Identification of Genes Involved in Biofilm Formation and Respiration via Mini-Himar Transposon Mutagenesis of Geobacter sulfurreducens

Abstract

Electron transfer from cells to metals and electrodes by the Fe(III)-reducing anaerobe Geobacter sulfurreducens requires proper expression of redox proteins and attachment mechanisms to interface bacteria with surfaces and neighboring cells. We hypothesized that transposon mutagenesis would complement targeted knockout studies in Geobacter spp. and identify novel genes involved in this process. Escherichia coli mating strains and plasmids were used to develop a conjugation protocol and deliver mini-Himar transposons, creating a library of over 8,000 mutants that was anaerobically arrayed and screened for a range of phenotypes, including auxotrophy for amino acids, inability to reduce Fe(III) citrate, and attachment to surfaces. Following protocol validation, mutants with strong phenotypes were further characterized in a three-electrode system to simultaneously quantify attachment, biofilm development, and respiratory parameters, revealing mutants defective in Fe(III) reduction but unaffected in electron transfer to electrodes (such as an insertion in GSU1330, a putative metal export protein) or defective in electrode reduction but demonstrating wild-type biofilm formation (due to an insertion upstream of the NHL domain protein GSU2505). An insertion in a putative ATP-dependent transporter (GSU1501) eliminated electrode colonization but not Fe(III) citrate reduction. A more complex phenotype was demonstrated by a mutant containing an insertion in a transglutaminase domain protein (GSU3361), which suddenly ceased to respire when biofilms reached approximately 50% of the wild-type levels. As most insertions were not in cytochromes but rather in transporters, two-component signaling proteins, and proteins of unknown function, this collection illustrates how biofilm formation and electron transfer are separate but complementary phenotypes, controlled by multiple loci not commonly studied in Geobacter spp.

Geobacter sulfurreducens is a member of the metal-reducing Geobacteraceae family and was originally isolated based on its ability to transfer electrons from internal oxidative reactions to extracellular electron acceptors such as insoluble Fe(III) or Mn(IV) oxides (5). G. sulfurreducens is also able to use an electrode as its sole electron acceptor for respiration, a phenotype which has many possible biotechnological applications (28, 29), and serves as a useful tool for direct measurement of electron transfer rates (2, 31). As G. sulfurreducens was the first Geobacteraceae genome sequence available (34) and the only member of this family with a robust genetic system (7), it serves as a model organism for extracellular electron transfer studies.

The proteins facilitating electron transfer to insoluble Fe(III) oxides by individual Geobacter cells and how these cells interact in multicellular biofilms are not fully understood. Many genes implicated in Fe(III) and electrode reduction were identified based on proteomic and microarray analysis of cultures grown with fumarate versus Fe(III) citrate as a terminal electron acceptor (9, 15, 35). More recently, similar expression data from Fe(III) oxide and electrode-grown cultures have also become available (8, 12, 16). In most extracellular electron transfer studies, outer membrane proteins (such as c-type cytochromes) have been the focus (4, 23, 27, 32), leading to targeted knockout studies of at least 14 cytochromes to date.

To reduce an insoluble electron acceptor, Geobacter spp. must achieve direct contact with the substrate (36). While contact with small Fe(III) oxide particles may be transient, growth on Fe(III)-coated surfaces or electron-accepting electrodes requires biofilm formation (31, 39). For example, when G. sulfurreducens produces an exponentially increasing rate of electron transfer at an electrode, this demonstrates that all newly divided cells remain embedded in the growing, conductive biofilm (2, 31). Thus, in addition to the need for an array of outer membrane cytochromes, there is also a need for control of both cell-cell contact and cell-surface contact.

While a genetic system for G. sulfurreducens has been developed, conjugal transfer of a plasmid or a transposon has not been reported (7). The broad-host-range cloning vector pBBR1MCS-2 has previously been electroporated into G. sulfurreducens, but its mobilization capabilities were not utilized (7). Similarly, a number of suicide vectors have been identified for G. sulfurreducens, but none have been used to deliver transposons for mutagenesis. mariner-based transposon mutagenesis systems have been successful in a variety of Bacteria and Archaea, producing random insertions (20, 25, 40, 41, 43, 46, 48, 49). For example, genes involved in Shewanella oneidensis cytochrome maturation were discovered using the modified transposon mini-Himar RB1 (3).

In this work, we describe a system for the conjugal transfer of the pBBR1MCS family of plasmids from Escherichia coli to G. sulfurreducens, which allowed transposon mutagenesis based on pMiniHimar RB1. Under strictly anaerobic conditions, a library of insertion mutants was constructed and screened to identify genes putatively involved in attachment and Fe(III) citrate reduction. Approximately 8,000 insertion mutants were isolated, with insertions distributed throughout the G. sulfurreducens chromosome. Subsequent characterization revealed mutants defective in metal reduction but unaffected in all aspects of electrode reduction, as well as mutants able to reduce metals but incapable of electrode reduction. These observations greatly expand the list of Geobacter mutants with defects in respiration or biofilm formation, and this library serves as a resource for further screening of extracellular electron transfer phenotypes.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

