Growth advantage in stationary-phase (GASP) phenotype in long-term survival strains of Geobacter sulfurreducens


Correspondence: Ming Tien, Department of Biochemistry and Molecular Biology, Pennsylvania State University, University Park, PA 16802, USA. Tel.: +1 814 863 1165; fax: +1 814 863 7024; e-mail:


Geobacter sulfurreducens exists in the subsurface and has been identified in sites contaminated with radioactive metals, consistent with its ability to reduce metals under anaerobic conditions. The natural state of organisms in the environment is one that lacks access to high concentrations of nutrients, namely electron donors and terminal electron acceptors (TEAs). Most studies have investigated G. sulfurreducens under high-nutrient conditions or have enriched for it in environmental systems via acetate amendments. We replicated the starvation state through long-term batch culture of G. sulfurreducens, where both electron donor and TEA were scarce. The growth curve revealed lag, log, stationary, death, and survival phases using acetate as electron donor and either fumarate or iron(III) citrate as TEA. In survival phase, G. sulfurreducens persisted at a constant cell count for as long as 23 months without replenishment of growth medium. Geobacter sulfurreducens demonstrated an ability to acquire a growth advantage in stationary-phase phenotype (GASP), with strains derived from subpopulations from death- or survival phase being able to out-compete mid-log-phase populations when co-cultured. The molecular basis for GASP was not because of any detectable mutation in the rpoS gene (GSU1525) nor because of a mutation in a putative homolog to Escherichia colilrp, GSU3370.


Geobacter sulfurreducens is a microorganism that has been identified growing in anaerobic subsurface environments and has been used for bioremediation of heavy and radioactive metals (Lovley & Anderson, 2000; Snoeyenbos-West et al., 2000). In bioremediation, high concentrations of acetate are introduced into a subsurface ecosystem, and Geobacter species preferentially use acetate as an electron donor (and carbon source) for respiratory metabolism where uranium or other metals are utilized as the terminal electron acceptor (TEA) (Holmes et al., 2002). This environment of high acetate concentration represents an artificial and transitory growth condition. Most of the time, G. sulfurreducens organisms exist in what has been termed a ‘famine’ environment (Kolter et al., 1993), where electron donor and/or electron acceptor are scarce. Understanding the survival strategies of G. sulfurreducens under famine states will provide a better understanding of how this organism exists in environmental locations such as contaminated waste sites and may yield insights into how to manipulate its effects on metal mobility.

The ability of several species of nonsporulating bacteria to survive in a laboratory-created famine condition has been described for bacteria grown for extended periods of time in batch culture (Hamilton et al., 1977; Thorne & Williams, 1997; Watson et al., 1998; Finkel & Kolter, 1999; Herbert & Foster, 2001; Finkel, 2006; Wen et al., 2009). Bacterial populations, after entering stationary phase and a death phase, enter a ‘survival phase’ where they are able to maintain a constant cell density and remain viable in unsupplemented batch culture for years (Zambrano et al., 1993; Finkel & Kolter, 1999; Finkel, 2006). Survival-phase organisms often demonstrate colony and cell morphologies distinct from their mid-log counterparts, including rounding of cells, clumping of cells, and different colors of colonies (Hamilton et al., 1977; Thorne & Williams, 1997; Finkel et al., 2000; Zinser & Kolter, 2004; Wen et al., 2009).

Survival during prolonged starvation involves transient changes in gene expression as well as permanent changes in genetic material (Finkel & Kolter, 1999; Finkel et al., 2000; Farrell & Finkel, 2003; Zinser & Kolter, 2004). In species that do not sporulate, it is believed that bacteria are naturally selected from the population when carrying mutations that allow these individuals to more efficiently utilize carbon sources, electron donors, and electron acceptors that are present at low abundance (Zinser & Kolter, 1999, 2000, 2004; Zinser et al., 2003). These mutations begin to arise as cultures attain stationary phase (Farrell & Finkel, 2003). Following stationary phase, bacteria go through a phase of declining population (death phase) until they reach a relatively stable population density. During this prolonged survival phase, some bacteria can demonstrate a growth advantage in stationary-phase (GASP) phenotype (Finkel, 2006) when grown in co-culture with cultures of the same bacteria grown to mid-log phase. To elucidate between the possibilities that mutants with a GASP phenotype either accumulated in long-term survival populations through mutation-and-repair processes or through dynamic equilibrium of the populations, researchers studying Escherichia coli co-cultured two aged cultures, each containing a separate antibiotic-resistance gene (Finkel & Kolter, 1999; Finkel et al., 2000). Support for the dynamic equilibrium model came from these experiments in the observation that while the total number of cells did not change during long-term survival, the percentage of the two resistance-tagged strains fluctuated over time. In E. coli, this GASP phenotype is frequently caused by reduced function of RpoS, which controls transcription of genes involved in stress response, because of well-documented mutation in the gene encoding it (Zambrano et al., 1993; Farrell & Finkel, 2003).

