All the strains of P. aeruginosa used in the heterologous expression experiments (Table 1) and Escherichia coli strains used for cloning and transformation were routinely grown aerobically in Luria-Bertani (LB) medium and incubated at 37°C with gentle agitation. G. sulfurreducens strain PCA was from our culture collection and was used as the wild-type background to construct all of the mutant derivatives (Table 1). All of the Geobacter strains were routinely cultured anaerobically in NB medium (Coppi et al., 2001) supplemented with 0.1% yeast extract and 1 mM cysteine and with 15 mM acetate and 40 mM fumarate (NBAF). When indicated, NBAF was replaced by a modified freshwater FW medium (Cologgi et al., 2011) supplemented with 1 μM Na₂SeO₄ to stimulate growth (Speers and Reguera, 2012) and with 15 mM acetate as electron donor and 40 mM fumarate as the electron acceptor (FWAF). Fe(III) oxides cultures contained FWA medium and 100 mM poorly crystalline Fe(III) oxides and, when indicated, 4 mM of the metal chelator nitrilotriacetic acid (NTA). Growth in the Fe(III) oxides cultures was measured periodically with the ferrozine method (Stookey, 1970) as the amount of 0.5 N HCl-extractable Fe(II) resulting from the reduction of Fe(III) (Lovley and Phillips, 1986). Unless otherwise indicated, all of the incubations were at 30°C.

The WT strain of G. sulfurreducens was grown in 10 ml of NBAF or FWAF and cultures were incubated either at 30 or 25°C. Exponentially growing cells were then harvested by centrifugation, and the pelleted cells were suspended in 2 ml of TRIzol® reagent (Invitrogen Life Technologies). Extraction of total cellular RNA was performed following manufacturer's recommendations. Genomic DNA was removed with RNase-free DNase I (QIAGEN) and the remaining RNA was purified using a QIAGEN's RNeasy column. Primers were designed to have the same melting temperature and PCR efficiency using the pcrEfficiency open source tool (http://srvgen.upct.es/efficiency.html; Mallona et al., 2011). Complementary strand synthesis was performed in a 20 μl reaction mixture containing 1 μg of total RNA, 10 pmol of gene-specific primer and dNTPs. The mixture was heat-treated at 65°C for 5 min followed by incubation on ice for at least 1 min before adding 1 μl of SuperScript® III reverse transcriptase (Invitrogen) and incubating at 50°C for 60 min. The reaction was inactivated by heating at 70°C for 15 min. PCR-amplification reactions (50 μl, total volume) contained 1 μl of cDNA (serial dilutions from first-strand synthesis reaction), 10 pmol of sense and antisense primers (Table 2), and 25 μl of SYBR green 2X master mix (Applied Biosystems, Life Technologies). qRT-PCR was performed in duplicate reactions under conditions (95°C for 30 s, 50°C for 30 s, and 50 cycles of 1 min at 70°C) that resulted in threshold crossings between 18 and 35 for all of the primer pairs. Fold changes in gene expression of the target genes were calculated in reference to the internal control recA as 2−ΔCT (Schmittgen and Livak, 2008), where ΔC_T is the C_T of the gene of interest minus the C_T of the recA internal control.

The P. aeruginosa strains and plasmids used for heterologous complementation are listed in Table 1. The wild-type PAK strain, the pilin-deficient PAKΔpilA and pilT-deficient PAKΔpilT derivatives, and plasmids pMMB67EH and pRK2013 were kindly provided by Matthew C. Wolfgang (University of North Carolina School of Medicine). The G. sulfurreducens pilT genes were cloned into the broad host range expression vector pMMB67EH using the In Fusion cloning system (Clontech Laboratories, Inc.). The pMMB67EH plasmids and its derivatives were electroporated in E. coli HB101 carrying the mobilizing helper plasmid pRK2013 and transformants were selected on LB agar containing 30 μg/ml carbenicillin and 50 μg/ml kanamycin. The resulting E. coli donor strains were mated with the PAKΔpilT strain on antibiotic-free LB agar plates for 2 h at 37°C. After the mating, the cells were suspended in 2 ml of LB medium and serially diluted and plated on LB agar containing 500 μg/ml carbenicillin and 25 μg/ml irgasan. The resulting strains were cultured with 300 μg/ml carbenicillin to maintain the plasmid and, when indicated, with 20 μM IPTG to promote the expression of the cloned genes.

