|Home | About | Journals | Submit | Contact Us | Français|
Despite their taxonomic description, not all members of the order Sulfolobales are capable of oxidizing reduced sulfur species, which, in addition to iron oxidation, is a desirable trait of biomining microorganisms. However, the complete genome sequence of the extremely thermoacidophilic archaeon Metallosphaera sedula DSM 5348 (2.2 Mb, ~2,300 open reading frames [ORFs]) provides insights into biologically catalyzed metal sulfide oxidation. Comparative genomics was used to identify pathways and proteins involved (directly or indirectly) with bioleaching. As expected, the M. sedula genome contains genes related to autotrophic carbon fixation, metal tolerance, and adhesion. Also, terminal oxidase cluster organization indicates the presence of hybrid quinol-cytochrome oxidase complexes. Comparisons with the mesophilic biomining bacterium Acidithiobacillus ferrooxidans ATCC 23270 indicate that the M. sedula genome encodes at least one putative rusticyanin, involved in iron oxidation, and a putative tetrathionate hydrolase, implicated in sulfur oxidation. The fox gene cluster, involved in iron oxidation in the thermoacidophilic archaeon Sulfolobus metallicus, was also identified. These iron- and sulfur-oxidizing components are missing from genomes of nonleaching members of the Sulfolobales, such as Sulfolobus solfataricus P2 and Sulfolobus acidocaldarius DSM 639. Whole-genome transcriptional response analysis showed that 88 ORFs were up-regulated twofold or more in M. sedula upon addition of ferrous sulfate to yeast extract-based medium; these included genes for components of terminal oxidase clusters predicted to be involved with iron oxidation, as well as genes predicted to be involved with sulfur metabolism. Many hypothetical proteins were also differentially transcribed, indicating that aspects of the iron and sulfur metabolism of M. sedula remain to be identified and characterized.
Biomining exploits acidophilic microorganisms to recover valuable metals (i.e., Cu and Au) from ores ( 52, 56, 57, 59) in biohydrometallurgical processes conducted at temperatures ranging from ambient ( 44, 71 ) to 80°C (17). Higher-temperature operations involve consortia of extremely thermoacidophilic archaea from the genera Sulfolobus, Acidianus, and Metallosphaera (49, 50). For example, Sulfolobus metallicus, certain Acidianus species ( 12, 47 ), and Metallosphaera sedula (29, 33) all can mobilize metals from metal sulfides. However, not all members of these genera are metal bioleachers. In fact, the three Sulfolobus species with completed genome sequences (S. solfataricus, S. acidocaldarius, and S. tokodaii) are apparently unable to effect metal sulfide oxidation. Since genome sequences are not yet available for metal- bioleaching extreme thermoacidophiles, genetic features characteristic of this physiological capability remain to be seen.
Biomining is based upon biological oxidation of iron (Fe) and reduced inorganic sulfur compounds (RISCs). Two Fe(III)-driven chemical leaching mechanisms are catalyzed by microorganisms regenerating Fe(III) from Fe(II), facilitating release of valuable metals bound in metal sulfides (52, 56, 57, 59). As accumulation of precipitating RISC compounds can slow leaching rates (52), microbial RISC-oxidizing capabilities are beneficial. Extreme thermophiles are particularly desirable, as problematic passivation on ore surfaces decreases at higher temperatures. Since chemolithotrophs use Fe(II) and/or RISCs as energy sources, nutritional supplement costs are relatively low, thus minimizing process economic impact.
Components of electron transport chains involved in Fe and RISC oxidation are differentially expressed as a function of substrate in both meso- and thermoacidophilic bioleachers (5, 36, 55). Bacterial Fe oxidation studies focused on mesophilic A. ferrooxidans have shown that electrons from Fe(II) can be transported along either a “downhill” or an “uphill” electron pathway, towards a lower (O2/H2O)- or higher [NAD(P)+/NAD(P)H]-potential redox couple, respectively (28). Both pathways involve electron transfer via cytochromes c and rusticyanin (Rus), a periplasmic blue copper protein (Bcp). The downhill and uphill pathways, encoded by the rus and petI operons, conclude with reduction of the final target by a cytochrome c oxidase (CoxABCD) and bc1 complex (PetI), respectively, and are highly transcribed on Fe(II) relative to elemental sulfur (S0) (7, 18, 72). In a proposed archaeal Fe oxidation model, based on Ferroplasma strains, the terminal oxidase combines select bc1 complex-like and cytochrome c oxidase-like components and involves association of a Bcp, sulfocyanin (15). Aspects of this model were noted previously in the nonleacher S. acidocaldarius (39, 41), while a correlation between cytochromes b, b558/562 (CbsAB), and Fe-S cluster proteins (possible quinol oxidase components) and the bc1 complex has been proposed (27, 39, 65, 75). Finally, a gene cluster containing terminal oxidase and electron transport chain components, highly transcribed on Fe(II) versus S0, was identified in S. metallicus (5).
