The majority of studies of FeRB carried out in pure culture have focused on soluble, complexed Fe forms (such as ferric citrate) or on Fe(III) oxyhydroxide minerals (ferrihydrite and goethite) as the Fe available for reduction. In contrast, most of the oxidized Fe by weight in natural sediments is associated with phyllosilicate clay minerals (48). Clay minerals are abundant and ubiquitous in soils and sediments (41). Along with microorganisms, clays provide some of the most catalytic surfaces in sedimentary environments, which are important to a variety of biogeochemical cycles (41, 43). However, few studies have been carried out on the interactions between these two reactive components of porous media.

Of the organisms currently available in pure culture which are known to conserve energy for growth from the reduction of Fe(III) minerals, members of the shewanellae and of the Geobacteraceae have been characterized in the most detail (21, 29). A small but expanding database has been collected which shows that microorganisms from both of these families catalyze the reduction of clay-bound Fe(III) (16). Microbial clay reduction has been demonstrated at temperatures and pHs common to soils and sediments (15-17). As with Fe oxide minerals, organic chelates and electron transfer agents increased the bioavailability of clay bound Fe(III) for reduction (16, 26). However, no past studies have provided evidence for the growth of bacteria with clay minerals as the sole electron acceptor. Further, it is unclear how respiration and growth on clay-bound Fe(III) compare to those on other Fe mineral forms.

In this study, we found that the respiration of structural Fe(III) bound in smectite clay minerals supports the growth of FeRB in pure culture and in enrichment cultures from two very different sedimentary environments. We also compared growth yields of a S. oneidensis strain MR-1 on a variety of Fe(III) forms with that on oxygen. Iron(III) oxide minerals are believed to be more reactive or available for microbial reduction than clay minerals. In contrast, we show that the growth rate and yield of S. oneidensis on smectite appear to be similar to growth on amorphous Fe(III) oxyhydroxide. These discoveries have important implications for contaminated subsurface and surface aquatic environments, where Fe(III)-bearing clay minerals are abundant and at times comprise the predominant electron acceptor available to microorganisms.

Bacterial cultures and cultivation methods. A pure culture of Shewanella oneidensis strain MR-1 was used which was isolated from the anoxic sediments of Lake Oneida, NY, and has been the subject of many physiological and genetic studies concerning the Shewanellae (29). S. oneidensis is a facultative anaerobe and an obligately respiratory bacterium, incapable of fermentative growth (39). Purified enrichment cultures consisted of FeRB consortia from two source inocula. Subsurface sediment samples were obtained from the saturated zone of unconsolidated alluvium (1 to 4 m below land surface) at the Field Research Center of the Department of Energy's Natural and Accelerated Bioremediation Research Program, Oak Ridge, Tenn. A second set of samples were collected at a 0.1-m depth in the surface soil of a rice paddy located in Nanjing, China. This rice field is flooded biannually, leading to prolonged anoxic conditions followed by alternating aerobic conditions between flood cycles. Inocula for enrichments were collected by taking cores or grab samples. All soil and sediment samples were collected and transported chilled to the lab. Sample handling and cultivation procedures were carried out under aseptic and strictly anoxic conditions.

Enrichment cultures were purified by successive transfer using the culture medium and methods described below. The enrichments were selective for respiratory FeRB, as Fe(III) oxide was added as the sole electron acceptor and acetate was added as the sole carbon source in a minimal medium throughout successive transfers. Dissimilatory Fe(III) reduction was shown repeatedly in the enrichments as the production of reduced Fe in acid extracts coupled to the depletion of acetate. Further, FeRB were observed to be abundant (10⁴ to 10⁶ cells g of wet sediment⁻¹) in the sediment used as an inoculum for the enrichment culture with a most probable number assay (Kostka et al., submitted for publication).

