Bacterial CS₂ Hydrolases from Acidithiobacillus thiooxidans Strains Are Homologous to the Archaeal Catenane CS₂ Hydrolase

Abstract

Carbon disulfide (CS₂) and carbonyl sulfide (COS) are important in the global sulfur cycle, and CS₂ is used as a solvent in the viscose industry. These compounds can be converted by sulfur-oxidizing bacteria, such as Acidithiobacillus thiooxidans species, to carbon dioxide (CO₂) and hydrogen sulfide (H₂S), a property used in industrial biofiltration of CS₂-polluted airstreams. We report on the mechanism of bacterial CS₂ conversion in the extremely acidophilic A. thiooxidans strains S1p and G8. The bacterial CS₂ hydrolases were highly abundant. They were purified and found to be homologous to the only other described (archaeal) CS₂ hydrolase from Acidianus strain A1-3, which forms a catenane of two interlocked rings. The enzymes cluster in a group of β-carbonic anhydrase (β-CA) homologues that may comprise a subclass of CS₂ hydrolases within the β-CA family. Unlike CAs, the CS₂ hydrolases did not hydrate CO₂ but converted CS₂ and COS with H₂O to H₂S and CO₂. The CS₂ hydrolases of A. thiooxidans strains G8, 2Bp, Sts 4-3, and BBW1, like the CS₂ hydrolase of Acidianus strain A1-3, exist as both octamers and hexadecamers in solution. The CS₂ hydrolase of A. thiooxidans strain S1p forms only octamers. Structure models of the A. thiooxidans CS₂ hydrolases based on the structure of Acidianus strain A1-3 CS₂ hydrolase suggest that the A. thiooxidans strain G8 CS₂ hydrolase may also form a catenane. In the A. thiooxidans strain S1p enzyme, two insertions (positions 26 and 27 [PD] and positions 56 to 61 [TPAGGG]) and a nine-amino-acid-longer C-terminal tail may prevent catenane formation.

INTRODUCTION

The sulfur compounds carbon disulfide (CS₂) and carbonyl sulfide (COS) play an important role in the earth's sulfur cycle. CS₂ and COS are released as breakdown products from organic matter, notably S-containing amino acids in soils (1) and dimethyl sulfide (DMS) in marine, mainly coastal and estuarine, environments (2). CS₂ is chemically and biologically converted to COS, which is highly stable when not in solution and is the most abundant sulfur species in the atmosphere (3). Anthropogenic CS₂ emissions account for approximately one-half of the total global emissions (2). This is due for a great part to the use of CS₂ as an organic solvent in the viscose and rayon industry, which brings with it a number of problems. First, CS₂ is toxic, causing vascular and coronary heart disease and affecting the central nervous system (4). Second, due to its low boiling point, large CS₂-polluted airstreams are created in factory plants and require treatment before release to the atmosphere.

Biofiltration is an effective and sustainable method to remove CS₂ from the contaminated airstreams (5–7). Several bacterial species that can grow chemolithoautotrophically on CS₂ at neutral pH have been identified from soil, sludge, and freshwater habitats (8–12). At acidic pH, so far only some (Acidi)thiobacillus strains, isolated from hot springs and volcanic areas, were shown to be able to use CS₂ (8, 13). Acidophilic Acidithiobacillus thiooxidans strains are currently in use in biotrickling filters (6, 14–16), due to the inherent acidification of the trickling filter upon growth on CS₂. The microorganisms convert CS₂ via a 2-step hydrolysis reaction (12, 17): CS₂ + H₂O → COS + H₂S and COS + H₂O → CO₂ + H₂S. The H₂S is subsequently oxidized, ultimately to sulfuric acid (18), yielding the energy required for growth, but also acidifying the biofilter trickling water. Large quantities of water are used to maintain the pH at levels tolerated by the CS₂-removing microorganisms. The use of new, more extreme acidophiles would reduce water usage as well as operational costs, making biofiltration more sustainable and effective (6, 19).

Biofiltration technologies can benefit from more in-depth knowledge about the molecular mechanism of bacterial CS₂ conversion. Although several bacterial CS₂-converting species are now known (9–13, 17, 20), their CS₂-converting enzymes have not been characterized at all. However, we previously purified the CS₂ hydrolase from the CS₂-converting hyperthermophilic archaeon Acidianus strain A1-3 and showed that it could convert CS₂ to COS, H₂S, and CO₂ via the hydrolysis reaction as described above (21). The CS₂ hydrolase appeared to be homologous to β-carbonic anhydrases (β-CAs), which catalyze the reversible hydration of CO₂ + H₂O ⇆ HCO₃⁻ + H⁺. The crystal structure of the CS₂ hydrolase revealed that the enzyme occurs as an octameric ring like the β-CA from the garden pea Pisum sativum (22). However, in the case of Acidianus CS₂ hydrolase, two of these rings interlock, forming a highly unusual hexadecameric catenane structure, both in the crystal form and in solution (21, 23). Intriguingly, despite the high homology with CAs, the Acidianus CS₂ hydrolase could not use CO₂ as a substrate, and CAs have not been found to use CS₂ as a substrate, although the conversion is theoretically possible (24–26). Recently, a COS hydrolase enzyme was purified from Thiobacillus thioparus strain THI15, which is also a β-CA homologue. It does not form a catenane structure. Instead, a weakly associated tetrameric ring is formed (27). Its physiological role is COS conversion, not CS₂ conversion (27). The COS hydrolase is not closely related to the Acidianus CS₂ hydrolase in the β-CA clade D cluster.

With the unusual catenane structure of the archaeal CS₂ hydrolase in mind, we set out to study bacterial CS₂ hydrolase enzymes from new strains of CS₂-degrading Acidithiobacillus thiooxidans isolated from sulfur-rich and highly acidic environments. Here, we report on the purification and characterization of two bacterial CS₂ hydrolases. We show that archaeal and bacterial CS₂ hydrolases are closely related and that the bacterial homologues also form catenane structures.

MATERIALS AND METHODS

Media and culture conditions.

