New Insight into Microbial Iron Oxidation as Revealed by the Proteomic Profile of an Obligate Iron-Oxidizing Chemolithoautotroph

Abstract

Microaerophilic, neutrophilic, iron-oxidizing bacteria (FeOB) grow via the oxidation of reduced Fe(II) at or near neutral pH, in the presence of oxygen, making them relevant in numerous environments with elevated Fe(II) concentrations. However, the biochemical mechanisms for Fe(II) oxidation by these neutrophilic FeOB are unknown, and genetic markers for this process are unavailable. In the ocean, microaerophilic microorganisms in the genus Mariprofundus of the class Zetaproteobacteria are the only organisms known to chemolithoautotrophically oxidize Fe and concurrently biomineralize it in the form of twisted stalks of iron oxyhydroxides. The aim of this study was to identify highly expressed proteins associated with the electron transport chain of microaerophilic, neutrophilic FeOB. To this end, Mariprofundus ferrooxydans PV-1 was cultivated, and its proteins were extracted, assayed for redox activity, and analyzed via liquid chromatography-tandem mass spectrometry for identification of peptides. The results indicate that a cytochrome c₄, cbb₃-type cytochrome oxidase subunits, and an outer membrane cytochrome c were among the most highly expressed proteins and suggest an involvement in the process of aerobic, neutrophilic bacterial Fe oxidation. Proteins associated with alternative complex III, phosphate transport, carbon fixation, and biofilm formation were abundant, consistent with the lifestyle of Mariprofundus.

INTRODUCTION

Iron (Fe) is one of the most abundant elements on Earth and a major component of the oceanic crust (1). The biologically catalyzed oxidation of Fe at circumneutral pH with oxygen (O₂) as the terminal electron acceptor has remained largely enigmatic, even though neutrophilic Fe oxidation is among the first chemoautotrophic microbial metabolisms described in the literature (2). This lack of data is due, in part, to obstacles such as culturing of fastidious microaerophilic, neutrophilic, Fe-oxidizing bacteria (FeOB); the relatively low cell densities in cultures; and the interference of Fe oxides with sample preparation. In addition to this, aerobic, neutrophilic FeOB have so far been elusive to genetic manipulation. Consequently, these challenges have impeded the ability to understand the mechanisms of neutrophilic Fe oxidation in the presence of O₂ and inhibited the development of molecular diagnostics targeting genetic markers for such a biological function (i.e., molecular probes targeting genes, transcripts, or proteins indicative of activity). Recent genomic analyses of microaerophilic, neutrophilic FeOB (3–6) have suggested genes that might be involved in Fe oxidation; however, evidence of expression of these genes in FeOB has not been shown. This is in contrast with the recent advancements in the elucidation of the mechanisms of Fe oxidation in aerobic, acidophilic bacteria (especially Acidithiobacillus ferrooxidans and Leptospirillum spp.) and neutrophilic, anoxygenic, phototrophic bacteria (Rhodopseudomonas palustris and Rhodobacter spp.) (see reference 7 for a review).

Mariprofundus ferrooxydans is a microaerophilic, chemolithoautotrophic FeOB that belongs to the Zetaproteobacteria class of the Proteobacteria and is known for its ability to biomineralize Fe in the form of extracellular twisted filaments (6, 8). It is one of only a few isolates from the marine environment that have been shown to be obligate chemolithoautotrophs that oxidize Fe at circumneutral pH, in direct competition with the abiotic chemical oxidation of Fe (6, 9, 10). M. ferrooxydans and related Zetaproteobacteria are the dominant biomineralizing organisms associated with Fe-rich hydrothermal ecosystems in the ocean (11, 12). Their presence in environmental samples has also been documented in several deep-sea sites, including the Juan de Fuca Ridge, Lō'ihi Seamount, Tonga Arc, Southern Mariana Trough (see reference 13 for a review), and, most recently, the Mid-Atlantic Ridge (14). Studies of coastal waters of Maine and China have shown evidence that this class of microorganisms is involved in microbiologically influenced corrosion of steel (10, 15). Zetaproteobacteria were also recently identified in the terrestrial environment in Fe-rich, saline-influenced, deep-subsurface waters of a geyser in Utah (16).

In addition to its ecological relevance, M. ferrooxydans is a useful organism to study because its genome was recently sequenced and annotated (3). However, the genome of M. ferrooxydans strain PV-1 contains 101 coding DNA sequences (CDSs) for energy production and conversion and at least 37 CDSs for different types of cytochromes (i.e., proteins involved in electron transfer) based on a Pfam search of the Integrated Microbial Genomes (IMG) database (17). Therefore, development of a proper model for how these organisms couple the oxidation of Fe(II) to a functioning electron transport chain (ETC) is difficult to elucidate without direct information about which proteins are expressed. In this study, a proteomic profile of M. ferrooxydans was generated to help answer fundamental questions about its physiology, with a focus on the expressed proteins that are part of the ETC. Potentially critical proteins involved in aerobic, neutrophilic Fe oxidation are identified and discussed.

MATERIALS AND METHODS

Culturing.

To generate enough biomass for protein extraction and characterization, a recently developed large-scale culturing method for M. ferrooxydans was used (18). The final salinity of the artificial seawater medium was decreased from 37 ppt to 35 ppt (average seawater salinity) to prevent the possible expression of salinity stress proteins. The large batch cultures were grown in 10 replicate 1-liter autoclavable polycarbonate bottles per experiment, with each bottle containing 800 ml of medium. Each culture vessel was inoculated with 40 ml of log-phase M. ferrooxydans PV-1 (ATCC BAA-1020) (∼10⁷ cells grown in a 100-ml batch culture) and incubated in the dark, horizontally at room temperature, without agitation. Filter-sterilized Fe(II) from a 100 mM FeCl₂ anoxic stock solution (initial concentration of 500 μM in medium) and air (4 ml) were added every 24 h. Cell counts to determine cell density were performed as described previously by Emerson and Moyer (11).

