Chemical and Biological Interactions during Nitrate and Goethite Reduction by Shewanella putrefaciens 200

Abstract

Although previous research has demonstrated that NO₃⁻ inhibits microbial Fe(III) reduction in laboratory cultures and natural sediments, the mechanisms of this inhibition have not been fully studied in an environmentally relevant medium that utilizes solid-phase, iron oxide minerals as a Fe(III) source. To study the dynamics of Fe and NO₃⁻ biogeochemistry when ferric (hydr)oxides are used as the Fe(III) source, Shewanella putrefaciens 200 was incubated under anoxic conditions in a low-ionic-strength, artificial groundwater medium with various amounts of NO₃⁻ and synthetic, high-surface-area goethite. Results showed that the presence of NO₃⁻ inhibited microbial goethite reduction more severely than it inhibited microbial reduction of the aqueous or microcrystalline sources of Fe(III) used in other studies. More interestingly, the presence of goethite also resulted in a twofold decrease in the rate of NO₃⁻ reduction, a 10-fold decrease in the rate of NO₂⁻ reduction, and a 20-fold increase in the amounts of N₂O produced. Nitrogen stable isotope experiments that utilized δ¹⁵N values of N₂O to distinguish between chemical and biological reduction of NO₂⁻ revealed that the N₂O produced during NO₂⁻ or NO₃⁻ reduction in the presence of goethite was primarily of abiotic origin. These results indicate that concomitant microbial Fe(III) and NO₃⁻ reduction produces NO₂⁻ and Fe(II), which then abiotically react to reduce NO₂⁻ to N₂O with the subsequent oxidation of Fe(II) to Fe(III).

In recent years, dissimilatory reduction of ferric (hydr)oxide minerals has been documented for a large number of microorganisms in a wide range of environments (31, 37) and has become recognized as an important constituent of the global carbon and iron cycles (31, 32). This process has a profound effect on groundwater geochemistry and places a key control on the fate of contaminant metals (12, 13, 19, 30, 60) and organic compounds (24, 42) in anoxic groundwater. As a consequence, processes that affect the rate and extent of microbial iron reduction in natural systems are of great interest. NO₃⁻ is a common groundwater contaminant in the United States (38) and may interfere with bioremediation schemes that seek to use microbial iron reduction to engender reductive immobilization of metal contaminants.

Shewanella putrefaciens 200 is a facultative anaerobe capable of utilizing O₂, NO₃⁻, NO₂⁻, Fe(III), Mn(IV), and a number of other compounds as terminal electron acceptors for carbon metabolism (14, 40, 41). Obuekwe and Westlake (40, 41) reported that NO₃⁻, NO₂⁻, and a number of other terminal electron acceptors can inhibit reduction of Fe(III)-phosphate by S. putrefaciens 200, and subsequent work (1, 51) has demonstrated that the presence of NO₃⁻ can inhibit microbial Fe²⁺ production during sediment incubations. Additional studies with S. putrefaciens and other bacteria (14) have also observed that NO₃⁻ or NO₂⁻ inhibit microbial iron reduction. Studies with S. putrefaciens 200 suggest that this inhibition may result from kinetically favorable transport of electrons to NO₃⁻ or NO₂⁻, a mechanism suggested by Arnold et al. to explain inhibition of Fe(III) reduction by oxygen (3-5, 14). Although the respiratory electron transport chain of S. putrefaciens is known to include a variety of b- and c-type cytochromes and quinones (36, 42, 55), the sequential arrangement of electron carriers has not been elucidated, and terminal NO₃⁻, NO₂⁻, or Fe(III) reductases have not been fully purified. While the NO₃⁻ and NO₂⁻ reductases for S. putrefaciens MR-1 have been partially purified by Krause and Nealson (27), MR-1 is a denitrifier, whereas the S. putrefaciens 200 used in our studies primarily reduces NO₃⁻ and NO₂⁻ to ammonia (Cooper, Coby, and Picardal, unpublished data). It is therefore likely that strains 200 and MR-1 have substantial enzymatic differences.

Although illuminating, previous studies provide only limited insight into the dynamics of interactions between dissimilatory iron and NO₂⁻ or NO₃⁻ reduction (NO_x⁻ reduction) when solid-phase, ferric (hydr)oxide minerals (e.g., ferrihydrite, goethite, lepidocrocite) are the Fe(III) source. The chemistry of ferric (hydr)oxide minerals (FeOOH) is fundamentally different from that of aqueous or microcrystalline Fe(III) used by others in the previous studies. Not only does the FeOOH provide a textured surface that sorbs metal cations, but Fe²⁺ adsorption to a FeOOH surface forms highly reactive ferrous-ferric surface complexes that have distinctive chemical properties that are often similar to those of green rust (16, 20, 22, 35, 52). Biogeochemical interactions in systems containing NO₃⁻ and solid-phase Fe(III) therefore clearly may be different than those previously described in systems using aqueous Fe(III). The objective of this study is to determine the nature of the biological and geochemical interactions between microbial ferric (hydr)oxide and NO₃⁻ reduction using S. putrefaciens 200 as a model microorganism capable of reducing both Fe(III) and NO₃⁻ and synthetic goethite as a model ferric (hydr)oxide mineral.

MATERIALS AND METHODS

Bacterial strain and growth conditions.