T4P from all the P. aeruginosa strains were purified following a previously described protocol (Castric, 1995). Briefly, 100 μl of an LB culture were plated on triplicate LB-agar plates supplemented with 20 μM IPTG and incubated overnight at 37°C. When needed, carbenecillin (30 μg/ml) was included in the LB-agar medium to maintain the pMMB67EH plasmid or any of its derivatives (Table 1). The cells were scraped off the plate and suspended in 1.5 ml of LB medium and 5 μl of the cell suspension was diluted in 2 ml water to measure its OD₆₀₀ and estimate the cell density. T4P fibers in the cell suspensions were sheared by passing the cells 4 times through a 23-gauge needle and the cells were then removed by centrifugation (2 times at 7500 × g for 15 min at room temperature). The T4P in the supernatant fluids were then precipitated by adding MgCl₂ to a final concentration of 100 mM and incubating on ice for 60 min. Once precipitated, the T4P were concentrated by centrifugation (12,000 × g for 15 min at 4°C), suspended in 100 μl of 0.1% SDS, and incubated overnight at 4°C to separate the pili bundles. The T4P content in the samples was quantified using the Bio-Rad protein assay Kit (Bio-Rad Laboratories, Inc.), following the manufacturer's recommendations for the standard procedure in microtiter plates, and its concentration (mg/ml) in each sample was normalized by the sample's cell density (in OD₆₀₀ units).

T4P purified in the 0.1% SDS solution were diluted appropriately to standardize for the cell density (OD₆₀₀) and then mixed with an equal volume of Tris-Tricine sample buffer. After incubation at 95°C for 5 min to depolymerize the fibers into the individual pilin subunits, the proteins in 15 μl of sample were separated electrophoretically in a 10–20% Mini-PROTEAN® Tris-Tricine Precast Gel (Bio-Rad). One lane was loaded with 4 μl of the Novex® Sharp Pre-stained Protein Standard (Invitrogen Life Technologies) as standards. The gel was stained overnight with Coomassie Brilliant Blue R-250 Staining Solution (Bio-Rad) and destained overnight with a solution of 50% methanol and 10% acetic acid.

Biofilm assays with strains of P. aeruginosa were performed under static conditions in microtiter plates, essentially as described elsewhere (Merritt et al., 2005) but using M63 minimal medium supplemented with 0.4% arginine as the sole carbon and energy source to stimulate biofilm development over a 24 h period (Caiazza and O'Toole, 2004). The medium was supplemented with carbenecillin (30 μg/ml) to maintain the pMMB67EH plasmid and with IPTG (20 μM) to heterologously express the cloned pilT genes. The biofilm biomass was then stained with 0.1% crystal violet, and allowed to dry before solubilizing the biofilm-associated dye with 95% ethanol and measuring its OD₅₈₀ (Merritt et al., 2005). To account for technical and biological variability, the biofilm assays included six technical replicates (wells) per strain tested and a minimum of three independent experiments. Outliers were excluded from the compiled biofilm assay results using the Quartile or Fourth-Spread method (Devore, 2000) and the data were plotted as a box-and-whisker graph using the Microsoft Excel software.

