Imaging of Cellular Oxidoreductase Activity Suggests Mixotrophic Metabolisms in Thiomargarita spp.

ABSTRACT The largest known bacteria, Thiomargarita spp., have yet to be isolated in pure culture, but their large size allows for individual cells to be monitored in time course experiments or to be individually sorted for omics-based investigations. Here we investigated the metabolism of individual cells of Thiomargarita spp. by using a novel application of a tetrazolium-based dye that measures oxidoreductase activity. When coupled with microscopy, staining of the cells with a tetrazolium-formazan dye allows metabolic responses in Thiomargarita spp. to be to be tracked in the absence of observable cell division. Additionally, the metabolic activity of Thiomargarita sp. cells can be differentiated from the metabolism of other microbes in specimens that contain adherent bacteria. The results of our redox dye-based assay suggest that Thiomargarita is the most metabolically versatile under anoxic conditions, where it appears to express cellular oxidoreductase activity in response to the electron donors succinate, acetate, citrate, formate, thiosulfate, H2, and H2S. Under hypoxic conditions, formazan staining results suggest the metabolism of succinate and likely acetate, citrate, and H2S. Cells incubated under oxic conditions showed the weakest formazan staining response, and then only to H2S, citrate, and perhaps succinate. These results provide experimental validation of recent genomic studies of Candidatus Thiomargarita nelsonii that suggest metabolic plasticity and mixotrophic metabolism. The cellular oxidoreductase response of bacteria attached to the exterior of Thiomargarita also supports the possibility of trophic interactions between these largest of known bacteria and attached epibionts.

IMPORTANCE The metabolic potential of many microorganisms that cannot be grown in the laboratory is known only from genomic data. Genomes of Thiomargarita spp. suggest that these largest of known bacteria are mixotrophs, combining lithotrophic metabolism with organic carbon degradation. Our use of a redoxsensitive tetrazolium dye to query the metabolism of these bacteria provides an independent line of evidence that corroborates the apparent metabolic plasticity of Thiomargarita observed in recently produced genomes. Finding new cultivationindependent means of testing genomic results is critical to testing genome-derived hypotheses on the metabolic potentials of uncultivated microorganisms. KEYWORDS Beggiatoa, Thiomargarita, chemolithotrophy S ulfide-oxidizing bacteria of the family Beggiatoaceae contain some of the largest known bacteria (1)(2)(3)(4)(5), with individual cells of the genus Thiomargarita spp. reaching millimetric diameters (3,5,6). Dense communities of these organisms on the seafloor make up some of the most spatially extensive microbial mat ecosystems on earth (4,7). The large sulfur bacteria are primarily chemolithotrophs or mixotrophs that live at interfaces between nitrate or oxygen and hydrogen sulfide (8)(9)(10)(11)(12). These bacteria can have a substantial influence on the biogeochemical cycling of sulfur, nitrogen, phosphorus, and carbon in diverse environmental settings (11,(13)(14)(15). Despite their large size and biogeochemical significance, the physiologies and ecologies of these bacteria remain incompletely understood, in part because the large sulfur bacteria are not presently isolated in pure culture. At present, Thiomargarita spp. can only be maintained long-term in their natural sediments, not in axenic isolation or even as mixed communities in a defined microbial medium. Therefore, approaches to studying Thiomargarita's physiology have focused on culture-independent methods. Schulz and de Beer used microsensors to measure O 2 and sulfide gradients influenced by Thiomargarita metabolism (16), and more recently, (meta)genomic and metatranscriptomic approaches have been used to investigate their genetic potential (17,18) and response to environmental perturbations (19). Here we add to these findings by using a novel approach to the study of large sulfur bacteria that employs tetrazolium redox dyes.