New Acidithiobacillus thiooxidans strains were isolated from sulfur-rich and highly acidic ecosystems including mixed hot spring samples (strain 2Bp), Solfatara, Naples (strain Sts 4-3), and Solfarata, Rome (strain S1p), or from samples from CS₂-removing biotrickling filters (strains G8 and BBW1). The strains were enriched and isolated with CS₂ as the sole energy source. Physiological details of the new strains will be described in a future article. Strains were grown in minichemostats, set up using 250-ml Schott bottles, with a culture volume of 150 ml. The medium was a basal salt mineral medium (MM) consisting of (in g · liter⁻¹) KH₂PO₄ (0.2), NH₄Cl (0.5), MgSO₄ · 7 H₂O (0.75), CaCl₂ · 2 H₂O (0.1), and 1 ml/liter trace element solution and acidified with 1% (vol/vol) H₂SO₄ (pH 0.75). Trace elements consisted of the following (in g · liter⁻¹): EDTA-Na · 2 H₂O (10.0), ZnSO₄ · 7 H₂O (2.2), MnCl₂ · 4 H₂O (1.02), FeSO₄ · 7 H₂O, (NH₄)₆Mo₇O₂₄ · 4 H₂O (0.22), CuSO₄ · 5 H₂O (0.32), and CoCl₂ · 6 H₂O (0.32). To maintain the extremely acidophilic strains, the sulfuric acid concentration was kept at 1%. The medium flow was 2.7 ml · h⁻¹ (D = 0.023 h⁻¹). The reactor temperature was 22°C. CS₂ gas (8.6 ± 0.05 μM) was bubbled through the reactors with a flow rate of 41.5 ± 0.1 ml · min⁻¹ and mixed into the reactor with a stir bar, stirring at 1,000 rpm. A. thiooxidans strains S1p and G8 were also grown in batch reactors, in MM acidified with sulfuric acid to pH 2. The reactor temperature was 25°C. CS₂-containing air was bubbled through the reactor with a flow rate varying between 132 and 140 ml · min⁻¹ and dispersed through the culture by stirring at 500 to 1,000 rpm. The concentration of CS₂ gas was varied between 10 nmol · ml⁻¹ for low-density cultures and 125 nmol · ml⁻¹ for high-density cultures. The μ_max values were determined by increasing the medium flow and simultaneously increasing the [CS₂] supplied to the reactors up to and beyond the point where S⁰ formation started becoming visible in the reactors.

Methylomicrobium alcaliphilum DSM19304 was grown at 28°C, with shaking at 200 rpm, as 10-ml cultures in 120-ml bottles closed with butyl rubber stoppers, on DSMZ medium 1180, and containing 15 ml CH₄. Halopiger xanaduensis DSM18323 was grown in Erlenmeyer flasks at 37°C and shaking at 200 rpm in the medium described in reference 28. The pH of the medium was adjusted to pH 8.0 before and after autoclaving with sterile 1 M NaOH. Mycobacterium marinum strain M was grown statically in 7H9 medium at 30°C.

Preparation of cell extracts.

Batch reactors with a culture optical density at 600 nm (OD₆₀₀) of 1.7 were harvested by centrifugation at 9,000 × g for 15 min. The pellets (about 4 g, wet weight) were washed with 20 mM Tris, pH 8, to raise the pH to 8.0. Pellets were resuspended in 15 ml 20 mM Tris, pH 8, containing EDTA-free protease inhibitors (Roche Diagnostics). Cells were broken by a French press (21), DNase I was added, and the suspension was centrifuged at 48,000 × g for 60 min. Supernatants were frozen at −20°C. Protein concentrations were determined using the Bio-Rad protein microassay. Using dry weight measurements of purified CS₂ hydrolase, we determined that bovine serum albumin (BSA) as a standard rather than IgG produced more accurate protein determinations for CS₂ hydrolase. Cell extracts from steady-state reactor-grown bacterial cells were prepared as follows: 30 to 50 ml was removed from the reactors and centrifuged at 4°C for 30 min at 12,000 × g. The cell pellets were washed with 15 ml sterile distilled water (sdH₂O) and resuspended in 0.5 ml 20 mM KP_i, pH 7. Approximately 350 μl glass beads (size, 80 to 110 μm) were added, and the cells were broken by bead beating for 2 times 2 min at 30 Hz (Retsch, Germany) with intermittent cooling on ice. The broken cell mixtures were centrifuged for 5 min at 16,000 × g, and the supernatants were either used directly for protein assays and kinetic measurements or stored at −20°C with a final concentration of 10% glycerol.

Purification of CS₂ hydrolases.

CS₂ hydrolases from A. thiooxidans strain G8 and S1p were purified by ammonium sulfate precipitation, followed by hydrophobic interaction chromatography (HIC) and anion-exchange chromatography. Chemicals used for chromatography were of ultrapure grade. Prior to ammonium sulfate precipitation, the Tris concentration was raised from 20 to 50 mM by adding an appropriate volume of 1 M Tris-HCl, pH 8. Extracts were stirred continuously and kept on ice, while finely ground (NH₄)₂SO₄ was added slowly to 30%, wt/vol. The extract was left on ice for 1 h and centrifuged for 45 min at 3,000 × g, and another aliquot of (NH₄)₂SO₄ was added until 60% saturation was reached.

For strain G8, (NH₄)₂SO₄ was added to 60%, wt/vol, followed by incubation and centrifugation as described above. The pellet was dissolved in 3 ml 50 mM Tris-HCl, pH 8, and centrifuged for 5 min at 13,500 × g. The resulting supernatant was filtered through a 0.45-μm filter unit (Millex-HV polyvinylidene difluoride [PVDF] Durapore). (NH₄)₂SO₄ was added from a 3.4 M stock to raise the concentration to 1 M, and the whole fraction was loaded onto a 1-ml HIC Hitrap Phenyl Sepharose 6 Fast Flow (High sub) column (GE Healthcare) equilibrated with 1.7 M (NH₄)₂SO₄ in 50 mM Tris-HCl (pH 8), using the Åkta system (GE Healthcare). Proteins were separated using a linear gradient from 1.7 to 0 M (NH₄)₂SO₄ over 10 column volumes at a flow rate of 2 ml · min⁻¹. Two-milliliter fractions were collected and assayed for CS₂ hydrolase activity (see below). The CS₂ hydrolase eluted at the very end of the gradient at 122 mM (NH₄)₂SO₄ together with at least 2 other dominant proteins. Active fractions were pooled and concentrated 3 times, and the buffer was exchanged to 20 mM Tris-HCl (pH 8) by centrifugation at room temperature (RT) through a Vivaspin 30-kDa cutoff spin column (Sartorius). Two milliliters was loaded onto a 1-ml Hitrap ANX-Fast Flow (High sub) column (GE Healthcare), and proteins were separated at RT using a linear gradient from 0 to 1 M NaCl over 25 column volumes and a flow rate of 1 ml · min⁻¹. The CS₂ hydrolase eluted at 117 mM (11.7%) NaCl.

