Abstract
Geobacter sulfurreducens utilizes extracellular electron acceptors such as Mn(IV), Fe(III), syntrophic partners, and electrodes that vary from +0.4 to −0.3 V vs. Standard Hydrogen Electrode (SHE), representing a potential energy span that should require a highly branched electron transfer chain. Here we describe CbcBA, a bc-type cytochrome essential near the thermodynamic limit of respiration when acetate is the electron donor. Mutants lacking cbcBA ceased Fe(III) reduction at −0.21 V vs. SHE, could not transfer electrons to electrodes between −0.21 and −0.28 V, and could not reduce the final 10% – 35% of Fe(III) minerals. As redox potential decreased during Fe(III) reduction, cbcBA was induced with the aid of the regulator BccR to become one of the most highly expressed genes in G. sulfurreducens. Growth yield (CFU/mM Fe(II)) was 112% of WT in ΔcbcBA, and deletion of cbcL (a different bc-cytochrome essential near −0.15 V) in ΔcbcBA increased yield to 220%. Together with ImcH, which is required at high redox potentials, CbcBA represents a third cytoplasmic membrane oxidoreductase in G. sulfurreducens. This expanding list shows how these important metal-reducing bacteria may constantly sense redox potential to adjust growth efficiency in changing environments.
Introduction
Life generates cellular energy by linking electron donor oxidation to acceptor reduction. Each electron source and sink has an inherent affinity for electrons, or redox potential, which defines the maximum amount of energy available in such coupled reactions. For example, the difference in midpoint potentials of NO3− and Fe(III) is more than half a volt, which is enough to generate an additional ATP per electron when acetate is the donor (E°’ of NO3−/NO2− = +0.43 V vs. E°’ of Fe(III) (oxyhydr)oxides/Fe(II) ∼ −0.2 V vs. Standard Hydrogen Electrode (SHE)) [1, 2, 3]. As the redox potential of soils and sediments can vary widely [4, 5, 6], adjusting electron transfer chains to use acceptors with more favorable potentials allows anaerobes to maximize growth in response to environmental conditions [3, 7, 8, 9].
The respiration of Fe(III) and Mn(IV) poses unique challenges. These elements exist as insoluble (oxyhydr)oxides near neutral pH, requiring diversion of electrons from inner membrane respiratory chains to electron-accepting surfaces outside the cell [10, 11]. Additional complexity arises from the number of metal oxide polymorphs that exist in nature, with nearly 30 Mn oxides and 15 Fe oxides described, each with their own characteristic redox potential [12, 13, 14, 15]. While all of these could appear to the cell as similar extracellular electron sinks, the higher redox potential of Mn(IV) compared to Fe(III) oxides (E°’∼ +0.5 to +0.3 V for Mn(IV) vs. +0.1 to −0.3 V for Fe(III) vs. SHE) predicts that bacteria should be able to recognize and prefer specific metal forms. Sequential reduction of Mn(IV) before Fe(III) was observed in sediments as early as 1966 [5] and in pure cultures of Geobacter metallireducens in 1988 [16], suggesting that biological mechanisms exist to differentiate between higher vs. lower potential materials outside the cell.
Geobacter spp. can reduce multiple oxidized metals [17, 18, 19, 20], directly transfer electrons to methanogens [21], and utilize electrode surfaces as electron acceptors [22]. The complex array of electron transfer proteins in Geobacter spp. may explain this flexibility, with multiple c-type cytochromes and extracellular appendages identified that facilitate reduction of extracellular compounds. In G. sulfurreducens, at least five triheme cytochromes are linked to periplasmic electron transfer [23, 24], five multi-protein cytochrome complexes aid electron transfer through the outer membrane [25], and both multiheme cytochrome nanowires and extracellular pili extend beyond the cell [26, 27, 28]. Some outer membrane cytochromes are necessary for reduction of specific oxyanions such as SeO32− [29], or use of Fe(III) vs. electrode surfaces [25, 30], but none explain how Geobacter might adapt its energy generation strategy to changes in redox potential.
The putative oxidoreductases ImcH and CbcL provide a possible mechanism for potential-dependent electron transfer [31, 32]. G. sulfurreducens requires the cytoplasmic membrane-localized seven-heme c-type cytochrome ImcH to respire extracellular acceptors above redox potentials of −0.1 V vs. SHE, and requires CbcL, a fusion of a diheme b-type cytochrome and a nine-heme c-type cytochrome, to use electron acceptors below −0.1 V vs. SHE. As imcH and cbcL are constitutively expressed [25, 32], the requirement for each appears to be controlled by ambient redox potential, somehow allowing cells to switch from ImcH-to CbcL-dependent electron transfer as conditions change [31, 32].
