importance (lines 213-217).

Significant differences in ORP were found between the wild-type and mutant strains. Figure 3B was revised to now at which time points there ORP values differed between the wild-type and mutant L. plantarum strains.

The Reviewer is correct in that iron (II) / iron (III) complexes influence ORP, but there are many other drivers of ORP reduction in the MRS medium, such as LAB consumption of oxygen with NADH oxidase (Zotta et al., 2017 – J. App. Micro) (see also response to Reviewer #1). This important facet of the experiment is now described in the manuscript (lines 219-223).

Figure 3D – did the authors include a strain that lacked the cluster?

No, we did not probe a strain that lacked the FLEET locus because this experiment was designed to probe the importance of specifically PplA for current production. Rather, we investigated strains lacking the FLEET locus using our Fe3+ reduction assay since this experiment allows testing a larger variety of species at the same time. We believe that the Reviewer’s suggestion would be a useful expansion of our experiment and will include this in our future studies.

Line 252 – unclear to me how ICP-MS can distinguish between iron inside a cell and associated with the cell surface.

We appreciate the Reviewer’s question here. While ICP-MS cannot spatially resolve iron that is intracellular vs. extracellular, we removed surface-associated metals by washing the cells twice in PBS before performing ICP-MS. This washing procedure has been shown in other bacteria to remove loosely bound metals (see Kumar et al., 2020 – Nat. Sci. Reports and Arauz et al., 2008 – J. Haz. Materials). An ICP-MS section was added to the Materials and methods (lines 811-821) to clarify these points.

Line 256 – have other metals been tested? Frequently manganese substitutes for iron in enzymes from non-iron dependent bacteria. Manganese oxide is also a known substrate for EET in Geobacter and Shewanella and could be nutritionally significant to LAB.

For simplicity, we focused on intracellular iron concentrations within the paper, but indeed ICP-MS was also able to detect other metals in L. plantarum cells. We now include the other findings from the ICP-MS analyses in Figure 4—figure supplements 1 and 2. Although quantities of barium and calcium were significantly lower and higher in the ndh2 mutant compared to the wild-type strain, respectively, these differences were not observed for wild-type L. plantarum when grown in EET-simulating culture medium (mMRS with DHNA and iron supplementation) compared to the mMRS control medium. In addition, these two metals are not expected to be redox-active metals for EET.

Line 262 – does this mean a large amount of the cells that grew are now dead?

The Reviewer makes an interesting observation. Indeed, the combination of a 4-fold increase in dry weight and only 2-fold increase in viable cells may indicate that a fraction of the biomass was dead or not metabolically active enough to form a colony at the time point sampled. In support of this idea, our measurements of cell density by OD600 suggest that cells reach stationary phase shortly after reaching maximum current density. Additionally, shortly after maximum current density (after day 2, Figure 3C), the rates of mannitol consumption, lactate production, and ethanol production decrease (Figure 4). Thus, we hypothesize that at the point of maximum current density, when dry weight and CFU measurements were taken, the cells are entering stationary phase and a fraction of them are metabolically inactive or dead. In the revised manuscript, we point out that this discrepancy could be due to the increased acid stress on L. plantarum resulting from EET condition (lines 337-340).

Figure 4 (and Figure 1) – I don't understand why maximum current output is completely decoupled from when biomass spikes. At 24h (maximum current), basically no growth has occurred. At 2 days and 5 days, when biomass is the greatest (in the WT culture), very little current is produced. This seems like direct evidence that EET is not coupled with metabolism in these bacteria. The ATP measurements were made at 24h, where OD of the mutant is decreasing (thus less ATP would be expected). Same issue with NAD/NADH measurement.

We apologize for introducing confusion here, which stemmed from a lack of clarity about which current density and biochemical measurements were from different experiments. We have clarified in the Figure Legends that the same set of experiments yielded the current density plots in Figure 3 and the metabolite and pH measurements shown in Figure 5. We have also clarified that another set of experiments were used to generate the current density shown in Figure 1C and the ATP and NAD/NADH ratios shown in Figure 4.