For the twitching motility assays, a thin layer (~1 ml) of molten LB medium with 1% agarose and 20 μM IPTG was deposited onto a sterile glass slide and allowed to solidify. The center of the agar was stab-inoculated with overnight LB-20 μM IPTG cultures of the P. aeruginosa strains incubated at 37°C with shaking (180 rpm). When needed, carbenecillin (30 μg/ml) was added to the liquid cultures to maintain the pMMB67EH plasmid. After stab-inoculation, the slide was placed inside a Petri dish surrounded by damp, sterile absorbent paper to maintain the humidity inside the chamber and the dish was then sealed with parafilm to prevent drying. After overnight incubation at 37°C, the areas of growth on the surface of the agar were gently removed with a coverslip and the slides were then stained with 1 ml Coomassie Brilliant Blue R-250 Staining Solution (Bio-Rad) for 2 h. The slides were briefly rinsed with 5 ml of a solution of 50% methanol and 10% acetic acid and then destained in 20 ml of the same solution for 4 h before replacing 15 ml of the solution with an equal volume of a solution of 5% methanol and 7.5% acetic acid and incubating overnight. After destaining, the agar layer on the slides was allowed to dry completely before imaging the stained twitching zones with a scanner.

G. sulfurreducens PCA was used to construct mutants carrying full deletions in each of the four pilT homologs and in the pilB gene, using the primers listed in Table 2. The general procedure was to replace the full length of the wild-type gene with the aaaC1 gentamicin cassette of plasmid pCM351 (Marx and Lidstrom, 2002) in the same gene orientation so as to prevent polar effects on downstream genes. For the construction of the pilT1 and pilT2 deletions, we PCR-amplified regions approximately 500 bp upstream and downstream of the target genes using the Phusion® high fidelity DNA polymerase (New England Biolabs) and cloned the fragments separately into the pCR2.1-TOPO vector (Invitrogen). The downstream fragments were then digested and cloned into the NdeI-XbaI site of pCR2.1-TOPO harboring the upstream fragments. This resulted in plasmids carrying the upstream and downstream regions and an NdeI site in between. The aaaC1 gentamicin-resistance cassette was then PCR-amplified from pCM351, digested with NdeI, and cloned into the NdeI site separating the upstream and downstream fragments. The cloned fragment was excised from the plasmid with EcoRI and XbaI and gel-purified prior to introduction into electrocompetent cells of G. sulfurreducens via electroporation, as described previously (Coppi et al., 2001). The pilT3, pilT4, and pilB deletion mutants were constructed by recombinant PCR (Coppi et al., 2001), using constructs generated by fusing together the upstream region of the target gene, the aaaC1 cassette, and the downstream region in a PCR reaction with the external primers. The final PCR product was gel-purified with a Zymoclean Gel DNA Recovery kit (Zymo Research Co.) and incubated at 70°C for 30 min with one volume of 2X GoTaq Green Master Mix (Promega) to produce a 3′ A-overhang prior to cloning into pCR2.1-TOPO. The cloned fragment was PCR-amplified and gel-purified before electroporation into G. sulfurreducens electrocompetent cells (Coppi et al., 2001). In all cases, transformants were selected on NBAF plates containing gentamicin (5 μg/ml) and confirmed by PCR, DNA sequencing, and Southern blot hybridization, as described previously (Ausubel et al., 1995).

The pCM158 plasmid was electroporated in the pilT4 (ΔpilT4::acc1) and pilB (ΔpilB::acc1) mutant strains to express the Cre-recombinase and excise the acc1 gentamicin-resistance cassette, which is flanked by the loxP sites that the Cre protein recognizes and recombines (Marx and Lidstrom, 2002). After confirming the excision of the gentamicin resistance cassette by PCR, the pCM158 plasmid was cured from the strains by culturing in the absence of kanamycin. The resulting strains, designated ΔpilT4 and ΔpilB, were electroporated with plasmids pRG5-pilT4 and pRG5-pilB to express the wild-type pilT4 and pilB genes, respectively, from the medium-copy number plasmid pRG5 (Kim et al., 2005). All cloning steps used the In-Fusion cloning system (Clontech) and selection of transformants was in agar-solidified NBAF medium with spectinomycin (150–300 μg/ml). The ΔpilT3::acc1 mutation was also introduced in the ΔpilT4 strain to generate the double pilT3 pilT4 mutant.

