Laboratory stocks of G. sulfurreducens PCA [lab strain resequencing described in Chan et al. (2015)], and mutants were resuscitated from laboratory stocks by streaking onto 1.5% agar containing minimal salts medium (NB – per 1 L: 0.38 g KCl, 0.2 g NH4Cl, 0.069 g Na2H2PO4^∗H2O, 0.04 g CaCl2^∗H2O, MgSO4^∗H2O, 10 mLs of mineral solution stock) with 20 mM acetate and 40 mM fumarate (NBFA), and picking colonies into liquid medium for each experiment. All G. sulfurreducens cultures and media were prepared anaerobically under 80% N₂, 20% CO₂ atmosphere.

Electrochemical bioreactor experiments contained NB medium with 20 mM acetate and additional NaCl salts added in place of fumarate. Cultures of Geobacter strains grown with excess acetate, were used as an inoculum as they approached an OD (600 nm) of 0.5. Polished graphite electrodes (1500 grit), with a surface area of 3 cm², were used as working electrodes. A small piece of platinum wire was used as a counter electrode and a calomel electrode connected via a Vycor frit salt bridge was used as a reference electrode. Bioreactors were maintained at a constant 30°C. Growth with freshly precipitated insoluble Fe(III) oxide (55 mM), Fe(III) citrate (55 mM) and MnOOH (30 mM) was performed in the same medium without the additional salt and 20 mM acetate as the electron donor. Detailed preparation of Fe(III) oxide and MnOOH stocks can be found in (Levar et al., 2017). The approach of FeCl₃ hydrolysis with NaOH used in our studies produces fresh Fe(III) oxide typically as ferrihydrite and akaganite, which reduces to magnetite during growth of G. sulfurreducens (Levar et al., 2017), but are referred as Fe(III) oxides for clarity. Fe(III) citrate media was prepared as follows: 13.7 g of ferric citrate was solubilized in just boiling water for 5 s, immediately added to ice cold water and pH adjusted to 6. Medium was finished similarly to above.

Fe(III) reduction was measured by monitoring accumulation of Fe(II) by means of a FerroZine (3-(2-Pyridyl)-5,6-diphenyl-1,2,4-triazine-p,p′-disulfonic acid) assay. As previously described, (Rollefson et al., 2009; Levar et al., 2014) 100 μL samples were extracted from 10 mL Balch culture tubes and diluted in 1 N hydrochloric acid until conclusion of the experiment when the FerroZine assay was performed in 96 well plate format. Mn(IV) reduction was monitored as described in Levar et al. (2017).

Cell attachment to surfaces was characterized using a crystal violet, 96 well plate assay as described previously (Rollefson et al., 2009, 2011). Growth medium contained 30 mM acetate, 40 mM fumarate. Incubation occurred for 72 h. Optical density at 600 nm was measured, wells were emptied, and cells bound to the plate were stained with 0.006% crystal violet for 15 min at room temperature. Excess dye was rinsed away with distilled water. 300 μL of 100% DMSO was used to solubilize the dye that remained attached to cells. Microwell plates with Nunclon were used in this study (Thermo Fisher Scientific).

Geobacter sulfurreducensΔpgcA was created using the markerless deletion method described in previously (Chan et al., 2015; Zacharoff et al., 2016). 1 kB up and downstream of GSU1761 (pgcA) was cloned into pk18mobsacB vector. This plasmid was mated into G. sulfurreducens via Escherichia coli strain S17-1. The first round of selection was performed on (200 μg/mL) kanamycin NBFA plates to obtain recombinant cells. Kanamycin resistant colonies were restreaked and plated on 10% sucrose for a second round of selection for recombination events that resulted in reversion to wild type or gene deletion. Colonies from this round of selection were patched onto plates with and without kanamycin. Colonies sensitive to kanamycin were screened for loss of GSU1761 using PCR. GSU1761 was also cloned into pRK2-Geo2 (Chan et al., 2015) backbone for growth complementation testing. This plasmid contains a constitutive promoter from the G. sulfurreducens gene acpP (GSU1604). GSU1761 was also cloned into the pBAD202/D-TOPO^® (Thermo Fisher Scientific) plasmid backbone which resulted in an arabinose inducible expression vector containing PgcA fused to a 6X – histidine tag on the carboxy terminus (pBAD202PgcA).

Shewanella oneidensis was electroporated with pBAD202PgcA plasmid after passage of the expression plasmid through a methylation minus E. coli K12 ER2925 strain (New England Biolabs, Ipswich, MA, United States). Transformants were selected on 50 μg/mL kanamycin infused LB plate. S. oneidensis was routinely cultured in lysogeny broth (LB) (Becton, Dickinson & Co., Franklin Lakes, NJ, United States). Plasmids and deletion strains were sequence confirmed via Sanger sequencing at UMGC, University of Minnesota. See Table 1 for strain designations.

