Production of pilus-like filaments in Geobacter sulfurreducens in the absence of the type IV pilin protein PilA


  • Editor: Stephen Smith

  • Present addresses: Richard H. Glaven, Naval Research Laboratories, Center for Biomolecular Science and Engineering, Washington, DC, USA.
    Hoa Tran, Department of Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA.

Correspondence: Anna Klimes, Department of Microbiology, University of Massachusetts Amherst, Amherst, MA 01003, USA. Tel.: +1 413 545 6748; fax: +1 413 577 4660; e-mail:


The pili of Geobacter sulfurreducens are of interest because of the apparent importance of the type IV pili in extracellular electron transfer. A strain of G. sulfurreducens, designated strain MA, produced many more pili than the previously studied DL-1 strain even though genome resequencing indicated that the MA and DL-1 genome sequences were identical. Filaments that looked similar to type IV pili in transmission electron micrographs were abundant even after the gene encoding PilA, the structural pilin protein, was deleted. The results of proteinase K treatment indicated that the filaments were proteinaceous. The simultaneous deletion of several genes encoding homologues of type II pseudopilins was required before the filaments were significantly depleted. The pilA-deficient MA strain attached to glass as well as the wild-type MA did, but strains in which three or four pseudopilin genes were deleted in addition to pilA had impaired attachment capabilities. These results demonstrate that there are several proteins that can yield pilin-like filaments in G. sulfurreducens and that some means other than microscopic observation is required before the composition of filaments can be unambiguously specified.


The type IV pili of Geobacter sulfurreducens are of interest because of their proposed role as conduits for extracellular electron transfer to insoluble electron acceptors such as Fe(III) oxides (Reguera et al., 2005) and electrodes (Reguera et al., 2006; Nevin et al., 2009). Fe(III) oxides are the most abundant natural electron acceptors for Geobacter species in a diversity of submerged soils, aquatic sediments, and the subsurface, where these organisms play an important role in the natural cycles of carbon and metals as well as the bioremediation of organic and metal contaminants (Lovley, 1991; Lovley et al., 2004). Extracellular electron transfer to electrodes offers a novel strategy for harvesting electricity from organic wastes and renewable biomass (Lovley, 2006, 2008) as well as environmental restoration (Zhang et al., 2010).

The most intensively studied strain of G. sulfurreducens is strain DL-1. The finding that deletion of the gene encoding PilA, the structural protein for type IV pili, resulted in the loss of the extracellular filaments of DL-1 suggested that most of the filaments were type IV pili (Reguera et al., 2005). Since those studies, strains of G. sulfurreducens that produce substantially more filaments than strain DL-1 have been identified (Yi et al., 2009; unpublished data), and continued examination of the genome sequence has revealed additional genes that could potentially encode filament proteins other than PilA. Therefore, the composition of filaments in G. sulfurreducens was investigated further.

Materials and methods

Bacterial strains and culturing conditions

Geobacter sulfurreducens strain DL-1 (Caccavo et al., 1994) was obtained from our laboratory culture collection. Geobacter sulfurreducens strain MA was selected after routine subculturing of strain DL-1 in a medium with acetate as the electron donor and fumarate as the electron acceptor, and exhibited increased attachment to glass. Both strains were routinely cultured under anaerobic conditions in NB medium with acetate (10 mM) and fumarate (40 mM) as described previously (Coppi et al., 2001) at 30 °C. For experiments requiring the production of filaments or biofilms, strains were grown in an acetate–fumarate freshwater medium (Coppi et al., 2001) at 25 °C. When required, chloramphenicol was used at a concentration of 15 μg mL−1, spectinomycin at 75 μg mL−1, and kanamycin at 200 μg mL−1.

DNA manipulations and construction of mutant strains

Genomic DNA extractions from G. sulfurreducens were performed using a MasterPure Complete DNA and RNA purification kit (Epicenter Technologies, Madison, WI). Plasmid purification, PCR product purification, and gel extractions were carried out using QIAprep Spin miniprep kits, QIAquick PCR purification kits, and QIAquick gel extraction kits (Qiagen Inc., Valencia, CA), respectively. Routine DNA manipulations were performed according to Sambrook et al. (1989). Ligations were carried out using Quick T4 DNA ligase (New England Biolabs, Beverly, MA) or a TOPO TA cloning kit (Invitrogen, Carlsbad, CA). PCR amplifications for cloning purposes contained the Platinum Pfx DNA Polymerase (Invitrogen).