S. putrefaciens is a gram-negative motile rod with an obligate respiratory metabolism (49). The strain used in these experiments, S. putrefaciens 200, was originally isolated from a Canadian oil pipeline by Obuekwe (39). The culture was maintained on a solid medium of nutrient agar containing 5 g of yeast extract/liter (Difco, Detroit, Mich.), as previously described (42). Liquid cultures were grown in a 2.5-liter Bioflow 3000 fermentor (New Brunswick Scientific Company, Edison, N.J.) in a medium that consisted (per liter) of 2.0 g of Na₂SO₄, 0.5 g of K₂HPO₄, 1.0 g of NH₄Cl, 0.198 g of CaCl₂ · 2H₂O, 0.1 g of MgSO₄ · 7H₂O, 19.35 mg of FeCl₃ · 6H₂O, 0.5 g of yeast extract, and 3 ml of 60% (wt/vol) aqueous sodium lactate. Following an initial period of aerobic growth, anaerobic reductase activity was induced by reducing the airflow to approximately 100 ml min⁻¹ and maintaining the cells under suboxic ([aqueous O₂ {O_2(aq)}] < 2.5 μmol/liter) conditions for a period of 12 h. It has been previously demonstrated that this culture technique induces anaerobic reduction systems for a wide range of terminal electron acceptors, including nitrate, nitrite, and Fe(III) (14). Cells were subsequently harvested by centrifugation and resuspended to a target optical density (A₆₀₀, ∼1.2) in low-ionic-strength, artificial groundwater medium (AGW) (12).

A nitrate-reducing enrichment culture was developed from nitrate-contaminated sandy sediments obtained from a Uranium Mill Tailings Remedial Action (UMTRA) site in Shiprock, New Mexico, in October 1999 during sediment collection by the U.S. Department of Energy. Sediments were collected with a backhoe, placed in whirl-pack bags, flushed with argon, placed in sealed argon-flushed bottles, and stored at 4°C aphotically for 1 month prior to use. Enrichment of nitrate-reducing cultures was done with a sulfate-free AGW medium with 10 mM lactate as the electron donor and 2.5 mM nitrate as the electron acceptor. The sulfate-free medium was created by substituting chloride salts for sulfate salts in the AGW medium previously described (12). To establish enrichments, sterile sulfate-free AGW medium was degassed with nitrogen and dispensed (50 ml) into 120-ml serum bottles in an anaerobic chamber (Coy Laboratory Products, Grass Lake, Mich.). After the addition of approximately 10 g of sediment, the bottles were crimp sealed with butyl rubber stoppers and statically incubated in the dark at room temperature. Transfers (10% volume) were done every 7 to 10 days into identical medium in anoxic serum bottles or tubes. After eight transfers, an enrichment culture was developed that was capable of complete utilization of 2.5 mM nitrate in 1 to 3 days. Subsequent experiments revealed that the enrichment culture was also capable of slow reduction of solid-phase Fe(III) oxides.

Fe(III) reduction and NO₃⁻ reduction experiments.

The AGW medium and goethite slurries used in these experiments have been described previously (12). When required, 1.0 ml of sterile NO₃⁻ or NO₂⁻ stock solutions of NaNO₃ and NaNO₂ (respectively) dissolved in Milli-Q water were added to the AGW medium (goethite free) or AGW + goethite slurry (NO_x⁻ + goethite). Although the general methodology employed in the Fe(III) and NO₃⁻ reduction experiments has been previously described (12), a summary of the procedure is presented here. Experiments were initiated by inoculating slurries with 2 ml of S. putrefaciens (or UMTRA enrichment culture) suspension under anoxic conditions (ca. 2 × 10⁶ cells ml⁻¹). For each experiment, a series of 150-ml (nominal volume) glass serum bottles were crimp sealed with butyl rubber stoppers and incubated horizontally on a shaker table at room temperature (95% N₂-5% H₂ headspace). Initial samples (t₀) were taken prior to inoculation, and subsequent samples were taken at regular intervals thereafter. During sampling, bottles were transferred to the anoxic chamber, shaken vigorously to resuspend any settled solids, and then immediately sampled with a 3-ml syringe (18-gauge needle). One 1.5-ml aliquot was extracted for 2 h in 2 M KCl for total ammonium-nitrogen (NH₃-N) measurements. The second 1.5-ml aliquot of slurry was placed into a separate microcentrifuge tube, and particles were separated by centrifugation (25 min at 14,900 × g). The supernatant was immediately sampled for pH, Fe²⁺_(aq), NH₃-N_(aq), NO₃⁻, and NO₂⁻. Aqueous Fe²⁺ was immediately analyzed via a modification of the ferrozine method (26, 53). Samples for NH₃-N, NO₃⁻, and NO₂⁻ were diluted 2× to 5× in Milli-Q water, frozen, and stored for later analysis. NO₃⁻ and NO₂⁻ were analyzed via ion chromatography (Dionex Series 4500i; column type, IonPac AS17 [4 by 250 mm]; eluent, 50 mM NaOH and Milli-Q H₂O gradient), and aqueous NH₃-N was analyzed colorimetrically via the phenate method (21). Extraneous supernatant was then discarded, and the pellet was resuspended and extracted for 2 h in 0.5 N HCl. Both slurries (KCl and HCl) were separated via centrifugation (10 min at 14,900 × g). The HCl supernatant was immediately analyzed for bound Fe²⁺ via the ferrozine method, and the KCl supernatant was diluted 2× in Milli-Q water, frozen, and stored for later analysis via the phenate method. Total 0.5 N HCl-extractable Fe²⁺ (Fe²⁺-HCl) was defined as the sum of aqueous and bound Fe²⁺, and the precision of the phenate analysis was monitored by comparing total and aqueous NH₃-N in the goethite-free bottles.