Bacterial strains and plasmids used in this study are described in Table 1. G. sulfurreducens PCA (ATCC 51573) was grown anaerobically at 30°C in a vitamin-free minimal medium containing 20 mM acetate as the electron donor and 40 mM fumarate as the electron acceptor (31). G. sulfurreducens had been maintained in medium containing 100 mM ferrihydrite as the electron acceptor to retain strong biofilm and metal reduction phenotypes. For G. sulfurreducens transposon mutants, 0.01% (wt/vol) Trypticase and 200 μg/ml kanamycin were added to growth medium. E. coli WM3064 (42) carrying pMiniHimar RB1 was grown in LB broth containing 50 μg/ml kanamycin and 30 μM 2,6-diaminopimelic acid (DAP) at 37°C. E. coli WM3064 carrying pBBR1MCS was grown in LB broth containing 34 μg/ml chloramphenicol and 30 μM DAP at 37°C, while 50 μg/ml kanamycin was added to cultures of E. coli WM3064 carrying pBBR1MCS-2, 100 μg/ml ampicillin was added to cultures carrying pBBR1MCS-4, and 10 μg/ml gentamicin was added to cultures carrying pBBR1MCS-5 (18, 19). G. sulfurreducens carrying pBBR1MCS was grown in anaerobic minimal medium containing 10 μg/ml chloramphenicol, while 200 μg/ml kanamycin was added to cultures of G. sulfurreducens carrying pBBR1MCS-2, 400 μg/ml ampicillin was added to cultures containing pBBR1MCS-4, and 20 μg/ml gentamicin was added to cultures containing pBBR1MCS-5.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant characteristics or description	Source or reference
Strains
G. sulfurreducens	Wild type (ATCC 51573)	5
E. coli
WM3064	Donor strain for conjugation: thrB1004 pro thi rpsL hsdS lacZΔM15 RP4-1360 Δ(araBAD)567 ΔdapA1341::[erm pir(wt)]	42
DH5α	Host for E. coli cloning	Invitrogen
Plasmids
pMiniHimar RB1	Plasmid carrying mini-Himar RB1; oriR6K oriT lacZ Kmr	3
pBBR1MCS	Mobilizable broad-host-range plasmid; lacZ Cmr	18
pBBR1MCS-2	Mobilizable broad-host-range plasmid; lacZ Kmr	19
pBBR1MCS-5	Mobilizable broad-host-range plasmid; lacZ Gmr	19
pGCOMP2505	GSU2505 in MCS of pBBR1MCS, Cmr	This study
pGCOMP1501	GSU1501 in MCS of pBBR1MCS-5, Gmr	This study
pGCOMP3361	GSU3361 in MCS of pBBR1MCS-5, Gmr	This study

Conjugal transfer of plasmid.

Wild-type G. sulfurreducens was used in filter mating experiments with E. coli WM3064 carrying pBBR1MCS, pBBR1MCS-2, pBBR1MCS-4, or pBBR1MCS-5 (18, 19). G. sulfurreducens was grown to an optical density at 600 nm (OD₆₀₀) of 0.3 to 0.4. The donor E. coli strain was grown overnight and then washed and resuspended in fresh LB to remove antibiotics. Cultures were mixed aerobically in a 1:1 ratio and vacuum filtered on a 0.45-μm-pore filter (Millipore). This filter was incubated on either an LB plate or a vitamin-free minimal medium plate supplemented with fumarate, acetate, and Trypticase for 1 to 24 h in a MACS MG500 anaerobic workstation (Don Whitley Scientific Limited, England) containing 5% H₂, 20% CO₂, and 75% N₂ gas. Filters were then suspended in 3 ml growth medium and vortexed to remove cells. Cells were plated and selected on growth medium using the appropriate antibiotic.

Transposon mutagenesis.

Wild-type G. sulfurreducens was used in filter mating experiments with E. coli WM3064 carrying pMiniHimar RB1 (3). G. sulfurreducens was grown to an OD₆₀₀ of 0.3 to 0.4. E. coli WM3064 was grown overnight in LB supplemented with 50 μg/ml kanamycin and 30 μM DAP, washed, and resuspended in LB. Cultures were mixed aerobically in a 1:1 ratio and vacuum filtered on a 0.45-μm-pore filter (Millipore). Mating filters were incubated anaerobically on minimal medium containing 0.01% (wt/vol) Trypticase for 4 h in a MACS MG500 anaerobic workstation (Don Whitley Scientific Limited, England) containing 5% H₂, 20% CO₂, and 75% N₂ gas. Filters were then placed in 3 ml growth medium and vortexed to remove cells. The cell suspensions were incubated with occasional shaking for 2 h and then further diluted in growth medium, plated on minimal medium containing 20 mM acetate, 0.01% (wt/vol) Trypticase, and 200 μg/ml kanamycin, and incubated anaerobically at 30°C for 1 week. Isolated colonies were picked anaerobically into 96-well plates containing growth medium supplemented with 20 mM acetate, 0.01% (wt/vol) Trypticase, and 200 μg/ml kanamycin and allowed to grow for 1 week prior to phenotype screening. In total, over 8,000 transposon mutants were cultured in 96-well plates and stored at −80°C. Frozen plates (stored at −80°C) could be revived and rescreened for other phenotypes [such as Fe(III) oxide reduction] after months of storage.

Phenotype screening.

A bolt replicator (V&P Scientific, San Diego, CA) was used to transfer transposon mutants from a master plate containing growth medium to 96-well, flat-bottom polystyrene microtiter plates (Nunc 167008) containing medium specific for auxotroph, biofilm, or Fe(III) reduction screening. Mutants with phenotypes of interest (described below) were streaked for isolation on agar plates to ensure culture purity and rescreened to verify phenotype. The OD₆₀₀ was monitored for 72 h to detect any growth defects, and a third phenotype screening was performed. Mutants were then transferred to vitamin-free minimal medium containing 25 mM acetate and approximately 100 mM Fe(III) oxide to test for the ability to reduce insoluble iron.

Screening. (i) Auxotroph screen.

To screen for amino acid auxotrophs, insertion mutants were transferred to medium lacking Trypticase and the OD₆₀₀ was measured to identify mutants that were unable to grow in the absence of Trypticase.

(ii) Biofilm formation assay.

To screen for attachment phenotypes, insertion mutants were grown with minimal medium containing 30 mM acetate for 72 h at 30°C. To identify mutants with low or high attachment phenotypes compared to that of the wild type, crystal violet (CV) biofilm assays were performed using a modification of a previously described protocol (37). Cells were stained with 200 μl of a 0.01% (wt/vol) CV solution and allowed to incubate for 15 min before wells were rinsed to remove unattached cells. Remaining cells were dried for 20 min at room temperature, and then the CV solution was solubilized with 200 μl of 100% dimethyl sulfoxide. The OD₆₀₀ was measured to identify mutants with low or high growth compared to that of the wild type. Prior to screening, electron donor and acceptor concentrations were optimized for G. sulfurreducens surface attachment. Acceptor limitation (30 mM acetate, 40 mM fumarate) resulted in greater attachment to the wells of a 96-well plate than donor limitation.