Comparison of the data shown in Figure 1C and 4C show maximum current is closely followed by maximum OD. After one day, the current density has reached its maximum, but no significant increase in planktonic biomass was observed. Between day one and day two, lower but significant current is produced, the cell density reaches ~90% of its maximum, and the consumption of mannitol and production of lactate also increases dramatically. Overall, these observations show that current production is followed by increases in growth and metabolic flux. We hypothesize that the delay observed between current production and the boost in growth and metabolites could be the time cells need to accumulate ATP and compounds needed to enter exponential growth (now included on lines 274-278). In future experiments, we seek to more precisely relate current produced and growth by continuously monitoring of L. plantarum cell density under EET conditions.

Figure 4C – Lines are undefined in the figure legend. Also, growth should be plotted on a semi-log scale, see: https://schaechter.asmblog.org/schaechter/2018/07/why-you-must-plot-your-growth-data-on-semi-log-graph-paper.html

We have addressed the reviewer recommendation and edited Figure 4C accordingly.

Figure 5 – based on the growth curve shown in Figure 4, the OC culture lags behind and does not reach maximum OD until day 5. What do these measurements look like at day 5 (or 6)? It seems likely that the authors are drawing conclusions from cells in two different states (stationary phase vs exponential phase) rather than focusing on the finding that having an external electron acceptor appears to accelerate the metabolism of cells.

The Reviewer is correct that at the time of ATP and NAD/NADH measurements, the cells performing EET and cells under OC were at different growth stages. In Figure 4, we probed how EET in L. plantarum affected growth, ATP and intracellular redox state. To show this connection, we chose to measure ATP and NAD+/NADH when EET rate is maximum (maximum current density). These measurements show that energy metabolism is greatly accelerated and changed by EET: cells exhibit higher viability, biomass and ATP accumulation and maintain a higher NAD+/NADH ratio. We agree that performing those measurements at other time points would provide a more dynamic picture of the relationship between EET, energy conservation and redox balancing. Our laboratories will address these questions in future studies.

Line 465 – Description of the experiment setup in 2B does not include addition of flavins for the iron reduction assay, so it is unclear why the authors claim that both flavins and quinones are necessary for EET in diverse LAB species.

We included both flavins and quinones in this statement because we show that flavins are required for EET in Figure 1—figure supplement 1D and Figure 1—figure supplement 2D. Flavins are required as cofactors for Ndh2 (as FAD – see Heikal et al., 2014 – Mol Micro) and PplA (as FMN – see Light et al., 2018 – Nature). L. plantarum and several other LAB which we found to be EET-active are auxotrophic for riboflavin (Burgess et al., 2004 – Appl Environ Microbio) so riboflavin was already supplemented in the growth medium as the vitamin itself (in CDM) or within yeast extract (in MRS). In the iron reduction assay, cells are metabolically active, but not growing, and would have sufficient holo-Ndh2 and holo-PplA carried over from their growth medium to perform EET. This is supported by new data in Figure 1figure supplement 1E where additional riboflavin supplementation into mMRS did not increase subsequent iron reduction. As such, we decided to not include riboflavin in the iron reduction assay used for Figure 2B.

Line 473 – Iron and electrode reduction is dependent on addition of DHNA. Are these cells likely to ever encounter 20 ug/mL DHNA in the environment? Is the EET activity observed relevant to normal physiology for these organisms, or a lab artifact caused by providing a non-native electron carrier? The argument beginning on line 491 underscores the artificial nature to me.