For N₂O analyses, 25.0 μl of headspace gas was withdrawn with a Hamilton gas-tight syringe and analyzed via gas chromatography with a Hewlett Packard 5890 Series II gas chromatograph equipped with a GS-Q megabore capillary column (conditions: oven temperature, 75°C; detector temperature, 275°C; pressure, 130 kPa) and electron capture detector. The instrument was calibrated with both laboratory and commercial N₂O standards (Scotty I analyzed gases), and gas concentration was defined as the average of data for three injections. The concentration of aqueous N₂O was calculated from temperature and headspace N₂O by Henry's law (59), and total N₂O was defined as the sum of headspace and aqueous N₂O. For N₂ analyses, 3.0-ml aliquots of headspace were withdrawn with a syringe and immediately injected into a Shimadzu model 14A gas chromatograph equipped with a CR501 Chromatopac, 1.0-ml sample loop, 3.0 m MS-5A 60/80, and 1.0 m porapak R 80/100 columns (2.5 min run time; isothermal at 75°C) and thermal conductivity detector. To guard against contamination from atmospheric N₂, the syringe was slowly compressed during injection, and excess volume acted as a purge for the sample loop. Check standards and He blanks were also analyzed at regular intervals. The instrument was calibrated with laboratory N₂ standards, and gas concentration was defined as the average of data for two injections. With the sole exception of the stable isotope studies (described below), the exact same methodology was used for all experiments with S. putrefaciens 200 and the UMTRA enrichment cultures. All experiments were conducted with the necessary controls.

Stable isotope experiments.

With noted exceptions, stable isotope experiments were conducted as described previously. Experiments were conducted under a He atmosphere in specially fabricated Pyrex media bottles (∼500 ml) equipped with a vacuum stopcock and crimp-sealed sampling port. After preparation of the experimental medium (100 ml) and replacement of the headspace with He, the bottles were allowed to equilibrate for 48 h, and t₀ samples were taken from all bottles prior to inoculation (biological and combined experiments) or NO₂⁻ addition (chemical experiments). Microbially reduced goethite (MrG) for the chemical experiment was harvested from previous NO₃⁻-free experiments, sterilized via pasteurization (25 min at 95°C), and washed three times with sterile, anoxic, lactate-free AGW medium prior use in isotope experiments.

Isotopic analytical requirements dictated that the entire ∼400-ml headspace be collected. Thus, four bottles (two inoculated and two uninoculated) were continuously monitored for aqueous parameters and headspace N₂O, while six isotope bottles were continuously monitored for headspace N₂O but sampled for aqueous parameters only at the initial and final times. Two isotope bottles were sacrificed at each sampling point, and procedural standards were randomly extracted to monitor the overall precision of the extraction process.

The isotope extraction was conducted in two stages. During the first stage, aqueous slurries were frozen immediately after wet chemical sampling was complete (2 to 4 h; standard commercial freezer) and the headspace-N₂O was subsequently removed and stored for isotope analysis. The headspace separation was achieved by cryogenically trapping the condensable gasses (N₂O, CO₂, H₂O) in a liquid-N₂ cold trap and slowly removing the incondensable gas (vacuum pumping) while cryogenically trapping any remaining headspace condensable gas. Next, the condensable gas was warmed and cryogenically transferred to a vacuum-tight storage vessel containing a few grains of degassed coconut charcoal. During the second stage, the condensable gas in the storage vessel was further purified on a more advanced vacuum line by cryogenically separating water and traces of incondensable gas from N₂O and CO₂. The purified N₂O and CO₂ was then cryogenically transferred to a glass ampoule containing excess granular Cu⁰ (ca. 2 g), sealed under vacuum, heated overnight at 500°C to reduce the N₂O to N₂, and stored for subsequent determination of δ¹⁵N values. Since the biological experiments did not produce the requisite 5 μmol total N₂O needed for accurate δ¹⁵N analysis, 5 μmol of standard N₂O was injected into the sampling port of the isotope bottle during the initial trapping of condensable gasses.

The values for δ¹⁵N of N₂O (δ¹⁵N-N₂O) for the biological experiment are based on dilution calculations utilizing the δ¹⁵N-N₂O value for the standard, the δ¹⁵N-N₂O value for a mixture of the sample and standard, and the mass ratio of sample to standard N₂O. The N₂O standard for δ¹⁵N-N₂O analyses was a 2.0-liter aliquot of medical-grade N₂O gas (UN1070, Air Products Corp.) stored at 3,000 kPa to prevent fractionation effects from N₂O liquid-gas phase changes at high pressure. Two different aliquots of standard gas were needed for these studies, and these standards had δ¹⁵N-N₂O values of 0.02 ± 0.17‰ (n = 9) (biological experiments) and −2.12 ± 0.40‰ (n = 9) (chemical and combined experiments). The fractionation effect associated with the previously described method for extracting headspace N₂O in equilibrium with AGW medium was calculated to be −2.8 ± 1.0‰. All reported δ¹⁵N-N₂O values account for this extraction bias. δ¹⁵N values of elemental nitrogen samples were determined off-line, manually, in dual-inlet mode with a Finnigan MAT 252 mass spectrometer with a cryogenic trap at the inlet. We used the ammonium sulfate nitrogen stable isotope standards NBS-N-1 and NBS-N-2 as internal standards for calibration and report our results relative to air nitrogen (air nitrogen δ¹⁵N ≡ 0‰), with a mass-spectrometric precision of ± 0.2‰.

RESULTS

NO₃⁻ inhibition of goethite reduction.

