H2 Is a Major Intermediate in Desulfovibrio vulgaris Corrosion of Iron

ABSTRACT Desulfovibrio vulgaris has been a primary pure culture sulfate reducer for developing microbial corrosion concepts. Multiple mechanisms for how it accepts electrons from Fe0 have been proposed. We investigated Fe0 oxidation with a mutant of D. vulgaris in which hydrogenase genes were deleted. The hydrogenase mutant grew as well as the parental strain with lactate as the electron donor, but unlike the parental strain, it was not able to grow on H2. The parental strain reduced sulfate with Fe0 as the sole electron donor, but the hydrogenase mutant did not. H2 accumulated over time in Fe0 cultures of the hydrogenase mutant and sterile controls but not in parental strain cultures. Sulfide stimulated H2 production in uninoculated controls apparently by both reacting with Fe0 to generate H2 and facilitating electron transfer from Fe0 to H+. Parental strain supernatants did not accelerate H2 production from Fe0, ruling out a role for extracellular hydrogenases. Previously proposed electron transfer between Fe0 and D. vulgaris via soluble electron shuttles was not evident. The hydrogenase mutant did not reduce sulfate in the presence of Fe0 and either riboflavin or anthraquinone-2,6-disulfonate, and these potential electron shuttles did not stimulate parental strain sulfate reduction with Fe0 as the electron donor. The results demonstrate that D. vulgaris primarily accepts electrons from Fe0 via H2 as an intermediary electron carrier. These findings clarify the interpretation of previous D. vulgaris corrosion studies and suggest that H2-mediated electron transfer is an important mechanism for iron corrosion under sulfate-reducing conditions.

Thus, microorganisms can stimulate iron corrosion by either accelerating reaction 1 or by contributing to one or more reactions that consume the electrons from reaction 1.
Sulfate reducers are often implicated in iron corrosion (1,(4)(5)(6). Desulfovibrio spp. have been the most studied pure culture isolates for investigating iron corrosion under sulfate-reducing conditions, dating back to the some of the earliest studies on microbial corrosion (1,3,7). Three mechanisms for Desulfovibrio species to consume electrons derived from Fe 0 oxidation have been proposed (Fig. 1A). The first mechanism proposed (8) was abiotic oxidation of Fe 0 coupled to proton reduction to generate H 2 : Fe 0 1 2H 1 ! Fe 12 1 H 2 (2) combined with consumption of the H 2 via sulfate reduction: Several mechanisms that might enhance H 2 production have been proposed (Fig. 1B). Hydrogenases released from moribund cells may accelerate reaction 1 by catalyzing H 2 production from Fe 0 (9). H 2 S generated from sulfate reduction may promote H 2 production from Fe 0 in two ways. Sulfide can react with Fe 0 to generate H 2 (1): Fe 0 1 H 2 S ! FeS 1 H 2 (4) and iron sulfide precipitates might facilitate electron transfer from the Fe 0 to H 1 , accelerating reaction 2 (2, 10). One of the most intriguing proposed mechanisms for Desulfovibrio species to participate in iron corrosion is direct microbial electron uptake from Fe 0 (1,(11)(12)(13)(14). D. ferrophilus (previously known as strain IS5) can grow with H 2 as the sole electron donor, but it was inferred to directly consume electrons from Fe 0 based on the observation that it reduced sulfate faster than several other H 2 -oxidizing sulfate reducers (11). However, this inference relies on the unsubstantiated assumption that direct electron transfer is faster than H 2 -mediated electron transfer from Fe 0 to microbes. Furthermore, possible adaptions in D. ferrophilus for enhanced growth on H 2 derived from Fe 0 -such as producing an extracellular hydrogenase to accelerate Fe 0 oxidation, a higher affinity for H 2 , or possibly a better capacity for attachment to Fe 0 -were not considered (3).
In subsequent studies, D. ferrophilus grew with pure Fe 0 as the electron donor, but not with stainless steel (15). This distinction is important because pure Fe 0 abiotically generates H 2 via reaction 1 (16,17), but stainless steel does not (18). In contrast to D. ferrophilus, stainless steel is an effective electron donor for Geobacter and Methanosarcina strains capable of direct electron uptake from Fe 0 (15,18,19). Notably, protease digestion of D. ferrophilus extracellular proteins did not affect sulfate reduction rates with Fe 0 as the electron donor (20), a result inconsistent with a microbe making direct electrical contact with Fe 0 because protease degrades outer-surface electrical contacts (21). Therefore, the evidence available to date suggests that D. ferrophilus most likely accepts electrons from Fe 0 via an H 2 intermediate (15).
