Lack of Specificity in Geobacter Periplasmic Electron Transfer

ABSTRACT

Reduction of extracellular acceptors requires electron transfer across the periplasm. In Geobacter sulfurreducens, three separate cytoplasmic membrane cytochromes are utilized depending on redox potential, and at least five cytochrome conduits span the outer membrane. Because G. sulfurreducens produces 5 structurally similar triheme periplasmic cytochromes (PpcABCDE) that differ in expression level, midpoint potential, and heme biochemistry, many hypotheses propose distinct periplasmic carriers could be used for specific redox potentials, terminal acceptors, or growth conditions. Using a panel of marker-free single, quadruple, and quintuple mutants, little support for these models could be found. Three quadruple mutants containing only one paralog (PpcA, PpcB, and PpcD) reduced Fe(III) citrate and Fe(III) oxide at the same rate and extent, even though PpcB and PpcD were at much lower periplasmic levels than PpcA. Mutants containing only PpcC and PpcE showed defects, but these cytochromes were nearly undetectable in the periplasm. When expressed sufficiently, PpcC and PpcE supported wild-type Fe(III) reduction. PpcA and PpcE from G. metallireducens similarly restored metal respiration in G. sulfurreducens. PgcA, an unrelated extracellular triheme c-type cytochrome, also participated in periplasmic electron transfer. While triheme cytochromes were important for metal reduction, sextuple ΔppcABCDE ΔpgcA mutants grew near wild-type rates with normal cyclic voltammetry profiles when using anodes as electron acceptors. These results reveal broad promiscuity in the periplasmic electron transfer network of metal-reducing Geobacter and suggest that an as-yet-undiscovered periplasmic mechanism supports electron transfer to electrodes.

IMPORTANCE Many inner and outer membrane cytochromes used by Geobacter for electron transfer to extracellular acceptors have specific functions. How these are connected by periplasmic carriers remains poorly understood. G. sulfurreducens contains multiple triheme periplasmic cytochromes with unique biochemical properties and expression profiles. It is hypothesized that each could be involved in a different respiratory pathway, depending on redox potential or energy needs. Here, we show that Geobacter periplasmic cytochromes instead show evidence of being highly promiscuous. Any of 6 triheme cytochromes supported similar growth with soluble or insoluble metals, but none were required when cells utilized electrodes. These findings fail to support many models of Geobacter electron transfer, and question why these organisms produce such an array of periplasmic cytochromes.

INTRODUCTION

In anaerobic respirations where the terminal electron acceptor is too large to cross the membrane, metabolically generated electrons are routed out of the cell via a process called extracellular electron transfer (1 to 3). To achieve this, bacteria combine cytoplasmic membrane quinone oxidoreductases, periplasmic carriers, trans-outer membrane conduits, and conductive extracellular wires in an electrical network linking intracellular biological reactions to extracellular events (2, 4). These unique electron transport chains alter metal redox states, directly exchange electrons with other bacteria, and provide new tools for bioelectronic applications (5 to 8).

Geobacter sulfurreducens is a model of extracellular electron transfer due to its ability to reduce acceptors such as Fe(III), Mn(IV), U(VI), V(V), Tc(VII), and electrodes (9). A central question raised by this versatility involves whether different redox proteins are required for each acceptor. To date, it is known that three separate inner membrane quinone oxidases are utilized depending on the redox potential of the terminal acceptor (10 to 13). In contrast, of five characterized outer membrane-spanning cytochrome conduits, specific complexes are linked to reduction of each type of metal or surface (14). Separate multiheme cytochrome nanowires are also used during growth with different terminal electron acceptors (15 to 17).

Some electron transfer mechanism is required to connect this array of inner and outer membrane proteins, as they are electrically separated by the ~30-nm-wide periplasm and cell wall (18, 19). Periplasmic redox carriers are often small promiscuous proteins able to form weak complexes and facilitate rapid electron transfer with multiple structurally unrelated proteins (20 to 25). A well-studied candidate for this role in G. sulfurreducens is the highly abundant 10 kDa triheme c-type cytochrome PpcA (Table 1). Deletion of ppcA causes defects in Fe(III) reduction (26, 27), and purified PpcA forms functional low affinity complexes with multiple periplasmic redox partners (23, 28, 29). However, G. sulfurreducens also produces four paralogs (PpcB, PpcC, PpcD, and PpcE) with highly similar heme arrangements and protein folds, and genetic evidence indicates these proteins also influence electron transfer (30, 31). For example, in growth-independent assays, deletion of ppcA impacts, but does not eliminate, insoluble Fe(III) oxide reduction, and has an even smaller effect on soluble Fe(III) citrate reduction (32). While PpcB can be more abundant during Fe(III) citrate growth, deletion of ppcB increases Fe(III) citrate reduction rates (31, 32). PpcE is rarely detected, but deletion of ppcE is reported to increase Fe(III) reduction rates (32). Deletion of PpcD is reported to increase Fe(III) citrate reduction, but not significantly affect insoluble Fe(III) reduction (32).

TABLE 1.

Summary of Ppc cytochrome characteristics^a

Protein	kDa	% amino acid identity to PpcA	Eappb (mV) vs. SHE	Potential window (mV)	Redox-Bohr effect	Expression level (RPKMe)(13)
						Mid-exp. fumaratec	30% reduced Fe(III) cit.d	70% reduced Fe(III) cit.d
PpcA	9.6 kDa	100%	−117 mV	285 mV	+	7371	4833	607
PpcB	9.6 kDa	77%	−137 mV	270 mV	−	66	167	3850
PpcC	9.6 kDa	62%	−143 mV	265 mV	NA	37	55	17
PpcD	9.6 kDa	57%	−132 mV	275 mV	+	813	79	3726
PpcE	9.7 kDa	65%	−134 mV	280 mV	−	109	151	24

^aTable adapted from Santos et al., 2015 (33).

^bE_app, apparent midpoint redox potential.