Our batch experiments with goethite as the Fe(III) source in the absence of NO₃⁻ (data not shown) yielded results similar to those of previous researchers who have investigated microbial reduction of iron (hydr)oxide minerals. Fe²⁺ production rates are initially rapid and then slow with time, possibly as a result of passivation of the cell and iron oxide surfaces by adsorbed Fe(II) (44, 45). It is important to note that approximately 50% of the microbially produced Fe²⁺ is associated with the mineral or cellular surface (precipitated, adsorbed, or a reactive surface complex). Data in Fig. 1 illustrate the effect of NO₃⁻ on the rate and extent of Fe²⁺ production during goethite reduction by S. putrefaciens 200. All slurries reduced NO₃⁻ (data not shown), but only the slurries containing initial concentrations of 0.0, 1.0, and 2.5 mM NO₃⁻ produced Fe²⁺ at concentrations significantly above those in the uninoculated controls. The presence of 1.0 mM NO₃⁻ resulted in a 50% reduction in the initial rate (∼28 h) of Fe²⁺ production (ca. 2× slower) and prevented significant Fe²⁺ production for a period of 8 h. The presence of 2.5 mM initial NO₃⁻ prevented significant Fe²⁺ production for over 200 h, and no Fe²⁺ production was observed in the 5 mM NO₃⁻ treatment. No measurable effect on the ultimate extent of iron reduction was observed for the 1 mM NO₃⁻ treatment.

FIG. 1.

Effect of NO₃⁻ on Fe(II) production (sum of aqueous and bound Fe) via dissimilatory reduction of synthetic goethite by S. putrefaciens 200. Data from experiments using no NO₃⁻ (•), 1.0 mM initial NO₃⁻ (○), 2.5 mM initial NO₃⁻ (□), 5.0 mM initial NO₃⁻ (▵), and uninoculated controls (▾) are plotted versus time. Solid lines connect the points for experiments with no NO₃⁻ and 1 mM NO₃⁻, and dotted lines connect all other points. Data represent average values of three replicates, and the error bars reflect standard deviation. Note the split axes.

Subsequent experiments were conducted with goethite plus one concentration of NO₃⁻ (2.5 mM initial NO₃⁻). This decision was made because the large sampling matrix prevented experimentation at multiple NO₃⁻ concentrations, and 2.5 mM represented a good midpoint between the 0.0 mM NO₃⁻ and 5 mM NO₃⁻ end members. These experiments show typical patterns of NO₂⁻ and N₂O production by S. putrefaciens 200 during NO₃⁻ reduction in the absence (Fig. 2a) and presence (Fig. 2b and c) of goethite and reveal that a very small amount of surface-associated Fe²⁺ (∼0.1 mM) is produced while NO₃⁻ is still present even though Fe²⁺ production rates are minimal until after the NO₂⁻ supply has been exhausted (Fig. 2b and c). More than 70% of the added NO₃⁻ was recovered as NO₃⁻, NO₂⁻, N_2, N₂O, or NH₃-N in all experiments with NO₃⁻ and goethite (data not shown). In these experiments, the proportion of nitrogen recovered was initially 100% and then decreased as experiments progressed and cell growth occurred. Since the modest amount of cell growth could not account for the unrecovered N, it is likely that products or complexes were formed that were not included in the N species quantified. Ammonia was the dominant end product of NO_x⁻ reduction in our experiments, and only minimal amounts of N₂ were produced (Fig. 3). Since the Fe²⁺-Fe³⁺ redox transition (E_H⁰ ≅ 0.8 V) is dominant in our system, this observation is consistent with a report by Samuelsson that ammonia is the dominant end product (and that N₂O is a minor by-product) of NO₃⁻ reduction by a different strain of S. putrefaciens at redox potentials greater than or equal to ∼0 mV (46).

FIG. 2.

NO₃⁻ reduction by S. putrefaciens 200 in the presence and absence of goethite. (a) Experiment with goethite absent, showing data for NO₃⁻ (○), NO₂⁻ (•), and N₂O (□). (b) Experiment with goethite present, showing data for NO₃⁻ (○) and NO₂⁻ (•). (c) Experiment with goethite present, showing data for N₂O (▾) and Fe(II)-HCl (□). Data represent average values of three replicates, and the error bars reflect standard deviation. Note the split axes and the dual y axes in Fig. 2a.

FIG. 3.

NH₃-N (circles) and N₂ (squares) production during 2.5 mM NO₃⁻ reduction by S. putrefaciens 200 in the presence (solid symbols) or absence (open symbols) of goethite. Whereas NO₃⁻ and biogenic NO₂⁻ were completely consumed in bottles lacking goethite, a mean NO₂⁻ concentration of 1.0 mM remained in bottles containing goethite at the conclusion of the experiment. Data represent average values of three replicates, and the error bars reflect standard deviation. Note the split y axis.

Goethite inhibition of NO₃⁻and NO₂⁻ reduction.

Results from NO₃⁻ reduction experiments conducted with 2.5 mM NO₃⁻ in the absence of goethite are summarized in Fig. 2a. Most NO₂⁻ was reduced within 48 h, but low levels persisted through the first 100 h. Observed trends in mass balance were statistically indistinguishable from experiments conducted in the presence of goethite, and trends in NH₃-N and N₂-N product distribution were similar as well. Examination of these data and comparison with results summarized in Fig. 2 reveal three important observations, namely (i) significantly more N₂O was produced when NO₃⁻ was reduced in the presence of goethite (Fig. 2c) than in the absence of goethite (Fig. 2a); (ii) in both systems, N₂O production was not observed until after all NO₃⁻ had been consumed (Fig. 2a to c); and (iii) the rates of NO₃⁻ and NO₂⁻ reduction were notably higher in the absence of goethite (Fig. 2a) than in the presence of goethite (Fig. 2b).