For strain S1p, the 30% (NH₄)₂SO₄ supernatant fraction was used to separate the proteins on the HIC column equilibrated with 1.3 M (NH₄)₂SO₄ in 50 mM Tris-HCl (pH 8) described above. The S1p CS₂ hydrolase eluted in a broad peak between 1.1 M and 179 mM (NH₄)₂SO₄, but the highest activity was seen at the lower salt concentrations, perhaps due to inhibition by (NH₄)₂SO₄. Pooled active fractions were concentrated 13 times, and the buffer was exchanged as described above. The ANX column was loaded with 100-μl samples of the concentrated fractions, and proteins were separated at RT using a linear gradient from 0 to 1 M NaCl over 18 column volumes and a flow rate of 1 ml · min⁻¹. The CS₂ hydrolase eluted in a well-separated peak at 202 mM NaCl. Fractions containing the purified enzyme were stored on ice or, if not used within 1 week, frozen at −20°C.

Protein gel electrophoresis.

SDS-PAGE (10 or 12% polyacrylamide, pH 8.3) and native PAGE (6 or 8% polyacrylamide, pH 8.3) was performed using a Mini-Protean III Cell (Bio-Rad) at RT. Fermentas unstained or prestained and the PageRuler Plus prestained molecular weight standards were used for SDS-PAGE. Invitrogen P/N 57030 markers were used for native PAGE. Ten micrograms of protein was loaded per lane. Proteins were visualized by Coomassie brilliant blue (CBB) G 250.

CS₂ and COS hydrolase activity measurements.

CS₂ hydrolase activity of purified fractions and on native PAGE gels was measured qualitatively as described previously (21) but at 30°C. CS₂ and COS hydrolase activities of purified CS₂ hydrolases were quantified by gas chromatography (29). To 120-ml bottles, 500 μl 20 mM HEPES (pH 7) was added. As COS was unstable in the buffer and its breakdown rate depended on the buffer volume, the latter was minimized and kept at 500 μl. This small volume simultaneously maximized the gas to liquid transfer of CS₂. Experiments were run against abiotic controls to check that substrate conversion was due to enzyme activity rather than chemical breakdown. Bottles were sealed with a gray butyl rubber stopper, and between 0.25 and 10 ml of CS₂-saturated air (final concentration in the buffer, between 8 and 800 μM CS₂) or between 0.125 and 4 ml COS (final concentrations in the buffer, between 5 and 473 μM COS) was injected. CS₂-saturated air was obtained by incubating an airtight 1-liter bottle containing 50 ml CS₂ for several hours. COS was injected from a COS gas bottle. The bottles were incubated at 30°C with shaking at 400 rpm for 15 min to ensure good gas transfer into the buffer. Purified CS₂ hydrolase (0.3 to 0.7 μg) in 100 μl 20 mM HEPES (pH 7) was injected. The formation of hydrolysis products from CS₂ (COS and H₂S) and COS (H₂S) was measured by gas chromatography, and production rates over the first 3 min were used to calculate enzyme activity. CS₂ and COS concentrations in the buffer were determined by measuring the concentration in the headspace of the bottles at the end of each experiment and converting it to liquid concentrations using the solubility of COS and CS₂ in 20 mM HEPES buffer (pH 7) at 30°C. The solubility of CS₂ or COS was determined by adding known amounts to a 120-ml bottle containing 60 ml buffer, incubating them for 30 min at 30°C with shaking at 400 rpm, and measuring the remaining CS₂ or COS in the headspace as well as the liquid phase by gas chromatography. The liquid/gas ratio at 30°C was 0.45:1 for COS and 0.94:1 for CS₂.

CS₂ hydrolase activity was also quantified using an H₂S Clark-type microsensor (Unisense A/S, Denmark), calibrated using an anaerobic Na₂S stock diluted in 20 mM HEPES (pH 7). A 6 mM CS₂ stock was prepared by mixing 200 μl CS₂ with 500 ml dH₂O in a 500-ml serum bottle by vigorous shaking for at least 30 min. Up to 0.1 μg cell extract in 20 mM HEPES (pH 7), 0.1 μg purified CS₂ hydrolase, or 100 μg bovine carbonic anhydrase was added to a 1-ml glass vessel containing 1 ml of 20 mM HEPES (pH 7) stirred at 500 to 1,000 rpm, in a 22°C water bath. The H₂S sensor was lowered into the vessel, CS₂ was added from the stock bottle to 600 μM CS₂, and the formation of H₂S was followed for approximately 30 s. Initial H₂S production rates were calculated to determine the V_max values for each cell extract.

The CS₂ hydrolase activity of bacterial strains was also tested by gas chromatography. A CS₂ stock bottle was prepared by adding 400 μl CS₂ to an empty 500-ml bottle sealed with a rubber seal and incubating it for 30 min at RT for the CS₂ to evaporate. Five- or 10-ml samples from late-logarithmic or early-stationary-phase bacterial cultures were added to 120-ml bottles, which were sealed with gray butyl rubber stoppers, and 0.25 ml CS₂-containing air was added from the stock bottle to obtain a CS₂ concentration of about 10 to 15 nmol · ml⁻¹ CS₂. The cultures were incubated at the temperature used for growth and shaken at 300 rpm. Headspace samples were taken at intervals for up to 24 h and analyzed for COS, H₂S, and CS₂ as described previously (29). The M. marinum strain M cultures were incubated with CS₂ statically and at RT. To make sure this strain could adapt to CS₂ conversion, fresh medium was added to the grown culture in addition to CS₂, and CS₂ conversion was followed for 24 h.

Genome sequencing.

Genomic DNA from A. thiooxidans strains G8 and S1p was isolated (30) and sequenced using next-generation sequencing. Strain G8 was sequenced by a combination of 454 titanium technology (31) (about 50 Mb in 450-nucleotide [nt] reads) and Illumina technology (32) (about 1.5 Gb in 75-nt reads) at the genome sequencing facilities of the departments of Human Genetics and Molecular Biology (Radboud University Nijmegen). The reads of both methods were combined using CLC bio software (Aarhus, Denmark) and assembled in about 200 contigs of >1,000 nt, including several large genome fragments of 100 kb. A. thiooxidans strain S1p was sequenced by Illumina sequencing only, resulting in a 1.7-Gb sequence in 75-nt reads. Sequence reads were trimmed to remove low-quality reads and ambiguous nucleotides, selecting those with a minimum read length of 30 nt. The resulting sequences (total, 0.8 Gb) were used for de novo assembly, yielding 265 contigs with an average length of 11.758 nt. Putative G8 and S1p CS₂ hydrolase genes were identified in both assemblies using the sequences of the CS₂ hydrolases from Acidianus strain A1-3 (GenBank accession number HM805096) and Sulfolobus solfataricus P2 (GenBank accession number AE006641.1) in BLASTP and motif searches.