^cMid-exp. fumarate culture means cells at midexponential (OD₆₀₀ ~0.3) stage in using fumarate as an electron acceptor.

^d30% and 70% reduced Fe(III) cit. means cells taken when 30% and 70% of Fe(III) citrate was reduced to Fe(II) as an electron acceptor.

^eRPKM, reads per kilobase mapped.

Biochemical differences further suggest unique roles for Ppc paralogs (Table 1). PpcA has two higher potential hemes near −100 mV versus standard hydrogen electrode (SHE), and only one below −150 mV, while other paralogs show the opposite pattern (33). PpcC exists in two unique conformations, depending on its oxidation state (34). PpcA and PpcD display a phenomenon coined the “redox-Bohr” effect, where reduction shifts the pKa of a heme propionate side chain more than one pH unit (35, 36). When certain PpcA and PpcD hemes are reduced, this shift can cause uptake of a proton. Protonation makes heme oxidation less favorable by ~50 mV, but if an adequate acceptor is available, oxidation drives proton release. At a cost of about 50 mV/e-, this could transport a proton across the periplasm along with the electron being delivered to extracellular acceptors, in a cycle proposed to increase energy generation when cells use PpcA or PpcD (35, 36).

While it is known that expression of ppcA alone in a mutant lacking all 5 paralogs will fully restore wild-type (WT) Fe(III) citrate reduction, data supporting functional roles for the remaining periplasmic carriers is incomplete and sometimes contradictory (27). In this work, a panel of marker-free single, quadruple, and quintuple ppc deletion mutants was constructed. Some quadruple mutants containing only one paralog showed defects in Fe(III) reduction, but these changes were correlated with low periplasmic cytochrome abundance. Standardizing periplasmic levels using the ppcA promoter, ribosomal binding site (RBS), and/or signal peptide showed for the first time that any ppc cytochrome, when present in the periplasm, will support wild-type reduction of both Fe(III) citrate and Fe(III) oxide. PpcA and PpcE from G. metallireducens were also capable of restoring metal respiration in G. sulfurreducens. No significant differences in growth rate, extent of reduction, or use of specific acceptors could be found in cells using any form of PpcA or PpcD, compared to PpcB, PpcC, or PpcE. PgcA, an unrelated extracellular triheme c-type cytochrome (37), was also shown to contribute to periplasmic electron transfer. Surprisingly, while triheme cytochromes were absolutely essential for wild-type-level metal reduction, the sextuple deletion mutant lacking ppcABCDE and pgcA still grew near wild-type rates when using anodes as electron acceptors. These results reveal broad promiscuity in this periplasmic cytochrome family and show that electron transfer to electrodes can use a Ppc-independent mechanism.

RESULTS

Single deletions of Ppc-family cytochrome genes do not affect soluble or insoluble Fe(III) reduction.

G. sulfurreducens encodes five homologous ~10 kDa triheme periplasmic cytochromes that range in pairwise identity from 46 to 77%. To test if any of these cytochromes were necessary for reduction of Fe(III) citrate or Fe(III) oxide, markerless single deletion mutants lacking ppcA, ppcB, ppcC, ppcD, or ppcE were compared to wild type (Fig. 1A and B). None of these single-deletion mutants exhibited a defect in Fe(III) reduction rate, or in the final extent of reduction for either Fe(III) form, indicating that no single gene was essential under the conditions tested.

FIG 1.

Soluble Fe(III) citrate and insoluble Fe(III) oxide reduction is not affected in G. sulfurreducens mutants lacking single Ppc-family cytochromes. (A) Fe(III) citrate reduction. (B) Fe(III) oxide reduction. A and B represent mean ± standard deviations of n = 3 technical replicates. (C) Heme stain of periplasmic fractions (16% tricine gel) from WT and single-deletion mutants, representative of 2 independent extractions.

Prior transcriptional and proteomic studies have shown that other PpcA paralogs are always expressed and present in the periplasm (Table 1) (31). Consistent with this, when periplasmic proteins were obtained, 10 kDa cytochromes could still be detected in each G. sulfurreducens single mutant (Fig. 1C). Deletion of ppcA caused the greatest decrease in the abundance of the pool, followed by ΔppcD and ΔppcB. In contrast, ΔppcC or ΔppcE showed little decrease in the abundance of 10 kDa periplasmic cytochromes.

Quadruple mutants reveal PpcA, PpcB, and PpcD are equally able to support WT Fe(III) reduction.

Quadruple markerless deletion strains were then constructed to better isolate the roles of individual cytochromes. Mutants lacking four ppc homologs were labeled according to the gene still remaining in its original genomic context. For example, ΔppcBCDE was referred to as ppcA⁺, ΔppcABCD as ppcE⁺, and the strain lacking all five (ΔppcABCDE) was referred to as Δppc5.

Three quadruple deletion strains (ppcA⁺, ppcB⁺, and ppcD⁺) retained wild-type rates and extents of both Fe(III) citrate and Fe(III) oxide reduction (Fig. 2A and B). In contrast, ppcC⁺ and ppcE⁺ showed strong defects with both acceptors. Deletion of all five ppc cytochromes further diminished, but did not eliminate, Fe(III) reduction. This Δppc5 strain grew 70% slower than wild type with Fe(III) citrate (Fig. 2A). When Fe(III) oxide was the acceptor, ppcC⁺, ppcE⁺, and Δppc5 showed an ~2-day lag phase and slower reduction rate, and stopped reduction at a lower Fe(II) concentration (Fig. 2B).

FIG 2.

Single periplasmic cytochromes are able to support Fe(III) citrate and Fe(III) oxide reduction, based on quadruple and quintuple ppc deletion strains. (A) Fe(III) citrate reduction (mean ± SD n = 3). (B) Fe(III) oxide reduction (mean ± SD n = 3). Curves are representative of 3 (Fe(III) citrate) and 2 (Fe(III) oxide) independent biological experiments. (C) Heme stain of the periplasmic fraction (16% tricine gel) showing nearly undetectable 10-kDa cytochrome in ppcC⁺ and ppcE⁺ mutants, representative of 5 independent extractions.

