Biochemistry and molecular biology of lithotrophic sulfur oxidation by taxonomically and ecologically diverse bacteria and archaea


  • Wriddhiman Ghosh,

    1. Department of Microbiology, University of Burdwan, West Bengal, India
    2. Department of Microbiology, Bose Institute, Kolkata, India
    Search for more papers by this author
  • Bomba Dam

    1. Department of Microbiology, Bose Institute, Kolkata, India
    2. Microbiology Laboratory, Department of Botany, Institute of Sciences, Visva-Bharati, Santiniketan, West Bengal, India
    Search for more papers by this author

  • Editor: Michael Galperin

Correspondence: Wriddhiman Ghosh, Department of Microbiology, Bose Institute, P-1/12, C. I. T. Scheme VII- M, Kolkata 700 054, India. Tel.: +91 33 23 55 94 16; fax: +91 33 23 55 38 86; e-mail:


Lithotrophic sulfur oxidation is an ancient metabolic process. Ecologically and taxonomically diverged prokaryotes have differential abilities to utilize different reduced sulfur compounds as lithotrophic substrates. Different phototrophic or chemotrophic species use different enzymes, pathways and mechanisms of electron transport and energy conservation for the oxidation of any given substrate. While the mechanisms of sulfur oxidation in obligately chemolithotrophic bacteria, predominantly belonging to Beta- (e.g. Thiobacillus) and Gammaproteobacteria (e.g. Thiomicrospira), are not well established, the Sox system is the central pathway in the facultative bacteria from Alphaproteobacteria (e.g. Paracoccus). Interestingly, photolithotrophs such as Rhodovulum belonging to Alphaproteobacteria also use the Sox system, whereas those from Chromatiaceae and Chlorobi use a truncated Sox complex alongside reverse-acting sulfate-reducing systems. Certain chemotrophic magnetotactic Alphaproteobacteria allegedly utilize such a combined mechanism. Sulfur-chemolithotrophic metabolism in Archaea, largely restricted to Sulfolobales, is distinct from those in Bacteria. Phylogenetic and biomolecular fossil data suggest that the ubiquity of sox genes could be due to horizontal transfer, and coupled sulfate reduction/sulfide oxidation pathways, originating in planktonic ancestors of Chromatiaceae or Chlorobi, could be ancestral to all sulfur-lithotrophic processes. However, the possibility that chemolithotrophy, originating in deep sea, is the actual ancestral form of sulfur oxidation cannot be ruled out.


A variety of redox reactions involving inorganic as well as organic compounds of sulfur drive the global cycling of this element. The broad range of oxidation states of sulfur (−2 to +6) has given rise to a wide variety of redox enzymes that transform its different compounds. Among the numerous biochemical reactions of the sulfur cycle, only some prokaryotic oxidative processes involving inorganic sulfur compounds (referred to as lithotrophic sulfur oxidation) can harness the generated reductants (such as NADPH) and ATP molecules to respiration and/or CO2 assimilation. Aerobic or anaerobic (often denitrifying) chemolithoautotrophic bacteria and archaea (Kelly, 1989; Kelly et al., 1997; Kelly & Wood, 2000a) use the reducing equivalents for both respiration and CO2 fixation, while anaerobic photolithoautotrophic bacteria use reductants mainly for CO2 fixation (Trüper & Fischer, 1982; Brune, 1989, 1995a; Imhoff, 1995, 2001a, b, c). To meet their carbon requirement, sulfur-oxidizing photo- or chemoautotrophic prokaryotes fix CO2 by various biochemical pathways (Thauer, 2007), whereas methylated sulfur compounds-oxidizing methylotrophic bacteria fix formaldehyde by more than one mechanism (de Zwart et al., 1997; Borodina et al., 2002).

Ecologically and taxonomically diverse sulfur-lithotrophic bacteria (Timmer-Ten Hoor, 1975; Jannasch et al., 1985; Bryantseva et al., 1999; Huber & Stetter, 1999; Imhoff, 2001d; Overmann, 2001; Sorokin et al., 2001a, 2003, 2004, 2007; Campbell et al., 2003, 2006; Ghosh & Roy, 2006a, b) and archaea (Golovacheva et al., 1985; Stetter et al., 1990; Karavaiko et al., 1994; Fuchs et al., 1996; Huber & Stetter, 2001; Reysenbach, 2001) oxidize a wide range of reduced inorganic sulfur species by discrete biochemical and molecular mechanisms. A clear biochemical multiplicity exists among diverged sulfur lithotrophs regarding their ability to oxidize different reduced sulfur compounds, and distinct enzymes, pathways and mechanisms of electron transport and energy conservation are also involved in the oxidation of the same sulfur compound by different phototrophic or chemotrophic species (Kelly, 1989; Brune, 1995a; Kelly et al., 1997; Friedrich, 1998; Friedrich et al., 2001, 2005; Kletzin et al., 2004; Dahl, 2008; Kletzin, 2008).

The biotic and abiotic chemical reactions involved in the oxidative transformation of inorganic sulfur compounds are well documented (Suzuki, 1999). The physiology and biochemistry of sulfur oxidation by phototrophic (Trüper & Fischer, 1982; Brune, 1989, 1995a) or chemotrophic (Kelly, 1982, 1989; Pronk et al., 1990; Takakuwa, 1992; Kelly et al., 1997; Friedrich, 1998) prokaryotes have also been extensively reviewed. These seminal works in combination with more recent discussions on the molecular genetics of the sulfur-lithotrophic processes of bacteria (Friedrich et al., 2001, 2005; Dahl, 2008; Grimm et al., 2008) and archaea (Kletzin et al., 2004; Kletzin, 2008) have shaped our overall understanding of the different mechanisms of microbial sulfur oxidation. The aim of the current review is to update this information and present an integrated molecular perspective of the common and divergent features of the sulfur oxidation substrates and systems encountered in ecologically and taxonomically diverse prokaryotes. In the course of our discussion we also touch upon issues related to the origin, evolution and interactions of the different molecular mechanisms of lithotrophic sulfur oxidation. These aspects of sulfur lithotrophy are central to our comprehension of the earliest metabolic strategies of life on Earth. The full range of capabilities and diversities of extant sulfur lithotrophs, depicted here in its maximum possible totality, provides a framework for future investigations on the nature of primordial metabolism.

Prokaryotic sulfur oxidation: photolithotrophy vis-à-vis chemolithotrophy

In nature, photolithotrophic bacteria, and chemolithotrophic bacteria and archaea, work in tandem to carry out the oxidative half of the sulfur cycle (Brüser et al., 2000a). Anaerobic photolithotrophic bacteria extensively oxidize H2S and other reduced sulfur compounds produced in anoxic water layers and sediments such as those found in thermally stratified meromictic lakes and brackish lagoons (Kämpf & Pfennig, 1980; Pfennig & Widdel, 1982; van Gemerden & Mas, 1995; Imhoff, 2001d). Sulfur-chemolithotrophic microorganisms, on the other hand, occupy habitats ranging from activated sludges and anaerobic digesters (Robertson & Kuenen, 1983; Katayama et al., 1995) to wastewater biofilms and water treatment plants having microaerophilic conditions (Howarth et al., 1999; Ito et al., 2004), from bacterial sulfur mats (Dubinina et al., 2004) to intertidal mud flats (Brinkhoff et al., 1999) and from crop lands, orchards and garden soils (Graff & Stubner, 2003; Deb et al., 2004; Ghosh et al., 2005) to plant rhizospheres (Ghosh & Roy, 2006b). Oxidations of inorganic sulfur compounds are some of the most important energy-yielding reactions for microorganisms living in critical environments such as acidic mine drainage (Harrison, 1984), hypersaline and/or hyperalkaline lakes (Sorokin et al., 2001b, 2006; Kompantseva et al., 2005), metal-rich marine geothermal fields and craters of volcanoes (Golovacheva et al., 1985; Huber & Stetter, 1989), volcanic hot springs, solfataras, subseafloor habitats such as deep-sea volcanic eruptions (Huber et al., 2003), submarine hydrothermal vents (Jannasch et al., 1985; Muyzer et al., 1995; Reysenbach et al., 2000b; Campbell et al., 2003, 2006; Takai et al., 2005, 2006), and even perennially ice-covered Antarctic lakes (Sattley & Madigan, 2006). Sulfur-chemolithoautotrophs act as primary producers in the food chains of deep-sea geothermal vent ecosystems completely cut off from light (Campbell et al., 2003, 2006). In such extreme ecosystems, sulfur-chemoautotrophic bacteria support quite large and diversified consumer communities and often occupy internal body parts of heterotrophic animals such as mussels and tubeworms (Campbell et al., 2003, 2006).

Photolithotrophic oxidation of reduced sulfur compounds is regarded as an ancient bacterial metabolism (Brocks et al., 2005). Anoxygenic photolithoautotrophy essentially involves a light-driven electron transport chain where excitation of the photosynthetic reaction center by appropriate wavelengths of light leads to cyclical movement of electrons from and to the reaction center via certain electron carriers specific for purple and green bacteria. NADH, the reductant for CO2 fixation, is produced via reverse electron transfer from the reduced carriers through redox enzymes, while the electrons thus removed from the photosynthetic electron transfer chain are eventually replaced by oxidation of an inorganic electron donor. Almost all anoxygenic photolithoautotrophic bacteria have at least limited abilities to use reduced sulfur species as electron source (Brune, 1989). On the other hand, chemolithoautotrophic growth via oxidation of sulfur compounds for both energy and electrons is achieved by taxonomically diverse nonphotosynthetic colorless sulfur bacteria and certain archaea belonging to the order Sulfolobales of Crenarchaeota (Kelly, 1989; Stetter et al., 1990; Fuchs et al., 1996; Huber & Stetter, 2001). The deepest phylogenetic branches of both Bacteria and Archaea include sulfur-chemolithoautotrophic extremophiles (Woese, 1987; Burggraf et al., 1992; Fuchs et al., 1996). This observation led to the postulation of sulfur chemolithotrophy as the earliest self-sustaining metabolism (Russell & Hall, 1997). Correspondingly, it is noteworthy that photolithotrophic sulfur oxidation has not yet been reported from any member of Archaea, or for that matter any other hyperthermophile. However, diverse groups of optimally adapted anoxygenic photolithotrophic bacteria thrive in moderately extreme temperature, pH or salinity conditions, and act as primary producers in such unusual habitats (Madigan, 2003). These organisms are ideal model systems for defining the physiochemical limits of photosynthesis and understanding the evolution of this process.

Anoxygenic photolithotrophy with reduced sulfur compounds as electron source

Anoxygenic photolithotrophic bacteria carry out light-dependent, bacteriochlorophyll-mediated, energy transfer processes with reduced sulfur compounds as electron donors (Imhoff, 1995). Such organisms mostly belong to the categories of ‘green sulfur bacteria’ (constituting a single coherent taxonomic group, viz. the phylum Chlorobi) (abbreviated henceforward as GSB) and ‘purple sulfur bacteria’ (members of the families Chromatiaceae and Ectothiorhodospiraceae within the class Gammaproteobacteria) (abbreviated henceforward as PSB). Some ‘purple nonsulfur bacteria’ (abbreviated henceforward as PNSB), distributed over the Alpha (e.g. species of Rhodospirillum, Rhodovulum, Rhodopseudomonas, etc.) as well as Beta (e.g. species of Rhodocyclus, Rhodoferax, etc.) classes of Proteobacteria, also follow similar metabolic strategies (Madigan, 1988; Madigan et al., 2000; Imhoff, 2001b, c). Besides these, three more types (the aerobic anoxygenic bacteriochlorophyll-containing bacteria, the obligately anaerobic anoxygenic phototrophic Heliobacteria, and the aerobic/anaerobic anoxygenic filamentous phototrophic green nonsulfur bacteria belonging to Chloroflexaceae) also have limited abilities to oxidize sulfur under a variety of metabolic conditions (Dahl, 2008). The physiology and biochemistry of sulfur oxidation by the last three groups of bacteria are very poorly understood and are not included in the present discussion. Sulfur-oxidizing capabilities of the representatives of the diverse groups of anoxygenic phototrophic bacteria are outlined in Table 1.

Table 1.   Sulfur-oxidizing capabilities of the representatives of the diverse groups of anoxygenic phototrophic bacteria
Taxonomic affiliationMetabolic featuresRepresentative
 ChlorobiObligately phototrophic; use S2−, S0 or S2O32− as electron donors for the reduction of CO2; S0 globules deposited outside the cells during oxidation of S2− and S2O32−; potentially mixotrophic, but can photoassimilate only a few simple organic substancesChlorobaculum tepidum
Chlorobaculum thiosulfatiphilum
Complex sulfur islands in the genome
APS reductase
ATP sulfurylase
Qmo complex
Vertéet al. (2002); Frigaard & Bryant (2008); Chan et al. (2008a, b)
 Gammaproteobacteria, ChromatiaceaeObligate or facultative photoautotrophs; members such as Rheinheimera spp. are nonphototrophic; all phototrophic members use S2− and S0 as photosynthetic electron donors; S0 globules deposited within the cells during oxidation of S2− (or S2O32−)Obligate phototrophs requiring strictly anoxic conditions. Generally unable to use S2O32−Allochromatium warmingii
Isochromatium buderi
Thiospirillum jenense
Imhoff et al. (1998a); Reinartz et al. (1998); Dahl (2008); Grimm et al. (2008); Sander & Dahl (2008)
Facultative phototrophs that can also grow chemo(litho-/organo-)(auto-/hetero-) trophically under reduced oxygen tension; capable of using many more sulfur substrates such as S2O32− or SO32−; photoassimilate a wide variety of organic substancesAllochromatium vinosumSoxXAYZB
APS reductase
ATP sulfurylase
 Gammaproteobacteria, Ectothiorhodospiraceae (Only members of Ectothiorhodospira, Halorhodospira and Thiorhodospira are phototrophic)S2− is oxidized by all the members. During S2− oxidation, S0 globules are usually deposited outside the cell, while polysulfides are produced under alkaline conditionsSome photoautotrophic species can also grow chemoautotrophically on sulfur compoundsEctothiorhodospira haloalkaliphila
Ectothiorhodospira shaposhnikovii
Kostanjevecki et al. (2000); Gorlenko et al. (2004); Dahl (2008); Sander & Dahl (2008)
Some spp. oxidize S2O32− to SO42−; S0 globules are sometimes deposited outside the cell during S2O32− oxidationEctothiorhodospira vacuolataSoxXAYZB
Halorhodospira halophilaSoxXAYZB + SoxEF
Dsr system
Few may be photoorganoheterotrophic, besides scarcely growing photolithoautotrophically with S2−Ectothiorhodosinus*mongolicum 
 AlphaproteobacteriaPhotoheterotrophy is always preferred under anaerobic conditionsMany purple nonsulfur bacteria can grow photolithoautotrophically with S2− and S2O32− as electron donorsRhodopseudomonas palustris
Rhodovulum sulfidophilum
Brune (1989), (1995a); Appia-Ayme et al. (2001); Imhoff (2001d)
 BetaproteobacteriaYet others can grow chemoorganotrophically or chemolithoautotrophically under aerobic or microaerobic conditionsRhodocyclus purpureus 
Aerobic anoxygenic bacteriochlorophyll-containing bacteria (no photolithoautotrophy with sulfur)
 All Alphaproteobacteria, except the betaproteobacterium Roseotales depolymeransPhotolithoheterotrophic growth oxidizing S2−, S2O32−, SO32− and S0 to SO42−Roseinatronobacter thiooxidansDahl (2008)
Heliobacteria (photoautotrophic growth not demonstrated in any species)
 The family Heliobacteriaceae phylogenetically related to the Gram-positive Bacillus and ClostridiumPhotoheterotrophic growth with few organic carbon sources +chemoheterotrophic growth in the dark by fermentation of pyruvate or lactate+S2− is oxidized to S0, which appears in the mediumHeliobacterium sulfidophilum
Heliobacterium undosum
Madigan (2001)
Filamentous anoxygenic phototrophs
 ChloroflexaceaePhotoheterotrophic or photoautotrophic growth with S2− as electron donor Chloroflexus aurantiacus
Oscillochloris trichoides
Hanada & Pierson (2002)

These phototrophic bacteria are incapable of using water as electron donor. Instead they require more reduced chemicals such as elemental sulfur and sulfide (and sometimes hydrogen, too) to serve this purpose. This kind of photosynthetic metabolism thus clearly differs from that encountered in cyanobacteria, algae and green plants (Madigan, 1988). Members of the anaerobic anoxygenic phototrophic Chlorobi, Chromatiaceae and Ectothiorhodospiraceae characteristically form intra- or extracellularly stored sulfur globules as an intermediate during the oxidation of thiosulfate or sulfide (Dahl, 1999; Dahl & Prange, 2006; Chan et al., 2008b), a property not shared by the PNSB. While the Chlorobi and Ectothiorhodospiraceae exhibit five to seven times higher affinities for sulfide than does Chromatiaceae (van Gemerden & Mas, 1995), Chlorobi and Chromatiaceae have equal affinities for polysulfides.

The predominantly aquatic and strictly anaerobic GSB fundamentally differ from the PSB in their strict phototrophic nature, and possession of distinctive bacteriochlorophylls and special light-harvesting structures (Overmann & Tuschak, 1997; Overmann, 2001). Although the GSB and PSB are basically similar in terms of their nutritional requirements, the former type, in contrast to the latter, can assimilate only a very few organic carbon compounds and can never grow chemotrophically with oxygen as the terminal electron acceptor (Overmann, 2001). In contrast, certain PSB such as Allochromatium vinosum (formerly known as Chromatium vinosum; Imhoff et al., 1998b) are capable of switching from anaerobic to aerobic sulfur oxidation and growing chemotrophically in the dark with thiosulfate under reduced oxygen tension (Kondratieva, 1989). While yet other species of Allochromatium (e.g. Allochromatium warmingii) do not have chemotrophic attributes, species belonging to Lamprobacter, Thiocystis, Thiohalocapsa, Thiolamprovum, Thiorhodovibrio and Halochromatium are all capable of chemoautotrophic growth (Dahl, 2008; Sander & Dahl, 2008). However, the GSB compensate for this apparent physiological inflexibility by virtue of their advanced light-harvesting potential, which allows them to occupy greater depths in stratified lakes far below the levels occupied by the PSB (Overmann, 2001).

Among anoxygenic phototrophic bacteria, the PSB and the green filamentous bacteria often appear in visible concentrations, forming colored microbial mats and blooms, in natural as well as anthropogenically influenced environments (Castenholz & Pierson, 1995; Madigan, 2003). Massive growth of phototrophic sulfur bacteria is also encountered in tidal seawater pools, hypersaline environments, stabilization lagoons and even geothermal hot springs. In contrast, phototrophic PNSB rarely emerge as massive formations in nature and are seldom the major constituents of phototrophic communities in colored microbial mats, although the metabolic diversity of these bacteria allows them to occupy a significantly wide range of environments (Madigan, 2003).

In nature, GSB populate planktonic habitats such as thermally stratified or meromictic lakes and brackish lagoons (van Gemerden & Mas, 1995), where light reaches anoxic water layers or sediments containing reduced sulfur compounds. Because of their inability to assimilatorily reduce sulfate, the GSB have a perpetual requirement of reduced sulfur sources; consequently, they often form stable syntrophic associations with sulfur- and sulfate-reducing bacteria (Pfennig & Widdel, 1982). Accordingly, closed sulfur cycles are often observed in ecosystems where sulfur or sulfate is reduced by chemotrophic components and the sulfide generated is reoxidized photosynthetically by the GSB (Overmann, 2001). The presence of GSB together with purple (sulfur as well as nonsulfur) bacteria has been demonstrated in diverse aquatic ecosystems, including perennially ice-covered meromictic Antarctic lakes (Burke & Burton, 1988; Karr et al., 2003). In the same way as PSB and GSB frequently coexist in aquatic habitats, their cocultures can also be grown with sulfide as the sole electron donor (Overmann & van Gemerden, 2000). Sulfide is oxidized by GSB to zero-valence sulfur, while polythionates are formed abiotically. As long as sulfide is present, the GSB cannot utilize polythionates, but the PSB can. Thus the higher sulfide affinity of the GSB is balanced by a similar fondness of the PSB for polysulfides (Overmann & van Gemerden, 2000; Overmann, 2001).