G. sulfurreducens was grown to mid-exponential phase in FWFA medium and serially transferred (10% v/v) three times before diluting the cells in fresh FWAF medium to an OD₆₀₀ of 0.04. Two hundred microliters of this cell suspension were dispensed in the wells of a 96-well plate, which was then sealed to prevent evaporation. The plate was incubated at 30°C inside a TECAN Sunrise™ Absorbance Reader (TECAN, Männedorf, Switzerland) housed in an anaerobic glove bag (Coy Labs) and planktonic growth was monitored periodically (OD₆₀₀) throughout the incubation. After 48 h, the plates were removed from the glove box, the supernatant was decanted, and the biofilms in each well were stained for 20 min with 200 μl of 0.1% crystal violet, gently washed 3 times with 200 μl of ddH₂O, and air-dried (Stepanovic et al., 2000). The biofilm-associated dye was released from the biofilms by adding 200 μl of 33% (v/v) acetic acid. After 20 min, the plate was mixed on an orbital shaker and the solubilized dye was transferred to a new microtiter plate to record the OD₅₈₀.

Loosely bound proteins of the outer membrane of G. sulfurreducens were first mechanically detached from cells grown to mid-exponential phase in FWAF medium. The proteins in the sheared fraction were then separated by SDS-PAGE, as previously described (Cologgi et al., 2011), and those containing heme groups were stained with N, N, N, N-tetramethylbenzidine (Thomas et al., 1976).

The MECs used in these studies were dual-chambered, H-type electrochemical cells equipped with graphite rod (Alfa Aesar,1.27-cm diameter, 99% metals basis, 12 cm²) anode and cathode electrodes, separated with a Nafion membrane (Ion Power, Inc., New Castle, DE), and set up as described previously (Speers and Reguera, 2012). Each chamber was filled with 90 ml of anaerobic mineral DB medium (Speers and Reguera, 2012) and an initial concentration of 1 mM sodium acetate was added to the anode chamber as electron donor to support the growth of the anode biofilms and current production. The WT and mutant strains of G. sulfurreducens used in the MECs experiments were first grown in DB medium with acetate (20 mM) and fumarate (40 mM) at 30°C, harvested by centrifugation, suspended in DB medium, and inoculated into the anode chamber, as previously described (Speers and Reguera, 2012). A VSP potentiostat (BioLogic, Claix, France) was used to set a 0.24-V potential at the anode electrode vs. a 3 M Ag/AgCl reference electrode (Bioanalytical Systems, Inc.) prior to cell inoculation and to monitor current production at the anode electrode throughout the experiment. The anode electrode was removed from the anode chamber when all the acetate was consumed, as indicated by the drop in current production to < 0.1 mA, and the biofilm cells were then differentially stained (green, live cells; red, dead cells) with the BacLight™ viability kit (Invitrogen), following the manufacturer's recommendations. After staining, the anode electrode was placed on a Lab-Tek® coverglass chamber (Nunc) filled with 3 ml of phosphate buffer saline and imaged with a FluoView FV1000 inverted confocal laser scanning microscope (CLSM) (Olympus) equipped with a UPLFLN 40x oil immersion objective (Olympus, NA 1.30). The green fluorescent dye (SYTO 9) was excited at 488 nm and emission was detected with a BA505-525 band pass filter. The red fluorescent dye (propidium iodide) was excited at 543 nm and emission collected with a BA560IF long pass filter. Vertical image stacks of 200 × 200 μm fields were collected every 1 μm and 3-dimensional image projections were generated with the FV10-ASW 3.0 software (Olympus). The biofilm biomass was estimated from the fluorescence emitted by the biofilm cells using the biovolume function of the COMSTAT software (Heydorn et al., 2000).