When periplasmic proteins from quadruple and quintuple mutants were compared (Fig. 2C), 10 kDa cytochromes were only detected in ppcA⁺, ppcB⁺, and ppcD⁺, the strains demonstrating wild-type growth. In contrast, cytochrome at 10 kDa was nearly undetectable in strains with defects (ppcC⁺, ppcE⁺, and Δppc5). Consistent with data from single mutants (Fig. 1C), where deletion of ppcA caused the largest decrease in cytochrome abundance, ppcA⁺ was most intense among the quadruple deletion strains, followed by ppcD⁺ and ppcB⁺. This showed that even cytochromes with much lower native expression or abundance, such as PpcB and PpcD, could still support metal reduction similar to the more abundant PpcA.

These data also suggested that the primary explanation for the failure of ppcC⁺ and ppcE⁺ strains to reduce Fe(III) could be related to their low abundance, rather than any biochemical specificity per se. As periplasmic fractions were routinely obtained from fumarate-grown cells, periplasmic fractions were also collected from all mutants grown in Fe(III) citrate (Fig. S1 in the supplemental material). Consistent with prior transcriptional studies reporting few differences in ppc paralog expression levels between these two conditions (Table 1), ppcA⁺, ppcB⁺, and ppcD⁺ strains contained detectable 10 kDa cytochrome, while ppcC⁺, ppcE⁺, and Δppc5 did not (Fig. 2C).

PgcA, previously identified as an extracellular c-type cytochrome, also contributes to periplasmic electron transfer.

The finding that the Δppc5 strain had significant residual extracellular electron transfer ability was unexpected. We designed enrichments for spontaneous mutants upregulating or increasing use of cryptic mechanisms in strains lacking some or all Ppc-family cytochromes. While incubations of quintuple Δppc5 strains did not readily yield faster-growing suppressor strains, a slow-growing strain lacking the three most abundant Ppc cytochromes (ΔppcABD) evolved near WT Fe(III) citrate reduction. Reisolation and resequencing identified a single 9-bp in-frame deletion within a gene encoding pgcA, a triheme cytochrome unrelated to any ppc homologs (Fig. 3A).

FIG 3.

Evidence that PgcA participates in periplasmic electron transfer. (A) Partial amino acid sequence of PgcA containing the lipobox domain (blue) and deleted nucleotides (red) in ΔppcABD pgcA^Δ37-39. Fe(III) reduction (mean ± SD n = 3) (B, C) and periplasmic cytochrome abundance (D) of the ΔppcABD pgcA^Δ37-39 strain (mean ± SD n = 3) showing increased Fe(III) citrate reduction by cells retaining PgcA in the periplasm, and expected Fe(III) oxide defect. Curves are representative of 3 independent experiments. (E) Proposed model of PgcA processing, from Grabowicz et al. (50). Lsp is shown as scissors.

PgcA is an extracellular triheme c-type cytochrome with a putative lipoprotein signal sequence (37). As the suppressor strain deleted three amino acids (Ala-Gly-Cys) within the N terminus “lipobox” that is typically the target of acylation before cleavage by the lipoprotein-specific signal peptidase (Lsp/SpII) (38), this strain was designated ΔppcABD pgcA^Δ37-39. We hypothesized that the mutated signal peptide was inhibiting translocation of PgcA out of the periplasm (Fig. 3E). Lipobox motifs typically allow proper lipoprotein anchoring in the outer membrane by acting as a recognition site for a series of enzymes, including Lgt, Lsp, and Lnt, that add acyl chains to maturing lipoproteins (38).

Periplasmic proteins from the suppressor strain ΔppcABD pgcA^Δ37-39 revealed a new abundant cytochrome near the molecular weight of PgcA (50 kDa) (Fig. 3D and Fig. S2). Ultracentrifugation of periplasmic fractions did not cause any significant change in the abundance of cytochromes, confirming that membrane-bound cytochromes were not present in these periplasmic fractions (Fig. S3). Deletion of the mutant pgcA^Δ37-39 from ΔppcABD pgcA^Δ37-39 eliminated this 50-kDa periplasmic cytochrome (Fig. 3D and Fig. S2) and produced a strain even more impaired than the parent ΔppcABD in Fe(III) citrate reduction (Fig. 3B). When native pgcA was deleted from Δppc5, the resulting Δppc5 ΔpgcA demonstrated the lowest rate of metal reduction of all mutants, less than 10% of wild type with Fe(III) citrate (Fig. 4A). These lines of evidence support a model where a change in PgcA localization increased its periplasmic abundance to increase its role in periplasmic electron transfer.

FIG 4.

Decrease in residual Δppc5 electron transfer ability by deletion of pgcA. (A) Fe(III) citrate reduction (mean ± SD n = 3). (B) Fe(III) oxide reduction (mean ± SD n = 3). Curves are representative of 3 independent biological experiments. (C) Heme stain of the periplasmic fraction (16% tricine gel), representative of 2 independent extractions.

The presence of extracellular PgcA is critical for rapid insoluble metal oxide reduction (37). According to the hypothesis that PgcA was being retained in the periplasm of ΔppcABD pgcA^Δ37-39, this predicted that pgcA^Δ37-39 cells should have a defect when insoluble Fe(III) oxide is the acceptor. Consistent with this model, both ΔppcABD pgcA^Δ37-39 and Δppc5 ΔpgcA strains reduced Fe(III) oxide poorly (Fig. 3C and 4B). Both mutants showed a significant lag phase and stopped reduction at a lower Fe(II) concentration compared to WT (Fig. 3C and 4B). The fact that cells still could initiate some metal reduction after a 5- to 6-day lag could be due to other extracellular cytochromes, such as the nanowire OmcS or OmcE proteins that also play a role in metal oxide reduction. Together, these data suggest that, as extracellular cytochromes are processed in the periplasm of wild-type cells, they may participate in periplasmic electron transfer. As pgcA is typically more highly expressed during reduction of insoluble metal oxides (31), it likely only plays a minor role during reduction of soluble compounds.