A homolog to the E. coli rpoS gene has been characterized in G. sulfurreducens and is involved in controlling expression of genes involved in response to oxygen exposure and nutrient limitation (Nunez et al., 2004, 2006), suggesting that reducing the activity of the gene product may improve competitiveness in growth-limiting conditions. In this study, we investigated whether G. sulfurreducens could demonstrate a survival phase in batch culture when the original TEA is exhausted. We also investigated whether during this stage the rpoS gene is involved in G. sulfurreducens survival. We indeed found that G. sulfurreducens demonstrates a survival phase and a GASP phenotype, but that mutation of the rpoS gene is not responsible for enabling long-term survival or providing a growth advantage.

Materials and methods

Bacterial cultures

Wild-type G. sulfurreducens strain PCA and derivative G. sulfurreducens strains carrying kanamycin- or spectinomycin-resistance genes in the nifD site (Coppi et al., 2001) were provided by Derek Lovley. Liquid batch cultures were grown in anaerobic serum bottles at 30 °C in NBAF basal medium containing 30 mM acetate and 50 mM fumarate as the TEA or in FWAFC medium containing 30 mM acetate and 50 mM Fe(III)-citrate (Sigma-Aldrich, St Louis, MO) as the TEA, as described previously (Coppi et al., 2001). The stability of the phenotype and genotype of the chromosomally encoded antibiotic-resistance gene strains was confirmed throughout the studies.

Bacterial co-cultures

Co-culture experiments and the inocula used in these experiments all used NBAF basal medium. Mid-log-phase (2 days old) cultures were used as the reference culture, and challenge cultures were cells grown between 2 and 370 days. These experiments were performed following principles of the two methods described by Zambrano et al. (1993). Via the first method, c. 50 μL of the challenge culture was transferred as a minority into 18 mL of the young reference culture. Using the second method, approximately equal numbers of colony-forming units (CFU) from challenge and reference cultures were mixed into 18 mL of fresh medium. Each method produced similar results.

A total of 20 experimental co-cultures were set up with death- or survival-phase (19–367 days old) challenge culture. One co-culture was set up with stationary-phase (9 days old) challenge culture. Five co-cultures were set up with mid-log-phase (2 days old) challenge culture; these served as controls with the expected outcome to be equal growth of both strains.

For the co-culture experiment described in Fig. 4c, approximately equal CFU of both a 60-day-old culture (death phase) and a 224-day-old culture (survival phase) were inoculated into the same bottle of 18 mL fresh NBAF basal medium.

Colony-forming unit enumeration

To determine CFU intermittently during batch culture, serial dilutions of samples from growing cultures were spread on NBAFYE agar-based plates containing sodium acetate and sodium fumarate, as described previously (Coppi et al., 2001). To differentiate between kanamycin-resistant and spectinomycin-resistant strains in co-culture, NBAFYE plates were amended with 200 μg mL−1 kanamycin or spectinomycin. Before use, plates were conditioned for at least 3 days in an anaerobic environment with 20% CO2/80% N2. Plates were incubated for 7–10 days at 30 °C in GasPak containers (BD Biosciences, San Jose, CA) containing Pack Anaero sachets (Mitsubishi Gas Chemical America, Inc., New York, NY) to create and maintain a microenvironment of 20% CO2/80% N2. All sampling and manipulation of cultures were performed using strict anaerobic techniques and, when needed, in an anaerobic glove box with 3% H2 in an N2 atmosphere.