On the other hand, low dissolved oxygen tension and high availability of light and simple organic nutrients, as found in eutrophicated stagnant water bodies, promote the proliferation of PNSB (Okubo et al., 2006). Nevertheless, these organisms are also capable of operating successfully under highly aerated and light-limited conditions. As such, they occur in large numbers (forming colored microbial mats or biofilms) and play crucial roles in the sulfur cycle in wastewater treatment plants, swine waste lagoons and tidal seawater pools (Hiraishi et al., 1989; Hiraishi & Ueda, 1995; Do et al., 2003; Okubo et al., 2006).

Photolithotrophic GSB

Photolithotrophic sulfur-oxidizing GSB require light as the energy source and use hydrogen sulfide, elemental sulfur and sometimes thiosulfate as electron donors for the reduction of CO2 (Pfennig & Trüper, 1992). Whereas some Chlorobi strains can utilize H2 (Overmann, 2000) or Fe2+ (Heising et al., 1999) as electron donors, species such as Chlorobaculum tepidum (formerly designated as Chlorobium tepidum; Imhoff, 2003) can only use reduced sulfur compounds as exogenous reductants to drive all their anabolic pathways (Chan et al., 2008a, b). Although all the genera of Chlorobi are potentially mixotrophic, they can photoassimilate only a very few simple organic substances in the presence of sulfide or thiosulfate (Imhoff, 2003; Chan et al., 2008a, b).

Although the physiology and ecology of GSB have long been studied in detail, comparatively less information regarding the biochemistry and genetics of their sulfur oxidation processes was available until recently (Frigaard & Bryant, 2008). However, current advances in genomics- and proteomics-based approaches have considerably facilitated the understanding of the molecular mechanisms of anaerobic sulfur oxidation in the GSB. While the completely annotated genome of Chlorobaculum tepidum has been used to propose models of sulfur oxidation (Eisen et al., 2002; Hanson & Tabita, 2003; Chan et al., 2008a), these novel perspectives have also demonstrated that sulfur metabolism in the GSB may actually encompass greater diversity than that appreciated traditionally (Frigaard & Bryant, 2008). This implies that the apparently similar sulfur oxidation phenotypes of the different Chlorobi members are governed by dissimilar sets of enzymes, and hence should be considered as cases of convergent evolution (Frigaard & Bryant, 2008).

Hydrogen sulfide is used as electron donor by almost all the GSB except the iron-oxidizing Chlorobium ferrooxidans. Sulfide is oxidized to sulfuric acid via transient deposition of sulfur globules outside the cells, with most of the strains oxidizing elemental sulfur after sulfide has been depleted (Brune, 1995a). A constitutive membrane-bound flavocytochrome c-sulfide dehydrogenase (FCSD) related to sulfide oxidation has been studied (Yamanaka et al., 1979) at the molecular level (Vertéet al., 2002) in Chlorobaculum thiosulfatiphilum (formerly Chlorobium limicola f. sp. thiosulfatophilum). In addition, homologues of sulfide : quinone reductase (SQR) and the reverse-acting dissimilatory sulfite reductase (Dsr) system (the latter has been proved to oxidize sulfane sulfur species to the level of sulfite in A. vinosum; Pott & Dahl, 1998; Dahl et al., 2005; Sander et al., 2006) are both reportedly widespread in the GSB, with the exception of a few species, for example, Chloroherpeton thalassium (Frigaard & Bryant, 2008). While all the members of Chlorobi typically use sulfur (S0) and sulfide (H2S/HS) as photosynthetic electron donors, a handful of species can additionally oxidize thiosulfate to support phototrophic growth. These include the strains (viz. 1630 and 9330) retained in the emended species C. limicola, and all the other strains of erstwhile C. limicola that are now classified as C. thiosulfatiphilum (all these were previously classified as C. limicola f. sp. thiosulfatophilum) (Trüper et al., 1988; Pfennig & Trüper, 1992; Imhoff, 2003). Similarly, strains of Chlorobaculum parvum (formerly affiliated to Chlorobium vibrioforme ssp. thiosulfatophilum; Imhoff, 2003) and Chlorobaculum tepidum (Pfennig & Trüper, 1992; Imhoff, 2003; Chan et al., 2008a, b), and also the type strain of Chlorobium clathratiforme (formerly Pelodictyon phaeoclathrathiforme) also have this capacity. The thiosulfate-utilizing forms of Chlorobi are easier to cultivate, and hence better studied, than those depending solely on sulfide (Overmann, 2001). These organisms oxidize thiosulfate via extracellular deposition of sulfur, which is subsequently oxidized to sulfate by most of them (Trüper et al., 1988). Although previous observations disputed the sulfur deposition step in Chlorobaculum tepidum (Chan et al., 2008a), the latest reports have confirmed the transient formation of elemental sulfur as an intermediate during thiosulfate oxidation by this bacterium (Chan et al., 2008b). It has also been demonstrated that in Chlorobaculum tepidum the elemental sulfur transiently produced during the oxidation of thiosulfate (or sulfide) is consumed concomitantly with the initial substrates. Again, acetate assimilation is carried out by this bacterium in tandem with thiosulfate oxidation, but the latter process is reportedly stimulated when cells are grown autotrophically (Chan et al., 2008a, b). The thiosulfate-oxidizing GSB possess unique characteristic features such as formation of thiosulfate during phototrophic sulfide oxidation, disproportion of elemental sulfur in the presence of light but in the absence of CO2, possession of high levels of thiosulfate sulfur transferase or rhodanese and the presence of soluble cytochrome c551, which plays a key role in the utilization of thiosulfate (Kusai & Yamanaka, 1973; Trüper et al., 1988; Brune, 1989; Vertéet al., 2002). While all thiosulfate-utilizing GSB identically possess a shortened version of the sox gene cluster (which typically governs sulfur compound oxidation in aerobic chemotrophic as well as anaerobic phototrophic Alphaproteobacteria; Friedrich et al., 2000, 2001; Mukhopadhyaya et al., 2000; Appia-Ayme et al., 2001), sulfide-utilizing strains possess some or all of the proteins such as FCSD, adenosine 5′-phosphosulfate (APS) reductase, ATP sulfurylase and the quinone-interacting membrane oxidoreductase (Qmo) complex (Frigaard & Bryant, 2008). Again, many of these putative sulfur oxidation-related genes are clustered in so-called sulfur islands in Chlorobaculum tepidum, where they are interspersed with novel genes encoding hypothetical and conserved hypothetical proteins crucial for sulfur-lithotrophic functions (Chan et al., 2008a, b). Molecular functions of these enzymes or enzyme systems are discussed in detail in subsequent sections dealing with the mechanisms of oxidation of different reduced sulfur species. In the present context, however, it would be pertinent to mention that mutations in these predicted sulfur oxidation genes reportedly yield pleiotropic phenotypes in Chlorobaculum tepidum, suggesting that sulfur oxidation, light harvesting and CO2 fixation are tightly integrated and interdependent processes in these bacteria (Chan et al., 2008a).

Photolithotrophic PSB

Photolithotrophic PSB include members of the families Chromatiaceae and Ectothiorhodospiraceae of Gammaproteobacteria. Both the families belong to the order Chromatiales and differ with regard to their deposition of elemental sulfur globules either outside (Ectothiorhodospiraceae) or inside (Chromatiaceae) the cellular limits during thiosulfate and sulfide oxidation (van Gemerden, 1986; Imhoff et al., 1998b; Imhoff, 1999, 2001a). The clear separation of Ectothiorhodospiraceae and Chromatiaceae is manifested in their phylogeny, in addition to which a genetic divergence is also observed between Chromatium species originating from freshwater and those of a truly marine and halophilic nature (Imhoff et al., 1998b).

Members of Chromatiaceae belong to three major phylogenetic branches that, respectively, consist of (1) marine and halophilic species classified under the genera Halochromatium, Marichromatium and Isochromatium, (2) freshwater Chromatium species together with species of Thiocystis and (3) species of the genera Thiocapsa and Amoebobacter, as recently reclassified, plus the new genus Thiolamprovum (Guyoneaud et al., 1998; Imhoff et al., 1998b). Former members of Chromatium that are now classified as species of Allochromatium (C. vinosum, Chromatium minutissimum and Chromatium warmingii) and Thermochromatium (Chromatium tepidum, Chromatium salexigens, Chromatium gracile and Chromatium buderi) are closely related to the second phylogenetic group (Imhoff et al., 1998b). Strains of Allochromatium are ecologically versatile and inhabit stagnant freshwater ditches, ponds and lakes containing hydrogen sulfide, and are also found in sewage lagoons, estuaries and salt marshes. They have no special requirement for salt, although some strains originating from marine and brackish water environment may tolerate or require low concentrations of salt. The metabolically versatile type species A. vinosum is the best-studied PSB. It partially oxidizes thiosulfate to tetrathionate at pH values <7.0 i.e. under neutral to slightly acidic growth conditions (Smith, 1966; Smith & Lascelles, 1966), but under alkaline conditions thiosulfate is oxidized to sulfate via formation of sulfur globules as an obligate intermediate (Grimm et al., 2008). Proteins essential for both these pathways have been purified and the molecular bases of the two apparently discrete processes determined (Hensen et al., 2006; Dahl, 2008). Sulfur oxidation by A. vinosum has been discussed below in detail under The branched thiosulfate oxidation pathway.

Ectothiorhodospiraceae, on the other hand, includes photo-, as well as chemoautotrophic halophilic and haloalkaliphilic bacteria distinguished by their characteristic lamellar intracellular membrane structures (van Gemerden, 1986; Imhoff, 1999, 2001a; Tourova et al., 2007). In the species Thiorhodospira sibirica part of the sulfur globules remains attached to the cells, or located in the periplasm, in such a way that it microscopically appears to be located inside the cells, although this is, in reality, not the case (Bryantseva et al., 1999). Sulfur-chemolithoautotrophs of the family include species of Thioalkalivibrio, Thioalkalispira and Alkalilimnicola, whereas photolithoautotrophic sulfur oxidizers are all included in the genera Ectothiorhodospira, Halorhodospira and Thiorhodospira (Sorokin et al., 2001a, 2002, 2004; Tikhonova et al., 2006; Tourova et al., 2007). A more recently described organism called Ectothiorhodosinus mongolicum, isolated from a soda lake having a pH of 9.4 and mineralization of 3.3%, is an anaerobe and facultative photoorganoheterotroph (Gorlenko et al., 2004). Although scant information is available on the biochemistry of sulfur lithotrophy in Ectothiorhodospiraceae, a membrane-localized cytochrome c552, involved in sulfide oxidation and closely related to the cytochrome subunit of the FCSD of C. thiosulfatiphilum (Vertéet al., 2002), has been characterized from Ectothiorhodospira vacuolata (Kostanjevecki et al., 2000). The genomic ORF encoding this cytochrome has been characterized together with the adjacent flavoprotein gene. Notably, although these two proteins of E. vacuolata are similar to FCSDs or SoxFE homologues of other sulfur lithotrophs (Chen et al., 1994; Reinartz et al., 1998; Appia-Ayme et al., 2001), unlike some of the latter types they are not flanked by any other sox gene and function as a discrete transcriptional unit regulated by sulfide and not thiosulfate (Kostanjevecki et al., 2000).

Photolithotrophic PNSB

Photolithotrophic PNSB are a highly diversified group of >15 genera distributed over the Alphaproteobacteria (e.g. species belonging to Rhodospirillum, Rhodocista, etc. of Rhodospirillaceae; Rhodopila of Acetobacteraceae; Rhodobacter, Rhodovulum, etc. of Rhodobacteraeae; Rhodopseudomonas, etc. of Bradyrhizobiaceae; Rhodomicrobium of Hyphomicrobiaceae; and Rhodobium belonging to Rhodobiaceae) and Betaproteobacteria (e.g. species of Rhodocyclus and Rhodoferax) (Hansen & Imhoff, 1985; Hiraishi et al., 1991; Imhoff, 1995, 2001b, c; Imhoff et al., 1998a; Madigan et al., 2000). All PNSB prefer photoheterotrophic growth under anaerobic conditions, and many species can grow photoautotrophically with hydrogen or sulfide as electron donor. Many members of the last category again do not completely oxidize sulfide to sulfate and instead form sulfur as the end product. Nonetheless, sulfate is the end product of sulfide oxidation by many Rhodovulum spp., Rhodopseudomonas palustris or Blastochloris sulfoviridis (Brune, 1995a; Imhoff, 2001d). Whereas sulfide tolerances are species specific (Imhoff, 1995), sulfide concentration plays a role in their distribution in different ecosystems (Karr et al., 2003). Growth of these organisms is arrested by sulfide concentrations much lower than the levels that are inhibitory for PSB (Hansen & van Gemerden, 1972). Regardless of their sulfide sensitivity, PNSB are metabolically highly diverse, and a good number of species are capable of growing chemoorganotrophically or chemolithoautotrophically under aerobic to microaerobic conditions in the dark (Kondratieva et al., 1992; Imhoff, 1995). These attributes not only allow them to occupy a wide range of habitats (Zhang et al., 2002; Do et al., 2003) but also provide a competitive edge over PSB in light-limited environments (Madigan, 1988). Thiosulfate is also used by many species, and oxidized either to tetrathionate or completely to sulfate (Brune, 1995a; Appia-Ayme et al., 2001; Imhoff, 2001d).

The photosynthetic electron transport and the oxidation–reduction processes of PNSB are integrated by complex mechanisms during photoautotrophic and photoheterotrophic growth (McEwan, 1994). The photosynthetic apparatus, enzymes involved in CO2 fixation, as well as pathways of anaerobic respiration, are all induced in these organisms upon a reduction in oxygen tension. Recently, there have been significant advances in the understanding of molecular properties of their photosynthetic apparatus and the control of the expression of genes involved in photosynthesis and CO2 fixation. In addition, anaerobic respiratory pathways of PNSB have been characterized and their interaction with photosynthetic electron transport described (McEwan, 1994).

Alphaproteobacteria such as Rhodovulum sulfidophilum [formerly (f.) Rhodobacter sulfidophilus, formerly Rhodopseudomonas sulfidophila] (Hansen & Veldkamp, 1973; Hiraishi & Ueda, 1995), Rhodobacter capsulatus and Rhodobacter sphaeroides, which are closely related to the chemolithotrophic species of Paracoccus, exhibit remarkable versatility in their anaerobic phototrophic metabolism. On the other hand, strains of Rhodospirillum rubrum, Rhodopseudomonas capsulata and Rhodopseudomonas sphaeroides oxidize sulfide to extracellular elemental sulfur only. Rhodopseudomonas palustris converts sulfide directly to sulfate without intermediate accumulation of any reduced sulfur species (Hansen & van Gemerden, 1972), while R. sulfidophilum is also capable of delivering the complete ‘eight-electron’ oxidation of sulfide or thiosulfate to sulfate (Appia-Ayme et al., 2001). Oxidation of sulfide, but not thiosulfate, is constitutive in R. sulfidophilum (Neutzling et al., 1985). Thiosulfate oxidation in this organism produces only sulfite as the free intermediate, although it is rendered by molecular mechanisms akin to processes encountered in the aerobic chemolithotrophic Alphaproteobacteria (typified by Paracoccus spp.), which do not form any detectable free intermediate during the oxidation of thiosulfate (as well as other reduced sulfur species). Although thiosulfate (or other sulfur compounds) oxidation by the periplasmically located thiosulfate-oxidizing multi-enzyme (TOMES or Sox) complex encoded by the conserved sox gene cluster (Kelly et al., 1997; Appia-Ayme et al., 2001; Friedrich et al., 2001; Bamford et al., 2002) produces no detectable free intermediate in Paracoccus spp. (presumably because the sulfur species remain bound to the multimeric complex during their oxidation; Friedrich et al., 2001), the same process produces sulfite as a free intermediate from both thiosulfate and sulfide in R. sulfidophilum (Neutzling et al., 1985; Appia-Ayme et al., 2001). By virtue of this unique property of the Sox system of R. sulfidophilum, sulfite could be established as a product as well as a substrate of the Sox complex.

The electron transfer pathways associated with thiosulfate and sulfide oxidation in R. sulfidophilum utilize cytochrome c2 and/or ubiquinone as the terminal oxidant. When grown autotrophically (but not heterotrophically) with thiosulfate as an electron donor, this organism was long found to express a c-type cytochrome (Neutzling et al., 1985) that was subsequently characterized as a heterodimeric (having a diheme 30-kDa SoxA and a monoheme 15-kDa SoxX subunits) constituent of the Sox complex essential for the oxidation of both sulfide and thiosulfate by this organism (Appia-Ayme et al., 2001). These findings simultaneously proved that there was only one pathway for the oxidation of both sulfide and thiosulfate in R. sulfidophilum (Appia-Ayme et al., 2001), and the mechanisms of sulfur photolithotrophy in purple sulfur and nonsulfur bacteria were clearly distinct (Friedrich et al., 2001).

Chemolithotrophic oxidation of reduced inorganic sulfur species

‘Their life processes are played out in a very simple fashion; all their activities are driven by a purely inorganic chemical process’; this is how Sergei Winogradsky described the unique life process of chemolithotrophy in 1887 [translation of the original text in German was done by Kelly & Wood (2000a)]. But perhaps he did not expect that 130 years after his epoch-making discovery we would come to answer fewer and ask more questions about this apparently ‘simple’ metabolic process.

Sulfur-chemolithotrophic prokaryotes are not only highly diversified in terms of their taxonomy and ecology, but are also multifaceted with respect to the physiology and biochemistry of their sulfur oxidation processes. Sulfur-chemolithotrophic bacteria exhibit differential abilities to utilize different reduced sulfur compounds as substrates (Wood & Kelly, 1987; Katayama et al., 1995; Kelly et al., 2000; Sorokin et al., 2001a, b; Ghosh et al., 2005; Ghosh & Roy, 2007a, b), and the efficacy of energy conservation from the same sulfur substrates by different organisms at their respective pH and temperature optima is also variable (Mason et al., 1987; Kelly, 1989). Species-dependent biochemical differences also pertain to the oxidative enzymes, pathways and electron transport mechanisms that different groups of bacteria use to metabolize any given sulfur compound (Lu et al., 1985; Lu, 1986; Lu & Kelly, 1988a, b; Pronk et al., 1990; Kelly et al., 1997; Friedrich, 1998). Sulfur-oxidizing capabilities of the representatives of the various metabolic categories of chemotrophic bacteria are outlined in Table 2. On the other hand, chemolithotrophic sulfur oxidation by archaea is relatively less understood in comparison to the bacterial counterparts. Nevertheless, it has been ascertained that the enzymatic and molecular basis of sulfur oxidation in archaea is totally different from those of bacteria (Kletzin et al., 2004; Kletzin, 2008). Again, more than one sulfur-oxidizing enzyme system have also been envisaged even within a single bacterium such as Thiobacillus denitrificans, which can adapt to varying physicochemical conditions in diverse environments (Beller et al., 2006). Thiobacillus denitrificans (Kelly & Wood, 2000b), which is also capable of carrying out the unique anaerobic (nitrate-dependent) oxidation of uranium oxide minerals [U(IV) to U(VI)] (Beller et al., 2006), differs from all other sulfur chemolithotrophs by its ability to conserve energy from the oxidation of inorganic sulfur compounds under both aerobic and denitrifying conditions. Although sulfur chemolithotrophs are mostly aerobic, using molecular oxygen as terminal electron acceptor, species of Beggiatoa and Thioploca, the haloalkaliphilic gammaproteobacterium Thioalkalivibrio and the moderately halophilic Thiohalomonas, and Sulfurimonas denitrificans (formerly Thiomicrospira denitrificans; Takai et al., 2006) can also anaerobically oxidize sulfur by coupling it to nitrate reduction (Timmer-Ten Hoor, 1975; Kelly & Wood, 2000b; Sorokin et al., 2003, 2004, 2007). Besides S. denitrificans, newly described members of Epsilonproteobacteria such as Sulfurimonas autotrophica and Sulfurovum lithotrophicum (Inagaki et al., 2003, 2004), Arcobacter sp. strain FWKO B, Thiomicrospira sp. strain CVO (Gevertz et al., 2000) and Sulfuricurvum kujiense (Kodama & Watanabe, 2004) are also facultatively anaerobic, nitrate-reducing, sulfur-oxidizing chemolithotrophs. Thiobacillus denitrificans is the best studied among the obligately sulfur-chemolithoautotrophic bacteria known to couple denitrification to sulfur compound oxidation. Like these aforesaid bacteria, several sulfur-oxidizing chemolithoautotrophic hyperthermophilic archaea can also use an extraordinary array of electron donors, including H2, Fe2+, H2S, S, S2O32−, S4O62−, sulfide minerals, CH4, various carboxylic acids, alcohols, amino acids and complex organic substrates, while electron acceptors include O2, Fe3+, CO2, CO, NO3, NO2, NO, N2O, SO42−, SO32−, S2O32− and S (Amend & Shock, 2001).

