Genetic Manipulation of Desulfovibrio ferrophilus and Evaluation of Fe(III) Oxide Reduction Mechanisms

We found that genes in the genome of D. ferrophilus could be replaced with a kanamycin resistance gene via double-crossover homologous recombination, as described in Materials and Methods. This methodological advance enabled evaluation of the composition and function of proteins with possible key roles in extracellular electron exchange.

For example, a previous study provided images that suggested that D. ferrophilus expressed polar flagella (1). These filaments as well as two types of pilus-like filaments (1) were thinner than the 30- to 50-nm-diameter membrane blebs that extend from the surface of D. ferrophilus grown under stress conditions (15). In our studies the putative flagella were evident with transmission electron microscopy of wild-type cells grown with sulfate as the electron acceptor (Fig. 2a and c), and the cells were motile (Fig. 2b). Analysis of the D. ferrophilus genome identified seven putative fliC genes (DFE_0352, DFE_0942, DFE_0943, DFE_1910, DFE_2105, DFE_2171, and DFE_2355) that might encode flagellin, the structural protein of flagella (see Fig. S1 and S2 in the supplemental material). Multiple fliC genes are also present in D. vulgaris Hildenborough and Desulfovibrio desulfuricans G20 (Fig. S2). Preventing flagellar expression by specifically eliminating all of the D. ferrophilus FliC genes would require identifying more antibiotic selection markers. No additional markers have been evaluated yet. However, the sigma factor FliA regulates the transcription of flagellar biosynthesis genes, including fliC, in other bacteria (17–19). Putative FliA-dependent promoter elements were identified for three (DFE_0352, DFE_1910, and DFE_2105) of the D. ferrophilus FliC genes as well as other genes encoding proteins (ycgR, flagellar protein YcgR; cheW, chemotaxis protein W; mcp, methyl-accepting chemotaxis protein; and genes adjacent to genes for putative flagellar assembly) associated with flagellar expression and motility (Fig. S4).

Therefore, the D. ferrophilus FliA gene (DFE_1249) was deleted (Fig. S3). The fliA-deficient mutant (ΔfliA strain) appeared to express shorter flagella than those in the wild-type strain, and many of the flagella were dissociated from the ΔfliA strain cells (Fig. 2d and e). The finding that deleting fliA did not completely eliminate the expression of apparent flagellar fragments is consistent with the finding that some of the FliC genes did not appear to be controlled by FliA, but rather by the sigma factor RpoN or as-yet-unknown regulators (Fig. S4). The apparent incomplete flagellar assembly in the ΔfliA strain was associated with the loss of motility (Fig. 2b). These results suggest that the largest-diameter filaments are flagella that are required for motility.

Two types of pilus-like filaments were apparent in previous images of D. ferrophilus (1), and these same filaments were again apparent in wild-type cells (Fig. 2a). One of the pilus-like filaments was relatively short and straight whereas the other type was longer and appeared to be more flexible. The short, straight filaments were previously found to be conductive, with a conductance (0.95 ± 0.07 nS) lower than the conductance of G. sulfurreducens e-pili measured under similar conditions (4.5 ± 0.3 nS) but possibly high enough to potentially play a role in long-range electron transport (1). The conductance of the longer, flexible filaments has yet to be evaluated.

In order to obtain information on the possible composition of the two pilus-like filaments, the D. ferrophilus genome was searched for putative pilin genes. Genes for possible type IV pilins were identified (Fig. 3). The genes DFE_1797 to DFE_1801 are located next to other genes for type IV pilus assembly and are predicted to be type IV pilins. Transcript abundance for the putative pilin gene DFE_1797 was ca. 4- to 6-fold higher in cells grown with Fe(III) oxide than in cells grown with sulfate or Fe(III) citrate as the electron acceptor, suggesting a possible role in Fe(III) oxide reduction (1).

A strain in which the genes DFE_1797 to DFE_1801 were deleted, designated the Δpil5 strain, continued to express flagella and the shorter, straight filaments but not the long flexible filaments (Fig. 4a). This requirement for putative pilin genes for D. ferrophilus to express these filaments suggested that the long, flexible filaments are type IV pili and that the shorter, straighter filaments, which were previously found to be conductive, were comprised of some other protein.