Table 1 provides summary data showing that the presence of goethite can inhibit the rate of dissimilatory NO₃⁻ and NO₂⁻ reduction by S. putrefaciens. This table summarizes results from experiments that compare the initial rates of Fe(III) and NO_x⁻ reduction across various experimental matrices. The presence of NO₃⁻ resulted in an ∼20-fold decrease in the rate of microbial goethite reduction, and the presence of goethite resulted in an ∼10-fold decrease in the rate of NO₂⁻ reduction and an ∼2-fold decrease in the rate of NO₃⁻ reduction. In addition, data presented in Fig. 4 indicate that the presence of goethite has a marked affect on the observed degree of N₂O production in slurries containing 2.5 mM NO₃⁻. Indeed, the N₂O produced in slurries containing goethite and 2.5 mM NO₃⁻ was even substantially higher than in incubations containing 5.0 mM NO₃⁻ lacking goethite.

TABLE 1.

Comparison of initial rates of NO₃⁻ reduction, NO₂⁻ reduction, and Fe²⁺ production across various experimental matrices

Matrix	NO3− reduction (μmol/liter/h)	NO2− reduction (μmol/liter/h)	Fe(II) production (μmol/liter/h)
Goethite only	NAa	NA	43.0 ± 24.0 (n = 6)
NO3− only	168 ± 72.5 (n = 6)	19.2 ± 2.1 (n = 6)	NA
Goethite and NO3−	72.7 ± 29.2 (n = 8)	2.5 ± 0.5 (n = 6)	2.3 ± 0.3 (n = 4)

^aNA, not applicable.

FIG. 4.

Effect of goethite on N₂O production rate with 5 mM initial NO₃⁻ without goethite (□), 2.5 mM initial NO₃⁻ without goethite (○), and 2.5 mM initial NO₃⁻ with goethite (•). Data represent average values of three replicates, and the error bars reflect standard deviation. Note the split x and y axes.

In order to determine if the observed mutual inhibition was due to a process unique to S. putrefaciens 200, similar experiments were also conducted with enrichment cultures. Data from these experiments (Fig. 5) indicate that the presence of goethite can also inhibit dissimilatory reduction of NO₃⁻ and NO₂⁻ by a natural enrichment culture under conditions that are analogous to those used in experiments with S. putrefaciens. These cultures produced 200 μM HCl-soluble Fe(II) and virtually no N₂O over the course of the experiments (data not shown). Although these experiments yielded different trends in NO₃⁻ reduction by-products, the data clearly demonstrate that inhibition of microbial NO₃⁻ and NO₂⁻ reduction by goethite is not limited to S. putrefaciens 200. Thus, this phenomenon may be important in a wide variety of sedimentary systems.

FIG. 5.

Effect of goethite on NO₃⁻ reduction by UMTRA enrichment culture with NO₃⁻ without goethite (○), NO₂⁻ without goethite (□), NO₃⁻ with goethite (•), and NO₂⁻ with goethite (▪). All experiments had 2.5 mM initial NO₃⁻. Data represent average values of three replicates, and the error bars reflect standard deviation. When not shown, error bars are smaller than the symbol size.

Stable isotope studies (δ¹⁵N-N₂O).

Data presented in Fig. 6 display the relationship between extent of reaction (percentage of available NO₂⁻ reduced) versus the δ¹⁵N-N₂O produced by the reduction of NO₂⁻ by (i) S. putrefaciens in the absence of goethite (microbial experiment), (ii) sterile microbially reduced goethite in the absence of S. putrefaciens (chemical experiment), and (iii) NO₃⁻ reduction by S. putrefaciens in the presence of goethite (combined experiment). This comparison and subsequent calculations assume that all N₂O is produced via NO₂⁻ reduction and that significant N₂O and ammonia production does not begin until all NO₃⁻ has been converted to NO₂⁻. This assumption is supported by experimental results (Fig. 2 to 4) and allows us to assume no fractionation between NO₃⁻ and NO₂⁻ in the combined experiment.

FIG. 6.

δ¹⁵N values of N₂O for microbial (○), chemical (•), and combined (⋄) experiments plotted versus extent of reaction.

Based on this assumption, the combined data have been normalized to account for the difference in the δ¹⁵N of the initial NO₃⁻ (+6.8 ± 0.2‰, combined experiment) and NO₂⁻ (2.0 ± 0.2‰, chemical and microbial experiments) in order to express all data on the same basis (NO₂⁻ reduction). The clear separation between the δ¹⁵N-N₂O_MICROBIAL (ca. +45 ± 27‰) and the δ¹⁵N-N₂O_CHEMICAL (ca. −25 ± 8‰) indicates that the stable isotopic signature of nitrogen in N₂O (δ¹⁵N-N₂O) can be used in our experiments to discriminate between N₂O of microbial origin and N₂O of chemical origin. Because we wanted to limit our arguments to cases where the NO₂⁻ concentration is at least 10× greater than the N₂O concentration, we have considered only δ¹⁵N-N₂O_MICROBIAL data points where at least 50 μmol NO₂⁻/liter remains (98% of original NO₂⁻). Even if all of the N₂O_MICROBIAL data points are considered, however, the new value of δ¹⁵N- N₂O_MICROBIAL (ca. +31 ± 29‰) is still notably different from that of the δ¹⁵N-N₂O_CHEMICAL (ca. −25 ± 8‰). The large degree of error in the values for δ¹⁵N- N₂O_MICROBIAL arises from the large standard dilutions needed to obtain the 5 μmol of N₂O needed for isotopic analysis (microbial experiments typically yielded less than 0.3 μmol N₂O). The observed value for the NO₂⁻-normalized δ¹⁵N-N₂O_COMBINED (ca. −28 ± 5‰) coincides closely with the observed value for δ¹⁵N-N₂O_CHEMICAL and therefore indicates that more than 95% of the N₂O produced during the combined reaction is of chemical origin.