The pilA-MAΔ mutant strain was generated by introducing the previously described pilA-specific mutagenic fragment (Reguera et al., 2005) into the MA strain using a previously described protocol (Coppi et al., 2001). The double mutant pilA/oxpG-MAΔ was produced by electroporating an oxpG-specific mutagenic fragment (Mehta et al., 2006) into the pilA-MAΔ mutant. To produce the quadruple mutant, a mutagenic fragment containing a spectinomycin resistance gene flanked by the 501 bp upstream of the pilA gene and by the 595 bp downstream of GSU1497 was introduced into DL-1–MA to generate the mutant pilA/GSU1497-MAΔ. The components of the mutagenic fragment were produced by PCR using the primers specified in Supporting Information, Table S1, restriction digested using enzymes specific to restriction sites introduced by the primers, and ligated to the spectinomycin cassette. A second mutagenic fragment was produced by recombinant PCR (Murphy et al., 2000) and was introduced into pilA/GSU1497-MAΔ. This second mutagenic fragment contained a chloramphenicol resistance gene flanked by the 530 bp upstream of the oxpG gene and the first 15 bp of the gene, and by the 462 bp downstream of the GSU1777 ORF and the terminal 35 bp of GSU1777. The primers used in constructing this mutagenic fragment are listed in Table S1. The quintuple gene disruption mutant was generated by electroporating a GSU0326-specific mutagenic fragment into the quadruple mutant described above. The GSU0326-specific mutagenic fragment contains the region 489 bp upstream of GSU0326 and the first 6 bp of the gene, a kanamycin resistance gene, and 430 bp downstream of GSU0326 along with the last 18 bp of the gene. The components of the mutagenic fragment were produced by PCR using the primers specified in Table S1, and restriction digested using the specified enzymes, and ligated to the kanamycin resistance cassette. PCR was used to verify that all constructs were integrated at the targeted loci following their introduction into each respective G. sulfurreducens strain.

Strain characterization

For transmission electron microscopy (TEM), cells were negatively stained with 0.2% uranyl acetate and examined using a JEOL 100S TEM operating under standard conditions at an 80 kV accelerating voltage.

To verify that the pilA-MAΔ mutant does not produce PilA, whole-cell lysates were separated by 12.5% sodium dodecyl sulfate-polyacrylamide electrophoresis (SDS-PAGE), immunoblotted, and probed with PilA-specific antiserum (Mehta et al., 2006). Immunoreactive protein bands were visualized using the One-Step Western Kit (GeneScript Co., NJ) according to the manufacturer's directions. For the identification of candidate filament proteins, loosely bound proteins were sheared from cells by blending in a Waring blender at a low speed for 2 min. Proteins were precipitated with 45% ammonium sulfate, separated by 12.5% SDS-PAGE, and stained with Coomassie blue.

To assess the attachment of strains to glass surfaces, glass tubes on which biofilms had developed were incubated for 5 min in a crystal violet staining solution (1.0% in distilled water) and washed twice in isotonic wash buffer. The stain was then dissolved in ethanol and absorbance was measured at 570 nm as described previously (Reguera et al., 2005).

For the visualization of biofilm development, cells were grown in culture tubes containing glass coverslips. At the stationary phase, coverslips were removed, washed twice in an isotonic wash buffer, stained with Cyto9 (Molecular Probes, Eugene, OR), and imaged using a Leica TCS SP5 microscope (Leica Microsystems GmbH, Wetzlar, Germany) under 20, 63, and 100 times objectives (numerical aperture 0.7, 0.9, and 1.4, respectively). Images were processed and analyzed using leica las af software (Leica Microsystems GmbH).

Illumina sequencing, genome sequence assembly, and polymorphism identification

A library for resequencing was constructed according to the Illumina standard genomic DNA library construction protocol. The quality of the resulting library was assessed by cloning and sequencing (Sanger); ∼20 clones were assessed to confirm correct inserts and adaptor pairs before sequencing on an Illumina sequencer. Once the library quality was confirmed, the libraries were sequenced on an Illumina GAII sequencer according to Illumina's standard protocol.

The Illumina output for each resequencing run was first curated to remove any sequences containing a ‘.’, which denotes an undetermined nucleotide. We then used mosaikaligner ( to iteratively align reads to the G. sulfurreducens (AE017180.1) reference sequence, where, in each iteration, a limit was placed on the number of alignment mismatches allowed. This limit iteratively increased from 0 to 5, and unaligned reads were used as input to the next iteration that had a more lenient mismatch limit. An in-house script (available upon request) was then used to compile the read alignments into a nucleotide-resolution alignment profile. Consistency and coverage were then assessed to identify likely polymorphic locations. The locations at which coverage was >10 × and for which indels were observed or the count of an SNP was greater than twice the count of the reference-sequence-matching nucleotide were considered to be likely polymorphic locations. Each of these potential mutations was also identified in multiple (4–48) strain resequencing experiments in our database. Potential mutations that were identified in multiple strains were assumed to be false positives; this assumption was borne out by the results of follow-up Sanger sequencing of over 25% of the possible mutations.