Multiple lines of evidence suggest ImcH and CbcL are not the only G. sulfurreducens oxidoreductases capable of routing electrons into the periplasm. The redox potentials of subsurface environments and microbial fuel cell anodes where Geobacter spp. typically dominate can be as low as −0.3 V vs. SHE, below the range where ImcH or CbcL are essential [33, 34]. Incubations of ΔcbcL with low-potential Fe(III) oxides such as goethite still produces Fe(II) [35], and ΔcbcL attached to electrodes still shows electron transfer below −0.2 V vs. SHE [32]. In addition, Geobacter genomes contain many uncharacterized gene clusters encoding a quinone oxidase-like b-type diheme cytochrome adjacent to a periplasmic multiheme c-type cytochrome, reminiscent of the two domains fused together in CbcL, and expression of some of these genes can be detected under metal-reducing conditions [36].
In this report, we identify CbcBA, a bc-type quinone oxidoreductase necessary for respiration near the thermodynamic limit of acetate oxidation. CbcBA is essential for extracellular metal and electrode reduction below −0.21 V vs. SHE, and is found within nearly every sequenced Geobacter genome [37]. We also provide evidence that use of CbcBA leads to lower growth yields, and may primarily act as a non-energy-conserving route for electron disposal. Unique from imcH and cbcL, cbcBA requires a σ54-dependent transcriptional activator for expression, and cbcBA is one of the most highly expressed genes during reduction of low potential Fe(III). Together, these cytochromes enable a branched electron transfer pathway that can operate at different redox potentials, allowing ImcH-dependent respiration when potential energy is plentiful, CbcL-dependent growth as energy becomes limiting, and use of CbcBA near the threshold able to support microbial life.
Materials and Methods
Bacterial strains and culture conditions
All strains and plasmids used in this study are listed in Table 1. G. sulfurreducens strains and mutants were grown in a minimal medium (referred to as NB) containing 0.38 g.L−1 KCl, 0.2 g.L−1 NH4Cl, 0.069 g.L−1 NaH2PO4.H2O, 0.04 g.L−1 CaCl2.2H2O, 0.2 g.L−1 MgSO4.7H2O, 10 mL of trace mineral mix, and buffered with 2 g.L−1 of NaHCO3 purged with N2:CO2 (80:20) atmosphere, incubated at 30 °C. Trace mineral mix was composed of 1.5 g.L−1 nitrilotriacetic acid as a chelator for growth, except when grown with Fe(III)-oxides, in which case minerals were prepared in 12.5 mL.L−1 of 7.7 M HCl to a final concentration of 0.1 M HCl, 0.1 g.L−1 MnCl2.4H2O, 0.5 g.L−1 FeSO4.7H2O, 0.17 g.L−1 CoCl2.6H2O, 0.10 g.L−1 ZnCl2, 0.03 g.L−1 CuSO4.5H2O, 0.005 g.L−1 AlK(SO4)2.12H2O, 0.005 g.L−1 H3BO3, 0.09 g.L−1 Na2MoO4, 0.05 g.L−1 NiCl2, 0.02 g.L−1 Na2WO4.2H2O, 0.10 g.L−1 Na2SeO4. Routine growth was performed in acetate-fumarate NB medium (NBFA) containing 20 mM acetate as the carbon source and electron donor and 40 mM fumarate as the electron acceptor. For solid medium, 1.5% agar was added to acetate-fumarate medium for growth on plates in an anaerobic workstation (Microbiology International, Maryland) under N2: CO2: H2 (75:20:5) atmosphere maintained at 30 °C. Every experiment was initiated by streaking fresh strains of G. sulfurreducens from −80 °C culture stocks. 200 µg.mL−1 kanamycin was used for G. sulfurreducens, 100 µg.mL−1 ampicillin and 50 µg.mL−1 kanamycin for Escherichia coli as indicated.
Strain construction and complementation
Deletion constructs were designed based on a strategy previously described [38]. Briefly, ∼1 kb upstream and downstream region of cbcBA (GSU0593-0594), and bccR (GSU0598) were amplified using primers listed in Table 2. Amplified upstream and downstream DNA fragments were fused using overlap extension PCR. Amplified fused DNA fragments were digested with restriction enzymes listed in Table 2, and ligated into digested and gel purified pk18mobsacB. The ligation product was transformed into UQ950 chemically competent cells. The resulting plasmid was sequence verified before transformation into S17-1 conjugation donor cells. Overnight grown S17-1 donor strain containing the plasmid was conjugated with G. sulfurreducens acceptor strain inside an anaerobic chamber on a sterile filter paper placed on an NBFA agar plate. After ∼4 h, cells scraped from filters were streaked on NBFA agar plates containing kanamycin. The positive integrants were streaked on NBFA + 10% sucrose plates to select for the wildtype or deletion genotype. Colonies from NBFA + 10% sucrose plates were patched on NBFA and NBFA + 200 µg.mL−1 to identify antibiotic sensitive, markerless deletion strains. The strains were verified by PCR for the gene deletion and final strains checked for off-site mutations via Illumina re-sequencing.