Although the δ¹⁵N-N₂O_MICROBIAL data presented in Fig. 6 are enriched in ¹⁵N with respect to the initial NO₂⁻ pool in all cases, the extent of enrichment varied substantially in different replicate experiments when >95% of the available NO₂⁻ had been consumed. Although the cause for this variation at the conclusion of the experiments is not clear, the variation does not materially affect our conclusion that the N₂O from the combined experiment is primarily of chemical origin.

DISCUSSION

NO₃⁻ inhibition of goethite reduction.

It has been previously reported that inhibition of iron (hydr)oxide reduction by NO₃⁻ also inhibited changes in the speciation of oxide-sorbed metals (12), and previous researchers have also reported NO₃⁻ inhibition of iron reduction by S. putrefaciens species. DiChristina (14) reported that 15 mM initial NO₃⁻ inhibited ferric chloride reduction by 46 to 92% and inhibited ferric citrate reduction by 2 to 22%. Similarly, Obuekwe and Westlake (40, 41) reported that the presence of 1 mM NO₃⁻ resulted in an ∼50% decrease in the rate of soluble Fe³⁺-phosphate (2% solution) reduction by a Pseudomonas sp. (later identified as S. putrefaciens 200). More recent experiments by Lee et al. with S. putrefaciens DK-1 also showed significant inhibition of 10 mM Fe(III) citrate reduction in the presence of 10 mM NO₃⁻ (28). These authors observed no inhibition until 25 h, after which Fe(II) production ceased and concentrations decreased by more than 30% over the course of their 95-h experiment. It is difficult to directly to compare data for extent of iron reduction due to differences in medium composition. Nevertheless, a comparison of reaction rates indicates that NO₃⁻ clearly inhibited the microbial reduction of crystalline iron (hydr)oxide minerals more strongly than it inhibits reduction of ferric citrate or the microcrystalline sources produced when ferric chloride is added to a circumneutral buffered medium.

Data in Fig. 1 also indicate that instead of simply slowing the rate of Fe²⁺ production, the initial presence of 1 mM NO₃⁻ appears to prevent any significant accumulation of Fe²⁺ for the first two sampling periods. This observation would, on the surface, seem contrary to the results of DiChristina (14) and Obuekwe and Westlake (40, 41). DiChristina's studies of competitive usage of NO₃⁻ and Fe(III) by S. putrefaciens 200 suggested only partial inhibition of Fe(II) production by 15 mM NO₃⁻. His studies were done with either ferric citrate or ferric chloride as the Fe(III) source. Ferric citrate is an aqueous complex, whereas an acidic solution of ferric chloride would immediately precipitate a ferric oxyhydroxides mineral at circumneutral pH. Thus, the experiments with ferric chloride were, in effect, done with an amorphous or microcrystalline Fe(III) oxyhydroxides, while the experiments with ferric citrate were done with a true aqueous Fe(III) source. Interestingly, his work showed that NO₃⁻ inhibition of Fe(II) production by microaerobically grown cells was 3- to 23-fold greater in the presence of this solid-phase Fe(III) (FeCl₃) than in the presence of Fe(III) citrate. Obuekwe's work, which also showed simultaneous NO₃⁻ reduction and Fe(II) production, utilized Fe(III) phosphate in which the Fe(III) is typically chelated by additional citrate. It would therefore appear that significant Fe(II) production occurs simultaneously with NO₃⁻ reduction when the Fe(III) is supplied in a chelated form. When Fe(III) is presented in the form of a solid-phase oxyhydroxide that would allow sorption of biogenic Fe(II), inhibition of Fe(II) production by NO₃⁻ is more substantial.

With respect to the N speciation and N balance in our system, our results are comparable to those of other researchers who have also reported incomplete recovery of N species during nitrate reduction to ammonia. Samuelsson and Rönner (47) reported that only 25 to 75% of NO₃⁻ added to isolates from the Baltic Sea was recovered as ammonia. In experiments with a different strain of S. putrefaciens (formerly Pseudomonas putrefaciens), Sammuelsson was unable to account for 6 to 91% of the NO₃⁻ consumed. In those experiments, only NH₄⁺, N₂O, and NO₂⁻ were measured, and the missing N was attributed to N₂ or some other unmeasured N species (46). Using S. putrefaciens strain MR-1, Krause and Nealson, however, found little or no ammonia production and suggested that the carbon/NO₃⁻ ratio may also play a role in determining whether S. putrefaciens produces gaseous compounds (N₂ or N₂O) or ammonia during dissimilatory NO₃⁻ reduction (27).