Ten milliliter cultures of the PgcA expressing strain of S. oneidensis were used to inoculate 1 liter of LB medium containing 50 μg/mL kanamycin. Cells were incubated at room temperature (25°C) at slow rotation speed to achieve microaerobic conditions. The use of non-baffled shake flasks also decreased the amount of oxygen in the medium. Growth was monitored at 600 nm until an optical density of 0.5 was achieved. At this time 3 mM (final concentration) of arabinose was added to induce PgcA expression. 100 μM FeCl₃ was also added at this time to increase the amount of bioavailable iron in the medium. Cells were pelleted 18 h after induction at 4,000 ×g. The pellet was washed with 100 mM Tris-HCl, 200 mM NaCl, pH 7.5 buffer. Resuspension and lysis via sonication (50% duty cycle, amplitude of 20%, 2 cm horn, for 30 min) was performed in the same buffer with lysozyme and DNase. Lysate was centrifuged at 30,000 ×g for 30 min. The soluble fraction was loaded on to a nickel affinity column. Protein was eluted with 300 mM imidazole. Concentrated eluent was further purified with gel filtration or anion exchange chromatography. Gel filtration was done using a 45 cm length, 1 cm diameter column filled with Sepharose 6B (Sigma–Aldrich, St. Louis, MO, United States). A flow rate of 1 mL/min was used during column equilibration and sample separation. Anion exchange separation was performed using HiTrap Q HP, 5 mL columns (GE Health Care, Uppsala, Sweden). A flow rate of 5 mL/min was used. Sample was loaded onto column in no salt 100 mM Tris-HCl. A gradient program was initiated using a mixture of 0.5 M NaCl, 100 mM Tris-HCl and no salt 100 mM Tris-HCl. Protein sample was monitored throughout purification using SDS-PAGE gel stained for total protein and for peroxidase activity based heme stain 3,3′,5,5′ tetramethylbenzidine (TMBZ) (Thomas et al., 1976; Smith et al., 2015).

Protein samples that resulted from nickel affinity purification were separated on a Bis-Tris, SDS, 12.5% polyacrylamide gel. Bands at 41 and 57 kDa were excised from the gel (Supplementary Figure S1). Trypsin digest and LCMS mass spectrometry using Thermo LTQ were performed on each of the band sizes (Center for Mass Spectrometry and Proteomics, University of Minnesota). PEAKS Studio software was used to analyze fragments (BSI Informatics Solutions).

Protein sample was dialyzed with 50 mM phosphate, ph 7.5, with 100 mM sodium fluoride to decrease background signal in the ultraviolet region (Greenfield, 2007). A JASCO-J815 spectropolarimeter was used to acquire circular dichroism spectra in the range of 185–600 nm. Samples were maintained at room temperature for the entirety of experimentation. Spectra were analyzed using K2D3 program (Pellegrini, 2015).

Ninety-six deep well plates were prepared with 20 mM Fe(III) oxide medium. Flavin mononucleotide (FMN) (0–200 μM), bovine serum albumin (BSA), horse heart cytochrome c, or purified PgcA (12 μM equivalent of each protein) were added prior to cell addition. 100 μL of 0.6 OD (600 nm) cells, either wild type G. sulfurreducens, or ΔpgcA strain, were mixed into the 1 ml wells. A negative control lacking cells was also included. Cells were allowed to reduce Fe(III) for 20 h in an anaerobic chamber with an atmosphere of 20% CO₂, 75% N₂, 5% H₂. A FerroZine assay was used for Fe(II) quantification, as described above. Preliminary Fe(II) measurements conducted over 4 h intervals verified that reduction was linear over this short incubation period.

Sequences used for Figure 6 were obtained from (Strain, locus, GI number); G. sulfurreducens, GSU1761, GI:637126441: G. uranirreducens, Gura_0706, GI:640548206: G. bemidjiensis, Gbem_1881, GI:642767873, Geobacter sp. FRC-32, Geob_3176, GI:643640481: Geobacter sp. M21, GM21_2329, GI:644869943: G. bremensis, K419DRAFT_01717, GI:2524445678: G. argillaceus, Ga0052872_00704, GI:2597449491: G. pickeringii, Ga0069501_111509, GI:2633859152: Desulfuromonas soudanensis WTL, Ga0081808_112930, GI:2637110285: G. sulfurreducens AM-1, Ga0098194_11, GI:2640720749: G. soli, Ga0077628_111213, GI:2649969705: G. anodireducens, Ga0133348_111806, GI:2689034555. After preliminary alignment by Clustal, sequences were trimmed to include only conserved repetitive/heme regions and re-aligned to obtain multifasta files as input for WebLogo3 using default parameters (Crooks et al., 2004).