Table 2.   Sulfur-oxidizing capabilities of the representatives of the various metabolic categories of chemotrophic bacteria
Representative speciespH/TempTaxonomic affiliationBiochemical characteristicsMolecular system(s)
  • *

    Beggiatoa alba has been found to possess a soxB gene (Meyer et al., 2007).

  • Names of facultative anaerobes are underlined.

Facultative chemolithoautotrophs
 Paracoccus spp.Neutrophilic/mesophilicAlphaproteobacteriaS2O32−, SO32−, S2− and S0 oxidized directly to SO42− by the periplasmic Sox complexSoxXAYZBCD
Kelly et al. (1997); Friedrich et al. (2001)
 Pseudaminobacter salicylatoxidansNeutrophilic/mesophilicAlphaproteobacteriaS2O32−, S4O62−, SO32−, S2− and S0 oxidized directly to SO42− without any intermediate formationSoxXAYZBCDMukhopadhyaya et al. (2000); Lahiri et al. (2006)
 Starkeya novellaNeutrophilic/mesophilicAlphaproteobacteriaOxidizes S2O32−, SO32−, S2− and S0; whole cells, but not cell-free extracts, also oxidize S4O62−
Direct S2O32− oxidation is mediated by a membrane-bound multienzyme complex
Soluble enzymes involved in sulfur oxidation include sulfite dehydrogenase, thiosulfate-cleaving enzyme and possibly a glutathione-dependent sulfur oxygenase
Charles & Suzuki (1966); Oh & Suzuki (1977); Kelly et al. (1997); Fukumori et al. (1989); Kappler et al. (2000, 2001, 2004)
 Magnetospirillum magnetotacticumNeutrophilic/mesophilicAlphaproteobacteriaH2S or S2O32− is oxidized aerobically, whereas S2O32− is also oxidized with N2O as the terminal electron acceptor. SO42− is always the end product
Formation of conspicuous sulfur globules during H2S as well as S2O32− oxidation
Bazylinski et al. (2004); Meyer et al. (2007)
 Tetrathiobacter kashmirensisNeutrophilic/mesophilicBetaproteobacteriaS4O62− and S2O32− are oxidized
S4I pathway for S2O32− oxidation
SoxBGhosh et al. (2005); Dam et al. (2007)
 Acidithiobacillus caldusAcidophilic/moderately thermophilicGammaproteobacteriaS4O62−, S2O32−, SO32−, S2− and S0 are oxidized
S4I pathway for S2O32− oxidation
TetH and DoxD
Hallberg et al. (1996); Rzhepishevska et al. (2007)
 Acidiphilium acidophilum (grows mixotrophically on mixtures of glucose and S2O32− or S4O62−; mixotrophic growth yields are higher than the sum of the heterotrophic and autotrophic yields)Acidophilic/mesophilicAlphaproteobacteriaS2O32−, S3O62−, S4O62− and S0 support growth
S4I pathway for S2O32− oxidation
GSSH-mediated sulfur dioxygenase pathway for sulfane sulfur species
Free S2− oxidized by SQOR
Kelly (1989); Pronk et al. (1990); Rohwerder & Sand (2003)
Obligate chemolithoautotrophs
 Acidithiobacillus ferrooxidansAcidophilic/mesophilicGammaproteobacteriaS4O62−, S2O32−, S0 and metal sulfides support growth
S4I pathway for S2O32− oxidation
S0 is an intermediate of S4O62− and S2− oxidation
GSSH-mediated sulfur dioxygenase pathway for sulfane sulfur species
Free S2− oxidized by SQOR
Hazeu et al. (1988); Kelly (1989); Pronk et al. (1990); Rohwerder & Sand (2003); Muller et al. (2004)
 Thermithiobacillus tepidariusNeutrophilic/moderately thermophilicGammaproteobacteriaSnO62−, S2O32−, S2− and S0
S4-intermediate in the oxidation of S2O32− and S3O62−
Lu & Kelly (1988a, b)
 Thiomicrospira crunogenaNeutrophilic/mesophilicGammaproteobacteriaS2O32− is oxidized by depositing S0 globules outside the cell under low pH and O2 conditions
Transient accumulation of SO32− and polythionates as products of S2O32− and S0 globule oxidation
Javor et al. (1990); Scott et al. (2006)
 Sulfurimonas denitrificansNeutrophilic/mesophilicEpsilonproteobacteriaNo intermediate formation during oxidation of thiosulfateSoxXAYZBCD
Sievert et al. (2008)
 Desulfocapsa sulfoexigensNeutrophilic/mesophilicDeltaproteobacteriaDisproportion of S2O32− with SO32− intermediate
Disproportion of S0
Thiosulfate reductase cleaving S2O32− into SO32− and S2−
Finster et al. (1998); Frederiksen & Finster (2003)
 Thiobacillus denitrificansNeutrophilic/mesophilicBetaproteobacteriaAerobic chemolithoautotrophic growth on S2O32−, S4O62− and SCN
Anaerobic chemolithoautotrophic growth on S2O32−, S4O62−, SCN, S2− or S0 using NO3, NO2 or
N2O as terminal oxidants
Chemostat cultures can repeatedly switch between aerobic and anaerobic growth modes
Accumulates S0 during anaerobic oxidation of S2O32− in batch cultures, but not in continuous-flow chemostats
SoxC (no SoxD)
APS reductase
ATP sulfurylase
(apparently multiple oxidation pathways are present for any given sulfur compound)
Justin & Kelly (1978a, b); Schedel & Trüper (1980); Kelly & Wood (2000b); Beller et al. (2006)
 Aquifex aeolicusNeutrophilic/hyperthermophilicAquificalesS2O32− alone cannot support chemolithoautotrophic growth, but is used alongside H2 for energy conservationSoxXAZYBC
Friedrich et al. (2001)
Chemolithoheterotrophs or mixotrophs
 Beggiatoa spp.*Neutrophilic/mesophilicGammaproteobacteriaIntermediary formation and deposition of S0 inside the cell during S2O32− and S2− oxidationDsr
APS reductase
ATP sulfurylase
Teske & Nelson (2004)
 Thiomicrospira sp. CVONeutrophilic/mesophilicEpsilonproteobacteriaS2− or S0 is oxidized with microaerobic O2, NO3 and NO2Gevertz et al. (2000)
 Silicibacter pomeroyiNeutrophilic/mesophilicAlphaproteobacteriaChemolithoheterotrophic growth using sulfur compounds as electron sourceSoxXAYZBCDFriedrich et al. (2005)

Conspicuously filamentous colorless sulfur-chemolithotrophic bacteria

The conspicuously filamentous colorless sulfur-chemolithotrophic bacteria such as Beggiatoa were some of the earliest autotrophic microorganisms to be isolated (Nelson & Jannasch, 1983; Kelly & Wood, 2000a). Like the photolithotrophic GSB and PSB, oxidation of thiosulfate or sulfide in the free-living as well as symbiotic sulfur-chemolithotrophic species of Beggiatoa, Thiothrix, etc. involve the formation of conspicuous globules or droplets of polymeric, water-insoluble sulfur inside the cells (Dahl, 1999; Pasteris et al., 2001; Dahl & Prange, 2006). Observation of this phenomenon apparently enabled Winogradsky to discover chemolithotrophy. Subsequent to their discovery, species of Beggiatoa were regarded as probable heterotrophs, or at best mixotrophs, but their ability to grow as sulfide-oxidizing chemolithoautotrophs is now well established (Nelson et al., 1986; Kelly & Wood, 2000a). Moreover, some sulfide-oxidizing marine strains of Beggiatoa have been found to be even obligately autotrophic (Hagen & Nelson, 1996). Whereas species of Leucothrix are known to grow chemolithoheterotrophically (or mixotrophically), filamentous sulfur chemolithotrophs belonging to the genus Thiothrix grow mixotrophically as well as autotrophically using inorganic sulfur compounds (Odintsova et al., 1993; Kelly & Wood, 2000a). Growth of Beggiatoa alba on acetate is greatly stimulated by the availability of sulfide, which is apparently used as an energy source, enabling surplus assimilation of acetate (Güde et al., 1981). Beggiatoa leptomitiformis grows mixotrophically on succinate and thiosulfate or tetrathionate, which are oxidized to generate ATP by oxidative phosphorylation (Grabovich et al., 1998). Interestingly, although Thiothrix and Beggiatoa are both members of Gammaproteobacteria, Thiothrix is not part of the monophyletic lineage comprising Beggiatoa and Thioploca (Teske & Nelson, 2004).

Small unicellular sulfur-chemolithotrophic bacteria

The small unicellular colorless sulfur chemolithotrophs were unknown at the time of Winogradsky. Fifteen years after the discovery of this unique process of energy conservation, in 1902, Alexander Nathansohn described the first Gram-negative bacterium growing autotrophically by oxidizing thiosulfate (Kelly, 1989). Another 2 years later, Beijerinck scientifically named the first sulfur-chemolithotrophic species Thiobacillus thioparus (Kelly, 1989). For almost a century thereafter sulfur chemolithotrophy was regarded as a conserved genetic trait and a key taxonomic character of the genus Thiobacillus, which until recently remained a heterogeneous assemblage of highly diversified bacteria.

The biochemically similar, neutrophilic obligate chemolithoautotrophs T. thioparus and Thiobacillus denitrificans (Kelly & Wood, 2000b) were the first two thiobacilli to be described (Beijerinck, 1904a, b; Beijerinck & Minkman, 1910). Another obligate sulfur chemolithoautotroph isolated from seawater during the early period of Thiobacillus research by Nathansohn was most possibly a strain of the halotolerant Thiobacillus neapolitanus (now Halothiobacillus neapolitanus) that was again based on the type strain Thiobacillus X (Kelly & Harrison, 1989; Kelly & Wood, 2000c). Halothiobacillus neapolitanus, which was later found to be present in a variety of habitats, despite having no salt requirement can tolerate >800 mM of NaCl in the medium (Sievert et al., 2000; Wood et al., 2005).

Prejudice stemming from the trend of early reports persuaded researchers to believe for quite a long time that sulfur lithotrophy was exclusively associated with the obligately autotrophic organisms. This notion was reversed with the report of the first facultative sulfur chemolithoautotroph Thiobacillus novellus (Starkey, 1935) [now called Starkeya novella (Kelly et al., 2000)] from natural soil samples. Thiobacillus intermedius was the next facultative sulfur chemolithoautotroph to be isolated from freshwater mud (London, 1963), followed by Thiobacillus perometabolis (London & Rittenberg, 1967), originally described as ‘nonautotrophic’ but later reported to grow autotrophically on reduced sulfur substrates (Katayama-Fujimura & Kuraishi, 1983). Another species, Thiobacillus rapidicrescens, very similar to T. perometabolis, was also reported (Katayama-Fujimura et al., 1983; Kelly & Harrison, 1989). Thiobacillus versutus (now Paracoccus versutus; Katayama et al., 1995), originally described as Thiobacillus A2 (Taylor & Hoare, 1969), was isolated from soil during anaerobic enrichments for the isolation of Thiobacillus denitrificans, and long remained the most widely studied facultative sulfur chemolithoautotroph. The acidophilic facultative chemolithotroph Thiobacillus acidophilus (now Acidiphilium acidophilum; Hiraishi et al., 1998) was first isolated from a supposedly pure culture of Thiobacillus ferrooxidans (Guay & Silver, 1975). Isolation of this species demonstrated that the extreme environment of acid mine water is inhabited not only by obligate sulfur chemolithoautotrophs but also their facultative counterparts such as A. acidophilum and gratuitously sulfur-oxidizing organoheterotrophs such as Acidiphilium cryptum (Harrison, 1984). On top of this, Huber & Stetter (1990) reported the facultative sulfur chemolithotroph Thiobacillus cuprinus from copper mine effluents, while yet another facultative chemolithotroph, Thiobacillus delicatus (Mizoguchi et al., 1976; Katayama-Fujimura et al., 1984; Katayama et al., 2006), was also isolated from mine water. However, T. delicatus was found capable of growing in less acidic or neutral pH ranges. The thermoacidophilic, but facultatively chemolithoautotrophic, Thiobacillus caldus (reclassified as Acidithiobacillus caldus; Kelly & Wood, 2000c), isolated from coal spoils, is also unique in its ecophysiological characteristics (Hallberg & Lindström, 1994).

The highly acidophilic and obligately chemolithoautotrophic species T. ferrooxidans (now classified as Acidithiobacillus ferrooxidans; Kelly & Wood, 2000c), one of the most widely studied acidophilic thiobacilli having three lithotrophic traits based on iron, sulfur or hydrogen, was first retrieved from acidic drainage of bituminous coal fields (Temple & Colmer, 1951; Hazeu et al., 1988; Drobner et al., 1990). Subsequently, a large number of strains were isolated from different coal mine water effluents, water of uranium, iron and copper mines, soils containing pyrite or marcasite, with low to very low pH. Many species such as Ferrobacillus ferrooxidans, Ferrobacillus sulfooxidans, etc. were created, but all were later identified as strains of A. ferrooxidans (Kelly & Tuovinen, 1972). Other acidophilic obligate chemolithoautotrophs such as Thiobacillus albertensis (Bryant et al., 1983) and Thiobacillus plumbophilus (Drobner et al., 1992) were isolated from acidic soil adjacent to sulfur stockpiles and lead-contaminated soil, respectively. Strains of the obligate chemolithoautotroph Thiobacillus thiooxidans (reclassified as Acidithiobacillus thiooxidans; Kelly & Wood, 2000c), originally isolated from natural field soil (Waksman & Joffe, 1922), were later reported from soil containing flowers of sulfur, sulfur springs, composts of soils, acidic soil of sulfur stockpile, corroding steel, corroding concrete and acid mine water, etc. A similar bacterium named Thiobacillus concretivorus was once listed as a distinct species in Bergey's Manual of Determinative Bacteriology (Parker, 1957) but was later recognized to be a member of T. thiooxidans (Kelly & Harrison, 1989).

Several years after the discovery of H. neapolitanus, another extremely halotolerant, obligately chemolithotrophic, neutrophilic and mesophilic species, Thiobacillus halophilus, was isolated from a hypersaline lake of the Australian wheat belt (Wood & Kelly, 1991). The obligately chemolithotrophic, mesophilic acidophile Thiobacillus prosperus, isolated from shallow geothermally heated sea floor, was the first halotolerant metal-leaching bacterium described (Huber & Stetter, 1989).

The neutrophilic, obligate sulfur chemolithotroph Thiobacillus tepidarius (reclassified as Thermithiobacillus tepidarius) was moderately thermophilic (Wood & Kelly, 1985; Kelly & Wood, 2000c), whereas other moderately thermophilic species such as Thiobacillus aquaesulis, isolated from thermal sulfur spring water (Wood & Kelly, 1988) were facultative chemolithotrophs capable of growing on complex rich media. Other moderately thermophilic sulfur chemolithotrophs included A. caldus (Hallberg & Lindström, 1994), Thiobacillus thermosulfatus (a facultative chemolithoautotroph) isolated from sewage sludge enriched with elemental sulfur (Shooner et al., 1996), and Thiobacillus hydrothermalis (this strictly autotrophic species despite being isolated from a deep-sea hydrothermal vent was found to be mesophilic) (Durand et al., 1993; Kelly et al., 1998).

Reorganization of the highly heterogeneous and artificial generic entity called Thiobacillus was not possible until the late 1990s when the development and application of the concepts of molecular phylogeny (Woese, 1987; Stackebrandt & Goebel, 1994) and polyphasic systematics (Vandamme et al., 1996) resulted in the restructuring of many such polyphyletic prokaryotic taxa. Since then there has been progressive widening of the phylogenetic and ecological spectra over which sulfur chemolithotrophs are distributed. The process of taxonomic reorganization led to the identification of the metabolically versatile facultative thiobacilli either as species of existing genera of Proteobacteria or as new genera created to accommodate the distinctive ones (Katayama et al., 1995; Hiraishi et al., 1998; Rainey et al., 1999; Kelly et al., 2000). The genus Thiobacillus retained a few of its species such as T. thioparus and Thiobacillus denitrificans, and was placed in the Betaproteobacteria (McDonald et al., 1997; Kelly & Wood, 2000b, c; Robertson & Kuenen, 2002). Whereas former species such as T. cuprinus, T. intermedius, T. perometabolis and T. thermosulfatus, etc. were collectively organized in a new genus Thiomonas also affiliated to the Betaproteobacteria (Moreira & Amils, 1997; Katayama et al., 2006), other obligately chemolithotrophic thiobacilli were identified as members of the Gammaproteobacteria and three new genera (Acidithiobacillus, Halothiobacillus and Thermithiobacillus) were created to accommodate eight such species (Kelly & Wood, 2000c). Thiobacillus thyasiris, isolated from the gill tissue of the marine bivalve mollusk Thyasira flexuosa (Wood & Kelly, 1989), was also transferred to the Gammaproteobacteria and classified as Thiomicrospira thyasirae (Wood & Kelly, 1993). Concurrent with this taxonomic reorganization of Thiobacillus, sulfur-chemolithotrophic attributes were increasingly reported from diverged, existing or newly created, generic and specific entities distributed over several phyla of Bacteria (Robertson & Kuenen, 1983; Das et al., 1996; Bazylinski et al., 2004; Deb et al., 2004; Ghosh et al., 2005, 2006; Hirayama et al., 2005; Ghosh & Roy, 2006a, 2007a, b) and Archaea (Stetter et al., 1990; Fuchs et al., 1996).

Like their wide taxonomic and ecological distribution, the unicellular sulfur-chemolithotrophic bacteria are equally diversified with regard to their nutritional attributes. They utilize a vast array of reduced sulfur compounds as substrates under varied pH and temperature conditions. Whereas obligate chemolithoautotrophs belonging to Acidithiobacillus, Halothiobacillus and Thermithiobacillus oxidize sulfur, thiosulfate, tetrathionate and other polythionates under extremely low pH, high salinity and high temperatures, respectively (Kelly & Wood, 2000c; Sievert et al., 2000), other obligately chemolithoautotrophic Gammaproteobacteria such as Thioalkalimicrobium, Thiohalomonas and Thioalkalivibrio utilize thiocyanate in addition to other reduced sulfur compounds at extraordinarily high pH levels and/or salt concentrations (Sorokin et al., 2001a, b, 2004, 2007). Alphaproteobacterial members are mostly facultatively autotrophic, and sometimes mixotrophic (e.g. Methylobacterium oryzae) or chemolithoheterotrophic (e.g. Silicibacter pomeroyi) (Kelly & Wood, 2000a; Moran et al., 2004; Anandham et al., 2007; Ghosh & Roy, 2007a). Under nutrient-limited mixotrophic conditions Xanthobacter tagetidis (Padden et al., 1998) and Pseudaminobacter salicylatoxidans (Lahiri et al., 2006) show increases in growth and cellular yield over the autotrophic levels. Obligately chemolithotrophic unicellular sulfur-oxidizing bacteria, on the other hand, belong to the Beta- (e.g. the thermophilic Thermothrix azorensis and Thiobacillus spp.), Gamma- (e.g. Acidithiobacillus and Thiomicrospira spp.) and Epsilonproteobacteria (e.g. S. denitrificans, S. lithotrophicum, S. autotrophica, etc.); although facultative, sulfur chemolithotrophy is not totally absent in the Betaproteobacteria (e.g. Tetrathiobacter kashmirensis and T. aquaesulis are facultatively sulfur chemolithoautotrophic, Thiobacillus rubellus and Thiobacillus strain Q are chemolithoheterotrophic, while species of Thiomonas are mixotrophic) (Mizoguchi et al., 1976; Gommers & Kuenen, 1988; Wood & Kelly, 1988; Odintsova et al., 1996; Moreira & Amils, 1997; Kelly & Wood, 2000c; Ghosh et al., 2005; Campbell et al., 2006). Again, unlike its phylogenetic relatives, the epsilonproteobacterium Thiomicrospira sp. CVO, isolated from oil-field brines, is a mixotroph (Gevertz et al., 2000). Facultative mixotrophy involving oxidation of thiosulfate and sulfur as energy and electron sources in tandem with use of diverse organic sources is also known in bacteria as diverged as Thermus scotodoctus and Thermus antranikianii belonging to the phylum DeinococcusThermus (Skirnisdottir et al., 2001). In comparison to the advances made with facultatively chemolithotrophic Alphaproteobacteria, the mechanisms of sulfur oxidation, particularly how it is coupled to energy conservation, is still not well understood in the obligately chemolithotrophic species of Beta-, Gamma- and Epsilonproteobacteria.

