Thiosulfate oxidation by Thiomicrospira thermophila: metabolic flexibility in response to ambient geochemistry

Summary

Previous studies of the stoichiometry of thiosulfate oxidation by colorless sulfur bacteria have failed to demonstrate mass balance of sulfur, indicating that unidentified oxidized products must be present. Here we present reaction stoichiometry and kinetics under variable pH conditions during the growth of Thiomicrospira thermophila strain EPR85, isolated from diffuse hydrothermal fluids at the East Pacific Rise. At pH 8.0, thiosulfate is stoichiometrically converted to sulfate. At lower pH, the products of thiosulfate oxidation are extracellular elemental sulfur and sulfate. We were able to replicate previous experiments and identify the missing sulfur as tetrathionate, consistent with previous reports of the activity of thiosulfate dehydrogenase. Tetrathionate was formed under slightly acidic conditions. Genomic DNA from T. thermophila strain EPR85 contains genes homologous to those in the Sox pathway (soxAXYZBCDL), as well as rhodanese and thiosulfate dehydrogenase. No other sulfur oxidizing bacteria containing sox(CD)₂ genes have been reported to produce extracellular elemental sulfur. If the apparent modified Sox pathway we observe in T. thermophila is present in marine Thiobacillus and Thiomicrospira species, production of extracellular elemental sulfur may be biogeochemically important in marine sulfur cycling.

Introduction

Sulfide oxidation is a critical step in the biogeochemical sulfur cycle and occurs via several complex reaction pathways, some of which are biologically mediated (Jorgensen and Nelson, 2004; Jorgensen, 1990; Jorgensen and Bak, 1991). The stepwise oxidation of sulfide to sulfate can occur via several reactive chemical species of intermediate oxidation state. These intermediates, including elemental sulfur, thiosulfate, and polythionates, can be biogeochemically important even when their steady-state concentrations are low (sub-µM) due to rapid, hidden (or ‘cryptic’) recycling (Thamdrup et al., 1994; Canfield et al., 2010). Elemental sulfur and thiosulfate can be oxidized, reduced, or disproportionated, resulting in multiple pathways and redox intermediates of the sulfur cycle (Troelsen and Jorgensen, 1982; Jorgensen and Nelson, 2004). However, pathways involving intermediate oxidation states of sulfur have been difficult to quantify, with many complex processes taking place simultaneously in the presence of several available oxidants and reductants (Flynn et al., 2014; Jorgensen, 1990; Jorgensen and Bak, 1991).

Two known enzymatic pathways mediate thiosulfate oxidation: i) the S₄ Intermediate pathway, which involves production and consumption of polythionates (Kelly et al., 1997), and ii) the multi-enzyme Sox pathway (Sox = sulfide oxidation) (Friedrich et al., 2001). The S₄ Intermediate pathway appears to be restricted to obligate chemolithotrophic strains within the genus Thiobacillus and involves the production of tetrathionate from thiosulfate via tetrathionate synthase (Lu and Kelly, 1988; Meulenberg et al., 1992). This process is likely coupled to enzymes that subsequently split tetrathionate to thiosulfate, sulfate, or elemental sulfur (Lu and Kelly, 1988; Pronk et al., 1990; Meulenberg et al., 1992).

The Sox pathway is versatile and has been shown to be active in physiologically diverse organisms, including green sulfur bacteria, purple sulfur bacteria, and colorless sulfur bacteria (Friedrich et al., 2001; Welte et al., 2009; Sauve et al., 2007). Complete thiosulfate oxidation via the Sox pathway results in sulfate as the sole product; however, incomplete pathways in which Sox(CD)₂ is absent can result in accumulation of elemental sulfur, either intracellularly (Grimm et al., 2008) or extracellularly (Gregersen et al., 2011; Van Gemerden, 1984). The colorless sulfur bacterium Thiomicrospira crunogena is known to use the Sox pathway to perform thiosulfate oxidation to sulfate and elemental sulfur (Scott et al., 2006; Sievert et al., 2008), but also produces tetrathionate under acidic conditions (Javor et al., 1990). The environmental conditions under which these various oxidation states are produced and/or consumed are poorly constrained for Thiomicrospira thermophila.

Here, we report new genomic and physiological evidence for a modified Sox pathway functioning in Thiomicrospira thermophila; a sulfur-oxidizing gammaproteobacterium found in marine settings and identified as an obligate aerobic chemolithomixotroph that can oxidize sulfide, thiosulfate and elemental sulfur (Kuenen and Tuovinen, 1981; Takai et al., 2004). Thiobacillus and Thiomicrospira spp. are widespread in marine sediments (Robertson and Kuenen, 1992; Kuenen et al., 1985; Brinkhoff and Muyzer, 1997) and in marine geothermal environments (Sievert et al., 2000; Sievert et al., 2008), particularly in diffuse flow hydrothermal vents (Brazelton and Baross, 2010; Brinkhoff and Muyzer, 1997; Huber et al., 2003). We investigated thiosulfate oxidation by T. thermophila strain EPR85, isolated from a seafloor hydrothermal diffuse flow vent on the East Pacific Rise at 9° N. T. thermophila is closely related to the well-studied strain T. crunogena XCL-2, which employs a pH-dependent Sox pathway for thiosulfate oxidation (Scott et al., 2006; Sievert et al., 2008) that has been shown to produce (at pH 6.3) elemental sulfur as well as consume it (at pH 8.5) (Javor et al., 1990). However, previous examination of T. crunogena could not account for all of the sulfur in the system, suggesting the production of undetermined sulfur products under all pH conditions studied (6.3 – 8.5). Problems with sulfur mass balance were also present in studies of two closely related strains of Thiobacillus that oxidize sulfide to elemental sulfur and sulfate (Stefess et al., 1996). In contrast, studies of T. thermophila have indicated that thiosulfate oxidation produced equal concentrations of elemental sulfur and sulfate, accounting for all sulfur in the system (Takai et al., 2004). These inconsistent results suggest that thiosulfate oxidation by the Sox pathway may be more complex than previously thought and may result in the production of as-yet unidentified sulfur products.

In this study we have identified the sulfur oxidation reactions mediated by T. thermophila in the context of specific sulfur cycling enzymes found in its genome. Under various culture conditions we replicated both the sulfur balance and imbalances in the studies discussed above, and identified the missing sulfur product as tetrathionate. We report data on the stoichiometry and kinetics of these reactions (Table 1) and provide a genetic context for the mechanisms potentially responsible for thiosulfate oxidation to multiple products in T. thermophila EPR85.

Table 1.

Chemical reactions observed in thiosulfate oxidation by T. thermophila. Inferred enzymes involved in thiosulfate and elemental sulfur oxidation for this study include rhodanese (rhd•S), SoxAX, SoxYZ, SoxB, Sox(CD)₂, and thiosulfate dehydrogenase (tsd) as indicated over each reaction. Free energy for the net chemical reaction is given in kJ/mole of reaction, with negative values indicating spontaneous reaction in the direction written.