Complementation was performed using the method described in Hallberg et al. [39]. Complement strains were constructed by first cloning cbcBA (GSU0593-94), cbcB (GSU0593), or cbcA (GSU0594) gene into the pRK2Geo2 vector. The cbcBA cluster with native ribosomal binding sites was cloned under the control of its native promoter (GSU0597). The resulting vectors were sequence verified, then subcloned into pTn7c147 between the n7L and n7R regions. Newly subcloned pTn7 vectors were transformed in MFDpir chemically competent cells [40]. Any DNA between n7L and n7R regions is integrated downstream of the glmS (GSU0270) site, surrounded by strong terminators [41]. A helper plasmid pJMP1039 (a derivative of pTNS3) expressing recombinase TnsABCD in MFDpir cells was utilized to recognize n7L and n7R regions in pTn7 vectors [41], and integrate DNA onto G. sulfurreducens chromosome downstream of glmS. A triparental mating strategy was used to create complement strains.
Integrating genes onto the genome minimizes growth-rate and biofilm defects encountered when using most plasmids in G. sulfurreducens.
Cyclic voltammetry
Three-electrode bioreactors contained 3 cm2 1500-grit polished polycrystalline graphite working electrodes (POCO AXF-5Q, TriGemini LLC, Illinois), platinum wire counter electrodes, Ag/AgCl reference electrodes [42, 43], and were autoclaved at 121 °C for 20 min. Anoxic conditions were maintained by constantly flushing reactors with anoxic humidified N2: CO2 (80:20) gas. Acetate (40 mM) served as the electron donor and carbon source, and poised electrodes (+0.24 V vs. SHE) served as the electron acceptor. Acetate-fumarate grown cells (acceptor limited, OD600 ≃ 0.5) were inoculated at 25% v/v inoculation into 30 °C stirred reactors. A 16-channel potentiostat (Biologic Science Instruments, France) constantly recorded anodic current over time. Cyclic voltammetry was applied by forward scanning electrode potential from −0.55 V vs. SHE to +0.24 V vs. SHE, and reverse scanned back to −0.55 V vs. SHE at 1 mV/s for two scans [42].
Growth with Fe(III) citrate
Minimal medium containing 20 mM acetate and 55 mM Fe(III) citrate was used in anaerobic Balch tubes, or in bioreactors when redox potential was measured over time [35, 43]. Media were autoclaved at 121 °C and immediately removed to cool at room temperature in the dark. Anaerobic tubes containing Fe(III) citrate medium were inoculated at 1:100 v/v from stationary phase cultures (OD600 ≃ 0.5) grown in NBFA. 0.1 mL of sample was taken at regular intervals and dissolved in 0.9 mL of 0.5 N HCl. Fe(II) concentrations were measured using a ferrozine assay [44].
Redox potential measurement
For monitoring redox potential, bioreactors were used in Open Circuit Potential (OCP) mode. Short (1cm) electrochemically cleaned platinum wires were used as sensing electrodes with a Ag/AgCl reference (+0.21 V vs. SHE). Platinum was cleaned in 0.5 M H2SO4 by holding the working electrode at +2.24 V vs. SHE, cycling electrode potential between +0.01 V and +1.34 V for 20 cycles and stopping at +1.34 V vs. SHE.
Fe(III) oxide reduction
Medium containing 20 mM acetate and either ∼50 mM akaganeite or ∼30 mM hydrous ferric oxide was supplemented with 0.69 g.L−1 NaH2PO4.H2O (to prevent formation of crystalline Fe(III) (oxyhydr)oxide while autoclaving) [14]. Fresh akaganeite was synthesized by slowly adding 25% NaOH dropwise over the course of 1 h into a stirring solution of 0.4 M FeCl3 to pH 7. The suspension was aged for at least one hour at pH 7, then washed with DI H2O via centrifugation. 1 mL of freshly synthesized akaganeite (β-FeOOH) (∼0.5 M) was added to 9 mL medium with 20 mM acetate as the carbon source before autoclaving [14, 35]. Hydrous ferric oxide was synthesized first as schwertmannite (Fe8O8(OH)6(SO4).nH2O) by adding 5.5 mL of 30% hydrogen peroxide to a solution of 10 g.L−1 FeSO4, then stirred overnight to stabilize. Schwertmannite solids were washed with DI H2O thrice by centrifugation. The resulting mineral was added to medium with 20 mM acetate as the carbon source before autoclaving. Autoclaving at neutral pH transforms the schwertmannite into ferrihydrite with an amorphous XRD-signature [14, 35, 39]. Iron oxide medium was inoculated with 1:100 v/v of cells (OD600 ≃ 0.5) grown in NBFA medium. Samples (0.1 mL) were dissolved in 0.9 mL 0.5 N HCl, and stored in the dark before measurement via ferrozine assay.