We designated genes DFE_1987, DFE_1988, and DFE_1992 as pseudopilins because unlike the DFE_1797 to DFE_1801 cluster, these genes were adjacent to genes that might have a role in a type II secretion system (Fig. 3). Strains lacking genes DFE_1987 and DFE_1988 (Δspp) or DFE_1992 (Δlpp) expressed the long flexible filaments but not the straight shorter filaments (Fig. 4b and c). These results suggest that the putative pseudopilins are required for the expression of the short, straight filaments that were previously (1) found to be conductive. Pseudopili are typically involved in the secretion of exoproteins (20, 21). Therefore, at present it is not possible to conclude whether the short straight filaments are comprised of pseudopilins or of other proteins that require pseudopili for extrusion outside the cell. For simplicity these filaments are referred to as pili in subsequent discussion.

Wild-type D. ferrophilus readily reduced Fe(III) with lactate as the electron donor but not in its absence (Fig. 5; Fig. S5). The phenotypes of various gene deletion mutants were consistent with the previous finding (1) that D. ferrophilus can reduce Fe(III) oxide via an electron shuttle, a mode of Fe(III) oxide reduction that does not require flagellum-based motility or e-pili. For example, strains that lacked the thin long putative type IV pili (Δpil5 strain), or the shorter, straighter pili (Δspp and Δlpp strains) reduced Fe(III) oxide as well as the wild type did (Fig. 5a). This result demonstrates that neither of the two types of pili emanating from D. ferrophilus is essential for Fe(III) oxide reduction. It is possible that both filament types might contribute to long-range electron transport and that eliminating the possibility for expression of just one of the filaments is not sufficient to inhibit Fe(III) oxide reduction because the other pilus type is still available. Evaluation of this possibility will require the development of genetic tools for making multiple gene deletions in the same D. ferrophilus strain. However, it is expected that a microbe like D. ferrophilus that produces an electron shuttle would not also require e-pili for Fe(III) oxide reduction. For example, G. uraniireducens, which effectively reduces Fe(III) oxide by producing an electron shuttle, does not express e-pili (6), in contrast to its close relatives G. metallireducens and G. sulfurreducens, which require e-pili (9, 10, 22–24) because they lack electron shuttles (7, 8).

Flagellum-based motility enhances Fe(III) oxide reduction by G. metallireducens and G. sulfurreducens, possibly because it enables cells to search for and establish contact with Fe(III) oxides (13, 25, 26). To determine the importance of flagellum-based motility for D. ferrophilus Fe(III) oxide reduction, the ability of the nonmotile ΔfliA strain to reduce Fe(III) oxide was evaluated (Fig. 5a). The nonmotile strain reduced Fe(III) oxide as well as the wild type did. This phenotype is also consistent with an electron shuttle mechanism for Fe(III) oxide reduction because electron shuttles eliminate the necessity for motility to intimately colocalize cells and Fe(III) oxides.

