Photoferrotrophs Produce a PioAB Electron Conduit for Extracellular Electron Uptake

Some anoxygenic phototrophs use soluble iron, insoluble iron minerals (such as rust), or their proxies (poised electrodes) as electron donors for photosynthesis. However, the underlying electron uptake mechanisms are not well established. Here, we show that these phototrophs use a protein complex made of an outer membrane porin and a periplasmic decaheme cytochrome (electron transfer protein) to harvest electrons from both soluble iron and poised electrodes. This complex has two unique characteristics: (i) it lacks an extracellular cytochrome c, and (ii) the periplasmic decaheme cytochrome c undergoes proteolytic cleavage to produce a functional electron transfer protein. These characteristics are conserved in phototrophs harboring homologous proteins.

marine productivity (9) and a potential process responsible for the deposition of the archean banded iron formations (BIFs) (1,(10)(11)(12)(13). Photoferrotrophs can also use electrons from a poised electrode to fix CO 2 with light, a process called phototrophic extracellular electron uptake (EEU) (14,15). Microbes capable of phototrophic EEU are good candidates for microbial electrosynthesis, a process in which microbes use electricity to produce biocommodities from CO 2 (16)(17)(18). Despite continued interest in microbes capable of photoferrotrophy and phototrophic EEU, the mechanisms underlying these processes are poorly understood.
The bacterial outer envelope is nonconductive to electrons and is impermeable to insoluble iron minerals/electrodes (19,20). Therefore, the ability of phototrophs to use the extracellular electron donors likely involves an extracellular electron uptake (EEU) process. Studies in the nonphototroph model bacterium Shewanella oneidensis suggest that the mechanism of electron transfer across the outer membrane (OM) employs a porin-cytochrome complex (21). The Shewanella complex is comprised of a porin (MtrB) and two decaheme cytochrome c proteins (MtrA and MtrC from periplasmic and extracellular sides, respectively) as an electron conduit. Deletion of the extracellular decaheme cytochrome c component of the conduit (MtrC and its paralogues, OmcA and MtrF) abrogates Shewanella's ability to reduce Fe(III) or transfer electrons to electrodes (22)(23)(24). Recently, it has been shown that the Shewanella MtrCAB conduit may also be employed for electron uptake (electron transfer in the opposite direction from its native function) from a cathode to an intracellular reduction reaction (18,25). In contrast to oxidation of insoluble iron/electrode, soluble iron oxidation could occur either extracellularly or in the periplasm of Gram-negative phototrophs (26)(27)(28). Evidence to date suggests that the oxidation of soluble iron in Rhodopseudomonas palustris TIE-1 (6) and Rhodobacter sp. strain SW2 (29) involves periplasmic iron oxidoreductases (PioA and FoxE, respectively). However, the oxidation of soluble iron could be extracellular because the product of this process is insoluble iron, which can be toxic if produced in the periplasm (10,13,30,31). Genetic studies in R. palustris TIE-1, the only genetically tractable phototroph (32) that can perform both photoferrotrophy and phototrophic EEU (6,7,15), identified the pioABC operon as required for both of these processes (6,15). The operon encodes PioA, a decaheme cytochrome c (cyt c), PioB, an outer membrane (OM) porin, and PioC, a high-potential iron-sulfur protein (HiPIP). A deletion mutant lacking pioABC cannot perform photoferrotrophy (6) and has a partial defect in phototrophic EEU (15). Beyond these genetic studies, only the iron-sulfur protein PioC has been biochemically characterized as a periplasmic electron transfer protein and proposed to shuttle electrons from PioA to the photosynthetic reaction center (33,34). However, where Pio proteins oxidize Fe(II) during photoferrotrophy and how they transfer electrons across the outer membrane from poised electrodes during phototrophic EEU are unknown.