Co-culture outcome classification

Co-cultures were classified based on the relative CFU growth of younger culture vs. older culture at day 100 of culture, with representative growth plots shown in Supporting Information, Fig. S1. In cultures with ‘equal growth’, strains showed < 1 order-of-magnitude difference in CFU density (Fig. S1a). ‘Outgrowth’ was identified when the older culture CFU density was > 1 order of magnitude more than that of the younger culture (Fig. S1b). The ‘death’ outcome was defined as cultures in which the younger culture's CFU density was ≥ 1 order of magnitude more than that of the older culture (Fig. S1c), and typically in these cases, the older culture displayed near or complete death (e.g. lack of growth on NBAFYE plates).

Cell staining and imaging

Samples from anaerobic cultures were stained using acridine orange to enumerate and differentiate between living and dead cells. Samples were viewed using a Zeiss Axioplan microscope with FITC, TRITC, and UV filters. We typically observe that direct cell counts of G. sulfurreducens using this acridine orange method provide counts c. 1.1 log greater than via CFU enumeration on NBAFYE plates, regardless of the age of the culture.

Transmission electron microscopy was performed using a JEOL JEM 1200 EXII microscope. To capture images of cell morphology, unstained cells were visualized using a DIC filter on an Olympus BX61 optical microscope. Intact bacterial colonies were imaged under a magnifier using a Nikon L12 camera affixed to the objective. To prepare colonies for visualization using scanning electron microscopy, slices of agar plates containing colonies were aseptically acquired and immediately fixed for 2 h in Karnovsky's fixative solution consisting of 2.5% paraformaldehyde, 1.5% glutaraldehyde, 4% sucrose, and 0.1 M cacodylate. Agar slices were then washed three times in 0.1 M cacodylate and dehydrated in solutions of increasing ethanol concentration. Samples were then subjected to critical point drying and sputter-coated with gold particles. Surface detail of colonies was imaged using either a JEOL JSM-5400 or Hitachi S-3500 N electron microscope made available through the Penn State Huck Institutes of the Life Sciences and Materials Characterization Laboratory facilities.

Ferrous iron concentration measurements

A 750-μL aliquot of culture was taken after 0, 8, and 11 days and acidified in an anaerobic glove box with 250 μL 2 N HCl. Samples were diluted 100-fold in 0.5 N HCl. To measure ferrous ion concentration, 50 μL of diluted acidified samples was added to 950 μL of a ferrozine solution (1 g L−1 ferrozine in 100 mM HEPES, pH 7), and the absorbance at 562 nm was determined spectrophotometrically and converted to concentration using an extinction coefficient of 27.9 mM−1 cm−1 (Stookey, 1970; Ross et al., 2009).

Gene amplification

PCRs to amplify the rpoS (GSU1525) gene and putative homolog to lrp (GSU3370) gene were performed from single colonies selected from NBAFYE agar plates or from liquid NBAF cultures. The kanamycin- and spectinomycin-resistance genes were amplified from single colonies to confirm the persistence of these genes throughout co-culture studies. Primers used for PCRs are listed in Table S1. To amplify the 16S gene, universal primers 530F (5′-GTGCCAGCMGCCGCGG-3′) and 1392R (5′-ACGGGCGGTGTGTRC-3′) were used as described previously (Geib et al., 2009).


For each sequencing reaction, 2 μL of PCR product was treated with 1 μL of ExoSAP-IT (USB Corporation, Cleveland, OH) according to the manufacturer's instructions. Primers used for sequencing were the same as for amplification of the gene. Sequencing was performed on an ABI Hitachi 3730XL DNA Analyzer at the Penn State Genomics Core Facility, University Park, PA. Chromatographic output was analyzed using BioEdit v.7.0.5 (Ibis Therapeutics, Carlsbad, CA); 16S sequences were compared with the full microbial database of the National Center for Biotechnology Information (NCBI) using their Basic Local Alignment Search Tool (blast), November 2009. To identify possible homologs to the E. coli Lrp protein in G. sulfurreducens, the amino acid sequence of E. coli Lrp (NCBI GeneID 949051) was used as input to the protein-blast server at the J. Craig Venter Institute Comprehensive Microbial Resource (March 2010) to probe the G. sulfurreducens peptide database. This result was supplemented with a protein-blast search against the G. sulfurreducens database at NCBI (March 2010).