The amino acid sequence of G. sulfurreducens PgcA predicts three c–type heme binding (CXXCH) motifs separated by repetitive elements (Figure 1). The amino acids threonine (T) and proline (P) alternate to form a string of 29 PT_X repetitions, followed by a heme motif, and a second PT_x-heme region (Figure 1). PT_x-rich tandem repeats are found in many G. sulfurreducens relatives, while PA_x-dominated repeats are found in strains such as G. uraniireducens and D. soudanensis. This general pattern could also be identified using PTRStalker to detect fuzzy tandem repeats (Pellegrini et al., 2012; Pellegrini, 2015), which detected many PgcA-like sequences in Geobacter genomes, and also predicted tandem repeats in extracellular cytochromes that did not contain PT_x or PA_x domains.

The number of hemes within putative PgcA homologs varies. Only one CXXCH motif is observable in G. metallireducens, while six occur in G. bemidjiensis. The presence of PT_X repeats in the G. sulfurreducens sequence was notable, as Lower et al. (2008) found that proline in the tripeptide S/T-P-S/T restricts flexibility and positions serine/threonine hydroxyl groups for hydrogen bonding with metal oxide surfaces. Hematite association has also been proposed near a short threonine-proline-serine motif near exposed heme groups in the S. oneidensis OmcA crystal structure (Edwards et al., 2012).

A markerless mutant lacking pcgA showed no defect in reduction of the soluble electron acceptor Fe(III) citrate, and expression of pgcA via a constitutive promoter in ΔpgcA cells also had no effect on growth (Figure 2A). In contrast, when insoluble metals such as Fe(III) oxide (Figure 2B) or Mn(IV) oxide were present (Supplementary Figure S2), reduction was severely impaired in the ΔpgcA strain. After 10 days of incubation, wild type G. sulfurreducens carrying an empty vector produced 36.3 mM Fe(II), while ΔpgcA produced only 9.0 mM Fe(II). Expression of pgcA from a constitutive promoter restored 75% of Fe(III) reduction activity, producing 27.4 mM Fe(II) in 10 days. Strains complemented with plasmids expressing pgcA showed higher variability, possibly reflecting known inhibitory effects of kanamycin carryover on extracellular respiration. A defect in Mn(IV) reduction was also observed in ΔpgcA mutants, where ΔpgcA produced 50% of wild type levels of Mn(II) (Supplementary Figure S2).

When cultivated using +0.24 V vs. SHE poised graphite electrodes as the electron acceptor, wild type and ΔpgcA cells demonstrated similar doubling times of 5.6 h (n = 3) vs. 5.5 h (n = 3) (Figure 3A). In addition, wild type and ΔpgcA cells reached a similar current density of 550 μA/cm² within 3 days of growth. Complementation of ΔpgcA in trans also resulted in similar growth. Further evidence that PgcA played no role at any stage of electron transfer to electrodes was obtained from cyclic voltammetry scans over a wide (-0.4 V to +0.3 V) potential range, which were similar at all redox potentials. Similar results were obtained at -0.1 V vs. SHE, consistent with cyclic voltammetry data (Figure 3B).

Geobacter strains lacking extracellular components can show increased binding to negatively charged surfaces, which has been correlated with defects in reduction of substrates. Mutants in the xap extracellular polysaccharide synthesis gene cluster show over 250% increases in attachment to negatively charged surfaces, and are also defective in binding poised graphite electrodes (Rollefson et al., 2009, 2011). In contrast, mutants showing 50–75% levels of attachment were not correlated with any reduction phenotypes. Binding was investigated in ΔpgcA cells grown to stationary phase with fumarate as the electron acceptor, and determined by a crystal violet attachment assay. Using polystyrene culture plates, ΔpgcA bound 79% as well as wild type, suggesting changes to the outer surface, but consistent with wild type-like interactions at electrodes.

PgcA was expressed in S. oneidensis under control of an arabinose-inducible promoter under microaerobic conditions, and successful incorporation of all three predicted hemes were determined by the pyridine hemochrome assay (Berry and Trumpower, 1987) and mass spectrometry. During purification, we consistently obtained both a large and small form, similar to what has been observed in Geobacter; mass spectrometry of excised gel bands revealed that these corresponded to 57 and 41 kDa forms of PgcA, where the short form was truncated at amino acid 127 leaving an N-terminal alanine. This smaller variant was similar to the dominant form of PgcA observed in supernatants of evolved G. sulfurreducens KN400 strains overexpressing PgcA (Tremblay et al., 2011), as well as the unidentified cytochrome previously recovered from G. sulfurreducens PCA supernatants (Lloyd et al., 1999). While these sizes and processing patters are comparable, suggesting similarities in proteases or cleavage mechanisms between the two organisms, until PgcA is purified directly from G. sulfurreducens the exact site or mechanism of processing under native conditions remains unclear.