Extremely thermoacidophilic archaeal RISC oxidation mechanisms have not been established. RISC electrons may enter the electron transport chain via a membrane-bound thiosulfate (S2O32−):quinone oxidoreductase (TQO), a sulfite:acceptor oxidoreductase, or possibly a sulfide:quinone oxidoreductase (38). Despite the annotation, tetrathionate hydrolases (TetH) expressed on S0, FeS2, and S2O32− substrates have been noted in A. ferrooxidans and Acidithiobacillus caldus (8, 35, 61). Certain terminal oxidase components have been expressed in extremely thermoacidophilic archaea grown on specific RISCs (36, 54). Although not directly linked to electron transport, sulfur oxygenase reductases (Sor) are believed to be important for breakdown of intracellular RISCs, particularly S0 or polysulfides (11, 37, 69). M. sedula growth on S0 and metal sulfides (i.e., FeS2 and chalcopyrite [CuFeS2]), has been documented (29). With S2O32− as a primary intermediate of acid-insoluble metal sulfide leaching, this may be a RISC substrate. However, there are no reports of M. sedula growth on S2O32− or other RISCs, such as sulfite (SO32−) or polythionates.
Although M. sedula presumably grows mixotrophically, autotrophy is critical in biomining environments, where relatively small amounts of organic carbon are available. Several members of the Sulfolobales, including M. sedula, purportedly fix carbon (CO2) via a modified 3-hydroxypropionate cycle, identified in Chloroflexus aurantiacus (31-34, 43, 67). In M. sedula, two enzymes of this cycle have been characterized biochemically (1, 33), while two additional enzymes' activities were detected in autotrophically grown cell extracts, suggesting their involvement in CO2 fixation (32).
Heavy metals, typically toxic at concentrations higher than trace amounts, are routinely encountered by biomining acidophiles. In contrast to the substantial amount of information about metal toxicity in neutrophiles (48), relatively little is known about such mechanisms in acidophiles, despite their higher resistance levels (16). Toxicity and mechanisms of resistance to mercury (Hg) have been investigated in S. solfataricus (14, 63, 64), while possible copper (Cu) resistance mechanisms have been studied in S. solfataricus and S. metallicus (19, 58). Continued study of heavy metal toxicity and resistance mechanisms in bioleachers is important, especially for cases like M. sedula, where tolerance to some metals, particularly Cu, may require improvement to make it competitive with other microbial biominers, such as S. metallicus and A. ferrooxidans (16, 30, 58).
Adhesion mechanisms of biomining microorganisms to acid-insoluble metal sulfides or S0 may be important to oxidation rates. Study of bioleachers' extracellular polymeric substances, often exopolysaccharides (EPS), has centered on A. ferrooxidans (6, 13, 23, 24, 51, 62). EPS production, composition, and properties were all observed to be a function of growth substrate in A. ferrooxidans (23, 24). In addition, significant homology between the components of the bacterial type II secretion, type IV pilus, and archaeal flagellar (Fla) systems has been suggested (53). Bacterial type IV pili and Fla systems have been implicated in adhesion (46, 66), and pilus-like structures, along with “wiggling of the cells at the ore,” were observed during initial characterization of M. sedula (29).
To better understand the defining features of extreme thermoacidophiles capable of metal sulfide oxidation, the genome sequence of M. sedula was completed and probed for clues with respect to five physiological areas (iron oxidation, sulfur oxidation, carbon fixation, metal toxicity, and adhesion) that are important to its current and future role in biomining operations. In addition, whole-genome transcriptional response analysis was used to identify specific open reading frames (ORFs) related to Fe oxidation.
M. sedula (DSM 5348T) was purified to clonality by aerobic cultivation at 70°C on a solid medium, prepared as described previously (64), and adjusted to a pH of 2.5 with sulfuric acid. High-molecular-weight genomic DNA was isolated, as described previously (26), and provided to the U.S. Department of Energy Joint Genome Institute (http://www.jgi.doe.gov/) for cloning and shotgun sequencing. A combination of small (average insert sizes, 3 and 8 kb) and large (40 kb, fosmid) insert libraries were prepared and used for analysis as indicated at http://www.jgi.doe.gov/.