Aerobic chemolithoautotrophic sulfur-oxidizing species of Archaea

Aerobic chemolithoautotrophic sulfur-oxidizing species of Archaea are largely restricted to the order Sulfolobales of Crenarchaeota (Stetter et al., 1990; Fuchs et al., 1996; Huber & Stetter, 2001). Species of Sulfolobus are facultatively aerobic and can grow not only in the presence of oxygen by oxidation of H2S or S0 to H2SO4 but also anaerobically by oxidation of H2 and reduction of S0 to H2S. Acidianus spp., on the other hand, are obligate aerobes oxidizing H2S, S0 and tetrathionate to H2SO4 in addition to sulfidic ores to metal sulfates. Whereas all the strains of Acidianus species are facultatively heterotrophic, only some strains of Sulfolobus possess the same capacity. Like Acidianus members, Sulfurisphaera species are also obligately aerobic and facultatively heterotrophic, oxidizing S0 and sulfidic ores to H2SO4 and metal sulfates, respectively, for chemolithoautotrophic growth (Huber & Stetter, 2001). Strictly aerobic, thermoacidophilic members of Sulfurococcus (Karavaiko et al., 1994) are facultatively lithotrophic and grow via oxidation of S0, pyrite, chalcopyrite and sphalerite to H2SO4 under both autotrophic and mixotrophic conditions (Reysenbach, 2001).

Sulfur-oxidizing chemolithoautotrophic hyperthermophilic archaea have mostly been isolated from marine hydrothermal systems, heated sediments, continental solfataras, hot springs, water heaters and industrial wastes. In recent times, an extremely halophilic, neutrophilic archaeon, identified as a member of the genus Natronorubrum of class Halobacteria belonging to the phylum Euryarchaeota, has been isolated from a mixed sediment sample from different hypersaline lakes by enrichment with acetate and thiosulfate as substrates. This archaeon is able to oxidize thiosulfate to tetrathionate during heterotrophic growth with acetate, a property not yet demonstrated for any of the known haloarchaea. The presence of a thiosulfate-oxidizing tetrathionate synthase associated with membranes, and dependent specifically on Cl, was detected in this archaeon (Sorokin et al., 2005).

Biochemistry and molecular biology of the oxidative transformations of different reduced sulfur compounds by different groups of Bacteria

Oxidation of thiosulfate and polythionates

Thiosulfate, one of the most abundant reduced inorganic sulfur species, plays a central role in the biogeochemical sulfur cycle, as it happens to be a common substrate oxidized by almost all sulfur lithotrophs. As thiosulfate can chemically decompose to sulfur and sulfite at pH values <4.0 (Johnston & McAmish, 1973) and the rate of the process is strongly dependent on thiosulfate concentration, batch culture studies are more feasible for the neutrophilic lithotrophs, while thiosulfate-limited chemostat cultures are suitable for studying the growth of acidophilic sulfur oxidizers. In contrast, tetrathionate is relatively acid-stable, though in the presence of other reduced sulfur species it often produces pentathionate (Johnston & McAmish, 1973). A wide variety of photo-, as well as chemo-, lithotrophic sulfur-oxidizing bacteria can use thiosulfate as an electron donor for their photosynthetic or respiratory energy-generating systems. Species distributed over diverse phylogenetic groups additionally utilize tetrathionate and other polythionates as energy and electron sources. Tetrathionate is also produced as an intermediate in the oxidation of thiosulfate (and in some cases trithionate) (Kelly, 1989; Kelly et al., 1997) by several Beta- and Gammaproteobacteria that follow the ‘Tetrathionate Intermediate (S4I) pathway’ (Trudinger, 1964; Wood & Kelly, 1985, 1987; Pronk et al., 1990; Meulenberg et al., 1993b; Hallberg & Lindström, 1994; Tano et al., 1996; De Jong et al., 1997a, b; Ghosh et al., 2005), but not by any Alphaproteobacteria, which as a group typically follow the ‘Paracoccus Sulfur Oxidation (PSO) pathway,’ where thiosulfate is converted directly to sulfate without the formation of any detectable intermediate (Kelly, 1982, 1989; Lu et al., 1985; Lu, 1986; Kelly et al., 1997; Friedrich et al., 2001). Nonetheless, certain species of Alphaproteobacteria can oxidize tetrathionate as a chemolithotrophic substrate, much like the S4I-forming bacteria, which can all grow with tetrathionate as a sole energy source (Kelly et al., 2000; Mukhopadhyaya et al., 2000; Ghosh & Roy, 2007a, b). Different thiosulfate-oxidizing enzymes (or enzyme systems) have been studied from taxonomically and physiologically diverse bacteria, and at least three biochemical pathways for the complete oxidation of thiosulfate are known from different sulfur-oxidizing chemo- or photolithotrophic bacteria (Kelly et al., 1997; Brüser et al., 2000a; Friedrich et al., 2001, 2005; Grimm et al., 2008).

Over and above these major processes, several other enzymic conversions of thiosulfate are known. Thiosulfate : cyanide sulfur transferases or rhodaneses (Cerletti, 1986) are widespread in the biological world and their activities have been reported from bacteria to man. Their physiological role has been debated for years, with several proposals being forwarded that range from detoxification of cyanide to the suggestion that the rhodanese-mediated thiosulfate-cleaving reaction, S–SO32−+2H2O↔S+H2SO3+2OH, is significant in the bioenergetic oxidation of thiosulfate by many chemolithotrophic species (Kelly, 1989; Suzuki, 1999).

Some Pseudomonas and Halomonas species also use the partial oxidation of thiosulfate to tetrathionate as supplemental energy source (Sorokin et al., 1999; Sorokin, 2003). Marine haloalkaliphilic heterotrophs belonging to the Gammaproteobacteria, particularly the Halomonas group, oxidize thiosulfate and, much less actively elemental sulfur and sulfide, incompletely to tetrathionate. Some denitrifying species from this group can also carry out anaerobic oxidation of thiosulfate and sulfide using nitrogen oxides as electron acceptors. Despite the low energy output of the oxidation of thiosulfate to tetrathionate it can be utilized for ATP synthesis by some of these organisms, although this potential is not always realized during their growth (Sorokin, 2003). Yet again, the deltaproteobacterium Desulfocapsa sulfoexigens can grow chemolithotrophically by disproportionation reaction of thiosulfate where sulfite is an intermediate (Frederiksen & Finster, 2003). Thiosulfate reductase, which cleaves thiosulfate into sulfite and sulfide, has also been detected in cell-free extract from thiosulfate disproportionating cultures of this bacterium and supposed to catalyze the first step in thiosulfate disproportionation.

The alphaproteobacterial Sox pathway

The alphaproteobacterial PSO pathway, governed by the conserved sox operon, operates in photo-, as well as chemolithotrophic Alphaproteobacteria, which convert both sulfur atoms of thiosulfate to sulfate without the formation of any free intermediate (neither tetrathionate nor sulfur globules) (Kelly, 1982, 1989). To understand this method of thiosulfate oxidation, a membrane-associated thiosulfate oxidizing system was first characterized from S. novella (Oh & Suzuki, 1977). Cell-free extracts of autotrophically grown cells were found to oxidize thiosulfate stoichiometrically to sulfate, while the related enzyme system was shown to be associated with cell membranes, instead of the periplasmic space as found later for P. versutus (Kelly, 1988). Oh & Suzuki (1977) postulated that rhodanese was involved in the initial attack on thiosulfate, cleaving it to sulfite and sulfur, following which sulfur was oxidized to sulfite by sulfur oxygenase and sulfite was oxidized to sulfate by sulfite : cytochrome c oxidoreductase. Oxygen reduction was envisaged to take place in the last step via cytochrome c oxidase. The unique feature of the S. novella scheme was the production of sulfur and its oxidation to sulfate by a glutathione-dependent sulfur oxygenase (Charles & Suzuki, 1966). On the other hand, the periplasmically located thiosulfate-oxidizing multienzyme complex system (TOMES) (comprising the so-called enzyme A, enzyme B and multiheme cytochromes) that was characterized from P. versutus later paved the way for the molecular understanding of the PSO pathway (Kelly, 1989; Kelly et al., 1997). Unlike S. novella, the TOMES of P. versutus did not involve any sulfur oxygenase to oxidize thiosulfate and eight molecules of cytochrome c were reduced for the oxidation of each molecule of thiosulfate to sulfate. All the oxygen atoms required to produce sulfate came from the dissociation of water molecules, and all the eight electrons released were transported to the terminal electron acceptor oxygen through the respiratory chain. Cytochrome-dependent complete oxidation of thiosulfate to sulfate was first demonstrated with crude enzyme preparations or cell-free extracts, while the multienzyme system that governs the sequence of reactions was identified later (Lu et al., 1985; Lu, 1986; Kelly, 1988). It was propounded that oxygen from the dissociation of water molecules was sequentially introduced to combine with the sulfur atoms of thiosulfate, with two electrons and two protons being released in each step. According to the P. versutus model, enzyme A initiates the oxidation process by initially binding with thiosulfate, following which, enzyme B (a sulfite : cytochrome c oxidoreductase) and multiheme redox center cytochromes c551 and c552.5 catalyze the production of sulfate, without releasing free intermediates, and transfer electrons to membrane-bound cytochrome c552 and eventually to oxygen by the action of a terminal cytochrome oxidase (Kelly, 1989).

The molecular mechanism of TOMES functioning was later elucidated by extensive genetic studies in Paracoccus denitrificans strain GB17 (now classified as Paracoccus pantotrophus; Rainey et al., 1999), where the entire DNA locus governing sulfur oxidation (Sox) by this pathway was first identified (Wodara et al., 1994, 1997; Friedrich et al., 2000; Rother et al., 2001). Concurrently, the putatively conserved operon soxTRS-VW-XYZABCDEFGH was characterized from facultatively chemo- as well as photolithotrophic Alphaproteobacteria (Mukhopadhyaya et al., 2000; Appia-Ayme et al., 2001; Friedrich et al., 2001, 2005; Kappler et al., 2001, 2004; Lahiri et al., 2006) and shown to govern the oxidation of thiosulfate, sulfite, sulfide and elemental sulfur via a multienzyme complex constituted by the thiosulfate-induced periplasmic proteins SoxXA (both c type cytochromes), SoxYZ (covalently sulfur-binding and sulfur compound-chelating proteins, respectively), SoxB (monomeric, dimanganese-containing protein that is similar to zinc-containing 5′-nucleotidases and thought to act as the sulfate thiol esterase component of Sox system), and Sox(CD)2 (sulfur dehydrogenase), all of which by themselves are catalytically inactive (Quentmeier et al., 2000, 2003, 2007; Appia-Ayme et al., 2001; Friedrich et al., 2001; Quentmeier & Friedrich, 2001; Bamford et al., 2002; Rother & Friedrich, 2002; Sauvéet al., 2007; Kappler et al., 2008). It is now understood that the Sox complex covalently binds thiosulfate by means of a cysteinyl residue located within a characteristic GGCGG pentapeptide at the C terminus of the SoxY subunit of the SoxYZ complex (Quentmeier & Friedrich, 2001) and oxidizes both the sulfur atoms of thiosulfate to sulfate by transferring electrons to small c-type cytochromes without the formation of any free intermediate (Fig. 1). Furthermore, it is recognized that sulfur species other than thiosulfate can also feed into this pathway at appropriate intermediate states via enzymatic or nonenzymatic conjugations to SoxY, while the intermediates remain covalently bound to the aforesaid cysteine residue (Fig. 1) (Friedrich et al., 2001; Sauvéet al., 2007).

Figure 1.

 Oxidation of thiosulfate and other reduced sulfur compounds by the Kelly–Friedrich pathway as established in Paracoccus spp. (Kelly, 1989; Kelly et al., 1997; Friedrich et al., 2001). Though cysteine residue of SoxY has been shown in its ionized state, the reactive form of this residue in the SoxAX-catalyzed reaction is yet to be confirmed and could be protonated (Sauvéet al., 2007). The electrons released in the reactions are accepted by a small c-type cytochrome. Sulfur species other than thiosulfate could feed into the Sox pathway at the appropriate intermediate state via either enzymatic (as for HS) or nonenzymatic (as for S0) conjugation to SoxY. In organisms lacking a SoxCD complex, the sulfur atom of the sulfane intermediate (SoxYZ)–S–S is plausibly fed into other sulfur oxidation pathways or into storage forms of sulfur. Long-chain sulfur compounds could be assembled on the SoxY carrier cysteine during thiosulfate oxidation if SoxAX is able to act upon sulfane intermediates such as (SoxYZ)–S–S and if the sulfane adduct interacts in a random order with partner proteins rather than being directed to react with SoxCD (Sauvéet al., 2007). The same could also take place in organisms lacking SoxCD (Sauvéet al., 2007).

The following few lines describe in a nutshell the mode of action of the Sox (or the erstwhile TOMES) complex (Kelly et al., 1997; Friedrich et al., 2001), which is now referred to as the ‘Kelly–Friedrich pathway’ (Bamford et al., 2002; Sauvéet al., 2007): initially, the heterodimeric triheme-containing SoxXA cytochromes oxidatively couple the sulfane sulfur of thiosulfate to a SoxY-cysteine-sulfhydryl group of the SoxYZ complex to form a cysteine S-thiosulfonate (thiocysteine sulfate) derivative (Appia-Ayme et al., 2001; Friedrich et al., 2001; Quentmeier & Friedrich, 2001), from which the terminal sulfone (-SO3) group is subsequently released as sulfate by the activity of the hydrolase SoxB (Quentmeier et al., 2003). SoxXA exhibits unique cysteine persulfide coordination to the heme group at its active site (Bamford et al., 2002), while the S–S bond formation involves both the copper and heme centers (Kappler et al., 2008). Eventually, the sulfane sulfur (–S) of the residual SoxY-cysteine persulfide is further oxidized to cysteine-S-sulfate by Sox(CD)2, from which the sulfonate moiety is again hydrolyzed by SoxB, regenerating SoxYZ in the process (Friedrich et al., 2001). The eight electrons released in these two oxidative steps are transferred to a small c-type cytochrome for delivery to the electron transfer chain.

In the course of further research, the Sox complex has been found widely distributed in Bacteria (Friedrich et al., 2001), even though its viability has so far been established only in facultatively sulfur-oxidizing organisms growing at neutral or near-neutral pH. Again, besides thiosulfate, this molecular mechanism has been proved to oxidize other reduced sulfur species such as sulfide (HS), elemental sulfur (S8), sulfite (HSO3) and tetrathionate (O3S–S–S–SO3) (Wodara et al., 1997; Mukhopadhyaya et al., 2000; Appia-Ayme et al., 2001; Lahiri et al., 2006). The modular nature of the Sox system allows such diverse sulfur species to be fed into the pathway as appropriate intermediates (Sauvéet al., 2007) (Fig. 1). In some cases this may result in the formation of adducts containing long chains of sulfur atoms before the terminal sulfane or sulfone groups. There are yet other possibilities that could lead to the accumulation of longer chain sulfur species on SoxY. Because of the modular arrangement of the Sox system, the reactions are not necessarily ordered, and instead of being oxidized by SoxCD, a sulfane intermediate could be oxidatively conjugated to a thiosulfate molecule by SoxXA (Quentmeier & Friedrich, 2001; Sauvéet al., 2007) (Fig. 1). However, irrespective of their origin, long-chain sulfur intermediates can be broken down by repeated iterations of the (SoxCD plus SoxB)-mediated step. In this connection, Sauvéet al. (2007) have aptly pointed out that the SoxYZ complex is the most remarkable and pivotal component of the Sox system, as it participates in every reaction of the pathway. It carries multiple types of sulfur compounds containing either a terminal sulfane or sulfone group, which in turn may contain variable numbers of zero-valent sulfur atoms before them. In addition, SoxY interacts with multiple partner proteins, namely SoxAX, SoxB and SoxCD (or alternatives in organisms lacking this complex; see discussion below), as well as possibly other Sox proteins such as the flavocytochrome SoxFE and thioredoxins. Sauvéet al. (2007) have also determined the crystal structure of the SoxYZ complex from P. pantotrophus and revealed that the SoxYZ complex carries the Sox pathway intermediates on an unusual swinging arm structure in which the carrier cysteine is bracketed by two universal molecular joints. The structure also identifies an apolar pocket that may protect labile pathway intermediates from adventitious reactions. However, it is imperative to mention that the so far elucidated molecular attributes of the Sox constituents (depicted in Fig. 1) (Appia-Ayme et al., 2001; Friedrich et al., 2001; Bamford et al., 2002; Sauvéet al., 2007; Kappler et al., 2008) cannot account for the oxidation of tetrathionate. Friedrich et al. (2001) mooted the role of a previously described enzyme tetrathionate hydrolase (or polythionate hydrolases in general, which are not sox gene products) (Steudel et al., 1987) and explained tetrathionate oxidation by postulating that such hydrolysis yields sulfane-monosulfonates or polysulfide sulfate esters [O3S–S–(S–)x]2− that decompose spontaneously to sulfur and thiosulfate (Steudel, 1989) and can subsequently be oxidized by the Sox complex. Again, Lahiri et al. (2006) have hypothesized that some gene(s) of this operon might have a hitherto unknown attribute of hydrolyzing tetrathionate (O3S–S–S–SO3) at its weakest and most polar S–S bond to produce SO4 and O3S–S–S2− and the latter compound can subsequently be taken care of by the Sox system, but the relevant proteins still remain to be identified.

Albeit coexpressed with the Sox structural genes, the cytochrome c-encoding soxE and flavocytochrome c sulfide dehydrogenase (oxidoreductase) or FCSD-encoding gene soxF, as well as the two thiol esterase-encoding genes soxG and soxH, are reportedly not essential for thiosulfate (or other reduced sulfur compounds) oxidation in P. pantotrophus (Friedrich et al., 2001; Rother et al., 2001). Nevertheless, recent reports on SoxFE function in P. pantotrophus contest these notions (Quentmeier et al., 2004; Bardischewsky et al., 2006b). SoxF homologues, or catalytic flavin subunits of flavocytochrome c (Chen et al., 1994), an enzyme known to oxidize sulfide to elemental sulfur or polysulfides, have been described from a number of sulfur photolithotrophs including A. vinosum and R. sulfidophilum (Brune, 1989, 1995a; Reinartz et al., 1998). SoxE, which contains two c-type heme attachment motifs, has been indicated as the redox partner of the SoxF flavoprotein, and genes coding the two are located side by side in the sox clusters of P. pantotrophus, R. palustris and Chlorobaculum tepidum, whereas they form a separate transcription unit in R. sulfidophilum (Appia-Ayme et al., 2001). Such gene pairs are again located in isolation in the genomes of the phototrophic sulfide oxidizer E. vacuolata (Kostanjevecki et al., 2000). As only some of the sox gene products are required for complete thiosulfate oxidation in the TOMES/Sox system, additional conserved sox genes such as soxEF could be specifically involved in sulfide oxidation. Expression of the sox operon, in its turn, is regulated by the soxSRT cluster, whose products SoxS (a periplasmic thioredoxin), SoxR (a transcriptional repressor belonging to the ArsR family of metalloregulatory proteins) and SoxT (a permease-like transport protein) act in collaboration with the auxiliary proteins, SoxV and SoxW (involved in the biosynthesis or maintenance of the Sox complex) via a mechanism that is yet to be fully understood (Bardischewsky & Friedrich, 2001a, b; Appia-Ayme & Berks, 2002; Rother et al., 2005; Bardischewsky et al., 2006a; Lahiri et al., 2006; Mandal et al., 2007; Orawski et al., 2007).

