Nitrite as a causal factor for nitrate‐dependent anaerobic corrosion of metallic iron induced by Prolixibacter strains

Abstract Microbially influenced corrosion (MIC) may contribute significantly to overall corrosion risks, especially in the gas and petroleum industries. In this study, we isolated four Prolixibacter strains, which belong to the phylum Bacteroidetes, and examined their nitrate respiration‐ and Fe0‐corroding activities, together with two previously isolated Prolixibacter strains. Four of the six Prolixibacter strains reduced nitrate under anaerobic conditions, while the other two strains did not. The anaerobic growth of the four nitrate‐reducing strains was enhanced by nitrate, which was not observed in the two strains unable to reduce nitrate. When the nitrate‐reducing strains were grown anaerobically in the presence of Fe0 or carbon steel, the corrosion of the materials was enhanced by more than 20‐fold compared to that in aseptic controls. This enhancement was not observed in cultures of the strains unable to reduce nitrate. The oxidation of Fe0 in the anaerobic cultures of nitrate‐reducing strains occurred concomitantly with the formation of nitrite. Since nitrite chemically oxidized Fe0 under anaerobic and aseptic conditions, the corrosion of Fe0‐ and carbon steel by the nitrate‐reducing Prolixibacter strains was deduced to be mainly enhanced via the biological reduction of nitrate to nitrite, followed by the chemical oxidation of Fe0 to Fe2+ and Fe3+ coupled to the reduction of nitrite.

In addition to SRB and methanogens, three facultatively anaerobic nitrate-reducing bacteria (NRB), namely Paracoccus denitrificans, Bacillus licheniformis, and Pseudomonas aeruginosa, have been found to enhance the corrosion of carbon steel in the presence of nitrate (Ginner et al., 2004;Jia et al., 2017;Till et al., 1998;Xu et al., 2013).
Seawater is often injected into oil-and gas reservoirs for enhanced oil and gas recovery. However, the high sulfate content of seawater can lead to corrosion and reservoir souring. To mitigate it, nitrate injection has been introduced during the last decades to promote the growth of NRB which can outcompete SRB for carbon sources (Veshareh & Nick, 2019). We previously isolated an NRB from an oil field in Northern Japan, that corroded Fe 0 under anaerobic conditions (Iino, Ito, et al., 2015). This strain was classified as Prolixibacter denitrificans sp. nov., and is the first Fe 0 -corroding NRB belonging to the phylum Bacteroidetes (Iino, Ito, et al., 2015).
However, only two strains, Prolixibacter bellariivorans strain F2 T (Holmes et al., 2007) and P. denitrificans MIC1-1 T (Iino, Ito, et al., 2015), have been isolated thus far. To explore the potential environmental functions and diversity of Prolixibacter strains, we were interested in isolating more Prolixibacter strains and characterizing their ability to corrode Fe 0 under anoxic conditions. In this study, four strains belonging to the genus Prolixibacter were newly isolated from oil-and gas fields and crude oil storage tanks in Japan.
We found that some but not all the Prolixibacter strains enhanced the corrosion of Fe 0 . The basis for the phenotypic differentiation between the Prolixibacter strains was also investigated.
The medium was dispensed by 20 ml into each 50-ml serum bottle. Dissolved air in SPYSw, NYPSw, and YSw media was removed by flushing with N 2 :CO 2 (4:1 [vol/vol]), and dissolved air in YPSw media was removed by flushing with H 2 :CO 2 (4:1) at an approximate pressure of 0.15 MPa. The bottles were sealed with butyl rubber stoppers. The pH of the medium was adjusted to 7.0 with 10 mM NaHCO 3 .

| Specimens for bacterial isolation
Crude oil emulsion samples were collected from an oil-production well in Akita Prefecture and two crude oil storage tanks, one in Kagoshima Prefecture and the other in Miyagi Prefecture, Japan.
A corrosion-scale sample was collected from the inner surface of a cast-iron pipe for brine transportation at a natural-gas-and iodineproduction plant in Chiba Prefecture, Japan. Each of these samples was kept in a transparent oxygen-barrier plastic bag containing an AnaeroPack-Anaero sachet (Mitsubishi Gas Chemical) until inoculation in fresh media.