Engineering ppcC and ppcE for increased periplasmic abundance shows these cytochromes can also support WT Fe(III) reduction.

Strains containing only ppcC or ppcE under the control of their native promoters reduced Fe(III) poorly and demonstrated a lack of 10-kDa periplasmic cytochromes in periplasmic extracts (Fig. 2, Table 1). Thus, we sought to more fairly test the properties of individual cytochromes by developing a standardized expression method. Beginning with the Δppc5 ΔpgcA mutant containing the lowest background levels of metal reduction (Fig. 4), ppcC and ppcE were cloned downstream of the ppcA promoter-RBS sequence, as ppcA is one of the most highly expressed genes in G. sulfurreducens (13). As controls, ppcA and ppcB were cloned downstream of the same promoters using the identical protocol. All constructs were integrated between terminators in a neutral site downstream of glmS (Fig. 5A).

FIG 5.

Fe(III) reduction by PpcC- and PpcE-containing strains is similar when periplasmic cytochrome abundance is restored. (A) Cytochrome reexpression strategy, combining the ppcA promoter and RBS (highlighted) with different signal peptides (SP). (B, D, F) Fe(III) citrate reduction (mean ± SD n = 3). (C, E, G) Heme-stained periplasmic fractions (16% tricine gel). While PpcC abundance increased with use of the ppcA promoter (B, C), PpcE required both the ppcA promoter and PpcA signal peptide (D, E). Fusion of the PpcE signal peptide to PpcA decreased PpcA abundance (F, G).

Expression of ppcA in this system successfully rescued wild-type Fe(III) reduction and produced abundant periplasmic PpcA in Δppc5 ΔpgcA (Fig. 5B and C). Similar results were obtained when ppcB was expressed using the same strategy (Fig. S4). When ppcC was expressed under the control of this ppcA promoter system, wild-type Fe(III) reduction was also rescued, and moderate levels of periplasmic PpcC were detected (Fig. 5B and C). However, expression of ppcE under the same conditions (Δppc5 ΔpgcA Tn7::ppcE) only partially improved metal reduction, and produced barely detectable periplasmic cytochrome around 10 kDa (Fig. 5B and C).

Multiple modifications were tested to identify the cause of low PpcE abundance. Recoding ppcE to eliminate rare G. sulfurreducens codons did not improve the abundance of PpcE or rescue wild-type Fe(III) citrate reduction (Fig. S5). However, when the PpcE signal peptide was replaced with the PpcA signal peptide sequence (ppcA_SP-ppcE), metal reduction improved to near wild type, and PpcE in the periplasm increased (Fig. 5D and E). To further test the hypothesis that PpcE signal peptides affected periplasmic protein levels, we generated a chimeric protein replacing the signal peptide of PpcA with that of PpcE (ppcE_SP-ppcA) (Fig. 5F and G). The PpcE signal peptide strongly decreased PpcA abundance, even though the gene remained under the control of the ppcA-promoter RBS (Fig. 5G). This decrease only caused a small defect in Fe(III) citrate reduction (Fig. 5F), again showing that large amounts of a Ppc cytochrome were not needed for rapid electron transfer.

Finally, PpcC and PpcE constructs were introduced into the Δppc5 background (as PgcA is necessary for Fe(III) oxide reduction) to test their ability to participate in Fe(III) oxide reduction. Both the PpcC and PpcE reexpression strains regained Fe(III) oxide reduction to rates similar to the wild type (Fig. 6A). In addition, periplasmic levels of each cytochrome in the Δppc5 background were similar to what was obtained in the Δppc5 ΔpgcA background (Fig. 6B). These combined results show that PpcB, PpcC, PpcD, and PpcE can support wild-type rates and extents of both soluble and insoluble metal reduction in G. sulfurreducens.

FIG 6.

Expression of G. sulfurreducens ppcC and ppcE cytochromes in Δppc5 also rescues Fe(III) oxide reduction. (A) Fe(III) oxide reduction (mean ± SD n = 3). Curves are representative of 3 independent biological experiments. (B) Heme staining of periplasmic fraction (16% tricine gel). The gel is representative of 2 independent extractions.

Expression of periplasmic cytochromes from other organisms.

As all Ppc paralogs, and even an unrelated extracellular cytochrome, could support periplasmic electron transfer during metal reduction in G. sulfurreducens, a panel of cytochromes were tested for their ability to be targeted to the periplasm and restore electron transfer in the Δppc5 ΔpgcA background. These included the PpcA and PpcE homologs from G. metallireducens (81% and 62% identity to PpcA in G. sulfurreducens), a structurally related tetraheme c-type cytochrome from Desulfovibrio vulgaris (Fig. S6C), and the tetraheme CctA (Fig. S6D) involved in Shewanella oneidensis periplasmic electron transfer (Fig. 7).

FIG 7.

Successful heterologous expression of periplasmic cytochromes from other species in G. sulfurreducens. (A) Fe(III) citrate reduction (mean ± SD n = 3). Curves are representative of 2 independent experiments. (B) Heme stain of periplasmic fractions (16% tricine gel) showing proper localization of introduced cytochromes, representative of 2 independent extractions.

All heterologous constructs used the G. sulfurreducens ppcA promoter, ribosomal binding site, and signal peptide sequence, and were detected in the periplasm (Fig. 7B). However, only the homologs from G. metallireducens fully rescued Fe(III) citrate reduction in Δppc5 ΔpgcA (Fig. 7A). CctA and DVU3171 did not improve Fe(III) citrate reduction.

Deletion of all six periplasmic cytochromes has little effect on electrode reduction.