Statistical analysis

Statistical analysis was performed using GraphPadPrism version 5. Stated P-values were determined using an unpaired two-tailed t-test with Welch's correction unless otherwise stated. The two-tailed P-value from Fisher's exact test was used to analyze the 2-by-2 contingency table summarizing data in Table 1.

Table 1. Co-culture outcomesa
Challenge growth phaseReference strainChallenge strainResult
OutgrowthEqual growthDeath
  1. a

    The co-culture competition experiments were performed as described in 'Materials and methods'. Mid-log cultures were challenged by a low inoculum of ‘Experimental culture’, which was either mid-log or from very old cultures (death/survival) harboring different selectable markers. Cells were then enumerated on antibiotic-containing plates.

Mid-logKanamycinSpectinomycin1 (20%)4 (80%) 
Death/survivalKanamycinSpectinomycin8 (50%)6 (38%)2 (12%)
SpectinomycinKanamycin 3 (100%) 
SpectinomycinWild type 1 (100%) 


Long-term survival of Geobacter sulfurreducens

Like bacteria that display GASP (Finkel, 2006), G. sulfurreducens demonstrated lag, log, stationary, death, and long-term survival growth phases when grown in batch cultures for more than 1 year (Fig. 1a and b). The cultures progressed from lag through log and into stationary phase within 5 days and remained at a relatively constant density in stationary phase for about 10 days before demonstrating a death phase (Fig. 1a and b). In cultures grown with ferric citrate as the TEA, greater than 93% of the ferric citrate was reduced by approximately day 12 of culture as measured by ferrous iron accumulation (Fig. 1a, inset). The death phase, characterized by a gradual and constant decrease in cell density at a rate of c. 1.68 × 106 cells mL−1 per day, persisted for c. 3 months. By about 4 months of batch culture, the population became stabilized at a density of c. 1–2 × 106 cells mL−1 and remained at this level thereafter with no significant difference in cell density (> 0.05 for each set of consecutive time points, without Welch's correction), encompassing the survival phase. Growth-phase durations were similar when grown in batch cultures using fumarate (Fig. 1b) or iron citrate (Fig. 1a) as TEA.

Figure 1.

Long-term batch culture of Geobacter sulfurreducens. Anaerobic bottles of medium containing acetate as electron donor and TEA of either ferric citrate (a) or fumarate (b–f) were inoculated and kept in batch culture at 30 °C with intermittent sampling. Plots of live cells using acridine orange cell counts (a, c) or CFU on agar plates using serial dilution (b) are shown. A plot of reduced iron in ferric citrate cultures is shown in the inset of (a); the dotted line represents the maximum ferrous iron that could be produced from reduction of the starting concentration of ferric citrate. Data points in (a) and (b) represent the mean of single measurements from four or five separate bottles. The plot in (b) is representative of 10 separate experiments. In (c), two bottles each were inoculated with a 51 (solid squares, solid line)- or 529 (open circles, dotted line)-day-old culture. Transmission electron micrographs of samples taken from cultures grown for 2 days (d), 6 months (e), or 20 months (f) are shown. Error bars represent standard error of the mean.

Resuscitation of cells from prolonged starvation

Cells in the survival phase grown under batch culture conditions maintained a stable cell density, even up to 23 or more months. The viability of these cells was confirmed by (i) live/dead staining, (ii) growth in liquid medium, and (iii) growth on agar-containing plates. Although they displayed an extended lag in growth compared with cells from late stationary phase, we observed that survival-phase cells could be recovered in liquid medium (Fig. 1c) and on agar plates (data not shown). These results indicate cells remained viable even after one-and-a-half years in unsupplemented batch culture.

Cellular morphology

Cells from liquid culture in various stages of growth were examined by light microscopy. Cells from 6-month-old cultures (termed here ‘early survival phase’) (Fig. 1e) displayed lysed cells and many more electron-dense cells compared with cells grown to mid-log phase (2 days old) (Fig. 1d). After 20 months of culture (termed here ‘late survival phase’), unlysed cells typically appeared rounded and electron dense via transmission electron microscopy (Fig. 1f).