Complementing the seminal molecular genetic investigations, whole-genome sequence analyses have also contributed significantly in revealing the actual range of sox genomic diversity. sox gene homologues are now known to be ubiquitous in the domain Bacteria but totally absent in Archaea. However, alternative proteins involved in inorganic sulfur metabolism have been described from sulfur-chemolithotrophic archaea (Kletzin et al., 2004). Again in the domain Bacteria, complete versions of the conserved sox operon having all the structural, regulatory and so-called dispensable genes, were detected in the members of Alphaproteobacteria alone. Interestingly, such sox operon-bearing Alphaproteobacteria included chemolithoautotrophs (Magnetospirillum magnetotacticum, Methylobacterium extorquens) photolithoautotrophs (R. palustris), chemolithoheterotrophs (S. pomeroyi) and even nonlithotrophs (Bradyrhizobium japonicum) (Friedrich et al., 2001, 2005; Kaneko et al., 2002; Moran et al., 2004). The sox gene clusters of species outside Alphaproteobacteria, on the other hand, were mostly incomplete and generally deficient in soxCD, for example, abbreviated sox clusters found in the phyla Chlorobi, Aquificae and Deinococcus–Thermus; and many species of Beta- and Gammaproteobacteria (Friedrich et al., 2001, 2005). However, it is interesting to note that whole-genome sequences published very recently have revealed complete sox clusters in several species belonging to the Beta- and Gammaproteobacteria.

While the purple sulfur gammaproteobacterium A. vinosum has five sox genes clustered in two independent loci, soxBXA and soxYZ (Hensen et al., 2006), organization and identities of the sox genes of GSB such as C. thiosulfatiphilum (soxF2XYZAorf106BW) (Vertéet al., 2002) and Chlorobaculum tepidum (soxFXYZAB) (Eisen et al., 2002) are also significantly different from those of photo- or chemolithotrophic Alphaproteobacteria (Friedrich et al., 2001). The truncated Sox systems of the PSB and GSB, proved or putatively, catalyze the oxidation of the sulfone sulfur of thiosulfate (Hensen et al., 2006; Dahl, 2008; Grimm et al., 2008). Want of SoxCD, which oxidizes the SoxY-bound sulfane sulfur to sulfate in Alphaproteobacteria (Friedrich et al., 2001), is apparently responsible for the characteristic formation of elemental sulfur as an intermediate of thiosulfate oxidation by these anoxygenic photolithotrophic bacteria (Dahl, 2008; Grimm et al., 2008). Notwithstanding exceptions such as Thiomicrospira crunogena, the same reason potentially also applies to the formation of sulfur globules during thiosulfate oxidation by soxCD-less chemolithotrophs such as species of Magnetospirillum, Beggiatoa and Thiothrix (Grimm et al., 2008).

Grimm et al. (2008) has further suggested that want of soxCD could be the plausible reason why batch cultures of the facultatively anaerobic chemolithoautotrophic betaproteobacterium Thiobacillus denitrificans (strain DSM 807) accumulates finely dispersed membrane-associated elemental sulfur during nitrate-dependent oxidation of thiosulfate (Schedel & Trüper, 1980). soxXYZAB genes have been found organized in a single cluster in Thiobacillus denitrificans ATCC 25259, and another complement of soxA, as well as soxH, soxE, soxF and soxW are dispersed in the genome (Beller et al., 2006). Although the Thiobacillus denitrificans genome encompasses an ORF corresponding to soxC, which is rather more closely related to sulfite dehydrogenase (sorA) of S. novella than the sulfur dehydrogenase of P. pantotrophus, there is no gene corresponding to the alphaproteobacterial soxD or sorB. However, it is worth remembering that in continuous-flow chemostat cultures with limiting thiosulfate, Thiobacillus denitrificans accumulates no intermediate and the entire supplied thiosulfate ion is stoichiometrically oxidized to sulfate (Justin & Kelly, 1978a, b). As such, sulfur deposition by Thiobacillus denitrificans could also be a function of an atypical response to environmental conditions (including artificial conditions imposed in laboratory experiments), and this unique organism may have potentially different mechanisms for the oxidation of thiosulfate or other reduced sulfur compounds under different oxygen tensions (Beller et al., 2006). This distinctive aspect of sulfur oxidation in Thiobacillus denitrificans is supported by its possession of multiple oxidation pathways for certain sulfur compounds and multiple copies of a number of genes associated with sulfur compound oxidation (Beller et al., 2006). It has also been aptly pointed out by Beller et al. (2006) that although the clustered organization of the genes encoding SoxXA, SoxB and SoxYZ in Thiobacillus denitrificans apparently suggests their active involvement in sulfur oxidation, the low primary sequence identity of these putative proteins with the corresponding sequences of P. pantotrophus could indicate biochemical functions that deviate from those of P. pantotrophus counterparts. Unlike P. pantotrophus and other Alphaproteobacteria, reactions of thiosulfate, sulfite and sulfide oxidation in Thiobacillus denitrificans (and other obligately chemolithotrophic sulfur oxidizers) at least partly involve membrane-associated processes (Kelly, 1989; Beller et al., 2006). Hence, deduction of the functions of the putative sox genes of this bacterium solely on the basis of bioinformatic comparison could be treacherous.

Thiosulfate oxidation via formation of tetrathionate intermediate

Thiosulfate oxidation via formation of tetrathionate intermediate (S4I) is observed in several Beta- and Gammaproteobacteria, particularly the obligately chemolithotrophic species of Acidithiobacillus, Thermithiobacillus and Halothiobacillus (Trudinger, 1961a, b, 1964, 1965; Guay & Silver, 1975; Pronk et al., 1990; Kelly et al., 1997). Similar processes have also been reported from facultatively chemolithotrophic Alphaproteobacteria such as A. acidophilum, Betaproteobacteria such as T. aquaesulis and T. kashmirensis and Gammaproteobacteria such as A. caldus (Meulenberg et al., 1992, 1993a, b; Hallberg et al., 1996; Dam et al., 2007). Although it appears that oxidation of thiosulfate to sulfate via acid-stable intermediates such as tetrathionate is preferred by acidophilic species because thiosulfate is unstable under acidic conditions (Johnston & McAmish, 1973), such explanations cannot justify tetrathionate formation by neutrophilic organisms such as Thermithiobacillus, Halothiobacillus or Tetrathiobacter (Trudinger, 1961a, b, 1964; Wood & Kelly, 1987; Ghosh et al., 2005).

Plausible enzymatic steps of the S4I pathway have been postulated from physiological and biochemical studies with T. tepidarius (Lu & Kelly, 1988a, b), A. acidophilum (Meulenberg et al., 1992, 1993a, b), A. ferrooxidans (Sinha & Walden, 1966; Guay & Silver, 1975), A. thiooxidans (Okuzumi, 1965, 1966a, b), A. caldus (Hallberg et al., 1996; Bugaytsova & Lindström, 2004) and T. kashmirensis (Dam et al., 2007). In all the proposed models of the S4I pathway, thiosulfate is said to be oxidized to tetrathionate in the periplasmic space by the enzymic action of thiosulfate dehydrogenases that use c-type cytochromes as cosubstrates and/or themselves contain c-type hemes. Although several thiosulfate-oxidizing and tetrathionate-forming thiosulfate dehydrogenases (possessing variable structural features) have been identified or characterized from the periplasmic or soluble fractions of thiosulfate-grown cells of both neutrophilic and acidophilic sulfur chemolithotrophs (Silver & Lundgren, 1968; Lu & Kelly, 1988b; Meulenberg et al., 1993a; Visser et al., 1996; Nakamura et al., 2001), no gene encoding this enigmatic enzyme has been identified. However, a membrane-bound, tetrathionate-forming, thiosulfate : quinone oxidoreductase (TQO), involved in sulfur chemolithotrophy (and coupling the sulfur oxidation process to quinone reduction rather than that of cytochromes), has been characterized from the thermoacidophilic archaeon Acidianus ambivalens (formerly known as Desulfurolobus ambivalens; Fuchs et al., 1996) (Muller et al., 2004). Similarly, the gene for a periplasmic thiosulfate dehydrogenase, having a pH optimum of 4.25 and oxidizing thiosulfate up to tetrathionate, has also been characterized from A. vinosum (Hensen et al., 2006).

Although oxidation of tetrathionate, both as a substrate and as an intermediate of thiosulfate oxidation, has been studied extensively, there have been conflicting reports regarding the characteristics of the corresponding enzymes. Whereas a periplasmic localization of this protein has been indicated in species such as A. acidophilum, A. ferrooxidans and A. thiooxidans (Steudel et al., 1987; Meulenberg et al., 1993b; Tano et al., 1996; De Jong et al., 1997a, b), studies with T. tepidarius, A. caldus and T. kashmirensis have shown that tetrathionate hydrolysis yielding sulfite as one of the intermediates takes place in the cytoplasm or in close vicinity to the inside of the cell membrane, followed by oxidation of sulfite to sulfate in the same cellular compartment (Kelly, 1989; Hallberg et al., 1996; Kelly et al., 1997; Dam et al., 2007) (Fig. 2). Yet again, Bugaytsova & Lindström (2004) reported tetrathionate hydrolase to be localized in the periplasmic fraction of A. caldus. Recent biochemical studies together with results of mutation in the molybdoenzyme biosynthetic gene locus in T. kashmirensis has indicated that oxidation of thiosulfate to tetrathionate in this bacterium is rendered by a periplasmic oxidase system, while complete oxidation of tetrathionate requires the participation of active membranes. Sulfite is oxidized in the cytoplasm by a sulfite dehydrogenase involving a ubiquinone–cytochrome b complex transferring electrons from SO32− to oxygen, and impairment of this cytoplasmic event negatively regulates the oxidation of tetrathionate to sulfite (Dam et al., 2007).

Figure 2.

 Chemolithotrophic thiosulfate oxidation by the S4I pathway as proposed in Thermithiobacillus tepidarius (Kelly, 1989; Kelly et al., 1997), and later supported by studies in Acidithiobacillus caldus (Hallberg et al., 1996) and Tetrathiobacter kashmirensis (Dam et al., 2007). Thiosulfate is oxidized to tetrathionate in the periplasmic space while tetrathionate hydrolysis yielding sulfite as one of the intermediates takes place in the cytoplasm or in close vicinity of the inside of the cell membrane, followed by oxidation of sulfite to sulfate in the same cellular compartment. Red and blue circles represent sulfur and oxygen atoms, respectively.

Long-chain sulfur globules are deposited in the periplasm during oxidation of tetrathionate by A. ferrooxidans (Steudel et al., 1987; Pronk et al., 1990). Detailed analysis of the chemical nature of these intermediary forms of sulfur (Steudel et al., 1987) has revealed that they consist of long-chain (even up to S13) polythionates and some S80 elemental sulfur. A set of reactions has been postulated (Steudel et al., 1987) to explain the formation of polythionates and elemental sulfur from tetrathionate with the highly reactive sulfane-monosulfonic acids (HS2SO3) as the key intermediates. Steudel et al. (1987) thus suggested that the hydrolysis of tetrathionate in A. ferrooxidans started with the formation of sulfane-monosulfonate according to the equation S4O62−+H2O→HS2SO3+HSO4.

The unstable sulfane-monosulfonates have been proposed as intermediates in chain elongation reactions (e.g. 2HS2SO3↔HS4SO3+HSO3 and 2HS4SO3↔HS8SO3+HSO3), leading to the production of other polythionates (tri-, penta- and hexathionate). Formation of trithionate from the oxidation of tetrathionate has been reported in A. ferrooxidans (Sinha & Walden, 1966), and a dismutation of tetrathionate to trithionate and pentathionate has been demonstrated in A. thiooxidans (Okuzumi, 1965, 1966a). Elongation of these sulfane-monosulfonic acids eventually leads to the formation of elemental sulfur and sulfite (HS8SO3↔S8+HSO3). While oxidative condensation of two sulfane-monosulfonic acids can form polythionates (2HS2SO3+1/2O2→S6O62−+H2O), the latter can also be formed anaerobically along with hydrogen sulfide (2HS2SO3↔S5O62−+H2S).

Notably, the postulated reactions leading to the formation of polythionates and elemental sulfur from S3-sulfane-monosulfonic acids are reversible, explaining the transient deposition of intermediary sulfur in continuous cultures of A. ferrooxidans, and therefore do not implicate long-chain polythionates or elemental sulfur as obligatory intermediates of tetrathionate oxidation. For the same reason these reactions can also catalyze, in principle, into a cyclic pathway (Pronk et al., 1990) involving formation of trithionate from oxidation of S3-sulfane-monosulfonic acids (Sinha & Walden, 1966) and subsequent hydrolysis of the former to thiosulfate (Okuzumi, 1966b).

Correspondingly, from experiments with intact cells of A. ferrooxidans and A. acidophilum, Hazeu et al. (1988) and Meulenberg et al. (1992) described the process of hydrolytic enzymatic decomposition of tetrathionate. Properties of purified tetrathionate-decomposing enzymes from different sources, viz. A. thiooxidans (Tano et al., 1996), A. ferrooxidans (Sugio et al., 1996; De Jong et al., 1997a) and A. acidophilum (De Jong et al., 1997b), have been found consistent with the attribute of tetrathionate hydrolysis. Again, HS2SO3 has been shown to be able to dissociate to S2SO32−, and both forms chemically react with tetrathionate, leading to the formation of thiosulfate and pentathionate (Wentzien et al., 1994). Yet again, Meulenberg et al. (1992) suggested that tetrathionate hydrolysis formed thiosulfate, sulfur and sulfate in equimolar amounts according to the equation: S4O62−+H2O→S2O32−+S0+SO42−+2H+. Physiological findings of Hallberg et al. (1996) are also consistent with this model. In strains of A. thiooxidans (Tano et al., 1996) and A. ferrooxidans (Sugio et al., 1996) tetrathionate hydrolase is said to decompose tetrathionate to thiosulfate and sulfate by the disproportionation reaction: 4S4O62−+5H2O→7S2O3+2SO42−+10H+.

Although very little is categorically known about the genetic basis of the S4I pathway, all or some of the genes required to assemble a fully functional Sox complex have lately been reported from the genomes of a few (obligately as well as facultatively) sulfur-chemolithotrophic Proteobacteria that involve polythionate intermediates in the oxidation of thiosulfate and sulfur. Discrete soxXYZA, soxB and soxCD clusters have been identified in the gammaproteobacterium T. crunogena (Scott et al., 2006), which oxidizes thiosulfate by depositing sulfur globules outside the cell under low pH and oxygen conditions and also transitorily accumulates sulfite and polythionates as products of oxidation of thiosulfate and sulfur globules (Javor et al., 1990). soxB homologues have also been detected in quite a few species of Halothiobacillus and Thiobacillus (Petri et al., 2001; Meyer et al., 2007) that oxidize different reduced forms of inorganic sulfur under different growth conditions via transient accumulation of polythionates, sulfur or sulfite (Trudinger, 1961a, b, 1964; Kelly & Wood, 2000c). Very recently, our laboratory has also been successful in partially sequencing a sox gene cluster from T. kashmirensis (unpublished observation). Low primary sequence identity of the putative sox gene products of T. kashmirensis with the corresponding Sox proteins of P. pantotrophus suggests that the former may have biochemical functions deviated from those of the P. pantotrophus counterparts. These findings augment the speculation that alternative functions of Sox homologues could well be associated with the different transformations of the S4I pathway.

Contrary to this, it is yet again pertinent to recapitulate that another model S4I-operating bacterium A. ferrooxidans has no sox gene (Muller et al., 2004; Ramírez et al., 2004); instead, its genome encompasses a duplicate set of doxDA genes that are known to encode the two subunits DoxD and DoxA of TQO in A. ambivalens and other archaea (Muller et al., 2004). Although the catalytic involvement of the DoxDA or TQO homologues in the oxidation of thiosulfate by A. ferrooxidans has not yet been experimentally established, a close relationship between the apparently Sox-independent thiosulfate oxidation mechanisms of these two acidophilic prokaryotes has been envisaged (Muller et al., 2004). What is more, a novel cluster encompassing five cotranscribed genes, ISac1, rsrR, rsrS, tetH and doxD, encoding a transposase, a two-component response regulator (RsrR and RsrS), tetrathionate hydrolase and DoxD, respectively, has recently been attributed with the oxidative metabolism of tetrathionate, thiosulfate and pyrite in A. caldus (Rzhepishevska et al., 2007).

Reported possession of Sox homologues by tetrathionate-forming Beta- and Gammaproteobacteria makes it pertinent to check whether any mechanistic correlation was plausible between the Sox proteins and the S4I reactions. Thiol groups located in cell membranes have long been indicated in H. neapolitanus as necessary for the oxidation of thiosulfate to sulfate as well as the initial reactions of tetrathionate oxidation (Trudinger, 1965). Lees (1960) suggested that some cell membrane-seated disulfide group concerned with substrate uptake could react with thiosulfate to form sulfenyl-thiosulfate by the reaction R–S–SR+S2O32−→R–S–S–SO3+RS. A chemical reaction between cystine and thiosulfate to form either cysteine-S-thiosulfate, or cysteine-S-sulfonate and sulfide, has also been reported. According to the above equation, thiols on the bacterial membrane would be formed aerobically from disulfides only after addition of thiosulfate. The possible role of thiol groups in the metabolism of thiosulfate and tetrathionate and an explanation for the quantitative oxidation of thiosulfate to tetrathionate in the presence of excess amounts of thiol-binding reagents were illustrated by the set of five reactions listed below (Vishniac & Trudinger, 1962; Trudinger, 1965) assuming sulfenyl-thiosulfate as the normal precursor of sulfate. A thiosulfate oxidizing enzyme [designated as E in Reactions (1)–(5))] was envisaged to remove two electrons from thiosulfate (possibly cysteine-S-thiosulfate or R-S-thiosulfate) to form an enzyme-bound intermediate [Reaction (1)], which in turn could combine either with another thiosulfate molecule to form tetrathionate [Reaction (2)] or with a thiol [Reaction (3)]. Thiols have also been said to react with tetrathionate by displacing either thiosulfate or sulfite to form sulfenyl-thiosulfate or the next higher homologue, respectively [Reaction (4)]. The sulfenyl-thiosulfate formed by [Reaction (3)] being the normal precursor of sulfate [Reaction (5)], the relative concentrations of specific thiol groups required for Reactions (3) and (4) (said to be located in the more inner quarters of the membrane than the uptake-related disulfide) can determine which way thiosulfate oxidation would proceed after the unstable enzyme-bound thiosulfate is formed. The fate of the enzyme-bound thiosulfate can thus be modified by thiol-binding reagents or by oxidation of thiol groups to disulfides, whereas in the presence of excess amounts of thiol-binding reagents, thiosulfate is quantitatively oxidized to tetrathionate (Vishniac & Trudinger, 1962; Trudinger, 1965 and references therein). Glutathione has also been implicated as the source of the thiol group necessary for Reaction (3), as in its absence thiosulfate oxidation reactions are known to proceed up to tetrathionate formation (Trudinger, 1961a):

image((3) and (5))

Sulfenyl-thiosulfate, the central substrate of this scheme, or cysteine-S-thiosulfate for that matter, can be regarded as analogous to the complex obtained after SoxXA (cytochrome c) oxidatively couples the sulfane sulfur of thiosulfate to a SoxY-cysteine-sulfhydryl group of SoxYZ from where the terminal sulfone group is released by the putative sulfate thiol esterase, SoxB. Again, SoxG and SoxH proteins, though not part of the TOMES complex of Alphaproteobacteria, are also potential thiol esterases (Friedrich et al., 2001). Sox proteins of Beta- and Gammaproteobacteria have characteristically low levels of identity with the typical alphaproteobacterial homologues involved in the Sox pathway. This indicates that many of them could encode functions significantly modified from the typical activity of their alphaproteobacterial counterparts. In such a scenario, if any of these three thiol esterase-like Sox proteins in tetrathionate-forming bacteria accomplish specific conversion of the thiol R1S to disulfides, this might be crucial in determining whether any thiosulfate oxidation process would proceed in the way it is observed in Alphaproteobacteria or move towards S4I formation. However, the correctness of this speculation remains wide open for experimental verification.