The genome of G. sulfurreducens contains four genes annotated as pilT (pilT1, GSU0146; pilT2, GSU0230; pilT3, GSU0436; and pilT4, GSU1492) that code for proteins containing the four conserved signature sequences (Walker A and B motifs and Asp and His boxes) of the secretion ATPase protein superfamily that PilT retraction ATPases belong to Savvides (2007). Indeed, all of the PilT paralogs encoded in the genome of G. sulfurreducens contain a Walker A motif with a conserved lysine for ATP-binding, a Walker B motif with a conserved glutamate residue for Mg²⁺-binding and ATP hydrolysis, and Asp and His boxes with conserved aspartic and histidine residues, respectively, which are essential for PilT activity (Figure 1; Walker et al., 1982; Savvides, 2007; Chiang et al., 2008; Jakovljevic et al., 2008). As shown in Table 3, the putative PilT proteins are 43–52% identical and 61–70% similar to each other, and PilT3 and PilT4 had the highest identities and similarities to the PilT ATPases that power pilus retraction in N. gonorrhoeae (PilT_Ng), M. xanthus (PilT_Mx), and P. aeruginosa strain K (PilT_Pa) (Whitchurch et al., 1991; Wolfgang et al., 1998; Jakovljevic et al., 2008). PilT3, in particular, was 56% identical and 74% similar to PilT_Pa, whereas PilT4 was 77% identical and 87% similar to PilT_Mx.

The genetic arrangement of the pilT genes in G. sulfurreducens and its transcriptional expression profile with neighboring genes is shown in Figure 2. pilT1 (GSU0146) is flanked by the recA and recX genes (Figure 2A), a gene arrangement that is conserved in other sequenced genomes in the Geobacteraceae family (Geobacter metallireducens, Geobacter uraniireducens, and Pelobacter carbinolicus). Furthermore, pilT1 was co-transcribed with the recA gene (Figure 2B), consistent with a role for PilT1 in the cellular recombination machinery. The pilT2 gene (GSU0230), on the other hand, is located downstream of a gene (GSU0231) encoding a hypothetical protein (Figure 2A) but transcribed independently from it under the conditions tested (Figure 2B). Downstream of pilT2, and in opposite orientation, is the alkK gene (GSU0229), which encodes a putative medium-chain fatty acid CoA ligase. This arrangement is unique to G. sulfurreducens. Further, pilT2 was not conserved in all of the Geobacter spp. with sequenced genomes. We found, for example, a pilT2 homolog in G. metallireducens (Gmet0260) but none in G. uraniireducens. The lack of conservation of pilT2 in other Geobacter genomes contrasts with the high conservation of the pilin-encoding gene, pilA (Reguera et al., 2005) and argues against its role as the PilT ATPase of the T4P apparatus in these bacteria.

The pilT3 gene (GSU0436) is downstream of a gene (mshE) encoding a putative mannose-sensitive haemagglutinin (MSHA) biogenesis/general secretory system II protein that is homologous to the pilin polymerization ATPase of MSHA pili in Vibrio cholerae (Jones et al., 2015). The mshE pilT3 arrangement is conserved in the genomes of G. metallireducens and G. uraniireducens. The two genes are unlikely to be co-transcribed in G. sulfurreducens because, although reverse transcription produced a faint band spanning both genes, the band was smaller than in the controls amplified from genomic DNA and thus it was more likely the product of non-specific binding of one or both primers (Figure 2B).