Protein extraction.

To isolate proteins from large-scale cultures of M. ferrooxydans, recently developed protein extraction protocols to overcome interference from Fe oxides were used (18). A brief description is provided here. Crude protein extracts were obtained as follows: a total of 8 liters (0.8 liters of culture in 10 replicates) of late-log-phase cultures per experiment (n = 2) was harvested by filtration on 47-mm, 0.22-μm-mesh, black Whatman-Nuclepore polycarbonate filters (GE Healthcare Life Sciences) with a 10-μm-pore-size support polycarbonate filter (Isopore; EMD Millipore). The filters were incubated in 80 ml of a 0.1 N NaOH–2% (wt/vol) SDS solution on ice for 10 min, with vigorous vortexing for 30 s every 2 min to lyse cells and release proteins. The mixture was centrifuged twice at 14,000 × g for 10 min at 4°C to pellet cellular debris and Fe oxides. The clarified supernatant was concentrated and buffer exchanged with 40 mM Tris base (pH 8.5) by using Macrosep concentrators (Pall) with a 3-kDa-molecular-mass-cutoff membrane according to the manufacturer's instructions. SDS was removed with SDS-Out (Thermo Fisher) according to the manufacturer's instructions. The above-described procedure resulted in a concentrated, slightly pink crude extract containing both insoluble and soluble proteins, which was then frozen at −20°C until further analysis.

For the insoluble (membrane) fraction, a total of 8 liters (0.8 liters of culture in 10 replicates) of late-log-phase cultures per experiment (n = 2) was harvested as described above, and the filters were then immersed in a total volume of 250 ml of cold, filter-sterilized 0.2 M oxalic acid (pH 3.0) for 2 h with mixing on a shaker table to dissolve the Fe oxides. A cell separation protocol (18) was then performed to remove stalks from cells by using Nycodenz density gradient medium (Axis-Shield, Oslo, Norway). The pellet containing cells was resuspended in 5 ml of 0.5 M NaCl–40 mM Tris base buffer at pH 8.5 and sonicated for 10 cycles of 30-s pulses followed by 30 s of cooling by using a Branson 450 digital sonifier (Branson Ultrasonics) to lyse the cells. The resulting protein extract was centrifuged at 14,000 × g for 10 min at 4°C to pellet cell debris. The supernatant was loaded into 13- by 51-mm, thin-wall, polyallomer tubes (Beckman Coulter) and centrifuged at 100,000 × g for 2 h at 4°C by using an Optima Max XP ultracentrifuge (Beckman Coulter) to fractionate soluble (supernatant) and membrane (pellet) proteins. The reddish pellet containing the membrane fraction was washed with 0.5 M NaCl–40 mM Tris base buffer for two additional ultracentrifugation cycles and was subsequently resuspended in 1 ml of 0.5% Ultrol-grade n-dodecyl-beta-d-maltoside (Calbiochem) to solubilize proteins. The solubilized membrane fraction was quantitated by using the RC DC protein assay (Bio-Rad) according to the manufacturer's instructions, using a UV-1601 spectrophotometer (Shimadzu Scientific Instruments). All samples were stored at −20°C.

SDS-PAGE and gel assays.

For the full proteomic profile, 30 μg of duplicate crude protein extracts was loaded onto a NuPAGE Novex 4 to 12% Bis-Tris precast polyacrylamide gel (Life Technologies) and run at 100 V under nonreducing conditions. Gels were stained with Bio-Safe Coomassie stain (Bio-Rad) for 1 h and destained with high-performance liquid chromatography (HPLC)-grade water for 2 h. The duplicate gel lanes were each cut into 11 slices in same areas of molecular mass by using sterile scalpels, stored in HPLC-grade water at 4°C, and immediately submitted for proteomic analysis. Membrane fractions and additional crude fractions were run on 12% TGX polyacrylamide gels (Bio-Rad) at 90 V under reducing and nonreducing conditions (i.e., no dithiothreitol [DTT] and no heating). Membrane fractions were run under nonreducing conditions to avoid precipitation of proteins and to increase the sensitivity of the heme stain. Gels were stained and destained as described above and visualized with a Gel Doc XR+ imaging system (Bio-Rad) equipped with a white light transilluminator. Gel images were analyzed with Image Lab software (Bio-Rad). In-gel, heme-protein staining was performed according to methods described previously by Francis and Becker (19). An in-gel, Fe oxidation assay was performed as follows: after SDS-PAGE, the gel was washed briefly with ultrapure water, fixed with 12.5% trichloroacetic acid (TCA) for 15 min, washed again with ultrapure water for another 15 min (water changed every 5 min), and immersed in a saturated ferrocyanide solution (14.5 g in 50 ml of ultrapure water) for 1 h. The gel was then washed repeatedly with ultrapure water for at least 1 h. A blue band in the gel indicated the oxidation of ferrocyanide to ferricyanide.

Peptide identification and protein analysis.