Transcriptomic analysis using RNA-seq
Total RNA was extracted from G. sulfurreducens fumarate-grown cultures in exponential phase. For cells grown with Fe(III) citrate, RNA was extracted from cultures at exponential growth phase when ∼30% or ∼70% of Fe(III) citrate was reduced. Cells were collected using vacuum filtration to minimize inhibition from Fe(III)/Fe(II) in the medium. Electrode biofilms were scraped from electrodes immediately after disconnecting them from the potentiostat. All cell pellets were washed in RNAprotect reagent (Qiagen) and stored at −80 °C before extraction using RNeasy with on column DNase treatment (Qiagen). Ribosomal RNA was depleted using Ribozero (Illumina) before sequencing on the Illumina Hiseq 2500 platform in 125-bp pair-ended mode. Residual ribosomal RNA sequences were removed using Bowtie2 [45] before analysis. Duplicate rRNA-depleted biological samples were analyzed for each strain and condition using Rockhopper [46], with our re-sequenced G. sulfurreducens genome as reference [38]. Expression was normalized by reads mapped by the upper quartile of gene expression values, and full RNA-seq data are in Supplementary table 1.
CFU and yield measurements
Growth of G. sulfurreducens strains was measured by counting colony-forming units (CFUs). A drop plate method adapted from Herigstad et al. [47], was used to count cells on NBFA agar medium. Briefly, 100 μL of samples were serially diluted 1:10 in liquid medium, and 10 µL of each dilution was plated on NBFA agar plates inside an anaerobic chamber (Coy laboratory products, Michigan) with an N2: CO2: H2 (75:20:5) atmosphere. Total Fe(III) reduced was measured using a ferrozine assay, so cellular yield could be calculated as CFU per mM Fe(III) reduced as cells were actively growing.
Results
The cbcBA gene cluster encodes a b-and c-type cytochrome expressed late in Fe(III) reduction
The G. sulfurreducens genome contains at least six putative inner membrane quinone oxidoreductase gene clusters. Five encode both b- and c-type cytochrome domains: Cbc1 (GSU0274, cbcL), Cbc3 (GSU1648-GSU1650, cbcVWX), Cbc4 (GSU0068-GSU0070, cbcSTU), Cbc5 (GSU0590-GSU0594, cbcEDCBA), Cbc6 (GSU2930-GSU2935, cbcMNOPQR) [48], and one contains only a multiheme c-type cytochrome (GSU3259, imcH) [31]. The b- and c-type cytochrome CbcL (Cbc1) is essential for growth below redox potentials of about −0.1 V vs. SHE [32], while the c-type cytochrome ImcH is essential for respiration as redox potential rises above this point [31]. Among these b- and c-type cytochrome gene clusters, Cbc5 is the most conserved cytochrome-containing gene cluster among Geobacter species [37].
Bioinformatic [49, 50, 51] and transcriptomic analyses [25, 52] place cbcBA in an operon with a σ54-dependent promoter upstream of GSU0597 and a transcriptional terminator downstream of cbcB (Figure 1A). This operon encodes two hypothetical proteins (GSU0597 and GSU3489), a RpoN-dependent response regulator (GSU0596), a quinone oxidoreductase-like di-heme b-type cytochrome (CbcB) [53], and a seven-heme c-type cytochrome (CbcA) (Figure 1C). An inner membrane localization of CbcBA is predicted by PSORT [54], with CbcB integrated into the inner membrane and CbcA exposed in the periplasm anchored by a C-terminal transmembrane domain. Cell fractionation studies also report a cytoplasmic membrane association of CbcA [55], implying that CbcBA is located at the inner membrane.
Divergently transcribed from this operon is GSU0598, a putative σ54-dependent transcriptional regulator, which we have named bccR (for bc-type cytochrome regulator) (Figure 1A). BccR belongs to the RpoN-dependent family of regulators that bind −12 /−24 elements [56]. BccR contains a response receiver domain, a σ54 factor interaction domain, and a C-terminal helix-turn-helix domain [57] (Figure 1C).
The cbcBA operon (GSU0597-GSU0593) had near zero expression when fumarate was the electron acceptor, but low expression was detected in electrode-grown biofilms [25] (Figure 1B). When growing with Fe(III) citrate as the electron acceptor, expression of the cbcBA operon remained low during the first 20 h of growth (Figure 1B), or as the first ∼30% Fe(III) was reduced (Figure 2A). However, cbcBA was dramatically upregulated after 30 h of growth (Figure 1B), as ∼70% of Fe(III) became reduced (Figure 2A). The level of cbcBA expression (>12 000 RPKM) was higher than 99% of G. sulfurreducens genes at this stage (SI figure 1).
CbcBA is essential for complete reduction of Fe(III) citrate
To determine if CbcBA was involved in extracellular electron transfer, a markerless deletion of GSU0593-94 (ΔcbcBA) was created. The ΔcbcBA mutant did not show any defect with fumarate as the electron acceptor (SI figure 2). However, the extent of Fe(III) reduction by ΔcbcBA was lower. Mutants lacking cbcBA never reduced the final 8-10% of Fe(III) citrate (Figure 2A), regardless of the amount of electron donor provided or length of incubation.
The putative quinone oxidoreductase ImcH is essential for reduction of high potential electron acceptors such as freshly prepared Fe(III) citrate [31], while the bc-cytochrome CbcL becomes essential as Fe(III) is reduced and redox potential drops [32, 35]. As the type of Fe(III) reduction defect observed for ΔcbcBA was similar to ΔcbcL, mutants lacking cbcL and cbcBA were directly compared. The ΔcbcBA strain ceased reduction of Fe(III) after 92.7 ± 1.4% (n=10) of Fe(III) citrate was reduced, whereas ΔcbcL only reduced 84.6 ± 1.0% (n=11) of Fe(III) citrate (Figure 2A, 2B). This suggested that CbcBA became necessary in the final stages of Fe(III) reduction.