	Net chemical reactions	Reactions	ΔG (kJ/mole rxn)
(ia)	S2O32− → SO32− + E·S		36.6
(ib)	S2O32− + 1/2O2 → S + SO42−		−229.6
(ic)	Sx2− → Sx-1 + S2−
(ii)	S + H2O + 3/2O2 → SO42− + 2H+		−529.9
(iii)	SO32− + 1/2O2 → SO42−		−266.2
(iv)	2S2O32− + 2H+ + 1/2O2 → S4O62− + H2O		−242.5

Results

Stoichiometry and mass balance

When buffered at the high end of its pH tolerance (8.0), T. thermophila mediates complete oxidation of thiosulfate to sulfate with no significant buildup of elemental sulfur in the media (Figure 1a). Maximum accumulation of moderate amounts of free elemental sulfur (S₈) in the media (<150 µM) and Tris-extractable solid elemental (polymeric) sulfur (<250 µM) (Table S1) is concurrent with the period of fastest thiosulfate consumption. Dissolved oxygen does not become limiting during growth (Table S1).

Figure 1.

Time series data for (A) thiosulfate oxidation at high pH (8.0) buffered with Tris; (B) thiosulfate oxidation at low pH (5.5) buffered with Bis-Tris; and (C) thiosulfate oxidation in unbuffered media. Error bars based on triplicate experiments are shown for all analyses.

In contrast, growth at pH buffered at the low end of tolerance (pH ~5.5) results in accumulation of free elemental sulfur and sulfate in the media, consistent with incomplete thiosulfate oxidation to sulfate. Experimental data (Figure 1b) suggest that thiosulfate oxidation and elemental sulfur oxidation occur simultaneously, as described by the combination of the chemical reactions (ib) and (ii) from Table 1. Over the course of the whole 8 hour experiment, total sulfur, as measured by the sum of known reaction products ([2*[S₂O₃]+[SO₄]+[S]), decreased from ~23mM to 15mM (Figure 1b). Abiotic controls at low pH (3.5 – 5.6), in media buffered with both citrate buffer and Bis-Tris buffer, show no loss of thiosulfate and concentrations of possible products such as tetrathionate, sulfate, or elemental sulfur were negligible for up to 73 hours (Table S2).

Growth in unbuffered media results in a complex combination of reactions that depends on pH. As elemental sulfur accumulates in the media, the stoichiometry is consistent with a combination of incomplete thiosulfate oxidation to sulfate (reactions (ib and ii) from Table 1). During this initial phase, the rate of elemental sulfur oxidation (reaction (ii)) is less than elemental sulfur production (reaction (i)), causing an accumulation of elemental sulfur in the media and a decrease in pH from 7.5 to ~6. However, a pronounced change in reaction stoichiometry occurs with the onset of elemental sulfur consumption (t~8 hrs, Figure 1c), which corresponds to a threshold pH of 6. Here, total sulfur, as measured by the sum of known reaction products ([2*[S₂O₃]+[SO₄]+[S]), decreased from ~11mM to 6mM over 6 hours (Figure 1c), suggesting production of an additional as-of-yet unaccounted for sulfur product. On the contrary, abiotic controls in unbuffered seawater at pH 6.0 to 7.9 indicate no change in thiosulfate or sulfate concentrations over the course of 144 hours at both 35°C and 4°C (Table S2).

To address the issue of a missing sulfur product(s) in both buffered and unbuffered media, we looked for other sulfur phases that could plausibly be present, including Tris-extractable solid elemental (polymeric S) sulfur, sulfide, polysulfides (here referring to the charged dissolved species S_x²⁻), and polythionates. Only minimal concentrations (<300 µM) of Tris-extractable solid polymeric sulfur were present and polysulfides were not detected. Sulfide concentrations never exceeded 20 µM (Table S1), with the maximum values coinciding with a peak in elemental sulfur concentration. However, polythionates, primarily in the form of tetrathionate (S₄O₆²⁻), were detected by cyanolysis at concentrations up to 1 mM, with continuous production taking place after pH stopped decreasing (Figure 2a). The abundance of tetrathionate is sufficient to account for the apparent missing sulfur (5.5 mM total) in this experiment (Figure 2a) and tetrathionate concentrations never decreased under any experimental conditions examined. Thus, once pH drops below 5.7, metastable tetrathionate is present, elemental sulfur consumption begins, and competition for thiosulfate takes place between reactions (1) and (4). Similarly, production of tetrathionate occurs in experiments buffered at low pH (5.5) and accounts for the apparent missing sulfur (7.7mM total) in those experiments (Figure 2b).

Figure 2.

Stoichiometry of growth experiments including polythionate concentrations determined with the cyanolysis method. (A) Growth in unbuffered media follows reactions 1 and 2 until pH drops below 5.5, only below which tetrathionate production (reaction 4) begins. pH was adjusted back to 7.5 at 6 hours (arrow), resulting in a return to reactions 1 and 2. (B) Growth at low pH (5.5; Bis-Tris buffer) occurs with simultaneous production of elemental sulfur, sulfate, and tetrathionate via reactions 1, 2, and 4 (Table 1). Error bars based on triplicate experiments are shown for all analyses.

Physiology and growth rate

Specific growth rate of T. thermophila correlates with pH throughout our experiments. At pH 8.0, where thiosulfate is oxidized completely to sulfate, the maximal specific growth rate of 0.49 hr⁻¹ is attained (Figure 3). Specific growth rate decreases with pH, corresponding to a change in reaction stoichiometry (Figure 3). In unbuffered media, the progression of reaction 2 (Table 1) results in a decrease in pH due to proton production. In every unbuffered experiment, pH eventually reaches a constant value during the thiosulfate oxidation phase of growth, which is coincident with the proton-consuming onset of tetrathionate production (reaction iv, Table 1). This growth phase is characterized by the slowest rates observed (Figure 3, open symbols).

Figure 3.

Specific growth rate as a function of starting pH of the media. Start pH determines which thiosulfate-consuming reaction will be mediated. Highest growth rates are observed when thiosulfate is completely oxidized to sulfate in Tris buffer (pH 8.0) (solid triangle) and lowest when thiosulfate is oxidized to tetrathionate (open squares), thiosulfate is oxidized to equal parts elemental sulfur and sulfate (open circles) or elemental sulfur is oxidized in the absence of thiosulfate (cross). Thiosulfate oxidation to elemental sulfur and sulfate with simultaneous oxidation of elemental sulfur to sulfate supports intermediate growth rates (solid diamonds). Specific growth rates in Bis-Tris buffer (pH 5.5) are shown in blue; unbuffered media data are shown in black; citrate buffer data are shown in green.

