With pyruvate as an electron donor, D. reducens is capable of reducing both soluble (i.e., Fe(III)-citrate) and solid-phase (i.e., HFO) Fe(III); killed, lysed, and no-cell controls did not reduce Fe(III), nor could citrate alone support soluble Fe(III) reduction (Fig. S1). Figure 1A and B shows increasing concentrations of Fe(II) over time, as well as pyruvate consumption, buildup of acetate, H2 and small amounts of butyrate and formate from pyruvate oxidation, and an increase in cell counts during reduction of Fe(III)-citrate and of HFO, respectively. In the case of Fe(III)-citrate reduction, there is no evidence for the formation of a Fe(II) precipitate, while in the case of HFO – despite the lower rate and extent of the reduction – the solid phase shifted in color from orange–brown to black as the reduction proceeded, suggesting the accumulation of a Fe(II)-bearing solid phase. This was confirmed by SAED patterns and HR-TEM diffractograms, which identified the solid phase as magnetite (Fig. 2).
Figure 1. Evolution of D. reducens cultures in the presence of Fe(III)-citrate (A) or HFO (B) or in the absence of Fe(III) (C): concentrations of H2 (full triangles), pyruvate (full diamonds) and acetate (full squares) are to be read on the left axis; concentrations of Fe(II) (open squares), butyrate (open circles), formate (crosses) and cell count (full circles) are to be read on the right axis.
Download figure to PowerPoint
Figure 2. TEM image (A), selected area electron diffraction (SAED) patterns with the strongest reflections indicated in red (B) and high resolution TEM (HR-TEM) image showing the product of HFO reduction to be orthorhombic magnetite (Fe3O4) (C). The insets in the HR-TEM image correspond to the HR-TEM diffractogram (Fourier transform, FFT) and the Fourier filtered image (IFFT) from the selected particle.
Download figure to PowerPoint
During the reduction of both soluble and solid-phase Fe(III), pyruvate – the electron donor – was completely consumed within a few days. At the same time, the acetate concentration in solution built up to reach a plateau, and cell growth reached stationary phase. Conversely, Fe(II) accumulation did not follow the same time scale and continued even after the depletion of the electron donor [after 25 days, 8 mm Fe(III) citrate and 4.4 mm HFO were reduced to Fe(II) (Fig. S2)], suggesting that pyruvate oxidation and TEA reduction are temporally uncoupled (Fig. 1). These observations are consistent with the utilization of pyruvate as a substrate for fermentative growth, rather than as an electron donor for Fe(III) respiration. If fermentation is the dominant metabolism, is there any benefit to the cells from Fe(III) reduction?
We address this question by considering pyruvate consumption by a fermentation culture, in the absence of Fe(III) (Fig. 1C). Comparing Fig. 1A,B to Fig. 1C, we found that in the absence of Fe(III), pyruvate is initially consumed at approximately the same rate as in the presence of Fe(III), but after 1 day of incubation, the consumption rate decreases, resulting in the persistence of a significant amount of pyruvate (~8 mm) even after 10 days. In contrast, Fe(III)-containing cultures deplete pyruvate completely after 4–7 days (Fig 1A,B). Consistent with the smaller amount of consumed pyruvate, lower concentrations of organic acids accumulate in the absence of Fe(III). In contrast, cell counts are comparable in the presence or absence of Fe(III). Because the decrease in the rate of pyruvate consumption in the Fe(III)-free culture temporally coincides with the H2 concentration plateau, we surmised that the accumulation of significant (~4 mm, corresponding to a partial pressure (pH2) of ~104 Pa) H2 in the headspace may inhibit further fermentation of pyruvate. Fermentation inhibition due to high pH2 (in the order of 104 Pa, or less in some cases) has been previously reported for several micro-organisms: Clostridium acetobutylicum (Yerushalmi et al., 1985), Pyrococcus furiosus (Schäfer & Schönheit, 1991), Caldicellulosiruptor saccharolyticus (van Niel et al., 2003), and anaerobic consortia used for H2 production (Valdez-Vazquez et al., 2006). To support this hypothesis, we evaluated the effect of removing H2 by purging the headspace of a Fe(III)-free D. reducens fermentation culture after pyruvate consumption had stopped. Consistently with the hypothesis that H2 accumulation in the serum bottle inhibited further pyruvate fermentation, we observed a renewed buildup of the gas and resumption of pyruvate consumption (Fig. 3).