The Reviewer is correct in stating that it is unlikely for L. plantarum to encounter 20 µg/mL DHNA in its native environment. DHNA is found in concentrations of 0.089-0.44 μg/mL in commercial fermented beverages, and under laboratory conditions, microbes can synthesize and secrete DHNA leading to concentrations of 0.37-48 μg/mL (lines 153-156). To test if EET in L. plantarum is relevant under these physiological concentrations, we probed whether L. plantarum can perform EET with a sub-physiological DHNA concentration of 0.01 μg/mL. Indeed, L. plantarum produced significant current density and a significant decrease in pH (new Figure 1—figure supplement 3). Far from being artificial, these results show that the concentrations of DHNA in environments known to support L. plantarum growth have significant effects on its physiology and mechanisms to outcompete neighboring organisms.

Line 559 – none of the references here provide any quantitation quinone concentration in the digestive tract. (Relates to the comment above from line 473).

As mentioned above, we have included in the manuscript the physiological levels of DHNA found in environments in which L. plantarum is known to reside. To our knowledge, no paper has quantified DHNA concentrations in the gastrointestinal tract. Other forms of quinones have been quantified in the GI tract. For instance, menaquinone concentrations of 5.5±2.7-14.54±0.29 μg/g were quantified in distal colonic contents (Conley and Stein, 1992 – Am J of Gastroenterol), and of 25.3-34.4 nmol/g were quantified from fecal samples (Karl et al., 2017 – Am J Clin Nut). Although we mainly focused on DHNA in this manuscript, it is reasonable to anticipate that L. plantarum can perform EET in the GI tract by responding to different forms of quinones due to their shared molecular structures. We will consider testing this hypothesis in our future studies.

Line 637 – is ferrozine toxic to these cells?

While not included in the data for this manuscript, we found that ferrozine does not inhibit L. plantarum growth. Therefore, there is no evidence that ferrozine is toxic to L. plantarum, nor do we expect that compound to impact the ferrihydrite reduction assay, which takes place over three hours and requires metabolically active, but not actively growing cells.

Line 646 – Were these mutants complemented?

Efforts for mutant complementation were made, but restoration of EET activity was variable, potentially because critical epigenetic regulatory conditions were not provided or because of the challenges of over-expression of membrane-bound proteins (Kang and Tullman-Ercek, 2018 – Methods). We are currently in the process of mutating all genes in the L. plantarum EET pathway and are including complementation analysis in order to identify the effects of gene proximity and synteny on EET activity.

Line 653 – the suicide vector should be identified and referenced here.

The text has been revised to specify the suicide vector along with a citation.

General comments:

Unclear why Figure 1C shows twice as much current twice as fast as Figure 3C in what appears to be identical conditions?

We observe somewhat variable current production depending on the exact conditions used for inoculation of bioreactors. Although we maintained identical culturing conditions for L. plantarum before introducing the cells into bioelectrochemical reactors, the inoculum did show mild growth differences. This can cause significant differences in the electrochemical output (current density). Thus, we suggest the differences in the current density production by the wild-type strain between Figure 1C and Figure 3C result from possible differences on the initial number and activity of cells. Future work will seek to uncover how initial conditions affect EET in L. plantarum.

Reviewer #3 (Recommendations for the authors):

Figure 1A – though the tubes are aligned with the plot below, it may be helpful for some readers to have labels on this image to clarify the conditions of each.

Numbered labels are now provided in Figure 1A to connect the image and the graph.

Figure 1C – for the top line, the "+DHNA+mannitol" label should be on one line

Figure 1C has been revised to include this suggestion.

Figure 4A – labeling should be more consistent ("anodic" vs. "EET")

We have revised the Figure accordingly and replaced anodic with EET.

Figure 4A – "log" should be removed from the y-axis label

We thank the reviewer for noticing this. We have corrected this typo.

Figure 4C – points on this plot should be connected with straight lines

We have addressed this suggestion accordingly, and we have also plotted Figure 4C in a semi-logarithmic scale following another reviewer’s suggestion.

Figure 5D – axis label is cut off

We thank the reviewer for noticing this and we have corrected this typo.

Figure 5G – FLEET should not be mentioned in the model at all. Instead, Ndh2 should be labeled

We have edited the Figure following the recommendations of the reviewer.