CbcBA is required for reduction of Fe(III) citrate below −0.21 V vs. SHE
Because the >3000-fold up-regulation of cbcBA occurred after more than half of Fe(III) citrate was reduced (SI figure 1), induction of cbcBA did not appear to be due to the presence of Fe(III) per se. To more accurately determine the energy available during each stage of Fe(III) reduction, we measured redox potential continuously during growth with a potentiostat [35, 58]. Redox potential titrations and voltammetry determined the midpoint potential of the Fe(II) citrate/Fe(III) citrate half-reaction in our medium to be −0.043 V vs. SHE (SI figure 3). This is lower than values calculated in literature, likely due to high levels of chelating carboxylic acids in commercial Fe(III) citrate combined with electron donors, creating bi- or tri-dentate complexes with lower redox potential than the 1:1 ratios assumed in standard calculations [59, 60, 61].
When wildtype (WT) cells were inoculated into freshly prepared Fe(III) citrate (>99% oxidized), redox potential dropped rapidly from +0.15 V, and stabilized days later at −0.27 V vs. SHE when nearly 100% of Fe(III) was reduced. Considering the formal redox potential of CO2/acetate is −0.28 V, cells utilized nearly all the free energy available. In contrast, ΔcbcL ceased Fe(III) reduction near −0.15 V vs. SHE (equivalent to 38 mM Fe(III) reduced) [35]. Under the same conditions, ΔcbcBA stabilized at −0.21 V vs. SHE (equivalent to ∼46 mM Fe(III) reduced). Each mutant produced these same endpoint potentials independent of inoculation size or incubation time (SI figure 4), or when the concentration of Fe(III) citrate was increased to 80 mM [35].
Complementation of ΔcbcBA requires both cbcB and cbcA
To test if cbcB or cbcA alone were responsible for this inability to reduce Fe(III) below −0.21 V vs. SHE, single genes were integrated into the chromosome under control of the cbcBA operon’s promoter [39]. When ΔcbcBA::cbcB or ΔcbcBA::cbcA strains were grown with Fe(III) citrate, reduction still ceased at the same extent and redox potential (Figure 3A).
However, when both cbcBA genes were integrated and expressed in the ΔcbcBA strain, the extent of Fe(III) reduction was restored to WT levels (Figure 3A). Based on these results, all subsequent experiments were conducted with mutants lacking both genes. BccR is necessary for expression of cbcBA. A response regulator is divergently transcribed upstream of cbcBA bc-type cytochrome operons in all examined Geobacter genomes [25, 62]. When bccR (GSU0598) was deleted, ΔbccR ceased reduction of Fe(III) at −0.21 V vs. SHE, the same potential as ΔcbcBA (Figure 3B). RNAseq revealed that expression of cbcBA was no longer upregulated in ΔbccR during Fe(III) citrate reduction (Figure 3C) consistent with BccR being an activator of the cbcBA operon. Deletion of bccR did not affect other putative quinone oxidoreductases, in particular imcH or cbcL, which were constitutively expressed at much more moderate (∼500 RPKM) levels.
While the largest effect of bccR deletion was downregulation of cbcBA operon (Figure 3C), ΔbccR showed upregulation of hgtR (GSU3364) when Fe(III) was the electron acceptor. HgtR is a RpoN-dependent repressor involved in downregulating acetate oxidation when hydrogen is the electron donor [56, 63]. The increase in hgtR expression by more than 1 000x in ΔbccR implies a possible role for HgtR in down-regulating the TCA cycle during reduction of Fe(III) as acetate oxidation becomes thermodynamically limited.
Double mutants show that imcH, cbcL, and cbcBA are required within different redox potential windows
If one inner membrane cytochrome is needed in order to lower redox potential enough to activate the next, then double and triple markerless deletion mutant strains should still show the phenotype of their dominant missing pathway. All single, double, and triple mutant strains lacking imcH failed to initiate Fe(III) citrate reduction when inoculated into fresh >+0.1 V vs. SHE medium, and did not lower the redox potential more than 20 mV over the following 60 h (Figure 4). The dominance of ΔimcH in all backgrounds corroborates data showing ImcH to be essential for electron transfer in fresh Fe(III) citrate, Mn(IV) oxide, and electrodes at redox potentials above 0 V [31, 35], and showed that the presence or absence of cbcBA did not alter this behavior.
Like the single ΔcbcL mutant, the ΔcbcL ΔcbcBA double mutant containing imcH initially reduced Fe(III), then ceased reduction at −0.15 V vs. SHE. This provides additional evidence that ImcH can function down to a redox potential of −0.15 V, and that only CbcL can lower redox potential beyond this point, regardless of whether CbcBA is present (Figure 4). The phenotype of ΔcbcBA (Figure 2, 3) similarly implies that CbcL is essential until −0.21 V vs. SHE, at which point CbcBA is required for electron transfer (Figure 3).