Results and discussion

Pilus-like filaments of the pilAΔ mutant

A previous study (Reguera et al., 2005) indicated that the deletion of the gene for the type IV pilin protein PilA prevented filament production in the DL-1 genome strain of G. sulfurreducens. However, an additional study of this strain revealed that filaments with a length and a diameter similar to the type IV pili could be occasionally observed in a very small proportion of cells in this strain (Fig. 1a, b); most grids contained cells with no filaments. We speculate that the scarcity of filamented cells is what precluded their detection until now.

Figure 1.

 Pilus-like filaments (arrows) in the pilAΔ mutant strain of DL-1 (a, b), the hyper-filamented strain MA (c), and in the pilA-MAΔ mutant strain (d) as visualized by TEM. Cells were negatively stained with 0.2% uranyl acetate and examined using a JEOL 100S TEM operating under standard conditions at an 80 kV accelerating voltage. Scale bars: a=0.25 μm; b=0.125 μm; c, d=0.25 μm.

In order to further evaluate whether G. sulfurreducens might produce pilin-like filaments from proteins other than PilA, studies were conducted with the MA strain of G. sulfurreducens, which routinely produces more abundant filaments than strain DL-1 and thus provided a more convenient study system. Resequencing of the MA strain with Illumina sequencing technology failed to reveal any mutations, indicating that this strain did not differ from the wild-type DL-1 strain at the genomic level.

When pilA in strain MA was deleted, the PilA protein could no longer be detected (Fig. S1), but the pilA-deficient strain produced abundant filaments (Fig. 1d) that were morphologically indistinguishable from those produced by the wild-type strain MA (Fig. 1c). Treating the pilA-deficient cells with proteinase K removed most of the filaments and the few filaments that remained were generally shorter than those typically seen on untreated cells (Fig. 2). These results suggested that the filaments were proteinaceous.

Figure 2.

 Filaments are sensitive to proteinase K treatment, as evidenced by TEM. Following proteinase K treatment, most pilA-MAΔ cells display no filaments (a) although some short filaments (arrow) are apparent (b). Filaments are visible in the untreated control pilA-MAΔ cells (c). Experimental cells were incubated with 100 μg μL−1 proteinase K for 4 h at 30°C, while control cells were incubated under the same conditions in the absence of proteinase K. Scale bars=0.2 μm.

Identifying genes required for filament production

Proteins other than PilA can form pilin-like filaments in other microorganisms. Pseudopilins, which function in type II secretion, share sequence homology with the type IV pilins (Bally et al., 1992; Nunn & Lory, 1993; Pugsley, 1993). A number of pseudopilins form pilus-like filaments known as pseudopili. For example, overexpression of the psuedopilin protein PulG in Klebsiella oxytoca or Escherichia coli resulted in the production of bundled filaments of PulG (Sauvonnet et al., 2000). Overexpression of the pseudopilin genes xcpT (from Pseudomonas aeruginosa), gspG (from E. coli K12), epsG (from Vibrio cholerae), exeG (from Aeromonas hydrophila), or outG (from Erwinia chrysanthemi) in E. coli producing the pullulanse secretion of K. oxytoca resulted in the production of pseudopili (Vignon et al., 2003) as did overexpression of xcpT in P. aeruginosa (Sauvonnet et al., 2000).

The pseudopilin gene oxpG was identified previously in G. sulfurreducens, and shown to play a role in outer membrane protein secretion (Reguera et al., 2005; Mehta et al., 2006). Deletion of oxpG in the pilA-deficient MA strain had little impact on filament production (Fig. 3a).

Figure 3.

 Disruption of multiple pseudopilin genes results in decreased filament production, as visualized by TEM. While abundant filament production is apparent in the pilA/oxpGΔ mutant (a), few filaments are visible in cultures of the quadruple mutant ΔpilA/GSU1497/oxpG/GSU1777-MAΔ (b). Filaments are rarely seen in cultures of the pilA/GSU1497/oxpG/GSU1777/GSU0326Δ mutant (c). Scale bars=0.5 μm.

The blast program (blastp) revealed that the G. sulfurreducens genes GSU1777 and GSU0326 have high degrees of similarity to the pseudopilin gene xcpT (E values 1e-76, 1e-36, respectively). The deletion of neither GSU0326 nor GSU1777 along with the adjacent GSU1776 had any detectable impact on filament production (data not shown).