Most batch cultures harbored organisms that grew into pink colonies (Fig. 2m and n). All colonies grown from mid-log-phase cultures (data not shown) were pink and contained rod-shaped cells with an average length of 4.23 ± 0.11 μm (n = 50), as well as many diplo-rod-shaped cells that presumably represent recently divided cells in the exponential phase of growth. Cells from pink colonies from survival-phase cultures (Fig. 2m and n) differed from those from pink colonies from mid-log-phase G. sulfurreducens in that survival-phase rod-like cells were on average significantly shorter (2.60 ± 0.04 μm, < 0.0001, n = 390).

Figure 2.

Colony and cell morphologies of survival-phase Geobacter sulfurreducens. Images of colonies and bacteria from isolates with white-pinpoint (a–d), clear-mucoid (e–h), clear-normal (i–l), and pink-normal (m–p) colony morphologies. (a, e, i, m) Magnified image of colonies on NBAFYE plates, scale bar = 1 mm. (b, f, j, n) Image of magnified colonies on NBAFYE plates, scale bar = 2 mm. In (j), arrowheads point to location of clear colonies. Inset shows representative image of colonies (thin arrowhead, magnified five times). (c, g, k, o) Differential interference contrast micrograph of individuals from colony, scale bar = 5 μm. (d, h, l, p) Scanning electron micrograph of surface of colony, scale bar = 1 μm.

Ten percent (3 of 30) of batch cultures in survival phase transiently contained colonies of abnormal morphology. This was not a stable phenotype; in each of the three culture bottles in which they appeared, nonpink colonies were detected at only one or two consecutive time points but then were not observed again in the same culture bottle. The colonies were confirmed to consist only of G. sulfurreducens by 16S rRNA gene amplification and sequencing (data not shown). These results demonstrated that the colonies were not other species. Overall, colonies that were not pink contained cells that had an average length significantly shorter from that of cells from pink colonies of survival-phase cultures (2.26 ± 0.04 μm, < 0.0001, n = 360).

The cells that made up these unique colonies displayed distinct shapes and properties. One culture contained white colonies that were convex and pinpoint (Fig. 2a and b). White colonies consisted of rod-shaped cells that tended to gather in clumps and had wrinkled sides (Fig. 2c and d). These were the shortest cells, significantly shorter on average than cells from pink colonies (1.98 ± 0.07 μm, < 0.0001, n = 93). Two cultures contained clear colonies that were smooth, glossy, circular, and were either raised and about one-quarter the size of pink colonies (Fig. 2e and f) or were flat, mucoid, and appeared as microcolonies (Fig. 2i and j). The cells from clear, typical-sized colonies that were transiently produced from one of these cultures were slightly shorter rod-shaped cells (2.52 ± 0.16 μm average length, n = 34) and more rounded than cells from pink colonies from survival-phase cultures (Fig. 2g and h). The third culture showed clear microcolonies that contained rod-shaped cells that were on average significantly shorter (2.33 ± 0.04 μm, < 0.0001, n = 233), more rounded, and often resembled coccobacilli (Fig. 2k and l).

GASP phenotype of G. sulfurreducens

To determine whether any of the G. sulfurreducens from survival phase develop a ‘GASP’ phenotype, challenge cultures from various growth phases were co-inoculated with reference mid-log-phase cultures. Cultures were differentiated from each other by the presence of an antibiotic-resistance-conferring gene. Detailed definitions of co-culture outcomes are described in 'Materials and methods' and Fig. S1.

When the kanamycin-resistant challenge culture and spectinomycin-resistant reference cultures were both from mid-log phase (grown for 2 days), we observed largely equal growth; four of the five co-cultures displayed equal growth. Only one co-culture test with these younger cells displayed mild outgrowth (Table 1). This shows that strains with antibiotic-resistance genes did not inherently survive better or worse than strains with a different antibiotic-resistance gene.