Tetrazolium dyes are widely used to measure the metabolic activity of cells, including both bacteria (20)(21)(22) and eukaryotes (23), growing on a variety of metabolic substrates. In their oxidized form, soluble tetrazolium salts generally result in a colorless or weakly colored solution. Cellular oxidoreductases, such as NADH dehydrogenase, reduce tetrazolium salts to an intensely colored formazan product that can be observed qualitatively or measured quantitatively in a variety of colorimetric metabolic assays (24)(25)(26). These dyes can be used to measure catabolic metabolic activity even in the absence of observable cell division. Typically, tetrazolium dyes are applied to bacteria in culture and measured via the color response of the bulk medium and culture in a microplate (27). We sought to apply a tetrazolium dye approach to investigate metabolism in Thiomargarita spp. However, Thiomargarita spp. are not in pure culture, and these large cells are covered with communities of smaller attached bacteria (12). Our initial attempts to analyze our custom tetrazolium microplate assays via spectrophotometry failed to differentiate between the metabolism of Thiomargarita and that of the bacteria attached to it. We then employed a microscopy-based approach to image the tetrazolium color change associated with individual Thiomargarita cells so as to differentiate metabolic responses of Thiomargarita cells from those of attached epibiont cells. Here we report on the cellular oxidoreductase response of Thiomargarita cells to several organic and inorganic substrates under oxic, hypoxic, and anoxic conditions.

RESULTS AND DISCUSSION
Tetrazolium dye staining and the metabolism of lithotrophic substrates. Thiomargarita sp. cells exhibited apparent metabolic responses to a variety of substrates, as indicated by formazan staining that was localized within the cell (Fig. 1) and that increased in intensity over the duration of the 7-day experiment ( Fig. 1A and B, 2, and 3). Maximum staining generally occurred within 4 to 6 days ( Fig. 2 and 3).
Intense color changes were restricted to the Thiomargarita cell itself and did not extend to the sheath (Fig. 1B, C, and J), except in cases where epibiont bacteria were stained as described below or in the case of sulfide treatments, also described below. Thiomargarita cells sometimes underwent spontaneous collapse, as is observed to occur in unamended freshly collected natural Thiomargarita populations. Collapsed or otherwise damaged cells at the time of incubation never showed a color change (Fig. 1D), though some cells that did stain early in the experiment were sometimes observed to collapse in the later stages of the experiment. These observations suggest that formazan staining is a good indicator of initial cell viability, but that oxidoreductase activity alone was not sufficient to prevent eventual cell collapse in some specimens. Staining of controls, which included cell rinse water (not shown), diatom frustule (Fig. 1E), and empty sheaths (Fig. 1F), exhibited very low intensities compared with Thiomargarita cell staining (Table 1; Fig. 2 and 3).
Sulfur bacteria such as Thiomargarita are known for their ability to oxidize H 2 S by using O 2 or nitrate as an electron donor (3,12). The intense tetrazolium dye staining of cells exposed to H 2 S under both oxic and anoxic conditions was highly statistically supported as being distinct from control staining (P Ͻ 0.005, Student t test), while staining under sulfidic hypoxic conditions was marginally statistically distinct from control staining (P Ͻ 0.05, Student t test) ( Table 1). The genomes of Thiomargarita spp. Staining of metabolically active cells was spatially separate and distinct from that of the surrounding sheath material (C). Collapsed or damaged cells at the time of incubation did not exhibit a color change (D), showing exposure to acetate and hydrogen, respectively, under anoxic conditions. Staining of metabolically active cells was distinct in intensity from that of control diatom frustules (E) and control sheath material (F). Upon exposure to H 2 S, the extracellular medium assumed an orange hue that was readily distinguishable from the purple color change in the Thiomargarita sp. cells (G) and in biofilms of attached epibiont bacteria (arrows). Under anoxic conditions, cells and medium showed the most extensive response, with very little color change observed with no additional electron donor added (H) and a strong staining response with the addition of other substrates such as H 2 under anoxic conditions (I and J). In some cases, staining of extracellular bacteria was present as a diffuse stained cloud composed of small cells within the well. A zone characterized by absence of staining and cells in the immediate vicinity of the Thiomargarita cells suggests some sort of inhibition of these small bacteria (arrows in panel K). In other cases, such as exposure to thiosulfate under anoxic conditions, as shown here, stained filamentous epibionts could be observed anchored to the Thiomargarita cell/sheath (L). All images were taken at ϫ400 magnification. Scale bar in panel A,~100 m.