Here, we show that PioA undergoes novel postsecretory proteolysis of its N terminus to produce a decaheme-attached PioA (holo-PioA C , where PioA C represents the C terminus of PioA). The holo-PioA C is an iron oxidoreductase that forms a membrane-associated protein complex with PioB. The holo-PioA C B complex acts as an electron conduit in TIE-1 to accept electrons from the extracellular oxidation of both Fe(II) and from the poised electrodes (that mimic insoluble iron minerals). Importantly, this process employs a single decaheme cyt c (holo-PioA C ) and a porin (PioB), with no apparent extracellular electron transfer protein. Therefore, the PioAB complex is distinct from other porin-cytochrome c systems found in nonphototrophic bacteria. The postsecretory proteolysis of the N-terminal extension of PioA-like homologs to produce a functional holo-PioA C and the formation of holo-PioA C B electron conduit are conserved in phototrophs such as Rhodomicrobium vannielii and Rhodomicrobium udaipurense. Together, these results suggest that the PioAB system can serve as a molecular signature for photoferrotrophy and phototrophic EEU.

RESULTS AND DISCUSSION
Production and purification of recombinant PioA from Escherichia coli. The pioA gene (Rpal_0817) encodes a 540-amino-acid (aa) decaheme cyt c with a canonical signal peptide (SP; aa 1 to 40), a 200-residue N terminus, and the C terminus (Fig. 1A). The C terminus contains a hydrophobic region (HR; aa 241 to 263) followed by 10 sites for c-type heme (CXXCH) attachment (Fig. 1A). PioA has a large (ϳ200 aa) N terminal region not present in previously characterized decaheme cyt c homologs such as MtrA and MtoA (see Fig. S1 in the supplemental material). This N-terminal extension is conserved in PioA homologs annotated in the genomes of different phototrophic bacteria and does not harbor any clear protein domains (Fig. S2). Previous translational fusion studies suggest that a full-length PioA with its N terminus is produced in TIE-1 (35). Here, we wanted to investigate the role of the N terminal extension in these decaheme homologs by studying PioA. Although PioA expression in TIE-1 is upregulated only under photoferrotrophy, its slow growth (doubling time of 80 Ϯ 10 h) and low biomass yields under this condition make it difficult to purify sufficient protein for biochemical analysis. We heterologously expressed PioA with a C-terminal His tag in the E. coli (Δccm) RK103 strain carrying the ccmABCDEFGH genes under an inducible promoter to ensure heme attachment (36). For correct localization of the protein in E. coli's periplasm, we replaced the PioA signal peptide with a validated signal peptide of the cytochrome c 4 gene (36). We confirmed production of the predicted full-length apo-PioA (ϳ54 kDa) in total E. coli lysate by immunoblotting with antibodies specific to the N-terminal 200-aa region (anti-PioA N ) and the C terminus (anti-PioA C ) of PioA (Fig. 1B). However, an ϳ34-kDa heme-containing polypeptide was observed upon enrichment using hexahistidine affinity purification (Fig. 1C). Heme stains were used to detect covalently bound (c-type) heme in the protein (37). The polypeptide was immunodetected with anti-PioA C but not with anti-PioA N antibodies (Fig. S3A). These results suggest that the 34-kDa protein is PioA after removal of the N-terminal 200-aa region. Here, the recombinant holo-PioA is termed holo-PioA C r . TIE-1 produces a 34-kDa holo-PioA C that exists as a free periplasmic protein and in a complex with PioB. To investigate the biological relevance of PioA N-terminal processing in the native photosynthetic host, we isolated soluble and membrane fractions of photoautotrophically grown wild-type (WT) and ΔpioABC mutant strains of TIE-1. We observed two heme-containing polypeptides: an ϳ34-kDa band in the soluble fraction and a larger (Ͼ170-kDa) band in the membrane fraction of WT TIE-1 (Fig. 1D). Both bands reacted with anti-PioA C antibodies ( Fig. 1E) but not with anti-PioA N antibodies (Fig. 1F). Mass spectrometry analysis of these bands identified only the C-terminal peptides of PioA, consistent with processing of the N-terminal region in these heme-attached forms of PioA. These results suggest that TIE-1 produces only the 34-kDa heme-attached C terminus of PioA (here termed holo-PioA C ). Mass spectrometry analysis of the Ͼ170-kDa protein band identified it as PioA C B complex (Fig. S4A). This band also demonstrates heat modifiability (Fig. S4B), a feature reported for bacterial porins (38)(39)(40)(41)(42). Indeed, the holo-PioA C B migrates as an ϳ120-kDa band without heat treatment and shifts to Ͼ170 kDa when the sample is heated to 90°C for 3 min (Fig. S4B). The 120-kDa holo-PioA C B band (unheated) likely represents the observed size of holo-PioA C (ϳ34 kDa) plus the expected size of PioB (ϳ87 kDa), further indicating that the complex is composed of holo-PioA C (that lacks the N-terminal region) and PioB. Overall, these results suggest that TIE-1 produces a 34-kDa holo-PioA C that exists in two forms, a free soluble periplasmic protein and a membrane-associated protein in complex with PioB (holo-PioA C B).