Sequence analysis of T. thermophila draft genome

The genome (estimated length of 2.65 mbp) was partially assembled into 40 contigs, with the largest 20 accounting for 97.6% of the bases of the scaffold assembly. The G+C content of the assembly is 50.3%. For comparison, the T. crunogena XCL-2 genome has a total length of 2.43mbp with a G+C content of 43% (Scott et al., 2006). The lengths of the largest 19 contigs in T. thermophila are indicated in Figure S1 and are referred to here as nodes, numbered for reference. Sulfur oxidation genes of interest are annotated (Figure S1).

The genome of T. thermophila has homologues for soxAXZYBCDL, all of which are necessary for the Sox system to completely oxidize thiosulfate to sulfate. However, in contrast to the well-studied Paracoccus pantotrophus (Friedrich et al., 2001), which has all Sox genes clustered in the same region of the genome, T. thermophila has homologues of soxAXYZ in one cluster and soxB, soxL and sox(CD)₂ in three separate regions (Figure S1). T. crunogena shares this pattern of arrangement (Sievert et al., 2008) and the homologous Sox genes in T. thermophila are very closely related to those in T. crunogena XCL-2 (Table 2; Scott et al., 2006).

Table 2.

Homologous genes for sulfur oxidation enzymes found on the T. thermophila genome. Genbank accession number for sequence with highest similarity is shown with corresponding e-value/% similarity in brackets. Known function is included with references to protein families, if known.

Protein	Node#/p osition in node	Highest sequence similarity	Nearest neighbor	comments
SoxA	1 (241316–242146)	YP_390871.1 (1e-180/85.51%)	T. crunogena XCL-2	Sulfur oxidation protein with signal peptide and conserved cytochrome c
SoxZ	1 (242251–242589)	YP_390872.1 (1e-62/89.81%)	T. crunogena XCL-2	Sulfur oxidation protein (pfam08770)
SoxY	1 (242629–243096)	YP_390873.1 (6e-86/91.61%)	T. crunogena XCL-2	Sulfur oxidation protein with signal peptide
SoxX	1 (243134–243508)	YP_390874.1 (2e-81/93.55%)	T. crunogena XCL-2	Cytochrome c with signal peptide
Cytochrome c	1 (267914–268273)	YP_390898.1 (3e-57/86.00%)	T. crunogena XCL-2	Cytochrome c553 with signal peptide
Hypothetical protein (SoxA)	6 (119648–120526)	YP_003644808.1 (2e-128/66.41%)	Thiomonas intermedia	Sulfur oxidation with non-matching signal peptide (KEGG Adeh_2281; KEGG Anaek_1594)
Hypothetical protein (SoxX)	6 (120852–121649)	YP_003644809.1 (4e-38/61.50%)	Thiomonas intermedia K12	Sulfur oxidation protein with cytochrome c
rhodanese	6 (165983–166432)	WP_011371372.1 (3e-83/85.71%)	T. crunogena	sulfurtransferase
SoxB	5 (107062–108993)	YP_391815.1 (0.0/79.06%)	T. crunogena XCL-2	Sulfate thiol esterase UshA with signal peptide, SoxB (cd07411) with metallophosphatase binding site, C-terminal domain (pfam02872)
Sulfide dehydro-genase	5 (129755–131155)	WP_019591503.1 (3e-139/52.18%)	Thioalkalivibrio sp. ALE20	NAD(FAD) dependent dehydrogenase with NAD(P) binding domain (cl17500) and flavocytochrome c sulfide dehydrogenase (pfam204176)
SoxC	10 (68564–69925)	YP_390426.1 (0.0/94.48%)	T. crunogena XCL-2	Sulfite oxidase with molydopterin-binding domain (cl00199)
SoxD	10 (69906–70994)	YP_390427.1 (0.0/88.69%) WP_020286444.1 (7e-34/38.89%)	T. crunogena XCL-2 Osedax symbiont Rs2	Hypothetical protein containing cytochrome c (pfam00034) Diheme cytochrome c SoxD (pfam00034)
Sulfide dehydro-genase	9 (44242–45666)	WP_019896075.1 (0.0/79.64%) YP_001997639.1 (2e-128/47.10%)	T. halophila Chlorobaculum parvum NCIB 8327	Hypothetical protein with pyridine nucleotide disulfide oxidoreductase (pfam07992) Flavocytochrome c sulfide dehydrogenase (pfam09242)
Sulfite reductase	3 (111050–112147)	YP_392157.1 (0.0/77.81%)	T. crunogena XCL-2	Oxidoreductase containing FAD binding domain (pfam00667), cytochrome p450 – sulfite reductase (NADPH dependent reduction of SO3 to H2S)
Sulfite reductase	4 (51414–53828)	WP_004280698.1 (0.0/39.79%)	Acinetobacter lwoffii	Fe-regulated membrane protein PiuB containing CysJ sulfite reductase and flavoprotein with cytochrome p450 – sulfite reductase
Sulfide-quinone reductase	2 (33149–34273)	YP_391650.1 (0.0/85.29%)	T. crunogena XCL-2	Sulfide-quinone reductase containing pyridine nucleotide – disulfide oxidoreductase (pfam07992)
Thiosulfate dehydro-genase	16 (13271–14293)	WP_019556080.1 (9e-142/58.88%) WP_012969337.1 (8e-41/39.41%)	T. arctica Allochromatium vinosum DSM180	Hypothetical protein – haem binding cytochrome c (pfam00034) Thiosulfate dehydrogenase containing different signal peptide
rhodanese	13 (55570–55866)	YP_391186 (2e-63/95.88%)	T. crunogena	Rhodanese-like protein
rhodanese (SoxL)	19 (25750–26430)	YP_003262911 (6e-108/72.64%) YP_003444124 (7e-59/46.23%)	Halothiobacillus neapolitanus c2 Allochromatium vinosum DSM 180	Rhodanese (pfam PF00581) Sulfurtransferase (SoxL)

Also like T. crunogena, T. thermophila has two homologues of rhodanese, a thiosulfate sulfurtransferase specific to thiosulfate as a substrate that cleaves the sulfite moiety and transfers the outer sulfur to an acceptor substrate such as cyanide, lipoic acid, or thioredoxin (Smith and Lascelles, 1966; Nandi et al., 2000). Thus, rhodanese could be an alternative enzyme scavenging thiosulfate in T. thermophila. The catalytic site in rhodanese binds to both the inner and outer sulfur atoms in thiosulfate in an orientation that polarizes the S-S bond, allowing enzymatic cleavage of the S-S bond (Gliubich et al., 1996). A copy of rhodanese with a completely conserved catalytic site (CRSGQR motif) exists in node 13 of the assembly (Figure S1; closest relative T. crunogena (YP391186), Table 2).