Standard methods for the culture of anaerobic bacteria were modified for clay reduction experiments as in the work of Kostka et al. (15). For S. oneidensis, a minimal culture medium was prepared as described by Kostka and Nealson (14). For the enrichment cultures, a carbonate-buffered minimal medium formulated to resemble groundwater was prepared and dispensed into Hungate tubes as described by Widdel and Bak (50). The freshwater version of the Widdel-Bak basal medium, originally designed for sulfate-reducing bacteria, was modified by replacing any sulfates with chloride salts and adding FeCl₂ instead of dissolved sulfide as a reductant (2 mM final concentration) (1).

Culture medium was prepared, the pH was adjusted to 7, and the medium was dispensed into serum bottles. Carbon substrates (lactate for S. oneidensis, acetate for enrichments) were added from sterile, anoxic stocks to a 10 mM final concentration for each. Unless otherwise specified, smectite clay SWa-1 was added to a particle concentration of 4 g liter⁻¹. Amorphous Fe(III) oxyhydroxide (FeOOH) or ferric citrate was added to a 50 mM final concentration. Culture bottles were sealed with butyl rubber stoppers and incubated at 30°C in the dark. All manipulations of culture samples were carried out under strictly anoxic conditions within a Coy anaerobic chamber (90% N₂, 10% H₂). Inoculum cultures were grown anaerobically on Fe(III) to late log phase on the appropriate minimal medium. Heat-killed controls were heated by microwave radiation until boiling (11).

Preparation of oxidized Fe. Amorphous Fe(III) oxyhydroxide (surface area = 600 m² g⁻¹) was prepared as described by Schwertmann and Cornell (38). For all experiments with smectite clay, the 0.5-to 2-μm fraction of the ferruginous smectite SWa-1 from Grant County, Wash. (Source Clays Repository, The Clay Minerals Society), was used. The clay was Na⁺ saturated, fractionated, dialyzed, and freeze-dried prior to use (42). Lear and Stucki (20) reported the structural Fe content of the same dialyzed SWa-1 to be 3.549 mmol of Fe g⁻¹ (with less than 0.1 mmol of this Fe g⁻¹ present as Fe oxide impurities) and the surface area to be 720 m² g⁻¹. All solutions were made anoxic using an updated, commercially available version of the apparatus described by Stucki et al. (42). All Fe minerals were sterilized by heating via microwave radiation (11) before addition to the culture medium. Ferric citrate was prepared as described previously by Kostka and Nealson (14).

Determination of reduction and growth. The reduction of Fe(III) was measured as the production of Fe(II) in HCl extracts using the colorimetric reagent ferrozine under strictly anoxic conditions (14, 23). This method was previously validated for use in clay cultures by comparison to HF extracts and Mossbauer spectroscopy (15, 16). Cell numbers were determined by direct counting using acridine orange and epifluorescence microscopy as described previously (10, 23). Bacterial direct counting procedures were modified for solid-containing suspensions according to the work of Proctor and Souza (33). Biomass was determined by measuring the accumulation of particulate carbon using previously described methods (27).

Phylogenetic characterization of purified Fe(III)-reducing consortia. Genomic DNA was extracted with an ultraclean soil DNA kit (Mo Bio Laboratories, Inc, Solana Beach, Calif.). Extracted genomic DNA was used as a template for PCR amplification of nearly the entire 16S rRNA gene (~1,400 bp) with the bacterium-specific primers 8F and 1392R as previously described (3). The amplification products were subsequently cloned into Escherichia coli with the TOPO TA cloning kit (Invitrogen, Carlsbad, Calif.). The clones were screened by restriction analysis, and unique clones were sequenced with an automated sequencer (Applied Biosystems model 3100) using the terminator cycle sequencing method. The sequences were aligned to comparison strains in the Ribosomal Database Project in accordance with the secondary structure of the 16S rRNA molecule using the ARB software package (40). A phylogenetic distance matrix was calculated from the aligned sequences using distance and maximum likelihood methods.