Figure 6A – Label differences between the two kale juice bottles

We apologize for omitting some of the details of the juice preparation in that plot. We have indicated the differences between the two juices in Figure 6D by adding details on the preparation steps.

Authors should also provide more information for the definitions of YATP and the values in Table 1

We have provided the definition of each yield at the bottom of Table 1.

Line 137: Where the authors compare current density capacity for L. plantarum to that of S. oneidensis—are the quoted values for S. oneidensis using (an) endogenously produced shuttle(s), or are they for S. oneidensis + (an) added shuttle(s)? This should be clarified, because L. plantarum has no current density capacity without added DHNA.

We agree with the reviewer that this point should be clarified. The new data we provide from the literature to compare L. plantarum vs S. oneidensis (mA/mg-protein) corresponds to a study in which flavins were self-secreted. We have also clarified that S. oneidensis and G. sulfurreducens do not require addition of riboflavin and quinones, unlike L. plantarum (lines 141-147). However, because L. plantarum possesses genetic elements involved in EET (i.e. ndh2) and performs EET with levels of DHNA similar to those found in its environment (additional data are provided to show this) L. plantarum does perform endogenous EET.

Authors should offer a mechanism for the conditional pplA phenotype

We briefly discussed the differences in the conditional need of PplA for EET across lactic acid bacteria and now have included an additional hypothesis for the conditional need of PplA in L. plantarum (lines 551-535). We have preliminary data that suggest there is a conditional need for PplA in L. plantarum to perform EET depending on the culturing conditions and will address this fully in future work.

Switching between the terms “electron donor” and “carbon source” may be confusing for some readers; one term should be used consistently.

We agree with the Reviewer, and we now appropriately use the term “electron donor” throughout the manuscript.

Figure 3—figure supplement 1F is not discussed in the text.

We apologize for this omission. We have included a brief discussion of this supplementary Figure (lines 355-359). This figure now is Figure 3 — figure supplement 2C.

Line 204: typo in “pplA”

The spelling error has been corrected.

The ndh2 mutant should be included in the kale juice fermentation experiments to test the contribution of Ndh2 in this context.

While this is a helpful suggestion, we did not use the Ndh2 mutant in our kale juice fermentation experiment since this experiment focused on the metabolic impact of EET during fermentation. However, we envision future experiments addressing the EET mechanism under ecological conditions will utilize this mutant.

Since we do not have this data, we have revised the Abstract to avoid concluding that ndh2 is necessary for EET in kale juice (line 27).

Readers may not be familiar with the term “Crabtree-positive”

We agree with the Reviewer and have removed this jargon from the Discussion.

“SLP” should be defined on page 3

We have replaced the acronym ‘SLP’ with the term ‘substrate-level phosphorylation’.

The introduction does not provide enough detail for a broad audience to understand the metabolic pathways being discussed. It would help if some of the detail about L. plantarum metabolism (such as lines 318-321, 553-555) was moved from the results/discussion to the introduction. If L. plantarum is able to grow via respiration (as suggested in lines 304-308), this should also be mentioned in the introduction. The introduction should also be revised to ensure that terms are used unambiguously and to remove inaccurate generalizations. Some examples are given in the comments below.

We appreciate the Reviewer’s suggestions regarding the introduction. Regarding the question of providing additional background on L. plantarum and LAB metabolism, the Introduction text (lines 55-71 and 97-104) has been revised to incorporate definitions of homofermentation and heterofermentation as well as stating that L. plantarum is primarily homofermentative but also a respiratory-capable LAB.

It is not totally accurate to say that energy per se is generated or created during metabolism. The authors should revise these statements to say that energy is obtained or conserved or that ATP is generated.

We agree with the Reviewer. We have carefully revised the manuscript to consistently use the term ‘energy conservation.’