| Enrichment, isolation, and cultivation of bacterial strains from crude oil and corrosion-scale samples
Half a milliliter of each crude oil sample or 1.0 g of the corrosionscale sample was added to 20 ml of SPYSw and YPSw media, and cultivated at 25°C for 3 weeks. Each of the resultant cultures was diluted 40-fold in the same medium and cultivated again for 3 weeks.
This procedure was repeated several times. Finally, each of the enriched cultures was streaked on 1.5% (w/v) agar slants of the same medium, and cultivated anaerobically for 7 days to isolate a single colony.

| Growth characterization
For the growth tests of each bacterial isolate, a preculture was prepared by growing an isolate in the NYPSw medium described above at 25°C for 30 days. Then, 0.1 ml of the preculture was used to inoculate in 10 ml of Sw medium supplemented with various organic acids at 10 mM. The culture was then grown at 25°C for 30 days either aerobically or anaerobically. The resulting growth was determined by measuring an optical density at 660 nm.

| Microscopy
Routine microscopic observations were performed using an Optiphot microscope (Nikon) and an S4E stereomicroscope (Leica). Laser and Electron Laboratory). The SEM was also used to observe the surfaces and cross-sections of corroded Fe 0 foils.

| Determination of 16S rRNA gene sequences
For the phylogenetic analyses of bacterial isolates, cells grown in NYPSw medium at 25°C for 30 days were harvested, and genomic DNA was extracted from the cells by the method of Saito and Miura (1963), and quantified using Qubit dsDNA HS assay kit (Thermo Fisher Scientific). The 16S rRNA gene was amplified by Approximately 100 ng of genomic DNA was used as a template under the following cycling conditions: initial activation at 95°C for 1 min, followed by 25 cycles of denaturation at 95°C for 30 s, annealing at 50°C for 30 s, extension at 72°C for 60 s and a final extension step at 72°C for 2 min. The PCR product was purified with the QIAquick PCR purification kit (QIAGEN), and an almost-complete 16S rRNA gene sequence (1444 bp) was determined by Sanger sequencing on a SeqStudio genetic analyzer (Applied Biosystems) with the BigDye terminator v3.1 cycle sequencing kit (Applied Biosystems) and one of the following six primers: 27F, 520F (5′-GTGCCAGCAGCCGCGG-3′),

| Phylogenetic analyses
Following a previously described method (Iino et al., 2010), the 16S RNA gene sequences of 12 phylogenetically-related bacteria in the order Bacteroidales were selected. After alignment using the ARB program (Ludwig et al., 2004), the phylogenetic tree was inferred from an alignment of 1380-bp-long sequences of 16S rRNA genes, and constructed using the neighbor-joining method with CLUSTAL_X 2.0.10 (Saitou & Nei, 1987;Thompson et al., 1997). The numbers at the nodes denote the bootstrap percentages derived from 1000 replications.

| Fe 0 -corrosion test
The Fe 0 -corrosion activities of Prolixibacter strains were tested in a corrosion test medium consisting of Sw medium supplemented with 100 mM HEPES buffer (pH 7.0), 10 mM nitrate, and 0.05% (wt/vol) yeast extract. Fe 0 foils (purity >99.99%, 10 × 10 × 0.1 mm) were purchased from Sigma-Aldrich, while foils of SS400 carbon steel (10 × 10 × 2 mm), and SUS316L stainless steel (10 × 10 × 2 mm) were obtained from Nippon Steel Corp. Before use, the foil surface of SS400 carbon steel was polished with Emery polishing sandpaper (No. 400, 3M), washed extensively with distilled water, and dried by air blowing at room temperature. Each of the foils was immersed in 20-ml of corrosion test medium contained in a 50-ml serum bottle; the air was removed from the medium by flushing with N 2 :CO 2 (4:1), and the bottle was sealed with a butyl rubber stopper (Nichiden-Rika Glass). Then, the foil of Fe 0 , SS400 carbon steel, or SUS316L stainless steel were sterilized by autoclave at 121°C for 15 min. before the Fe 0 -corrosion test. Twenty milliliters of the medium were added anaerobically and aseptically to a 50-ml serum bottle containing a foil of Fe 0 , SS400 carbon steel, or SUS316L stainless steel. Subsequently, 0.2 ml of a bacterial preculture was added to the medium, and the culture was incubated at 25°C for 30 days.