In every experiment up to this point, as long as the cytochrome was detectable in the periplasmic fraction, ppcA, ppcB, ppcC, ppcD, or ppcE supported comparable rates and extents of reduction, and the small amount of background electron transfer activity observed in the absence of these five genes could be attributed to pgcA. Based on these results, extracellular respiration by G. sulfurreducens requires at least one of these periplasmic cytochromes for wild-type level reduction, and the Δppc5 ΔpgcA strain would be expected to be highly defective with all other extracellular electron acceptors.

Unexpectedly, the Δppc5 ΔpgcA strain grew only 8% slower when using a poised electrode (+0.24 versus SHE) as the sole electron acceptor, and reached the same maximal current as the wild type (Fig. 8A). During the exponential phase of growth, the Δppc5 ΔpgcA mutant actually produced current 12% faster than the wild type (expressed as μA produced per μg protein, n = 4, measured at 100 μA/cm²). Dividing the amount of biomass on electrodes at 100 μA/cm² by the integrated amount of current produced by the time of harvest revealed a 12–17% reduction in apparent yield of Δppc5 ΔpgcA (as protein/coulomb), consistent with the faster respiration rate (Fig. 8A).

FIG 8.

No known periplasmic cytochrome is important for electrode reduction. Comparison of wild-type versus Δppc5 ΔpgcA grown with an anode poised at +0.24 V versus SHE. (A) Chronoamperometry of WT and Δppc5 ΔpgcA. (B) 16% tricine gel of osmotic shock periplasmic fractions from biofilm and planktonic cells after 4 days of electrode growth, representative of 2 independent extractions. (C) Cyclic voltammetry of wild type and Δppc5 ΔpgcA. (D) First derivative of data from panel C. Data are representative of 4 independent biological replicates.

Periplasmic fractions recovered from both planktonic and anode biofilm cells did not reveal induction of any new periplasmic cytochromes that could explain the unexpected growth of Δppc5 ΔpgcA with electrodes (Fig. 8B). Cyclic voltammetry showed that the characteristic onset and midpoint potentials were similar in wild type and Δppc5 ΔpgcA, indicating electron transfer across a range of redox potentials was unaffected (Fig. 8C and D). The only qualitative difference observed was a reduced hysteresis between forward and reverse scans, a feature that could reflect lower electron storage capacity in cells lacking the normally highly abundant Ppc family cytochromes. Otherwise, there was no evidence that any of the six triheme cytochromes removed from this strain were necessary for electron transfer to electrodes.

Evidence for triheme cytochromes being necessary for oxidative stress protection.

Previous research reported a transient interaction between PpcA and the diheme cytochrome c peroxidase MacA (28). If periplasmic cytochromes provide reducing power to peroxidases, mutants should also have increased sensitivity to H₂O₂ stress. In lawns of cells exposed to 3% H₂O₂-soaked filter discs, the zone of inhibition was unchanged for any single ppc deletion mutant compared to wild-type cells (Fig. S7). Mutants that lacked most periplasmic cytochromes (ppcC⁺, ppcE⁺, Δppc5, and Δppc5 ΔpgcA) exhibited detectable larger zones of inhibition (Fig. S7). These data were consistent with multiple Ppc-family cytochromes, as well as PgcA, aiding H₂O₂ detoxification.

DISCUSSION

Every Geobacter genome contains between 4 and 6 PpcA paralogs that share similar heme packing and backbone structures, but have significant differences in redox potentials, microstates of partial oxidation, protonation behaviors, and surface charges near solvent-exposed hemes (30, 35). PpcA homologs from G. sulfurreducens and G. metallireducens show such large differences in midpoint potential and heme oxidation order that the two cytochromes are proposed to interact with different redox partners. In this study, we could find no direct evidence that these biochemically different proteins had unique roles, redox potentials, partners, or energetic benefits during reduction of both soluble and insoluble Fe(III). Instead, genetic data suggest the triheme cytochromes PpcA-E, PgcA, and PpcA and PpcE homologs from G. metallireducens are promiscuous enough to support complete Fe(III) reduction at similar growth rates. As none of these cytochromes were required for electron transfer to electrodes, another as-yet-unidentified periplasmic electron carrier is utilized during conductive biofilm growth.

The growth phenotypes of some deletion mutants differed from earlier insertional mutant data. For example, in prior washed cell U(VI) and Fe(III) reduction assays, ΔppcE, ΔppcBC, and ΔppcD were reported to show slightly faster Fe(III) reduction (32). In addition, a comprehensive deletion of all five ppc paralogs eliminated G. sulfurreducens’ ability to reduce Fe(III) citrate (27). Along with variation expected from growth-versus-cell-suspension assays, genetic factors could explain these differences. Tn-Seq data recently revealed many essential genes immediately up- and downstream of ppc paralogs that could be affected by antibiotic cassette insertions, such as the essential cytochrome biosynthesis genes GSU0613-0614 adjacent to ppcA/GSU0612, DNA helicase GSU0363 downstream of ppcCD/GSU0364-365, and the purine metabolism cluster GSU1758-1759 adjacent to ppcE/GSU1760 (39). Improvements in genetic tools and genomic resequencing allowed use of verified markerless deletions to better avoid polar effects. Also, variation in expression between laboratory strains is common, especially for pgcA (39). A higher background level of PgcA likely aided the finding that this cytochrome can contribute to periplasmic electron transfer.

The discovery of a ΔppcABD suppressor mutation in pgcA (pgcA^Δ37-39) that rescued Fe(III) citrate growth (Fig. 3) revealed a Lys⁻³-Ala⁻²-Gly⁻¹-Cys⁺¹ lipobox motif likely recognized by the Geobacter prolipoprotein diacylglyceryl transferase (Lgt) prior to cleavage by Lsp/SPII peptidase. We hypothesize that PgcA^Δ37-39 failed to translocate out of the periplasm due to a lack of acyl group signals. This lipobox motif could be useful for targeting secretion of future heterologous proteins to the cell surface. It is interesting to note that PgcA was assigned a periplasmic localization in earlier proteomic studies (periplasmic geobacter cytochrome A), raising the possibility that a significant amount of this cytochrome is always present in the periplasm (40).