Holo-PioA C produced natively in TIE-1 is the same size as holo-PioA C r produced heterologously in E. coli (as determined by SDS-PAGE). A similarly sized heme-stainable PioA (ϳ40 kDa) was previously observed from the soluble fraction of TIE-1 by Jiao and Newman (6). The size observed for holo-PioA C is comparable to the size of the previously characterized decaheme homologs MtrA and MtoA (21,31). Our biochemical results suggest that holo-PioA C is the functional component of a PioA C B complex in TIE-1. However, holo-PioA C lacks the N-terminal 200-aa region of the predicted PioA protein. The question remains as to how TIE-1 produces holo-PioA C . There are two possibilities: (i) full-length PioA is produced and processed to produce holo-PioA C , or (ii) an internal methionine closer to the first heme attachment site is the functional start codon. Interestingly, the internal hydrophobic region (HR) of the C terminus of PioA starts with a methionine residue (Met241), and Met241 is the only internal methionine from which a putative Sec signal peptide is predicted based on PredTat (43). It is therefore possible that Met241 is used as a start codon for PioA, with the Sec signal in the HR serving as a periplasmic signal peptide and with the resulting protein being ϳ34 kDa. There is, however, no observable canonical ribosome binding site (RBS) upstream of Met241. To directly test this possibility, we expressed the pioA gene encoding only the C terminus of PioA from the HR (241 to 540 aa; HR-PioA C ) with a C-terminal 6ϫHis tag in the ΔpioA mutant using a pRhokS-2 plasmid under a constitutive promoter, PaphII, along with a strong RBS (Fig. 1G) (44). We observed (by PioA C Western blotting [WB]) the expected size of HR-PioA C upon enrichment using hexahistidine affinity purification, suggesting that this version of the protein can be produced artificially in TIE-1 (Fig. 1H, WB:PioA C ). However, HR-PioA C was not heme attached as determined by heme staining (Fig. 1H, PioA C ). Because c-heme maturation will depend on the periplasmic transport of HR-PioA C , the above result suggests that the Sec signal within the HR is not a functional periplasmic signal and that Met241 does not serve as an alternative start codon for PioA. These results demonstrate the importance of the N terminus (signal peptide and 200-aa region) in the synthesis of holo-PioA C in TIE-1.
N-terminal processing of PioA is required for photoferrotrophy. To gain insights into the biological role of N-terminal processing, we engineered a series of N-terminal-PioA chromosomal deletion mutants in TIE-1 ( Fig. 2A). We tested the ability of these mutants to perform Fe(II) oxidation. The Δ240 mutant lacks the complete N terminal region that includes both the predicted signal peptide and the 200-aa region, the Δ43 mutant lacks only the signal peptide, and the Δ200 mutant lacks only the 200-aa region but contains the intact signal peptide. Photoautotrophically grown TIE-1 cells with hydrogen were used for cell suspension assays. We observed that the Δ200 mutant oxidized Fe(II) both in the cell suspension assays and during photoferrotrophy ( Fig. 2B and C, green line, and Fig. S5). In contrast, the Δ240 and Δ43 mutants, which lack the signal peptide that is essential for periplasmic export, were unable to oxidize Fe(II) or perform photoferrotrophy ( Fig. 2B and C, red and purple lines, respectively). Furthermore, only the Δ200 construct rescued the phenotype of a ΔpioA mutant when expressed in trans from a plasmid (Fig. 2D). The ability of the Δ200 mutant to oxidize Fe(II) suggests that holo-PioA C is a functional iron oxidase in TIE-1. It also demonstrates that the N-terminal 200-aa region of PioA is not directly involved in the ability of PioA to oxidize Fe(II).