The rapid reduction of soluble and solid Fe(III) forms was observed in cultures of Shewanella oneidensis strain MR-1 over 3 days (Table 1). Nearly all of the Fe(III) citrate was reduced, compared to half of the Fe(III) oxyhydroxide (FeOOH) and approximately one-third of the structural Fe(III) bound in smectite. A white, ferrous carbonate precipitate was produced over time in Fe(III) citrate cultures, whereas a black magnetic precipitate, presumably magnetite, was formed in Fe(III) oxyhydroxide cultures. With smectite as the sole electron acceptor, no solid form other than smectite was observed at the end of the experiments.

TABLE 1. Summary of growth yields for S. oneidensis strain MR-1 utilizing a variety of electron acceptors after 70 h in culturea

Addition of the humic acid analog anthraquinone disulfonate (AQDS) generally stimulated reduction of the FeOOH to a larger extent than clay-bound Fe(III) (Table 1). Control cultures (heat killed, exposed to HgCl₂, or cultured aerobically) showed little or no reduction of clay-bound Fe(III) to Fe(II). As has been demonstrated previously (16, 25), lactate was depleted and carbon dioxide was produced according to the 4:1 stoichiometry of Fe(III) reduced to carbon oxidized in all cultures (data not shown).

Growth of S. oneidensis, measured by an increase in cell number, paralleled the reduction of all Fe(III) forms tested (Table 1; Fig. 1 and 2). Little or no growth was observed in killed controls (with heat or HgCl₂) or in control cultures to which no electron acceptor had been added (Fig. 1). Interestingly, the increase in cell number exhibited a similar range for all electron acceptors tested (Table 1; Fig. 2). Within 3 days, cell densities of S. oneidensis were observed to increase to a range between 9.0 × 10⁷ and 16.5 × 10⁷ cells ml⁻¹. On the third day, cultures were sacrificed for particulate organic carbon analysis and the yield within each culture was determined. It was not possible to resolve the particulate organic carbon yield in smectite cultures due to the presence of background carbon impurities (data not shown). However, in contrast to the increase in cell number, accumulation of S. oneidensis biomass (measured as grams of C cell⁻¹) ranged over a factor of 5 depending upon the electron acceptor utilized [O₂, Fe citrate, and Fe(III) oxyhydroxide] (Table 1).

FIG. 1. Growth (A) and structural Fe(III) reduction (B) of S. oneidensis strain MR-1 with lactate as the electron donor and smectite as the electron acceptor. Results are expressed as means ± 1 standard deviation from triplicate cultures. Symbols: open (more ...)

FIG. 2. Growth of S. oneidensis strain MR-1 with lactate as the electron donor in culture treatments which differ according to the electron acceptor added. Results are expressed as the means ± 1 standard deviation from triplicate cultures. Symbols: open (more ...)

S. oneidensis conserved energy for growth by coupling the reduction of smectite to the oxidation of lactate (Fig. 1). S. oneidensis was then inoculated into minimal basal media supplemented with various particle concentrations of smectite as the sole electron acceptor. Not only was the degree of structural Fe(III) reduction proportional to the initial particle load, but a corresponding proportional increase in cell density was also observed (Fig. 3).

FIG. 3. Concentration of reduced structural Fe plotted against growth (as cell number) for S. oneidensis strain MR-1 cultures with lactate as the electron donor and where the particle load of smectite was varied as the electron acceptor. Growth is expressed as (more ...)