(iii) Fe(III) reduction.

To screen for the inability to reduce Fe(III) citrate, insertion mutants were transferred to minimal medium containing 10 mM acetate as the electron donor and 55 mM Fe(III) citrate as the electron acceptor. Clearing of the medium was monitored as an indication of Fe(III) citrate reduction.

In addition, sequenced transposon mutants were tested for the ability to reduce Fe(III) oxide. Mutants were transferred to minimal medium containing 20 mM acetate as the electron donor and 100 mM ferrihydrite as the electron acceptor. After two successive transfers (to remove potentially chelating citrate that could rescue certain mutations), cultures were inoculated into the same medium and samples were collected over 10 days and diluted 10-fold in 0.5 N HCl. To monitor production of Fe(II) over time, 50-μl samples were analyzed in microtiter wells with 300 μl of 2 g/liter ferrozine in 100 mM HEPES buffer, followed by A₅₆₂ measurement (30).

DNA sequencing.

For sequencing of transposon insertion sites directly from G. sulfurreducens genomic DNA, chromosomal DNA was isolated from mini-Himar RB1 mutants after overnight growth using the Wizard genomic DNA purification kit (Promega, Madison, WI). Purified DNA was precipitated in ethanol and resuspended in nuclease-free water. Sequencing was performed at the Biomedical Genomics Center at the University of Minnesota. Approximately 3 to 4 μg of DNA was submitted as a template with 12 pmol of primer DRB05-27 (TGACGAGTTCTTCTGAGCGG), using the following cycling parameters: 95°C for 5 min, followed by 99 cycles of 95°C for 30 s, 55°C for 20 s, and 60°C for 4 min.

Identification of transposon insertion site.

To determine the site of mini-Himar RB1 insertion, DNA sequences obtained from transposon mutant genomic DNA were compared to the complete genome of G. sulfurreducens (available at the Department of Energy Joint Genome Institute [JGI; http://www.jgi.doe.gov/]) using the BLASTN algorithm. The site of insertion in each mutant was confirmed using PCR analysis with primers designed to amplify DNA spanning the insertion.

Construction of complemented mutants.

GSU2505 was amplified from wild-type G. sulfurreducens using primers NHLF (GCAAGCTTATGAGACAAATCGGCAACCG; the HindIII site is underlined) and NHLR (TAGGATCCTCAGTCGTGCGTTACCTTGA; the BamHI site is underlined), GSU1501 was amplified from wild-type G. sulfurreducens using primers 1501S (CGAAGCTTATGGGTACGTTCATCAATGG; the HindIII site is underlined) and 1501E (TAGGATCCTCATTCCGGCCCGTTAGACT; the BamHI site is underlined), and GSU3361 was amplified using primers 3361S (TTAAGCTTTTGCGTGCCGTTCGACGAGAG; the HindIII site is underlined) and 3361E (TAGGATCCCTACCTCATCTCAACCACCC; the BamHI site is underlined). The following conditions were used for each: 30 cycles of 94°C for 1 min, 55°C for 1 min, and 72°C for 1.5 min, and then a final extension at 72°C for 10 min. Each product was digested with HindIII and BamHI. The GSU2505 produced was inserted into the HindIII and BamHI sites of pBBR1MCS, creating the vector pGCOMP2505. This construct was mated into the GSU2505 mutant (G. sulfurreducens with insertion coordinate 2762032) and selected on chloramphenicol. The GSU1501 product was inserted into the HindIII and BamHI sites of pBBR1MCS-5, creating the vector pCGOMP1501. This construct was mated into the two GSU1501 mutants (G. sulfurreducens with insertion coordinate 1646730 or 1646910) and selected on gentamicin. The GSU3361 product was inserted into the HindIII and BamHI sites of pBBR1MCS-5, creating the vector pCGOMP3361. This construct was mated into the two GSU3361 mutants (G. sulfurreducens with insertion coordinate 3693699 or 3694136) and selected on gentamicin.

Electrochemical analysis.

Carbon electrodes were polished using P1500 grit sandpaper (3M, Minneapolis, MN) and prepared as previously described (31). Bioreactors containing a carbon working electrode, a platinum counter electrode, and a saturated calomel reference electrode connected via a salt bridge were prepared as previously described (31). Sterile growth medium was added to autoclaved bioreactors, with anaerobic conditions generated by a constant flow of humidified N₂/CO₂ (80:20 [vol/vol]). Reactors were placed in a 30°C water bath and connected to a 16-channel potentiostat (VMP; Bio-Logic, Knoxville, TN) with software (EC-Lab v9.41) able to run differential pulse voltammetry, cyclic voltammetry, and chronoamperometry (CA) as previously described (31). Bioreactors were inoculated with 50% (vol/vol) of G. sulfurreducens approaching stationary phase (OD₆₀₀, 0.40 to 0.55) and incubated for 4 to 120 h with a potential of 0.24 V versus the standard hydrogen electrode (SHE) applied.

Confocal analysis.

A Nikon C1 spectral imaging confocal microscope (Nikon, Japan) was used to image biofilm-covered electrodes. Immediately after harvest, biofilms were washed with growth medium and stained with propidium iodide and SYTO 9 from a LIVE/DEAD BacLight bacterial viability kit (Invitrogen Corp., Carlsbad, CA). SYTO 9 is membrane permeable and stains cells in a population green, while propidium iodide stains cells with damaged membranes red. Electrodes in medium were placed on microscope slides with the coverslips elevated above the thickness of the electrode and viewed using the 488-nm and 561-nm lasers.

Biomass measurement.