Co-cultures inoculated with challenge cultures grown from 3 to 14 days (log and stationary phase) displayed no evidence of organisms with a growth advantage over the older cultures (Fig. 3a). In contrast, when the challenge culture was between 2 and 8 weeks old (death phase) at the time it was mixed with the mid-log-phase reference culture, individuals from this older culture began to demonstrate an outgrowth response compared with the younger culture (Fig. 3b, c and d). Consistent with a GASP phenotype, when co-cultures were inoculated with G. sulfurreducens cultures older than 60 days, equal growth or outgrowth of the older culture was generally observed (Fig. 3e and f). In each co-culture, the challenge culture exhibited an ability to grow (i.e. enter log phase), indicating that the ability to outgrow the younger reference culture was not due to an acclimation to the survival-phase conditions. The outgrowth phenotype was observed regardless of which antibiotic-resistance marker was carried on the older challenge culture, indicating that a particular antibiotic-resistance marker does not account for this phenomenon (Table 1).

Figure 3.

Acquisition of GASP by Geobacter sulfurreducens. Geobacter sulfurreducens containing a chromosomally located resistance gene for spectinomycin (closed circles, solid line) or kanamycin (open squares, dashed line) were co-inoculated into NBAF medium when cultures were grown to within 2 and 120 days (d), as indicated on each graph. CFU mL−1 were measured on NBAFYE plates at designated times after inoculation. An x symbol indicates the detection limit, with a downward-facing arrow symbolizing that the limit is an upward bound of CFU density.

Altogether, 20 experiments with death or survival-phase cells co-cultured with mid-log-phase cultures were studied. Of these, two displayed near-complete death of the older culture, 10 displayed equal growth, and eight displayed outgrowth of the older culture (Table 1). With equal growth and outgrowth both considered consistent with a GASP phenotype, the aged cultures showed statistically significant propensity to develop the GASP phenotype compared with mid-log cultures used as challenge (= 0.0055). In a separate experiment, we also determined whether much older survival-phase cultures exhibited a growth advantage when co-cultured with younger cells that had been grown to survival phase. Specifically, we selected two survival-phase strains (kanamycin-resistant strain K and wild-type strain W) that showed a GASP response when competed against mid-log-phase cells of spectinomycin-resistant strain S (Fig. 4a and b). When the two cultures of survival-phase cells were competed against each other, an advantage of the older culture, K, was observed (Fig. 4c).

Figure 4.

Selection for growth advantage may arise continuously in survival-phase Geobacter sulfurreducens. Selection for growth advantage may arise continuously in survival-phase G. sulfurreducens. Kanamycin-resistant strain K (open squares, dashed line) was grown for 49 days, and some of this culture was used co-inoculate NBAF medium along with 2-day-old spectinomycin-resistant strain S (filled circles, solid line) (a; also shown in Fig. 3d). The remaining strain K culture was allowed to grow for an additional 175 days. Wild-type strain W (open triangles, dotted line), grown for 60 days, was used to co-inoculate NBAF medium along with 2-day-old strain S (b). At the same time, strain K had been growing for 224 days, and it was co-inoculated with 60-day-old strain W into fresh NBAF medium (c). An x symbol indicates the detection limit, with a downward-facing arrow symbolizing that the limit is an upward bound of CFU density.

Molecular profile during survival phase

To investigate whether mutation in the RNA polymerase sigma factor S gene (rpoS) might contribute to GASP in G. sulfurreducens, DNA sequencing of the G. sulfurreducens rpoS homolog, gsu1525 (Fig. S2a), was performed on 51 individual colonies that arose from the outgrowth population demonstrating the GASP phenotype in co-culture competition experiments (i.e. behavior such as Fig. S1a, b, and c). The 51 colonies were collected from five separate co-culture experiments. No sequence differences were noted from any of the colonies when compared with the published G. sulfurreducens genome sequence (Methe et al., 2003), indicating that mutations in the rpoS coding region were not responsible for the GASP phenotype in these individuals (data not shown). On the population level, the rpoS gene also remained identical to the wild-type gene in batch cultures grown for 35, 65, 111, 135, and 344 days based on sequencing of the rpoS PCR product from a sample of liquid culture from each time point. In a study to be published elsewhere, our data comparing the relative protein expression levels from mid-log and long-term survival-phase G. sulfurreducens using iTRAQ analysis did not demonstrate any difference in the expression of the gene product of gsu1525 (data not shown).