Figure 3. Pyruvate (crosses) concentration in a serum bottle containing a fermenting culture (in the absence of Fe(III)) and effect of the replacement of all H2 (diamonds) accumulated in the headspace by N2 (at 5 days): pyruvate consumption resumes and fresh H2 is released in the headspace.
Download figure to PowerPoint
Hence, our findings strongly indicate that when Fe(III) is present, it can be used as an electron sink alternative to protons, which allows the bacteria to divert electrons derived from pyruvate fermentation to the metal, thus overcoming the obstacle imposed by the strong backpressure of H2 accumulated in the headspace. The role played by Fe(III) during pyruvate fermentation by D. reducens is consistent with reports of bacteria using metal reduction as a means of eliminating excess reducing equivalents derived from electron donor oxidation (Lovley, 2006).
The temporal decoupling of electron donor oxidation and TEA reduction observed not only supports the conclusion that Fe(III) reduction by D. reducens is not a respiratory process, but it also suggests electron storage. In order for the electrons to be slowly released to Fe(III) even after electron donor depletion, they must be stored within a molecule acting as a capacitor. The concept that bacterial cells may oxidize an available electron donor, accumulating electrons for later release to an electron acceptor has been proposed previously for G. sulfurreducens and S. oneidensis (Esteve-Nunez et al., 2008; Schuetz et al., 2009; Bonanni et al., 2012). In contrast to Geobacter sp. and Shewanella sp., however, D. reducens is unlikely to use c-type cytochromes as electron capacitors, as they are not upregulated during Fe(III) reduction (as described in detail below). The capacitor could be a compound accumulated extracellularly such as H2 or formate. Other plausible candidates are intracellular molecules (e.g., storage polymers, ferrodoxin, NADH, quinones). Very low concentrations of formate were detected in the reduction medium (Fig. 1A,B), and it was found not to be a suitable electron donor for Fe(III) reduction by D. reducens (Fig. S1b), excluding it from playing a significant part in Fe(III) reduction. H2 accumulates significantly in the system; however, if it were the capacitor providing the electrons for Fe(III)-reduction, its concentration would be expected to decrease over time, which is not the case. Furthermore, we performed a separate experiment to test the effect of substituting the H2-rich headspace of an HFO-reducing, pyruvate-fermenting culture with sterile N2. We found that removing H2 did not affect the reducing activity of the culture, proving that electrons are not stored in H2 (Fig. S3) and that electron transfer from the cells to the TEA does not include an intermediate H2 step. Having excluded the most likely candidates for the role of capacitor, we propose that it must be an intracellular molecule. Microscopy observations show that during stationary phase, the cell morphology of D. reducens evolves from rod-shaped to round bodies (that are not spores) (Fig. S4), within which dark regions are visible. We hypothesize that these may be reduced carbon species, by analogy to other several sulfate-reducing bacteria, for example, Desulfovibrio spp, (Fareleira et al., 1997), Desulfobulbus propionicus (Stams et al., 1983), Desulfonema magnum (Hai et al., 2004), that are known to accumulate storage polymers (e.g., polyglucose (PG), polyhydroxybutyrate (PHB), glycogen) in the presence of excess energy or carbon source, when other growth conditions are restrictive (Preiss, 1984; Fareleira et al., 1997). PG was also shown to serve as a source of electrons for sulfate reduction in Desulfovibrio and Desulfobulbus spp. (Stams et al., 1983). Figure S5 shows that, indeed, the round bodies store electrons and are capable of transferring them to Fe(III)-citrate.
During pyruvate fermentation, about 8.1 mm of electrons are unaccounted for if one considers the consumption of pyruvate and the production of acetate, butyrate, formate, H2, and biomass. In the presence of Fe(III)-citrate and of HFO, about 8.5 and 9.2 mm of electrons are unaccounted for, respectively. Carbon balance calculations, presented in Table 1, indicate that there is excess carbon in all cases. However, only in one instance is there sufficient excess carbon to store all the unassigned electrons. Indeed, in the case of Fe(III)-citrate – containing the highest carbon excess – , there is 15.5 mm of C that could represent a storage compound. If we consider PHB as a possible compound, it can store 2e− per molecule – 7.75 mm e− in total, in our system – and thus comes close to representing a sink for all the unassigned electrons (corresponding to 8.5 mm). Thus, it is possible that reduced polymers (such as polyhydroxybutyrate, PHB) could be synthesized and could serve as a source of electrons for Fe(III) reduction. However, in the case of HFO and fermentation alone, there must be another (or several other) electron storage molecule(s). Potential candidates are ferredoxin, quinones, and NAD (although the latter is less likely to be involved as it plays no role in the pathway of pyruvate fermentation – see pathway in Fig. S6). For most of these compounds, iron reduction [E0′(Fe3+/Fe2+) ~ 0–0.4 V, depending on the species: E0′(Fe(III)-citrate/Fe2+) ~ 0.37 V, E0′(Ferrihydrite/Fe2+) ~ 0 V (Thamdrup, 2000), E0′(Fe(OH)3/Fe2+) = 0.118 V (Dolfing et al., 2007)] is thermodynamically favorable [E0′ (Fdox/Fdred) = −0.398 V, E0′ (Ubiquinone/Ubiquinol) = 0.113 V, E0′ (Menaquinoneox/Menaquinonered) = −0.074 V, E0′ (NAD/NADH) = −0.320 V) (Thauer et al., 1977)].