Excised protein samples were submitted to the Children's Hospital of Los Angeles Proteomic Core Facility for trypsin digestion and liquid chromatography-tandem mass spectrometry (LC-MS/MS) on a Thermo LTQ-Orbitrap XL mass spectrometer equipped with an Eksigent Nanoliquid Chromatography 1-D Plus system. The resulting MS/MS spectra were searched against the M. ferrooxydans PV-1 proteome in the Uniprot database by using the Proteome Discoverer SEQUEST search engine. Alternatively, the proteome of M. ferrooxydans M34 in the IMG database was searched to cover gaps present in the PV-1 proteome. The criterion that was used for having a protein identified in the profile was the need for a minimum of two unique peptides that matched it, with each peptide established at >95% probability. Protein and peptide false discovery rates (FDRs) were 0% throughout the study. Predictions of subcellular localization were performed with CELLO, PSORTb, Phobius, and PRED-TMBB (20–23). Protein signatures were searched with InterProScan (24). Alignment and analysis of amino acid sequences were performed with Geneious version R6 (Biomatters Ltd.). Maximum likelihood (ML) phylogenetic trees were obtained from Geneious by using the PhyML plug-in (25) and based on 1,000 bootstrap replications. Statistical analysis was performed by using JMP version 11 (SAS Institute Inc.). The mass spectrometry proteomic data have been deposited at the ProteomeXchange Consortium (26) via the PRIDE partner repository with the data set identification numbers PXD001050 and PXD001439 (see Table S1 in the supplemental material for details).

DNA extraction and PCR.

Cultures of M. ferrooxydans PV-1 were harvested as described above. DNA was extracted by using the FastDNA Spin soil kit according to the manufacturer's instructions (MP Biomedicals). Extracted DNA was stored at −20°C until processing. Genomic gaps were amplified by PCR on a Veriti thermal cycler (Life Technologies) as follows: 1 step of denaturation at 95°C for 4 min; 35 cycles of denaturation, melting, and extension (95°C for 30 s, 51°C for 30 s, and 72°C for 3 min, respectively); 1 step of extension at 72°C for 10 min; and 1 final step of cooling at 4°C. Primers used for the gap containing the Cyc1_PV-1 gene were 79F (5′-GAAGCGATGGGAAATGTGAAT-3′) and 375F (5′-CACACTGGAAGATGTTCTGG-3′). Primers used for the gap containing the Cyc2_PV-1 gene were 555F (5′-ACTGATGGGTATCAACAACC-3′) and 92R (5′-CCTATCTGTACCGAGCATTC-3′). The amplicons were purified by using the QIAquick PCR purification kit (Qiagen). Amplicon sizes were checked in a 1% agarose gel via electrophoresis. Purified PCR amplicons were submitted for Sanger sequencing and primer walking (Laragen).

Nucleotide sequence accession numbers.

DNA sequences were deposited in GenBank under accession numbers KR106296, KR106297, KR091570, and BK009249.

RESULTS AND DISCUSSION

General proteomic profile results.

Overall, 825 proteins from a total of 2,866 protein-coding genes were identified by using the amino acid database generated from the genome of Mariprofundus ferrooxydans PV-1 (here PV-1) (see Data Set S1 in the supplemental material). The percent proteome coverage (28.7%) is well within the range obtained by previous proteomic studies employing LC-MS/MS (27–30). There were 200 hypothetical proteins identified (24% of all identified proteins), including the most abundant protein (SPV1_07114). Clusters of Orthologous Groups (COG) functional category codes were assigned to 709 proteins (86% of all identified proteins), including some proteins that were annotated as hypothetical.

Figure 1 illustrates the proportion of assigned COG functional categories and their relative abundances in the obtained proteomic profile (see Fig. S1 in the supplemental material for means with standard deviations). COGs J, C, O, and G were the groups with the highest number of assigned spectra. An analysis of means (ANOM) revealed that these four COGs had assigned spectrum means that were significantly different (>95% confidence) from the overall mean calculated for all the COG groups (see Fig. S1 in the supplemental material). COG J (translation, ribosomal structure, and biogenesis) was the functional category with the most assigned spectra (average of 20%). The three elongation factors required for translation in prokaryotes, Tu (SPV1_09894), Ts (SPV1_05739), and G (SPV1_09899), were among the most abundant proteins expressed in PV-1 (Table 1). Proteins associated with translational functions were also abundant, as seen in other proteomic studies involving bacteria (31, 32). The second largest COG represented proteins with function in energy production and conversion (COG C). Within this group, the most abundant proteins identified were associated with ATP synthase (SPV1_13824, SPV1_13814, and SPV1_13804) and ETC components such as a soluble cytochrome c (A37KDRAFT_02147), cytochrome cbb₃ oxidase subunits (SPV1_07401 and SPV1_07396), and molybdopterin oxidoreductase (SPV1_03948). The third largest COG contained proteins with posttranslational modification, protein turnover, and chaperone functions (COG O). The most abundant proteins within this group were the chaperone DnaK (SPV1_07796) and the chaperonin GroEL (SPV1_00170), which may have functions related to preventing the aggregation of newly synthesized polypeptides as well as preventing protein misfolding (33). The fourth largest COG represented proteins with function in carbohydrate transport and metabolism (COG G). Within this group, Calvin cycle proteins were among the most abundant in the profile, including RubisCO (SPV1_04963), transketolase (SPV1_05909), and fructose-biphosphate aldolase (SPV1_05929). As described above, there was a high proportion of proteins with no known function and no COG code assigned (Fig. 1; see also Fig. S1 in the supplemental material). Proteins in this group constituted 12.8% of all the assigned peptide spectra.

FIG 1.

Pie chart of Clusters of Orthologous Groups (COG) functional categories and their relative abundances in the M. ferrooxydans proteomic profile.

TABLE 1.