The branched thiosulfate oxidation pathway

The branched thiosulfate oxidation pathway (Fig. 3), comprehensively characterized from the model PSB A. vinosum, is rendered through the interaction of two spatially separated and characteristically distinct enzyme systems (Grimm et al., 2008). This mechanism has been arguably envisaged as the ground strategy of thiosulfate oxidation in lithotrophic bacteria that form conspicuous globules of polymeric, water-insoluble sulfur during this process (Dahl, 2008). These apparently include all the anoxygenic photolithotrophic GSB and PSB; the free-living as well as symbiotic strains of certain sulfur-chemolithotrophic bacteria such as Beggiatoa, Thiothrix and Thiocapsa roseopersicina (Dahl, 1999; Brüser et al., 2000a; Dahl & Prange, 2006); and ostensibly Thiobacillus denitrificans, which accumulates elemental sulfur (albeit not in the microscopically visible form) during anaerobic oxidation of thiosulfate in batch cultures (Schedel & Trüper, 1980) but not continuous-flow chemostat cultures (Justin & Kelly, 1978a, b).

Figure 3.

 The emerging mechanism of the ‘two-staged’ branched thiosulfate oxidation pathway in sulfur-storing anoxygenic photolithotrophic as well as facultatively anaerobic chemolithotrophic bacteria. The initial oxidation and covalent binding of thiosulfate to SoxYZ is envisaged as carried out by SoxXA, and SoxB would then hydrolytically release sulfate as in the Sox-mediated pathway (Hensen et al., 2006). As these organisms lack SoxCD, the sulfane sulfur atom still bound to SoxY cannot be directly oxidized further and is instead transferred to growing sulfur globules in the bacterial periplasm (Pattaragulwanit et al., 1998; Dahl & Prange, 2006). Extracytoplasmic sulfur globules, which are enveloped by the proteins SgpA, SgpB and SgpC (Pattaragulwanit et al., 1998), are reductively activated and transported to the cytoplasm, where they are further oxidized to sulfite by the reverse activity of DsrAB (Pott & Dahl, 1998; Dahl et al., 2005). The most up-to-date model for the interaction of Dsr proteins to achieve the oxidation of globular sulfur to sulfite (Dahl et al., 2005; Dahl, 2008; Grimm et al., 2008) states that DsrAB specifically interacts with DsrL on the one hand, and DsrMKJOP and DsrEFHC on the other. Electrons released from the oxidation of sulfide by DsrAB are fed into the photosynthetic electron transport via DsrC and DsrKMJOP with DsrJ as the terminal acceptor. Finally, sulfite is thought to be oxidized to sulfate by the sequential reverse action of the cytoplasmic AMP-dependent APS reductase and ATP sulfurylase (Brune, 1995b; Brüser et al., 2000a; Beller et al., 2006), although in quite a few phototrophic organisms a unique cytoplasmically located AMP-independent SAOR might essentially render this final oxidative step (Dahl, 1996; Sanchez et al., 2001). The invaluable perspectives and generous help extended by Dr Christiane Dahl toward the construction of this figure are deeply acknowledged.

Intermediary formation and deposition of sulfur during thiosulfate (and also sulfide) oxidation by these bacteria takes place either inside (e.g. Beggiatoa spp. and Chromatiaceae members) or outside (e.g. Chlorobi and Ectothiorhodospiraceae members) the cells. Laser Raman microprobe spectroscopy and laser scanning confocal microscopic studies to determine the presence and speciation of sulfur in marine bacteria such as the large, filamentous Thioploca and Beggiatoa, and the endosymbionts in the vesicomyid clam Calyptogena kilmeri, have shown that these organisms store elemental sulfur in vesicles that are submicrometers to several micrometers in diameter (Pasteris et al., 2001). While elemental sulfur in these vesicles is bonded in the common stable S8 ring configuration (Prange et al., 2002) and is of an extremely fine-grained microcrystalline solid form, liquid-like elemental sulfur or homogeneous complex sulfur compounds have also been reported from other organisms (Pasteris et al., 2001).

Early experiments involving strains of C. parvum grown on thiosulfate differentially labeled at the sulfane and sulfone sulfur positions had revealed an initial reductive cleavage of thiosulfate to sulfide and sulfite (Khanna & Nicholas, 1982). Studies with radioactive-labeled thiosulfate in PSB had also demonstrated that the more-reduced sulfane and the more-oxidized sulfone sulfur atoms are oxidized by different pathways (Smith & Lascelles, 1966; Trüper & Pfennig, 1966) and only the sulfane sulfur accumulates as stored sulfur [S0] before being further oxidized, whereas the sulfone sulfur is rapidly converted into sulfate in the periplasm. These findings underscored the existence of more than one mechanism for sulfide oxidation in these bacteria.

At the then point of perception it seemed that this kind of thiosulfate oxidation plausibly involved reductive cleavage of the molecule by the action of rhodaneses or thiosulfate reductases (Brune, 1989, 1995b; Dahl, 1999; Brüser et al., 2000a), both of which can function as thiosulfate sulfur transferase and release sulfite and sulfide in the presence of suitable reduced thiol acceptors such as glutathione or dihydrolipoic acid. However, the physiological role of thiosulfate : sulfur transferases in chemolithotrophic sulfur oxidation was never ascertained at the molecular level. Instead, accumulating molecular and genomic evidence pointed toward the involvement of a truncated Sox system (encompassing SoxXAYZB and characteristically lacking SoxCD) in the initial step of thiosulfate oxidation by the aforesaid bacteria (Petri et al., 2001; Eisen et al., 2002; Vertéet al., 2002; Hanson & Tabita, 2003; Beller et al., 2006; Meyer et al., 2007). However, it was only from seminal experiments in A. vinosum that the involvement of SoxXAYZB in the oxidation of the sulfone sulfur of thiosulfate was definitely established (Hensen et al., 2006). Consequently, deficiency of soxCD in A. vinosum, and logically in the other above-mentioned species, was correlated with the formation of elemental sulfur as an intermediate during thiosulfate oxidation by these organisms (Dahl, 2008; Grimm et al., 2008).

The branched thiosulfate oxidation scheme thus envisages that in the absence of SoxCD these anaerobic bacteria cannot directly oxidize the sulfane sulfur species any further and instead transfer it to growing sulfur globules transiently deposited in the periplasm, the same cellular compartment as that of the Sox proteins (Pattaragulwanit et al., 1998; Dahl & Prange, 2006; Hensen et al., 2006). The mechanism of transfer of SoxY-bound sulfane-sulfur to the growing sulfur globules is still unclear, but so far as the fate of the latter is concerned, it has long been pointed out (Kelly, 1989) that no sulfur oxygenase or dehydrogenase can operate in organisms that grow in the absence of oxygen, and hence the enzyme involved in the oxidation of sulfide, elemental sulfur, as well as the sulfane sulfur of thiosulfate or tetrathionate, to sulfite in Thiobacillus denitrificans could be a ‘siroheme sulfite reductase’ operating in its reverse (oxidative) direction: S2−+H2O→SO32−+6H++6e. Clinching evidence in this direction, however, came from a series of seminal research works by C. Dahl and her colleagues with A. vinosum as the model system. Dahl & Trüper (1994) first proposed a general scheme for dissimilatory sulfur metabolism in anaerobic phototrophic bacteria, where elemental sulfur (in the same way as sulfide) was oxidized to sulfite and then to sulfate as the final product. Subsequent studies showed that anaerobic oxidation of sulfide and sulfur in A. vinosum was brought about by the reverse-acting DsrAB (Pott & Dahl, 1998). Demetallated siroheme-binding proteins similar to DsrAB were already known to mediate the reduction of sulfite to sulfide in many sulfate-reducing prokaryotes (Wolfe et al., 1994; Marritt & Hagen, 1996; Hipp et al., 1997). The prosthetic siroamide-[Fe4S4] group of DsrAB is an amidated form of the typical siroheme, and the protein from A. vinosum is a cytoplasmic α2β2-structure encoded from the dsrABEFHCMKLJOPNRS operon along with 13 other proteins (Dahl et al., 2005; Dahl, 2008). All these proteins are either cytoplasmic or membrane-bound and cannot act directly on the extracytoplasmic sulfur globules, leading to the postulation that sulfur needs to be reductively activated, transported to and further oxidized in the cytoplasm (Pott & Dahl, 1998; Dahl et al., 2005). Accordingly, in the ensuing molecular model(s) (Pott & Dahl, 1998) sulfur globules were considered to be enveloped by the proteins SgpA, SgpB and SgpC (Pattaragulwanit et al., 1998), whereas sulfite reductase was said to bind siroheme prosthetic groups, allowing electron flow from the cytoplasm to the integral membrane protein DsrM via the iron–sulfur protein DsrK [a more detailed account of this mechanism (described below and shown in Fig. 3) has very recently been proposed and suggested as the most apposite model for the oxidation of stored sulfur; Grimm et al., 2008]. Pott & Dahl (1998) further illustrated that stored sulfur was reduced to the level of sulfide that could diffuse over the cytoplasmic membrane as either H2S or transported in an unknown form by an unknown mechanism, while DsrK, and then DsrM, plausibly accepted electrons from DsrAB, feeding it into the photosynthetic electron flow. Alternatively, it was also propounded (Pott & Dahl, 1998) that shuttling of sulfur from the extracytoplasmic globules to the cytoplasm could take place via a perthiol acting as a carrier molecule, and DsrM and DsrK could be involved in the regeneration of a thiol and sulfide from the hypothetical perthiol in the cytoplasm. Nevertheless, the process by which the substrate HS was made available from extracytoplasmically stored sulfur and how electrons resulting from the sulfite reductase reaction are fed into the electron transport chain remained unclear.

Subsequent investigations revealed that dsrABCNMKJOP represents a core molecular unit common to both sulfur oxidizers and sulfate reducers, whereas other dsr genes are specific for either sulfur-oxidizing (e.g. dsrEFH and dsrL) or sulfate-reducing (e.g. dsrD) bacteria (Grimm et al., 2008). At the same time, an absolutely essential function (linked with DsrAB) of the tightly held transmembrane complex DsrMKJOP was also demonstrated (Sander et al., 2006). Although the precise roles of DsrJ, DsrO and DsrP are yet to be determined, the first possibility proposed by Pott & Dahl (1998) has recently been substantiated by accommodating an essential participation of these transmembrane proteins in conjunction with the NADH : acceptor oxidoreductase activity of the iron–sulfur flavoprotein DsrL (Dahl et al., 2005; Dahl, 2008; Grimm et al., 2008). In this current model, electron is thought to flow through the DsrMKJOP complex from the cytoplasm into the periplasm (Fig. 3). The gene dsrL, which is essential for the oxidation of stored sulfur (Lübbe et al., 2006), encodes a cytoplasmic iron–sulfur flavoprotein having proven NADH : acceptor oxidoreductase activity and putative potentials for the reduction of disulfidic and persulfidic compounds by NADH (Dahl, 2008). Moreover, DsrL and DsrAB were copurified from the soluble fraction of A. vinosum, which together with the above-mentioned information helped attribute to DsrL the possible function of reductive release of sulfide from the putative carrier molecule that transports sulfur from the periplasmic globules to the cytoplasm (Dahl et al., 2005), followed by direct transfer of the sulfur onto DsrAB (Dahl, 2008; Grimm et al., 2008). Again, glutathione amide has been conjectured as a likely candidate carrying out this transportation of sulfur across the membrane into the cytoplasm and the whole process is regarded as more efficient (Dahl, 2008; Grimm et al., 2008) than the transport of sulfur as gaseous H2S proposed earlier (Pott & Dahl, 1998).

While copurification (from the membrane) of DsrAB and the transmembrane electron-transporting complex DsrMKJOP pointed out that the former interacted with membrane-bound Dsr proteins, detection of the other soluble cytoplasmic proteins DsrC and DsrEFH in the same fraction indicated the interactions of DsrC and DsrEFH with DsrAB (Dahl et al., 2005). On compiling all these contingent information it was concluded that DsrAB specifically interacts with the soluble protein DsrL on the one hand and with DsrMKJOP and DsrEFHC on the other, and electrons released from the oxidation of sulfide by DsrAB could be fed into the photosynthetic electron transport via DsrC and DsrKMJOP, a conduit that is analogous to the pathway postulated for sulfate reducers, operating in the reverse direction (Dahl, 2008). Under this scheme, DsrM could operate as a quinone reductase, DsrP as a quinol oxidase and the c-type cytochrome DsrJ as the terminal acceptor (Dahl et al., 2005) from where electron is transferred to the photosynthetic reaction center via high-potential iron protein (Grimm et al., 2008). It is, however, noteworthy that electron flow through the DsrMKJOP complex from the cytoplasm into the periplasm is not in accordance with the very low redox potential of the c-type hemes in DsrJ from sulfate reducers such as Desulfovibrio sp. (Pires et al., 2006), and although the potential for the A. vinosum DsrJ heme is yet to be determined, electron transfer from the periplasm to the cytoplasm, i.e. from DsrJ through the membrane to DsrK, could be equally (if not more) probable (C. Dahl, pers. commun.).

Condensing the above discussion, the essentials of the branched thiosulfate oxidation pathway may be summed up as illustrated in Fig. 3. Corresponding to the Sox-based mechanism typical of Paracoccus spp. (Friedrich et al., 2001), the SoxXA in these bacteria oxidatively couple thiosulfate to a cysteine-sulfhydryl group of the SoxYZ complex, from which sulfate is hydrolyzed by SoxB. Further oxidation of the sulfane sulfur of thiosulfate involves its transfer to the growing sulfur globules, followed by the oxidation of the latter to sulfite by Dsr proteins (Dahl, 1996; Kappler & Dahl, 2001; Dahl et al., 2005). Eventually, sulfite is indirectly oxidized to sulfate via the sequential reverse activities of (1) APS reductase (Apr), which catalyzes the oxidative binding of sulfite to AMP and generation of APS as product and (2) ATP : sulfate adenylyltransferase (ATP sulfurylase) or adenylylsulfate : phosphate adenylyltransferase (APAT), earlier known as ADP sulfurylase (Brüser et al., 2000a, b). In a second step, the AMP moiety of APS is transferred either to pyrophosphate by ATP sulfurylase, or to phosphate by APAT, resulting in the formation of ATP or ADP, respectively. Both ATP sulfurylase and APAT thus release sulfate from APS, and energy is produced by substrate phosphorylation either when the former enzyme transfers the AMP moiety of APS onto pyrophosphate generating ATP or when the latter replaces the sulfate by phosphate and produces ADP (Brune, 1995b; Brüser et al., 2000a, b).

Some or all of the above-mentioned enzymes or enzyme systems are present in several strains of sulfur-storing, soxCD-less, anaerobic anoxygenic phototrophic sulfur-oxidizing bacteria (e.g. A. vinosum, T. roseopersicina, Chlorobaculum spp., etc.) as well as facultatively aerobic chemolithotrophs such as Thiobacillus denitrificans, marine Beggiatoa, invertebrate symbionts and their free-living relatives (Brune, 1995b; Nelson & Fisher, 1995; Visser et al., 1997b; Dahl et al., 1999; Brüser et al., 2000a, b; Sanchez et al., 2001; Teske & Nelson, 2004; Beller et al., 2006; Dahl, 2008). Meyer et al. (2007) and Meyer & Kuever (2007a, b) have recently presented an extensive phylogenomic survey of the distribution of sox, sor, apr and dsr gene homologues that, respectively, encode the Sox enzyme system (SoxXAYZB±CD), sulfite dehydrogenase (SorAB), dissimilatory APS reductase (AprBA) and sulfite reductase (DsrAB, together with its functionally associated transmembrane complex DsrMKJOP) in taxonomically diverse bacteria. Interestingly, distribution of apr genes among the anoxygenic phototrophs is quite restricted, and despite their occurrence among most of the Chromatiaceae members (except marine Isochromatium and Marichromatium), they are totally absent in the members of Ectothiorhodospiraceae and selectively present in certain taxonomic subclusters of Chlorobi. What is more, reverse Apr activity is reportedly a dispensible mechanism of sulfite oxidation in A. vinosum and under photolithoautotrophic growth conditions, the same transformation is said to rely primarily on the activity of a cytoplasmically located sulfite : acceptor oxidoreductase (SAOR) (Dahl, 1996; Sanchez et al., 2001), which is also a molybdenum-containing protein like the prototypical periplasmic SorAB of S. novella (Kappler & Dahl, 2001; Kappler & Bailey, 2005). Absence of apr genes in several GSB and PSB in tandem with the detection of SAOR activity from cell extracts of some Marichroatium species (Trüper & Fischer, 1982) further support the implication of AMP-independent reactions in the sulfite oxidation by anoxygenic phototrophs (Meyer & Kuever, 2007b). Nonetheless, it is relevant to mull over the fact that the genomic data so far available for sulfur-storing anoxygenic phototrophs or facultatively anaerobic chemotrophs (Eisen et al., 2002; Beller et al., 2006; Meyer et al., 2007) identify no SAOR gene in these organisms. Consequently, how species belonging to Ectothiorhodospiraceae and Chlorobi, which lack APS reductase as well as SorAB, oxidize sulfite remains an interesting question and holds the potential for the discovery of some hitherto unknown sulfite-oxidizing enzyme.

The basic strategy of thiosulfate oxidation in all the sulfur-storing lithotrophs by and large conforms to the broad paradigm of the branched thiosulfate oxidation pathway epitomized in A. vinosum (Dahl, 2008; Grimm et al., 2008), but considerable mechanistic diversity can be expected among the different species or groups of species. For example, sulfur deposition by Thiobacillus denitrificans could be a function of an atypical response to environmental conditions, as this organism may have potentially distinct mechanisms for the oxidation of thiosulfate or other reduced sulfur compounds under different oxygen tensions (Beller et al., 2006). Similarly, a key difference between the thiosulfate oxidation processes of A. vinosum and Chlorobaculum tepidum concerns the formation and oxidation of periplasmic elemental sulfur globules. This process in A. vinosum is absolutely dependent on the sulfur globule proteins SgpA, SgpB and SgpC (Pattaragulwanit et al., 1998; Prange et al., 2004). Although Chlorobaculum tepidum accumulates elemental sulfur extracellularly, it does not have sgp gene homologues, and yet grows perfectly well on thiosulfate and sulfide (Chan et al., 2008b).

Notably, so-called sulfur islands have been postulated to govern the oxidation of thiosulfate and other sulfur compounds in Chlorobaculum tepidum. As mentioned earlier, this organism possesses a partial sox gene cluster, in addition to which it has another major sulfur island that encompasses a 32-kb genomic locus encoding homologues of the Dsr complex, APS reductase and ATP sulfurylase, Qmo (implicated in sulfate reduction; Pires et al., 2003), thioredoxin reductase, rhodanese-like protein and other catalytically crucial novel proteins (Eisen et al., 2002; Chan et al., 2008a, b). Two models have been proposed for how Chlorobaculum tepidum harvests the rest of the six electrons in the absence of soxCD. The one proposed from genomic evidence (Eisen et al., 2002) envisages cleavage of SoxY-bound sulfur to free sulfide, followed by periplasmic oxidation of the resulting sulfide by SoxF flavocytochrome c or SQR. The second model, proposed by Hanson & Tabita (2003), invokes transfer of the SoxY-bound sulfur to an unknown low-molecular-mass thiol for subsequent transport and cytoplasmic oxidation.