In enrichment cultures freshly purified from rice paddy soil and from contaminated subsurface sediment, Fe(III)-reducing consortia were shown to conserve energy for growth by coupling the reduction of structural Fe(III) in smectite to the oxidation of acetate. Growth paralleled the reduction of Fe(III) with smectite added as the sole electron acceptor, while little or no growth was observed in control cultures to which no electron acceptor had been added (Fig. 4). Representative growth curves are shown for parallel enrichment cultures enriched from the same subsurface sediment core (Fig. 4), and approximately twice the growth was observed for duplicate rice paddy enrichments [13 × 10⁷ cells ml⁻¹ and 2.5 mM smectite-bound Fe(III) reduced; data not shown]. Using a cloning and sequencing approach of the 16S rRNA genes in these enrichment cultures, the dominant members of the Fe(III)-reducing consortia were identified. In the subsurface enrichment (Ac032) (Fig. 5), five clones were obtained, and the dominant sequences retrieved showed the highest sequence similarity to “Geobacter akaganeitreducens” (three clones, 90% similarity) and Desulfitobacterium chlororespirans (two clones, 90% similarity). In the rice paddy enrichment (LacRPS) (Fig. 5), six clones were obtained and the dominant sequences retrieved showed the highest sequence similarity to Clostridium celerecrescens (three clones, 96% similarity) and Geobacter sulfurreducens (two clones, 98% similarity).

FIG. 4. Growth (A) and structural Fe(III) reduction (B) of an FeRB consortium purified from subsurface sediments with acetate as the electron donor and smectite as the electron acceptor. Results are expressed as means ± 1 standard deviation for triplicate (more ...)

FIG. 5. Phylogenetic tree for enrichment cultures of Fe(III)-reducing consortia and selected strains of eubacteria, based on a distance matrix analysis (see Materials and Methods). Consortia purified from rice paddy soil and subsurface sediments are designated (more ...)

Growth with smectite clay minerals as the sole electron acceptor. The growth of FeRB in culture has mostly been quantified by measuring the increase in cell number. Using the maximum cell number increase in comparison to control cultures, the range of growth measured for S. oneidensis coupled to smectite respiration reported here (1 × 10⁸ to 2 × 10⁸ cells ml⁻¹) is very similar to the range reported in previous studies for the growth of pure cultures of FeRB coupled to the respiration of synthetic Fe(III) oxide minerals (1 × 10⁸ to 2 × 10⁸ cells ml⁻¹) under similar culture conditions (7, 23, 25, 34). Our observation that microorganisms can gain energy for growth by catalyzing the respiration of smectite clay minerals is not surprising when the energy available from these reactions is considered (Table 2). The standard free energy at pH 7 that was calculated to be available from lactate oxidation coupled to smectite reduction (Table 2, equation 1) is very similar to that calculated to be available with Fe(III) oxyhydroxide or soluble ferric citrate as the electron acceptor (Table 2, equations 2 and 3). Our findings suggest that a similar amount of energy for growth is generated during Fe(III) reduction, regardless of the form of Fe being utilized in cultures of S. oneidensis.

TABLE 2. Stoichiometry and free energy of reactions related to the metabolism of S. oneidensis strain MR-1 with various electron acceptors

Earlier culture studies indicated that the availability of Fe(III) minerals for reduction by microorganisms was determined by their crystallinity or mineral form (22). Amorphous Fe(III) oxyhydroxides were shown to be rapidly and extensively reduced by microbes, whereas crystalline Fe(III) minerals were reduced slowly and incompletely. More recent studies by Kostka and Nealson (13) and Roden and Zachara (36) revised this view by showing that FeRB were capable of growth on the crystalline Fe(III) oxide minerals magnetite and goethite, respectively. It was further suggested that the potential for cell growth and Fe(III) reduction was determined by the Fe(III) oxide surface area and not by crystallinity (36). Though Fe-containing clay minerals are operationally defined as crystalline Fe(III) minerals, rates of microbial clay reduction have been observed to be comparable to rates of reduction for poorly crystalline or amorphous Fe oxide minerals (16). In this study, we observed that growth was highly correlated to the particle concentration of smectite added to cultures of S. oneidensis (Fig. 3). The percentage of structural Fe(III) reduced in smectite remained at ~20% throughout the range of particle loading tested (Fig. 3). Thus, our results support previous observations which suggested that reduction and growth on crystalline Fe(III) minerals is determined by mineral surface area. We extend this concept to smectite clay minerals and suggest that their high surface area, comparable to that of amorphous Fe(III) oxyhydroxides (~700 m² g⁻¹) (38, 41), results in an increased availability for microbial reduction and growth.