Additional genes that have been implicated with GASP response in E. coli include lrp (leucine-responsive regulatory protein) (Zinser & Kolter, 2000) and the ybeJ-gltJKL (Zinser et al., 2003). A blast search of the E. coli gene product of lrp against the G. sulfurreducens peptide database revealed two proteins with > 95% probability of matching the peptide sequence (smallest sum probability < 0.05). These included a putative transcriptional regulator (gsu2033) and a GntR-family transcriptional regulator (gsu3370) (Fig. S2b). As gsu2033 is a short protein and classified as a putative protein, it did not appear to encode a functional protein, so only the sequence of gsu3370 was screened from PCR products of three separate survival-phase liquid batch cultures as well as from PCR products from 45 separate colonies representing the subpopulation that outgrew the younger strain from 15 different co-culture experiments. No mutations were found in any region of the gsu3370 gene.


Our results show that long-term batch cultures of G. sulfurreducens exhibit the five characteristic stages of growth: lag (approximate days after inoculation: 0–2), log (days 2–5), stationary (days 5–15), death (days 15–122), and survival (days 122 and greater). This pattern is similar to that observed with other species, in particular E. coli (Finkel et al., 2000; Farrell & Finkel, 2003; Martinez-Garcia et al., 2003). The duration of each growth phase correlates with that of E. coli when the doubling time (20 min for E. coli, c. 6 h for G. sulfurreducens) of the two species is taken into consideration, i.e. G. sulfurreducens stationary phase is about 10 days, 18-times longer than the c. 14-h stationary phase of E. coli. Similar to that observed with E. coli, Listeria monocytogenes, and other organisms (Hamilton et al., 1977; Thorne & Williams, 1997; Finkel, 2006; Wen et al., 2009), colony and cell morphological changes were also observed for G. sulfurreducens during prolonged starvation when both electron donor and TEA were depleted. In survival phase, cells undergo rounding, decrease in size and aggregation. We also observed that colony color sometimes transiently changes from pink to white or clear and back to pink. These results are consistent with a dynamic population surviving in this stage of growth with the rise and fall of subpopulations of bacteria with unique morphologies and potentially unique genotypes driven by mutation and not by selection of a single mutant strain that overcomes the entire population and remains established (Zambrano et al., 1993; Finkel et al., 2000). Indeed, the challenge experiment described in Fig. 4c is analogous to the study performed to support the dynamic equilibrium model in E. coli long-term survival (Finkel & Kolter, 1999). It is not apparent how, or if, changes in morphology or color would confer a survival advantage. The decrease in cell size may be solely due to a consequence of starvation (skinny vs. fat cells). However, a decrease in size increases the surface area to volume ratio, thus facilitating the ability of the cell to come in contact with soluble nutrients.

Based on this model of dynamic population-density maintenance, the replication rate and death rate in survival-phase cultures of G. sulfurreducens would be equal. Our co-culture experiments might be explained by cellular turnover. While our results do not reveal the rate of this turnover, there must be, nevertheless, a basal rate of energy production to drive the replication process. This would require redox processes within the cell that can generate the energy required for cell growth and replication, implying that all components for redox (i.e. electron donor and TEA) were available in the unamended batch culture. As the original TEAs in this study [fumarate and iron(III) citrate] were exhausted by the end of stationary phase, we speculate that other components of the batch cultures presumably can act as TEAs. A possible source would be the components of dying cells. Geobacter sulfurreducens has been shown to be able to utilize an array of TEAs (Caccavo et al., 1994; Afkar & Fukumori, 1999; Lin et al., 2004). In the survival-phase batch culture system, it is possible that components from this list would be released from dying cells.