D. vulgaris is the most intensively studied sulfate reducer for biochemical and physiological investigations, and has served as a model sulfate reducer for many corrosion studies (7). Direct Fe 0 -to-microbe electron transfer has also been proposed for D. vulgaris (13,14,22), but as with the D. ferrophilus studies, the possibility of H 2 -mediated metal-tomicrobe electron transfer was not rigorously eliminated. Studies with Geobacter (17,18), Shewanella (23,24), and Methanosarcina (19) species have provided evidence for direct electron uptake from Fe 0 by (i) eliminating the possibility that H 2 was serving as an electron shuttle between Fe 0 and cells and (ii) demonstrating with gene deletions that outer-surface c-type cytochromes were required for electron uptake from Fe 0 . In contrast, no studies have previously been reported on D. vulgaris corrosion with strains that were unable to use H 2 (7). D. vulgaris lacks outer-surface cytochromes (25), and no other D. vulgaris outer surface electrical contacts are known. Unlike the microbes previously shown to directly accept electrons from Fe 0 (17-19, 23, 24), D. vulgaris does not directly reduce Fe(III) (26), an ability common to most microbes that can directly exchange electrons with extracellular electron donors and acceptors (27).
Higher rates of corrosion following the addition of riboflavin (14,28,29) led to the suggestion that riboflavin can function as an electron shuttle that Fe 0 reduces: with D. vulgaris oxidizing the reduced riboflavin with the reduction of sulfate: However, those studies did not determine whether Fe 0 could donate electrons to riboflavin or whether reduced riboflavin can serve as an electron donor for sulfate reduction. The alternative possibility that riboflavin might stimulate other aspects of microbial metabolism was also not evaluated. Furthermore, electron transfer via H 2 was still possible in those studies.
A rigorous strategy to evaluate the possibility of H 2 serving as an intermediary electron carrier is to determine whether strains unable to use H 2 as an electron donor can respire with Fe 0 as the sole electron donor (17-19, 23, 30). In instances in which the wildtype strain of interest can consume H 2 , this can be accomplished by deleting genes necessary for H 2 metabolism (17,23,30). A strain of D. vulgaris in which genes for all of the annotated hydrogenases on the genome were deleted is available as one of a large collection of mutant strains (31). We report here on studies on Fe 0 -dependent sulfate reduction conducted with this hydrogenase-deficient strain.

RESULTS AND DISCUSSION
Hydrogenase mutant unable to grow with H 2 as electron donor. The hydrogenase-deficient mutant grew as well as the parental strain in medium with lactate as the electron donor and sulfate as the electron acceptor ( Fig. 2A), but unlike the parental strain, the hydrogenase mutant did not grow in medium with H 2 as the sole electron donor (Fig. 2B). These results suggested that the hydrogenase mutant was a suitable strain to evaluate the role of H 2 as an intermediary electron carrier during growth with Fe 0 as the electron donor.
Hydrogenase mutant cannot reduce sulfate with Fe 0 as electron donor. The parental strain reduced sulfate with Fe 0 as the sole electron donor, but the hydrogenase mutant did not (Fig. 3A). The slight decline in sulfate over time in cultures with the hydrogenase mutant could be attributed to carry over of lactate with the inoculum because the final sulfate levels for the hydrogenase mutant with Fe 0 were the same as for the parental strain without Fe 0 (Fig. 3A). As expected from previous studies under similar conditions (17), H 2 accumulated in sterile controls (Fig. 3B), reflecting abiotic Fe 0 oxidation coupled to H 1 reduction. H 2 also accumulated in cultures inoculated with the hydrogenase mutant, further demonstrating the inability of this strain to consume H 2 . H 2 accumulated more in the hydrogenase mutant cultures than in the uninoculated control, probably due to the sulfide that was transferred along with the inoculum (see sulfide effect on H 2 production below). In contrast, the parental strain maintained low H 2 concentrations ( Fig. 3B), as expected for a microbe that can consume H 2 produced from Fe 0 (17). These results indicated that H 2 produced from Fe 0 was an important electron donor for sulfate reduction by the parental strain.