contain genes for the oxidation of H 2 S via a sulfide:quinone oxidoreductase and/or flavocytochrome c (17). In the treatments containing H 2 S, an orange color change was noted in the medium (Fig. 1G). The orange color change, which occurred only in the presence of H 2 S, was distinguishable from the typical formazan purple color change that occurs within the cells (Fig. 1G). This orange coloration is likely the result of the abiotic reduction of tetrazolium by H 2 S. Despite the sulfide treatments having a higher background than other treatments, a dark purple color change in the cell could be differentiated from the orange background ( Fig. 1G and 2C; Table 1).
Many sulfur-oxidizing bacteria are also known to be able to oxidize other sulfurcontaining substrates such as thiosulfate. Thiosulfate addition to the experimental wells resulted in an increasingly strong formazan staining response under anoxic conditions (P Ͻ 0.005, Student t test) (Fig. 3D). However, significant staining was not observed with thiosulfate addition under oxic and hypoxic conditions (Table 1). Instead, cells were observed to collapse (Fig. 2B). All currently available genomes of Beggiatoaceae contain , and other substrates under oxygenated conditions relative to that of controls. However, H 2 S (C) and citrate (not shown) did induce a statistically significant staining response under oxic conditions. Potentially significant staining was observed under hypoxic conditions in the presence of acetate (D), succinate, citrate, and H 2 S. Plotted here is the mean change in reciprocal intensity of luminance relative to that at day zero. Error bars indicate the standard error of the mean of the Thiomargarita incubations or controls. a.u., arbitrary units. the genes for thiosulfate oxidation via a partial sox system (soxABXY) for the oxidation of thiosulfate to intracellularly stored sulfur granules (17,18). Thiosulfate is not as strong a reductant as H 2 S, which may have left Thiomargarita cells susceptible to oxidative stress under oxic and hypoxic conditions. Indeed, cell collapse was common with a variety of metabolic substrates under oxic conditions. Oxic treatments with H 2 S were the only exception.
In addition to reduced sulfur compounds, some sulfur-oxidizing bacteria are also known to use H 2 as an electron donor for lithotrophic metabolism (28). Our staining results also showed strong statistical support (P Ͻ 0.005, Student t test) for formazan staining relative to controls when the cells were incubated in an anaerobic glove box containing 3% H 2 (Fig. 1D, I, and J and 3B). Cells incubated without other supplied electron donors in an anoxic chamber without H 2 showed no color change ( Fig. 1H and  3A). Recently, a chemolithotrophic strain of Beggiatoa sp., 35Flor, was found to use H 2 as an electron donor under oxygenated conditions (29). 35Flor was not observed to oxidize H 2 under anaerobic conditions, but it does not store nitrate to serve as an . With the addition of H 2 (B), succinate (C), thiosulfate (D), H 2 S, formate, acetate, citrate, and formate (not shown), a statistically significant staining response was observed under anoxic conditions. Plotted here is the mean change in the reciprocal intensity of luminance relative to that at day zero. Error bars indicate the standard error of the mean of the Thiomargarita incubations or controls. a.u., arbitrary units. electron acceptor as Thiomargarita does. In addition to the microsensor studies that show H 2 consumption by Beggiatoa sp. strain 35Flor, the genomes of Candidatus Thiomargarita nelsonii Thio36 and Bud S10 both contain genes for Ni-Fe hydrogenase (17,18), as do those of other sulfur-oxidizing bacteria (30,31). The source of H 2 for Thiomargarita in nature is not clear, but fermentation by epibionts or other bacteria in the environment is a possible source.
Organic acid metabolism. In addition to their canonical lithotrophic metabolism, some representatives of the family Beggiatoaceae are known to metabolize organic acids (32). Organic acid degradation may be a common trait among sulfur-oxidizing gammaproteobacteria, since the generation of NADPH through the tricarboxylic acid (TCA) cycle would coincide with the assimilation of organic acids and would thus reduce the energetic costs of carbon fixation (33,34). The genomes of Ca. Thiomargarita nelsonii Bud S10 and Thio36 and other members of the family Beggiatoaceae contain genes for a complete TCA cycle, including the gene for NADH dehydrogenase I (17,18,35,36), and while succinate dehydrogenase is not electrochemically favored under most anaerobic conditions, it is tightly coupled with respiratory nitrate and nitrite reduction (37). Nitrate was provided in the medium (100 M), and nitrate is also stored in the internal nitrate vacuole.