Table 1. Comparison of carbon and electron balances during pyruvate fermentation in the absence and presence of soluble or solid-phase Fe(III). This table shows that there is an abundance of electrons that are unaccounted for in all cases. Based on the C balance, only in the case of Fe(III) citrate reduction, there is sufficient excess carbon that could serve as a sink for these electrons
| ||Fermentation||Fermentation + Fe(III) citrate||Fermentation + HFO|
|Amount consumed or formed (mm)||e- donated/accepted (mm)||Carbon content (mm)||Amount consumed or formed (mm)||e- donated/accepted (mm)||Carbon content (mm)||Amount consumed or formed (mm)||e- donated/accepted (mm)||Carbon content (mm)|
|Acetate||8.0|| ||16.0||9.7|| ||19.3||11.2|| ||22.5|
|H2||4.0||7.9|| ||5.7||11.5|| ||4.8||9.6|| |
|CO2||7.4|| ||7.4||6.0|| ||6.0||6.5|| ||6.5|
|Fe(II)||–|| || ||6.7||6.7|| ||3.9||3.9|| |
|Biomassa|| || ||1.60|| || ||1.74|| || ||1.62|
|Balance|| ||8.1||0.40|| ||8.5||15.5|| ||9.2||2.8|
A reduced soluble compound is present in the spent medium of cultures fermenting pyruvate
AQDS was used as a TEA to test for the presence of soluble reduced compounds in spent HFO reduction medium. A sample from the HFO-growing culture was filter-sterilized at inoculation time (i.e., prior to the onset of HFO reduction), and another sample collected and filter-sterilized while HFO reduction was underway. Both samples were amended with 1 mm AQDS. After 48 h, measurement of residual AQDS revealed that 40% of the AQDS had been reduced to AH2DS in the medium collected when HFO reduction was ongoing, while none was reduced in the medium collected before cells started growing and reducing HFO. This result suggests that a reduced compound, potentially involved in Fe(III) reduction, is released into the medium by the cells during HFO reduction with pyruvate as an electron donor.
Filter-sterilized spent medium from a culture grown fermentatively in the absence of Fe(III) was also amended with AQDS. Consistently with the HFO spent medium, the fermentation medium reduced AQDS by about 30% in 3 days, with a total of 0.54 mm electrons transferred (0.27 mm of AQDS was reduced), indicating that the reduced soluble compound is released during pyruvate fermentation and not only in the presence of HFO.
Additionally, we aimed at probing whether the soluble compound is secreted in sufficient amount during pyruvate oxidation to account for all the Fe(III) reduction observed or whether it is replenished by the cells. To probe this question, we used cell-free fermentation spent medium and found that Fe(III) was reduced to a much lesser extent than AQDS: only 0.17 mm electrons were transferred without further reduction even after a month (Fig. S8). The limited efficacy of the fermentation spent medium, and presumably of the soluble compound, to reduce Fe(III) as compared to AQDS (Fig. S8) suggests that the electron shuttle is limited in quantity. Replenishment of the soluble compound by the cells could be achieved via redox cycling of the carrier or continuous release of its reduced form.
We probed the potential role of as an electron shuttle, because D. reducens is a sulfate-reducing bacterium, and trace amounts of this anion are present in the TE and YE. Sulfide produced by sulfate reduction could abiotically reduce Fe(III) and cycle back through sulfate reduction. If this were true, catalytic amounts of sulfate would be responsible for Fe(III) reduction. Because of the fastidiousness of MI-1 (see Fig. S9), we could not directly use WLP depleted of trace elements and yeast extract (likely to contain some sulfate), so instead we used, as an Fe(III)-reduction medium, cell-free spent fermentation medium stripped of S2− by precipitation as FeS. Fe(III)-citrate reduction tests by D. reducens were carried out in unmodified spent fermentation medium, spent fermentation medium stripped of S2−, and fresh WLP medium (Fig. S10). The reduction curves in the three cases overlap almost perfectly, indicating that removing the sulfide from the system does not impact Fe(III) reduction and thus that the small amount of present in the complete WLP medium does not play a significant role in Fe(III)-reduction by D. reducens.