Cyclic voltammetry detects a CbcBA-dependent electron transfer process with a midpoint potential of −0.24 V vs. SHE
All evidence up to this point that cbcBA was required at specific redox potentials was derived from soluble Fe(III) incubations, which could be non-physiological compared to environments where G. sulfurreducens uses a partner in syntrophy, or a solid electrode as the electron acceptor. To examine electron transfer in the absence of Fe(III), we grew G. sulfurreducens on graphite electrodes, and subjected the biofilms to cyclic voltammetry. During cyclic voltammetry, redox potential can be brought to a value too low to support acetate oxidation (−0.4 V vs. SHE) to obtain a baseline. When electrode potential is slowly increased, electron transfer from adherent cells is observed at a key onset potential as it becomes thermodynamically favorable, accelerates until a maximum electron transfer rate is reached, and follows the reverse trend as potential is decreased.
In theory, when a single event is rate-limiting in voltammetry, a Nernstian sigmoidal rise in current occurs over a ∼100 mV window, rising most steeply at the potential that most strongly affects the oxidation state of a key redox center. The potential-dependent responses of G. sulfurreducens cells during voltammetry are more complex than one-event models, and instead display at least three overlapping processes [32, 42, 64, 65, 66]. These three inflection points can be easily identified by displaying the first derivative of current increase as a function of applied potential.
Prior work described a change in voltammetry near −0.10 V vs. SHE when cbcL was deleted, which could be restored by cbcL complementation [32]. These experiments also detected a lower potential process independent of CbcL that increased with each subsequent voltammetry sweep. Impedance measurements by Yoho et al. [65] reported a similar low potential electron transfer process detectable within minutes of applying reducing electrode potentials. Based on our data, we hypothesized these unexplained features [32, 65] could be due to cbcBA activation during exposure to low potential electrodes.
To test this hypothesis, we first grew WT and ΔcbcL biofilms on electrodes as electron acceptors at +0.24 V vs. SHE, then subjected biofilms to voltammetry sweeps to reveal the low potential response below −0.2 V, and confirm loss of the middle −0.1 to −0.15 V process attributed to CbcL (Figure 5A). When cbcBA was deleted in the ΔcbcL background, the low potential electron transfer event disappeared, and all electron transfer below −0.15 V was eliminated. In the single ΔcbcBA mutant, only current below −0.2 V was eliminated, further linking cbcBA to activity in this low potential range (Figure 5A). By plotting the first derivative of voltammetry data, regions where changes in potential caused the steepest response(s) could be identified. According to these data, deletion of cbcBA eliminated an electron transfer process between −0.28 and −0.21 V, with a midpoint potential of −0.24 V vs. SHE.
CbcBA is essential for complete reduction of different Fe(III) (oxyhydr)oxides
With the evidence that G. sulfurreducens not only required cbcBA for electron transfer to soluble metals, but also to electrode surfaces, we then asked if cbcBA was involved in reduction of insoluble Fe(III) (oxyhydr)oxide particles found in the environment [13]. While common forms such as ferrihydrite, akaganeite, goethite, and hematite all have the same chemical formula (FeOOH), these minerals differ greatly in calculated redox potentials [67]. For example, freshly synthesized hydrous ferric oxide possesses a relatively high redox potential (+0.1 to 0 V, depending on age and surface area) [15, 68], while more crystalline hematite can be as low as −0.2 to −0.3 V [69]. These differences could affect the relative importance of cbcBA, especially if a lower-potential form is available.
To compare insoluble Fe(III) minerals, two different forms representing progressively lower redox potential acceptors compared to Fe(III) citrate were synthesized. First, single mutants were incubated with a freshly precipitated hydrous ferric oxide, which has an estimated redox potential of ∼0 V vs. SHE. Consistent with this acceptor having a potential near where both ImcH and CbcL have both been shown to operate, ΔimcH initially reduced Fe(III) slowly, until Fe(II) accumulated to 1-2 mM, then accelerated to reduce nearly the same total Fe(III) as reduced by WT cells (Figure 6A). The mutant lacking cbcL reduced only 50% of Fe(III), and ΔcbcBA reduced 90% of total Fe(III) compared to WT (Figure 6A). This pattern was similar to Fe(III) citrate, but showed increased importance of both cbcL and cbcBA.
The double deletion mutant ΔimcH ΔcbcBA (CbcL+) displayed the same lag as seen in ΔimcH but then also failed to reduce the last 10-15% of Fe(III) as seen for ΔcbcBA (Figure 6A, 6C). The double mutant ΔcbcL ΔcbcBA (ImcH+) ceased reduction similar to ΔcbcL, reducing 50% as much Fe(III) as WT (Figure 6C). Fe(III) reduction by double mutants aligned with the abilities of single mutants. Notably, even though concentrations of Fe(II) were much lower in hydrous ferric oxide incubations than in Fe(III) citrate, each cytochrome was necessary at the same phase of reduction, supporting the hypothesis that phenotypes were linked to the effective redox potential, not absolute Fe(III) or Fe(II) concentrations.