Although the Δ200 mutant was able to grow via photoferrotrophy, it had an extended lag phase compared to that of the WT (Fig. 2C). We examined whether this lag is due to the difference in PioA concentrations between these two strains. To approximate photoferrotrophy and obtain sufficient cells for total protein extraction, strains grown photoautotrophically with hydrogen (optical density at 660 nm [OD 660, ] of 1.0) were exposed to 3 mM Fe(II) and harvested after 22 h when all Fe(II) was oxidized (Fig. 2E). We observed a smaller amount of both free holo-PioA C and holo-PioA C B complex in the Δ200 mutant than in the WT ( Fig. 2F and G). We also assessed the relative transcript levels of pioA under these conditions and observed a slight increase in pioA transcripts in the Δ200 mutant compared to the level in the WT (Fig. 2H). This result rules out the possibility that the lower level of free holo-PioA C and holo-PioA C B complex in the Δ200 mutant could be due to lower protein expression in the mutant. Therefore, we propose that the N-terminal 200-aa region of PioA plays a role in maintaining the concentration of holo-PioA C in TIE-1 at wild-type levels. Edman degradation analyses of holo-PioA C purified from TIE-1 identified Ala243 as the N-terminal residue of the processed protein (Fig. 3). Cleavage at this site would produce a protein with a predicted molecular weight of 34 kDa, as calculated for the holo-PioA C .
N-terminally processed PioA is an iron oxidase. The fact that holo-PioA C is required for Fe(II) oxidation in TIE-1 suggests that it is an iron oxidoreductase. Thus, we affinity purified holo-PioA C r and assayed its iron oxidation capacity by UV-visible light (UV-Vis) spectroscopy under anaerobic conditions. The UV-Vis spectra of holo-PioA C r show the spectral signatures of typical c-type cytochromes ( Fig. 4A and Fig. S3B).
Covalently bound heme in holo-PioA C r was determined by observing a 550-nm ␣ peak in the pyridine hemochrome assay (Fig. 4A, inset) (45,46). The dithionite-reduced holo-PioA C r (Fig. 4A, purple) was reoxidized after addition of Fe(III) chloride solution, as indicated by a shift of the Soret peak back from 418 to 408 nm and disappearance of the ␣ and ␤ peaks at 551 and 523 nm, respectively (Fig. 4A, teal blue). This spectral characteristic of holo-PioA C r is similar to that of the reduced-MtrCAB complex after addition of Fe(III) citrate (21). As reported for the decaheme cyt c MtoA (31), the reduction of holo-PioA C r with Fe(II) chloride was pH dependent and only detected at a basic pH of 9 to 10 (Fig. 4B). These results demonstrate that holo-PioA C r can donate electrons to Fe(III) and accept electrons from Fe(II). Thus, the N-terminal 200-aa region Photoferrotrophs Use the PioAB Electron Conduit for EEU ® localization of PioA to the membrane. Because the ΔpioB mutant contains only soluble holo-PioA C , it serves as an ideal construct to investigate the role of periplasmic soluble holo-PioA C in Fe(II) oxidation. We found that ΔpioB cannot oxidize Fe(II) in cell suspension assays (Fig. 5B). Because periplasmic holo-PioA C is produced in the ΔpioB mutant, this result indicates that Fe(II) oxidation is unlikely to be carried out by soluble holo-PioA C in the periplasm. Another explanation for this result might be that PioB acts as an Fe(II) import porin that makes Fe(II) available to periplasmic PioA for iron , and, therefore, the two proteins are inserted into the OM as a MtrAB complex (21,47). To this end, we investigated the fate of PioB in the absence of PioA by using a ΔpioA mutant. Similar to observations of MtrAB, using immunoblotting with an antibody specific to PioB, we observed the presence of PioB in the soluble fraction but not in the membrane fraction of the ΔpioA mutant (Fig. 5C). This result suggests that the holo-PioA C and PioB localize together as a complex in the