As flavins have been reported to act as the electron shuttles for Fe(III) reduction carried out by S. oneidensis (von Canstein et al., 2008; Marsili et al., 2008), we performed a bulk spectrofluorometric measurement of cell-free spent fermentation medium to test for their presence. We found that after 24 h of growth, they are present at a concentration of approximately 170 nm. In order to verify that the fluorescing compound was indeed a flavin, and to identify its composition (whether RF, FMN or FAD), we analyzed the sample by HPLC. We found that a small FAD peak is present in the background (fresh WLP medium) but does not increase with cell growth. Conversely, FMN and RF, which are absent in fresh WLP medium, build up in the medium during cell growth (Fig. 6A), with RF representing 93% of the secreted flavin. To verify that flavins are present in the spent growth medium as a result of active release into the medium rather than cell lysis, we monitored their concentration over time by spectrofluorometry. We found that the flavin concentration curve follows the growth curve and that flavins are secreted by D. reducens during pyruvate fermentation regardless of the presence of Fe(III) in the system (Fig. 6B). Interestingly, the maximum concentration of flavins (i.e., 350 nm) measured in solution was found to be comparable with the amounts of extracellular flavins secreted by S. oneidensis MR-1 (i.e., 700 nm according to von Canstein et al., 2008, 250–500 nm according to Marsili et al., 2008) for solid-phase Fe(III) reduction.
Figure 6. (A) HPLC chromatograms of: a mix of FAD, FMN and RF standards (red curve), fresh WLP medium (blue curve), and filtered growth medium at 40 h of incubation (black curve). (B) Concentration of total flavins over time during fermentation in the presence (triangles) and absence (squares) of HFO, and growth curve for the culture without Fe(III) (diamonds). (C) Fe(II) accumulation due to HFO reduction by pyruvate fermentation cultures amended with 0 μm (diamonds), 1 μm (circles), and 10 μm (squares) of RF. (D) UV-VIS spectra of filtered spent growth medium from an HFO reducing culture spiked with 10 μm RF, measured anoxically (dashed curve) and after oxidation of the same sample (bold curve); the two arrows indicate the positions of the characteristic peaks of oxidized flavins (Macheroux, 1999). The appearance of diffuse peaks after oxidation clearly indicates the presence of flavins in the sample, present in their reduced form prior to oxidation.
Download figure to PowerPoint
Based on these observations, we suggest that flavins released from the cells as metabolites of pyruvate fermentation could act as electron carriers for Fe(III) reduction in D. reducens. To further validate this hypothesis, we compared the HFO reduction rate in the absence and in the presence of added RF (1 and 10 μm) and found that, indeed, spiking the culture with additional RF increases the rate of reduction (Fig. 6C). In addition, the UV–VIS spectrum of spent WLP medium during HFO reduction in the presence of 10 μm RF shows that the RF is reduced (Fig. 6D), proving that D. reducens is capable of reducing this compound.
Overall, we show, for the first time, that a Gram-positive bacterium is able to carry out solid-phase electron acceptor reduction through the intermediate of an endogenous soluble electron carrier, namely RF (and small amounts of FMN). Oxidized endogenous electron carrier is replenished by the cells by re-reduction and, during growth, de-novo secretion. Previous reports of solid-phase reduction by Gram-positive bacteria involving electron carriers only included exogenous molecules such as AQDS or an exudate from a Gram-negative bacterium (Milliken & May, 2007; Pham et al., 2008).