The ability of PgcA to aid Fe(III) reduction further underscored the promiscuity of periplasmic electron transfer (Fig. 3 and 4). While both are triheme c-type cytochromes, there is no significant amino acid sequence similarity between the 50-kDa PgcA and 10-kDa Ppc proteins, and PgcA contains long repetitive proline-threonine sequences between each heme. This raises the possibility that other outer membrane and extracellular multiheme cytochromes could participate in periplasmic electron transfer prior to secretion, explaining residual metal reduction activity in the Δppc5 ΔpgcA background and possibly the growth phenotype of Δppc5 ΔpgcA mutants on poised electrodes.

With these new data, the question remains—why does Geobacter express multiple ppc paralogs at such high levels when such a metabolic burden appears unnecessary? A similar strategy, where different abundant periplasmic cytochromes appear to have overlapping functions, is also observed in the versatile metal-reducing bacterium Shewanella oneidensis (25). One hypothesis involves the ability of periplasmic cytochromes to act as “capacitors,” accepting electrons to enable proton motive force generation until extracellular oxidants can be found (25, 41). Iron-starved G. sulfurreducens cells with fewer cytochromes have much lower rates of Fe(III) reduction, and cells subjected to on/off cycles of electrode polarization produce more net current while increasing cytochrome expression and electron storage capacity (41 to 43). Having multiple promiscuous carriers in the periplasm also increases the likelihood that new respiratory pathways can be acquired, as they could easily “plug in” to the Geobacter network (22). The fact that ppc paralogs from G. metallireducens fully complemented growth of mutants (Fig. 7) indicates such horizontal exchange is feasible.

At every step of the Geobacter electron transfer chain, proteins that initially appeared redundant were later found to have nonoverlapping roles. The inner membrane cytochromes ImcH and CbcL are both expressed constitutively, but only ImcH operates above ~0 V versus SHE (10 to 13). The porin-cytochrome complexes OmcB and ExtABCD are both produced by cells in conductive biofilms, but only ExtABCD appears able to direct electrons to the electrode (14). Nanowire cytochromes OmcS and OmcE are linked to metal reduction, while only OmcZ is used for electrode reduction (16). In contrast, the promiscuous Ppc family cytochromes show the opposite behavior, collecting electrons for distribution to any available acceptor, more similar to CctA and FccA in the Shewanella electron transfer network (25). Such versatility could greatly simplify full reconstruction of extracellular electron transfer in a heterologous host, and allow synthetic combinations of proteins from multiple species. In addition, the evidence that an undiscovered periplasmic carrier exists, which is only functional during electrode growth, provides a new target for engineering a separate communication network specifically for interaction with electrical surfaces.

MATERIALS AND METHODS

Medium conditions and inoculation.

Strains and plasmids used in this study are listed in Table 2 and Table S1. Cloning information can be found in Table S2. G. sulfurreducens was grown in defined anaerobic salt medium with 20 mM acetate as the electron donor, and 40 mM fumarate, 55 mM Fe(III) citrate, or 30 mM amorphous Fe(III)-(oxyhydr)oxide as the acceptor as described (11, 44 to 46). The medium was prepared with 0.38 g/L KCl, 0.2 g/L NH₄Cl, 0.069 g/L NaH₂PO₄·H₂O, 0.04 g/L CaCl₂, 0.2 g/L MgSO₄·7H₂O, and 10 mL/L of a trace mineral mix, adjusted to pH 6.8 and buffered with 2 g/L NaHCO₃. The trace mineral mix contained 1.5 g/L nitrilotriacetic acid (NTA), 0.1 g/L MnCl₂·4H₂O, 0.5 g/L Fe₂SO₄·7H₂O, 0.17 g/L CoCl₂·6H₂O, 0.10 g/L ZnCl₂, 0.03 g/L CuSO₄·5H₂O, 0.005 g/L AlK(SO₄)₂·12H₂O, 0.005 g/L H₃BO₃, 0.09 g/L Na₂MoO₄, 0.05 g/L NiCl₂, 0.02 g/L NaWO₄·2H₂O, and 0.10 g/L Na₂SeO₄. For media with Fe(III) citrate or Fe(III) oxide as the electron acceptor, the chelated trace mineral mix was replaced with nonchelated trace mineral mix, which omitted NTA and instead dissolved minerals in 0.1 M HCl.

TABLE 2.