OM, and, therefore, PioB is likely not present as a discrete porin in TIE-1. This result is further supported by the fact that we could detect only the holo-PioA C B complex and not free PioB in the membrane fraction of WT TIE-1 (Fig. 5C). Together, these results suggest that it is unlikely that soluble holo-PioA C performs periplasmic Fe(II) oxidation and supports the hypothesis that the holo-PioA C B complex catalyzes Fe(II) oxidation extracellularly. The holo-PioA C B complex is required for phototrophic EEU in TIE-1. In addition to Fe(II), TIE-1 can oxidize insoluble iron minerals, such as hematite (7), and electrodes poised at the potential of insoluble iron minerals (15). Because these insoluble electron donors cannot cross the bacterial envelope, they are most likely oxidized by TIE-1 extracellularly via the holo-PioA C B complex. Poised electrodes in bulk reactors have been used as proxies of natural interactions between microbes and minerals such as iron oxides (15,48,49). However, in such bulk reactors different factors, such as extracellular enzymes (50), presence of planktonic cells, mediators, and abiotic reactions, could affect the electrochemical signals (15,51). Therefore, we examined the involvement of the holo-PioA C B complex in EEU using a microfluidic bioelectrochemical cell (-BEC) developed in our laboratory (14). In the -BEC, an electrode poised at ϩ100 mV versus a standard hydrogen electrode (SHE) was used to mimic the redox potential of insoluble iron minerals (15). We monitored and compared the lightdependent electron uptake ability of WT TIE-1 and mutants. Intermediate electron uptake by the Δ200 mutant was observed compared to that of the WT TIE-1 (Fig. 6A to C, green and blue, respectively; Table S1). This corroborates the observation that the Δ200 mutant contains a smaller amount of holo-PioA C B complex (Fig. 2G). We did not detect electron uptake by either the ΔpioABC or ΔpioA mutant. Similar to findings with respect to Fe(II) oxidation, ΔpioB mutant could not perform phototrophic EEU (Fig. 6A to C, light brown). This result shows that the holo-PioA C B complex, but not free periplasmic holo-PioA C , is responsible for phototrophic EEU in TIE-1. Phototrophic EEU has been studied in bulk bioelectrochemical systems (BESs) (15) and in -BECs (14). In Photoferrotrophs Use the PioAB Electron Conduit for EEU ® BESs, both planktonic cells (70% electron uptake) and biofilm-attached cells (30% electron uptake) contribute to phototrophic EEU (14,15). The -BEC collects data only from biofilm-attached cells. Therefore, our results from the -BEC suggest that in biofilm-attached cells, the PioABC system is essential for phototrophic EEU. This supports previous results from bulk BESs where the ΔpioABC mutant lost 30% electron uptake ability (15). Overall, our results indicate that the holo-PioA C B complex serves as an electron conduit in TIE-1 to take up electrons from poised electrodes that mimic insoluble iron minerals. Postsecretory processing of PioA and holo-PioA C B complex formation is a conserved trait in other phototrophs. Homologs of PioA/MtrA and PioB/MtrB occur as operons in the genomes of many proteobacteria (15,21). Sequence alignment of decaheme homologs separates them out into two groups, one without the N-terminal extension (MtrA-like) and other with the N-terminal extension (PioA-like) (Fig. S2B). Additionally, phylogenetic analysis clusters bacteria with PioA-like homologs in a distinct clade (Fig. 7A, highlighted clade). Interestingly, this clade includes phototrophic bacteria known to be photoferrotrophs such as TIE-1 (32) and Rhodomicrobium vannielii (52). Although the N-terminal extensions vary in length and are not conserved at the level of amino acid sequence, the presence of a larger N terminal region and an internal  hydrophobic region (HR) upstream of the first heme-binding motif are conserved