The pilT4 gene (GSU1492) is flanked by genes encoding proteins with predicted or confirmed roles in pili biogenesis and its regulation such as the pilin assembly ATPase PilB (Steidl et al., 2016), the pilin biogenesis protein PilC, the PilS sensor, the PilR transcriptional regulator (Juarez et al., 2009), and the pilin subunit PilA (Reguera et al., 2005; Figure 2A). This genetic arrangement is conserved in other Geobacteraceae (G. metallireducens, G. uraniireducens, and P. carbinolicus) and is similar to the organization of the pil region of M. xanthus (Wall and Kaiser, 1999). The genes flanking pilT4 were also very conserved among Geobacter species. The PilB protein encoded by GSU1491 was, for example, 94% identical and 97% similar to Gmet1393 from G. metallireducens, and 88% identical and 96% similar to Gura2682 from G. uraniireducens. Similarly, the PilC protein encoded by GSU1493 was 89% identical and 95% similar to Gmet1395 from G. metallireducens, and 81% identical and 90% similar to Gura2680 from G. uraniireducens. PilB and PilC proteins are the most highly conserved components of the bacterial T4P apparatus and related systems such as the bacterial type II secretion system and archaeal motility systems and both have been proposed to form the minimal functional motor unit of the pilus (Burrows, 2012). Reverse Transcription (RT)-PCR analysis using a primer set spanning the intergenic pilB pilT4 failed to amplify a full-length product, suggesting that, under the conditions tested, the two genes are not co-transcribed (Figure 2B). Yet, despite their independent transcription, the expression profiles of pilB and pilT4 in both rich (NBAF) and mineral (FWAF) medium were similar and matched well the expression profile of pilA regardless of the incubation temperature (30°C or pili-expressing temperatures of 25°C; (Reguera et al., 2005; Cologgi et al., 2011; Figure 3). Indeed, the pilT4, pilB, and pilA genes were up-regulated under all of the conditions tested, whereas pilT1, pilT2, and pilT3 were all down-regulated. Hence, based on gene and protein homologies, gene clustering, overall conservation in other Geobacter spp., and expression profiles with other pil genes, pilT4 is arguably the most likely candidate to encode for the PilT motor of the T4P apparatus in G. sulfurreducens.

We gained insights into the functionality of each of the PilT proteins encoded in the genome of G. sulfurreducens in heterologous complementation experiments that tested the ability of each pilT gene to complement the phenotypes associated with a mutant of P. aeruginosa strain K (PAK) carrying an in-frame deletion in the pilT gene (PAKΔpilT). The inability of the PAKΔpilT mutant to retract the T4P leads to hyperpiliation, a phenotype that can be restored by genetic complementation of the mutation with the wild-type pilT copy in trans (Sundin et al., 2002). PilT3 was the only PilT homolog of G. sulfurreducens that rescued the hyperpiliated phenotype of the PAKΔpilT mutant (p < 0.01), reducing the piliation per cell to WT levels (Figures 4A,B). The hyperpiliation of the PAKΔpilT mutant also increases the adhesion of the cells to surfaces and co-aggregation and promotes the formation of denser biofilms under static conditions, a phenotype that is restored by genetic complementation (Chiang and Burrows, 2003). We observed a similar restoration of the biofilm phenotype when the PAKΔpilT strain was genetically complemented with the pilT3 gene of G. sulfurreducens (Figure 4C). Thus, the PilT3 protein of G. sulfurreducens is able to power pilus retraction and control the levels of piliation and biofilm formation in P. aeruginosa.

Twitching motility in P. aeruginosa requires the activity of PilT and its paralog, PilU (Whitchurch et al., 1991; Whitchurch and Mattick, 1994). This mode of surface translocation involves cycles of pilus extension, adhesion of the pilus tip to the surface, and retraction of the pilus fiber to pull the cell (Skerker and Berg, 2001). The PilT motor hydrolyzes ATP to energize the retraction of the surface-attached pilus and provides the force needed to move the cell forward (Merz et al., 2000). As a result, the inactivation of PilT results in a non-twitching phenotype (Whitchurch and Mattick, 1994). Thus, we also investigated the ability of the PilT proteins of G. sulfurreducens to restore the twitching motility defect of the PAKΔpilT mutant in subsurface twitching motility assays on glass slides (Figure 4D). Interestingly, not only PilT3 but also PilT4 rescued the non-twitching phenotype (Figure 4D). Examination of the colony edges at higher magnification (Figure 4E) revealed the smooth and thin expansion zones in the strains genetically complemented with pilT4, and to a lesser extent with pilT3, that are characteristic of the WT twitching motility phenotype (Semmler et al., 1999). Thus, the heterologous expression studies demonstrated that both PilT3 and PilT4 can function as motor ATPases in P. aeruginosa, but PilT4's activity is restricted to twitching motility.

