Selenium controls transcription of paralogous formate dehydrogenase genes in the termite gut acetogen, Treponema primitia


E-mail; Tel. (+1) 626 395 4182; Fax (+1) 626 395 2940.


The termite gut spirochete, Treponema primitia, is a CO2-reductive acetogen that is phylogenetically distinct from other distantly related and more extensively studied acetogens such as Moorella thermoacetica. Research on T. primitia has revealed details about the role of spirochetes in CO2-reductive acetogenesis, a process important to the mutualism occurring between termites and their gut microbial communities. Here, a locus of the T. primitia genome containing Wood-Ljungdahl pathway genes for CO2-reductive acetogenesis was sequenced. This locus contained methyl-branch genes of the pathway (i.e. for the reduction of CO2 to the level of methyl-tetrahydrofolate) including paralogous genes for cysteine and selenocysteine (Sec) variants of formate dehydrogenase (FDH) and genes for Sec incorporation. The FDH variants affiliated phylogenetically with hydrogenase-linked FDH enzymes, suggesting that T. primitia FDH enzymes utilize electrons derived directly from molecular H2. Sub-nanomolar concentrations of selenium decreased transcript levels of the cysteine variant FDH gene. Selenium concentration did not markedly influence the level of mRNA upstream of the Sec-codon in the Sec variant FDH; however, the level of transcript extending downstream of the Sec-codon increased incrementally with increasing selenium concentrations. The features and regulation of these FDH genes are an indication that T. primitia may experience dynamic selenium availability in its H2-rich gut environment.


The hindguts of many wood-feeding, phylogenetically lower termites harbour a rich diversity of microbial symbionts (Bignell, 2000). The combined metabolic activities of termite gut microorganisms underlie the stepwise conversion of the cellulosic and hemicellulosic fractions of wood lignocellulose into acetate, which serves as the primary carbon and energy source fuelling termite metabolism (Breznak, 1982; Odelson and Breznak, 1983; Brauman et al., 1992). In addition to acetate and CO2, H2 is an important metabolite released during the fermentation of cellulose and xylan by termite gut flagellate protozoa (Odelson and Breznak, 1985a,b). The standing concentration of free H2 produced in the termite gut can approach saturation, one of the highest levels associated with natural environments (Schink et al., 1983; Ebert and Brune, 1997; Hoehler et al., 2001; Pester and Brune, 2007). Possibly due to the striking availability of this electron donor as well as the atypically low competition for this substrate by methanogenic Archaea, hydrogenotrophic, CO2-reducing acetogenic bacteria thrive in this environment (Odelson and Breznak, 1983; Breznak and Switzer, 1986; Leadbetter et al., 1999). These bacteria, which employ the Wood-Ljungdahl pathway for energy metabolism and CO2 fixation, are responsible for up to 20–30% of the gut acetate pool (Breznak and Switzer, 1986).

Termite hindgut microbial populations often feature an abundance of spirochetes, a division of bacteria comprising their own phylum (Spirochaetes). Termite gut spirochetes belong mainly to the genus Treponema (Paster et al., 1996; Lilburn et al., 1999; Paster and Dewhirst, 2000) and they can account for 50–80% of all bacteria present in the gut of some termite species (Paster et al., 1996; Warnecke et al., 2007). Although a role for termite gut spirochetes in the process of acetogenesis was postulated more than 30 years ago (Breznak, 1973), only in the last decade were the first spirochete strains isolated from the termite gut and described as comprising a novel species, Treponema primitia, capable of acetogenic growth on H2 + CO2 (Graber et al., 2004). Treponema primitia thus provided the first direct link between spirochetes and the process of CO2-reductive acetogenesisoccurring in the termite gut environment (Breznak and Switzer, 1986).

Gene-centred studies on termite gut communities, enabled in part by analyses of T. primitia, have since reinforced the prediction that spirochetes have an important role in termite gut CO2-reductive acetogenesis (Salmassi and Leadbetter, 2003; Ottesen et al., 2006; Pester and Brune, 2006; Warnecke et al., 2007). These results underscore the continued relevance of physiological and gene-based studies on T. primitia, which is phylogenetically distinct from other acetogenic organisms and is the first and currently the only known chemolithotrophic spirochete (Salmassi and Leadbetter, 2003; Ottesen et al., 2006; Pester and Brune, 2006; Warnecke et al., 2007).

A previous study on T. primitia identified the nucleotide sequence of genes encoding methenyl-THF cyclohydrolase (MTHFC) and two partial nucleotide sequences of flanking genes encoding formyl-THF synthetase (FTHFS) and a bifunctional FolD-type methylene-THF dehydrogenase/cyclohydrolase (MTHFD/C) (Salmassi and Leadbetter, 2003). All three enzyme functions are important components of the ‘methyl-branch’ of the Wood-Ljungdahl pathway (Drake and Daniel, 2004). Based on phylogenetic analysis, it was proposed that these Wood-Ljungdahl pathway genes were acquired by an ancestral spirochete from a member of the phylum Firmicutes, i.e. via lateral gene transfer (Salmassi and Leadbetter, 2003). However, other genes associated with the methyl- and carbonyl-branches of the Wood-Ljungdahl pathway in T. primitia remained to be identified. Herein, a T. primitia genome insert library was screened for clones containing the MTHFC gene, which led to the identification of a region of the genome that included genes for all of the methyl-branch enzymes of the Wood-Ljungdahl pathway (i.e. those required for the reduction of CO2 to methyl-tetrahydrofolate). An analysis of two genes encoding formate dehydrogenase (FDH), the first enzyme in the methyl-branch of the Wood-Ljungdahl pathway, and their implications for the physiological ecology of T. primitia is discussed in detail.