All of the organic acids tested here yielded positive staining results under anoxic conditions, but staining results with organic acid exposure were more variable under hypoxic and aerobic conditions. We observed a positive staining response with the addition of citrate under oxic, hypoxic, and anoxic conditions and succinate under hypoxic and anoxic conditions ( Table 1). The utilization of exogenous organic acids requires specific inner membrane transporters. The genomes of a few members of the family Beggiatoaceae encode putative citrate transporters (Ga0060138_111857, BGP_2612), and a partial putative citrate transporter is present in Ca. Thiomargarita nelsonii Thio36 (Thi036DRAFT_00068870). Succinate is usually transported by proteins encoded by either one of two three-gene clusters, dctABD, where dctA is a permease and dctBD is a two-component system for the activation of transcription (38), or kgtPSR (39), or via a tripartite ATP-independent periplasmic (TRAP) transporter, dctPQM (40). The two available Thiomargarita genomes contain a dctB gene but lack dctAD, although dctB is located near the terminal end of a contig in the genome of Bud S10. While freshwater strains of Beggiatoa (Beggiatoa leptomitoformis D-402 and B. alba B18LD) possess putative dctPQM genes, annotated genes for C 4 -dicarboxylate transporters are lacking in all marine strains. Thus, the capacity to transport succinate via known transporters appears to be lacking in marine members of the family Beggiatoaceae. However, the marine strains do possess a number of uncharacterized TRAP transporters that might serve as future targets for the characterization of C 4 -dicarboxylate metabolism. Acetate is another organic acid that is thought to be metabolized by certain sulfur bacteria. Acetate can be metabolized by the TCA and glyoxylate cycles, as well as putative alternate pathways that make carboxylic acids as intermediates for biosynthesis, but that may not yield energy to the cell (41). The microelectrode respiration experiments of Schulz and de Beer showed the stabilization of oxygen-sulfide gradients around Thiomargarita when acetate was added to the experiments, although acetate itself was not found to steepen the oxygen gradient (16). Our formazan dye staining results show no statistical support for acetate use under oxic conditions, which is consistent with the results of Schulz and de Beer. However, our results also show marginal statistical support for acetate metabolism under hypoxic conditions and strong support for acetate use under anoxic conditions (Table 1; Fig. 2D). Acetate metabolism is preceded by the activation of acetate by acetate kinase and phosphotransacetylase (42) (known as the ack-pta pathway) and/or via acetyl coenzyme A synthetase (known as the ACS pathway). The ack-pta pathway occurs in the freshwater strains of members of the family Beggiatoaceae but not in the marine strains. However, almost all members of the family Beggiatoaceae possess the ACS pathway (18) and an acetate/cation symporter (e.g., Ga0063879_05139) (36). These genomic features are consistent with the ability of Thiomargarita spp. to take up and metabolize acetate from the environment, but our staining results suggest that this may occur only under anoxic and perhaps hypoxic conditions in Thiomargarita spp. The activity of cellular oxidoreductases that reduce tetrazolium to formazan under acetate exposure, as we observed here under hypoxic and anoxic conditions, suggests that acetate is metabolized by a pathway such as the TCA cycle that contributes to energy production, as well as biosynthetic intermediates. In future studies, isotopically labeled organic substrates and their assimilation into Thiomargarita could be used to determine the relative contributions of those substrates to biosynthesis versus energy production.