features of PioA-like homologs (Fig. S2B). Our data from TIE-1 has established that the N-terminal region of PioA is important to produce a functional iron oxidoreductase (holo-PioA C ) and, hence, the holo-PioA C B complex. To investigate the functional parallels of holo-PioA synthesis and PioAB complex formation in phototrophs, we studied Rhodomicrobium vannielii and Rhodomicrobium udaipurense. Both of these organisms have PioABC homologs. R. vannielii can perform phototrophic Fe(II) oxidation (52) while the ability of R. udaipurense to oxidize Fe(II) is unknown. The ability of R. vannielii and R. udaipurense to perform phototrophic EEU is also unknown. First, we studied the ability of R. vannielii DSM 162 (53) and R. udaipurense JA643 (54) to oxidize Fe(II) and perform phototrophic EEU. Both R. vannielii DSM 162 and R. udaipurense JA643 oxidized Fe(II) in cell suspension assays (Fig. 7B, chartreuse green and teal blue, respectively). We also observed that these phototrophs, similar to TIE-1, can perform phototrophic EEU from an electrode poised at ϩ100 mV versus SHE in a -BEC reactor ( Fig. 7C and Table S1) (14). Photoautotrophically grown R. vannielii DSM 162 and R. udaipurense JA643 with hydrogen were used for cell suspension assays, and their ability to oxidize Fe(II) suggests that the iron oxidase (PioA homologs) is produced under this growth condition. To investigate whether these phototrophs, like TIE-1, also synthesize holo-PioA C and holo-PioA C B complex, cell fractions of photoautotrophically grown R. vannielii DSM 162 and R. udaipurense JA643 were analyzed. Heme staining of the soluble and membrane fractions from these bacteria showed band patterns similar to those of TIE-1. We observed two heme-stainable bands, an ϳ34-kDa band in the soluble fraction and a Ͼ170-kDa band in the membrane fraction ( Fig. 7D and E), for both phototrophs. The Ͼ170-kDa bands from R. vannielii DSM 162 and R. udaipurense JA643 were confirmed to contain PioA and PioB homologs by mass spectrometry analysis (Fig. S6). Interestingly, these bacteria also contain a smaller holo-PioA (ϳ34 kDa) in the soluble fraction than expected (ϳ48 kDa) from their annotated sequences (Fig. 7D). Furthermore, only the C-terminal peptides of the PioA homologs were detected by the mass spectrometry of these heme-stainable PioAB complexes (Ͼ170 kDa). Together, these results suggest that the PioA-like homologs in R. vannielii DSM 162 and R. udaipurense JA643, similar to the PioA of TIE-1, undergo postsecretory proteolysis of the N-terminal region to produce a smaller holo-PioA C .
A holo-PioA C B complex as the cornerstone for photoferrotrophy and phototrophic EEU. Our biochemical and genetic studies show that the holo-PioA C B complex is responsible for extracellular oxidation and electron uptake from both soluble Fe(II) and poised electrodes (that mimic insoluble iron minerals). The molecular mechanism of electron transfer by a porin and a single decaheme protein, as in the holo-PioA C B complex in TIE-1, is not well understood. To investigate whether the dimensions of predicted models of holo-PioA C and PioB could support the formation of a holo-PioA C B complex to allow EEU to occur across the OM, we performed in silico analysis of PioA C and PioB using RaptorX, a web-based server for protein structure prediction (55). The predicted structure of holo-PioA C has the dimensions of ϳ100 by 40 by 30 Å (Fig. S7A and B), and the predicted structure of PioB has an estimated pore diameter of ϳ30 to 40 Å (Fig. S7C and D). Although cryo-electron microscopy (cryo-EM) or crystal structures will be necessary to validate the topology and structure of PioA C B, these in silico structure predictions suggest a holo-PioA C B model ( Fig. 8A and Fig. S7E) where holo-PioA C inserts through the entire length of PioB and supports EEU across the OM (ϳ40 to 50 Å in width) of TIE-1.