As in P. aeruginosa, the pili of G. sulfurreducens also promote cell-cell aggregation and biofilm formation (Reguera et al., 2007; Cologgi et al., 2014). Thus, we investigated the biofilm-forming abilities of G. sulfurreducens mutants carrying pilT deletions in reference to the WT strain (Figure 5A). As controls we also included a pili-defective pilB mutant, which makes thin biofilms, and its genetically complemented strain pilB+, which expresses the wild-type pilB gene in trans and restores pili assembly and biofilm formation (Steidl et al., 2016). Deletion of pilT1, pilT2, and pilT3 had no significant effect on biofilm formation in G. sulfurreducens. By contrast, a pilT4 mutant formed denser biofilms on plastic surfaces (p < 0.001), a phenotype associated with hyperpiliation in this bacterium (Reguera et al., 2007; Cologgi et al., 2014). The enhanced biofilm phenotype of the pilT4 mutant was restored in the genetically complemented strain, pilT4+, but not in the control pilT4p strain, which carries the empty pRG5 expression vector in the same pilT4 mutant background (Figure 5A). The enhanced biofilm-forming abilities of the pilT4 mutant cannot be explained by differences in cell growth, because planktonic growth rates for this mutant were comparable to those measured in all of the strains tested (~0.12 OD₆₆₀/h). Rather, the dense biofilm phenotype observed in the pilT4 mutant is consistent with the increased adhesion and co-aggregation reported for hyperpiliated strains of G. sulfurreducens (Reguera et al., 2007) and other bacteria (Deziel et al., 2001; Chiang and Burrows, 2003).

The T4P of G. sulfurreducens are also required for biofilm formation on anode graphite electrodes poised at a metabolically oxidizing potential and for optimal current production in MECs (Reguera et al., 2006; Steidl et al., 2016). Hence, we grew the pilT3 and pilT4 mutants in the anode chamber of a MEC fed with 1 mM acetate and monitored current production in reference to the WT strain (Figure 5B). The pilT4 mutant attached rapidly to the electrode and initiated current production before any other strain. The average attachment or lag phase in triplicate MECs for the pilT4 mutant was, for example, 5 ± 1 h, which is 4–5 times shorter than the WT (24 ±11 h). By contrast, the pilT3 mutant was delayed in attachment (33 ± 3 h average lag phase in triplicate MECs; Figure 5B). Yet once the cells attached and the exponential phase of biofilm growth started (corresponding to the linear phase of current production) the pilT4 and pilT3 mutant biofilms produced current at rates comparable to the WT strain and also reached similar levels of current maxima (Figure 5B). This contrasts with the severe defect in current production of the pilA mutant (Figure 5B), which carries a deletion in the pilin-encoding gene pilA that prevents it from making conductive pili (Reguera et al., 2005; Cologgi et al., 2011) and is also unable to secrete the matrix-associated cytochrome OmcZ_S required for electron transfer to the underlying electrode (Steidl et al., 2016). Furthermore, coulombic efficiencies in the WT and pilT4 biofilms were similar (80–90%), indicating that on average the WT and pilT4 anode biofilms converted the same amount of acetate into electricity (80–90%). However, the pilT4 mutant accumulated a much greater biofilm biomass on the electrode (23.3 ± 4.2 μm³/μm²) than the WT (11.9 ± 1.5 μm³/μm²) to produce the same amount of current as the WT (Figure 5C). This indicates that, on average, less current was generated per cell in the pilT4 biofilms than in the WT biofilms.