It is generally accepted that under most conditions, microorganisms can generate more energy from dissimilatory NO₃⁻ reduction than from dissimilatory Fe(III) reduction. Consequently, NO₃⁻ inhibition of microbial iron (hydr)oxide reduction may result from differences in the rate of electron transport to NO_x⁻ and Fe(III) that have evolved to allow preferential use of the most favorable oxidant (1, 14, 41, 50, 51). Alternate explanations, however, are also possible. Since a small amount of Fe(II) was produced in our goethite-containing bottles as NO₃⁻ reduction proceeded (Fig. 2c), NO₃⁻, NO₂⁻, and Fe²⁺ were simultaneously present in the same system. Under suitable conditions, Fe²⁺ can chemically reduce both NO₃⁻ to NO₂⁻ (11) and NO₂⁻ to N₂O and/or ammonia (10, 40, 57). Thus, concomitant reduction of NO₃⁻ and Fe(III) may create an apparent inhibition of microbial iron reduction by oxidizing Fe²⁺ to Fe(III) via chemical NO_x⁻ reduction. Results from previous studies testing this possibility are somewhat contradictory but do reveal two general themes. First, the chemical rate of NO₂⁻ reduction via oxidation of aqueous Fe²⁺ (produced via the microbial reduction of ferric chloride or ferric citrate) is too slow to account for a significant degree of NO_x⁻ inhibition of microbial Fe(III) reduction (14). Second, the chemical rate of NO₂⁻ reduction via oxidation of surface-associated Fe²⁺ [produced via the microbial reduction of iron (hydr)oxide minerals] is notably faster than the rate of NO₂⁻ reduction by aqueous Fe²⁺ (57). This finding that Fe²⁺-Fe(III) moieties (e.g., Fe²⁺ adsorbed to FeOOH minerals, green complexes, and/or ferrous precipitates) can reduce NO₂⁻ more rapidly than aqueous Fe²⁺ is supported by previous geochemical research into the catalytic effect of these moieties on the rate of NO_x⁻ reduction to N₂O and/or ammonium (22, 23, 52). This information, coupled with the observation that goethite reduction in the presence of NO₃⁻ produced notably more N₂O than NO₃⁻ reduction alone (Fig. 4), suggests that concomitant NO₃⁻ and goethite reduction in our systems does result in chemical reoxidation of Fe²⁺ coupled to NO₂⁻ reduction to N₂O. However, the extremely small amount of N₂O produced by this reaction indicates that most NO₂⁻ in our system is reduced enzymatically.

Goethite inhibition of NO₃⁻ and NO₂⁻ reduction.

While our work agrees with previous studies with S. putrefaciens 200 which indicate that the presence of NO₃⁻ will substantially inhibit microbial Fe(III) reduction (14), the observation that goethite inhibits NO₃⁻ and NO₂⁻ reduction is unprecedented and differs from the results of DiChristina (14) and Lee et al. (28), who reported that the presence of Fe(III)-chloride and/or Fe(III)-citrate had no notable effect on the observed rates of NO₃⁻ and NO₂⁻ reduction by S. putrefaciens. This disparity likely arises from significant differences in the nature of the Fe(III) source used in these experiments and the time scale of the experiments described by DiChristina (minutes to hours) and by the present work (hours to weeks). In addition, the observation that natural enrichment cultures can also show goethite inhibition of NO₃⁻ reduction (Fig. 5) indicates that this process may be important in natural systems with high iron content and low flow rates and thus merits significant further investigation.

Several mechanisms can potentially explain the goethite inhibition of NO_x⁻ reduction in both S. putrefaciens 200 and the UMTRA enrichment culture.

(i) The presence of goethite particles might have a toxic effect. Direct goethite toxicity is probably not a factor, since goethite is not known to be toxic, especially to microorganisms capable of dissimilatory iron reduction. To examine the prospect that goethite particles could have abraded the cells and either increased cell mortality or made cells more susceptible to other toxic effects, cells were incubated (5 days at 30°C with no carbon substrate and in the same experimental medium) in the presence and absence of goethite and a series of other sediments. In all cases, cells incubated in the presence of a solid surface yielded a higher final cell number than cells incubated in the absence of a solid surface (data not shown). Thus, cell abrasion probably does not contribute significantly to the observed goethite inhibition of microbial NO_x⁻ reduction.

(ii) If all NO_x⁻-reducing cells are coated by a dense layer of goethite crystals, such a coating might reduce the diffusive flux of NO_x⁻ to the cell surface. There is currently no evidence, however, that such dense coatings are formed or that they would limit NO_x⁻ diffusion to the cell surface.

(iii) If rates of electron transport to NO_x⁻ and Fe(III) are similar, goethite inhibition of NO_x⁻ reduction could result from simple kinetic competition between the two electron acceptors, a mechanism previously suggested to explain inhibition of Fe(III) reduction by oxygen (4). However, given the very small amount of Fe(III) reduction observed in the presence of NO_x⁻, the extent of inhibition shown in Table 1 probably cannot be explained via a competitive mechanism alone.

(iv) Prolonged exposure to elevated concentrations of NO₂⁻ may be toxic (2). This mechanism, however, would not explain why NO₂⁻ reduction proceeded much more slowly in cultures containing goethite than in cultures absent goethite. Oxygen uptake studies on cells that had been cultured aerobically in the presence of NO₂⁻ indicated that NO₂⁻ showed little toxicity at concentrations up to 5 mmol/liter (data not shown). When combined with the observation that the NO₂⁻ reduction rate in the absence of goethite and presence of high (∼2 mM) NO₂⁻ concentrations (Fig. 2a) was markedly greater than the NO₂⁻ reduction rate in the presence of goethite and similar NO₂⁻ concentrations (Fig. 2b), these experiments indicate that NO₂⁻ toxicity could not explain the observed goethite inhibition of NO_x⁻ reduction.