When a lower potential Fe(III) mineral (akaganeite) was used, the lag by ΔimcH was shorter (Figure 6B), consistent with less Fe(II) needing to accumulate to reduce redox potential and activate CbcL. Mutants lacking cbcL initiated growth, but only reduced 26% of Fe(III) compared to WT. Cells lacking cbcBA only reduced 65% of WT Fe(III) (Figure 6B). The extent of Fe(III) reduction by the double mutant ΔcbcL ΔcbcBA (ImcH+) was the same as Fe(III) reduction by ΔcbcL, and reduction by ΔimcH ΔcbcBA (CbcL+) was equivalent to reduction by the single mutant ΔcbcBA (Figure 6D).
These results across different electron acceptors, Fe(III) forms, and Fe(II) concentrations were consistent with ImcH, CbcL, and CbcBA each having a role at a different redox potential. In all cases, ImcH was essential when redox potential was above ∼0 V, CbcL was needed for reduction of moderately low potential acceptors (to about −0.2 V), and CbcBA was necessary for reduction closest to the thermodynamic limit. As lower potential electron acceptors such as akaganeite were used, CbcBA became more important for complete reduction.
While double mutants containing either imcH or cbcL demonstrated growth under at least one condition, double mutants containing only cbcBA failed to reduce Fe(III) (Figure 6C, 6D), and the same ΔimcH ΔcbcL mutant also failed to grow at any potential on electrodes (SI figure 5). The inability of cells containing only cbcBA to grow raised the possibility that CbcBA-dependent electron transfer conserves much less energy than when ImcH or CbcL is involved, possibly to the point where it cannot produce enough energy to support growth by G. sulfurreducens (SI figure 6). It also suggested that these are the only three options supporting Fe(III) reduction in this organism.
Inner membrane cytochrome background affects growth yield
Similar to how oxygen-limited E. coli induces separate terminal oxidases with a lower proton pumping stoichiometry, an explanation for different quinone oxidoreductase-like genes in Geobacter could be generation of variable amounts of proton motive force in response to environmental conditions [70, 71]. Support for this hypothesis can be found in slower growth rates of electrode-reducing ΔimcH cells [31] and higher cell counts per mol Fe(II) in ΔcbcL cells [35]. However, strains in these prior experiments still contained cbcBA, which could have been contributing to phenotypes.
If cells containing ImcH translocate more protons than when a CbcL or CbcBA-dependent pathway is in use, then forcing cells to only use ImcH and not transition to use of the other pathways should increase ATP production and growth yield. In agreement with this prediction, cbcL deletion led to higher cell numbers at the end of Fe(III) reduction (Figure 7A). Cell counts increased further when both cbcL and cbcBA were deleted. When accounting for how much Fe(III) was reduced, these differences were even more pronounced (Figure 7B). Growth yield of ΔcbcBA increased 112 ± 25% compared to WT, yield of ΔcbcL increased 152 ± 32%, and yield of ΔcbcBA ΔcbcL (ImcH+) more than doubled, to 223 ± 59% (Figure 7B). This supported higher net ATP generation by ImcH-utilizing cells compared to those using CbcL or CbcBA.
While CbcL and CbcBA negatively affected yield, data showed that these genes positively affect viability as Fe(III) became limiting. Near the end of Fe(III) reduction, viability of ΔcbcBA dropped over 50%, and ΔcbcL dropped by 68%. Cells lacking both cbcL and cbcBA had the worst survival, losing over 85% of cell viability within 24 h. A decrease in proton translocation stoichiometry would not only lower growth yield, but would also allow G. sulfurreducens to continue conserving energy as Fe(III) reduction becomes less favorable. Because we have been unable to demonstrate growth with extracellular electron acceptors by cbcBA-only strains, and CbcBA is necessary when less than 0.07 V/electron is available (7 kJ/e−), we hypothesize that CbcBA participates in an electron disposal route that primarily meets maintenance requirements when conditions are near thermodynamic limits.
Discussion
Long before the isolation of metal-reducing bacteria, higher potential Mn(IV) in sediments was shown to be reduced before lower potential Fe(III) [5]. In this report, we provide a molecular explanation for how a microorganism can choose the most thermodynamically beneficial acceptor amid a collection of minerals that lie beyond the cell membrane. Our data supports a model where redox potential controls which of three different inner membrane respiratory pathways are used, removing the need to sense the solubility or chemistry of complex extracellular metal oxides in a changing environment.
Data from this study, combined with prior genetic observations [31, 32, 35] are consistent with G. sulfurreducens utilizing ImcH to achieve high growth rates and yields when redox potential is above −0.1 V. As redox potential decreases below this level, cells are increasingly dependent on CbcL, which lowers growth rate and yield but continues generating energy. As both of these cytochromes are constitutively expressed, this model predicts that CbcL should have a mechanism to prevent it from functioning at higher potentials. When redox potential approaches −0.2 V vs. SHE, induction of cbcBA provides a means for cells to respire if CbcL cannot function, and the energy available to the organism approaches zero (Figure 4). The fact that cbcBA is not expressed until it is needed is consistent with it supporting the lowest growth yields.