Biomass attached to electrodes was determined using the Pierce bicinchoninic acid protein assay kit (Thermo Scientific, Rockford, IL). Immediately after electrochemical analysis, carbon working electrodes were dipped in growth medium to remove planktonic cells. The carbon working electrode was then removed from the platinum wire and incubated in 1 ml of 0.2 M NaOH at 96°C for 20 min to remove the attached biomass. The protein present in the 1-ml NaOH sample was measured, indicating the biomass attached to the electrode.

RESULTS

Development of a transposon mutagenesis system in G. sulfurreducens.

A system for the conjugal transfer of plasmids into Geobacter spp. has not been described. However, the broad-host-range plasmid pBBR1MCS-2, which has been electroporated into G. sulfurreducens (7), can be mobilized by E. coli WM3064, as this DAP auxotroph contains mobilization genes on its genome. Filter matings between E. coli WM3064 carrying pBBR1MCS-2 and G. sulfurreducens were successful when filters were incubated on either LB or Geobacter medium plates, with the highest numbers of transconjugants (>10⁷ from a mating of 10⁹ G. sulfurreducens cells) occurring when matings were performed in an anaerobic chamber on Geobacter medium supplemented with Trypticase. Related plasmids (pBBR1MCS-1, pBBR1MCS-4, and pBBR1MCS-5) were also transferred into G. sulfurreducens using the appropriate antibiotic selection.

G. sulfurreducens is typically described as a strict anaerobe, but recent data have shown an ability to both tolerate and even utilize small amounts of oxygen (26). This tolerance was exploited in developing a protocol in which Geobacter cultures were captured on a filter with E. coli. The filter could be aerobically rinsed free of antibiotics and other medium, placed on solid medium, and transferred into an anaerobic chamber. Anaerobic incubation beyond 4 h did not significantly increase the number of transconjugants. After matings, cells were recovered on selective medium, and putative transconjugants were picked and inoculated into liquid medium (in 96-well plates) in an anaerobic chamber.

Matings between G. sulfurreducens and E. coli WM3064 carrying pMiniHimar RB1 (a nonreplicating plasmid in G. sulfurreducens) were performed using the conditions developed for conjugal transfer (4-h anaerobic filter matings; anaerobic colony picking), producing up to 10⁵ mutants (containing insertion of the 2.2-kb transposable element) in a single mating. Some preliminary sequencing using genomic DNA to identify the location of transposon insertions produced mixed sequencing data, suggesting either mixed cultures or clonal cultures carrying multiple insertions. Thus, after mating, filters were shaken gently in kanamycin-free medium for 2 h (less than half the doubling time of G. sulfurreducens) to facilitate separation of cells before dilution and plating. In addition, all mutants were reisolated from single colonies and phenotypes were reverified before sequencing. These steps dramatically reduced the occurrence of mutants with ambiguous sequencing results. If cultures produced poor sequencing data or insertion sites were uncertain in any way, mutants were discarded prior to characterization. Transposon locations determined via direct sequencing of genomic DNA averaged ∼900 bp, which was sufficient to locate the insertion site. Insertions were verified via PCR analysis, using primers spanning the putative insertion site.

Construction and screening of a mini-Himar RB1 insertion library in G. sulfurreducens.

To verify the method's ability to produce insertional mutations with identifiable phenotypes, mutants were screened for an inability to grow in the absence of Trypticase. Putative auxotrophs were restreaked and reisolated, and DNA from these cultures were used to test protocols for identifying the transposon insertion site. For example, a mutant (13A2) with a transposon insertion in GSU3097 was identified that was unable to grow in the absence of Trypticase. This gene was annotated as hisH (a component of imidazole glycerol phosphate synthase) in a cluster of genes putatively involved in histidine biosynthesis (Fig. 1A). PCR analysis of this region confirmed the presence of a 2.2-kb insertion in this mutant (data not shown). Addition of histidine restored growth, which was consistent with a defect in histidine biosynthesis (Fig. 1B). No mutants recovered from the auxotroph screen were in similar regions of the genome, suggesting random incorporation of the transposable element (Table 2).

FIG. 1.

(A) View of G. sulfurreducens chromosomal DNA with coordinates 3399456 to 3405516 showing location of transposon insertion for the GSU3097 mutant, located among putative histidine biosynthesis genes (hisF, imidazole glycerol phosphate synthase subunit; hisA, phosphoribosylformimino-5-aminoimidazole carboxamide ribotide isomerase; hisH, imidazole glycerol phosphate synthase subunit; hisB, imidazoleglycerol-phosphate dehydratase; hisC, histidinol phosphate aminotransferase; hisD, histidinol dehydrogenase; and hisG-2, ATP phosophribosyltransferase). (B) Growth curves (representative of three replicates) for the GSU3097 mutant in the presence and absence of 0.01% (wt/vol) Trypticase or supplemented with 0.01% (wt/vol) histidine. Lines connect data points collected automatically every 2.75 h by a spectrophotometer housed inside an anaerobic glove box.

TABLE 2.