Formate is both a waste product of fermentation and a potential electron donor that can be used for both aerobic and anaerobic respiration (43). The genomes of Ca. Thiomargarita nelsonii Thio36 and Beggiatoa strain PS both contain genes that code for subunits of formate dehydrogenase. In addition to anaerobic respiration, the energyyielding reaction under anaerobic conditions is the formate-assisted cleavage of pyruvate via a pyruvate formate lyase enzyme. The genomes of Ca. Thiomargarita nelsonii Thio36 and Bud S10 both contain three annotated pyruvate formate lyase genes, and these genes occur in most other representatives of the family Beggiatoaceae. The transport of exogenous formate could be mediated by a formate/nitrate transporter (FocA), which is present in most marine strains of Beggiatoaceae, including both Ca. Thiomargarita nelsonii genomes.
Although Thiomargarita spp. are thought to be more oxygen tolerant than other marine vacuolate sulfur bacteria, such as Ca. Maribeggiatoa and Ca. Marithioploca (16), the results presented here suggest that while Thiomargarita spp. may tolerate oxygen exposure, their metabolism(s) is most versatile under anoxic conditions. This is perhaps unsurprising, given that Thiomargarita bacteria on the Namibian shelf are found primarily in sediments that are anoxic for much of the year (44). During the collection of the samples used in our experiments, no oxygen was detectable in the lower water column, as measured by Winkler titration, and in core-top waters, as measured with a handheld oxygen meter. Limited tolerance to O 2 is also suggested by recent genomic results for most members of the family Beggiatoaceae. All Beggiatoaceae strains have cbb 3 -type cytochrome c oxidase, and some also possess bd-type cytochromes, both of which are specific to hypoxic conditions. However, both Ca. Thiomargarita nelsonii Thio36 and Beggiatoa strain PS possess the more oxygen-tolerant cytochrome c oxidase (coxABC, cyoE), which suggests greater O 2 tolerance in some strains of Beggiatoaceae. Catalase, which is used to ameliorate oxidative stress, is not present in the genomes of marine Beggiatoaceae and occurs only in the freshwater strains. Most Beggiatoaceae genomes contain a superoxide dismutase and a cytochrome c peroxidase, while most marine strains also possess a desulfoferredoxin. The enzymes coded for by these genes may provide protection against O 2 and reactive oxygen species during periodic exposure to O 2 . Additionally, H 2 S, which can scavenge reactive oxygen species (45), may serve as an extracellular antioxidant under conditions in which sulfide is fluxing into oxygenated waters. Under anoxic conditions, oxidized forms of inorganic nitrogen, whether exogenous or stored within the vacuole, serve as terminal electron acceptors (3,46). Ca. Thiomargarita nelsonii Bud S10 (17) and Thio36 (18) both possess a complete denitrification pathway that includes both membrane-bound (nar) and periplasmic (nap) nitrate oxidoreductases and the capacity to reduce nitrite to ammonium (nirBD). Thus, our staining results are consistent with the genomes and the canonical knowledge of nitrate being used as an electron acceptor by certain members of the family Beggiatoaceae.
Epibiont microbial cells. In our experiments, we undertook imaging of individual formazan-stained Thiomargarita cells in order to differentiate Thiomargarita metabolism from that of attached bacteria. These attached bacteria exhibited staining spatially distinct from that of Thiomargarita under the microscope that could not be differentiated with a typical microplate assay. Our microscope-based imaging approach had the added benefit of allowing us to observe the discrete staining of epibiont biofilms and filaments associated with Thiomargarita ( Fig. 1G and L). In some cases, stained filamentous epibionts could be observed anchored to the Thiomargarita cell/sheath (e.g., Fig. 1L, arrows). In particular, dense accumulations of filamentous bacteria (Fig. 1L) exhibited intense staining when exposed to H 2 under anoxic conditions. While filamentous bacteria were observed in some control wells and in zones of the well distal to Thiomargarita, their accumulation was far denser in the vicinity of Thiomargarita cells (Fig. 1L).
We also observed under anoxic conditions in the presence of H 2 S that small bacteria that stained purple were present throughout the medium, except in clear cell-free zones we observed immediately surrounding Thiomargarita cells and chains (Fig. 1K,  arrows). Additional studies are needed to identify these epibiont bacteria and determine the nature of their trophic interactions (if any) with Thiomargarita spp.