List of the 25 most abundant proteins identified in the M. ferrooxydans proteomic profile (n = 2)

Gene product	Locus tag	Molecular mass (kDa)	COG	Avg no. of peptides identified (SE)
Hypothetical protein	SPV1_07114	40		850 (250)
ABC transporter	SPV1_07119	37	ABC-type phosphate transport system	587 (27)
Chaperonin, 60-kDa subunit	SPV1_00170	57	Chaperonin GroEL (HSP60 family)	448 (100.5)
Translation elongation factor Tu	SPV1_09894	43	GTPases-translation elongation factors	388 (28.0)
Ribulose-bisphosphate carboxylase	SPV1_04963	51	Ribulose 1,5-bisphosphate carboxylase	278 (84.5)
Hypothetical protein	SPV1_06074	22	Opacity protein, related surface antigens	269 (21)
Elongation factor Ts	SPV1_05739	31	Translation elongation factor Ts	218 (21.5)
FoF1 ATP synthase subunit beta	SPV1_13824	50	FoF1-type ATP synthase, beta subunit	215 (0.5)
Flagellar protein	SPV1_01952	37	Flagellin and related hook associated	207 (0)
FoF1 ATP synthase subunit alpha	SPV1_13814	55	FoF1-type ATP synthase, alpha subunit	207 (54.5)
Transketolase	SPV1_05909	71	Transketolase	185 (7.5)
Translation elongation factor G	SPV1_09899	76	Translation elongation factors (GTPases)	173 (23.5)
Fructose-bisphosphate aldolase	SPV1_05929	37	Fructose/tagatose bisphosphate aldolase	169 (27)
Ribosomal protein L7/L12	SPV1_12170	13	Ribosomal protein L7/L12	167 (38)
DNA-directed RNA polymerase	SPV1_12180	154	DNA-directed RNA polymerase, beta subunit	158 (4.5)
Cyc1 homolog	A37KDRAFT_02147	28	Cytochrome c553	152 (9)
Cytochrome oxidase, cytochrome c subunit	SPV1_07401	33	cbb3-type cytochrome oxidase, cytochrome c subunit	151 (36.5)
Hypothetical protein	SPV1_07396	25		148 (21)
Glutamine synthetase type I	SPV1_00922	45	Glutamine synthetase	147 (0.5)
FoF1-type ATP synthase, subunit b	SPV1_13804	17	FoF1-type ATP synthase, subunit b	147 (16.5)
Molybdopterin oxidoreductase	SPV1_03948	83	Anaerobic dehydrogenases	128 (2)
Molecular chaperone DnaK	SPV1_07796	69	Molecular chaperone	121 (6)
DNA-directed RNA polymerase	SPV1_12175	149	DNA-directed RNA polymerase, beta subunit	121 (3)
Hypothetical protein	SPV1_00467	14		117 (3.5)
Ribosomal protein S8	SPV1_09814	15	Ribosomal protein S8	116 (9.5)

Redox proteins.

A molybdopterin oxidoreductase associated with alternative complex III (ACIII) was identified in PV-1 through preliminary proteomic work (3) and was denoted a “Mob” (Mo-binding) complex, a complex seen in other microaerophilic, neutrophilic FeOB such as Gallionella capsiferriformans, Sideroxydans lithotrophicus (4), and Leptothrix ochracea. In this report, the mob cluster of genes is formally renamed act to be consistent with other reports on ACIII genes in other organisms (34). The act cluster of genes is located upstream of the genes that encode the putative bc₁ complex and includes seven genes: actAB1B2CDEF (see Fig. S2 in the supplemental material). In the proteomic profile reported here, all the gene products belonging to this cluster of genes were identified (see Data Set S1 in the supplemental material), indicating that they were all expressed in PV-1 cells growing on Fe(II). In addition to this, the gene product annotated as “molybdopterin oxidoreductase” (ActB1) is among the top 25 most abundant proteins in the whole profile (Table 1), suggesting that it is functionally relevant to PV-1.

Following the recent discovery of ACIII, there are questions regarding its function even though there have been important advancements from the model organisms Chloroflexus aurantiacus, a filamentous, anoxygenic phototroph, and Rhodothermus marinus, a marine heterotroph (34–36). The actB genes in these microorganisms encode a large protein that contains domains with homology to a molybdopterin-guanine dinucleotide-containing catalytic subunit in the complex of iron-sulfur molybdoenzyme (CISM) family (domain 1) and to an iron-sulfur protein in the CISM family (domain 2). Despite this, it has been proposed that ActB in these organisms does not contain a molybdopterin cofactor (36, 37). Interestingly, in FeOB, actB is split into two genes, actB1 and actB2. Here, we identify the presence of the CX₂CX₃CX₂₇C motif at the N terminus of ActB1 in PV-1 (see Fig. S3 in the supplemental material), which differentiates it from domain 1 of ActB in C. aurantiacus and R. marinus and suggests the presence of an additional [4Fe-4S] cluster and possibly a molybdopterin cofactor (38). Remarkably, these conserved cysteine residues are also seen in ActB1 proteins of other neutrophilic FeOB that also share high levels of synteny with PV-1 in their act cluster of genes. Since domain 1 of ActB and ActB1 are similar to other proteins in the CISM family, a maximum likelihood (ML) phylogenetic tree was produced (Fig. 2) based on an amino acid alignment using proteins from this family, including FdhA (formate dehydrogenase, alpha subunit) and NapA (nitrate reductase, subunit A). The ML phylogenetic tree shows that ActB1 is closely related to domain 1 of ActB, but it forms a distinct clade that is notably represented by microaerophilic, neutrophilic FeOB from different classes of Proteobacteria, suggesting that it is functionally different from ActB proteins in heterotrophic (R. marinus and Myxococcus xanthus) and phototrophic (C. aurantiacus) bacteria. In comparison, the act cluster of genes is not conserved in all Fe-reducing bacteria (e.g., Shewanella spp.). Different Fe-reducing Geobacter species have the act cluster, but they show variability in its composition (i.e., some have actB, and others have actB1 and actB2) and in the order of the genes (34). It is noteworthy that ActB1 and domain 1 of ActB from different species of Geobacter form a distinct clade in the protein phylogenetic tree and seem to be closely related to ActB1 in FeOB.