pMiniHimar transposon mutants isolated in this study

Mutant	GSU locusa	Annotationb	Coordinatec	Attachmentd	Fe(III) citratee	Fe(III) oxidef
13A2	3097	hisH	3401903		+	+
13D8	3253	Response regulator	3566024	+	−	+
51G10	0785	Nickel-dependent hydrogenase, large subunit	846918	+	−	+
54B11	0782	Nickel-dependent hydrogenase, small subunit	843038	+	−	+
21B3	0685	Radial SAM domain protein	722345	+	−	+
48B1	1074	Hypothetical protein	1163782	+	−	+
56F4	1197	RNA methyltransferase	1300712	+	−	+
58F8	0506	Methylamine utilization protein	537753	+	−	+
59C5	0007	PAS/PAC signal transduction histidine kinase	11967	+	−	+
63E10	1330	Metal ion efflux protein	1455965	+	−	+
70B1	0274	Cytochrome c family protein	283160	+	−	+
70B4	1486	MttB family protein, TatC	1629929	+	−	+
2C3	2505g	NHL repeat domain protein	2762032	−	+	+
7A2	1012	Hypothetical protein	1094418	−	+	+
13B6	0881	Sensor histidine kinase	941495	−	+	+
15E9	1999	Hfq protein	2191532	−	+	+
47D9	1928	Sensor histidine kinase/response regulator	2109312	−	+	+
63H3	2889	Hypothetical protein	3175621	−	+	+
70A12	1665	Rhomboid family protein	1829414	−	+	+
71F12	1508	Hypothetical protein	1653572	−	+	+
79G11	2759	Potassium efflux system protein	3038775	−	+	+
13E5	1492	Twitching motility protein PilT	1636675	++	+	+
20B5	1501	ABC transporter, ATP binding protein	1646730	++	+	−
20C6	2948	Hypothetical protein	3247587	++	+	+
21H3	0170	GreA/GreB family protein	187315	++	+	+
22G11	1499	Hypothetical protein	1645371	++	+	−
26D6	2898	High-molecular-weight cytochrome c	3186221	++	+	+
26E12	1226	Hypothetical protein	1329442	++	+	+
33B8	3361	Transglutaminase domain protein	3693699	++	+	+
34E4	3129	MATE efflux family protein	3430078	++	+	−
36H2	0351	NADH dehydrogenase 1, N subunit	381393	++	+	+
61C5	1891	Response regulator	2068995	++	+	+
62F6	1891	Response regulator	2068944	++	+	+
67B2	1501	ABC transporter, ATP binding protein	1646910	++	+	−
68D1	0599	Sensor histidine kinase	631524	++	+	+
68H10	2925	Hypothetical protein	3225163	++	+	+
72B5	1432	TPR domain protein	1569891	++	+	+
73F4	3361	Transglutaminase domain protein	3694136	++	+	+
77B8	3438	Hypothetical protein	3783890	++	+	+

^aGSU locus designation assigned by JGI for the open reading frame disrupted by the pMiniHimar RB1 transposon.

^bAnnotation assigned by JGI.

^cCoordinate of the transposon insertion in the G. sulfurreducens genome.

^dHigh (++) or low (−) levels of biofilm adhesion in crystal violet assay compared to that of the wild type (+).

^eAbility to reduce Fe(III) citrate.

^fAbility to reduce Fe(III) oxide.

^gInsertion just upstream of the open reading frame.

In this report, we detail results from two phenotypic screens, attachment and Fe(III) citrate reduction. Mutants were screened for deficiencies in adhesion to 96-well plates using a CV assay (37), following incubation under electron acceptor-limited conditions. After a determination of the OD₆₀₀ (to identify low-growth cultures which could produce false negatives), mutants with attachment levels as high as 275% and as low as 35% of that of the wild type were identified (Fig. 2). To identify genes involved in Fe(III) reduction, mutants were cultivated in Fe(III) citrate and detected based on the color difference between Fe(III) and fully reduced Fe(II) after 10 days.

FIG. 2.

Biofilm formation by G. sulfurreducens attachment mutants, based on CV staining of cells adherent to 96-well plates after 72 h of growth in fumarate-limited medium in an anaerobic glove box. The mean absorbance of each mutant in a minimum of three trials is expressed relative to that for wild-type G. sulfurreducens (mean absorbance, ∼0.39). Error bars are the standard errors of the mean for three replicates.

Approximately 500 cultures were initially identified with detectable alterations in these two phenotypes. After a second round of reisolation, rescreening, sequencing, and PCR verification, 39 mutants were retained with strong, consistent phenotypes of interest to the current study. An additional factor in choosing mutants for further study was measurement of growth rates and yield (OD₆₀₀) in 96-well plates using a spectrophotometer housed in an anaerobic chamber (Fig. 3). This additional screen allowed for identification of mutants with general metabolic deficiencies. An example of a mutant with markedly slower growth and final OD is shown in Fig. 3 (containing an insertion in GSU0170), compared to other mutants which demonstrated nearly wild-type behavior. While some genes or gene clusters were disrupted more frequently than others, no mutant was isolated with an identical insertion, and alignment of insertion regions from 15 sequenced mutants revealed only a TA motif at the insertion site (as noted in references 21 and 22). A list of phenotypes and corresponding sites is shown in Table 2.

FIG. 3.

Growth curves showing screening process used to identify metabolic defects in the initial library of transposon mutants. Wild-type (WT) G. sulfurreducens is shown in red, and a selection of transposon mutants discussed in subsequent sections are shown, including GSU1330 (orange), GSU1501 (yellow), GSU2505 (purple), GSU3097 (green), and GSU3361 (blue). Examples of mutants with growth rates similar to those of the wild type are shown in black, while a mutant (GSU0170) with a decreased growth rate is shown in brown.

Electrochemical analysis and confocal microscopy of selected transposon mutants.

The screening tools used to discover mutants were based on general consequences of respiration or cell surface alterations. However, electron transfer to solid electron acceptors by Geobacter spp. is a complex process requiring initial attachment, followed by proper networking of daughter cells within the biofilm matrix (39). Improper localization of proteins on cell surfaces can also interfere with reduction of insoluble electron acceptors (33), and these intertwined phenotypes are difficult to study using heterogeneous electron acceptors such as Fe(III) oxides. Use of an electrode surface as an electron acceptor offers a consistent, geometrically defined surface, which can be poised at a fixed potential to act as a homogenous electron acceptor. Defects in initial attachment can be separated from defects in cell-cell attachment, and both biofilm characteristics and respiration rates can be directly measured.