Although CbcBA and CbcL both have type I b-type diheme quinone oxidoreductase domains, they share no sequence homology, and have a different number of transmembrane helices predicted to coordinate their hemes. CbcB has four transmembrane domains, with 3 conserved histidines linked to b-heme coordination, based on alignments with characterized diheme proteins. While CbcA is a separate protein, a fourth histidine for binding a b-type heme appears to be located in its C-terminal domain. This pattern, where a b-type cytochrome is coordinated by a domain from a periplasmic enzyme is also seen in [NiFe] hydrogenases related to CbcB [72, 73].
CbcL has a different domain structure, with six transmembrane helices. One histidine capable of b-heme coordination is found in each of the first three transmembrane domains, but an additional two histidines arranged similar to those in formate dehydrogenase are in the fifth transmembrane domain [74]. The presence of five heme-coordinating residues could enable more than one b-heme binding configuration in CbcL, and provide a mechanism for preventing electron transfer until a key redox potential is reached. This hypothesis lacks precedent in other model systems and illustrates the need to biochemically characterize these putative quinone oxidoreductases.
Another feature of CbcBA is its consistent location in a regulated operon that is amongst one the most conserved cytochrome-encoding regions in Geobacter, occurring in 93 out of 96 Geobacteraceae, and 119 out of 134 Desulfuromonadales. Unlike imcH and cbcL, cbcBA is expressed only under low potential conditions (Figure 3). Our data here help explain studies that detected cbcBA expression in cells harvested after Fe(III) oxide reduction, but not higher-potential Mn(IV) oxide reduction [36]. Upregulation of cbcBA in electrode-grown biofilms is also consistent with G. sulfurreducens biofilms having low-potential regions farther from electrodes [25, 75, 76]. We predict that moderate cbcBA expression reported in electrode biofilms is an average of high expression in upper leaflets with low levels in the rest of the biofilm [76, 77]. Considering all of these studies, the radical change in cbcBA expression during growth with the same electron acceptor highlights a need to control or account for redox potential when cells are harvested for RNA extraction (SI figure 1).
Such fine-tuning of respiration is not found in all metal reducing organisms. Shewanella oneidensis uses one inner membrane quinone dehydrogenase, the tetraheme c-type cytochrome CymA [78] for reduction of acceptors that differ in redox potential by over 0.6 V, including fumarate, nitrate, DMSO, Fe(III), and Mn(IV) [79]. This may be explained by the fact that Shewanella partially oxidizes organic compounds to derive most of its ATP via substrate-level phosphorylation, and uses extracellular electron transfer primarily for electron disposal [80]. In contrast, Geobacter completely oxidizes substrates and requires chemiosmosis for ATP generation. Having multiple options for coupling electron flow to proton extrusion may allow Geobacter to utilize all available electrons and compete under such varied conditions as laboratory enrichments selecting for rapid growth, energy-limited aquifers selecting for persistence, and electrodes that create redox-stratified biofilms [75, 81, 82].
Nearly every important biological respiration can be easily identified by a highly conserved functional gene, such as mcr for methanogenesis, dsr for sulfate reduction, or amo for ammonia oxidation. Tools for molecular detection of metal-reducing bacteria are lacking, and prediction of extracellular electron transfer in uncultivated organisms is difficult, due to poor sequence similarity between multiheme cytochromes and poor conservation of cytochrome content between organisms [25, 37, 83, 84]. Unlike most Geobacter c-type cytochromes, the sequence of the b-type cytochrome CbcB is highly conserved, possibly because its donor (menaquinone) and acceptor (CbcA) remains more constant. This reduced rate of genetic drift allows CbcBA homologs near BccR-like regulators to be easily identified in other Deltaproteobacteria (such as metal-reducing Anaeromyxobacter) where the b-heme protein is typically annotated as ‘thiosulfate reductase’-like. Homologous cbcBA clusters annotated as hypothetical proteins are also present in metal-reducing genera such as the Calditrichaeota (Caldithrix) and Bacteroidetes (Prolixibacter, Marinilabiliales, Labilibaculum), making cbcBA a possible marker for extracellular electron transfer in more distant phyla. Based on the presence of cbcBA homologs in genomes from uncultivated organisms within the Verrucomicrobia, and a family of cbcB-cbcA gene fusions within Chloroflexi genomes, undiscovered organisms capable of extracellular respiration still remain buried deep within anoxic sediments and metagenomic bins.
Conflict of interest
The authors declare no conflict of interest.
Acknowledgements
We thank the University of Minnesota Genomics Center for RNA sample processing and sequencing, and the University of Minnesota Supercomputing Institute for bioinformatic resources. This work was supported by the Office of Naval research grants N00014-16-1-2194, and N00014-18-1-2632.
References:
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