FIG 2.

Maximum likelihood phylogenetic tree of Act proteins and molybdopterin-binding proteins (i.e., NapA and FdhA). Numbers on branches indicate bootstrap values based on 1,000 replicates. The gray shading highlights both marine and freshwater FeOB. GenBank accession numbers are in parentheses.

It was originally reported (3) that the genome of PV-1 contained a putative ccoNOP operon (SPV1_10291, SPV1_10301, and SPV1_10306) that encodes the subunits CcoN, CcoO, and CcoP of the cytochrome cbb₃ oxidase. Here, we present evidence that some of these proteins are expressed in the proteomic profile although in relatively reduced abundances and without any peptide hits for the catalytic subunit CcoN (SPV1_10306). Surprisingly, the proteome of PV-1 revealed that the most abundant cytochrome cbb₃ oxidase was not associated with the above-described cluster of genes but instead was associated with a previously unidentified cluster that consists of genes encoding the catalytic subunits CcoN (SPV1_07406) and CcoO (membrane-bound cytochrome c) (SPV1_07401) and one cytoplasmic membrane protein with no conserved domain in known families of proteins (SPV1_07396) (Table 1). All components of this putative respiratory complex were identified in the proteomic profile. This putative operon is missing a gene for CcoP, which is a membrane diheme cytochrome c. Interestingly, this gene cluster in PV-1 shares nearly complete synteny with a gene cluster (locus tags Slit_2653 to Slit_2655) in the freshwater, neutrophilic, FeOB Sideroxydans lithotrophicus (4) and with gene clusters of other uncultivated Zetaproteobacteria (39). Furthermore, this operon is also associated with a large cluster of putatively redox protein-encoding genes (Fig. 3). The identification of cytochrome cbb₃ oxidase, which is a member of the class of respiratory heme-copper oxidases, is consistent with the microaerophilic nature of PV-1, since this type of terminal oxidase has a higher affinity for O₂ than the typical complex IV (40). Because the chemical kinetics of Fe(II) oxidation in seawater are sufficiently slow at reduced O₂ concentrations (41, 42), marine FeOB can gain access to the available soluble pool of reduced Fe. Therefore, as a result of microaerobic conditions, the spontaneous chemical oxidation of Fe(II) can be retarded, and PV-1 can compete to oxidize it instead, eventually transferring the electrons from Fe to O₂.

FIG 3.

Gene neighborhood around the Cyc1_PV-1 gene. The region within the dashed box indicates the gap that has been sequenced in this study. The DNA sequence of this gene has been deposited in GenBank under accession number BK009249.

Based on its positioning within the most abundant proteins (Table 1), a cytochrome c₄ is suggested to be one of the most important electron transfer proteins in M. ferrooxydans. This cytochrome c₄ (A37KDRAFT_02147) was identified by using an amino acid sequence database generated from the draft genome of the closely related bacterium M. ferrooxydans M34 (see Data Set S2 in the supplemental material). It was subsequently determined that there is a sequencing gap (∼3 kb) in the permanent draft genome of PV-1 that flanks the cytochrome c₄ gene (SPV1_07306; partial sequence) at bp 269. This gap was subsequently sequenced, and the DNA sequence was deposited in the NCBI GenBank database under accession number KR106296 (99.7% nucleotide sequence identity to A37KDRAFT_02147).

The gene encoding cytochrome c₄ in PV-1 (here and provisionally named the Cyc1_PV-1 gene) is in the same vicinity and downstream of the newly identified cytochrome cbb₃ oxidase gene cluster discussed above (SPV1_07396, SPV1_07401, and SPV1_07406) (Fig. 3). Its amino acid sequence reveals two conserved CXXCH motifs, which suggests that this is a diheme cytochrome c (see Fig. S4 in the supplemental material). A protein signature search using InterProScan revealed that the Cyc1_PV-1 protein is a member of the cytochrome c₄ family of high-redox-potential proteins (InterPro identification number IPR024167) with two separate domains containing a single heme motif each. The domains were conserved, with 31.6% identical residues, adding support to the classification of the Cyc1_PV-1 protein as a cytochrome c₄. Analysis with CELLO and PSORTb indicates that the Cyc1_PV-1 protein is periplasmic and contains a signal peptide. The Cyc1_PV-1 protein is a homolog of the soluble, diheme Cyc1 in Acidithiobacillus ferrooxidans (33% pairwise identity and 54% pairwise similarity) (43). Other neutrophilic FeOB that encode homologs of the Cyc1_PV-1 protein include Sideroxydans lithotrophicus ES-1, Burkholderiales bacterium GJ-E10, uncultivated Gallionella species, and uncultivated Zetaproteobacteria. There is evidence originating from physiological studies on Pseudomonas aeruginosa and Vibrio cholerae that cytochrome c₄, in general, is preferentially used as the electron donor for cytochrome cbb₃ oxidase (44, 45).