MALDI-TOF MS.

Purified CS₂ hydrolase from pieces excised from SDS-PAGE gels was extracted, digested with trypsin, and analyzed using matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) as described previously (33).

Phylogenetic analysis.

The carbonic anhydrase(-like) protein sequences most homologous to the CS₂ hydrolase of Acidianus strain A1-3 and A. thiooxidans strains G8 and S1p were retrieved via Blast. They were aligned using MUSCLE (34), and phylogenetic analysis was performed in MEGA 5.05 (35).

Stopped-flow spectrometry.

Carbonic anhydrase activity was measured by stopped-flow spectrometry (21). Per assay, up to 30 μg purified CS₂ hydrolase was used. The positive control consisted of up to 63 ng bovine carbonic anhydrase (Sigma) per assay.

Analytical ultracentrifugation.

The oligomeric state of the CS₂ hydrolases in solution was investigated by analytical ultracentrifugation at A₂₈₀^{1 cm} of 0.2 (S1p) and 0.67 (G8) as described in reference 21.

Structure modeling.

The hexadecameric structure of the Acidianus A1-3 CS₂ hydrolase (Protein Data Bank accession number 3TEO) was used as a template to form the hexadecameric Zn-containing CS₂ hydrolase from 3TEN (which is an octamer). The YASARA & WHAT IF Twinset (36) was used for homology modeling and subsequent analysis of the A. thiooxidans G8 and S1p CS₂ hydrolases. Sequence identity between the models and the Acidianus A1-3 CS₂ hydrolase template was 49% for A. thiooxidans G8 CS₂ hydrolase and 46% for the CS₂ hydrolase from strain S1p. Access routes from outside the enzymes to the active sites were calculated using MOLE 2.0 online (37).

Nucleotide sequence accession numbers.

The sequences of the bacterial CS2 hydrolases were deposited in the GenBank database under accession numbers KC902814 and KC902815.

RESULTS

Purification of the CS₂ hydrolases from A. thiooxidans strains G8 and S1p.

From a variety of sulfur-rich and highly acidic ecosystems, we isolated five new extremely acidophilic CS₂-converting A. thiooxidans strains (2Bp, Sts 4-3, S1p, G8, and BBW1). SDS-PAGE analysis of the soluble proteins in cell extracts from cells grown in chemostats with CS₂ as the sole energy source indicated that in all strains there was one hugely dominant protein present, of approximately 24 kDa (Fig. 1A). This size corresponds to the size of the monomeric CS₂ hydrolase from Acidianus A1-3, which is also present in high abundance in CS₂-grown cells (21). Therefore, we postulated that these highly abundant proteins in the A. thiooxidans extracts might be the bacterial CS₂ hydrolases.

Fig 1.

Purification of CS₂ hydrolases from CS₂-converting Acidithiobacillus thiooxidans strains. (A) SDS-PAGE (12% polyacrylamide) of cell extracts from CS₂-converting Acidithiobacillus thiooxidans strains 2Bp (lane 1), Sts 4-3 (lane 2), S1p (lane 3), G8 (lane 4), and BBW1 (lane 5) grown in minichemostats, with 10 nmol · ml⁻¹ CS₂ as the energy source and at a D of 0.02. (B) Cell extract from A. thiooxidans strains G8 (lane 1) and S1p (lane 3), and purified CS₂ hydrolases of strains G8 (lane 2) and S1p (lane 4). Ten micrograms of protein was loaded per lane.

The highly abundant protein of strain S1p was approximately 1 kDa higher in mass than those of the other 4 strains as well as the Acidianus A1-3 CS₂ hydrolase. To investigate potential differences between the CS₂ hydrolases, we purified the CS₂ hydrolase from both strain G8 and strain S1p by ammonium sulfate precipitation followed by HIC and anion-exchange chromatography (Fig. 1B). The purified CS₂ hydrolases both corresponded to the highly abundant protein present in the A. thiooxidans cell extracts.

Identification of the CS₂ hydrolase genes of A. thiooxidans strains G8 and S1p.

In order to identify the genes encoding the purified CS₂ hydrolase enzymes, genomic DNA of A. thiooxidans strain G8 and S1p was sequenced by next-generation sequencing platforms and assembled. Using the sequences of the Acidianus A1-3 and S. solfataricus P2 CS₂ hydrolase genes, homologues were identified in the assemblies of both A. thiooxidans strains and aligned with the most homologous genes identified in the NCBI databases. The putative G8 CS₂ hydrolase (predicted mass, 22.6 kDa) had 49% amino acid identity and the S1p CS₂ hydrolase (predicted mass, 24.3 kDa) 46% amino acid identity to the Acidianus A1-3 enzyme, and they had 50% amino acid identity with each other. The active-site residues that are characteristic of β-CAs are perfectly conserved (Fig. 2). The translated S1p gene was 18 amino acids longer than the translated G8 gene and 13 amino acids longer than the Acidianus A1-3 CS₂ hydrolase, corresponding to the larger mass of the S1p enzyme observed on SDS-PAGE (see above). MALDI-TOF MS confirmed that the purified CS₂ hydrolases were encoded by the identified Acidianus A1-3 CS₂ hydrolase homologous genes: for the purified CS₂ hydrolase of strain G8, peptides from the trypsin digest covered 60% of the identified (translated) gene. For strain S1p, the coverage was 90%.

Fig 2.

Amino acid alignment of CS₂ hydrolases from Acidianus strain A1-3, Sulfolobus islandicus strains HEV 10/4 and LS 2.15, S. solfataricus P2, A. thiooxidans strain G8, Acidithiobacillus caldus strain SM-1, and A. thiooxidans strain S1p. Underlining indicates residues forming the catalytic center of the enzymes. Boldface indicates perfectly conserved amino acids. Gray box, FF motif.

Phylogenetic analysis of CS₂ hydrolases and homologous enzymes.