However, the quantity of H 2 that accumulated in abiotic Fe 0 -only controls or in the presence of Fe 0 and the hydrogenase mutant was not sufficient to account for the amount of sulfate that the parental strain reduced with Fe 0 as the electron donor. For example, on day 14 the parental strain had reduced 3.4 mM sulfate (50% of the time zero concentration of 6.8 mM), which would require 13.6 mM H 2 (4:1 stoichiometry of H 2 oxidized per sulfate reduced, reaction 2). Only about half that much H 2 accumulated in the hydrogenase mutant cultures (Fig. 3B). One possibility for this disparity is that because rapid H 2 uptake by the parental strain maintained low H 2 concentrations (Fig. 3B), H 2 production from Fe 0 (reaction 1) was more thermodynamically favorable, possibly accelerating H 2 generation over that in the hydrogenase mutant cultures in which H 2 accumulated. We could not devise an experimental approach to abiotically mimic the expected rapid removal of H 2 at the Fe 0 surface. However, the alternative possibility that sulfide produced during the growth of the parental strain on Fe 0 accelerated H 2 production could be evaluated.
Sulfide stimulates H 2 production from Fe 0 . Sulfide that the parental strain generated from sulfate reduction with Fe 0 as the electron donor is also likely to have promoted H 2 production (Fig. 4). Parental strain sulfide production was evident from the intense black precipitates indicative of iron sulfides on the Fe 0 (Fig. 4A). In contrast, there was only a small amount of iron sulfide on the Fe 0 of the hydrogenase mutant cultures, which could be attributed to sulfide transferred along with the inoculum (Fig. 4A). Sulfide was added to sterile medium, generating black iron sulfide precipitates (Fig. 4B), to assess the possible sulfide impact on H 2 production. Adding sulfide stimulated H 2 generation (Fig. 4C). One potential source of more H 2 was the reaction of sulfide with Fe 0 (reaction 3) in which there is a 1:1 stoichiometry for sulfide reacted and H 2 produced. However, within 300 h the addition of 1.25 mM sulfide produced 2.7 mmol/L H 2 (Fig. 4C), more than twice that expected from reaction 3. This result suggested that, as previously proposed (2, 10), iron sulfide precipitates also facilitated electron transfer from Fe 0 to H 1 (reaction 1), leading to additional H 2 formation. The addition of 10-fold more sulfide only increased H 2 an additional ;2-fold (Fig. 4C), further demonstrating a lack of defined stoichiometry between sulfide additions and H 2 formation.
Culture supernatants lack extracellular hydrogenase activity. Hydrogenases released from some microbes can accelerate H 1 reduction with Fe 0 (30, 32, 33), and Role of H 2 in D. vulgaris Corrosion of Iron mBio hydrogenase activity has been detected in supernatants of moribund D. vulgaris cultures (9). However, supernatants from D. vulgaris cultures grown either with H 2 or Fe 0 did not stimulate H 2 production from Fe 0 over that in abiotic controls (see Fig. S1).
Stainless-steel studies confirm importance of H 2 as intermediary electron carrier. The inability of the hydrogenase mutant to reduce sulfate with Fe 0 as the electron donor contrasts with electroactive microbes such as Geobacter sulfurreducens (17) or Shewanella oneidensis (23), which continue to utilize Fe 0 as an electron donor even after gene deletions have eliminated the capability for H 2 uptake. Both G. sulfurreducens and S. oneidensis are capable of direct electron uptake as evidenced from an inhibition of Fe 0based respiration when genes for key outer-surface c-type cytochromes are deleted (17,23,24). Thus, the lack of sulfate reduction by the D. vulgaris hydrogenase mutant when Fe 0 was the electron donor suggests that it is incapable of direct electron uptake from Fe 0 .
This conclusion was further supported with the results of studies in which stainless steel was provided as the electron donor. Unlike pure Fe 0 , H 2 production from stainless steel is minimal (18). However, microbes capable of direct electron uptake from Fe 0 can extract electrons from stainless steel to support anaerobic respiration (18, 19, 24). D. vulgaris did not reduce sulfate with stainless steel as the electron donor (Fig. 3C).