Critica, Generation, and Glimmer software programs were used for coding region detection and gene identification. TMHMM 2.0 (http://www.cbs.dtu.dk/services/TMHMM/) was used to predict transmembrane helices in translated sequences. SignalP v2.0b2 (http://www.cbs.dtu.dk/services/SignalP-2.0/) was used to predict the presence and location of N-terminal signal peptides.
For comparative genomics, M. sedula gene sequences and annotations were downloaded from the JGI microbial genome website (http://genome.ornl.gov/microbial/msed/28feb07/). Other sequences referenced were obtained from GenBank, with the exception of A. ferrooxidans ATCC 23270 sequences, which were obtained from The Institute for Genome Research's Comprehensive Microbial Resource. Basic Local Alignment Search Tool (BLAST) searches were conducted using a BLOSUM62 matrix with either the NCBI protein-protein BLAST program (BLASTP) against the Swissprot or nr database or the ORNL Microbial BLAST server's BLASTP program against the Msed database. “Significant similarity” or “significant homology” was defined as ≥20% identity and an Eval value of ≤10−5, unless otherwise noted.
Samples used for scanning electron microscopy were grown aerobically at 70°C in a minimal salts medium (60) containing 0.1% FeS2 as the sole energy source; prior to use, the FeS2 was sterilized by baking for 3 days at 140°C. Glutaraldehyde-fixed cells treated with osmium tetroxide were applied to poly-l-lysine-coated, gold-palladium-treated coverslips, followed by ethanol dehydration. Following critical-point drying, gold-palladium-coated cells were visualized with a Hitachi S-3000N variable pressure scanning electron microscope (http://bsweb.unl.edu/labs/blum/index.html).
An oligonucleotide microarray was developed using a minimum of one 60-mer probe (designed using OligoArray 2.1 software, synthesized by Integrated DNA Technologies) for each identifiable protein-encoding ORF in the M. sedula genome. Cells (DSMZ 5348) were grown aerobically at 70°C in an orbital shaking oil bath at 70 rpm on DSMZ 88 medium (pH 2), supplemented with 0.1% yeast extract (YE). An initial density of mid-106 cells/ml was used in four 1-liter bottles containing 300 ml of the same medium. Approximately 7.5 g/liter of FeSO4·7H2O was added to two of these cultures prior to inoculation. At harvest, cultures in mid-exponential phase (~24 h of growth, mid-107 cells/ml) were quickly chilled and then centrifuged at 9,510 × g for 15 min at 4°C. RNA was extracted and purified (RNAqueous; Ambion), reverse transcribed (Superscript III, Invitrogen), repurified, labeled with either Cy3 or Cy5 dye (GE Healthcare), and hybridized to one of two microarray slides (Corning). Slides were scanned on a Perkin-Elmer scanner, and raw intensities were quantitated with ScanArray Express 2.1 software. Normalization of data and statistical analysis were performed using JMP Genomics 6.0.2 software (SAS Institute, Cary, NC).
The 18 contigs of the draft genome sequence were made available publicly on 27 November 2006, with the final assembly released on 25 April 2007 and listed under GenBank accession no. CP000682 (http://genome.jgi-psf.org/finished_microbes/metse/metse.info.html).
The M. sedula genome (46% G+C) encodes 2,258 proteins, 35% of which are annotated as either “hypothetical protein” or “protein of unknown function.” However, almost 90% of these have their strongest similarity (“top hit”) to Sulfolobus species (786 to S. solfataricus, 829 to S. tokodaii, 327 to S. acidocaldarius, 14 to “S. islandicus,” and 3 to S. shibatae), with 35 ORFs being most similar to genes in Acidianus species. This raises some interesting questions. Will the genes responsible for the biomining capabilities of M. sedula be found within the 10% of gene sequences with strongest similarities to non-Sulfolobales or nonarchaeal microorganisms? Or, conversely, will subtle differences in the Sulfolobales' gene sequences result in the distinction between thermoacidophilic leachers and nonleachers?
The Fe oxidation mechanism in M. sedula is, as yet, unknown. The organization and composition of at least five respiratory clusters found in the genome sequence, two of which are predicted to be highly transcribed during growth on Fe(II) or FeS2, differ from those of the bacterial rus and pet operons. No PetC-like ORFs (cytochrome c subunit of the bc1 complex) are apparent. This is consistent with reports that archaeal genomes of Ferroplasma and Sulfolobus species do not contain cytochromes c (68) and suggests that acidophilic archaeal and bacterial Fe oxidation do not occur via the same mechanism. The gene neighborhood organization of respiratory clusters in M. sedula provides support for the Ferroplasma hybrid terminal oxidase model.