This paper discusses detailed metabolic pathways and the terms used to describe these pathways should be used carefully and precisely to avoid confusion for the reader. For instance, the authors should precisely define “substrate-level phosphorylation” and “fermentation” and should not use these terms interchangeably. “Substrate-level phosphorylation” refers to a reaction that generates ATP directly, without the use of the proton motive force and the membrane-associated ATP synthase enzyme. “Fermentation” refers to the pathways that restore the NAD+/NADH ratio and allow growth with substrate-level phosphorylation as the sole mechanism of ATP generation.

Per the Reviewer’s helpful suggestion, the Introduction text has been revised to define substrate-level phosphorylation and fermentation (lines 50-54).

Lines 50-51 and 373-374: Some organic compounds can also be respired (for example fumarate).

The Reviewer is correct here. The introduction (lines 41-49) includes a broad definition of anaerobic respiration, which includes using both organic and inorganic compounds. We acknowledge that the original text (lines 373-374) may lead the reader to think that respiration is only associated with the use of inorganic electron acceptors. To avoid this confusion, we now state that another major difference in fermentation and anaerobic respiration is the use of the endogenous versus external organic electron acceptors (lines 377-378).

Lines 51-52 should say “Fermentative bacteria produce ATP solely via substrate-level phosphorylation” because substrate-level phosphorylation also occurs when bacteria are growing by respiration (i.e., in glycolysis).

We have revised this sentence (lines 50-52). Instead of using the word ‘solely’, we prefer to employ the word ‘mainly’ since ATP can also be generated via respiration in some primarily fermentative bacteria, like lactic acid bacteria, under specific conditions.

Lines 71-73: It is unclear how the property of being “exergonic” or “endergonic” is relevant in the context here.

We agree with the Reviewer that providing these details is not relevant and have removed them.

Lines 85-86: “…EET is associated with non-fermentative respiratory organisms…” Does this refer specifically to endogenous EET (i.e., without an added mediator?). This should be clarified because there is a long history of research on fermentation+EET in the presence of added mediators (see next comment).

The Reviewer makes a good point, and we revised the sentence to clarify that we are describing that endogenous EET is mainly associated with respiratory species.

Related to the above point: In the introduction or discussion, the authors should mention or acknowledge other work examining how growth in the presence of an electrode affects fermentative metabolisms. Some examples are:

Glasser NR, Kern SE, Newman DK. Phenazine redox cycling enhances anaerobic survival in Pseudomonas aeruginosa by facilitating generation of ATP and a proton-motive force. Mol Microbiol. 2014 Apr;92(2):399-412. Doi: 10.1111/mmi.12566.

Emde R, Schink B. Enhanced Propionate Formation by Propionibacterium freudenreichii subsp. Freudenreichii in a Three-Electrode Amperometric Culture System. Appl Environ Microbiol. 1990 Sep;56(9):2771-6. Doi: 10.1128/aem.56.9.2771-2776.1990.

And other references discussed in reviews such as:

Moscoviz R, Toledo-Alarcón J, Trably E, Bernet N. Electro-Fermentation: How To Drive Fermentation Using Electrochemical Systems. Trends Biotechnol. 2016 Nov;34(11):856-865. Doi: 10.1016/j.tibtech.2016.04.009.

Vassilev I, Averesch NJH, Ledezma P, Kokko M. Anodic electro-fermentation: Empowering anaerobic production processes via anodic respiration. Biotechnol Adv. 2021 May-Jun;48:107728. Doi: 10.1016/j.biotechadv.2021.107728.

In the original Discussion, we state that fermentation can be modulated using electrodes and cite the Moscovitz reference. However, we acknowledge this statement was brief and have expanded this discussion and included the Vassilev and Emde and Schink references in the revised manuscript (line 594 and lines 495-496). We also included the Glasser et al. reference (lines 93-96) as an example of a primarily respiratory organisms that also ferments performing EET.

Comparing Figure 1A to Figure 3A, the extent of Fe3+ reduction seems quite different. Were the conditions the same for these two experiments? If so, an explanation should be added to the text.