Conclusions. Our observations of oxidoreductase-mediated formazan staining in Thiomargarita cells exposed to a variety of organic and inorganic substrates are consistent with recent genomic results that suggest metabolic plasticity in Ca. Thiomargarita nelsonii. There are, however, some limitations and caveats in drawing broad conclusions from our results. It is possible that the incubation medium formulations or concentrations we used here, or other aspects of the experimental design, resulted in responses from Thiomargarita bacteria that do not represent their metabolic activities in nature. Yet the results presented here are broadly consistent with genomic data currently available for Ca. Thiomargarita nelsonii and consistent with the sedimenthosted habitat of these bacteria that is anoxic for much of the year. We used both the spherical Thiomargarita cells that are typical of Thiomargarita namibiensis and the cylindrical forms that are more typical of Ca. Thiomargarita nelsonii for this study (1). However, the large number of cells used in the experiment, and the difficulty in amplifying the Thiomargarita 16S rRNA gene because of the multiple introns contained therein, prevented a detailed phylogenetic characterization of the cells used for the assay (47). Therefore, we cannot say that our results apply broadly to the multiple candidate Thiomargarita species that can sometimes co-occur in sediments off Namibia (1,48). While cultivation will ultimately be necessary for rigorous testing of the physiologies of Thiomargarita spp., for now, these results provide additional evidence in support of genomic and microsensor findings of metabolic versatility and/or mixotrophy in the genus Thiomargarita and suggest cultivation approaches that include both reduced sulfur compounds and organic substrates perhaps under extremely low oxygen conditions. Tetrazolium/formazan redox dyes have long been used to study bacteria, primarily in plate-based assays. Our expansion of the use of these dyes to the microscopic examination of individual stained cells may be broadly useful for assaying metabolism in mixed communities and other uncultivated microorganisms and for validation of genome-derived assessments of physiological capabilities.

MATERIALS AND METHODS
Sample collection. Thiomargarita sp. cells were collected from organic-rich sediments on the Atlantic shelf near Walvis Bay, Namibia (23°00.009=N 14°04.117=E) with a multicorer on board the R/V Mirabilis. All of the cells used in the experiments described here were collected at a depth of 1 to 3 cm beneath the sediment-water interface. Thiomargarita sp. cells were stored in their host sediments with overlying core top water in closed 50-ml centrifuge tubes at 4°C and protected from direct light for 3 weeks.
Incubation experiments with tetrazolium staining. Chains and clusters of Thiomargarita sp. cells were rinsed three times in 0.2-m-filtered artificial seawater before being added to an incubation medium in 96-well microplates. The purpose of these rinsing steps was to remove loosely attached smaller bacteria from the exterior of Thiomargarita cells and sheaths. These washing steps did not remove tightly attached or sheath-embedded epibionts, as confirmed by microscopy and described above. One Thiomargarita cell, cell cluster, or cell chain was added to each individual well of the 96-well plate. Each substrate treatment contained eight wells with Thiomargarita cells, as well as four control wells without Thiomargarita cells. Two of these control wells contained 20 l of the first saltwater bath used to rinse the Thiomargarita cells, added to 180 l of medium, as a control that contains cells found loosely attached to Thiomargarita cells. Wells containing empty diatom frustules picked by pipette from the same samples as the Thiomargarita cells were also used as negative controls to monitor the color change of xenic biological surfaces from the same environment. A third control type consisted of empty mucus sheaths that were produced by Thiomargarita cells but no longer contained the cells. These empty sheaths are common in Thiomargarita-rich sediments and can be readily identified by the shape of the chains that is preserved as void space in the sheath material.