The newly identified CS₂ hydrolases cluster within a group of β-CA homologues that contain the two confirmed CS₂ hydrolases from Acidianus strain A1-3 and S. solfataricus P2. This group is divided into archaeal and bacterial enzymes, with three cyanobacterial exceptions: three enzymes, from Calothrix sp., Cyanothece sp., and Nostoc punctiforme, fall in the archaeal instead of the bacterial group (Fig. 3). Previously, we identified two phenylalanine residues, F77 and F78, in the Acidianus A1-3 sequence that were present in the other confirmed CS₂ hydrolases. These residues form part of a long hydrophobic tunnel, the only access to and from the active center. Amino acid F78 was shown to be crucial for activity of the CS₂ hydrolase of Acidianus A1-3 (21). As the F77 and F78 residues were not conserved in β-CA homologues outside the cluster that contained the CS₂ hydrolases, we proposed that they may be signature residues for CS₂ hydrolase enzymes. In support of this hypothesis, this FF motif was also conserved in the G8 and S1p CS₂ hydrolases. In fact, in all the enzymes of both the bacterial and the archaeal branches of the putative CS₂ hydrolase cluster within the β-CAs, the FF motif is conserved. There are two exceptions: the enzyme from Nostoc punctiforme has a VF motif, and that of Haloquadratum walsbyi DSM16790 has a YF motif instead, but the F78 residue, important for activity in the Acidianus A1-3 enzyme, is still conserved.

Fig 3.

Phylogenetic analysis of putative and confirmed CS₂ hydrolases. Amino acid sequences were retrieved from GenBank and through BLAST and aligned using the MUSCLE aligner as implemented in Mega5.0. All putative CS₂ hydrolases are β-CA homologues and are usually annotated as such in GenBank. The evolutionary history was inferred by using the maximum likelihood method based on the Dayhoff matrix-based model. The tree with the highest log likelihood (−3645.9759) is shown. The percentage of trees (500 replicates) in which the associated taxa clustered together is shown next to the branches. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. All positions containing gaps and missing data were eliminated. There were a total of 183 positions in the final data set. A phylogenetic tree based on DNA sequences showed comparable clustering. Underlining indicates putative CS₂ hydrolases from strains with confirmed CS₂ conversion activity. Boldface indicates confirmed CS₂ hydrolases.

CS₂ conversion by microorganisms present in the CS₂ hydrolase phylogenetic tree.

Three strains present in the putative CS₂ hydrolase cluster were grown and tested for CS₂ conversion by adding 10 nmol · ml⁻¹ CS₂ gas to the headspace of a 10-ml grown culture. Of the three strains, Methylomicrobium alcaliphilum 20Z (DSM19304) converted CS₂ immediately to H₂S with COS as an intermediate. After 1 h, all CS₂ was depleted. Although CS₂ conversion was immediate and effective, we did not observe growth of M. alcaliphilum on CS₂ instead of CH₄, even after incubation periods of more than 1 month. Halopiger xanaduensis SH-6 (DSM18323), on the other hand, was not able to convert CS₂ over a period of at least 3 h. A small, nontransient increase in COS was seen over time, but this occurred also in cultures that had been inhibited or killed by heat treatment or addition of sodium azide or potassium cyanide, as well as in Escherichia coli cultures, and did not significantly reduce the concentration of CS₂ in the headspace. Mycobacterium marinum strain M was also tested for CS₂ conversion activity. However, under the conditions tested, this strain did not show CS₂ hydrolase activity, not even after 24 h of incubation of a growing culture with CS₂.

Substrate specificity of purified CS₂ hydrolases from A. thiooxidans strains G8 and S1p.

As the CS₂ hydrolase enzymes are highly homologous to the β-CAs, it is possible that β-CAs may be able to hydrate CS₂ as well as CO₂. Although the possibility of this conversion was predicted quantum chemically (24–26), it has never been experimentally confirmed. In addition, we tested potential CS₂ hydrolase activity of bovine erythrocyte α-CA, the γ-CAs Cam from Methanosarcina thermophila (38), Cab from Methanobacterium thermoautotrophicum (39), and Cam from Pelobacter carbinolicus (kindly made available by R. Siva Sai Kumar and James G. Ferry) and the β-CA from Streptococcus pneumoniae (40, 41), but none of these could convert CS₂. Reversely, CS₂ hydrolases may be able to hydrate CO₂ instead of CS₂. We have already shown that for the CS₂ hydrolase from Acidianus strain A1-3 this was not the case (21). Here, we tested the substrate specificity of the A. thiooxidans CS₂ hydrolases by stopped-flow spectrophotometry. Like the Acidianus CS₂ hydrolase, the A. thiooxidans enzymes could not convert CO₂ to HCO₃⁻ (Fig. 4).

Fig 4.

Stopped-flow spectrophotometry to measure carbonic anhydrase activity of purified CS₂ hydrolases from A. thiooxidans strains G8 and S1p compared with carbonic anhydrase from bovine erythrocytes. Symbols: open circles, strain G8; open triangles, strain S1p; solid squares, bovine carbonic anhydrase; solid line, chemical control.

The newly purified CS₂ hydrolases did convert CS₂ and COS to H₂S (Table 1). With CS₂ as the substrate, the A. thiooxidans G8 CS₂ hydrolase had a nearly 10-fold-higher catalytic efficiency (K_cat/K_m) than the Acidianus A1-3 and A. thiooxidans S1p CS₂ hydrolases. This was due to both a higher V_max (131 versus 32 nmol product · min⁻¹ · μg enzyme) as well as a higher affinity (lower K_m of 46 versus 93 μM CS₂). The difference in catalytic efficiency between the strain G8 CS₂ hydrolase and the strain S1p CS₂ hydrolase was even more pronounced with COS as the substrate, due to the much higher affinity of the strain G8 CS₂ hydrolase for COS.

Table 1.

Comparison of kinetic constants^a of the conversion of CS₂ (to H₂S and COS) and COS (to H₂S) by purified CS₂ hydrolases from Acidithiobacillus thiooxidans strains S1p and G8 with those of Acidianus A1-3 CS₂ hydrolase

Strain	CS2			COS
	Km	Vmax	Kcat/Km	Km	Vmax	Kcat/Km
A. thiooxidans S1p	93 ± 21	32 ± 2	8.4 × 106	74 ± 11	34 ± 2	1.1 × 107
A. thiooxidans G8	46 ± 6	131 ± 4	6.4 × 107	14 ± 2	97 ± 3	1.6 × 108
Acidianus A1-3b	130 ± 3	40 ± 0	7.3 × 106	22 ± 3	74 ± 3	8.0 × 107

^aK_m values are expressed in μM substrate in the buffer ± standard error of the mean (SEM); V_max is given in nmol product · min⁻¹ · μg enzyme ± SEM. K_m, V_max, and K_cat/K_m values (in S⁻¹ · M⁻¹) were determined by nonlinear regression using GraphPad Prism 5.04 (n ≥ 9).