The conditions differed between Figures 1A and 3A. The iron reduction assay results shown in Figure 1A were obtained for L. plantarum after growth in mMRS. The results shown in Figure 3A were obtained after L. plantarum growth in mMRS supplemented with DHNA and ferric ammonium citrate. The inclusion of DHNA and ferric ammonium citrate in the mMRS growth medium was found to result in increased L. plantarum EET activity (data previously in Figure 2—figure supplement 2C). To clarify the rationale for the change in mMRS growth medium, we moved that supporting data to Figure 1—figure supplement 1 and revised the Results section to better describe these results and how they influenced our experimental methods (lines 123-133).

Lines 104-106: Can DHNA itself be used as an electron shuttle? This should be addressed.

Yes, DHNA can be used as an electron shuttle (Mevers, 2018 — eLife; Glasser NR et al., 2017 – Annu Rev Microbiol). DHNA can also be synthesized into menaquinone or demethylmenaquinone, which can transfer electrons in an electron transfer chain. However, our current data does not allow us to assess which of these roles DHNA is playing. Future work will elucidate the role of DHNA and flavins in EET in L. plantarum.

Line 137: Where the authors compare current density capacity for L. plantarum to that of S. oneidensis—are the quoted values for S. oneidensis using (an) endogenously produced shuttle(s), or are they for S. oneidensis + (an) added shuttle(s)? This should be clarified, because L. plantarum has no current density capacity without added DHNA.

We revised the comparison of EET among species to indicate that, unlike L. plantarum, Shewanella oneidensis and Geobacter sulfurreducens can synthesize riboflavin and quinones and do not require the addition of either for EET activity (lines 146-147).

Line 139: “current production was not dependent on the carbon source or growth medium, and current increased after supplementation of riboflavin in the growth medium”. Both of these claims are hard to evaluate because the plots in figure 1 and figure1—figure supplement 2 are shown for different lengths of time, different carbon sources (electron donors), and different media. To make it easier for the reader to compare, it would help to have the name of the medium as a title for each current density plot.

We acknowledge the lack of clarity here and have revised the text to more accurately convey our point (lines 148-152). Our intention is to qualitatively, rather than quantitatively, evaluate current production with different media. We have followed the Reviewer’s suggestion and added a title for each plot of Figure 1—figure supplement 2 to make it easier for the reader to understand the medium used.

To judge whether riboflavin truly increases current, we should compare figure 1c to figure 1—figure supplement 2D. In figure 1c, with no riboflavin added, current increases to ~120 uA/cm2 at around 1 day. In the supplemental figure, this increase is not seen until after riboflavin is added. Without a control, it is hard to know whether this increase was dependent on the riboflavin addition.

We agree with the Reviewer on this point. We provide additional data in Figure 1—figure supplement 2D that includes a control in which riboflavin was not added. This control shows that the increase in current is due to the addition of riboflavin.

Except for Figure 1—figure supplement 2D, all the experiments performed in bioelectrochemical reactors contain riboflavin in the media (as part of the vitamin stock solution in the CDM). This is now explained in the manuscript text (lines 148-152).

Line 154: Please define homofermentative.

The Introduction text has been revised to include definitions of both homofermentation and heterofermentation (lines 60-64).

Line 160: “the percentage of” should be removed

The legend text of Figure 2 has been revised to incorporate this suggestion.

Figure 2—figure supplement 2C: I am confused about how this experiment controls for the possibility that reduction of soluble iron (i.e., the ferric ammonium citrate ), which wouldn’t necessarily require EET, contributes to the higher levels of ferrous iron observed when DHNA and ferric ammonium citrate are both added. Maybe a control with no ferrihydrite added would tell us whether ferric ammonium citrate is being reduced.