Outer surface, multiheme, c-type cytochromes are typically essential electron transport components for Fe(III) oxide reduction by Gram-negative bacteria whether they rely on electron shuttles or on e-pili to extend their range of electron transport beyond the immediate cell surface (2, 27, 28). To determine whether any of the previously recognized (15) multiheme cytochromes thought to be localized near the cell outer surface (Fig. 1) might be involved in Fe(III) oxide reduction, three strains were constructed: strain Δ448-450 (deletion of DFE_0448 to DFE_0450), strain Δ461-462 (deletion of DFE_0461 and DFE_0462), and strain Δ464-465 (deletion of DFE_0464 and DFE_0465). None of these strains was defective in Fe(III) oxide reduction (Fig. 5b), indicating that none of the putative outer surface cytochromes is essential for Fe(III) oxide reduction. One possibility for the continued Fe(III) oxide reduction in the cytochrome deletion mutants is that cytochromes with genes in different clusters might compensate for the loss of deleted cytochromes. However, it was also previously found that all six of the putative outer surface cytochrome genes whose transcript abundance was quantified (DFE_0465 transcripts were not evaluated) did not have higher gene transcript abundance during growth on Fe(III) oxide than during growth on sulfate (1). Outer surface cytochromes are not required for sulfate reduction. Thus, it would be expected that a cell might specifically regulate the production of biosynthetically expensive cytochromes to minimize their expression during sulfate reduction and increase their production during Fe(III) oxide reduction if the cytochromes were essential to reduce Fe(III) oxides. Alternative strategies that do not require outer surface cytochromes for the reduction of electron shuttles are possible, as exemplified by Gram-positive bacterial reduction of electron shuttles with flavoproteins (2, 29). Therefore, the lack of cytochrome gene deletion impact on Fe(III) oxide reduction may further reflect reliance of D. ferrophilus on electron shuttling as its primary mechanism for Fe(III) oxide reduction.

D. ferrophilus reduction of the electron shuttle within the cell, rather than with putative outer surface cytochromes, would also be consistent with the observation that, following the addition of an exogenous electron shuttle, D. ferrophilus reduces Fe(III) oxide occluded within porous beads (1) much more slowly than Geobacter species (6–8), which are thought to reduce electron shuttles at the outer cell surface. A requirement for an electron shuttle to move across the outer membrane would make electron shuttling slower than if the electron shuttle was reduced with cytochromes at the outer cell surface.

Few electroactive microbes are currently genetically tractable, limiting the exploration of the diversity of mechanisms for extracellular electron exchange (2). Thus, the demonstration of a simple strategy for making gene deletions in D. ferrophilus provides an opportunity to evaluate mechanisms for electroactivity in a microbe that has substantial differences from intensively studied electroactive microbes like Shewanella and Geobacter species.

The phenotypes of mutations that prevented flagellum-based motility, as well as pilus and cytochrome expression, are consistent with the concept that D. ferrophilus primarily relies on a soluble electron shuttle for electron transfer to Fe(III) oxides. Further development of genetic tools to permit deleting multiple regions of the chromosome in the same strain is required to determine whether functional redundancy of outer surface cytochromes or the two different types of pili might explain the finding that no individual cytochrome or pilus type evaluated was essential for Fe(III) oxide reduction. However, it has already been demonstrated that D. ferrophilus can reduce Fe(III) oxides that it cannot directly contact (1), and thus, electrical contacts via pili or outer surface cytochromes may not be necessary. The results emphasize that the presence of genes for multiheme c-type cytochromes cannot be simply interpreted as evidence for involvement of c-type cytochromes in extracellular electron exchange. They also can function in intermediary intracellular electron transfer (30), intracellular reduction of metal ions (31), and temporary intracellular electron storage to permit continued respiration when electron acceptors are unavailable (32).

In addition to its ability to grow with Fe(III) oxide as an electron acceptor, another unique feature of D. ferrophilus (previously known as strain IS5) is its ability to rapidly corrode metallic iron (33). This was originally attributed to direct electron uptake from Fe(0) (33). However, direct electron uptake was only inferred (34, 35). Subsequent studies concluded that D. ferrophilus did not directly accept electrons from Fe(0) (1). Instead, its electron donor was the H₂ that is generated from Fe(0) and serves as an electron shuttle between Fe(0) and the cells. However, it has been suggested that D. ferrophilus can directly accept electrons from negatively poised electrodes without H₂ serving as an intermediary electron carrier (36–38). Expansion of the genetic approach described here should make it feasible to more rigorously address these additional questions about D. ferrophilus extracellular electron exchange by generating the appropriate mutant strains.