First, wild-type G. sulfurreducens was analyzed by inoculating cultures into bioreactors containing polished carbon electrodes, which were poised to act as electron acceptors at +0.24 V versus SHE. After inoculation of a wild-type electron acceptor-limited culture, CA revealed an exponentially increasing rate of electron transfer to the electrode, as has been described previously (2, 31). To prepare a baseline set of images for comparison to mutant cultures, electrodes were harvested during growth and imaged using confocal laser scanning microscopy (CLSM). The use of polished (P1500 grit), flat electrodes caused biofilms to be highly similar at all imaged locations, with the exception of sites at the very edge of the electrode, near the magnetic stir bar. For each electrode, multiple stacks near the center of each electrode were collected (two independent electrodes analyzed per time point), and representative images are shown spanning early to late growth.

A series of CLSM images collected over a span of 120 h allowed for a demonstration of events occurring as current production rates at electrodes increased (Fig. 4). Within the first 24 h of colonization, electrodes appeared completely covered with at least a monolayer of cells. As biofilms became thicker, films remained uniform, with no clear “pillars” or higher-order structures appearing. During the first 48 h, the rate of electron transfer to electrodes increased exponentially, indicating that each new layer of cells attached to the growing biofilm was also capable of donating electrons to the electrode. After 72 h, the rate of electron transfer did not increase further, as has been reported previously (31). However, the biofilm continued to thicken over the next 48 h, indicating that beyond a certain level, additional cells could not contribute to additional electron transfer to the surfaces.

FIG. 4.

CLSM of wild-type biofilm formation by G. sulfurreducens grown using a carbon electrode as the electron acceptor (+0.24 V versus SHE). Cells are shown at 4 h (A), 18 h (B), 72 h (C), and 120 h (D) after inoculation. Images similar to wild-type film formation at intermediate time points can be seen in Fig. 5G and H and Fig. 6D and E. Top panels are maximum projections (bar = 20 μm); bottom panels are side projections (bar = 10 μm). “Live” cells stain green, while permeable cells stain red.

Mutants with strong biofilm or Fe(III) reduction phenotypes were then analyzed using the same electrochemical approach and imaged using CLSM. In addition, voltammetry data (cyclic voltammetry and differential pulse voltammetry, described previously [31]) were collected for each mutant for comparison with wild-type behavior. Each mutant was grown in at least four independent replicate electrode experiments, with representative data and confocal images shown in Fig. 5 and 6.

FIG. 5.

Representative CLSM images and CA of biofilm formation by G. sulfurreducens transposon mutants on carbon electrodes poised at +0.24 V versus SHE. Shown are the GSU1501 mutant at 24 h (A) and 48 h (B), the GSU3361 mutant at 24 h (D) and 60 h (E), and the GSU1330 mutant at 24 h (G) and 60 h (H). Top panels are maximum projections (bar = 20 μm); bottom panels are side projections (bar = 10 μm). “Live” cells are green, while permeable cells are red. CA for each mutant is shown in the right column (C, F, and I) along with wild-type G. sulfurreducens. Arrows indicate CLSM time points.

FIG. 6.

Representative CLSM images of biofilm formation of the GSU2505 mutant and complemented strain on carbon electrodes poised at +0.24 V versus SHE. Both strains were imaged at 24 h (A and D) and 72 h (B and E). Top panels are maximum projections (bar = 20 μm); bottom panels are side projections (bar = 10 μm). “Live” cells are green, while permeable cells are red. CA for GSU2505 mutant (C) and complemented (D) strain is shown in comparison to wild-type G. sulfurreducens. Arrows indicate CLSM time points.

For example, two separate mutants were identified in the CV assay as having a high level of attachment (∼260% that of the wild type) but no defect in Fe(III) citrate reduction. Both of these mutations mapped to separate regions of open reading frame GSU1501. This gene encoded a putative ATP-binding protein in a gene cluster containing a permease domain for a hypothetical ABC transporter. While the GSU1501 mutant was identified based on a strong attachment phenotype (in 96-well plates), it failed to demonstrate any significant capacity for electron transfer to the electrode surface (maximum of 10 μA, compared to nearly 900 μA in the wild type) (Fig. 5C). Both insertion mutants were tested separately for growth on electrodes and produced identical phenotypes.

Imaging of electrodes at different time points showed very few GSU1501 mutant cells attached to electrodes at 24 and 48 h in contrast to that for wild-type G. sulfurreducens (Fig. 4 and 5A and B). In addition, most mutant cells attached to the electrodes were permeable to the propidium iodide stain (Fig. 5), which suggested that these cells were not viable. When biomass was recovered from electrodes and used to express electron transfer rates as a function of the attached protein, the GSU1501 mutants produced a maximum of only 100 μA/mg protein compared to a maximum of over 4,000 μA/mg protein produced by wild-type cultures. Consistent with this strong defect in electron transfer to an electrode surface was the observation that these mutants were unable to reduce Fe(III) oxides [even though this mutant reduced Fe(III) citrate at wild-type levels].

While the same open reading frame (GSU1501) was disrupted in two separate mutants, complementation with this gene alone did not restore the wild-type phenotype. When GSU1501 was expressed in trans with a constitutive promoter, complemented strains increased their maximum rate of current production only marginally (∼20 μA maximum) (data not shown). The presence of multiple genes upstream and downstream of GSU1501, including additional ATP binding cassette subunits, suggested that the insertion was polar, producing the observed phenotype as the sum of multiple defects in this cluster.

Another gene identified more than once in the initial screen, based on a high level of attachment in the CV assay (230% of that of the wild type), also demonstrated unexpected behavior when grown as a biofilm with an electrode as the electron acceptor. Both mutants contained insertions located in different sites within a gene (GSU3361) encoding a protein putatively attached to the inner membrane by a single transmembrane domain, with a periplasmic domain bearing similarity to transglutaminases. After 24 h of growth, each of these mutants colonized electrode surfaces and demonstrated electron transfer rates similar to those of the wild type (Fig. 5D).