Oxidation of elemental sulfur as a ‘bottleneck’ in the sulfur cycle

Oxidation of elemental sulfur is also one of the most energy-yielding reactions in several lithotrophic organisms. As discussed above, intermediary production of elemental sulfur during microbial oxidation of reduced sulfur compounds is a widespread phenomenon. Anaerobic, anoxygenic and phototrophic species of Chlorobi, Chromatiaceae and Ectothiorhodospiraceae, along with some aerobic chemotrophic species of Beggiatoa, Thiothrix, Thiobacillus, Thiomicrospira and free-living relatives of invertebrate symbionts, form intra- and extracellularly stored sulfur globules as an intermediate during thiosulfate oxidation (Nelson & Fisher, 1995; Howarth et al., 1999; Imhoff, 1999, 2001a, 2003; Robertson & Kuenen, 2002; Teske & Nelson, 2004). Again, neutrophilic Betaproteobacteria such as T. thioparus (reported to have soxB in its genome) and Thiobacillus denitrificans (having soxXAYZB and dsrAB+dsrMKJOP) profusely deposit elemental sulfur when grown on ammonium thiocyanate (Kelly & Wood, 2000b). The acidophilic obligate chemolithoautotroph A. ferrooxidans also produces elemental sulfur during the oxidation of thiosulfate, trithionate, tetrathionate and sulfide (Pronk et al., 1990). Sulfur is produced from sulfide by an oxidative step, whereas its production from tetrathionate is initiated by a hydrolytic step, followed by a series of less-understood chemical reactions (Hazeu et al., 1988). These facts are all the more quizzical in view of the fact that the A. ferrooxidans genome harbors no sox or dsr homologue. The deltaproteobacterium D. sulfoexigens can grow chemolithotrophically by disproportionation reaction of elemental sulfur (Finster et al., 1998), and though the initial step of this process has not been identified, dissimilatory sulfite reductase detected in sulfur-disproportionating cultures could be responsible for the same through its reverse oxidative activity (Frederiksen & Finster, 2003).

Although the envisaged mechanism involving the partial Sox system (minus SoxCD) in conjunction with the Dsr proteins (Dahl et al., 2005) succeeds in explaining the formation and subsequent oxidation of sulfur globules in some Beta- and Gammaproteobacteria as well as Chlorobaculum spp., the mechanism of oxidation of sulfane sulfur to sulfate by chemotrophic bacteria that neither possess soxCD or dsr genes, nor deposit sulfur, remains a thought-provoking question (Friedrich et al., 2005). On the other hand, the aerobic obligately sulfur-chemolithotrophic gammaproteobacterium T. crunogena and the versatile epsilonproteobacterium S. denitrificans possess arrays of sox genes not organized as operons but dispersed in their genomes as discrete sox clusters (Scott et al., 2006; Sievert et al., 2008). Thiomicrospira crunogena also possesses a putative sulfide : quinone reductase (sqr) gene that catalyzes the oxidation of sulfide to elemental sulfur in several lithotrophic bacteria. Intriguingly, despite the presence of sox homologues including soxCD, T. crunogena, which has neither dsr nor apr genes, deposits sulfur globules outside the cell under low pH and oxygen conditions (Javor et al., 1990). Respiring cells of this chemolithotrophic bacterium produce sulfur globules from the sulfane sulfur of thiosulfate <pH 7, and consume the globules >pH 7. The switch in metabolism is said to be immediate and reversible upon titration of the culture. The consumed sulfur globules remain in a membrane-bound form and are reportedly not oxidized unless the medium is depleted of thiosulfate. Thiol-binding agents and inhibitors of protein synthesis reportedly block globule uptake. Transitory accumulations of sulfite and polythionates are said to be the reaction products of thiosulfate and sulfur globules (Javor et al., 1990). How Sox proteins in this bacterium render oxidation of thiosulfate and sulfide via sulfur deposition is an interesting question.

Again, the epsilonproteobacterium ‘Candidatus Arcobacter sulfidicus,’ isolated from coastal marine sediments having oxygen-sulfide chemocline (Wirsen et al., 2002), is capable of mesophilic, chemolithoautotrophic growth on sulfide with oxygen (and sometimes nitrate and elemental sulfur) as the electron acceptor, yielding filamentous sulfur. Although the metabolic attributes of most of the Acrobacter spp. are poorly understood, many of the cultured representatives are from marine environments with a well-defined geochemical interface between dissolved oxygen and sulfide concentrations. Production of filamentous sulfur mats by both vibrioid and filamentous sulfur-oxidizing chemolithoautotrophic Epsilonproteobacteria (Taylor & Wirsen, 1997; Taylor et al., 1999) closely related to the genus Arcobacter (Wirsen et al., 2002) is an important phenomenon in deep-sea hydrothermal vents. In situ filamentous sulfur formation in marine habitats is very much analogous to the process of sulfur production under laboratory conditions by enrichment cultures of a chemolithoautotrophic sulfur-oxidizing bacteria isolated from coastal marine seawater (Taylor & Wirsen, 1997). Production of filamentous sulfur might also facilitate colonization of surfaces in marine habitats by ‘C. Arcobacter sulfidicus’ (Moyer et al., 1995).

A biochemical model for the oxidation of sulfur in Acidithiobacillus and Acidiphilium has been proposed by Rohwerder & Sand (2003) where extracellular elemental sulfur containing stable octasulfane ring system, and forming poorly soluble orthorhombic crystals (Steudel, 2000), is nonenzymically activated by glutathione (GSH) to yield glutathione persulfide (GSSH) and mobilized as persulfide sulfane sulfur by the action of the thiol groups of specialized outer-membrane proteins. The sulfane sulfur of this GSSH, but not free sulfide, is oxidized by periplasmic sulfur dioxygenase and the resulting sulfite is oxidized to sulfate by a SAOR, which probably uses cytochromes as electron acceptors. Free sulfide is oxidized by a separate periplasmic dehydrogenase, viz. sulfide : quinone oxidoreductase (SQOR), which uses quinones as electron acceptors. Notably, the accepted reaction mechanism for SQOR also involves free polysulfides, rather than elemental sulfur, as the initial oxidation product (Griesbeck et al., 2002). The water-soluble polysulfides can reportedly be mobilized enough to cross the outer membrane, whereas elemental sulfur precipitates in the periplasm, explaining why elemental sulfur is only formed extracellularly in species such as R. capsulatus that do not oxidize sulfide further to sulfite or sulfate (Griesbeck et al., 2002). However, as polysulfides are not stable in acidic and neutral solutions and immediately decompose to elemental sulfur and sulfide (Steudel, 1996, 2000), elemental sulfur would accumulate in the periplasm even if polysulfides were the initial product of sulfide oxidation. In the model proposed for Acidithiobacillus and Acidiphilium, the zero valence sulfur formed from sulfide, whether as the initial product or after decomposition of polysulfides, does not precipitate because it reacts with the thiol groups of the outer-membrane proteins and forms persulfide sulfur. If no further oxidation occurs, the persulfide sulfur is possibly transported out of the cell through the same reversible transport mechanism that facilitates the entry of extracellular sulfur to the cell, an argument supported by the fact that inhibition of complete sulfide oxidation in Acidithiobacillus causes extracellular accumulation of elemental sulfur (Chan & Suzuki, 1993).

Oxidation of sulfide

Anaerobic oxidation of hydrogen sulfide to sulfur in the presence of sunlight is a characteristic of phototrophic purple nonsulfur Alphaproteobacteria such as R. capsulatus (Schutz et al., 1997, 1999) and GSB such as Chlorobium luteolum (formerly Pelodictyon luteolum; Imhoff, 2003) and species of Chlorobium and Chlorobaculum (Shahak et al., 1992). Whereas R. capsulatus utilizes hydrogen sulfide as the only sulfur substrate for phototrophic growth and oxidizes it only up to sulfur, PSB such as A. vinosum completely oxidizes hydrogen sulfide to sulfate (Friedrich, 1998). Oxidation of hydrogen sulfide to sulfur has been attributed to FCSD (Fukumori & Yamanaka, 1979; Visser et al., 1997a) as well as SQR (Schutz et al., 1997, 1998), with the two enzymes being reported from both phototrophic and chemotrophic sulfur oxidizers. Interestingly, cytochromes without flavin groups have also been proposed to mediate electron transfer from sulfide to the reaction center in some PSB (Brune, 1989). In A. vinosum phototrophic growth by oxidation of hydrogen sulfide to sulfate essentially depends upon SQR, whereas flavocytochrome c is said to be dispensable (Reinartz et al., 1998). In R. sulfidophilum, on the other hand, the Sox system is indispensable for in vivo oxidation of sulfide (Appia-Ayme et al., 2001). SQR (Schutz et al., 1997) is also essential for phototrophic growth of R. capsulatus with the partial oxidation of hydrogen sulfide to sulfur (Schutz et al., 1999). Again, SQR has been characterized from Chlorobium spp. (Shahak et al., 1992) and, more interestingly, from the cyanobacterium Oscillatoria limnetica (Arieli et al., 1991, 1994).

Again, the preferred sulfur substrate utilized by symbiotic bacteria as energy source is sulfide (Cavanaugh et al., 2004). Bathymodiolus thermophilus and Calyptogena magnifica, which detoxify sulfide by conversion to the less reduced sulfur compound thiosulfate, directly supply the latter to their symbiotic partners (Nelson & Fisher, 1995; Cavanaugh et al., 2004). SQR activity has also been detected in the aerobic chemolithotroph P. pantotrophus (Schutz et al., 1998), even though the enzyme does not account for chemolithotrophic oxidation of hydrogen sulfide in this bacterium, this being rather a function of the Sox complex (Friedrich et al., 2001). Again, an FCSD homologue, SoxFE, is present in the sox operon of P. pantotrophus, and similar flavoproteins are observed in isolation in the genomes of different sulfur-oxidizing bacteria (Kostanjevecki et al., 2000). Interestingly, although SoxFE is not essential for oxidation of reduced sulfur compounds by P. pantotrophus (Friedrich et al., 2001), the monomeric form of SoxF alone has recently been shown to possess sulfide dehydrogenase activity, i.e. sulfide-dependent cytochrome c reduction at the optimum pH of 6.0, which is reportedly inhibited by sulfur and cyanide, besides being fully inactivated by sulfite (Quentmeier et al., 2004). A strong cofactor interaction with the apoprotein and an activation/variation of the protein during the redox cycles have also been observed (Quentmeier et al., 2004). Deletion of soxF allegedly lowers the growth rate of P. pantotrophus by >50% with either thiosulfate or sulfide as the substrate (Bardischewsky et al., 2006b). Although SoxF does not affect the thiosulfate-oxidizing activity of the reconstituted Sox enzyme system (Friedrich et al., 2001) it is now said to enhance chemotrophic thiosulfate oxidation in vivo and act on some component or condition present in whole cells and cell-free extracts but not in the reconstituted system (Bardischewsky et al., 2006b).

Oxidation of sulfite to sulfate

Two distinct pathways for the oxidation of sulfite have been identified, one involving APS reductase and ATP sulfurylase and looked upon as the reversion of the initial steps of the dissimilatory sulfate reduction pathway (Dahl, 1996; Brüser et al., 2000a; Sanchez et al., 2001), and the other concerning the direct oxidation of sulfite to sulfate by a type of mononuclear molybdenum enzyme known as sulfite oxidoreductase not found in any sulfate reducer (Kappler & Dahl, 2001). Mononuclear molybdenum enzymes, in their turn, fall into three distinct groups, viz. the xanthine oxidase, sulfite oxidase (SO) and dimethyl sulfoxide reductase families, all of which are associated with the maintenance of redox balance within the cell (Kisker, 2001). The SO family, in particular, comprises both plant assimilatory nitrate reductases and sulfite-oxidizing enzymes found in all the three domains of life, out of which the latter can directly catalyze the two-electron oxidation of the highly reactive, and hence toxic, sulfite to sulfate (Aguey-Zinsou et al., 2003) [SO32−+H2O=SO42−+2H++2e] with oxygen and/or heme-coordinated iron ions as the final electron acceptor (Hille, 1996; Kappler & Dahl, 2001).

All enzymes of the SO family contain a single Moco and catalyze reactions involving the transfer of oxygen atom to or from an available electron lone pair of a substrate (Hille, 1996). Depending on their ability to transfer electrons to molecular oxygen, the sulfite-oxidizing enzymes are further divided into two categories: the eukaryotic SOs and the prokaryotic sulfite dehydrogenases, even though both types are together referred to as SAORs. As such, mainly three types of sulfite-oxidizing enzymes have been characterized so far: the homodimeric, heme b and molybdenum-containing enzymes from humans, rats and birds (Garrett & Rajagopalan, 1994; Kisker et al., 1997); the homodimeric, molybdenum-containing enzymes from plants (Eilers et al., 2001); and a third heterodimeric, heme c and molybdenum-containing bacterial enzyme (Kappler et al., 2000) that cannot transfer electrons to molecular oxygen and is therefore classified as a sulfite dehydrogenase. Notably, the N-terminal heme-binding domains are characteristic features of animal and bacterial SOs, but not the plant counterparts. SO from Arabidopsis thaliana (Eilers et al., 2001), although homodimeric and homologous to animal SOs (Hille, 2003), contains only Moco and no heme-binding domain or any alternatively bound heme, and hence lacks activity with cytochrome c that requires heme as a mediator for electron transfer from the Moco (Eilers et al., 2001). The third type of SO, i.e. the typical bacterial sulfite dehydrogenase SorAB of the chemolithotrophic alphaproteobacterium S. novella, is a heterodimeric complex of a catalytic molybdopterin and a c-type cytochrome subunit (Kappler & Dahl, 2001). It is now proven that S. novella can oxidize sulfite by such a sulfite dehydrogenase encoded by its sorAB genes (Kappler et al., 2001) located distantly from the sox operon (Kappler et al., 2004). Another case of SO diversity is presented by the molybdopterin-containing, Fe/heme-minus, DraSO protein of the polyextremophilic bacterium Deinococcus radiodurans encoded from the so-called draSO gene, which is very similar to the SO of A. thaliana and is not surrounded by any thiosulfate- or respiratory chain-oxidation gene, and has no direct association with such pathways (D'Errico et al., 2006). However, despite considerable similarity to bacterial and animal SO homologues, the A. thaliana SO and DraSO lack the N-terminal heme-binding domain and are consequently incapable of electron transfer from sulfite to cytochrome, indicating that the electron acceptor for their in vivo functioning could be oxygen. Bioinformatic screening of different prokaryotic genomes revealed that several other bacterial SOs lack the heme-binding domain, but none has been biochemically characterized except the one from D. radiodurans (D'Errico et al., 2006).

The lithotrophic alphaproteobacterial molybdoenzyme encoded by the soxC genes is a dehydrogenase containing a Moco-binding domain of the SO family that interacts with a monoheme or diheme cytochrome c SoxD to form the α2β2 heterodimer Sox(CD)2 that mediates a unique oxidative six-electron transfer per mole of thiosulfate, without which only 2 mol of electrons plus sulfur or polysulfide is produced (Friedrich et al., 2001). As no free intermediate is observed in alphaproteobacterial sulfur oxidation, it is presumed that Sox(CD)2 acts upon protein-bound sulfur atoms. Although the Sox multienzyme system of P. pantotrophus oxidizes sulfite in vitro, Sox(CD)2 is apparently not involved in that reaction (Friedrich et al., 2001). Moreover, the structure and function of Sox(CD)2 reportedly differ from those of other prokaryotic SOs (Friedrich et al., 2005), leading to its particular designation as a sulfur dehydrogenase.

Chemolithotrophic utilization of thiocyanate

Processes generating and transforming C1-organosulfur compounds such as thiocyanate (N≡C–S) are environmentally significant. Oxidation of the sulfur atom of thiocyanate, released by hydrolysis via cyanate or carbonyl sulfide, is in essence comparable with the oxidation of sulfide (Kelly & Baker, 1990). Like methylated sulfides, methane sulfonate, carbon disulfide and carbonyl sulfide, the biological and chemical interconversions of thiocyanate are essentially influenced by autotrophic sulfur oxidizers, methylotrophs, methanogens and sulfate-reducing bacteria. Thiocyanate is primarily produced as a waste product of coke and metal plants, and in nature it is formed during biological cyanide detoxification (Kelly & Baker, 1990). Besides these, microorganisms can utilize thiocyanate as an energy, carbon, nitrogen or sulfur source after it is hydrolyzed to sulfide, ammonia and CO2, and complete oxidation of thiocyanate to sulfate, ammonia and CO2 yields eight electrons. Like degradation of other C1 sulfur compounds, CNS degradation requires the primary action of specific enzyme(s) to release the sulfane atom for further microbial oxidation (Kelly & Baker, 1990; Sorokin, 2003).

CNS-containing wastewaters can be treated by acclimated bacterial sludge containing a high density of thiocyanate-oxidizing autotrophs (Hung & Pavlostathis, 1999), or heterotrophs if an alternative carbon source is available (Karavaiko et al., 2000). Such biosystems can remove millimolar amounts of CNS at neutral or alkaline conditions (Sorokin et al., 2001b). In natural alkaline environments such as soda lakes, thiocyanate can be used as the nitrogen source and the energy source under highly alkaline conditions by alkaliphilic obligately organoheterotrophic and obligately lithoautotrophic sulfur-oxidizing bacteria, respectively (Sorokin et al., 2001b).

The research group of D.Y. Sorokin has extensively investigated the alkaliphilic sulfur-oxidizing bacteria and reported several incompletely (e.g. Thioalkalivibrio nitratireducens and nitrite-reducing Thioalkalivibrio denitrificans) (Sorokin et al., 2003) as well as completely denitrifying (Thioalkalivibrio thiocyanodenitrificans) (Sorokin et al., 2004) sulfur-oxidizing bacteria that can utilize thiocyanate as electron donor. Thiohalomonas denitrificans and Thiohalomonas nitratireducens, two novel taxa of obligately chemolithoautotrophic, moderately halophilic, thiodenitrifying Gammaproteobacteria, have recently been identified from hypersaline habitats and have been reported to grow anaerobically as complete denitrifiers, and aerobically under micro-oxic conditions, while sulfate is the final product of thiosulfate and sulfide oxidation, and nitrite and N2O are the intermediates of reduction of nitrate to N2 (Sorokin et al., 2007). Of the two species, Thiohalomonas nitratireducens is additionally capable of complete denitrification of nitrate in the presence of thiocyanate as electron donor above and beyond thiosulfate.

Among the neutrophilic sulfur-oxidizing bacteria, the ability to grow with thiocyanate as an electron donor for energy generation and CO2 fixation was for long limited to a few strains of T. thioparus (Katayama & Kuraishi, 1978; Kelly & Harrison, 1989; Katayama et al., 1992) and Thiobacillus denitrificans (Kelly & Wood, 2000b), but recent studies have reported the same ability for the alphaproteobacterium Paracoccus thiocyanatus and its close phylogenetic relatives (Katayama et al., 1995; Ghosh & Roy, 2007b). A detailed enzymic study of the thiocyanate-oxidizing systems of these Paracoccus species is still wanting. Thiobacillus thioparus and Thiobacillus denitrificans, the two physiologically similar chemolithotrophs share the characteristic property of growth on thiocyanate as a sole source of energy (Kelly & Harrison, 1989; Katayama et al., 1992). It was Beijerinck (1904a) who first reported profuse deposition of elemental sulfur by these bacteria when grown in 0.25% ammonium thiocyanate. Thiocyanate consumption has long been reported as a taxonomic characteristic of both Thiobacillus denitrificans and T. thioparus (Hutchinson et al., 1965, 1967, 1969). However, no quantitative assessment of thiocyanate-dependent growth, or an exact study of the biochemistry of thiocyanate metabolism, has been carried out for Thiobacillus denitrificans, although it has long been shown that under both aerobic and nitrate-dependent conditions thiocyanate is used by Thiobacillus denitrificans and quantitatively converted to sulfate and elemental sulfur (Van Der Walt & De Kruyff, 1955). On the other hand, the two distinct pathways of microbial degradation of thiocyanate with either H2S or NH3 as the first product (Sorokin et al., 2001b) is known from the autotrophic thiocyanate-oxidizing bacterium T. thioparus. It has been postulated that thiocyanate is degraded via cyanate (N≡C-O), which is converted to ammonia and CO2 by the specific enzyme cyanase (Happold et al., 1958). The liberated sulfide is utilized as an electron donor and energy source: CNS+H2O→CNO+H2S. Although the first enzyme in this pathway ought to break the C–S bond, the identity of such enzyme(s) is still uncertain, in addition to which, no direct proof of cyanate being the intermediate during bacterial thiocyanate oxidation has been put forward. However, formation of cyanate from thiocyanate has been reported in a mixed bacterial population from thiocyanate-degrading sludge (Hung & Pavlostathis, 1997). Strains of T. thioparus have also been reported to degrade thiocyanate via carbonyl sulfide (O=C=S) using thiocyanate hydrolase, which has substantial homology to nitrile hydratase (Katayama et al., 1992, 1998), which also breaks nitrile bonds (N≡C). A thiocyanate hydrolase enzyme has recently been characterized as the primary enzyme initiating thiocyanate degradation from a novel obligately chemolithoautotrophic halophilic sulfur-oxidizing bacterium Thiohalophilus thiocyanoxidans (Bezsudnova et al., 2007). By poorly understood mechanisms, the carbonyl sulfide produced is hydrolyzed to sulfide and CO2, with sulfide being eventually oxidized to sulfate. A similar two-stage hydrolysis via carbonyl sulfide has been observed during carbon disulfide (S=C=S) degradation by strains of T. thioparus that are also able to oxidize thiocyanate (Smith & Kelly, 1988) and it has been indicated that hydrolytic cleavage of CS2 and CNS to sulfide proceeds in this bacterium through a common pathway involving carbonyl sulfide as intermediate.