T. thermophila contains additional homologues of genes involved in sulfur-cycling, including sulfide dehydrogenase, which is co-located with soxB on the genome, NADPH-dependent sulfite reductase and cytochrome c proteins, sulfide: quinone reductase, and thiosulfate dehydrogenase (Figure S1). Flavocytochrome-c sulfide dehydrogenase (pfam09242) oxidizes hydrogen sulfide to elemental sulfur coupled to the reduction of cytochrome c (Chen et al., 1994; Visser et al., 1997b). The highest-scoring matches for sulfide dehydrogenase are not T. crunogena, but Thioalkalivibrio sp. ALE20, Thiomicrospira halophila, and Chlorobaculum parvum (Table 2). Sulfite reductase reduces sulfite to hydrogen sulfide coupled to the oxidation of cytochrome c (Crane and Getzoff, 1996). Closest matches for sulfite reductase are T. crunogena and Acinetobacter lwoffii (Table 2). The sulfide:quinone reductase in T. thermophila is homologous to that of T. crunogena (Table 2, Figure S1) that mediates polysulfide formation from hydrogen sulfide by reduction of quinone (Reinartz et al. 1998). T. thermophila also has a gene coding for a putative thiosulfate dehydrogenase (Figure S1) that mediates tetrathionate production from thiosulfate (32.6% similarity to thiosulfate dehydrogenase from Allochromatium vinosum; Table 2; Denkmann et al., 2012).

No genes with significant homology to dissimilatory sulfite reductase (dsr) or sulfur reductase (sre) were identified in the T. thermophila genome. These genes could potentially code for enzymes mediating sulfite reduction to hydrogen sulfide and sulfur, or elemental sulfur reduction to hydrogen sulfide, respectively. Homologues for the sorAB genes that function as alternatives to the sox(CD)₂ genes (Kappler et al., 2001) to oxidize sulfite to sulfate, were also not found. Homologous genes for thiosulfate reductase (dvu, phs; Aketagawa et al., 1985; Bang et al., 2000) that mediate the disproportionation of thiosulfate to sulfite and hydrogen sulfide were not found. Homologous genes for tetrathionate reductase (ttr; Hensel et al., 1999) mediating the reduction of tetrathionate to thiosulfate were also not found. However, the failure to identify homologous sequences to these genes in T. thermophila does not unequivocally rule out their presence in unaligned portions of the genome (i.e., the genome was not closed).

Discussion

We have demonstrated that in T. thermophila strain EPR85 the rate of incomplete thiosulfate oxidation to elemental sulfur and sulfate (reaction ib in Table 1) is decoupled from the rate of elemental sulfur oxidation to sulfate (reaction ii in Table 1), with the rate of the latter being a strong function of pH. We have also confirmed that the stoichiometric imbalance of sulfur reported in this study is due to the production of tetrathionate at low pH (<6.0). The imbalance previously reported in Thiomicrospira and Thiobacillus strains is also likely due to tetrathionate production. Analysis of genomic DNA indicates the presence of enzymes for the complete Sox pathway. Based on these findings, we hypothesize that the observed pH dependence of elemental sulfur consumption is consistent with a modified Sox pathway that we outline below, which, if confirmed, represents the first report of elemental sulfur accumulation in bacteria that have genes for sox(CD)₂.

Thiosulfate consumption

The Sox pathway is a common multienzyme pathway for thiosulfate oxidation found in both photosynthetic purple and green sulfur bacteria and neutrophilic lithoautotrophic sulfur bacteria (Friedrich, 2001; Kelly et al., 1997). This pathway requires 7 essential proteins (SoxXYZABCD) to complete thiosulfate oxidation to sulfate (Friedrich et al., 2001). The combined activity of the periplasmic SoxYZ, SoxAX and SoxB proteins partially convert thiosulfate to one sulfate (Welte et al., 2009; Friedrich et al., 2000, 2001; Lu et al., 1985; Wodara et al., 1994; Figure 4; Table 1, enzymatic reaction ib). The remaining enzyme-attached sulfane group is oxidized to a second sulfate by the combined activity of Sox(CD)₂ and SoxB proteins (Figure 4; Wodara et al., 1997; Denger et al., 2008; Quentmeier et al., 2000; Table 1 reaction ii). T. thermophila strain EPR85 has homologous genes encoding for all the Sox enzymes (soxAXYZBCD) needed to completely oxidize thiosulfate to sulfate. Abiotic controls performed at all experimental conditions indicate negligible chemical reaction rates over similar time periods (Table S2) and indicate that the observed differences in thiosulfate oxidation products (e.g., sulfate, elemental sulfur, or tetrathionate) are due to microbially-mediated reactions. However, our observations of incomplete thiosulfate oxidation to sulfate associated with formation of elemental sulfur at low pH conditions (Figure 1a, b) suggest that the kinetic rate of thiosulfate oxidation to elemental sulfur (red path in Figure 4) is enhanced compared to elemental sulfur oxidation to sulfate (blue path in Figure 4).

Figure 4.

Variations on the Sox pathway inferred to exist in T. thermophila based on this study. When T. thermophila is grown buffered at high pH (8.0), sulfur oxidation is consistent with the classic Sox pathway of P. pantotrophus (Friedrich et al., 2001), shown here annotated with the oxidation state of each sulfur atom in blue. Charged species are indicated in purple. In unbuffered experiments once tetrathionate formation begins, the thiosulfate substrate would have to be divided between SoxAX and thiosulfate dehydrogenase to explain the observed chemistry (tsd; orange arrows). When grown buffered at low pH (5.5), sulfur oxidation by T. thermophila, appears to mediate the second part of the pathway (blue arrows) at a slower rate than the first part (red arrows), allowing elemental sulfur to accumulate in the media (see text).

Here, we consider a kinetic rate law that describes the net chemical effect of microbial growth via a multienzyme pathway. The rate of thiosulfate consumption was fitted to the following equation:

$$r=k{C}^q$$ (3)

where r is the rate in mmol/hr calculated for each timestep, k is the rate constant and q is a constant in hr⁻¹ (Walas, 1999). The power constant (q) is determined by fitting the rate data for S₂O₃ consumption as a linear function of the concentration of elemental sulfur. In the case where q is not equal to one, the integrated form of equation 3 is:

$$k=\frac{C_0^{q-1}}{(t-t_0)(q-1)}[{\left(\frac{C_0}{C}\right)}^{q-1}-1]$$ (4)

which can be linearized in the following form:

$${\left(\frac{1}{C}\right)}^{q-1}-{\left(\frac{1}{C_0}\right)}^{q-1}+k(q-1)(t-t_0)$$ (5)

Rate data for reaction i (Table 1) from four datasets during which thiosulfate reaction to elemental sulfur occurs simultaneously with elemental sulfur oxidation, while mass balance is conserved, were used to determine this constant (q = 0.421; Figure 5a). Thus, the rate of thiosulfate oxidation at all pH conditions tested is not first order, but rather a function of thiosulfate concentration as a power of the concentration of DCM-soluble solid elemental sulfur (k = 0.207; Figure 5b). That time series data from all conditions tested fit the same slope confirms that variable rates of the subsequent reactions (ii) and (iv) in Table 1 do not affect the rate law describing reaction (i) and that the rate is therefore independent of pH. In other words, the mechanism by which thiosulfate is incorporated into the sulfur oxidizing pathway is the rate-limiting step in thiosulfate consumption for T. thermophila.