The change in metabolic strategy appears to confer surviving G. sulfurreducens a selective advantage over their younger counterparts. Our results show that the advantage in stationary phase, GASP, is observed in c. 90% (18 of 20) of death- or survival-phase cultures. This frequency of GASP phenotype is similar to that observed in other microorganisms (Zambrano et al., 1993; Farrell & Finkel, 2003; Martinez-Garcia et al., 2003; Silby et al., 2005; Dong et al., 2009; Pradhan et al., 2010). The transition from mid-log phase to stationary and eventually to starvation phase involves not only changes in gene expression but also changes in the predominant genotype encoded by the population (Thorne & Williams, 1997; Finkel et al., 2000; Zinser & Kolter, 2004). While the total number of cells in our co-cultures varied little during long-term survival, the proportion of each resistance-tagged strain fluctuated by about 10% when both persisted. In E. coli and Citrobacter rodentium, the genes rpoS and/or lrp have been positively identified as the locus of mutation causing GASP (Zambrano et al., 1993; Farrell & Finkel, 2003; Dong et al., 2009). The rpoS gene has been shown in G. sulfurreducens to be important for optimal growth on poorly crystalline Fe(III) oxide, recovery from oxygen exposure, and survival in stationary phase (Nunez et al., 2004). Although they seem likely candidates for the molecular basis of GASP in G. sulfurreducens, in the present study, neither the rpoS gene nor a putative lrp homolog (gsu3370) was mutated in any of the survival-phase cultures or individual G. sulfurreducens colonies demonstrating a GASP response. However, our data do not preclude the fact that the parent cultures received as frozen stocks may already have acquired a mutation in these genes that differs from the original environmental isolate organism or that the published sequence genome sequence may reflect a mutation acquired through laboratory passaging.

Furthermore, a mutation that deletes the G. sulfurreducens rpoS gene (gsu1525) is not deleterious to the organism as it is in E. coli (Nunez et al., 2004), and the regulon of rpoS differs between these two organisms (Nunez et al., 2004; Santos-Zavaleta et al., 2011). These findings suggest that the affected gene product(s) that results in the G. sulfurreducens GASP phenotype may be part of the set of genes that is not highly conserved in the rpoS sigmulon. The amino acid similarity between G. sulfurreducens and the portion of E. coli rpoS that is duplicated in the GASP phenotype organisms (final six residues of gsu1525) is quite poor, so it is unclear what effect this sort of mutation would have on the function of G. sulfurreducens gsu1525 gene product. As determining the genetic basis of naturally selected GASP has been difficult in other organisms, the ongoing genomic sequencing of the survival-phase G. sulfurreducens in our laboratory likely presents the most rapid method for identifying mutations that cause the results observed in the present study. As tools to study the genetic variation within these subpopulations over time are defined for G. sulfurreducens, further exploration of population dynamics during long-term survival of this organism can be carried out.

Morita points out that environmental bacteria live in conditions of both feast and famine, but that famine may be the norm (Morita, 1988; Kolter et al., 1993). Organisms thus require adaptations to survive such conditions. A recent study suggests that E. coli GASP organisms thrive in nutrient-rich microenvironments when competed against wild-type cells (Lambert et al., 2011), so it may be that the most relevant G. sulfurreducens organisms in acetate-amended contaminated sites are in fact GASP mutants. Long-term batch cultures such as those used in our study mimic such conditions of famine in the natural environment. While our model closed system may not mimic the natural environment, it does provide a system for monitoring changes in the population and a window into the selection process. The biomarkers for GASP in G. sulfurreducens identified in our studies may be added to the developing set of biomarkers useful for monitoring the metabolic status of this and other Geobacter species in situ at contaminated sites (Yun et al., 2011). While our studies focused on survival without TEA, these long-term cultures may also mimic conditions that likely exist following depletion in electron donors after application to sites of bioremediation in that in both cases organisms are forced into a famine state following a feast. The latter scenario is relevant to G. sulfurreducens because this species is representative of the primary genus of metal-reducing organisms that can catalyze immobilization of uranium and technetium at sites in the United States including Rifle, Hanford, and the Oak Ridge Field Research Center (Anderson et al., 2003; McKinley et al., 2007; Michalsen et al., 2007).


The authors wish to thank Greg Ning, Missy Hazen, Ruth Haldeman, Maria Klimkiewicz, and Nicole Zembower for assistance with bacterial imaging; Lorraine Santy for use of a fluorescence microscope; and Jia Wen and Stephen Knabel for helpful discussions. Derek Lovley provided the G. sulfurreducens strains used to begin this work. This research was supported by the National Science Foundation Environmental Molecular Sciences Institute program (CHE-0431328) through the Penn State Center for Environmental Kinetics Analysis and by the U.S. Department of Energy grant DE-FG02-07ER64399.