Biochemistry and molecular biology of sulfur oxidation by thermophilic chemolithoautotrophic archaea

Oxidation of elemental sulfur

Oxidation of elemental sulfur is one of the central bioenergetic processes of the thermophilic archaea (Kletzin et al., 2004). The bulk of the information about these pathways is derived from research on the thermoacidophilic archaeon A. ambivalens (formerly Desulfuromonas ambivalens; Fuchs et al., 1996), which grows optimally at 80 °C and pH 2.5 (Kletzin, 1989, 1992; Kletzin et al., 2004; Muller et al., 2004). This organism oxidizes S0 to H2SO4 under aerobic conditions, whereas under anaerobic conditions it uses hydrogen as the electron source for reduction of S0 to hydrogen sulfide (Laska et al., 2003).

Aerobic Archaea such as Acidianus and Sulfolobus use elemental sulfur as the electron donor and oxidize it via sulfite and thiosulfate by means of a pathway involving both soluble and membrane-bound enzymes. This pathway is reportedly coupled to the aerobic respiratory chain linking sulfur oxidation and oxygen reduction at the level of the respiratory heme copper oxidase. Quite the opposite is observed in the facultatively anaerobic chemolithotrophic species of Acidianus and Pyrodictium, where elemental sulfur is the electron acceptor in a short electron transport chain consisting of a membrane-bound hydrogenase and a sulfur reductase.

Thermoacidophilic archaea oxidize S0 with a cytoplasmic sulfur-disproportionating enzyme that is not found in Bacteria (Sun et al., 2003). A sulfur oxygenase was first described from Acidianus brierleyi (Emmel et al., 1986), following which a cytoplasmic sulfur oxygenase-reductase (SOR) catalyzing a sulfur disproportionation coupled to an oxygenase reaction converting sulfur to sulfite, thiosulfate and hydrogen sulfide was described from A. ambivalens (Kletzin, 1989, 1992). The two proteins, which constitute a unique family of low potential mononuclear nonheme iron proteins (Urich et al., 2004, 2005), probably function identically and produce sulfite, hydrogen sulfide and thiosulfate from sulfur and molecular oxygen by the reaction 5S0+O2+4OH→HSO3+S–SO3+2HS+H+. The X-ray crystallographic structure of the SOR from A. ambivalens has been determined, confirming it to be constituted of 24 monomers forming a large hollow sphere enclosing a positively charged nanocompartment with apolar channels providing access for linear sulfur species (Urich et al., 2006). A cysteine persulfide and a low-potential mononuclear nonheme iron site constitute its active sites, accessible from the inside of the sphere, while the iron is thought to be the site of both sulfur oxidation and sulfur reduction. However, no link of this well-studied enzyme with energy metabolism has been reported (Kletzin, 1992). The significance of SOR in energy metabolism, if any, is not clear, as the cytoplasmic homologue of A. ambivalens cannot couple sulfur oxidation to electron transport or substrate-level phosphorylation (Urich et al., 2004). On the other hand, a different SOR homologue described from Acidianus tengchongensis and localized in the membrane alongside active SAORs and thiosulfate : acceptor oxidoreductases is conjectured to possess this attribute (Chen et al., 2005). Genes homologous to these have also been found in the genomes of not only the Archaea Sulfolobus tokodaii and Ferroplasma acidarmanus, but also the hyperthermophilic bacterium Aquifex aeolicus (Friedrich et al., 2005).

However, A. ambivalens reportedly did not grow on thiosulfate as the sole sulfur substrate (Muller et al., 2004), and SOR did not couple S0 oxidation with the aerobic electron transport chain. Consequently, it remained unresolved whether thiosulfate was a primary product of the enzyme SOR or its formation resulted from the rapid non-enzymic reaction between S0 and sulfite (Kletzin, 1989). Moreover, the fate of the products of SOR-mediated S0 oxidation also remained unknown until other enzymes coupling sulfite, thiosulfate or sulfide oxidation to proton transport across the membrane were identified.

Oxidation of thiosulfate and tetrathionate by A. ambivalens

Acidianus ambivalens was subsequently shown to possess a membrane-bound, tetrathionate-forming, TQO catalyzing the oxidation of thiosulfate to tetrathionate (Muller et al., 2004). Although several thiosulfate-oxidizing and tetrathionate-forming thiosulfate dehydrogenases associated with the S4I pathway have been identified or characterized from the periplasmic or soluble fractions of thiosulfate-grown cells of both neutrophilic and acidophilic chemolithotrophs, no molecular genetic basis of this obscure enzyme having variable structural features is yet available. TQO of A. ambivalens, which distinctively couples sulfur compounds oxidation to quinone reduction, is thus the first tetrathionate-forming thiosulfate-oxidizing enzyme to be characterized at the molecular level. The two 16- and 28-kDa subunits of TQO are encoded by a bicistronic operon doxDA, homologues of which are also present in the genomes of Sulfolobus solfataricus and S. tokodaii (Muller et al., 2004).

The model of sulfur oxidation in A. ambivalens (Fig. 4), as propounded by Kletzin et al. (2004), envisages that elemental sulfur is oxidized by a cytoplasmic SOR, yielding thiosulfate, sulfite and hydrogen sulfide. The first two products in their turn are substrates for TQO and SAOR (Zimmermann et al., 1999), which couple their oxidation to energy conservation. Under this scheme APS reductase and APAT are said to be involved in ATP generation from sulfite by substrate-level phosphorylation, while SQORs could catalyze the oxidation of sulfide. Very recently, Kletzin (2008) has updated all this information on the oxidation of sulfur and its inorganic compounds by A. ambivalens. That article should be referred to for the latest appreciation of the molecular mechanism of archaeal sulfur oxidation.

Figure 4.

 Model for sulfur oxidation in Acidianus ambivalens. Reproduced from Kletzin et al. (2004), with the author's consent and kind permission from Springer Science and Business Media.

Wide distribution of sox genes in the domain Bacteria and the place of Sox in the global scheme of sulfur oxidation

Genes encoding the Sox multienzyme complex, or at least some of its components, have now been detected from representatives of almost all the major phylogenetic branches of Bacteria, but not Archaea (Friedrich et al., 2001, 2005; Petri et al., 2001; Meyer et al., 2007). Notably, none of the mechanisms of thiosulfate oxidation found in Bacteria bear any similarity with the function of the thiosulfate-converting enzymes identified from members of Archaea. As such, archaeal sulfur chemolithotrophy could be a convergently evolved process. While most of the sulfur-lithotrophic bacteria possess sox genes, several species, having no sulfur-oxidizing phenotypes, also harbor sox homologues in their genomes. Besides Proteobacteria, several species belonging to the phyla Aquificae, Deinococcus–Thermus, Chlorobi and Spirochaeta possess sox genes or sox gene clusters. Again, within Proteobacteria, such homologues have been reported from species distributed over the classes Alpha, Beta, Gamma, Delta and Epsilon. However, it is noteworthy that the sox operons of the facultatively sulfur-lithotrophic Alphaproteobacteria alone consist of well-defined regulatory genes, whereas equivalent gene clusters of obligate sulfur lithotrophs (photosynthetic or chemosynthetic) have no regulatory element. Expression of sulfur oxidation in obligate lithotrophs is perhaps constitutive without the requirement of any tight regulation. Probably owing to this, sox genes of obligate sulfur oxidizers are often not even organized in a single genomic locus.

The widespread conservation of sox homologues in environmentally diverged sulfur-lithotrophic bacteria coincides with the proven (Hensen et al., 2006; Grimm et al., 2008) or presumed (Scott et al., 2006; Sievert et al., 2008) involvement of SoxXA, SoxYZ, and SoxB±Sox(CD)2 in distinct sulfur oxidation processes. While the detection of soxB genes in sulfur-chemotrophic symbionts such as Endoriftia persephone and Olavius algarvensis (Meyer et al., 2007) is remarkable, their abundance in organic sulfur compounds-degrading Roseobacter spp. (Buchan et al., 2005) is equally interesting. All these findings outwardly encourage the conjecture that the Sox system could be a fundamental and primordial molecular mechanism of sulfur oxidation.

Conversely, it has been argued that Sox-based chemolithotrophy dependent on oxygen or nitrate respiration cannot be ancient because such terminal electron acceptors could not have appeared in the highly reducing environment of the early Proterozoic, the supposed era of origin of sulfur chemolithotrophy (Broda, 1977; Kelly & Wood, 2000a; Meyer et al., 2007). Moreover, such an environment of the Earth as it was then is not considered to support anoxic formation of thiosulfate (from pyrites by abiotic or biotic means) (Canfield & Teske, 1996), the central substrate of the Sox system. In agreement with this, SoxB sequence-based phylogenies (Petri et al., 2001; Meyer et al., 2007) indicate that the homologues of the anoxygenic photolithotrophic GSB and PSB were laterally acquired from different chemotrophic donors and horizontal gene transfer (HGT) could have been the main driver of the global spread of sox genes. What is more, a large majority of the genomically identified sox homologues are functionally nonviable, while the putatively translated polypeptide sequences of others have such low levels of identity with the Sox proteins of Alphaproteobacteria that their catalytic significance becomes extremely doubtful.

Notwithstanding these serious questions on any primitive status of the Sox mechanism, this debate has acquired an all-new dimension with the identification of complete sets of sox structural genes in extremophilic epsilonproteobacterial genera whose members inhabit deep-sea geothermal vents and subsea floors and oxidize sulfur using a wide range of electron acceptors other than oxygen and nitrate (Campbell et al., 2006; Scott et al., 2006; Nakagawa et al., 2007; Sievert et al., 2008).

Origin and evolution of sulfur lithotrophy

Corroborating the hypothesis that the extant mechanisms of sulfur chemolithotrophy are evolutionarily recent, phylogenies of the dissimilatory sulfite reductase and APS reductase genes have illustrated close relationships among bacterial and archaeal homologues (Hipp et al., 1997; Wagner et al., 1998). This implied that these genes originated in the last common ancestor of the three domains of life, from which the first bifurcation in the universal phylogenetic tree is said to have given rise to the bacterial descent alongside the last common ancestor of Archaea and Eukarya (Gribaldo & Cammarano, 1998). More recent phylogenetic analyses based on dsr and APS reductase gene sequences have also underscored the ancient nature of coupled sulfur and sulfate reduction/sulfide oxidation pathways (Boucher et al., 2003; Meyer & Kuever, 2007a, b), and the origin of these processes has been predicted by some workers to date back to the early Proterozoic era (3.47 Gyr ago) when the environment of the Earth was highly reducing in nature (Shen & Buick, 2004; Meyer et al., 2007).

Integrating the aforesaid phylogenetic information with molecular fossil (unique hydrocarbon biomarkers) records, which confirm the presence of GSB and PSB in rocks dated 1.64 Gyr ago (Brocks et al., 2005), Meyer et al. (2007) have outlined a putative course of evolution of the different sulfur-lithotrophic processes. Their hypothesis states that amidst the generally reducing environment of the early Proterozoic era, anoxic sulfidic oceans (Canfield, 2005) plausibly supported the growth of planktonic ancestors of the modern GSB and PSB. Because thiosulfate, oxygen or nitrate, the prerequisites of extant chemolithotrophic pathways, could have appeared in the biosphere only after it became relatively less reducing (Broda, 1977; Kelly & Wood, 2000a), present-day chemolithotrophy must have originated from inorganic oxidative metabolic processes that had already developed in anaerobic geological times. This implicates anaerobic, anoxygenic photolithotrophy as not only the predecessor of aerobic sulfur chemolithotrophy but also as the most ancient of all the sulfur oxidation functions (Kelly & Wood, 2000a). Under the environmental conditions prevailing in the early Proterozoic era the anaerobic anoxygenic phototrophs plausibly converted the abundant sulfide and sulfur by the reverse-operating Dsr system. Subsequently, atmospheric oxygen increased during the end of the Proterozoic era, and kept doing so until 1.05 Gyr ago, resulting in decreased levels of sulfides and increased proportions of less-reduced inorganic sulfur species such as thiosulfate (Canfield & Teske, 1996; Canfield, 2005), which plausibly ushered the evolution and diversification of nonphotosynthetic, facultatively aerobic or even strictly aerobic, sulfur-oxidizing bacteria. Commencement and flourishing of the oxidative half of the sulfur cycle on a global basis perhaps did not start until the concluding periods of Proterozoic era, i.e. 0.75–0.62 Gyr ago (Canfield & Teske, 1996). It was only during this geological period that the more sophisticated pathways (akin to the Sox system) for using less-reduced inorganic sulfur species as electron donors could have originated and evolved in aerobic ancestors of nonphotosynthetic sulfur-oxidizing bacteria (plausibly the Proteobacteria) that lacked the reverse Dsr pathway. The reverse Dsr pathway, in its turn, remained conserved in the ancestors of the few extant facultatively anaerobic sulfur-chemolithoautotrophic bacteria. These latter types with relict dsr loci plausibly selected the branched thiosulfate oxidation pathway. Concurrently, to add to their bioenergetic fitness and broaden their substrate-utilization spectrum, primordial anaerobic sulfur-oxidizing photolithotrophs could have acquired components of the more recent Sox apparatus by HGT from the sulfur chemolithotrophs.

As opposed to the notion that putative planktonic ancestors of modern phototrophic GSB and PSB were the most primitive sulfur oxidizers, the possibility of chemolithotrophy, originating in the deep-sea environment, as the ancestral form of all sulfur oxidation processes also cannot be ruled out. Deep-oceanic vents are considered some of the most ancient colonized habitats on Earth (Reysenbach et al., 2000a), while metabolic strategies of hyperextremophilic Bacteria and Archaea (which include chemolithoautotrophy, mixotrophy or even chemoorganoheterotrophy, but not photolithotrophy) represent the most ancient phenotypes of living organisms (Schonheit & Schaeferr, 1995; Gevertz et al., 2000; Campbell et al., 2006).

The perception of chemolithoautotrophy as the earliest self-sustaining metabolism (Wächtershäuser, 1990a; Stetter, 1992; Russell & Hall, 1997) has gained currency ever since the discovery that the most deeply rooted phylogenetic branches of Bacteria and Archaea encompass chemolithoautophic extremophiles (Woese, 1987; Burggraf et al., 1992; Stetter, 1992; Fuchs et al., 1996). It is now known that sulfur chemolithotrophy in many hyperthermophiles is potentially anaerobic or microaerophilic, and based on the reduction of a wide range of electron acceptors (e.g. Fe3+, CO2, CO, NO2, NO, N2O, SO42−, SO32−, S2O32−, S0, etc.) other than O2 and NO3, which emanate from volcanic or geothermal activities in their habitats (Schonheit & Schaeferr, 1995; Amend & Shock, 2001; Campbell et al., 2006). Chemolithoautotrophy in these organisms is said to have evolved alongside CO2-fixing mechanisms similar to either of the two extant autotrophic pathways, viz. the acetyl-coenzyme A pathway (or the Wood–Ljungdahl pathway, found in bacteria and archaea, that converts CO2 to acetate by the action of the key enzyme acetyl-coenzyme A synthase/CO dehydrogenase) or the reductive or reverse TCA cycle (or the Arnon cycle, found only in anaerobic or microaerophilic microorganisms such as GSB, some members of Deltaproteobacteria, Aquificales and archaeal Thermoproteaceae groups, plus the epsilonproteobacterial sulfur chemolithoautotrophs of deep-sea vents) (Evans et al., 1966; Fuchs et al., 1980; Wächtershäuser, 1990b; Pereto et al., 1999; Lindahl & Chang, 2001; Wirsen et al., 2002; Campbell et al., 2003, 2006; Hugler et al., 2003; Russell & Martin, 2004; Smith & Morowitz, 2004). Either way it is reasonable to conclude that the first autotrophic processes utilized cyclic processes involving simple organic acids and were akin to those seen in extant sulfur-, and sulfate-reducing bacteria and archaea (Kelly & Wood, 2000a). Parallel to these notions, it has been postulated that the first chemolithotrophs could have been chemolithotrophic heterotrophs (and not chemolithotrophic autotrophs) that used inorganic energy sources in tandem with prebiotic organic molecules (Kelly & Wood, 2000a) and chemolithotrophy paved the way for autotrophy to evolve from pre-existing sugar-metabolizing pathways. If this is correct, the extant obligate chemolithoautotrophs can be considered to have evolved via secondary loss of heterotrophy coinciding with the evolution of autotrophy. Concurrently, abundant organic molecules excreted by extremophilic sulfur chemolithoheterotrophs might have led to the evolution of chemo-organoheterotrophic lineages through loss of chemolithotrophic potentials. The fact that the Calvin cycle is not used for inorganic carbon assimilation by hydrogen- or sulfur-oxidizing hyperthermophiles (Maden, 1995; Schonheit & Schaeferr, 1995) negates the possibility that it is a primordial pathway (Evans et al., 1966; Maden, 1995). However, it is noteworthy that a novel (or even primordial) ribulose bisphosphate carboxylase has been reported from Pyrococcus and other archaea, and if they are indeed relicts and not products of HGT, then the Calvin cycle could be more ancient than currently supposed (Kelly & Wood, 2000a).

Although the availability of free oxygen would have been crucial for evolution of present-day sulfur-oxidizing chemotrophs that require aerobic or even microaerophilic conditions, oxygen is not essential for many of the isolated sulfur-reducing/sulfur-oxidizing Epsilonproteobacteria, especially the deeply branching Nautiliales, which are obligate anaerobes, or others that use various alternative electron acceptors (Campbell et al., 2006). Because the ecological niches and metabolic characteristics (e.g. thermophilic growth, anaerobic metabolism and autotrophy via the rTCA cycle) of the members of Epsilonproteobacteria, Deltaproteobacteria and Aquificales are equivalent to those of the anaerobic and photolithotrophic GSB (Campbell et al., 2006), the mode of lithotrophic sulfur oxidation encountered in the former groups is also a strong candidate for the most ancient status. It is interesting to note that the Epsilonproteobacteria and Deltaproteobacteria are not only phylogenetically closest to the Aquificae and Chlorobi, the 16S rRNA gene-based divergence (i.e. origin) of Epsilonproteobacteria also dates back to 1.37 billion years ago (Gupta, 2000; Sheridan et al., 2003), which is contemporary with the geological age when molecular fossils of anoxygenic sulfur phototrophs were detected (Brocks et al., 2005).

Whatever may be the bottomline of this debate, it is indisputable that the delineation of the origin and evolution of sulfur lithotrophy holds the key to our understanding of the earliest metabolic strategies of life on Earth. As such, these issues should be given foremost priority among all the avenues of future research on prokaryotic sulfur oxidation.


This study is a tribute to our teacher the late Dr Pradosh Roy who taught us everything. We owe our present endeavors to our mentor Dr Sujoy Kumar DasGupta. The invaluable perspectives offered by Dr Christiane Dahl of the Institut für Mikrobiologie und Biotechnologie, Rheinische Friedrich-Wilhelms-Universität Bonn, Germany, on the topic of phototrophic sulfur oxidation are profoundly acknowledged. We thank Dr Arnulf Kletzin, Institute of Microbiology and Genetics, Darmstadt University of Technology, Germany, along with the Springer Science and Business Media for allowing the reproduction of the molecular model for sulfur oxidation in A. ambivalens. We also thank Dr Cornelius G. Friedrich, Lehrstuhl für Technische Mikrobiologie, Universität Dortmund, Germany. Extensive academic assistance provided by Srimati Baishali Ghosh is sincerely acknowledged. W.G. was fiscally assisted by the Burdwan University in the first half of this study and by the Bose Institute and Council for Scientific and Industrial Research (CSIR), Government of India in the concluding periods. We thank Sri Satyabrata Pal for diagrammatic artwork.