Figure 5.

Rate data during thiosulfate oxidation to elemental sulfur and sulfate. (A) The power constant of the rate law is determined by the linear relationship between the log of the rate of thiosulfate consumption and the log of the normalized elemental sulfur concentration, giving a q of 0.421. Initial thiosulfate oxidation rate data (via the simultaneous mediation of reactions (1–3) in Table 1) from experiments ranging in pH from 7.5 to 5.0, each run in triplicate, were used to determine q. (B) The rate constant k for a power law is determined by the linear relationship between 1/[S₂O₃]^q-1 over time. The slope of the linear relationship in this plot corresponds to k(q-1) and the rate constant k for reaction (1) is 0.207.

Elemental sulfur formation

It is thought that the accumulation of elemental sulfur as an intermediate in the Sox pathway only takes place in organisms that lack both Sox(CD)₂, which converts the inner sulfur of thiosulfate directly to sulfate, and the dsr operon, which oxidizes elemental sulfur to sulfate (Dahl et al., 1993; Dahl et al., 2005). In Allochromatium vinosum, a rhodanese-like SoxL protein has been implicated in the formation of intracellular sulfur globules, cleaving long chains of polysulfides from SoxYZ (Welte et al., 2009; Table 1 reaction ib). In the case of T. thermophila strain EPR85, elemental sulfur is extracellular (Figure 6). Formation of sulfur globules via a partial Sox pathway has also been observed in colorless sulfur bacteria such as Beggiatoa (Mussmann et al., 2007) and members of the Chlorobiaceae family of green sulfur bacteria (Gregersen et al., 2011). T. thermophila has a homologue of a rhodanese-like enzyme SoxL, but apparently no homologues for the dsr genes that would oxidize elemental sulfur (Table 2).

Figure 6.

TEM images of cell pellets with elemental sulfur from the media. (A) Small amounts of amorphous elemental sulfur (arrow) in the media when grown in Tris buffer at pH 8.0. The smooth edges above and below are the edges of the copper grid on the holder. No biofilm was visible. Scale bar represents 500 nm. (B) Crystalline monoclinic elemental sulfur exhibiting a characteristic coffin shape (arrow) formed during growth in citrate buffer at pH 5.5. No biofilm was present. Scale bar represents 500 nm. (C) Sheets of biofilm trapped in the grid holder, containing nucleating amorphous elemental sulfur (arrow), formed after 4 hours of growth in unbuffered media. Biofilm (bold arrow) appears to effectively trap sulfur and allow growth of larger globules without leading to crystal formation, presumably aiding the cells in re-harvesting the trapped sulfur as other substrates become limiting. Scale bar represents 500 nm. (D) A single vibrioid cell (dark shape on the left of the image marked with arrow) surrounded by sulfur and imbedded in degrading biofilm network. Note the different scale bar (200 nm). Image taken after 22 hours of growth in unbuffered media when sulfur is actively scavenged and tetrathionate is being produced.

Formation of elemental sulfur by an organism that has genes for the Sox(CD)₂ enzyme is unexpected given that pH-dependent inhibition of the activity of Sox(CD)₂, responsible for converting the inner sulfur in thiosulfate to sulfate, has not been previously documented. However, it has been shown that the optimal pH for sulfite dehydrogenase activity of the Sox(CD)₂ complex is 8.0, with a 10% decline at pH 7.3 and 8.5 (Lu and Kelly, 1984). In addition, the crystal structure of Sox(CD)₂ in Paracoccus pantotrophus indicates that after the addition of the third oxygen atom onto the SoxY-bound sulfane, the newly created SoxY-S-SO_3⁻ complex is displaced by OH⁻, regenerating the oxidized molybdenum cofactor (Zander et al., 2011). At low pH, this displacement reaction could be inhibited. In addition, given that Sox(CD)₂ releases 6 H⁺, its activity may be inhibited when pH is already low. Our observations suggest a pH-dependent effect on Sox(CD)₂ activity that could also explain the observed elemental sulfur accumulation at low pH observed in T. crunogena (Javor et al., 1990) and Thiobacillus (Stefess et al., 1996).

In T. thermophila, the production of large quantities of extracellular elemental sulfur coincides with low pH and decreased growth rates (Figure 3). In media buffered at high pH (8.0), the rate of production is inferred to equal consumption such that elemental sulfur does not appreciably accumulate in the media (Figure 7a). In media buffered at low pH (5.5), elemental sulfur consumption occurs simultaneously with production that always exceeds consumption (Figure 7b). In unbuffered media initially at pH 7.5, elemental sulfur production and consumption occur simultaneously until pH decreases below 5.5, at which point the rate of consumption exceeds production corresponding to the onset of stationary phase (Figure 7c). In hydrothermal diffuse flow settings and on chimney surfaces bathed in hydrothermal fluids diluted with oxygenated seawater where T. thermophila is commonly found, these neutral pH values are common, particularly in fluids below 45°C (Foustoukos et al., 2009). However, in these environments, turbulent advective flow near chimney orifices frequently results in the outer surfaces of chimneys being subjected to higher temperature and more acidic fluids. Under these conditions, T. thermophila can oxidize thiosulfate to elemental sulfur and then maintain cell biomass by utilizing that sulfur.

Figure 7.

Schematic depicting the relative degree to which reactions 1, 2, and 4 (Table 1) occur over time at various pH conditions (indicated above each figure). Time series progression at high (A), low (B), and variable (C) pH conditions.

It is significant that the activity of SoxL is thought to produce polysulfides (Welte et al., 2009). Small concentrations (<40 µM) of sulfide are observed to form and remain at constant levels in the media throughout growth experiments buffered at pH 5.5 (Table S1). A mechanism for elemental sulfur formation via a polysulfide precursor may play a role in the creation of extracellular elemental sulfur in a variety of thiosulfate-utilizing bacteria. The exact mechanism for transferring polysulfide to extracellular sulfur globules like those produced by T. thermophila is currently unknown. Polysulfides, as charged dissolved species, would be more likely to be excreted extracellularly and rapidly convert to elemental sulfur (initially solid polymeric sulfur and then further to stable S₈ rings) and sulfide (Steudel, 1996). Alternatively, it is possible that polysulfide degrades to elemental sulfur and sulfide in the periplasm and nanocrystalline uncharged elemental sulfur diffuses across the cell membrane as in acidophilic Archaea (Boyd and Druschel, 2013). Our observations in this study are consistent with the nature of either of these mechanisms; however, further study is needed to confirm this.