Another noteworthy cytochrome c, encoded by A37KDRAFT_01357, was identified in the proteomic profile as an abundant cytochrome c with localization in the outer membrane (see Data Set S2 in the supplemental material). This protein (here and provisionally named the Cyc2_PV-1 protein) is a distant homolog of Cyc2 (corroborated by PSI-BLAST), the iron-oxidizing outer membrane cytochrome c from Acidithiobacillus ferrooxidans (46–48). It was also the fourth most abundant cytochrome c overall in the profile (the Cyc1_PV-1 protein and two cytochrome c subunits of the cbb₃-type oxidases were first, second, and third, respectively). The Cyc2_PV-1 protein was identified by using the M34 proteomic database. There is another sequencing gap (∼3 kb) in the position where the Cyc2_PV-1 gene should be located in the PV-1 genome (Fig. 4), preventing the identification of the expressed protein in PV-1. The gene neighborhood around the Cyc2_PV-1 gene is highly conserved in both PV-1 and M34, providing further evidence that the gene is missing from the assembled PV-1 genome due to the sequencing gap. This gap was subsequently sequenced, and the DNA sequence was deposited in the NCBI database under accession number KR106297 (88% nucleotide sequence identity to A37KDRAFT_01357). There is an additional homolog of Cyc2 in PV-1 (SPV1_10541); however, this cytochrome c is present in relatively low abundance in the proteomic profile.

FIG 4.

Gene neighborhood around the Cyc2_PV-1 gene. The region within the dashed box indicates the gap that has been sequenced in this study. The DNA sequence of this gene has been deposited in GenBank under accession number KR091570.

The Cyc2_PV-1 protein was also identified in the membrane fraction of PV-1 cells (see Data Set S3 in the supplemental material) at >200 kDa, which corresponded to the area where in-gel assays showed both heme peroxidase activity and ferrocyanide oxidation activity (see Fig. S6 and S8 in the supplemental material). An interesting aspect of this cytochrome c is that it shares similarities to an outer membrane cytochrome c, Cyc2, that is a known Fe-oxidizing protein from A. ferrooxidans (46). Jeans et al. (49) previously pointed out, when comparing Cyc2 to an Fe-oxidizing protein from Leptospirillum, Cyt₅₇₂, that the N terminus is highly conserved (41% and 65% pairwise identity and similarity, respectively), while the rest of the sequences are highly divergent (overall 18% and 35% pairwise identity and similarity, respectively). The amino acid alignment (Fig. 5; see also Fig. S5 in the supplemental material for the full alignment) of the Cyc2 and Cyc2_PV-1 proteins reveals that there are both a highly conserved N-terminal region (51% and 71% pairwise identity and similarity, respectively) and a conserved heme motif (CXXCH) between the two proteins (overall 20% and 42% pairwise identity and similarity, respectively). A signal peptide is predicted by both SignalP and Phobius. A predictor for transmembrane beta-barrel proteins (23) indicates that it is possible that the Cyc2_PV-1 protein is an outer membrane protein. There are homologs of the Cyc2_PV-1 protein present in the genomes of both freshwater and marine FeOB, including S. lithotrophicus ES-1, Gallionella capsiferriformans ES-2, Ferriphaselus sp. strain R-1, and cultivated and uncultivated Zetaproteobacteria. BLAST results also indicate matches to endosymbionts in the deep-sea tubeworms Riftia pachyptila and Tevnia jerichonana as well as in deep-subsurface, Fe(II)-rich fluids where Mariprofundus-like organisms were identified (16). Shared similarities to Cyc2, the putative localization to the outer membrane, and positive hemoprotein in-gel identification results make this cytochrome c a candidate for the first electron acceptor from Fe. Results reported recently by Summers et al. (50) showed that PV-1 is able to grow on the surface of electrodes, supporting the hypothesis of the presence of an outer membrane protein that is able to accept electrons (3, 51). We propose that the Cyc2_PV-1 protein is a strong candidate for such a function in the outer membrane of PV-1.

FIG 5.

Amino acid alignment comparing the Cyc2_PV-1 protein to two iron oxidases, Cyc2 and Cyt₅₇₂. The heme-binding motif around residues 12 to 16 is enclosed in a box. Columns with a black background contain identical amino acids in all three sequences. Columns with a gray background contain identical amino acids in only two of the sequences. The alignment was produced in Geneious by using the CLUSTAL aligner. Signal peptides (predicted in the case of the Cyc2_PV-1 protein) were removed. This alignment has been trimmed. The full alignment is shown in Fig. S5 in the supplemental material.

Hemoprotein staining.

Proteins extracted from PV-1 were run on a gel. The nonreduced, unheated crude fraction that was stained with the heme stain (Fig. 6A) revealed two particular regions in the gel that contained peroxidase activity: an area between 25 and 27 kDa with possible multiple bands and an area with a doublet at >200 kDa. The band(s) at 25 to 27 kDa was also seen in a sample that was reduced and heated (Fig. 6B), indicating that a protein(s) with a covalently bound heme prosthetic group(s) may be present in this area. These results further support the LC-MS/MS results identifying an abundant cytochrome c₄, Cyc1_PV-1, in this area of the gel. The theoretical molecular mass of the Cyc1_PV-1 protein is 26.4 kDa, taking into account the removal of the signal peptide in the mature protein and the incorporation of two heme cofactors (600 Da each) covalently bound to the protein, which agrees with the results shown in Fig. 6A and B. A distinctive band with heme peroxidase activity is also seen near the 40-kDa-molecular-mass marker (Fig. 6B; see also Fig. S7 in the supplemental material); the position of this band correlates with the molecular mass of the Cyc2_PV-1 protein, which is calculated to be between 42.1 kDa and 42.7 kDa, if the covalently bound heme cofactor (600 Da) and the predicted signal peptide (different lengths predicted) are taken into account.

FIG 6.

Heme-stained SDS-PAGE gels of crude extracts under nonreducing (A) and reducing (B) conditions. Molecular masses in kilodaltons are indicated on the left. Approximately 2 μg of protein was loaded into each well.