| Chemical analyses
After cultivation, culture fluids (100 µl) containing oxidized iron were acidified with 50 µl of 6 N HCl, and reduced with 100 µl of 1 M ascorbic acid for the quantification of total iron (ferrous and ferric ions).
The iron ion concentration in each of the acidified solutions was determined colorimetrically using o-phenanthroline, as described by Sandell (1959). For the quantification of nitrate, nitrite, and ammonium ions, the culture fluids were centrifuged at 20,400× g for 10 min. The supernatant was recovered, and filtered through a 0.2µm pore membrane filter. Nitrate, nitrite, and ammonium ions in the culture were quantified using a high-performance liquid chromatography (HPLC) system (model HIC-20Asuper; Shimadzu Corp.) equipped with a conductivity detector (model CDD-10ADsp), a Shim-Pack cation column (IC-C4), and a Shim-Pack anion column (IC-SA2). Corrosion products on the surface of iron coupons were also analyzed using an XRD analyzer with CuKa radiation ranging from 2 θ = 5 to 100° at a scanning rate of 1°/min (Rint1500; Rigaku).

| Measurement of corrosion potential and corrosion current of an Fe 0 electrode
Electrochemical analyses were conducted at 25°C in an electrochemical cell (8 ml in capacity) equipped with three electrodes, as described by Okamoto et al. (2014), with slight modifications as follows. The working electrode was an Fe 0 foil with a surface area of 3.14 cm 2 , which was placed on the bottom of the electrochemical cell, while the counter and reference electrodes were a platinum wire and an Ag/AgCl/(saturated KCl) electrode, respectively.
A filter-sterilized corrosion test medium was used as an electrolyte into which a cell suspension was injected to a final optical density at 660 nm of 0.02 to start the measurement. The corrosion potential of the working electrode was measured continuously, except that, every 8 h, the corrosion potential of the working electrode was swept at ±25 mV versus the corrosion potential for the measurement of the corrosion current of the working electrode.

| Characterization of Prolixibacter strains isolated from crude oil-and corrosion-scale samples
Previously, P. denitrificans MIC1-1 T was isolated from a crude oil emulsion sample collected from an oil well in Akita Prefecture, Japan (Iino, Ito, et al., 2015;Iino, Sakamoto, et al., 2015). In this study, a similar method was used to isolate pure bacterial cultures from three crude oil emulsion samples and one corrosion-scale sample collected from four different locations (Table A1). Bacteria capable of growing anaerobically in two different media, SPYSw and YPSw, were screened by repeated "dilution and growth" cycles as described in the Materials and Methods. A total of 76 pure cultures were obtained, which were characterized by 16S rRNA gene sequencing. As shown in Table A1, 16 Prolixibacter strains whose 16S rRNA gene sequences showed >95% identity to that of P. denitrificans MIC1-1 T were isolated using SPYSw medium, while only one Prolixibacter strain was obtained when YPSw medium was used for the screening.
The cells of all the isolates were mainly rods with a width of approximately 0.3-0.5 µm and a length of approximately 1.2-6.5 µm and had rough cell surfaces ( Figure A1). Spherical cells with a size of 0.6-0.8 µm and long rod cells with a length of 15 µm or more were observed sometimes. Cells usually occurred singly or in pairs.
Motility and spore formation were not observed during phasecontrast microscopy. The cell pellets of strains AT004 and KGS048 collected using centrifugation were salmon pink, while those of strains NT017 and SD074 were beige. The cells of strains AT004, KGS048, NT017, and SD074 were stained Gram-negatively by conventional Gram staining (Table A2).
All Prolixibacter strains grew anaerobically, with the same growth yields, in Sw medium supplemented with 0.1% (wt/vol) yeast extract (YSw medium) and YSw medium devoid of NH 4 Cl (ammonium-free YSw medium) ( Figure 2). Ammonium was formed upon the growth of these strains in an ammonium-free YSm medium (Table 2), indicating that this compound was generated by the catabolism of amino acids and other nitrogen-containing compounds present in yeast extract. Thus, yeast extract served as sources of carbon, energy, and nitrogen for the growth of the Prolixibacter strains in ammonium-free YSw medium. The anaerobic growth of P. denitrificans MIC1-1 T and three newly isolated strains (AT004, KGS048, and SD074) was enhanced in the presence of nitrate ( Figure 2) showing that nitrate respiration improved the growth yield of these strains.
On the other hand, neither growth stimulation by nitrate nor the reduction of nitrate was observed in strain NT017 and P. bellariivorans JCM 13498 T (Figure 2 and Table 2), indicating that these two strains did not respire nitrate. The ammonium concentrations in the nitrate-amended cultures of the nitrate-reducing (NR + ) strains were significantly higher than those in the nitrate-free cultures of the same strains (p < 0.05, Student's t-test), while such trends were not observed in the strains unable to reduce nitrate (NR -) (Table 2).
Thus, it seems that the NR + strains reduced nitrate not only to nitrite but also to ammonium. The sum of the nitrite and ammonium concentrations formed during the cultivation of the NR + strains were always smaller than the concentrations of nitrate metabolized by these strains. This stoichiometric anomaly could be interpreted as either that ammonium was assimilated by hosts, or that nitrate was also converted to other products than nitrite and ammonium, for example. nitric oxide.
Yeast extract in YSm medium could be replaced by d-glucose as sole carbon and energy sources, but not by simple organic acids, including lactate, pyruvate, and acetate (Table A2).
Based on the phylogenetic positions shown in Figure 1, the phenotypic properties shown in Table A2, and the nitrate-reducing activities described above, strains AT004 and KGS048 are considered to belong to P. denitrificans. Strain NT017, whose 16S rRNA gene sequence was 98.9% identical to that of P. denitrificans MIC1-1 T , differed from P. denitrificans by its cell color and the absence of nitrate respiration. Strain SD074 was considered to be a new species because of the low identity of its 16S rRNA gene sequence compared with those of P. bellariivorans and P. denitrificans.