^bData from reference 21.

We compared the kinetic properties of the purified enzymes with those of cell extracts from the five new CS₂-converting A. thiooxidans strains growing in chemostats supplied with 10 nmol · ml⁻¹ CS₂. Under these conditions, these strains all produced large amounts of CS₂ hydrolase (Fig. 1). The K_m values varied between 81 μM for strain G8 and 130 μM CS₂ for strain 2Bp. The V_max of cell extracts from strain S1p was higher than those of cell extracts from the 4 other strains tested, which is in contrast to the results obtained with the purified CS₂ hydrolases. However, due to the lower affinity of the purified CS₂ hydrolase to CS₂, strain S1p would require larger amounts of CS₂ hydrolase when growing on the low CS₂ concentration supplied to the reactor to be able to gain as much energy for growth as strain G8. Indeed, strain S1p had relatively more CS₂ hydrolase present in cell extracts than strain G8 (Fig. 1).

Catenane formation in the A. thiooxidans CS₂ hydrolases.

To investigate whether the bacterial CS₂ hydrolases form the highly unusual catenanes observed for the Acidianus A1-3 CS₂ hydrolase (21), we compared cell extracts from five CS₂-converting A. thiooxidans strains and two purified CS₂ hydrolases with the Acidianus enzyme by native PAGE. Four of the five A. thiooxidans strains had CS₂ hydrolases of a size similar to that of the Acidianus 16-mer form of the native enzyme. This suggests that, like the archaeon Acidianus, these A. thiooxidans strains contain the interlocked, double-ring catenane form of the enzyme rather than the single octameric ring. However, strain S1p CS₂ hydrolase migrated faster than the other CS₂ hydrolases, both in cell extract and in purified form (Fig. 5), suggesting that the S1p CS₂ hydrolase may not form hexadecameric catenanes. To confirm this finding, purified CS₂ hydrolases from strains S1p and G8 were analyzed by analytical ultracentrifugation (AUC) (Fig. 6). For the CS₂ hydrolase of strain G8, a small peak with sedimentation coefficient at 8.4 S and a large peak at 13.8 S were identified, corresponding to an 8-mer and 16-mer, respectively. The CS₂ hydrolase of strain S1p yielded only one peak, at 7.8 S. From these data we conclude that the CS₂ hydrolases from A. thiooxidans strains G8, 2Bp, Sts 4-3, and BBW1 share the unique hexadecameric catenane structure with the CS₂ hydrolase from Acidianus A1-3. However, the A. thiooxidans S1p CS₂ hydrolase forms only the single octameric ring.

Fig 5.

Native PAGE (8% polyacrylamide) of three purified CS₂ hydrolases and of cell extracts from five CS₂-converting A. thiooxidans strains and of Acidianus A1-3, stained for protein (top) or CS₂ hydrolase activity (bottom). Ten micrograms of protein was loaded per lane. Lanes: M, marker; lane 1, cell extracts of strain 2Bp; lane 2, cell extracts of strain Sts 4-3; lane 3, cell extracts of strain S1p; lane 4, purified CS₂ hydrolase of strain S1p; lane 5, strain G8 cell extract; lane 6, strain G8 purified CS₂ hydrolase; lane 7, strain BBW1 cell extract; lane 8, Acidianus A1-3 cell extract; lane 9, Acidianus A1-3 purified CS₂ hydrolase. Arrows indicate the expected positions of 8-mer (about 192-kDa) and 16-mer (about 384-kDa) configurations.

Fig 6.

Analytical ultracentrifugation results for purified CS₂ hydrolases from A. thiooxidans strains S1p and G8, fitted as the concentration distribution (c) (arbitrary units) of sedimentation coefficients (s) in Svedberg units (S). For strain G8 (solid line), two species are observed, corresponding to octamers (at 8.3 S) and hexadecamers (13 S). For strain S1p (dashed line), only one peak was observed, corresponding to an octameric form of the enzyme.

Structure models of the A. thiooxidans CS₂ hydrolases.

To compare the structural homology between the three CS₂ hydrolases and to find a structural basis for the differences in oligomerization between the A. thiooxidans CS₂ hydrolases, we used YASARA to create structural models of the A. thiooxidans G8 and S1p enzymes using the hexadecameric Acidianus A1-3 CS₂ hydrolase as a template. The CS₂ hydrolases of both enzymes could be modeled as catenane hexadecamers (Fig. 7A, B, and C). The models show a highly conserved structural homology in the active sites of the enzymes (Fig. 8). In addition, the long and narrow hydrophobic tunnel that forms the only access to the active site in the Acidianus A1-3 CS₂ hydrolase is also predicted to be the only available access route in the modeled structures of A. thiooxidans strains G8 and S1p CS₂ hydrolases (Fig. 7D, E, and F). The residues that form these tunnels are partly conserved: the F77F78 motif critical for activity in the Acidianus A1-3 enzyme is perfectly conserved, and some of the residues at the outer entrance of the tunnel are still hydrophobic but smaller, potentially creating a slightly wider tunnel at those positions (Fig. 7G, H, and I). The tunnel in the Acidianus A1-3 enzyme is formed by residues from 3 monomers, with the third monomer contributing one isoleucine residue (I201) from the C-terminal extending arm. However, in both the A. thiooxidans strain G8 and the strain S1p CS₂ hydrolase models, the C-terminal domain of the third monomer is positioned away from the tunnel. As the C-terminal domain of the enzyme from strain G8 is shorter and that of the strain S1p enzyme is longer than the C terminus of the Acidianus A1-3 CS₂ hydrolase, the model may not be accurate in this position. It therefore remains to be seen if a third monomer contributes to the tunnel formation in the A. thiooxidans CS₂ hydrolase.

Fig 7.

Crystal structure and models of CS₂ hydrolases. (A) Resolved crystal structure of the CS₂ hydrolase from Acidianus strain A1-3 and structural models of the CS₂ hydrolases from A. thiooxidans strain G8 (B) and strain S1p (C), modeled using the hexadecameric structure of the Acidianus A1-3 CS₂ hydrolase as a template. The two octameric interlocking rings forming the hexadecamer are shown in different colors. (D, E, F) Details of octamers showing the narrow hydrophobic entrance tunnels (pink) to the Zn-containing active site in the Acidianus CS₂ hydrolase (D) and in the models of the G8 (E) and S1p (F) CS₂ hydrolases. The Zn atoms are indicated by pink spheres. (G, H, I) Views of the tunnels looking in from the entrances to the active sites, showing the three monomers (two shades of gray and blue) and the residues (red) forming the tunnel in the Acidianus CS₂ hydrolase (G) and in the models of the G8 hydrolases (blue, cyan, and green monomers) (H) and the S1p CS₂ hydrolases (yellow, orange, and red monomers) (I).