Role of H 2 in D. vulgaris Corrosion of Iron mBio
Electron shuttles do not promote Fe 0 -dependent sulfate reduction. An alternative proposed electron transfer mechanism in Fe 0 corrosion is that flavins shuttle electrons between Fe 0 and D. vulgaris (Fig. 1A). An observed increase in Fe 0 corrosion when riboflavin is added to D. vulgaris cultures has been offered as evidence for flavin shuttling (14,28,29). However, the riboflavin amendments were to complex medium in which lactate was provided as an electron donor in addition to Fe 0 . It was not demonstrated that the riboflavin additions increased rates of Fe 0 -dependent sulfate reduction. In order to examine the possibility of electron shuttles facilitating electron transfer between Fe 0 and D. vulgaris, studies were conducted under defined conditions with Fe 0 as the sole electron donor for sulfate reduction and either riboflavin or the known electron shuttle anthraquinone-2,6-disulfonate (AQDS) (34,35). Riboflavin or AQDS did not accelerate Fe 0 -dependent sulfate reduction in the parental strain and did not enable the hydrogenase mutant to reduce sulfate with Fe 0 as the electron donor (Fig. 3D). The midpoint potentials of AQDS (2184 mV) and riboflavin (2208 mV) are probably too positive for the reduced form of these molecules to support the reduction of sulfate to sulfide (midpoint potential, 2217 mV). Therefore, the enhanced D. vulgaris Fe 0 corrosion with riboflavin amendments (14,28,29) is likely to represent an impact of riboflavin on some aspect of growth or metabolism other than enhancement of electron transfer from Fe 0 via an electron shuttle.
D. vulgaris attaches to Fe 0 electron donor. The turbidity of D. vulgaris growing on Fe 0 was very low compared to the turbidity in H 2 -grown cultures when a comparable amount of sulfate had been reduced (Fig. 5A). A portion of the cells in the D. vulgaris parental strain culture have a mutation that can, in the short-term (;100 h), delay attachment to glass surfaces (36), but confocal scanning laser microscopy revealed that cells colonized Fe 0 ( Fig. 5B and C). Individual cells were distributed across the Fe 0 surface, without apparent cell stacking, in a manner similar to the surface growth of Geobacter (17), Shewanella (23), and Methanosarcina species (19) with Fe 0 serving as the sole electron donor. The attachment of cells should be advantageous because it enables H 2 uptake at the point of production where localized H 2 concentrations are Role of H 2 in D. vulgaris Corrosion of Iron mBio higher than in the bulk surrounding environment. Furthermore, localized conditions at the cell/Fe 0 interface are likely to accelerate Fe 0 oxidation (Fig. 5D). For example, attached D. vulgaris oxidizing H 2 can make Fe 0 oxidation more thermodynamically favorable, both by removing a product of the reaction (H 2 ) and resupplying a reactant (H 1 ) near the Fe 0 surface. Sulfide produced at the Fe 0 surface can further accelerate H 2 production. A common practice in Fe 0 corrosion studies has been to infer that corrosion rates faster than that observed from abiotic H 2 generation are indicative of corrosion mechanisms other than H 2 serving as an intermediary electron carrier between Fe 0 and cells (3). However, the possibilities for attached H 2 -consuming cells to accelerate H 2 production from Fe 0 illustrate the limitations to that reasoning.
Implications. Understanding how D. vulgaris promotes Fe 0 oxidation is important because it is the microbe that has been used to develop much of the existing mechanistic framework to describe how sulfate reducers corrode Fe 0 (7). The results demonstrate Role of H 2 in D. vulgaris Corrosion of Iron mBio that the primary mechanism for D. vulgaris to reduce sulfate with Fe 0 as an electron donor is with H 2 serving as an electron shuttle between Fe 0 and the cells. Sulfate was not reduced in the absence of genes required for H 2 uptake, even when previously proposed organic electron shuttles were added. All the microbes that have been previously shown to be capable of direct electron uptake from Fe 0 have outer-surface c-type cytochromes known to be involved in extracellular electron exchange with other donors/acceptors (17-19, 23, 24). D. vulgaris lacks outer-surface c-type cytochromes (25). Direct electron uptake from extracellular electron donors by routes other than cytochromes is possible (37). For example, several methanogen species that lack outer-surface c-type cytochromes appear to directly accept electrons from Geobacter metallireducens (38)(39)(40)(41). However, the results presented here demonstrate that D. vulgaris does not function as an electrotroph with Fe 0 as the electron donor. If D. vulgaris is representative of the sulfate reducers most responsible for the corrosion of ferrous metals, then potent hydrogenase inhibitors might provide a targeted approach to mitigate iron corrosion. Microbes other than sulfate reducers also contribute to corrosion (3,33,42,43). Elucidating the mechanisms by which a diversity of microbes accelerate corrosion is essential for understanding why corrosion takes place, predicting corrosion rates under various environmental conditions, and developing strategies for corrosion prevention. The studies reported here further demonstrate that construction of appropriate mutants is a powerful approach to distinguish between a complexity of potential corrosion mechanisms.