(v) Chemical oxidation of biogenic Fe²⁺ by NO₂⁻ results in the nucleation of ferric (hydr)oxide minerals on the surface of a cell, forming coatings which impede transport of NO_x⁻ into the cell. Such a hypothetical mechanism would proceed via three steps. Firstly, S. putrefaciens concomitantly reduces NO₃⁻ and, to a lesser extent, goethite, producing NO₂⁻ and small amounts of aqueous Fe²⁺. Some of the microbially produced Fe²⁺ then sorbs to the cellular surface, forming a reactive complex that can catalyze the oxidation of Fe²⁺ to Fe(III) via the chemical reduction of NO₂⁻ to N₂O. Most of the NO₂⁻ produced is reduced biologically to ammonia and/or N₂, with N₂O as a minor by-product. Lastly, the Fe(III) precipitates as an iron oxyhydroxide mineral. This mineral can nucleate on the cell surface and thereby inhibit transport of NO₃⁻ and NO₂⁻ to NO_x⁻ reductases inside the cell. Since chemical NO₂⁻ reduction via Fe²⁺ oxidation is known to produce N₂O (22, 23, 52, 57), this mechanism is supported by the increased N₂O production observed to occur during concomitant NO_x⁻ and goethite reduction (Fig. 4). In addition, the data presented in Fig. 6 provide compelling evidence that N₂O is being produced via chemical NO₂⁻ reduction via Fe²⁺ oxidation. An inhibitory mechanism involving the presence of Fe oxyhydroxide coatings on the cell surface is supported by the recent work of Liu et al. (29), who observed the formation of Fe mineral coatings on S. putrefaciens after exposure of cells to Fe²⁺. These coatings coincided with a lag phase during which reduction of Fe(III) citrate was inhibited. The mineral coatings disappeared as cells recovered from the lag phase, and the authors suggested that the cells had developed a mechanism for coating removal. Formation and subsequent removal of such coatings in our experiments would explain the long lag period seen in Fig. 2b and c (between ∼50 and 300 h), during which very little NO₂⁻ or Fe(III) was reduced.

At the current time, however, a definitive mechanism of NO_x⁻ reduction inhibition by goethite cannot be completely substantiated by the data generated in these experiments. Since all cultures in our experiments were grown identically prior to resuspension in slurries containing or lacking goethite, levels of nitrate or nitrite reductase were initially induced equally in all reactors, regardless of whether goethite was present or absent.

δ¹⁵N₂O studies of interactions between microbial NO_x⁻ and goethite reduction.

We are not aware of any reported values for the isotopic fractionation between NO₂⁻ and N₂O for the chemical reduction of NO₂⁻ by Fe²⁺. However, it is possible to compare our values for biological NO₂⁻ reduction to N₂O with values reported in the literature. Because the actual δ¹⁵N-N₂O value is somewhat dependent on the δ¹⁵N of the starting NO_x⁻ species, these comparisons were made on the basis of the difference in δ¹⁵N between reactant and product (Δ¹⁵N = [δ¹⁵N-NO_x⁻ species] − [δ¹⁵N-N₂O]). The key difference between our microbial studies and previously reported Δ¹⁵N values for N₂O formation during denitrification and NH₄⁺ oxidation (7, 33, 34, 56, 58) is that we report a negative Δ¹⁵N value (enrichment in ¹⁵N with respect to the source), whereas previous studies reported positive Δ¹⁵N values (depletion in ¹⁵N with respect to the source). Considering these reports, it should be noted that the N₂O produced during denitrification is produced as an intermediate or final product, whereas the N₂O produced during nitrate reduction to ammonia is thought to be produced as a minor side reaction (6, 54). As these processes are quite different, it is not surprising that our Δ¹⁵N values display a trend not reported in the literature. The production of ¹⁵N-enriched N₂O during reduction of NO_x⁻ by S. putrefaciens 200 can easily be explained if (i) active biological NO₂⁻ reduction to ammonia by S. putrefaciens favors ¹⁴N over ¹⁵N and creates a pool of ¹⁵N-enriched NO₂⁻ and (ii) the N₂O-producing side reaction utilizes this remnant pool of ¹⁵N-enriched NO₂⁻ to produce N₂O that is depleted in ¹⁵N relative to the remnant pool but that is still enriched in ¹⁵N relative to the original NO₂⁻.

Conclusion

This work has demonstrated not only that NO₃⁻ inhibits microbial reduction of crystalline Fe³⁺ sources more strongly than that of aqueous Fe³⁺ sources but also that the presence of crystalline Fe³⁺ (e.g., goethite) inhibits microbial NO₃⁻ and NO₂⁻ reduction. Increased N₂O production in the presence of goethite and stable isotopic evidence both indicate that abiotic NO₂⁻ reduction to N₂O coupled to Fe²⁺ oxidation is occurring concomitantly with microbial NO_x⁻ and goethite reduction.

The implications of this process can be dramatic in a number of ways. Most relevantly, it provides for a series of geochemical reactions that can potentially inhibit a microbiological process by nucleating minerals on an active cellular surface. While novel, the idea that the formation of minerals on a cell membrane can inhibit intracellular processes is not without precedent. Both gram-negative and gram-positive bacterial cell membranes are documented to be ideal sites for mineral nucleation (8, 9, 15, 17, 18, 25, 48), and previous researchers have demonstrated that the production of Fe²⁺ and ferrous minerals may be responsible for the cell passivation commonly observed during batch mode iron reduction experiments (29, 43). Further studies of the metabolic and environmental implications of abiotic Fe²⁺ oxidation occurring concomitantly with microbial goethite and NO_x⁻ reduction are needed.