However, in all incubations (for both mutants), there was a sudden decrease in electron transfer rates by 48 h (Fig. 5F). Imaging of biofilms prior to this sudden decrease showed films of similar thickness and morphology to wild-type cultures, but imaging of biofilms immediately after the decrease revealed that the layer of cells closest to the electrode was permeable to propidium iodide. This indicated a sudden event which damaged the membranes of cells closest to electrodes, triggered at a time when biofilms reached an electron transfer rate ∼50% of the maximum (Fig. 5E). This event was highly repeatable in both mutants, and biofilms could not be restored to high rates of electron transfer after this precipitous drop occurred (e.g., by addition of an electron donor, changing of the medium, or further incubation).

Complementation with only GSU3361 in trans with a constitutive promoter produced cultures which did not display this catastrophic decline in electron transfer. The current production rate decreased as cultures approached ∼50% of wild-type rates and only reached current densities of ∼70% of the wild type (see Fig. S1 in the supplemental material). Confocal images of these complemented strains also confirmed that cells throughout the biofilm consistently stained as viable. However, biofilms formed by the complemented strains appeared more compact at the base and extended in tufts for as much as 40 μm beyond the base of the biofilm. This intermediate response, which was not seen in any other wild-type or mutant culture, could be due to improper expression levels of GSU3361 from the constitutive promoter or residual polar effects from the small hypothetical gene downstream of GSU3361.

An example of a mutant which demonstrated a strong phenotype in liquid culture, but not on electrode surfaces, contained an insertion in GSU1330. This mutant was initially identified as having a decreased ability to reduce Fe(III) citrate. As GSU1330 was similar to outer membrane metal ion efflux proteins [most similar to Cu(I)/Ag(I) RND exporters] (10, 13), the kinetics of metal reduction were investigated further. In liquid culture, metal reduction slowed as Fe(II) accumulated above 10 mM and ceased at levels above 35 mM; however, this mutant was able to reduce Fe(III) oxide at wild-type levels and rates (data not shown). As free Fe(II) does not accumulate in Fe(III) oxide medium, but rather adsorbs to Fe(III) surfaces and leads to formation of magnetite (36), this further argued that GSU1330 was part of an Fe(II) export system. Consistent with a role for this protein in metal ion export rather than respiration, this mutant colonized electrodes and demonstrated peak electron transfer rates similar to those of wild-type cultures (Fig. 5G to I).

Some mutants originally identified as having a decreased level of attachment in the CV assay also had altered respiratory phenotypes. A mutant containing an insertion upstream of GSU2505 (part of a cluster of outer membrane c-type cytochromes including OmcS [GSU2504] and OmcT [GSU2503]) (32), was able to reduce Fe(III) citrate, but electrochemical analysis showed a defect in the maximum rate of electron transfer. While wild-type cultures always produced maximum rates of >800 μA/electrode, the GSU2505 mutant produced a maximum value of approximately 200 μA (Fig. 6C). Unlike the behavior of mutants such as the GSU1501 mutant, CLSM imaging showed only slightly altered levels of attachment and biofilm formation on electrodes compared to those of the wild type, suggesting that the defect was not related to the ability of cells to attach to surfaces or to each other but to transfer electrons within films (Fig. 6B). When corrected for attached biomass, the GSU2505 mutant generated a maximum of approximately 1,500 μA/mg protein, also consistent with high levels of attachment but poor rates of long-distance electron transfer.

Unlike the other mutants, the insertion site of the GSU2505 mutant was upstream of an open reading frame rather than within a coding sequence. DNA containing only GSU2505, the first gene downstream of the insertion, was expressed in trans with a constitutive promoter in the transposon mutant. Complementation with this fragment restored electron transfer and also produced biofilms with similar morphologies (less dense and clumped cells) (Fig. 6D to F). Based on biomass recovered from electrodes, the rate of electron transfer in the complemented mutant increased to 3,800 μA/mg protein, nearly that of the wild type.

The mutant containing an insertion upstream of GSU2505 was the only strain which also demonstrated altered phenotypes in the more detailed electrochemical analyses (e.g., low-scan-rate cyclic voltammetry). Wild-type G. sulfurreducens has been shown to respond to applied potential in what is commonly described as a wave, rising steeply at a characteristic midpoint potential at all stages in biofilm growth (31). When the derivative of this wave is plotted, three peaks are evident in wild-type biofilms at approximately −0.25 V, −0.15 V, and −0.05 V. The GSU2505 mutant demonstrated a similar midpoint potential and peak width at half the height of the primary catalytic feature at −0.15 V (Fig. 7). However, differences in the two other regions of the current-potential curve were evident. Both features were regained in the complemented strain, namely, the small peak at −0.25 V and the broad shoulder at the higher potential (−0.05) (Fig. 7). These observations suggested alterations in levels of redox-active proteins accessible to the electrode but not a change in the rate-controlling step (at −0.15 V).

FIG. 7.

(A) Cyclic voltammetry of G. sulfurreducens biofilms 72 h after inoculation, showing wild-type (black), GSU2505 mutant (red), and complemented mutant (blue) strains. (B) First derivatives of cyclic voltammograms of biofilms.

DISCUSSION

A system for the introduction of plasmids through conjugation, coupled with an efficient transposon mutagenesis method for G. sulfurreducens, has increased the number of known genes linked to extracellular electron transfer and biofilm formation. In addition, the use of a poised electrode surface has demonstrated that these phenotypes can be separated from each other, resulting in mutants with defects in one aspect (electron transfer) but not the other (biofilm formation).

As transposon mutagenesis has not been reported in this organism, preliminary experiments were performed to verify the method and maximize the likelihood of single random insertions causing observed phenotypes. For example, additional cell separation and reisolation steps were essential for obtaining pure cultures, and all mutants were screened for metabolic defects (by measuring growth rates and yields) which could confound interpretation of attachment and respiration data. The method identified amino acid auxotrophs (such as the histidine auxotroph with an insertion in GSU3097), as well as mutants with biofilm and electron transfer phenotypes, and produced a library which can be used in future experiments.