The visible spectrum of PgcA had an absorbance maximum at 405 nm in the oxidized state. The pyridine hemochrome assay extinction coefficient at 408 nm was 137,000 M^-1cm^-1, consistent with the incorporation of three hemes (protein concentration was determined based on the predicted extinction coefficient at 280 nm). PgcA was rapidly oxidized and reduced by ferricyanide and sodium dithionite, respectively. The reduced protein shifted to a maxima at 417 nm (γ, or Soret), and demonstrated peaks at 518 nm (β), 552 nm (α). No additional changes in the absorbance spectrum in the 650–700 nm range were observed which can sometimes detect a His-Met coordination that results in a low-spin iron (Figure 4A) (Roldán et al., 1998; Ghosh et al., 2005).

Because of the significant amount of proline-rich repeats in PgcA, analyses of protein secondary structure was conducted using circular dichroism spectroscopy. Proline-induced backbone rigidity can create unique secondary structures which alter regular alpha-helix/beta-sheet patterns, such as in the case of the collagen triplex helix. CD spectra were recorded in millidegrees (mdeg) from 200 to 240 nanometer wavelengths of fully oxidized, truncated (41 kDa) PgcA in pH 7.5 phosphate buffer with 100 mM sodium fluoride (Greenfield, 2007). No evidence of intrinsic disorder or unique secondary structures were detected in the experimental conditions (using K2D3). The 41 kDa form of PgcA was composed of 70.5% alpha helical character and 5.2% beta-sheet character (Figure 4B), (Greenfield, 2007; Pellegrini, 2015). The alpha helix relative to beta sheet composition of PgcA was significantly more helical than the 10% alpha helix value estimated for OmcS (Qian et al., 2011) and the 13% value reported for OmcZ (Inoue et al., 2010), but is consistent with the alpha helical bias observed for other hemeproteins (Smith et al., 2010).

As some extracellular cytochromes (such as OmcS) show an affinity for Fe(III) oxides, PgcA was incubated with freshly prepared Fe(III) oxide media, as well as biologically reduced Fe(II) oxide at pH 6.5, where both of these minerals have a net positive charge (Kosmulski, 2011). Using absorbance at 410 nm to monitor soluble protein concentrations, as-purified PgcA showed an ability to bind the oxidized, but not reduced mineral (Figure 4C). After incubation with Fe(III), all PgcA added to solution was removed by the pelleting of metal oxide particles. As all PgcA was precipitated with the Fe(III) oxide, it appeared that both short and long forms of the protein had an affinity for the oxidized form of Fe(III) oxide. In contrast, incubation with magnetite [a mixed Fe(II)-Fe(III) mineral] removed less than 10% of PgcA from solution, and magnetite also did not reduce PgcA, based on spectroscopy. When BSA or horse heart cytochrome c were incubated with Fe(III) or Fe(II)-Fe(III) oxides, both proteins remained in the supernatant (>90%).

Purified PgcA was used to determine if PgcA added extracellularly could rescue the inability of ΔpgcA to reduce Fe(III), or accelerate activity in the wild type. As a control in these experiments, purified PgcA was compared with additions of a known electron shuttle, flavin mononucleotide, as well as proteins not expected to facilitate electron transfer [bovine serum albumin and horse heart c-type cytochrome (Supplementary Figures S3, S4)] (Hartshorne et al., 2007; Shi et al., 2012; Kotloski and Gralnick, 2013). All cells were pre-grown to a state of electron acceptor limitation, washed and incubated with Fe(III) oxide and acetate, and the accumulation of Fe(II) monitored for 20 h.

Under these conditions, wild type G. sulfurreducens provided with 20 mM acetate and 5% hydrogen produced 2.2 mM Fe(II), while the ΔpgcA mutant only produced 0.13 mM Fe(II) over the same time period (Figure 5). When purified PgcA was added, rates of Fe(III) reduction doubled in the wild type, but increased nearly 20-fold in the ΔpgcA mutant. Addition of horse heart c-type cytochrome or bovine serum albumin at similar concentration had no stimulatory effect on either culture.

When wild type cells were incubated with 20 mM acetate, 5% hydrogen, Fe(III) oxide and increasing amounts of flavin mononucleotide, rates of Fe(III) reduction improved until FMN concentrations reached 50 μM. Levels above 50 μM produced similar levels of stimulation. Addition of 50 μM FMN accelerated metal reduction by wild type cells similar to addition of PgcA, and stimulated reduction in ΔpgcA mutants more than added PgcA alone. When both FMN and PgcA were added to the pgcA mutant, the effects were additive, resulting in the highest observed levels of stimulation, over 40-fold faster than the mutant alone.