Tetrathionate production in T. thermophila

Previous studies on thiosulfate oxidation by Thiobacillus and Thiomicrospira strains have demonstrated variable reaction stoichiometry and mass imbalances suggesting that these bacteria have complex strategies for maintaining growth during variable environmental conditions (Javor et al., 1990; Stefess et al., 1996). In addition to thiosulfate oxidation, for example, T. crunogena changes from elemental sulfur production at pH 6.3 to consumption at pH 8.5 (Javor et al., 1990). The sum of products in batch growth of T. crunogena was less than the sum of reactants, indicating the presence of additional undetermined sulfur products in previous experiments (Javor et al., 1990). Similar discrepancies in sulfur cycling mass balance were observed for two closely related strains of Thiobacillus that oxidize sulfide to elemental sulfur and sulfate (Stefess et al., 1996). Multiple oxidation pathways of thiosulfate have also been documented in Allochromatium vinosum, which variably produces elemental sulfur and sulfate via the Sox pathway or produces tetrathionate at low pH (Hensen et al., 2006). Given the similarities in observed stoichiometry and enzymes utilized in thiosulfate oxidation between A. vinosum and T. thermophila, we investigated the occurrence of tetrathionate to determine if the sulfur imbalance was a result of polythionate formation in T. thermophila.

The tetrathionate-producing thiosulfate dehydrogenase of T. thermophila closely matches that of A. vinosum (Table 2). The purple sulfur bacterium A. vinosum has also been shown to oxidize thiosulfate to tetrathionate under neutral to acidic pH conditions using a thiosulfate dehydrogenase (Hensen et al., 2006; Pott and Dahl, 1998). In addition, previous studies of Thiobacillus tepidarius reported fast rates of tetrathionate production via thiosulfate dehydrogenase (Lu and Kelly, 1988). More importantly, though, a recent review of thiosulfate dehydrogenase genes identified 2 conserved heme-binding motifs: the necessary catalytic cysteine, and conserved distal ligands that assist in heme-binding (Denkmann et al., 2012). The homologous gene in T. thermophila shares all conserved regions with only minor substitutions in the heme-binding motifs (Figure S2).

Below pH 6, T. thermophila abruptly switches to tetrathionate production from thiosulfate (Table 1 reaction iv), elemental sulfur consumption from the media, and continued processing of thiosulfate to sulfate (Figure 2a). Comparison of experiments, both with unbuffered and buffered pH (5.5 and 8.0), indicates that this shift in tetrathionate formation correlates with pH (Figure 8). The stoichiometry of sulfate production indicates that the elemental sulfur lost after this switch cannot account for all sulfate produced (about 1mM excess SO₄; Figure 1c); rather, some fraction of thiosulfate consumed must still be processed, albeit at a slow rate. This switch coincides with a change from decreasing to nearly invariant pH in the medium, possibly as a result of simultaneous and comparable reaction rates of elemental sulfur oxidation that consumes H⁺ and of tetrathionate production that produces H⁺ (reactions ii and iv, Table 1). Given the similarities in reported stoichiometry of thiosulfate oxidation and tetrathionate production in other colorless sulfur bacteria such as T. crunogena (Javor et al., 1990; Scott et al., 2006), Sulfurimonas denitrificans (Sievert et al., 2008), and Thiobacillus strains (Stefess et al., 1986), the genetic mechanism responsible for our observations of extracellular elemental sulfur formation and simultaneous tetrathionate production may be more widespread than previously thought.

Figure 8.

Inferred and/or measured tetrathionate (see text) as a function of pH in repeat experiments. Closed symbols represent composition during exponential growth; open symbols are composition during stationary phase. Media tested included unbuffered media (squares), Bis-Tris buffer at pH 5.5 (triangles), citrate buffer at pH 5.5 (diamonds), and Tris buffer at pH 8.0 (circles). Measured values are shown with error bars.

Conclusions

Thiomicrospira thermophila strain EPR85 has a wide range of metabolic options for coping with the temporal variability in fluid composition that characterizes seafloor diffuse-flow hydrothermal environments. In particular, pH is the dominant control over the type of reactions that are mediated and pH below 5.5 can cause the onset of stationary phase growth, even when thiosulfate is abundant. At pH below 6.0, thiosulfate is oxidized to tetrathionate, accounting for the missing sulfur observed in previous experiments. Thiosulfate is converted to elemental sulfur and sulfate at pH 5.5–8, with the amount of elemental sulfur accumulating in the media increasing as pH decreases. The observed sulfur oxidation chemistry is consistent with the Sox pathway, for which homologues of all genes necessary were found in the genome of T. thermophila strain EPR85. The pH dependency of elemental sulfur accumulation suggests a pH-dependent limitation on the activity of Sox(CD)₂, resulting in incomplete thiosulfate oxidation. This is the first report of obligate formation of elemental sulfur as an intermediate in an organism that does not contain the dsr genes (Grimm et al., 2008; Dahl et al., 2002).

When pH is lower than 5.5, tetrathionate production is upregulated. T. thermophila strain EPR85 has a homologous gene for thiosulfate dehydrogenase, which can mediate tetrathionate production while consuming H⁺, partially counteracting the decrease of pH in the medium. Simultaneous with tetrathionate production is the shift to elemental sulfur consumption exceeding production, even if thiosulfate is still present in the media. Thus, T. thermophila strain EPR85 can mediate several sulfur oxidation reactions simultaneously, resulting in large geochemical changes with very little increase in biomass.

Experimental Procedures

Bacterial strain and growth conditions

Thiomicrospira thermophila strain EPR85 was isolated from fluids collected from a diffuse flow vent on the East Pacific Rise (EPR; 9°50’N, 104°17’W), at a depth of approximately 2500 m in May 2005 (Cruise AT 11–26, Alvin dive 4103). Primary enrichment cultures were initiated shipboard by adding 1 mL of fluids to 10 mL of growth medium designed to enrich for chemolithoautotrophic thiosulfate oxidizing bacteria. The medium used was 142-A (Crespo-Medina et al., 2009). Cultures were incubated at 30°C in the dark. Growth was initially determined by the observation of a change in color of the pH indicator in the medium and then confirmed by direct microscopic counts after staining the cells with acridine orange. For isolation of single colonies, liquid cultures of strain EPR 85 were inoculated on Petri dishes containing medium 142-A solidified with 15 g L⁻¹ of noble agar (Sigma). Stocks for long-term storage were prepared by adding 150 µL of 100% sterile glycerol (Fisher Scientific) to 850 µL of overnight cultures, and were stored at –80°C.