Conclusions. The results support a model (Fig. 8B) for the synthesis and function of key proteins involved in photoferrotrophy and phototrophic EEU whereby the cytoplasmic PioA precursor (540 aa) is exported to the periplasm via the Sec pathway with the concomitant removal of its signal peptide (1 to 40 aa) (56). The periplasmic apo-PioA (500 aa) is then proteolytically processed at the Ala243 site by a periplasmic/ membrane protease to facilitate heme attachment and produce holo-PioA C (297 aa, 34 kDa). Holo-PioA C exists in two forms, in a free periplasmic form and in a complex with PioB. The free periplasmic form of holo-PioA C is required for the stability and incorporation of PioB in the OM. Based on the dependency of MtrB on MtrA for its OM localization (21,47) and the discovery of MtrB with an opposite orientation in the OM (57), it was proposed that the folding and insertion of MtrB to the OM follow a different pathway than a typical ␤-barrel OM porin in Shewanella (57). Here, we propose that the orientation of PioB ( Fig. 8 and Fig. S7) and its folding and insertion to the OM in TIE-1 are similar to those of MtrB. The holo-PioA C B complex catalyzes electron uptake from both extracellular electron donors such as Fe(II) and poised electrodes (that mimic insoluble iron minerals). Subsequently, electrons from the holo-PioA C B complex are transferred to the photosynthetic reaction center in the inner cytoplasmic membrane, most likely via PioC (periplasmic soluble iron-sulfur protein), as previously proposed (33,34).
Although the 500-residue apo-PioA (54 kDa) is most likely secreted into the periplasm, we did not detect this apo-PioA in TIE-1. This lack of detection could be due to the rapid proteolysis of apo-PioA in TIE-1. In E. coli, we observed both the 54-kDa apo-PioA and the 34-kDa holo-PioA C r ( Fig. 1B and C), detecting heme attachment only in the 34-kDa C-terminal PioA. These results suggest that the proteolysis of apo-PioA occurs before heme maturation in both TIE-1 (and other photoferrotrophs) and E. coli. We could not detect a PioAB complex containing full-length apo-PioA; so heme maturation of PioA is likely important for formation of the PioAB complex. Our data demonstrate that the deletion of the N-terminal 200-aa region of apo-PioA does not abrogate production of holo-PioA C in TIE-1. However, it causes a decrease in the cellular concentration of holo-PioA C and holo-PioA C B complex, likely leading to a lag in photoferrotrophy (Fig. 2C) and an overall reduction in phototrophic EEU (Fig. 6A to C). The 200-aa region of the N terminus of PioA may play a role in maintaining the concentration of holo-PioA C in TIE-1 at wild-type levels. This region could control the C-terminal PioA secretion rate and/or access to the cyt c maturation pathway (i.e., CcmF/H synthetase) (58,59). The exact function of the N-terminal region will require further investigation. Although the ΔpioB mutant can produce soluble holo-PioA C , it did not oxidize Fe(II) (Fig. 5A and B) or perform phototrophic EEU (Fig. 6A to C). This indicates that both of these processes in TIE-1 are catalyzed extracellularly by holo-PioA C B complex (33,60). The model describes the postsecretory steps in the synthesis of holo-PioA C and demonstrates an electron uptake mechanism via PioA C B complex that enables R. palustris TIE-1 to utilize both Fe(II) and poised electrodes that mimic insoluble iron minerals as electron donors for photosynthesis. PioC, a periplasmic high-potential iron-sulfur (Fe-S) protein, shuttles electrons from the inner face of the holo-PioA C B complex to photosynthetic reaction center. In silico-predicted models in panel A and in Fig. S7 in the supplemental material were used to demonstrate folded forms of holo-PioA C , PioB, and the holo-PioA C B complex in this model. Solid arrows, steps in protein transport, processing, and complex formation; dashed arrows, the path of electron transfer; OM, outer membrane; ICM, inner cytoplasmic membrane; Q, ubiquinone pool; bc 1 , bc 1 complex; RC, photosynthetic reaction center.