Fig 8.

Superposition of active-site residues of the resolved structure of the Acidianus A1-3 CS₂ hydrolase (gray) and the models of the A. thiooxidans G8 (blue) and S1p CS₂ hydrolases (red). The Zn residue is shown as a green sphere.

From the structure of the Acidianus A1-3 CS₂ hydrolase as well as in the models of the A. thiooxidans strain G8 and strain S1p enzymes, it appears that the N-terminal α-helical extending arms of the monomers are responsible for most of the interaction between the 2 rings in the catenane structure (Fig. 9). In particular, in the Acidianus structure the residue K22 just beyond the N-terminal helix of one octamer is in close proximity to the Q49 and P50 on a small 4-residue loop (residues 49 to 52, QPDD) of the other octamer. In the model of the A. thiooxidans G8 CS₂ hydrolase, this small loop is one residue larger (residues 50 to 55, RGDRD) but points in a different direction from that of the Acidianus A1-3 loop. Although the CS₂ hydrolase of A. thiooxidans S1p could also be modeled as a catenane double ring, there is a much larger loop corresponding to the sequence insertion at residues 56 to 61, TPAGGG, that is not present in the Acidianus A1-3 or A. thiooxidans G8 CS₂ hydrolase (Fig. 2). The A. thiooxidans S1p CS₂ hydrolase has a second 2-residue insertion (26P-27D [Fig. 2]) that is part of a loop in the other octamer (residues 21 to 28, RRRYGPD [Fig. 9]). This loop protrudes into the space close to where the TPAGGG loop is positioned. It is therefore possible that these insertions in the A. thiooxidans S1p CS₂ hydrolase protein are interfering with catenane formation.

Fig 9.

Structure detail of the proximity of two octamers in the hexadecameric form of the CS₂ hydrolase, shown as an overlay of the models of the A. thiooxidans strains G8 (two shades of blue, one for each octamer) and S1p (red and ochre) onto the Acidianus strain A1-3 resolved structure (two shades of gray). In the Acidianus CS₂ hydrolase, the smallest distance between the 2 octamers is between a small loop of 4 residues between Gln49 and Asp52 (QPPD) and the Lys22 on the other octamer (residues in gray). In the model of the G8 CS₂ hydrolase, the G8 loop is 6 residues long (RGDRGD, in blue). In the model of the S1p, the loop is 10 residues long (HVDTPAGGGD, in ochre). The S1p enzyme is modeled to have an additional loop in the second octamer (RRRYGPD).

In addition, the C-terminal tail of the A. thiooxidans S1p CS₂ hydrolase is nine amino acids longer and its sequence differs from that of the Acidianus A1-3 CS₂ hydrolase C-terminal tail, making it difficult to model (Fig. 7C). The C-terminal tails of the Acidianus A1-3 CS₂ hydrolase as well as the A. thiooxidans G8 modeled CS₂ hydrolase contain several residues that make stabilizing interactions with the core of the structure, where potential interactions between the 2 octamers take place, and therefore it seems likely that the C-terminal tail is not very flexible. Whether the C-terminal tail of the A. thiooxidans S1p CS₂ hydrolase adopts a similar conformation remains unclear. Because of the differences in length and sequence, it is possible that this tail can also interfere with hexadecameric catenane formation.

DISCUSSION

In this study, we identified the bacterial CS₂ hydrolase enzymes from five A. thiooxidans strains, purified the CS₂ hydrolases from A. thiooxidans strains G8 and S1p, and identified the encoding genes. The enzymes were highly homologous to each other and to the CS₂ hydrolase from Acidianus strain A1-3, both on DNA and on protein levels. All putative CS₂ hydrolases enzymes are phylogenetically most related to a group of β-CAs that contain mostly bacterial rather than archaeal enzymes (21). This suggests that the gene evolved in bacteria and spread via lateral gene transfer to archaeal species. Indeed, two of the putative CS₂ hydrolases were plasmid encoded. Also, transposon-related sequences surrounding the (putative) CS₂ hydrolase were identified in the available genome sequences of both archaeal species (S. solfataricus P2, Sulfolobus islandicus strains L.S.2.15 and HVE10/4, and Haloquadratum walsbyi) and bacterial species (Nitrosomonas sp. AL212 and Mycobacterium marinum M). Acidithiobacillus and the archaeal Sulfolobus and Acidianus species occupy the same sulfur-rich acidic environments such as hot springs and solfataras (42, 43). Gene transfer between these species from the two domains of life is possible and likely (44–46).

In addition to the acidophilic species that are represented in the (putative) CS₂ hydrolase phylogenetic tree, most other species were isolated from saline and/or alkaline environments. Saline marshes and estuaries are rich sources of CS₂ (47, 48). Also, the haloalkaline soda lakes harbor an active sulfur cycle with a high biodiversity of sulfur-oxidizing bacteria (SOB) that respire mainly on sulfide, thiosulfate (S₂O₃²⁻), and polysulfide (49). Thioalkalivibrio paradoxus, isolated from a Kenyan soda lake, was shown to be able to respire on CS₂ when grown on thiocyanate (SCN⁻) (50). We could not find a CS₂ hydrolase homologue in the available genome sequence of this species (which at the time of writing is still wrongly labeled as the Thioalkalivibrio thiocyanoxidans genome [D. Sorokin, personal communication]). However, the Thioalkalivibrio paradoxus genome does contain a homologue of the recently described COS hydrolase from T. thioparus strain THI 115, one of only three close homologues currently in the databases. T. thioparus strain THI 115 converts SCN⁻ to COS and NH₃. COS is subsequently converted to H₂S by COS hydrolase. This enzyme can also convert CS₂ at a lower rate (27). It is possible that the CS₂ hydrolase activity observed for Thioalkalivibrio paradoxus is due to activity of the COS hydrolase homologue. Interestingly, the closely related Thioalkalivibrio nitratireducens, isolated from the Egyptian soda lake Lake Fazda (51), contains both a CS₂ hydrolase and a close COS hydrolase homologue. Thioalkalivibrio paradoxus and Thioalkalivibrio nitratireducens both convert SCN⁻, but they do not produce COS or CS₂ as intermediates (D. Sorokin, personal communication). Therefore, the function of the CS₂ hydrolase homologues in these strains remains unclear at present.