| Fate of nitrate during Fe 0 corrosion
As has been observed in our previous study (Iino, Ito, et al., 2015), P. denitrificans MIC1-1 T grown in corrosion test medium containing 10 mM nitrate corroded Fe 0 to extents more than 20-fold higher than that in the aseptic control. When yeast extract in corrosion test and Prolixibacter sp. SD074, also oxidized Fe 0 extensively (Table 3).
Conversely, the two NRstrains, Prolixibacter sp. NT017 and P. bellariivorans JCM 13498 T did not show such an activity confirming the results from the electron microscopic and electrochemical studies.
None of the six strains enhanced Fe 0 corrosion in the presence of sulfate in place of nitrate (Table 3).
In Fe 0 -foil-containing corrosion test media inoculated with the NR + strains, nitrate was reduced by 50% over 30 days (Table 3), similarly to the results in Table 2. On the other hand, the nitrite concentrations were lower, while the ammonium concentrations were higher in the Fe 0 -containing cultures ( concomitantly with the reduction of nitrite to ammonium. To clarify this point, time-course changes in the concentrations of nitrate, nitrite, and ammonium in the cultures of the four NR + strains in the presence or absence of Fe 0 were examined for four weeks. As shown in Figure 6, a similar trend was observed among the four strains for the concentration changes of nitrate, nitrite, and ammonium. In all the cultures (Figure 6a-d,f-i), the nitrate concentrations decreased sharply during the first week, followed by gradual decreases.
In the Fe 0 -non-amended cultures (Figure 6a-d), the ammonium concentrations decreased during the first 2 to 4 days, indicating that a portion of ammonium present in the corrosion test medium was used as a nitrogen source during the growth of these strains.
Subsequently, the ammonium concentrations increased. The nitrite concentrations in the same cultures increased continuously until the end of the cultivation period.
TA B L E 2 Nitrate reduction and/or the production of nitrite and/or ammonium during the anaerobic growth of Prolixibacter strains cultured in ammonium-free YSw medium in the presence and absence of nitrate and ammonium. Note: Each of the six Prolixibacter strains was grown anaerobically at 25°C for 30 days under an atmosphere of N 2 :CO 2 (4:1) in a 50-ml serum bottle containing 20 ml of ammonium-free YSw medium or the medium supplemented with 10 mM nitrate and/or 2.8 mM ammonium. The concentrations of nitrate, nitrite, and ammonium in each culture or in an aseptic control at day 30 were determined, and those in the aseptic control were subtracted from the respective values in each culture. Data represent means and standard deviations (n = 3).
In the Fe 0 -amended cultures (Figure 6f-