Some disrupted genes have been identified and discussed in previous Geobacter work, such as GSU0785 and GSU0782, which are predicted to encode large and small subunits (hybL and hybS) of a nickel-dependent hydrogenase necessary for hydrogen-dependent reduction of Fe(III) (6). As the Fe(III) reduction screen was conducted in the presence of hydrogen, this phenotype was expected. Another example of a previously studied locus was GSU1492, coding for the PilT subunit of one of the pilus clusters of G. sulfurreducens. This mutant demonstrated increased attachment to surfaces, suggesting that pilT (typically involved in retraction of type IV pili) could also be involved in release of Geobacter from surfaces (1).

While these findings show how mutagenesis may complement other approaches, most mutants were identified with insertions in genes never described as significantly abundant or differentially regulated in genome-wide studies (8, 9). Notable among these was a c-type cytochrome GSU0274 mutant, which was identified based on a defect in Fe(III) citrate reduction. This cytochrome has never been described as differentially regulated or essential in any other microarray or proteomic experiments. The overall lack of identified cytochrome genes was not surprising, as we selected for mutants unable to demonstrate any Fe(III) reduction over a long period of time, and many cytochrome mutants in Geobacter spp. show residual background activity or an ability to adapt via expression of other cytochromes to rescue their phenotype (17, 24).

The mutants chosen for further characterization also demonstrated how the 96-well-based attachment assay was not a predictor of electrode-attached biofilm phenotypes in Geobacter species but did detect cells with altered outer surfaces. For example, both of the GSU1501 mutants were identified based on a “high attachment” CV phenotype but showed almost no attachment to a carbon electrode. As these mutants were able to reduce Fe(III) citrate, but unable to reduce Fe(III) oxide, the inability to transfer electrons to the surfaces of carbon electrodes and Fe(III) oxide may require similar attachment mechanisms which are facilitated by this gene cluster. While strong defects in metal or electrode reduction in Geobacter are typically due to deletions of key cytochromes or in protein export pathways, GSU1501 was a component of an ABC transporter not identified in any previous study. A substrate-binding domain was not located near this transporter (suggesting a role for this protein in export) (38), which indicates for the first time that export of a small molecule may be involved in the assembly or localization of proteins involved in biofilm formation and metal reduction.

Similarly, both of the GSU3361 mutants, first identified by their high attachment phenotype, did not display increased attachment to electrodes or enhanced current generation (Fig. 5D to F). In these mutants, growth and respiration were initially similar to those of the wild type, followed by a sudden decrease which correlated with the apparent death of cells nearest the electrode (Fig. 5E and F). Recent studies have shown that the interior of a thick Geobacter biofilm can represent a markedly different environment than the exterior (e.g., a lower pH at the base) (11, 44), which may explain the dependence on biofilm age in triggering the striking phenotype. A drop in cell wall or membrane integrity could be related to the predicted periplasmically targeted transglutaminase domain in GSU3361, which is similar to domains implicated in protein cross-linking (47), and strengthening of cell walls and membranes (14, 45). In addition, the fact that the entire flux of electrons to the electrode was compromised when only the cells closest to the electrode were damaged provides an independent observation of the importance of the cell-electrode interface layer in transmitting electrons to the surface from cells more distant from the electrode.

A mutant with a reduced capacity for Fe(III) citrate reduction that generated wild-type current levels (GSU1330) was also identified. Past studies had suggested that GSU1330 played a role in electron transfer to electrodes, as it is part of a gene cluster strongly upregulated in cells grown on the anode of a microbial fuel cell (12). In addition, GSU1330 was downregulated in an outer membrane cytochrome mutant with a decreased capacity for current production (16). However, no effect on electron transfer was observed when the GSU1330 mutant was examined under our controlled electrode growth conditions (Fig. 5I). The hypothesis that the disruption was in an Fe(II) efflux protein was supported by the observation that the GSU1330 mutant was defective only in the reduction of soluble Fe(III) [where high levels of Fe(II) accumulate] but not insoluble Fe(III) [where soluble Fe(II) binds the iron oxide]. Based on these observations, not all genes significantly upregulated or downregulated when cells are grown on electrodes are necessarily involved in electrode respiration.

An example of the opposite phenotype [a mutant deficient in current generation but not in Fe(III) reduction] was also characterized (GSU2505) (Fig. 6C). The nearest gene downstream of the insertion was annotated as encoding a putative NHL domain protein but was located just upstream of genes for the well-studied outer membrane cytochromes OmcS and OmcT. Deletion of omcS has been reported to reduce rates of current production in microbial fuel cells (12, 32). In this study, the GSU2505 transposon mutant was also partially defective in respiration to electrodes. Electrodes held at constant potential, combined with biofilm imaging and electrochemical analyses, were able to show that this was not due to an attachment or biofilm defect per se, nor was it due to a defect in the primary mechanism of transferring electrons to the outer surface (e.g., as evidenced by the similar midpoint potentials in cyclic voltammetry). Instead, data pointed to a defect in electron transfer between cells. As complementation with only GSU2505 (which contains no putative redox-active domains) restored respiration, electrochemistry, and attachment phenotypes (Fig. 6E and F), and the phenotypes of this mutant and OmcS mutants were similar, GSU2505 may be needed for proper expression or assembly of proteins such as OmcS.

Implications.

A reliable and efficient method for transposon mutagenesis and screening under anaerobic conditions in G. sulfurreducens using mini-Himar RB1 was used to identify many genes previously not known to be involved in biofilm formation and Fe(III) reduction. We have also demonstrated how these two phenotypes can be separated by analysis using poised-potential electrodes and biofilm imaging. This library extends the study of extracellular electron transfer beyond key cytochromes, implicating such factors as RNA processing, two-component regulatory networks, small exported molecules, and posttranslational processing in electron transfer. Expansion and further screening of this same library for mutants defective in reduction of insoluble metals, reduction of compounds relevant to bioremediation [such as U(VI)], and attachment to more environmentally relevant surfaces will complement ongoing studies aimed at understanding the physiology of Geobacter species.