Genes for four Bcp-like proteins are found in the genome; two are annotated as hypothetical sulfocyanins (SoxE). Figure Figure11 displays a multiple-sequence alignment of these Bcps, along with similar Bcp-like sequences from other microorganisms. Almost all the Bcp sequences appear to contain four ligands important for Cu binding and the characteristically high positive redox potential of Rus in A. ferrooxidans (4, 76). Both Msed0323 and Msed0826 contain two recognized protein signatures for sulfocyanins, while the two remaining Bcps from M. sedula (Msed0966 and Msed1206) contain rusticyanin protein signatures (25). The SoxE-like sequences appear to align fairly well with each other, but the same cannot be said for the Rus-like sequences. The relevance of poor alignment between Rus-like sequences to protein function is not clear. The archaeal Rus-like proteins cannot have the same cellular location as their bacterial periplasmic counterpart. The Msed0966 product is more similar to the Rus found in A. ferrooxidans (Afe3186) than that of Msed1206. Both have predicted N-terminal signal peptides, but the Msed1206 protein also contains a single, predicted C-terminal hydrophobic transmembrane region. Although the other sequenced Sulfolobus species contain multiple SoxE homologs, only S. solfataricus (Sso1870) appears to contain a Rus-like (Msed0966-like) sequence. However, Sso1870 does not appear to contain a recognized Rus protein signature or the four ligands noted as important to Cu binding and redox potential.
Three sets of ORFs with low similarity to the PetAB subunits of the bc1 complex are found in the M. sedula genome (Msed0288 and Msed0291, Msed0322 and Msed0321, and Msed0501 and Msed0500) (Fig. (Fig.2).2). In each case, a Rieske protein (SoxL/SoxF) pairs with a cytochrome b (SoxC/SoxG/SoxN). In the first two cases, the Rieske protein and cytochrome b are clustered with cytochrome oxidase (Cox) subunits I and II (SoxAB or SoxHM) at the same loci, suggesting close association between quinol- and Cox-like subunits. In the case of locus 3, there is another Cox I-Cox II pair located further downstream (locus 4) which lacks a corresponding Rieske-cytochrome b (PetAB-like) pair. However, it is not clear if loci 3 and 4 should be grouped into a single, more diffuse, locus. Both loci contain a cytochrome b558/562 (CbsAB; Msed0503 and -0504 and Msed0477 and -0478). Limited information available makes it difficult to determine whether CbsAB could substitute for quinol oxidase or Cox components of a terminal oxidase; however, previous studies have suggested that CbsAB may play a role in extracellular electron transport similar to that of the bc1 complex (27, 65, 75). Locus 5 encodes DoxBCE (Msed2030 to Msed2032), the SoxAB-like terminal oxidase components from Acidianus ambivalens, and appears to be missing quinol oxidase-like subunits. Interestingly, Msed1191 appears to encode a lone quinol oxidase fusion protein (cytochrome b N-terminal region and Rieske protein C-terminal region).
Because locus 1 soxB (Msed0290) was reported to be up-regulated on S0, compared to FeS2 or YE (36), and DoxBCE expression and copurification with TQO subunits (DoxDA) were noted under S0-oxidizing conditions in A. ambivalens (45, 54), it seems likely that the locus 1 and 5 respiratory clusters in M. sedula would also be stimulated during growth on RISCs. Locus 2 soxM (Msed0324) was reported as being up-regulated on YE, compared to FeS2 or S0 (36). Locus 3 cbsA (Msed0504) was found to have higher rates of transcription on FeS2 than YE or S0 (36); therefore, the entire locus is predicted to be up-regulated on Fe(II). Interestingly, locus 3 follows a proton efflux ATPase (Msed0505). Previous A. ferrooxidans studies suggested that this type of ATPase could function in either direction to regulate a cytochrome oxidase on the basis of available ATP, restoring proton motive force or supporting NADP(H) reduction (18). Locus 4 contains a gene cluster with homology to fox genes in S. tokodaii and S. metallicus, which were demonstrated to be up-regulated on Fe(II) versus S0 (5). While locus 3 components appear to be present in both S. acidocaldarius and S. solfataricus genomes (27, 65), neither genome contains fox homologs (excepting FoxHJ in S. solfataricus and FoxH in S. acidocaldarius). This is not entirely surprising, as neither organism appears to be capable of growth on Fe(II). In light of this observation, the fox gene cluster is likely essential for metal mobilization. Because S. tokodaii was recently reported to grow on Fe(II) (5) and its genome does not seem to encode Rus, it is not clear whether rusticyanin is essential for archaeal Fe oxidation or whether an alternative redox protein, such as sulfocyanin (SoxE), may perform an equivalent function. It is interesting that locus 4 includes a gene for a Cox II homolog (soxH; Msed0480) which contains conserved Cu-binding ligands similar to locus 2 SoxH (Msed0320) of the SoxM supercomplex, which is thought to interact with sulfocyanin (Msed0323), also of the SoxM supercomplex (39). Taking these data together, the predicted Fe oxidation capacity encoded in locus 4 could involve sulfocyanin, rusticyanin, or both. Because Msed1191 is unique to the BLAST database (as of September 2007), no similar sequence could be identified to predict what type of substrate could induce transcription of this ORF.