The PioAB system potentially catalyzes electron transfer from extracellular solid substrates, and its broad distribution suggests that this phenotype may also be widespread (15,21). Interestingly, we found that PioA-like (decahemes with extended N termini) homologs are conserved only in phototrophic bacteria ( Fig. 7A and Fig. S2). Although MtrA-like decaheme homologs and the MtrCAB electron conduit have been previously characterized, these proteins are found in nonphototrophic organisms such as Shewanella that use extracellular minerals as their terminal electron acceptors. Typically, these proteins allow electrons produced inside the cytoplasm to be transferred to the terminal electron acceptor. However, the MtrCAB conduit in Shewanella requires its extracellular decaheme component, MtrC, or its paralogues (OmcA and MtrF) to reduce Fe(III) or to transfer electrons to electrodes (23)(24)(25). In contrast, PioA-like homologs are found in phototrophic bacteria, and the PioAB conduit (with no apparent extracellular decaheme cyt c component) likely allow electrons to be transferred across the outer membrane from a variety of extracellular electron donors [such as soluble Fe(II), insoluble iron, and insoluble iron mineral proxies such as poised electrodes] (14). Subsequently, PioC transfers electrons from the PioAB conduit to the photosynthetic reaction center (33,34). The transferred electrons are utilized to produce NAD(P)H that is required for carbon fixation via the Calvin-Bassham-Benson cycle in TIE-1 (14).

MATERIALS AND METHODS
Bacterial strains and plasmids. A complete list of strains, plasmids, and primers used in this study are described in Table S2 in the supplemental material. Bacteria were grown using medium and culture conditions as previously described (14,15), and details are provided in Text S1.
Fractionation and preparation of soluble and membrane fractions of TIE-1. Photoautotrophically grown R. palustris TIE-1 strains with hydrogen in fresh water (FW) medium (10) were used for fractionation. Fractionation was done as previously described (6) with some modifications (see Text S1 in the supplemental material for a complete description of the methods used).
Protein expression and purification. Affinity purification of proteins from E. coli and R. palustris TIE-1 were performed as previously described (61). E. coli RK103 and the R. palustris TIE-1 ΔpioA mutant were used as the expression hosts (see Text S1 in the supplemental material for a complete description of the methods used).
Antibody production and immunoblots. Antibody production and immunoblotting are described in detail in Text S1.
Cell suspension assay. All cell suspensions were performed in an anaerobic chamber (80% N 2 , 15% CO 2 and 5% H 2 ; Coy Laboratory, Grass Lake, USA) at room temperature as previously described (6,28,62). TIE-1 and other strains were inoculated from a prephotoautotrophic culture in FW medium with hydrogen (80% H 2 , 20% CO 2 ) and grown to an OD 660 of ϳ0.3. The cells were harvested by centrifugation (10,000 ϫ g for 5 min), washed three times with HEPES buffer (50 mM HEPES, pH 7.0, 20 mM NaCl), resuspended, and concentrated to an OD 660 of ϳ0.9 in HEPES buffer supplemented with 1 mM Fe(II) and 5 mM nitrilotriacetic acid (NTA). One hundred microliters of the cell suspensions was aliquoted in the 96-well plate. To start the assay, the plate was placed under a 60-W incandescent light at a distance of 25 cm. Fe(II) measurement at the initial time point (time zero [T 0 ]) was taken before the light source was turned on. All of the Fe(II) measurements were performed using a ferrozine assay (63).
RNA-preparations and RT-qPCR. RNA-preparations and reverse transcription-quantitative PCR (RT-qPCR) were done as previously described (14,15), and details are provided in Text S1.
Microfluidic bioelectrochemical cell and conditions. The microfluidic bioelectrochemical cells (-BECs) were assembled and used as previously described (14) (see Text S1 in the supplemental material for a complete description of the methods used).