F I G U R E 4
Scanning electron micrographs showing the surface and cross-sections of Fe 0 foils incubated anaerobically with NR -Prolixibacter strains. Fe 0 foils were incubated for 30 days in the corrosion test medium in the presence of Prolixibacter sp. NT017 (a-c), and P. belleriivorans JCM 13498 T (d-f). The aseptic controls are also shown (g-i). a, d, and g: the surface of Fe 0 foils (×300 magnification). b, e, and h: the surface of Fe 0 foils (×1000 magnification). c, f, and i: the cross-section of Fe 0 foils (×300 magnification). Bar = 10 µm for a, c, d, f, g, and i. Scale bar: 1 µm for b, e, and h. Scanning electron micrographs were obtained from two independent experiments Fe 0 foils may not be 100%, and/or (ii) nitrite was reduced not only chemically by Fe 0 and Fe 2+ but also biologically to either ammonium or nitric oxide. In any case, the results in Figure 7 indicate that sufficient amounts of nitrite were consumed to account for the observed amounts of oxidized Fe 0 .
The chemical corrosion of Fe 0 by nitrite under the current experimental setup was also examined. As shown in Table A4, nitrite oxidized Fe 0 in a concentration-dependent manner.

| Corrosion of SS400 carbon steel and SUS316L stainless steel by Prolixibacter strains
Two NR + strains, P. denitrificans MIC1-1 T and Prolixibacter sp. SD074, and one NRstrain, Prolixibacter sp. NT017, were used to evaluate the corroding activities on SS400 carbon steel and SUS316L stainless steel (Table A5) under anaerobic conditions. Both NR + strains corroded SS400 carbon steel more intensively than Fe 0 (Table 4).
However, these two strains did not corrode SUS316L stainless steel, indicating that they could not accept electrons through the passive film formed on the surface of stainless steel. As expected, the Fe 0non-corroding Prolixibacter sp. NT017 corroded neither SS400 carbon steel nor SUS316L stainless steel.

| DISCUSS ION
In this study, four bacterial strains closely related to P. denitrificans MIC1-1 T and P. bellariivorans JCM 13498 T were newly isolated from crude oil-or corrosion-scale samples (Tables A1 and 1).
Phylogenetically, these four strains formed a monophyletic lineage together with P. bellariivorans JCM 13498 T and P. denitrificans MIC1-1 T with a bootstrap value of 100% (Figure 1). The pairwise 16S rRNA sequence similarities among these four strains and two previously isolated strains, P. denitrificans MIC1-1 T and P. bellariivorans JCM 13498 T , were >95.8%, which was higher than the cutoff value for the delimitation of prokaryotic genera (95%) (Stackebrandt & Goebel, 1994;Tindall et al., 2010). Thus, the four newly isolated strains belonged to the genus Prolixibacter. Among them, strains AT004 and KGS048 were considered to belong to P. denitrificans based on their 16S rRNA gene sequences and phenotypic traits.
Nitrate-reducing activity was observed in some but not all strains in the genus Prolixibacter. The three strains belonging to P. denitrificans, namely strains MIC1-1 T , AT004, and KGS048, were all nitrate reducers, while Prolixibacter sp. NT017, a close relative of P. denitrificans, was nitrate non-reducer (Table 2). Although P. bellariivorans JCM 13498 T and Prolixibacter sp. SD074 formed the second clade in the genus Prolixibacter (Figure 1), the former was nitrate non-reducer and the latter was nitrate-reducer. Thus, there was no concordance between the phylogenetic relationships and the nitrate-reducing phenotype. The anaerobic growth of the NR + strains was promoted by nitrate, while that of the NRstrains was not ( Figure 2). The growth improvement with nitrate may be due to nitrate respiration, which provides more ATP than the fermentation, and due to nitrate assimilation, which provides more organic nitrogen to the hosts. The nitrogen assimilation in the NR + strains was also supported by nitrogen stoichiometry analysis in Figure 6.
Nitrate-reducing Prolixibacter strains enhanced Fe 0 corrosion, which has been demonstrated by electron microscopic studies (Figures 3 and 4), electrochemical studies ( Figure 5), and the biochemical analysis of corrosion products (Table 3, Figure 6). On the other hand, the enhancement of Fe 0 -corroding activity was not observed in the NR -Prolixibacter strains.
MIC can be classified into two types, namely chemical MIC (CMIC) and electrical MIC (EMIC), according to Fe 0 -corrosion mechanisms (Enning et al., 2012). We propose that the NR + Prolixibacter strains enhanced Fe 0 corrosion mainly through CMIC, as shown in Figure A2, based on the results obtained between days 7 and 28, during which over 70% of the corrosion products were produced.
One evidence that nitrite formed by biological nitrate reduction was the major causal agent of Fe 0 corrosion came from the results