M. sedula encodes a pair of proteins (Msed0363 and Msed0364) with significant similarity to DoxDA subunits of the TQO in A. ambivalens (45). At least two sulfite:acceptor oxidoreductase-like proteins exist in M. sedula: a putative sulfite oxidase (Msed0362) on the strand opposite doxDA and an A. ferrooxidans CysI-like (Afe3210) putative sulfite reductase (Msed0961) near a putative rusticyanin; however, neither appears to be membrane associated. Of the five ORFs in M. sedula (Msed0353, -0558, -1039, -1323, and -2059) with significant similarity to putative sulfide:quinone oxidoreductase proteins in A. ferrooxidans (Afe1293 and Afe2761), only Msed1039 is predicted to contain a signal peptide. The M. sedula genome encodes a membrane-associated protein (Msed0804) with significant similarity to TetH found in both A. ferrooxidans (Afe2996) and Acidithiobacillus caldus (ABP38225). No counterparts can be identified in either S. solfataricus or S. acidocaldarius, although Sto1164 appears to be similar to the TetH of A. ferrooxidans. Neither an archaeal nor a bacterial Sor sequence is evident in the genome of M. sedula. While Sor is also apparently absent from the genomes of S. solfataricus and S. acidocaldarius, it is present in S. tokodaii and “F. acidarmanus.”
Putative versions of genes for adenosine phosphosulfate (APS) reductase and sulfate adenylyl transferase (SAT) are colocated in M. sedula (Msed0962 and Msed0963), on the strand opposite a putative sulfite reductase gene (Msed0961). Depending on operational direction, these proteins could link RISCs to substrate-level phosphorylation. Operating oxidatively, as seen in chemolithotrophs, APS “reductase” converts SO32− to APS, followed by SAT oxidation of APS to sulfate and ATP (74). Operating reductively, as seen in heterotrophs and autotrophs, requires function of SAT and then APS reductase, converting sulfate to SO32−. A sulfite reductase further reduces SO32− to sulfide, and then cysteine is formed via an o-acetyl-l-serine(thiol) lyase (70) or a cysteine synthase. Similar to M. sedula, the A. ferrooxidans 23270 genome also encodes a collocated APS-SAT pair, although the SAT sequence does not appear to be similar to that of M. sedula. SAT is usually encoded by two genes, one for an ATP sulfurylase and one for an APS kinase, although there are instances of their fusion (74). However, in M. sedula, only the ATP sulfurylase component is found, suggesting that APS, and not phosphoadenosine phosphosulfate, is formed. This is supported by the APS reductase sequence, which contains four cysteines (for 4Fe-4S center coordination) characteristic of APS reductase but missing in phosphoadenosine phosphosulfate reductase (70). Similarly, the CysI-like sulfite oxidoreductase subunit (Msed0961) is found in A. ferrooxidans, along with a flavoprotein subunit (CysJ) (70); however, in M. sedula, the CysJ-like component appears to be absent. Because the APS-SAT pair appears in nonleaching Sulfolobus species, the importance of these proteins in a bioleaching microorganism is unclear.
Finally, one putative rusticyanin mentioned previously (Msed0966) is located on the strand opposite of the APS-SAT pair in M. sedula. The clustering of this gene relative to those involved with sulfur metabolism may be coincidental; however, it may also suggest a physiological link between Fe and RISC oxidation important for bioleaching. Preliminary support for this hypothesis is based on an A. ferrooxidans study, which showed rusticyanin to be up-regulated during early exponential growth phase on a sulfur substrate (55).