In this study, T. thermophila strain EPR85 was grown routinely in batch culture with air in the headspace at 35°C on a shaker table. Unbuffered seawater salinity medium contained the following (L⁻¹): 25g NaCl, 0.2g NH₄Cl, 1.23g MgCl·6H₂O, 0.4g CaCl₂·2H₂O. After autoclaving, the following filter-sterilized solutions were added (by volume): 10 wt% K₂HPO₄ (0.25g in 500mL), 10 wt% NaHCO₃ (2.52g in 500mL), 1 wt% trace metal solution SL-10, and ~6 wt% Na₂S₂O₃·5H₂O (16g in 500mL). The trace metal solution SL-10 is made as follows (in 1L): 1.5g FeCl₂·4H₂O dissolved in 10mL 25% HCl, 70mg ZnCl₂, 100mg MnCl₂·4H₂O, 6mg H₃BO₃, 190mg CoCl₂·6H₂O, 2mg CuCl₂·2H₂O, 24mg NiCl₂·6H₂O, 36mg Na₂MoO₄·2H₂O. The thiosulfate stock solution was made fresh often and was not added to media until just prior to inoculation to prevent potential degradation in the media. Bicarbonate, phosphate, and trace metals were added just prior to inoculation as well. For the citrate buffer experiments, the salts were added to a buffer base made by mixing 115 mL 0.1M citric acid with 885 mL 0.1M sodium citrate. The Bis-Tris buffer was made at 0.1 M and pH was set to the desired value with HCl. The Tris buffer was made by mixing 557 mL 0.1M Tris-HCl with 443 mL 0.1M Tris base. The pH was measured prior to inoculation.

Culture stock stored at −80°C was routinely pre-grown in gas-tight serum bottles containing 15mL unbuffered medium and transferred to 50mL volumes twice prior to the start of each experiment. The headspace volume in the gas-tight bottles was at least 2/3 of the total volume to ensure that oxygen gas would not be limiting. The inoculum culture was transferred in exponential growth phase at the start of each experiment to limit the lag phase. Serum bottles were constantly agitated on a Thermo Forma 420 orbital shaker table set at 35°C. Each experiment was grown in triplicate and accompanied by an abiotic control. The abiotic controls never produced any change in thiosulfate concentration, pH or cell abundance.

Analytical techniques

Thiosulfate and sulfate concentrations were monitored by ion chromatography (Metrohm 881) after filtering out biomass and elemental sulfur using an anion column (Metrosep ASupp5-100) with suppression (0.1M sulfuric acid). Conductivity and UV-Vis detectors (Metrohm 887) were employed in series. The eluent solution was 3.2mM Na₂CO₃/1mM NaHCO₃ with 2.5 vol% acetone as a modifier. The UV-Vis detector (215 nm) was used exclusively for thiosulfate analysis. Sulfate is not UV-detectable and dissolved species concentrations were analyzed by conductivity detector. pH was monitored using a Mettler Toledo Ag/AgCl electrode. Sulfide was measured using the method of Cline (1969) as follows: 1mL sample was fixed with excess Zn acetate (1mL, 0.3M); 1mL of that solution was added to 0.08mL of mixed diamine reagent, and allowed to sit for up to 1 hour; after the color change this solution was diluted to a volume of 12.5mL with deionized water, transferred to a plastic cuvette and absorbance at 664nm was analyzed on a UV-Vis spectrometer (Thermo Evolution 60). A blank of sulfide-free media was also fixed in Zn acetate and processed in the same manner. Polythionates and thiosulfate were analyzed by the cyanolysis method (modified from Kelly and Wood, 1994) described in the supplemental material.

Elemental sulfur (S₈) was extracted from water samples using dichloromethane in glass tubes without letting solvent touch the caps. Three sequential extractions were combined, each with two times the volume of the water sample, to ensure complete extraction. All transfers were made with glass pipettes and quartz cuvettes were used to allow detection in the UV range. The extracted DCM was then measured on a UV-Vis spectrometer (Thermo Evolution 60) by scanning from 200–700nm. Elemental sulfur has a distinctive double peak (Meyer, 1976), which in DCM occurs at 270nm and 285nm. Both peaks are linear with respect to sulfur concentration. The 270nm peak was chosen because at very low concentrations the 285nm peak is sometimes indistinct. The scan of a blank (sulfur-free DCM) was subtracted from each sample and the peak height at 270nm was normalized to the background at 550nm, as there were differences in baseline in samples analyzed at different times. Polymeric sulfur is insoluble in organic solvents (Steudel and Eckert, 2003). We developed a protocol to measure polymeric sulfur by converting it to DCM-soluble S₈ rings. Based on the stability of commercially available polymeric sulfur (Crystex, Eastman Chemical Company), after the first triple extraction we treated the remaining sample with Tris base (pH 9.36) to raise the pH and provide excess amines in solution. The sample was then autoclaved for 30 minutes and allowed to sit overnight to promote conversion of polymeric sulfur to stable S₈. Elemental sulfur (S₈) was then quantified after a second triple extraction in DCM. Although we cannot quantify the yield of this process, no insoluble residue was visible after this treatment.

Samples taken for cell counts were fixed in 4 vol% formaldehyde in phosphate buffered saline (PBS: 10.9 g K₂HPO₄, 3.2 g NaH₂PO₄, 90 g NaCl in 1L DIW)) to a dilution factor of 2. Due to the production of elemental sulfur solids in the media, optical density could not be used to determine biomass abundance. Cells were stained with DAPI (2µL of 1ppb DAPI in PBS per mL of sample) and allowed to sit for 10 minutes prior to filtering through 13mm black polycarbonate 0.22µm filters (Millipore). Variable volumes (100–400µL) were carefully pipetted onto a single spot on the filter to concentrate the biomass for imaging. By varying the volume used, the optimal cell density for counting was achieved (<100 cells/field of view). The filters were imaged on a Nikon Eclipse 80i autofluorescence microscope at 1000X magnification. Five to ten representative fields of view were counted and averaged. The cell density (cells/mL) was adjusted for the factor (f), the volume filtered (v), and the dilution (d) caused by the fixative:

$$\mathrm{C}=\mathrm{c}\mathrm{f}\mathrm{d}/\mathrm{v}$$ (1)

where the fraction (f) is the area of the spot size pipetted onto the filter divided by the area of the field of view counted. Counting with microscopy became difficult once elemental sulfur started forming, as the cells begin to attach and form clumps with the sulfur. As a result the error of counting increases with elemental sulfur concentration, as shown in Figure S3. Specific growth rates, reported in hr⁻¹, were calculated according to:

$$\mathrm{\mu }=\text{ln}(\mathrm{X}/{\mathrm{X}}_{\mathrm{i}})/(\mathrm{t}-{\mathrm{t}}_{\mathrm{i}})$$ (2)

Separate growth rates were calculated if the reaction stoichiometry shifted during an experiment.

Nucleic acid analysis

Cells were grown in 50mL unbuffered media and harvested during exponential phase by filtering out excess elemental sulfur through a 5µm filter before pelleting by centrifugation. Genomic DNA was extracted using the Promega Wizard Genome DNA Purification Kit, following the manufacturer’s instructions. Quality and concentration of extracted DNA was verified on a Nanodrop 2000c and by a 2 wt% agar gel to confirm the size and purity of the product.