Another candidate gene was derived from a comparison of loosely bound, outer surface protein preparations from strain DL-1 and the highly filamented pilA-deficient MA strain (Fig. S2). Matrix-assisted laser desorption/ionization MS indicated that a band found in the pilA-deficient MA strain, but not in the DL-1 strain, contained the protein product of GSU1497, which is annotated as a hypothetical gene, and has no significant similarity to any known proteins. The deletion of GSU1497 resulted in a significant decrease of this protein (Fig. S1b). However, because the deletion of GSU1497 in strain MA or the pilA-deficient strain of MA had little impact on the production of filaments (data not shown), our data do not clearly support its involvement in filament production.

Because none of the single or the double gene disruptions resulted in significant inhibition of filament production, mutants deficient in multiple pilin and pseudopilin candidate genes were generated. Filament production was clearly reduced in the quadruple mutant pilA/1497/oxpG/1777-MAΔ (Fig. 3b) compared with the single mutant pilA-MAΔ (Fig. 1c) or the double pilA/oxpG-MAΔ mutant (Fig. 3a). The quintuple mutant pilA/1497/oxpG/1777/0326-MAΔ also produced filaments, but they were extremely rare (Fig. 3c); while 47% of pilA/1497/oxpG/1777-MAΔ cells had one or more pilus-like filaments, only 9% of pilA/1497/oxpG/1777/0326-MAΔ cells produced a filament. Because the disruption of multiple pseudopilin genes, along with the hypothetical gene GSU1497, inhibited filament production, it appears likely that the encoded proteins comprise, or are required for the production of, the filaments produced by the pilAΔ and pilA-MAΔ mutants. Further studies are underway in our laboratory to further characterize the specific roles of the pseudopilin genes in filament production. The genes involved in the production of the rare filaments associated with the quintuple mutant ΔpilA/1497/oxpG/1777/0326Δ also remain to be identified.

Filaments requirement for attachment

The deletion of pilA in the DL-1 strain slightly inhibited the attachment of cells to glass (Reguera et al., 2007) and had no impact on attachment to graphite (Nevin et al., 2009). In a similar manner, the deletion of pilA in strain MA did not affect attachment to glass culture tubes (Fig. 4a) or coverslips (Fig. 4b, c). Both strains formed morphologically similar biofilms on glass coverslips with pillars over 40 μm in height, and cells covered 77.3±9.4% (MA strain) or 86.0±3.0% (PilA-deficient MA) of the surfaces (Fig. 4b, c).

Figure 4.

 Strains defective in multiple pseudopilin genes exhibit decreased attachment to glass surfaces. Attachment to glass test tubes is higher in the WT MA strain and the pilA-MAΔ mutant than in the pilA/GSU1497/oxpG/GSU1777Δ (4Δ) and pilA/GSU1497/oxpG/GSU1777/GSU0326Δ (5Δ) mutants, as determined by crystal violet-binding assay (a). Values are expressed as a ratio of OD570 nm/OD600 nm normalized to planktonic density as described previously (Clark et al., 2007). The MA strain (b) and pilA-MAΔ (c) produced morphologically similar biofilms on glass coverslips, but the pilA/GSU1497/oxpG/GSU1777/GSU0326Δ mutant (d) failed to produce biofilms under the same conditions. Scale bars=50 μm.

However, the quadruple pilA/1497/oxpG/1777Δ mutant and the quintuple pilA/1497/oxpG/1777/0326Δ mutant were defective in attachment (Fig. 4a, d). The quintuple mutant formed a single monolayer of cells covering only 1.5±0.7% of the glass surface (Fig. 4d). These findings suggest that one or more of the non-PilA filaments are important for attachment, at least in the absence of PilA.


These results demonstrate that pilin-like filaments of G. sulfurreducens can be comprised of proteins other than PilA. Although these filaments look similar, the fact that they are composed of different proteins suggests that other properties may not be the same. For example, the conductivity of filaments, believed to be composed of PilA, is considered to allow PilA pili to act as conduits for extracellular electron transfer to Fe(III) oxides (Reguera et al., 2005) and electrodes (Reguera et al., 2006; Nevin et al., 2009). Whether any of the other filaments detected in this study are also conductive is not known. The finding that the MA strain described here and the recently described KN400 strain of G. sulfurreducens (Yi et al., 2009) produce substantially more filaments than the DL-1 strain, coupled with the possibility that different strains may produce different proportions of various filaments that look similar, but have other dissimilar properties, indicates that mere visual observation is insufficient to provide information on the composition of G. sulfurreducens filaments.


This research was supported by the Office of Science (BER), US Department of Energy, Cooperative Agreement No. DE-FC02-02ER63446, and Office of Naval Research Grant N00014-10-1-0084.