Sequences of genes for the two previously characterized enzymes of the 3-hydroxypropionate cycle (catalyzing steps 1, 2, and 4) can be located in the genome (Fig. (Fig.3).3). Potential sequences of genes for previously detected enzymatic activities of propionyl coenzyme A (propionyl-CoA) synthase (step 3) and fumarate hydratase (step 9) activities were also found. Although no propionyl-CoA synthase gene is annotated in the M. sedula genome, there are multiple AMP-dependent synthetase/ligase candidates, with Msed1353 having the most sequence similarity to the N-terminal propionyl-CoA synthases from both C. aurantiacus (AAL47820) and Roseiflexus RS-1 (YP_001277513) (20). It is interesting that both bacterial enzymes are two to three times the size of the Msed1353 protein, which raises the prospect that a multisubunit propionyl-CoA synthase complex exists in M. sedula. Querying amino acids ~900 to 1100 and ~1300 to 1500 of both bacterial enzymes against the M. sedula genome results in hits to ORFs automatically annotated as enoyl-CoA hydratases (or 3-hydroxybutaryl-CoA dehydratases) and alcohol dehydrogenases, respectively. Two classes of fumarate dehydratase were also identified (Msed1115 and Msed1116, class I; Msed1462, class II). Candidates for the six additional enzymatic steps (steps 5, 6, 8, 10, 11, and 14) of the cycle can be identified in the genome sequence, with the predicted assignment of Msed0381 to steps 10, 11, and 14 based on its sequence similarity to the gene for the single enzyme in C. aurantiacus (mcl) responsible for malyl-, β-methylmalyl-, and S-citramalyl-CoA lyase activity (20, 42). Recent information suggests that at least part of the pathways in C. aurantiacus and M. sedula evolved independently (1). Hence, it is not surprising to find that the genome is devoid of sequences with similarity to encoded enzymes responsible for transferase activity in C. aurantiacus (21, 22) (steps 7 and 10). Because components of this pathway identified in M. sedula are not clustered, genome location is of limited use in determining the remaining enzymes participating in the pathway.
Microbial P-type ATPases, specifically CPX (Cys-Pro-X)-ATPase, have been proposed to mediate efflux of heavy metal cations, such as Cu, Zn, and Cd (16, 48, 58). A CPX-ATPase has been identified in M. sedula (Msed0490), with a C-P-C (Cys-Pro-Cys) pattern located in the fifth of seven predicted transmembrane regions. This gene has significant homology to a P-type ATPase in S. solfataricus (CopA) implicated in Cu(II), and possibly also cadmium, cation efflux (19). Ettema et al. (19) demonstrated that the metal efflux process involved a metallochaperone (CopM) having a 32-bp overlap with CopA and a new type of archaeal transcriptional regulator (CopT). M. sedula contains ORFs with significant similarity to both CopM and CopT (Msed0491 and Msed0492). However, the copMA overlap is limited to 10 bp, and the location of copT (on the opposite strand) makes the gene organization reminiscent of versions in the S. tokodaii and S. acidocaldarius genomes.
With respect to the polyphosphate mechanism proposed for Cu(II) detoxification in S. metallicus (58), Msed0740 is annotated as a probable polyphosphate/NAD kinase (ppnk), and Msed0981 has 37% similarity to the Ppx previously characterized in S. solfataricus (9). Searching for possible metal-phosphate complex exporters, no significant hits to E. coli phosphate transport systems (PstSCAB or Pit) were identified in M. sedula. However, as found with other genome-sequenced acidophiles (2, 58), use of an S. cerevisiae Pho84 sequence query produced results: three hits with similarities of 25 to 30% (Msed1512, Msed1094, and Msed0866). Using the top Pho84-like sequence in A. ferrooxidans (Afe0861) as a template, four hits with similarities of 30 to 32% resulted (the previous three plus Msed0846). All four hits belong to the major facilitator superfamily of substrate transporters. Msed1521 encodes a hypothetical protein downstream of Msed1512 with weak similarity to a phosphate transport regulation protein (COG1392, PFAM01865). Msed1968 (COG0704) encodes a putative phosphate uptake regulator.
A gene corresponding to an arsenite transporter (ArsB) was also identified (Msed2086; COG1055). The As(III) membrane potential-driven pump is typically found along with a repressor protein (ArsR) and sometimes with an As(V) reductase (ArsC), an As resistance-boosting APTase (ArsA), and/or a secondary regulation protein (ArsD) (16). In M. sedula, however, ArsB appears to stand alone. Msed1005 potentially encodes an ArsR-like regulatory protein with a TRASH domain, but it is unclear whether this protein might be involved in the regulation of ArsB, MerAH (Msed1241 and Msed1242), or some other metal resistance mechanism.