From the extracted genomic DNA, the 16S DNA sequence was amplified by PCR (Phusion) using the universal bacterial primers 27F (AGAGTTTGATCCTGGCTCAG) and 1492R (GGTTACCTTGTTACGACTT) on an Eppendorf Mastercycler. PCR product was cleaned using the Wizard SV Gel and PCR Cleanup kit (Promega) and the quality of PCR product was confirmed with Nanodrop and by agarose gel. Product was sent to Eurofins MWG Operon for sequencing. To ensure that we had uncontaminated genomic DNA from our strain prior to getting the genome sequenced, we cloned our blunt-end PCR product into a plasmid using the Zero Blunt TOPO PCR cloning kit (Life Technologies) and grew the plasmid in a kanamycin-resistant strain of E. coli (OneShot TOP10 Chemically Competent Cells; Life Technologies). The plasmid-bearing E. coli was grown on LB broth agar plates dosed with kanamycin at 37°C overnight. Eight single colonies were transferred to 10mL of LB broth dosed with 10µL kanamycin and grown on a shaker table at 37°C overnight. The full 10mL culture was pelleted and the plasmid DNA extracted using the Pure Yield Plasmid Mini Prep System (Promega). Five of these were sent for sequencing and all the sequences aligned against T. crunogena XCL-2 as a reference.

Genomic DNA was sent to the Genome Technology Access Center at Washington University School of Medicine for paired-end Illumina sequencing. The first 10 bp from the 5’ end of 101bp length sequences were clipped from the raw Illumina data. Replicate sequences were removed and the quality of the remaining sequences was confirmed with FastQC (Babraham Bioinformatics). Any sequences with a Phred score of <30 or a sequence length of less than 50bp were removed using Sickle (Joshi and Fass, 2011). The remaining sequences were assembled using SPAdes (Bankevich et al., 2012) with a range of K-mer lengths (15, 21, 33, 55, 70) tested with very little difference between them when the quality was checked with Quast (Gurevich et al., 2013). The total number of contigs in each assembly ranged from 37–40, with L75 score ranging from 9–11, and zero uncalled bases. The assembly used has a total length of 2.65 mbp and 40 contigs total, with the largest 20 accounting for 97.6% of the bases of the assembly. The G+C content of the assembly is 50.3%.

The assembly was examined for genes involved in all known sulfur oxidation pathways (MetaCyc) using a series of BLAST queries. These included NADPH-dependent sulfite reductase (EC.1.8.1.2), sulfite oxidoreductase (EC.1.8.2.1), thiosulfate dehydrogenase (EC.1.8.2.2), sulfide dehydrogenase (EC.1.8.2.3), tetrathionate reductase (EC.1.8.5.2), sulfide:quinone reductase (EC.1.8.5.4), dissimilatory sulfite reductase (EC.1.8.99.1), thiosulfate reductase (EC.1.97.1), and rhodanese (EC.2.8.1.1). To confirm those BLAST hits with e value <10⁻²⁰, the amino acid sequences were compared to the translated amino acid sequence from the assembly. In cases where the closest match was a hypothetical protein, several of the closest matches were examined to determine the homologous gene. In the case of rhodanese, which has been non-selectively annotated in the past, and SoxL, which has only one gene annotated with this identity but is similar to many genes annotated as hypothetical proteins, protein alignments were constructed to confirm the active site motif and conserved regions. In addition, protein alignments were constructed to confirm similarity between multiple genes that gave BLAST hits of the same protein. Global protein alignments with free end gaps were constructed using Blosum62.

Phylogenetic Analysis

We obtained a nearly full-length sequence (1514 bp) of the 16S rRNA gene from T. thermophila strain EPR85 and, for comparison, Thiomicrospira crunogena XCL-2 (Scott et al., 2006). These sequences were aligned to release 117 of the SILVA SSU reference database (Quast et al. 2013) using the SILVA Incremental Aligner (SINA) (Pruesse et al. 2012). This method utilizes a combination of k-mer searching and partial-order alignment to optimize the accuracy of alignment. We then constructed a phylogenetic tree in ARB (Ludwig et al. 2004) these sequences and other Gammaproteobacteria using the “All-Species Living Tree” Project (LTP) (Yarza et al. 2008; Munoz et al. 2011) as a template. 16S rRNA gene sequences in the LTP are derived from type strains of all classified strains of Bacteria and Archaea. Tree construction was done using the maximum likelihood algorithm RAxML (Stamatakis 2006). T. thermophila strain EPR85 and T. crunogena XCL-2 were inserted into the tree using the ARB parsimony algorithm.

Accession numbers in Genbank

The sequence for the 16S rRNA gene was submitted to GenBank under the accession number KP749905. Sequences for the SoxC and SoxD genes that code for proteins involved in the Sox sulfur oxidation pathway were submitted to GenBank under accession numbers KP749906 and KP749907, respectively.

TEM imaging

To investigate the nature of elemental sulfur produced by T. thermophila under different pH conditions, we imaged the sulfur-biomass clusters with a Tecnai Spirit TEM operated at 120 kV with a 2K 4-megapixel CCD TEM camera. Culture samples at different stages of growth were pipetted (100µL) onto copper grids, allowing excess water to wick onto a kimwipe. The loaded grid was then stained with methylene blue. Only material that attached to the edge of the grid was visible with this treatment and samples with a high abundance of EPS produced better images as a result.

Supplementary Material

figS1. Figure S1.

Genome scaffold lengths labeled by node # referenced in the text. Locations of specific homologous genes are indicated by boxes in their order on the scaffold (cyt-c: cytochrome c; sqr: sulfide quinone reductase; sr: sulfite reductase; sd: sulfide dehydrogenase; rhd: rhodanese; tsd: thiosulfate dehydrogenase; SoxABCDLXY: enzymes involved in the Sox pathway). Genes indicated by parentheses are second copies of genes with weaker affinities.

figS2. Figure S2.

Sequence alignment of thiosulfate dehydrogenase (tsd) homologous proteins. Included is the A. vinosum and Tint_2892 genes of the alignment in Denkmann et al. (2012), the closest match with the T. thermophila gene annotated as a hypothetical protein in T. arctica, and the gene from T. thermophila from this study. The heme-binding motifs are highlighted in grey; distal ligands are in blue, and the catalytic cysteine in red.

figS3. Figure S3.

Phylogenetic tree of the Gammaproteobacteria created in ARB based on a 16S rRNA gene alignment from the SILVA “All-Species Living Tree” Project (Munoz et al., 2011; Yarza et al., 2008). Sequences from Thiomicrospira thermophila strain EPR85 and Thiomicrospira crunogena XCL-2 were inserted using the ARB parsimony algorithm.

tableS1. Table S1.

Stoichiometry of all growth experiments. Polythionates (S₃O₆ and S₄O₆) were determined using cyanolysis. *Indicates S₂O₃ concentrations determined with the cyanolysis method. All other S₂O₃ concentrations were determined by ion chromatography.

tableS2. Table S2.

Stoichiometry of all abiotic control experiments. Polythionates (S₄O₆) were determined using cyanolysis. *Indicates S₂O₃ concentrations determined by the cyanolysis method. All other S₂O₃ concentrations were determined by ion chromatography. All experiments were performed at 35°C unless otherwise noted. Measurements below detection are indicated as b.d. and as n.a. if not analyzed.