The incubation medium was designed to maintain metabolizing Thiomargarita cells and to provide basic elemental constituents and nutrients based on similar base medium recipes for phylogenetically related taxa (34). The incubation medium included the following constituents (liter Ϫ1 ): 34 g of NaCl, 0.112 g of CaCl 2 , 0.008 g of NH 4 NO 3 , 0.5 g of KCl, 1.46 g of MgSO 4 , 20 ml of 1 M MOPS propanesulfonic acid] buffer (pH adjusted to 7.8, final concentration, 20 mM), 1 ml of 1,000ϫ potassium phosphate buffer (pH 7.6), 1 ml of 1,000ϫ vitamin B 12 , 1 ml of a 1,000ϫ vitamin solution, and 10 ml of a 100ϫ trace element solution. The vitamin solution contained (liter Ϫ1 ) 10 mg of riboflavin and 100 mg each of thiamine HCl, thiamine pyrophosphate, L-ascorbic acid, D-Ca-pantothenate, folic acid, biotin, lipoic acid, nicotinic acid, 4-aminobenzoic acid, pyridoxine HCl, thiotic acid, NAD, and inositol dissolved in 100 ml of 10 mM KPO 4 buffer (pH 7). The trace metal solution contained (liter Ϫ1 ) 0.1 g of FeCl 2 · 2H 2 O, 0.03 g of H 3 BO 3 , 0.1 g of MnCl 2 , 0.1 g of CoCl 2 · 6H 2 O, 1.5 g of nitrilotriacetic acid, 0.002 g of CuCl 2 · 6H 2 O, 0.024 g of NiCl 2 · 6H 2 O, 0.144 g of ZnSO 4 · 7H 2 O, 0.036 g of NaMoO 4 , 0.025 g of Na-vanadate, 0.010 g of NaSeO 3 , and 0.01 g of NaWO 4 · 2H 2 O. NaHCO 3 was added to each base medium to a final concentration of 3 mM for the oxic and anoxic stocks or 40 mM for the hypoxic treatments. The medium was sterilized by filtration through 0.22-m membrane, and the final pH was 7.9 to 8.0.
Potential electron donors acted as experimental variables and included H 2 , H 2 S, thiosulfate, succinate, acetate, citrate, and formate. All electron donors except H 2 and H 2 S were supplied at a final concentration of 1 mM. H 2 was supplied by shaking the plates in a Coy anaerobic chamber containing a 3% H 2 , 97% N 2 atmosphere. H 2 S was supplied by the daily addition of 10 l of the freshly neutralized 4 mM sodium sulfide delivered by syringe. The tetrazolium redox dye mixture we used, which is known by the commercial name Dye H (Biolog catalog no. 74228), was added to the medium at a final 1ϫ working strength just prior to cell incubation.
Microplates were placed on orbital shakers at 50 rpm. One microplate was maintained under benchtop atmospheric conditions, one microplate was placed in a Coy hypoxic chamber with 5% atmospheric level O 2 and~5% total CO 2 , one microplate was placed in an H 2 -free anaerobic chamber (NextGen; Vacuum Atmospheres Company, Hawthorne, CA), and one microplate was placed in a Coy anaerobic chamber containing a mixture of 97% N 2 and 3% H 2 . All plates were maintained in plastic containers with an open aperture to allow free exchange of gases for the week-long duration of the experiment. The plastic chambers that housed the microplates contained moistened paper towels to inhibit plate evaporation. A Unisense oxygen microsensor was used to confirm that O 2 was present, and wells were well mixed at depth in oxic treatments, including those that contained H 2 S. Image processing. Cells were imaged with an Olympus IX-81 inverted microscope equipped with a long working distance 40ϫ objective (numerical aperture, 0.6; working distance, 2.7 to 4.0 mm) and a 17.28-megapixel DP73 color camera. Images were collected with CellSens dimension (Olympus, Japan) software under constant (manually set) exposure and white balance settings. ImageJ was used to subtract the background and convert the image to an XYZ color profile. A 40-by 40-pixel region of interest representing an area of cytoplasm with a low density of sulfur globules was chosen for quantification of dye change in each cell. The average pixel intensity of this region was then measured in the luminance channel (Y). Reciprocal intensity was calculated as described in reference 49 for quantification of chromogen intensity. The change in intensity relative to that at time zero was reported, and a Student's t test was used to determine the likelihood that the imaged intensities were distinct from the mean of the controls by chance alone ( Table 1). Copying of the selection region in ImageJ from image to image in the time series for an individual cell ensured that the same area of the cell was measured at each time point. Uniform adjustment of color levels was used for presentation of images in Fig. 1, but this adjustment was performed after image analysis.