The nonreduced, unheated membrane fraction of PV-1 (see Fig. S6 in the supplemental material) was subjected to hemoprotein staining after SDS-PAGE. In this fraction, each of the doublet bands at >200 kDa contained the Cyc2_PV-1 protein (see Data Set S3 in the supplemental material). Other redox proteins in this region included a cytochrome cbb₃ oxidase subunit and ActC (bottom band); however, Cyc2 was the only redox protein identified in both the bottom and top bands. These doublet bands can also be seen in a crude extract gel that was heme stained (Fig. 6A). An in-gel ferrocyanide oxidation assay under nonreduced, unheated conditions indicated that this doublet is particularly active (see Fig. S8 in the supplemental material). Additionally, there was a region of peroxidase activity between ca. 80 kDa and 150 kDa (see Fig. S6 in the supplemental material). This region contained most of the gene products of the act cluster of genes (including the pentaheme cytochrome c, ActA), with the iron-sulfur protein molybdopterin oxidoreductase (ActB1) being the most abundant protein based on the assigned number of spectra (see Data Set S4 in the supplemental material). All other Act proteins were identified, with the exception of ActE. Cytochrome cbb₃ oxidase subunits were also identified in this region. Since the sample was not reduced, nor was it heated, disulfide bonds would still be present; therefore, it is possible that these proteins migrated in complexes. In other experiments, a reduced membrane fraction of PV-1 was also heme stained after SDS-PAGE (see Fig. S7 in the supplemental material) and indicated the presence of a ca. 42-kDa heme-containing protein that matches the molecular mass of the Cyc2_PV-1 protein. However, this fraction was not analyzed by LC-MS/MS due to the concentrated presence of incompatible detergents in the sample {e.g., 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS) in ReadyPrep reagent 3}.

Phosphate transport.

It has been well documented in Escherichia coli, Pseudomonas aeruginosa, and Bacillus subtilis that expression of the phosphate-specific transport operon (pst) is induced when inorganic phosphate (P_i) levels become low (52–54). The Gram-negative bacteria E. coli and P. aeruginosa both have a pstSCAB-phoU operon, while the Gram-positive bacterium B. subtilis lacks phoU. The most abundant protein in the proteomic profile of PV-1 is encoded by a gene with the locus tag SPV1_07114, which shares similarities to the polyphosphate- and P_i-selective porins O and P, respectively. SPV1_07114 is located 83 bp from a cluster of pst genes (Fig. 7). This gene does not have homologs in the pst operon of E. coli, P. aeruginosa, or B. subtilis. Thus, it is not clear what function this gene might have or if it is related to the uptake of P_i. Transcriptomic results indicate that the intergenic region of SPV1_07114 and pstS is transcribed, suggesting that the former gene is part of an operon with pstS and possibly with the other genes (see Fig. S9 in the supplemental material). This cluster of genes is conserved in other Zetaproteobacteria, including cultivated Mariprofundus sp. strain EKF-M39 and uncultivated Zetaproteobacteria from the Lō'ihi Seamount (39). Amino acid sequence analysis with CELLO, PSORTb, and PRED-TMBB predicted that this protein is located in the outer membrane. A protein signature search on InterProScan matched a signal peptide and a porin domain (InterPro identification number IPR023614), adding support to this prediction. The second most abundant protein in PV-1 is an ABC transporter P_i-binding protein (SPV1_07119) with predicted localization in the periplasm. This protein is classified as belonging to the PstS family of ABC transporters (InterPro identification number IPR005673). A BLAST search with Uniprot reveals that it has homology to other PstS proteins. The other 4 genes downstream of pstS have domains that identify them in the following order: pstC, pstA, pstB, and phoU. All gene products from this cluster of pst genes were identified in the proteomic profile, with the exception of PstA, which is a permease that is a homolog of PstC.

FIG 7.

Cluster of phosphate-specific transport (pst) genes in M. ferrooxydans and comparison to known pst operons in E. coli and B. subtilis. Genes with the same predicted function have the same color or pattern.

It is well known that iron oxyhydroxides have a strong affinity for the binding and sorption of P_i, and this can result in the removal of P_i from solution (55). There is evidence that active iron oxidation results in greater P_i sorption than with simply adding preformed Fe(III) oxides to a P_i-containing medium (56). Furthermore, biogenic oxides produced by FeOB can play an important role in controlling P_i dynamics, as was shown for the freshwater Fe oxidizer Leptothrix ochracea (57). Therefore, it is plausible that the pool of soluble P_i (initial concentration of 279 μM) in the culture medium used to grow PV-1 for these experiments became depleted as the organisms accumulated more and more iron oxides. To confirm this, the amount of P_i was monitored during a 3-day period. Figure S10 in the supplemental material shows the depletion of P_i over time in control samples that had Fe added on a daily basis. This trend was even more pronounced when PV-1 cultures were actively growing at adequate P_i concentrations (undiluted [i.e., 1:1 P]; 279 μM) and producing more flocculent material (i.e., twisted stalks) containing iron oxyhydroxides. Together, these data support that the high abundances of SPV1_07114 and PstS may indicate that PV-1 was undergoing P limitation at the time of harvest, during the late log phase.

Carbon fixation.