We also ruled out compensatory effects between PilT3 and PilT4 by testing the ability of a pilT3 pilT4 double mutant to colonize and generate current in the MECs. The double mutant was phenotypically indistinguishable from the pilT4 single mutant, having shorter lag phases but otherwise producing current like the WT (Figure 5B) and forming denser biofilms on the anode electrodes (Figure 5C). Hence, the pilT4 mutation was dominant over pilT3, not only rescuing the colonization defect of the pilT3 cells, but also promoting faster attachment and the formation of denser biofilms. This is again consistent with the aggregative nature of hyperpiliated strains of G. sulfurreducens (Reguera et al., 2007; Cologgi et al., 2014), which is expected for cells defective in pilus retraction (Sundin et al., 2002).

As G. sulfurreducens also expresses pili to access and reduce Fe(III) oxides (Reguera et al., 2005), we tested the ability of the pilT3, pilT4, and double pilT3 pilT4 mutants to grow with Fe(III) oxides when provided as sole electron acceptor in reference to the WT strain (Figure 6A). The strains were first grown in Fe(III) oxide media with 4 mM of NTA, a metal chelator that solubilizes the Fe(III) from the Fe(III) oxides and alleviates the need for cells to directly contact the oxide minerals (Reguera et al., 2005). Exponentially growing cells were then transferred to Fe(III) oxides medium without NTA. The chelating activities of the NTA carried over in the inoculum (ca. 0.4 mM) stimulated initial growth in the cultures, as indicated by the low levels of reduced iron (Fe[II]) solubilized by all of the strains during the first 4–5 days (Figure 6A). However, whereas the WT continued to grow exponentially beyond this initial phase with similar doubling times (2.9 ± 0.1 days), the pilT3 mutant plateaued for approximately 4 days before resuming Fe(III) reduction at rates similar to the WT strain (Figure 6A). This contrasts with a Tyr3 mutant, which carries alanine replacements in the pilin's three tyrosines that increase the electrical resistance of the T4P (Steidl et al., 2016) but is able to resume exponential growth without delay after the NTA-dependent phase, albeit at slower rates (doubling time, 10.5 ± 0.8 days). The delay observed in the pilT3 mutant is consistent with a strain that has a transient defect in adhesion, as we previously observed during the colonization of anode electrodes in MECs (Figure 5B). However, the pilT3 defect was chemically rescued in the presence of the metal chelator NTA, which provided a soluble, reducible form of Fe(III) for cellular respiration throughout the experiment (Figure 6B). By contrast, the reductive activities of the pilT4 mutant never fully recovered after the NTA-dependent phase (Figure 6A). Furthermore, as we observed during the colonization of electrodes (Figure 5B), introducing the pilT3 mutation in the pilT4 mutant (pilT3 pilT4 mutant) did not change the pilT4 mutant phenotype. Thus, they are unlikely to have complementary roles or compensate for each other's inactivation. And, as with pilT3, the defects of the single pilT4 and double pilT3 pilT4 mutants were chemically rescued with NTA (Figure 6B).

We also ruled out pleiotropic effects of the mutations in c-cytochrome expression by examining the type and abundance of outer membrane c-cytochromes (Figure 6C). There were no measurable differences in the heme-containing proteins mechanically sheared from the outer membrane of the WT, pilT3, and pilT4 strains and all included the OmcS c-cytochrome that is required for optimal reduction of Fe(III) oxides (Mehta et al., 2005). Hence, the Fe(III) oxides experiments support our early conclusion that PilT4 is the retraction ATPase in G. sulfurreducens and highlight a previously unrecognized role for pilus retraction in the respiration of Fe(III) oxides.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.