(mM)
Prolixibacter denitrificans  presented in Figure 6, which showed the reduction of nitrite in parallel to the Fe 0 oxidation. The amount of nitrite consumed between days 7 and 28 was sufficient to produce the observed amounts of oxidized Fe 0 (Figure 7). This was further demonstrated by nitriteinduced chemical Fe 0 corrosion, as shown in Table A4. There was no evidence for biotic oxidation of Fe 0 or Fe 2+ coupled to the reduction of nitrate because the nitrate reduction by the NR + strains was not enhanced by the presence of Fe 0 and Fe 2+ (Figure 6). On the other hand, the NR + strains may catalyze the oxidation of Fe 2+ coupled to the reduction of nitrite which was observed previously (Schaedler et al., 2018), because the oxidation of Fe 0 by the NR + strains formed both Fe 2+ and Fe 3+ in the average ratio of 3:2 (Table 3), while only a scarce amount of Fe 3+ was formed in the abiotic oxidation of Fe 0 (Table A4).
P. denitrificans MIC1-1 T and Prolixibacter sp. SD074 induced the corrosion of SS400 carbon steel in the presence of nitrate.
F I G U R E 6 Nitrate reduction and the accumulation of nitrite and ammonium in the cultures of NR + Prolixibacter strains. NR + Prolixibacter strains were grown under an atmosphere of N 2 :CO 2 (4:1) in corrosion test medium either in the presence (a to e) or absence (f to j) of an Fe 0 foil. In f to j, the concentrations of oxidized iron in the aseptic controls were between 0 mM at day 0 and 0.2 mM at day 28. The strains used were P. denitrificans MIC1-1 T (A and F), P. denitrificans AT004 (b and g), P. denitrificans KGS048 (C and H) Prolixibacter sp. SD074 (d and i), and aseptic control (e and j). The corrosion rates by these strains calculated by millimeters per year were 0.15 mm/year, while the rate in the aseptic control was below 1.0×10 −3 mm/year (Text A1). All of the iron-corrosive Prolixibacter strains were isolated from either a crude oil well or crude oil storage tanks. Nitrate is widely used to prevent MIC because it enhances the growth of NRB, which competitively inhibit the growth of Fe 0 -corroding sulfate-reducing bacteria (Gittel et al., 2009;Schwermer et al., 2008;Telang et al., 1997 Technology.

CO N FLI C T O F I NTE R E S T
None declared.

F I G U R E 7
The relationship between Fe 0 oxidation and nitrite reduction. The data in Figure 6 are rearranged to show the relationships between the increment in the concentration of oxidized iron (open circles) and the consumed concentration of nitrite (filled triangle) during three weeks after day 7. The latter concentration was calculated as described in the text. The strains used were P. denitrificans MIC1-1 T (a), P. denitrificans

E TH I C S S TATEM ENT
None required.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data are provided in full in this paper apart from the 16S rRNA

TE X T A1 . E S TI M ATI O N O F CO R ROS I O N R ATE S I N M M PE R Y E A R
The amount of dissolved Fe 0 from one side of a SS400 foil (10 × 10 × 2 mm) immersed in 20-ml cultures of the nitrate-reducing Prolixibacter strains was 9.0 mM per month. Since the molecular weight of Fe 0 is 55.8, 9.0 × 55.8 × 0.02 mg (≈10.0 mg) of Fe 0 were dissolved per month. The specific gravity of Fe 0 is 7.85 g/cm 3 . Thus, 1.3 × 10 -3 cm 3 of Fe 0 per month, or 15 × 10 −3 cm 3 of Fe 0 per year were dissolved from the SS400 foil with a surface area of 1 cm 2 . These values corresponded to a corrosion rate of 0.15 mm/year.   Note: The concentrations of ferrous and ferric ions in each culture at day 30 were determined as described previously (Iino, Ito, et al., 2015) except that 2.6 mM ammonium was supplemented to the corrosion test medium. Data represent means and standard deviations (n = 3).   Figure A2 Proposed mechanism of anaerobic Fe 0 corrosion induced by nitrate-reducing Prolixibacter strains in the corrosion test medium. Electrons generated by the metabolism of yeast extract (C 4 H 7 O 2 N) are used to reduce nitrate to nitrite by nitrate reductase. Nitrite is further reduced to ammonium either biologically or chemically by the reaction coupled to the oxidation of Fe 0 to Fe 2+ . A part of Fe 2+ is oxidized to Fe 3+ mainly by biological action