Prior to performing the ferrihydrite reduction assay, the cells were washed twice (in PBS), thereby removing any ferric ammonium citrate that was present in the mMRS growth medium (line 744). Moreover, there was very little color change in the ferrihydrite reduction assay when wild-type L. plantarum was grown in mMRS containing ferric ammonium citrate but not DHNA (noting that DHNA was still included in the ferrihydrite assay medium) (Figure 2—figure supplement 2C). Therefore, our data show that ferric ammonium citrate is not contributing to the color formation in the ferrihydrite reduction assay. Additionally, we agree that it is notable how L. plantarum exposure to DNHA and ferric ammonium citrate during growth results in an improved capacity for that organism to reduce extracellular ferrihydrite and that the expression of ndh2 and pplA is induced under those conditions (lines 199-206). We are pursuing these questions in studies investigating the transcriptional control and downstream regulation of EET in L. plantarum.

Line 194: more background should be provided for the FLEET-inducing conditions

The Results section was revised to clarify and provide background on what was intended by the term “FLEET-inducing conditions” (lines 126-133). For clarity, we have also removed that term from the text and explained directly which culture conditions were used.

Lines 304-309: More background information is needed to understand the motivation behind these experiments. Were oxygen and nitrate available as electron acceptors?

To clarify further the motivation behind this experiment for the reader we have included a more detailed explanation (lines 295-301). Neither oxygen nor nitrate were used as electron acceptors, as their presence would be expected to compete with the anode as an electron acceptor.

Line 305: “…needed for PMF generation in aerobic…”

We have corrected this typo.

Line 414: “previous results”; does this refer to results in this paper or a prior publication? If a prior publication, it should be cited here.

The text has been revised to incorporate this suggestion.

https://doi.org/10.7554/eLife.70684.sa2

Article and author information

Author details

  1. Sara Tejedor-Sanz

    1. Department of BioSciences, Rice University, Houston, United States
    2. Biological Nanostructures Facility, The Molecular Foundry, Lawrence Berkeley National Laboratory, Berkeley, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review and editing
    Contributed equally with
    Eric T Stevens
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1192-8256
  2. Eric T Stevens

    Department of Food Science & Technology, University of California‐Davis, Davis, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review and editing
    Contributed equally with
    Sara Tejedor-Sanz
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0648-8281
  3. Siiliang Li

    Department of BioSciences, Rice University, Houston, United States
    Contribution
    Formal analysis, Investigation, Validation, Visualization, Writing – review and editing
    Competing interests
    No competing interests declared
  4. Peter Finnegan

    Department of Food Science & Technology, University of California‐Davis, Davis, United States
    Contribution
    Data curation, Investigation, Software, Writing – review and editing
    Competing interests
    No competing interests declared
  5. James Nelson

    Department of Electrical and Computer Engineering, University of California‐Davis, Davis, United States
    Contribution
    Data curation, Investigation, Writing – review and editing
    Competing interests
    No competing interests declared
  6. Andre Knoesen

    Department of Electrical and Computer Engineering, University of California‐Davis, Davis, United States
    Contribution
    Data curation, Methodology, Resources, Software, Supervision, Writing – review and editing
    Competing interests
    No competing interests declared
  7. Samuel H Light

    Department of Microbiology, University of Chicago, Chicago, United States
    Contribution
    Conceptualization, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8074-1348
  8. Caroline M Ajo-Franklin

    1. Department of BioSciences, Rice University, Houston, United States
    2. Biological Nanostructures Facility, The Molecular Foundry, Lawrence Berkeley National Laboratory, Berkeley, United States
    Contribution
    Conceptualization, Funding acquisition, Methodology, Supervision, Writing – original draft, Writing – review and editing
    For correspondence
    cajo-franklin@rice.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-8909-6712
  9. Maria L Marco

    Department of Food Science & Technology, University of California‐Davis, Davis, United States
    Contribution
    Conceptualization, Funding acquisition, Methodology, Supervision, Writing – original draft, Writing – review and editing
    For correspondence
    mmarco@ucdavis.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3643-9766

Funding

National Science Foundation (1650042)

  • Eric T Stevens

Office of Naval Research (0001418IP00037)

  • Caroline M Ajo-Franklin

U.S. Department of Energy (DE-AC02-05CH11231)

  • Sara Tejedor-Sanz
  • Eric T Stevens
  • Caroline M Ajo-Franklin
  • Maria L Marco

U.S. Department of Agriculture (W4122)

  • Maria L Marco

Cancer Prevention and Research Institute of Texas (RR190063)

  • Caroline M Ajo-Franklin
  • Siiliang Li

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by the National Science Foundation grant #1650042, Office of Naval Research grant 0001418IP00037 (CMAF), and the USDA National Institute of Agriculture Multi-State Project (W4122). Work at the Molecular Foundry was supported by the Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. James Nelson was supported by the Rodgers University fellowship in Electrical and Computer Engineering.