Organic compounds such as humic acids are believed to facilitate the reduction of Fe(III) minerals by serving as an electron shuttle or by chelating and solubilizing Fe(III), thereby making the Fe more available for reduction. Our results concur with past studies to show that the reduction of smectite (26) or Fe(III) oxyhydroxide (21) is enhanced in the presence of the humic acid analog AQDS (Fig. 1; Table 1). However, we observed a minimal enhancement of cell growth in the presence of AQDS (Fig. 1 and 2).

Growth with smectite as the sole electron acceptor was also observed in Fe(III)-reducing consortia enriched from two very different environments (contaminated subsurface sediment and rice paddy soil) where Fe-rich clay minerals are abundant and potentially important electron acceptors (41, 52). The 16S rRNA gene sequences retrieved from these consortia were dominated by those closely related to the δ-Proteobacteria (Geobacteraceae) and low-G+C gram-positive members of the Bacteria (Desulfitobacterium and Clostridium). Retrieval of Geobacteraceae sequences is not unexpected, since these organisms have been established as important members of FeRB consortia in sediments (5, 37, 46). Detection of low-G+C gram-positive members of FeRB consortia is more surprising and intriguing. Members of the low-G+C Bacteria (Bacillus, Desulfitobacterium, and Desulfotomaculum) been shown to be capable of Fe(III) respiration (6, 30, 31, 32, 45, 47). Gram-positive organisms are thought to be more resistant to environmental extremes such as desiccation in soils and sediments.

Growth rate and yield of S. oneidensis strain MR-1 on various electron acceptors. S. oneidensis is a facultative anaerobe which grows well on oxygen and is incapable of fermentative growth (29, 39). Thermodynamic calculations of free energy are often used as the basis for comparing the energetics of Fe(III) respiration to aerobic respiration by FeRB of the Shewanellae family, though few quantitative growth data are available to support this comparison. Soluble Fe(III) forms such as ferric citrate are also believed to be more available for reduction and therefore may provide more energy for growth than solid Fe(III) minerals. To our knowledge, growth rates or yields of FeRB on solid versus soluble Fe forms have not been compared, and no studies have compared growth coupled to Fe(III) respiration versus aerobic respiration.

We observed growth yields of S. oneidensis as the increase in total cell carbon in the following order: O₂ > ferric citrate Fe(III) oxyhydroxide. When normalized per mole of electron acceptor (e−) utilized, growth yields were nearly identical for cells growing on Fe(III) oxyhydroxide (0.17 g of C mol of e−⁻¹) compared to ferric citrate (0.20 g of C mol of e−⁻¹), while the growth yield on oxygen was larger by a factor of 5 (Table 1). By measuring an increase in cell protein, Bazylinski et al. (4) observed a growth yield, similar to that in our study, of 0.40 g of cell C mol of Fe(III) reduced⁻¹ for Geobacter cells with ferric citrate as the electron acceptor.

In support of the growth yield data, cell sizes visualized under epifluorescence microscopy appeared to be substantially larger in cultures grown on soluble electron acceptors (oxygen or ferric citrate) than in those grown with solid electron acceptors [Fe(III) oxyhydroxide or smectite]. Therefore, we conclude that cell counts are not always directly proportional to growth yield in cultures of S. oneidensis. Such an observation suggests that the increase in biomass should be measured more often for these anaerobes. Given that previous perceptions on the growth of FeRB are heavily dependent upon direct counts in pure cultures, this observation could well revise our views on the energy obtained for growth via Fe(III) respiration. This conclusion is supported by the observation that both the predicted thermodynamic energy yield (Table 2) and the measured cell yield (Table 1) were higher by a factor of 5 when S. oneidensis cells were grown on oxygen than on various Fe(III) forms as the electron acceptor.