Strains used in this study

Strains	Description or relative genotype	Source
G. sulfurreducens
Wild type (WT)		Lab collection
DB1864	ΔppcA	This study
DB867	ΔppcB	This study
DB1483	ΔppcC	This study
DB1040	ΔppcD	This study
DB1854	ΔppcE	This study
DB1853	ΔppcBCDE (ppcA+)	This study
DB1960	ΔppcACDE (ppcB+)	This study
DB1917	ΔppcABDE (ppcC+)	This study
DB1887	ΔppcABCE (ppcD+)	This study
DB1850	ΔppcABCD (ppcE+)	This study
DB1862	ΔppcABCDE (Δppc5)	This study
DB1244	ΔppcABD	This study
DB1977	ΔppcABD pgcAΔ37-39	This study
DB1989	ΔppcABD ΔpgcA	This study
DB1546	ΔpgcA	37
DB1976	Δppc5 ΔpgcA	This study
DB1900	Δppc5 Tn7::ppcA	This study
DB2161	Δppc5 Tn7::ppcC	This study
DB2173	Δppc5 Tn7::ppcASP-ppcE	This study
DB2005	Δppc5 ΔpgcA Tn7::ppcA	This study
DB2006	Δppc5 ΔpgcA Tn7::ppcB	This study
DB2064	Δppc5 ΔpgcA Tn7::ppcC	This study
DB2007	Δppc5 ΔpgcA Tn7::ppcE	This study
DB2065	Δppc5 ΔpgcA Tn7::ppcEopt	This study
DB2062	Δppc5 ΔpgcA Tn7::ppcASP-ppcE	This study
DB2148	Δppc5 ΔpgcA Tn7::ppcESP-ppcA	This study
DB2149	Δppc5 ΔpgcA Tn7::ppcASP-Gmet ppcA	This study
DB2210	Δppc5 ΔpgcA Tn7::ppcASP-Gmet ppcE	This study
DB2150	Δppc5 ΔpgcA Tn7::ppcASP-DVU3171	This study
DB2063	Δppc5 ΔpgcA Tn7::ppcASP-cctA	This study
E. coli
UQ950	Cloning strain
S17-1	Conjugation donor strain
MFDpir	Conjugation donor strain
DB1325	MFDpir conjugation donor strain containing helper plasmid pmobile-CRISPRi	42
	UQ950 containing pRK2-Geo2-lacZa	51
DB1777	UQ950 containing pTn7m-kan-lacZa	This study
DB1880	UQ950 containing pTn7-Geo7	17
DB1284	S17-1 containing pDppcA	This study
DB1283	S17-1 containing pDppcB	This study
DB968	S17-1 containing pDppcC	This study
DB1003	S17-1 containing pDppcBC	This study
DB943	S17-1 containing pDppcD	This study
DB1851	S17-1 containing pDppcE	This study
DB1782	S17-1 containing pDpgcA	This study
DB1927	S17-1 containing pDpgcA_wo1760	This study
DB1914	S17-1 containing pGeo7::ppcA	This study
DB1916	MFDpir containing pGeo7::ppcB	This study
DB2056	S17-1 containing pGeo7::ppcC	This study
DB1945	S17-1 containing pGeo7::ppcE	This study
DB2057	S17-1 containing pGeo7::ppcEopt	This study
DB2058	S17-1 containing pGeo7::ppcASP-ppcE	This study
DB2133	S17-1 containing pGeo7::ppcESP-ppcA	This study
DB2134	S17-1 containing pGeo7::ppcASP-Gmet ppcA	This study
DB2144	S17-1 containing pGeo7::ppcASP-Gmet ppcE	This study
DB2135	S17-1 containing pGeo7::ppcASP-DVU3171	This study
DB2192	MFDpir containing pGeo7::ppcASP-cctA	This study

To make Fe(III) oxide, 10 g of FeSO₄·7H₂O was added to 1 L of water, and 5.32 mL of 30% H₂O₂ added with stirring overnight. This produced schwertmannite (Fe₈O₈(OH)₆(SO₄)·nH₂O), which was washed in distilled water and stored until needed. After addition of schwertmannite to medium and autoclaving, the Fe(III) ages to an amorphous Fe(III)-(oxyhydr)oxide, allowing generation of a highly repeatable iron oxide form between experimental replicates.

For electrode bioreactor media, 40 mM acetate was added as the electron donor and a poised graphite electrode (+0.24 V versus SHE) used as the acceptor. Fifty mM NaCl was added for osmotic balance to replace salts present in typical fumarate or Fe(III)-citrate growth medium. All media were flushed with N₂:CO₂ (80:20) gas mix passed through a heated copper column to remove oxygen.

All experiments were initiated by streaking out frozen stocks onto anaerobic 1.7% agar medium containing 20 mM acetate as the electron donor and 40 mM fumarate as the electron acceptor in a MACS MG-500 gloveless anaerobic chamber (Don Whitley Scientific) with N₂:CO₂:H₂ (75:20:5). Trypticase peptone (0.1%) and cysteine (1 mM) were added to promote recovery on the solid medium. Single colonies were picked and propagated in triplicate 1 mL liquid medium with 20 mM acetate and 40 mM fumarate, inoculated 1:10 into 10 mL medium for use in experiments.

For metal reduction experiments, cultures at 0.6 OD₆₀₀ were inoculated 1:100 into medium containing 20 mM acetate as the electron donor and 55 mM Fe(III) citrate or 30 mM Fe(III) oxide as the electron acceptor. All cultures were incubated at 30°C.

Fe(III) reduction assay.

Fe(III) citrate and Fe(III) oxide medium samples were diluted 1:10 into 0.5 N HCl for each time point. The solution was diluted further with 0.5 N HCl when needed. Samples were analyzed for Fe(II) using a modified FerroZine assay (47). The FerroZine solution contained 2 g/L FerroZine and 23.8 g/L HEPES with pH adjusted to 7.0. Three hundred microliters of the FerroZine solution were added to 50 μL of the diluted samples in 96-well plates. The plates were read at 625 nm in a BioTek Synergy multimode reader.

Growth with poised electrode as electron acceptor.

Three-electrode bioreactors consisted of a 3-cm² graphite working electrode set at +0.24 V versus SHE, a platinum counter electrode, and a calomel reference electrode. The graphite working electrode was polished with P1500 sandpaper and sonicated before each use. Each bioreactor was inoculated with 12 mL of medium with 40 mM acetate and 50 mM NaCl, and flushed with humidified N₂:CO₂ (80:20) gas overnight, before inoculation of 4 mL OD₆₀₀ 0.5 cells. Cells were grown in 30°C under constant stirring. Biomass attached to anodes was determined by removing electrodes during exponential growth (as they reached 100 μA/cm²), boiling in 0.2 N NaOH, and determining the total protein concentration using the bicinchoninic acid (BCA) assay.

Gene deletion and complementation.

A sucrose-SacB counterselection method was employed for construction of scarless deletion strains (44). Up- and downstream fragments (~750 bp each) of the target gene were joined by overlapping PCR and ligated into pK18mobsacB. The plasmid was purified and Sanger-sequenced to verify the target region after transformation into E. coli UQ950. Once confirmed, the plasmid was transformed into E. coli S17-1 or MFDpir to be conjugated with G. sulfurreducens. One milliliter of each G. sulfurreducens recipient strain and S17-1 or MFDpir donor strain culture were centrifuged together, decanted, and resuspended in the residual supernatant. This cell suspension was placed on sterilized 0.22-um pore size filter paper on agar medium with 20 mM acetate and 40 mM fumarate overnight. Merodiploids were selected on agar medium containing 20 mM acetate and 40 mM fumarate with 200 μg/mL kanamycin, and integration of the plasmid at the target site verified by PCR. Colonies with the integrated sacB-containing plasmid were subjected to sucrose-counter selection on solid agar medium containing 20 mM and 40 mM fumarate with 10% sucrose, to screen for recombination of homologous regions, which should delete the target gene in 50% of events. PCR of reisolated antibiotic-sensitive colonies using flanking primers was performed to identify deletion strains.