MATERIALS AND METHODS
Microbial strains. Desulfovibrio vulgaris strains JW710 and JW5095, which were constructed in the laboratory of Judy Wall, University of Missouri (31,44), were provided from a repository of D. vulgaris mutants by Valentine V. Trotter and Adam M. Deutschbauer of the Lawrence Berkeley Laboratory. Strain JW710 is a platform strain for a markerless genetic exchange system in D. vulgaris (44). The upp gene encoding uracil phosphoribosyltransferase has been deleted, to enable utilization of the upp gene as a counterselectable marker (44). Strain JW5095 was constructed by markerless deletion of all the hydrogenases that have been described in the D. Cells were routinely grown with sodium L-lactate as the electron donor (20 mM) and sodium sulfate (20 mM) as the electron acceptor. Growth was monitored by inserting culture tubes directly into a spectrophotometer and determining the A 600 value. Growth with H 2 as the sole electron donor was evaluated with 5 mM sodium acetate as a carbon source and H 2 (140 kPa) as the sole electron donor. Cultures were incubated horizontally with shaking at 25 rpm and were routinely repressurized with H 2 to compensate for any H 2 consumption.
To evaluate growth with Fe 0 as the potential electron donor, cells were grown in NB medium with Fe 0 granules (2 g; 1 to 2 mm in diameter; Thermo Scientific) as the sole electron donor, 5 mM sulfate as the electron acceptor, and 5 mM sodium acetate as a carbon source. A 10% inoculum of a mid-log-phase culture of lactate-grown cells served as the inoculum. When specified, 50 mM riboflavin or 50 mM AQDS was added from concentrated anaerobic stock solutions. For studies with 316L stainless steel as the potential electron donor for sulfate reduction, five stainless steel cubes (5 mm Â 3 mm Â 3 mm) replaced the pure Fe 0 . The stainless-steel cubes were polished with sand paper, and the pure Fe 0 and stainless steel were presterilized with ethanol as previously described (15). Impact of added sulfide or culture supernatant on H 2 production. A final concentration of either 1.25 or 12.5 mM sodium sulfide was added to sterile Fe 0 -containing medium to determine whether sulfide stimulated H 2 production. Culture filtrates were prepared by filtering late-log-grown cultures (Fe 0grown or H 2 -grown) through a 0.2 mM PES filter in a Coy anaerobic glove bag (gas phase, 7:20:73 H 2 / CO 2 /N 2 ) into pressure tubes with 2 g of Fe 0 . Tubes were resealed and flushed with N 2 /CO 2 (80:20) for 5 min. Controls were sterile NB medium.
Analytical methods. For sulfate determinations, culture aliquots (0.1 mL) were anaerobically withdrawn with a syringe and needle, filtered (0.22 mm; polyvinylidene difluoride), and analyzed with a Dionex ICS-1000 with an AS22 column and AG22 guard with an eluent of 4.1 mM sodium carbonate and 1 mM sodium bicarbonate at 1.2 mL/min. H 2 concentrations in the headspace were monitored on an Agilent 6890 gas chromatograph fitted with a thermal conductivity detector. The column was a Supelco Role of H 2 in D. vulgaris Corrosion of Iron mBio Carboxen 1010 plot capillary column (30 m Â 0.53 mm) with N 2 carrier gas and 0.5-mL injections. The oven temperature was 40°C, the inlet was splitless at 5.5 lb/in 2 and 225°C, and the detector had a makeup flow of 7 mL/min and a temperature of 225°C. Confocal microscopy. For confocal microscopy, Fe 0 was gently removed from the pressure tube, soaked in isotonic wash buffer for 10 min, drained, stained for 10 min (Live/Dead BacLight bacterial viability kit (Thermo Fisher); 1 mL staining with 3 mL of each stain per mL), and destained for 10 min in isotonic wash buffer. Fe 0 pieces were then mounted on petri plates with an antifade/glycerol mixture. Cells were visualized with a 100Â objective on a Nikon A1R-SIMe confocal microscope with NIS-Elements software.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. FIG S1, DOCX file, 0.1 MB.

ACKNOWLEDGMENTS
This study was made possible by using the public available Desulfovibrio vulgaris strains JW710 and JW5095 made in the laboratory of Judy Wall, University of Missouri. Thomas R. Juba constructed strain JW5095. We thank Valentine Trotter and Adam Deutschbauer of the Berkeley National Laboratory for providing strains used in this study. Confocal microscopy was performed in the Light Microscopy Facility and Nikon Center of Excellence at the Institute for Applied Life Sciences, UMass Amherst. We thank the reviewers for helpful comments that improved the manuscript.