PV-1 has two different forms of ribulose-biphosphate carboxylase (RubisCO), the enzyme involved in the first step of the Calvin-Benson-Bassham (CBB) cycle: form IAq and form II (3). The proteomic results indicate that form II is one of the most abundant proteins expressed, which is consistent with the autotrophic lifestyle of PV-1 (Table 1). In contrast to this result, form IAq was not identified in the proteomic profile. The CbbO and CbbQ activation proteins associated with form II RubisCO are present in the profile but at lower abundances (see Data Set S1 in the supplemental material). In general, form II RubisCO is differentiated from form IAq by having higher turnover rates and a lower affinity for carbon dioxide (CO₂), and by being functional at lower concentrations of O₂ (58). Therefore, form II is more prevalent in suboxic environments with high CO₂ concentrations (>1.5%). The large bottles used for culturing of PV-1 in this study had a concentration of 30% CO₂ (gas) and <2% O₂ (gas) in the headspace, which may explain the preference for form II RubisCO. Similar environments with high CO₂ and low O₂ concentrations are seen in the Lō'ihi Seamount, where PV-1 was isolated. The hydrothermal vents at the summit of the Lō'ihi Seamount are located in an O₂-minimum zone, and the vent fluids are the source of the CO₂ and Fe(II) that support the luxurious growth of mats of FeOB (59). It is possible that form II is predominant in FeOB that colonize these CO₂-rich deep-sea environments or that the form I enzyme is expressed only during times when O₂ concentrations fluctuate toward higher levels. Other abundant proteins involved in carbon fixation were identified, including transketolase, phosphoglycerate kinase, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), phosphoribulokinase, and fructose 1,6-bisphosphate aldolase, indicating that PV-1 has a fully functional CBB cycle.

Stalk formation.

The formation of a helical stalk is a defining morphological characteristic of PV-1. It has been proposed that the stalk serves at least two functions, in excreting Fe oxides from the cell surface and in helping the cell maintain its position in the dynamic gradients of O₂ and Fe(II) that it requires for growth (6, 8). Compositionally, the stalk is believed to incorporate an organic component, possibly a polysaccharide, that forms an integral part of the biomineral matrix (8, 60). The molecular mechanism for stalk formation is not yet known; however, there are two identified groups of genes, SPV1_00467 to SPV1_00517 and SPV1_07411 to SPV1_07616, that are particularly rich in permeases, glycosyltransferases, and capsular polysaccharide synthesis enzymes (60). Some gene products from these two groups of genes are present in the proteomic profile (see Data Set S1 in the supplemental material); thus, these proteins could be good candidates for subsequent molecular approaches to study stalk formation. It is also possible that stalk formation is a more complex process that directly involves ETC proteins, since Fe oxidation produces the insoluble Fe oxyhydroxides present in the stalks. However, the localization (e.g., outer membrane, periplasm, or inner membrane) of the process of Fe oxidation in PV-1 is still not known.

Model of Fe oxidation and the electron transport chain.

It has been shown that PV-1 contains most of the oxidized Fe in the stalks and not in the cell (8). The localization of Fe(II) oxidation on the cell surface, as opposed to the periplasm, inner membrane, or cytoplasm, may be beneficial to the cell by preventing encrustation due to the constant production of insoluble iron oxyhydroxides; however, this process requires extracellular electron transport (EET) from the cell exterior to the internal ETC components in the periplasm and inner membrane. The schematic diagram shown in Fig. 8 presents an emergent picture of EET in PV-1 based on this new proteomic data set. This scheme generally follows a scheme that was presented previously but based primarily on genomic data (3, 51). The proteomic profile presented here confirms that some of the genes hypothesized to be important in Fe oxidation are likely expressed and that their associated proteins are relatively abundant. This profile also identifies novel proteins that could play an important role, such as Cyc1_PV-1 and Cyc2_PV-1. Electrons removed from Fe(II) at the cell surface are donated by an outer membrane cytochrome c (Cyc2_PV-1) → periplasmic cytochrome c₄ (Cyc1_PV-1) → cbb₃-type cytochrome oxidase → O₂, generating a proton motive force in the process that can drive the production of ATP through ATP synthase. An “uphill” pathway of electron transport is also a possibility, with the electron from the Cyc1_PV-1 protein being donated to the cytochrome bc₁ complex (complex III) (SPV1_03843 to SPV1_03873) and then transported to the quinone pool and ultimately accepted by NADH dehydrogenase to produce reducing power in the form of NADH. However, peptides for this bc₁ complex are present at relatively low levels in this proteomic profile. Alternatively, it is hypothesized that this function could potentially be performed by ACIII, which has been shown to be an analog of the bc₁ complex (34–36) and is represented in the profile by a large number of peptides. A recent report (61) presents a remarkably similar model, including reverse electron transport via ACIII, in an uncultivated nonphotosynthetic member of the Chromatiaceae that is capable of growing on biocathodes.

FIG 8.

Proposed electron transport chain of M. ferrooxydans based on proteomic analysis. OM, outer membrane; IM, inner membrane; Act, alternative complex III; Q, quinone pool; cyt cbb₃, cytochrome cbb₃ oxidase. The asterisk indicates that other soluble components may be involved in the electron transfer between the Cyc2_PV-1 and Cyc1_PV-1 proteins.

In summary, evaluation of protein expression in an aerobic, chemolithoautotrophic FeOB was performed. The analysis of the proteome of an important marine FeOB, M. ferrooxydans, helped confirm predictions made through comparative genomics (3, 4) but also revealed new insights into its physiology that were not previously identified. The results presented here motivate further research into the characterization of the proteins identified (e.g., Cyc1_PV-1 and Cyc2_PV-1) and the development of genetic markers for quantifying this process in environmental samples. More rigorous experiments need to specifically confirm the interactions and functions of the different complexes in the ETC. The data set presented here provides a starting point for such work. The quest for the isolation of aerobic, chemolithoautotrophic FeOB with alternate metabolisms continues to date. The isolation of such organisms would allow comparative proteomics or transcriptomics and would help pinpoint the specific genes that are essential for neutrophilic Fe oxidation.

We dedicate this article to the memory of the late Katrina J. Edwards, a great explorer of the oceans, mentor, and pioneer geomicrobiologist.