Of the three species that were tested for CS₂ conversion, only M. alcaliphilum was able to convert CS₂. This methanotroph was isolated from the highly alkaline soda lake Shara-Nur, in Central Asia (52, 53). CS₂ conversion by methane-grown cultures was immediate, suggesting that the enzyme was not induced but already present in the cell. Why this organism requires the presence of CS₂ hydrolase activity is not clear at present. However, another Methylomicrobium species, M. kenyense strain AMO1 isolated from a Kenyan soda lake, was also found to have CS₂ conversion activity (54). These authors suggested that CS₂ could be a suicide substrate for methane monooxygenase (MMO), the enzyme that converts methane to methanol as the first step in the energy-generating pathway of methanotrophs. MMO is mechanistically and evolutionary homologous to the ammonium monoxygenase (AMO) from nitrifiers. Their substrates are often interchangeable, and they share alternative substrates and inhibitors (55). The nitrifier Nitrosomonas europaea was shown to have CS₂ conversion activity, but the mechanism of CS₂ conversion was not investigated (56). CS₂ is one of the oldest known inhibitors of nitrification (57). The inhibitory action has been attributed to the CS₂ molecule forming a complex with a nucleophilic amino acid close to the active center of AMO, thereby chelating the Cu cofactor from the active center (58, 59). If MMO is similarly inhibited by CS₂, then both methanotrophs and nitrifiers would benefit from an enzyme capable of CS₂ removal. Indeed, in addition to the methanotroph M. alcaliphilum, two Nitrosomonas strains and one Nitrosococcus strain were found to contain CS₂ hydrolase homologues in their genomes (Fig. 3). We therefore propose that these CS₂ hydrolases could function as a detoxification mechanism for environmental CS₂ in nitrifiers and methanotrophs.

COS is an ecologically relevant alternative substrate for CS₂ hydrolase, as it is the most abundant sulfur species in the atmosphere. The major sinks for COS are vegetation and soil (2). Biological conversion of COS by plants via their carbonic anhydrases has been described (60, 61), and COS-degrading soil microorganisms, including 4 Mycobacterium spp., were readily isolated from Japanese soils (62). It is therefore possible that (some of) the CS₂ hydrolase homologues identified in the phylogenetic tree convert mainly COS rather than CS₂ in their natural environments. Interestingly, neither the two CS₂ hydrolases from A. thiooxidans strains S1p and G8 nor the archaeal CS₂ hydrolase (21) or the T. thioparus COS hydrolase (27) converted CO₂. This suggests that their function evolved specifically for CS₂ and COS conversion and that they are not just broad-specificity CAs.

All (putative) CS₂ hydrolase proteins presented in Fig. 3 but two contain the conserved FF residues that are crucial in forming the hydrophobic access tunnel to the active center. As they are not conserved in CAs or in the T. thioparus COS hydrolase, the FF residues were proposed to be a CS₂ hydrolase-specific motif (21). However, H. xanaduensis and M. marinum, which have CS₂ hydrolase homologues containing the FF motif, do not convert CS₂ under the conditions that we tested. It is possible that the amino acids surrounding the FF motif are also important for CS₂ hydrolase activity: in all confirmed bacterial and archaeal CS₂ hydrolases as well as in M. alcaliphilum, the sequence stretch that is perfectly conserved is NFFGT, but in H. xanaduensis the sequence is NFFDT. In the structural model of the Acidianus CS₂ hydrolase, the N76 and G79T80 residues are positioned away from the hydrophobic tunnel that the FF residues are a part of, but it is possible that the larger D79 residue causes a conformational change that affects the positioning of the crucial F78 residue. Although both the CS₂ hydrolase homologues present in M. marinum strain M contain the NFFGT motif, they are also the only two CS₂ hydrolase homologues with an extended C terminus, 56 amino acids longer than that of the CS₂ hydrolase from A. thiooxidans strain S1p. This may have profound effects on oligomerization and therefore enzyme activity (see below).

The Acidianus A1-3 CS₂ hydrolase exists as a 16-mer interlocked ring (catenane) structure. Catenanes are extremely rare in biology. We are aware of only 3 reports: the gp5 capsid protein of bacteriophage HK97 (63), the bovine mitochondrial peroxiredoxin III (64), and a Pyrobaculum aerophilum citrate synthase in which the catenanes are formed by disulfide bridging between the N- and C-terminal ends of each of the homodimer chains (65). The molecular mechanism of the catenane assembly in the Acidianus A1-3 enzyme is not understood. The N- and C-terminal domains of the monomers, which are crucial for interlinking dimers into octamers (21), also reside in the field of interaction between the 2 octameric rings (Fig. 9). Interestingly, the bacterial CS₂ hydrolases of four of the five isolated A. thiooxidans strains also existed in the 16-mer catenane form. The reason why these enzymes adopt a catenane conformation is unclear at present. However, cells growing on low concentrations of CS₂ produce vast amounts of enzyme, probably due to the poor affinity of the CS₂ hydrolase to CS₂. This results in extremely high intracellular concentrations with very dense packing of the enzyme, similar to what has been reported for the most abundant enzyme on earth, RuBisCo, that can be expressed to about 40% of total protein in cells and is packed in cell structures called carboxysomes (66). Dense packing of CS₂ hydrolases can be obtained both with octameric rings and with 16-mer catenanes, but the exclusion volume is a little bit smaller for catenanes than for single rings. Also, highly densely packed Acidianus strain A1-3 CS₂ hydrolase in crystals consisted of pure 16-mer catenanes (21).

The A. thiooxidans S1p enzyme was found only in the 8-mer form, both by AUC and by native PAGE analysis. The S1p CS₂ hydrolase monomer is 11 amino acids longer at the C terminus than the monomer from strain G8 and 9 amino acids longer than the monomer from Acidianus A1-3. In addition, the S1p enzyme has an insertion at residues 56 to 61 that the other two CS₂ hydrolases lack. Modeling the G8 and S1p enzymes on the Acidianus A1-3 CS₂ hydrolase structure indicated that in the S1p CS₂ hydrolase, both the elongated C-terminal tail and the insertion may interfere with the formation of the 16-mer catenane by filling the space where the other octamer should reside. It is possible that the N- and C-terminal areas of the Acidianus A1-3 and G8 CS₂ hydrolases, which are situated in the interaction field between the two octameric rings, may play a role in the formation of the rare catenane structure. Whether this is indeed the case is the subject of future investigations.