Figure Figure44 depicts M. sedula attached to FeS2 and apparent EPS production. In support of this observation, the Msed1805-1857 locus encodes proteins potentially involved in EPS production. Queries from A. ferrooxidans (Afe1738) and Vibrio cholerae (Vps21) produced hits with significant similarity to Msed1810. All three sequences are annotated as a UDP-glucose (or mannose) 6-dehydrogenase (EC 1.1.22). Msed1809, which shares a 3-bp overlap with Msed1810, has lower, but significant, similarity to Afe1739, although both sequences are annotated as encoding UDP glucose 4-epimerase. The Msed1808 protein, annotated as a hypothetical protein, shares significant similarity with glycosyltransferases of Staphylococcus aureus (CapM) and V. cholerae (Vps32) implicated in capsular polysaccharide biosynthesis (40, 73). Msed1811, annotated as encoding a UDP-glucose pyrophosphorylase (EC 18.104.22.168), is not related to the EPS query sequences used, but, like Msed1808-10, it may encode enzymes involved in the synthesis of biofilm components, specifically glucuronic acid via the Leloir pathway (3). EPS of FeSO4- and FeS2-grown cells of A. ferrooxidans contain Fe(III) complexed with glucuronic acid, which is believed to impart an overall positive charge to the EPS, promoting attachment to FeS2, which tends to maintain an overall negative charge at low pH (62). Additional ORFs located further downstream in this 1800s region have annotations suggestive of a role in formation of other EPS components (i.e., UDP-galactose, dTDP-rhamnose), but specific biochemical functions and cellular roles of the encoded proteins remain to be seen.
Genome analysis reveals at least four gene clusters with similarities to the type II/IV/Fla systems. Msed1324 to Msed1330 appear to encode seven flagellar proteins. The Msed1324 protein, annotated as FlaJ, contains eight predicted transmembrane regions and shares a 14-bp overlap with Msed1325. Msed1325 and Msed1326 (FlaIH) encode two cytoplasmic subunits related to ATPases (COG0630 and COG2874), while Msed1327 to Msed1329 overlap each other by 1 bp (Msed1327 and Msed1328) or 23 bp (Msed1328 and Msed1329) and are all predicted to have one N-terminal transmembrane helix. Msed1327 and Msed1328 correspond to FlaFG, while the Msed1329 protein shares no identity with characterized proteins. Msed1330 is annotated as encoding a flagellin protein and has similarity to the major preflagellin/prepilin FlaB genes (COG1681). Outside of this cluster, there are at least three other pairs of FlaIJ-like protein genes (Msed1197 and -1198, Msed2104 and -2105, and Msed0650 and -0651). In the first two cases, flaI-like genes are annotated as encoding type II secretion system proteins, while their overlapping flaJ-like genes are annotated as encoding hypothetical proteins. In the last case, however, both genes are annotated as hypotheticals and share no overlap. Although all three FlaI-like proteins are closely related to the Msed1325 product, none of the three FlaJ-like proteins are similar to the Msed1324 product. Msed1198 has significant similarity to Msed2105, but neither is similar to Msed0651. Also of note is a prepilin/preflagellin peptidase gene located upstream of one of these FlaIJ pairs (Msed2090). With 237 amino acids and six predicted transmembrane helices, this FlaK-like sequence fits the characteristic size range and α-helix predictions for bacterial type II/IV systems (53).