D. ferrophilus IS5 was obtained from DSMZ-German Collection of Microorganisms and Cell Cultures GmbH. It served as the parent (wild-type) strain for the construction of mutants. D. ferrophilus was anaerobically grown at 30°C in DSMZ 195c medium with modifications as previously described (1). Sodium dl-lactate (60 mM) served as the electron donor with either sodium sulfate (30 mM) or poorly crystalline Fe(III) oxide (100 mmol/L) as the electron acceptor. The poorly crystalline Fe(III) oxide was prepared as previously described (39) by slowly neutralizing a solution of ferric chloride and collecting and washing the resulting precipitate. This form of Fe(III) oxide is highly available for microbial reduction and is typically employed when evaluating the ability of isolates to reduce Fe(III) oxides (40). The gas phase was N₂/CO₂ (80:20). Cysteine (1 mM) was added as a reductant when sulfate was the electron acceptor.

To construct the ΔfliA strain, ~1-kb upstream and downstream regions of the fliA gene were amplified by PCR with primer pairs fliA-up1/fliA-up2 and fliA-do1/fliA-do2 (see Table S1 in the supplemental material), respectively, with the genomic DNA of D. ferrophilus IS5 as the template. A kanamycin resistance gene was amplified by PCR with a primer pair, km-F/km-R (Table S1), and the plasmid pBBR1MCS-2 (41) as the template. The PCR products were digested with restriction enzymes, ligated with T4 DNA ligase, and cloned in the plasmid pBluescript (Stratagene). Correct sequences of the cloned PCR products were confirmed by DNA sequencing. The plasmid containing the upstream and downstream regions of the fliA gene and the kanamycin resistance gene was linearized by XbaI. The fliA gene was replaced with the kanamycin resistance gene via double-crossover homologous recombination (Fig. S3).

Transformation of D. ferrophilus was carried out by electroporation. Electrocompetent D. ferrophilus cells were prepared from cultures grown in the lactate-sulfate medium. All manipulations were conducted in an anaerobic glove bag (7% H₂/20% CO₂/73% N₂). The cultures (10 mL) at the mid-log phase (~0.3 value of optical density at 600 nm) were harvested in a centrifuge tube with an O-ring screw cap by centrifugation at 4°C. The cells were washed with a buffer (1 mM MgCl₂/175 mM sucrose/1 mM HEPES, pH 7) twice. The washed cells were resuspended in 50 μL and kept on ice. The linearized plasmid (1 μg) was mixed with the washed cells, and the mixture was electroporated in an electroporation cuvette (0.1-cm gap) at 1.5 kV by a MicroPulser electroporator (Bio-Rad). The cells were recovered in the lactate-sulfate medium and incubated at 30°C for 24 h. Transformants were selected with the lactate-sulfate medium containing 1.5% agar supplemented with Geneticin (G418) (400 μg/mL). Typically, 30 to 60 colonies per plate were recovered. Proper recombination was confirmed by PCR with primer pairs fliA-P1/fliA-P2, fliA-P1/fliA-P3, and fliA-do2/km-V (P4 and P5 in Fig. S3).

Other mutants were constructed as described above. Primers used for the strain constructions are shown in Table S1.

Growth of D. ferrophilus with sulfate as the electron acceptor was monitored by measuring the optical density at 600 nm. Reduction of Fe(III) was monitored by measuring the concentration of Fe(II) by the ferrozine assay as described previously (39). The motility assay was conducted on the lactate-sulfate medium containing 0.5% or 0.7% agar at 30°C in the anaerobic glove bag. The cultures (5 μL) at the mid-log phase were spotted on the agar surface. Photographs were taken after 4 days (0.5% agar) or 9 days (0.7% agar) of incubation. Transmission electron microscopy was performed as described previously (1).

Sequence similarity was analyzed by NCBI blastp (protein-protein BLAST) (https://blast.ncbi.nlm.nih.gov). Genome sequences were utilized from NCBI for D. ferrophilus IS5 (NCBI reference sequence NZ_AP017378.1), D. vulgaris Hildenborough (NCBI reference sequence NC_002937.3), and D. desulfuricans G20 (NCBI reference sequence NC_007519.1). Sequence alignments were constructed by Clustal Omega Multiple Sequence Alignment (https://www.ebi.ac.uk/Tools/msa/clustalo/). Protein localization was predicted by PSORTb (https://www.psort.org/psortb/).