Senior and Reviewing Editor

  1. Gisela Storz, National Institute of Child Health and Human Development, United States

Publication history

  1. Preprint posted: May 26, 2021 (view preprint)
  2. Received: May 26, 2021
  3. Accepted: December 24, 2021
  4. Version of Record published: February 11, 2022 (version 1)
  5. Version of Record updated: February 14, 2022 (version 2)

Copyright

© 2022, Tejedor-Sanz et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 3,617
    Page views
  • 575
    Downloads
  • 5
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Sara Tejedor-Sanz
  2. Eric T Stevens
  3. Siiliang Li
  4. Peter Finnegan
  5. James Nelson
  6. Andre Knoesen
  7. Samuel H Light
  8. Caroline M Ajo-Franklin
  9. Maria L Marco
(2022)
Extracellular electron transfer increases fermentation in lactic acid bacteria via a hybrid metabolism
eLife 11:e70684.
https://doi.org/10.7554/eLife.70684
  1. Further reading

Further reading

    1. Biochemistry and Chemical Biology
    2. Chromosomes and Gene Expression
    Radhika A Varier, Theodora Sideri ... Folkert Jacobus van Werven
    Research Article

    N6-methyladenosine (m6A) RNA modification impacts mRNA fate primarily via reader proteins, which dictate processes in development, stress, and disease. Yet little is known about m6A function in Saccharomyces cerevisiae, which occurs solely during early meiosis. Here we perform a multifaceted analysis of the m6A reader protein Pho92/Mrb1. Cross-linking immunoprecipitation analysis reveals that Pho92 associates with the 3’end of meiotic mRNAs in both an m6A-dependent and independent manner. Within cells, Pho92 transitions from the nucleus to the cytoplasm, and associates with translating ribosomes. In the nucleus Pho92 associates with target loci through its interaction with transcriptional elongator Paf1C. Functionally, we show that Pho92 promotes and links protein synthesis to mRNA decay. As such, the Pho92-mediated m6A-mRNA decay is contingent on active translation and the CCR4-NOT complex. We propose that the m6A reader Pho92 is loaded co-transcriptionally to facilitate protein synthesis and subsequent decay of m6A modified transcripts, and thereby promotes meiosis.

    1. Biochemistry and Chemical Biology
    2. Cancer Biology
    Jacob M Winter, Heidi L Fresenius ... Jared Rutter
    Research Article

    The tumor suppressor gene PTEN is the second most commonly deleted gene in cancer. Such deletions often include portions of the chromosome 10q23 locus beyond the bounds of PTEN itself, which frequently disrupts adjacent genes. Coincidental loss of PTEN-adjacent genes might impose vulnerabilities that could either affect patient outcome basally or be exploited therapeutically. Here we describe how the loss of ATAD1, which is adjacent to and frequently co-deleted with PTEN, predisposes cancer cells to apoptosis triggered by proteasome dysfunction and correlates with improved survival in cancer patients. ATAD1 directly and specifically extracts the pro-apoptotic protein BIM from mitochondria to inactivate it. Cultured cells and mouse xenografts lacking ATAD1 are hypersensitive to clinically used proteasome inhibitors, which activate BIM and trigger apoptosis. This work furthers our understanding of mitochondrial protein homeostasis and could lead to new therapeutic options for the hundreds of thousands of cancer patients who have tumors with chromosome 10q23 deletion.