To assess whether specific aspects of bioinformatics analysis of M. sedula genome sequence have a physiological basis, transcriptional profiling using a whole-genome oligonucleotide microarray was pursued. The addition of FeSO4 to a YE-supplemented medium triggered differential transcription of 88 genes (≥2.0-fold), or ~4% of the genome (Fig. (Fig.5;5; see Table S1 in the supplemental material). Of these, 53 were more highly transcribed in the presence of FeSO4, including several M. sedula ORFs predicted to be involved with Fe oxidation, sulfur metabolism, and adhesion. Six ORFs, most with only a general “major facilitator superfamily” annotation, were up-regulated in the presence of FeSO4+YE compared to YE (Msed0467, -0907, -1001, -1095, -1181, and -1355). Table Table11 lists differentially transcribed ORFs predicted to be involved with Fe oxidation, sulfur metabolism, and adhesion, while Fig. Fig.66 summarizes sequence-based predictions and preliminary transcriptome implications for these same three physiological functions. With respect to Fe oxidation, a putative rusticyanin (Msed1206), the locus 3 respiratory cluster (Msed0500 to Msed0504) expected to be up-regulated on Fe(II), and a unique putative quinol oxidase fusion protein (cytochrome b N-terminal and Rieske protein C-terminal domains) were highly transcribed in the presence of FeSO4. Several fox genes of locus 4 (specifically Msed0477 to Msed0480) were also highly transcribed on FeSO4, although levels in the presence of YE alone were at the upper end of the dynamic response range of the array. The exception to this was subunit I of the terminal oxidase (SoxB; Msed0484), which was more highly transcribed in the presence of FeSO4+YE. The implied involvement of M. sedula locus 4 genes in Fe oxidation builds on previous reports on the closely related autotrophic biomining microorganism S. metallicus (5, 30), and the influence of a heterotrophic substrate (YE) compared to FeSO4 has not previously been investigated. ORFs potentially involved in sulfur metabolism (Msed0960 to Msed0963; CysGIHN-like sequences, respectively) were up-regulated in the presence of FeSO4. With the addition of a sulfur compound in the +6 valence state, these ORFs, including the APS-SAT pair, may function in the reductive direction, reducing sulfate to sulfide. It is interesting that a cysteine synthase (CysM) gene-like ORF (Msed1607) was down-regulated on FeSO4+YE. This may be an attempt to limit accumulation of cysteine, which could lead to Fenton reaction oxidative damage to DNA (70), although it is not immediately apparent how excess sulfide is alternatively processed. Additionally, Msed1197 and Msed1198, similar to genes for cytoplasmic ATPase and transmembrane subunits, were stimulated in the presence of FeSO4 compared to YE alone. This FlaIJ pair was one of four Fla clusters predicted to be involved with adhesion, through either flagellum formation or EPS secretion. Although the three remaining Fla clusters may not function as predicted, it is also possible that their transcription has yet to be induced by the appropriate substrate. EPS production, composition, and properties have been observed to be a function of substrate in A. ferrooxidans (13, 24, 62). Finally, nearly half of the differentially transcribed ORFs in the YE-versus-YE+FeSO4 contrast were annotated as encoding “hypothetical proteins” or “proteins of unknown function,” indicating that novel proteins are involved in Fe oxidation metabolism in M. sedula. Efforts to understand Fe oxidation and other elements of M. sedula physiology using functional genomics approaches are under way.
Comparing genome sequences of extremely thermoacidophilic nonleachers, mesoacidophilic bioleachers, and M. sedula provides a number of preliminary insights into the genotype that supports high-temperature biomining. Such comparisons between M. sedula and genome-sequenced Sulfolobales indicate that the fox gene cluster, two putative versions of Rus, and TetH appear to be factors distinguishing thermoacidophilic bioleachers from nonbioleachers. However, it is still not clear whether bioleachers have acquired these genes or nonleachers have lost them (or their functionality). For example, the genome of the apparent nonleacher S. solfataricus encodes a protein similar to a putative Rus in M. sedula but is lacking a Rus protein signature and apparent Cu-binding ligands. S. tokodaii was recently demonstrated to have some degree of Fe oxidation capability and contains several fox genes (5), as well as a TetH, but is lacking a Rus-like sequence. Finally, the terminal oxidase components (locus 3) shown to be up-regulated on Fe(II) here, as well as in another study (36), were originally identified in the apparent nonleacher S. acidocaldarius (27). The differential transcription of significant numbers of hypothetical proteins in YE+FeSO4-grown versus YE-grown M. sedula cells confirms there are many other proteins involved in Fe oxidation and/or sulfur metabolism which may help to further distinguish between bioleachers and nonleachers in the future.
The future success of the biomining industry will depend upon the performance of cellular biocatalysts. The availability of the M. sedula genome sequence has advanced both functional genomics and genetics efforts in this area. To understand performance, upcoming functional genomics studies will focus on M. sedula oxidation of RISCs and reduced Fe substrates and how these processes relate to cell adhesion to solid inorganic substrates. Functional genomics analyses can also help guide genetic tools in target selection. Genetic tools in development for M. sedula (Y. Maezato and P. Blum, unpublished data) will be important for examining and enhancing performance of pathways and metal-microbe interactions essential to biomining operations at elevated temperatures.
We gratefully acknowledge the U.S. Department of Energy and the Joint Genome Institute for providing the genome sequence for M. sedula.
This work was supported in part by a grant from the U.S. National Science Foundation Biotechnology Program (no. BES-0317886). K.S.A. acknowledges support from an NIH T32 Biotechnology Traineeship.
Published ahead of print on 14 December 2007.
†Supplemental material for this article may be found at http://aem.asm.org/.