Investigation of the involvement of c-type cytochromes in Fe(III) reduction by D. reducens
The necessary involvement of cells for Fe(III)-reduction to proceed implies an enzymatic step that allows the transfer of electrons from the capacitor to the carrier and potentially directly to soluble Fe(III)-citrate. In all the Gram-negative bacteria whose mechanism of Fe(III) reduction has been investigated and described, including the cases that involve electron shuttles and capacitors (von Canstein et al., 2008; Esteve-Nunez et al., 2008; Marsili et al., 2008; Schuetz et al., 2009; Bonanni et al., 2012), multiheme c-type cytochromes play a seminal role in the process (Shi et al., 2007). Their key role was, in most cases, proven by deletion mutagenesis (Myers & Myers, 1997; Leang et al., 2003; Lloyd et al., 2003; Mehta et al., 2005; Bucking et al., 2010; Coursolle & Gralnick, 2012). We performed gene expression and proteomic experiments to probe the involvement of c-type cytochromes in Fe(III) reduction by D. reducens, although the lack of genetic system for this micro-organism did not allow genetic confirmation. Our targets were the only two genes present in the genome of strain MI-1 that encode for proteins containing binding domains for heme c (CXXCH): nrfH and nrfA (Junier et al., 2010). NrfH and NrfA are predicted to be tetraheme cytochromes. RT-PCR carried out with D. reducens-specific primers for nrfH and nrfA and showed that the target genes appear to be downregulated during Fe(III)-citrate reduction as compared to fermentation (Fig. 7A). A very weak band is visible for one of the Fe(III)-citrate reduction duplicate samples. Weak bands are also detectable for nrfA in the case of Fe(III) reduction. In contrast, the bands for both genes are significantly more pronounced under fermentation conditions, although the nrfH band appears slightly weaker. qRT-PCR was performed to confirm these results quantitatively (Table 2). The results obtained with this method were consistent with the visual results of the RT-PCR electrophoresis gel: the expression level of nrfA and nrfH relative to 16S rRNA () is approximately two orders of magnitude lower during Fe(III) reduction than during fermentation.
Table 2. qRT-PCR results: comparative expression of nrfA and nrfH relative to 16S rRNA under different conditions () and relative to fermentation ()
|Conditions|| nrfA || nrfH |
| || || || |
|Fermentation + Fe(III)||9.0E-05||7.6E-01||1.6E-05||6.8E-01|
|Fe(III)-citrate + pyruvate||2.8E-06||2.4E-02||5.4E-07||2.3E-02|
Figure 7. Expression of the two c-type cytochromes encoded in the genome of D. reducens, NrfA and NrfH. (A) Electrophoresis gel showing the product of RT-PCR for nrfA and nrfH: from left to right in duplicate lanes: pyruvate fermentation, Fe(III)-citrate reduction, fermenting culture amended with iron at mid exponential growth phase. (B) and (C) SDS-PAGE gels stained with Coomassie-blue or TMBZ, respectively, of total protein extract from cultures under two different conditions; the tested conditions, each in duplicate lanes, are: pyruvate fermentation amended with Fe(III)-citrate at mid exponential growth phase and pyruvate fermentation. The bands reacting with the heme-stain in the fermentation amended with Fe(III) sample do not correspond to the predicted sizes for NrfA or NrfH, 48 kDa and 17 kDa, respectively.
Download figure to PowerPoint
With the goal of further validating these expression data, which point to the lack of involvement of c-type cytochromes in Fe(III) reduction by D. reducens with pyruvate as an electron donor, we conducted SDS-PAGE with total protein extracts from cells fermenting pyruvate or reducing Fe(III)-citrate, which was added to a mid-exponential fermentation culture. Several bands reacting to the peroxidase-specific stain are visible for the fermentation culture amended with Fe(III) (Fig. 7C); however, the positions of these bands (~27–20 kDa) do not correspond to the predicted sizes for either NrfA or NrfH, 48 and 17 kDa, respectively. The lane corresponding to the fermentation sample exhibits weak staining at the upper edge of the gel, suggesting that a peroxidase-type protein may not have migrated very far into the gel. Because NrfA and NrfH are predicted to form a transmembrane complex, based on the presence of transmembrane helices in the amino acid sequence, we considered it possible that the peroxidase-staining and non-migrating band in the fermentation sample lane could be a membrane complex containing NrfA or NrfH, or both. Therefore, we excised this band, as well as the peroxidase-reacting bands in the fermentation amended with Fe(III) sample, and analyzed them by LC-MS/MS to search for peptides forming significant coverage of the sequence of NrfH or NrfA (Table S1). We were unable to find any peptides belonging to c-type cytochromes or to identify the protein(s) with known peroxidase activity in any of the bands.
We conclude that, in contrast to Gram-negative IRB, there is no evidence for the involvement of c-type cytochromes in Fe(III) reduction by D. reducens. Interestingly, no c-type cytochrome nor orthologs of nrfA or nrfH were identified in the draft genome sequence of Desulfotomaculum hydrothermale Lam5T, also reported to be capable of reducing Fe(III) with pyruvate as the electron donor (Haouari et al., 2008; Amin et al., 2013).