Suppressor analysis.

Replicate cultures of mutants with significant Fe(III) citrate reduction defects (ΔppcABD, ΔppcABCDE, ΔppcABCDE ΔpgcA) were grown with 20 mM acetate and 40 mM fumarate, then inoculated 1:100 into medium with 20 mM acetate and 55 mM Fe(III) citrate. If growth was detected, cultures were subcultured with Fe(III) citrate, and tubes demonstrating growth faster than parent cultures then streaked onto plates of agar medium containing 20 mM acetate and 40 mM fumarate to isolate colonies. Individual colonies were rescreened in medium with 20 mM acetate and 55 mM Fe(III) citrate to identify clonal suppressor strains. Genomic DNA of these strains was resequenced along with the parent, and breseq version 0.28.0 used to identify mutations compared to the parent.

Fractionation of periplasmic proteins.

A protocol for releasing periplasmic proteins via osmotic shock was adapted from Ross et al. (48). Periplasmic fractions for Fig. 1C, 2C, 3D, 4C, 5C, 5E, 5G, 6B, and 7B and Fig. S2, S3, S4B, and S5B are from stationary-phase fumarate-grown cultures. Cultures were harvested and adjusted by OD₆₀₀, so all extractions began with the same number of cells. Fig. S1 shows periplasmic fractions from Fe(III) citrate-grown cultures that reduced Fe(III) to ~50 mM.

Cultures were centrifuged at 5,200 × g for 10 min, pellets were resuspended in 1 mL 50 mM Tris, 250 mM sucrose, and 2.5 mM EDTA, pH 8.0, and equilibrated at room temperature for 5 min. This suspension was centrifuged at 16,000 × g at 4°C for 10 min and the supernatant carefully removed. The pellet was rapidly resuspended in 200 μL of ice-cold 5 mM MgSO₄ with gentle mixing on ice, causing rapid influx of water to the periplasm, osmotically disrupting the EDTA-destabilized outer membrane and releasing periplasmic proteins. After 30 min, cells and debris were removed by centrifugation at 16,000 × g for 10 min at 4°C. For electrode-grown cells, biofilms were collected by rinsing off four electrodes with a pipette tip in a tube containing 500 μL of 50 mM Tris, 250 mM sucrose, and 2.5 mM EDTA, pH 8.0, and resuspended in 5 mM 300 μL of MgSO₄ solution. Fifteen milliliters of planktonic cultures were collected in 1 mL of 50 mM Tris, 250 mM sucrose, and 2.5 mM EDTA, pH 8.0, and resuspended in 280 μL of MgSO₄ solution. Electrode-grown cells were collected at a stationary phase.

The supernatant containing the periplasmic fraction was boiled at 95°C for 5 min with SDS loading buffer (omitting β-Mercaptoethanol) and separated on a tricine-SDS-PAGE gel (adapted from Ross et al. [49]). The 16% resolving gels were made from the solution containing 5.33 mL acrylamide/bis-acrylamide 19:1 30% (wt/vol), 4.3 mL of 2.5 M Tris buffer (pH 8.8), 0.22 mL of dH₂O with 100 μL of 30 mg/mL ammonium persulfate (APS), and 6 μL of N,N,N′,N′-tetramethylethylenediamine (TEMED) added to polymerize. The 7% resolving gels were made from 2.33 mL acrylamide/bis-acrylamide 19:1 30% (wt/vol), 5.6 mL of 2.5 M Tris buffer (pH 8.8), and 1.91 mL of dH₂O with 150 μL of 30 mg/mL APS and 7 μL of TEMED. The stacking gels were made from the solution containing 0.66 mL acrylamide/bis-acrylamide 19:1 30% (wt/vol) solution, 0.76 mL 2.5 M Tris buffer (pH 8.8), 3.42 mL of dH₂O with 150 μL APS 30 mg/mL, and 5 μL TEMED. All heme-staining periplasmic fraction gels are 16% tricine gels, except Fig. 3D and Fig. S1A. For detection of c-type cytochromes, the gel was dark-incubated for 1 h in a solution containing 0.0227 g of 3,3′,5,5′-tetramethylbenzidine dissolved in 15 mL of methanol and mixed with 35 mL 0.25 M sodium acetate (pH 5.0) for a total 50 mL. The gel was visualized upon the addition of 1.5 mL of 3% H₂O₂ for 10 to 15 min.

To test if periplasmic fractions contained contaminating membranes, we performed ultracentrifugation of periplasmic fractions at 177,000 g at 4°C for 45 min (Fig. S3). The samples of pre-and postultracentrifugation did not show any significant difference.

Hydrogen peroxide tolerance assay.

An assay for hydrogen peroxide (H₂O₂) sensitivity was conducted using an H₂O₂ disc diffusion assay. Ten milliliters of media containing melted 0.5% molten agar (“0.5% top agar”) was mixed 1:100 from liquid culture of OD₆₀₀ 0.6 poured on top of the solid medium containing 1.7% agar and 0.1% Trypticase. After 3 h, autoclaved BBL Blank Paper Discs (6 mm diameter) were placed on top of the agar and spotted with 10 μL of 400 mM H₂O₂. The diameter of the zone of inhibition was measured after 2.5 days of incubation at 30°C in a MACS MG-500 gloveless anaerobic chamber (Don Whitley Scientific). The zone of inhibition around each filter disc was measured by ImageJ.