In agreement with the yield results (Table 1), growth rates of S. oneidensis calculated from the increase in cell number (Fig. 2) were higher for the soluble electron acceptors (0.489 × 10⁷ [O₂] and 0.520 × 10⁷ [Fe citrate] cells ml⁻¹ h⁻¹) in comparison to the solid electron acceptors (0.125 × 10⁷ [smectite] and 0.174 [FeOOH] cells ml⁻¹ h⁻¹) tested. Growth rates for smectite and FeOOH were similar to those observed in previous studies of S. putrefaciens growing on FeOOH (25). In contrast, our growth rates were 1 to 2 orders of magnitude higher than those observed in past studies of Shewanella strains growing on the crystalline Fe(III) oxide minerals goethite (36) and magnetite (13). It appears that the growth rate of Shewanella on smectite more closely resembles rates on poorly crystalline Fe(III) oxides (FeOOH) than those on crystalline Fe(III) oxides. We suggest that at least 20% of the Fe(III) bound in smectite is as available as FeOOH for relatively rapid growth. However, the fact that the growth rate of S. oneidensis on smectite was 25% lower than that on FeOOH suggests that there is some metabolic cost associated with smectitic Fe(III) utilization.

Biogeochemical significance. The form and concentration of reactive Fe(III) minerals are of paramount importance to the environmental significance of microbial Fe(III) reduction in sedimentary environments. Clay minerals are particularly important because they are highly reactive and account for a large fraction of Fe-containing minerals in nature (16, 41, 43, 48). In aquatic and marine sediments, silicates or clay minerals comprise 65% of all Fe minerals (48). Geochemical evidence has shown that Fe(III) in clay minerals is rapidly reduced and may constitute a significant fraction of the redox-active Fe from terrestrial environments to the deep sea (43, 52). Evidence also suggests that Fe-rich smectites comprise an important electron acceptor available for dissimilatory metal reducing metabolism in some surface (marine and aquatic) (12) and terrestrial subsurface sediments (52). In some contaminated subsurface sediments within the Department of Energy complex, iron-rich clay minerals are the primary electron acceptor available for microbial Fe(III) reduction (52). Thus, the microbial reduction of Fe-rich clay minerals is thought to have a significant impact on nutrient cycles, agricultural productivity, and the environmental fate of contaminants (8, 9, 41, 43, 51). However, the role of microbial clay reduction in natural environments has not been extensively determined.

Here we have demonstrated that FeRB, in a well-characterized pure culture and in purified enrichment cultures, can conserve energy for growth by coupling the reduction of structural Fe(III) bound in clay minerals to the oxidation of organic acids, lactate, or acetate. This is the first description of any organism capable of such metabolism. Given their abundance and ubiquity, Fe-containing clay minerals may be important and previously overlooked electron acceptors for the growth of bacteria in natural environments. Our results showing rapid bacterial growth via smectite respiration support past geochemical evidence (8, 9, 16, 17, 48) to indicate that a substantial portion (20 to 50%) of the Fe(III) bound in smectite is easily accessible to FeRB. Furthermore, our results show that the growth rate and yield on smectitic Fe(III) are comparable to those on poorly crystalline Fe(III) oxide minerals (FeOOH), which suggests that FeRB respire smectite in parallel with FeOOH. In other words, our data suggest that bacteria do not preferentially reduce Fe oxides over clay minerals. Microbial clay reduction may therefore be an important, but little-studied, process limiting natural and contaminant biogeochemical cycles.

This research was funded by the Natural and Accelerated Bioremediation Research (NABIR) program, Biological and Environmental Research (BER), the U.S. Department of Energy (grant DE-FG02-00ER62986), and by the U.S. Department of Agriculture (grant 98-35107-6313).

We thank Jun Wu for preparing the smectite clay and for assistance with the iron extractions. We acknowledge Lita Proctor and her lab personnel for their assistance with the epifluorescence microscopy as well as the technical support provided by the FSU Biological Science Imaging Resource Center. We also thank Jean Cowan of the Dauphin Island Sea Lab in Mobile, Ala., for help with the total organic carbon analyses.