A 52 kb contig was assembled from nucleotide sequence information generated from fosmid and genomic DNA templates (Fig. 1C). A local gap that included the 5′ end of the gene for formyl-THF synthetase in the fourfold-coverage genome insert library of T. primitia necessitated the use of inverse PCR to extend the sequence information beyond this gene. Collateral, unpublished evidence from our laboratory suggests that even partial treponemal FTHFS genes may be toxic to Escherichia coli, an issue that may underlie the gap in the coverage of the genome library used in this study. In total, 43 open reading frames (ORFs) were identified for investigation based on criteria of ORF size (> 90 codons, except tRNA) and homology (expect value < 1 e−3) to predicted proteins or genes in NCBI databases (Table 1).

Figure 1.

Treponema primitia ZAS-2 open reading frames (ORFs) located near genes for enzymes of the methyl-branch of the Wood-Ljungdahl pathway.
A. Putative ORFs based on criteria of ORF size (> 90 codons, except Sec-tRNA) and homology (expect value < 1 e−3) to predicted proteins or genes in NCBI databases. Direction of arrows indicates transcriptional orientation; ORF numbers refer to Table 1. Grey arrows, genes for folate-dependent enzymes of the Wood-Ljungdahl pathway. Bold open arrows, genes for formate dehydrogenases encoding cysteine (fdhFCys) or selenocysteine (fdhFSec) residues corresponding to amino acid position Sec140 of E. coli fdhF. Black arrows, hydrogenase components and FDH accessory protein. Diagonal striped arrows, genes for selenoprotein synthesis. Stippled arrows, ORFs located in a region of G+C content atypically low for this organism (Graber et al., 2004).
B. Plot of running mol% G+C content. Bracket, low GC region; horizontal line, the T. primitia genome 50–51% GC average.
C. Clones identified from a genomic DNA library and an inverse PCR product used as template for sequencing.

Table 1. Treponema primitia open reading frames (ORFs) identified in this study and closely matching protein homologues.
ORF No.aPutative gene product/functionClosest identified GenBank matchb (GenBank accession No.)Relationship to known organismsc
Amino acid span (identity, similarity)
  • a. 

    ORF numbers refer to gene map given in Fig. 1.

  • b. 

    Protein homologues located in NCBI databases were determined using blast algorithms (Altschul et al., 1997).

  • c. 

    Based on the indicated number of amino acid residues analysed, protein homologies are given in identical residues (identity) and identical residues plus conservative substitutions (similarity).

  • NA, not applicable.

 1CooCCO dehydrogenase maturation factor (ZP_01232264)149 (54%, 75%) Clostridium difficile
 2unkPredicted membrane protein (NP_782287)171 (36%, 63%) Clostridium tetani
 3CbiOABC-type cobalt transport ATPase (NP_623801)215 (41%, 66%) Thermoanaerobacter tengcongensis
 4CbiQABC-type cobalt transport permease (NP_349700)208 (25%, 43%) Clostridium acetobutylicum
 5MTHFR5,10-Methylene-tetrahydrofolate reductase (NP_662255)309 (33%, 52%) Chlorobium tepidum
 6FTHFSFormate-tetrahydrofolate ligase (YP_516838)564 (71%, 82%) Desulfitobacterium hafniense
 7MTHFCMethenyl-tetrahydrofolate cyclohydrolase (ZP_01373036)168 (47%, 64%) Desulfitobacterium hafniense
 8MTHFD/CMethylene-THF dehydrogenase/methenyl-THF cyclohydrolase (YP_360698)281 (48%, 66%) Carboxydothermus hydrogenoformans
 9SpollEStage II sporulation E family protein (YP_845967)397 (45%, 65%) Syntrophobacter fumaroxidans
10Sensor proteinPutative Fe–S PAS/PAC sensor protein (YP_845966)315 (44%, 61%) Syntrophobacter fumaroxidans
11unkHypothetical protein (ZP_01322618)39 (66%, 74%) Burkholderia pseudomallei
12unkHypothetical protein (YP_392995)104 (67%, 83%) Thiomicrospira denitrificans
13unkProtein of unknown function (ZP_00580765)324 (52%, 74%) Shewanella baltica
14ATPaseSMC protein-like (YP_580626)899 (30%, 51%) Psychrobacter cryohalolentis
15HelicaseHelicase domain protein (YP_099262)1681 (56%, 71%) Bacteroides fragilis
16unkHypothetical protein (XP_672950)165 (23%, 46%) Plasmodium berghei
17RepressorPutative transcriptional repressor protein (NP_273600)225 (29%, 48%) Neisseria meningitidis
18unkHypothetical protein (NP_773562)161 (40%, 67%) Bradyrhizobium japonicum
19unkHypothetical protein (YP_697488)177 (54%, 74%) Clostridium perfringens
20unkHypothetical protein (YP_503002)163 (39%, 58%) Methanospirillum hungatei
21unkHypothetical protein (YP_318696)181 (28%, 44%) Nitrobacter winogradskyi
22unkHypothetical protein (ZP_01354572)278 (39%, 53%) Clostridium phytofermentans
23MoeAMolybdenum cofactor biosynthesis protein (BAA76925)404 (36%, 58%) Clostridium perfringens
24unkHypothetical protein (NP_348615)173 (47%, 59%) Clostridium acetobutylicum
25FDHFormate dehydrogenase H (CAJ70215)716 (61%, 76%) Clostridium difficile
26HycBElectron transport protein (ZP_00907687)184 (45%, 63%) Clostridium beijerincki
27FdhDFormate dehydrogenase accessory protein (ZP_01014099)101 (39%, 61%) Rhodobacterales sp.
28HycBElectron transport protein (YP_077119)189 (48%, 61%) Symbiobacterium thermophilum
29HydAIron only hydrogenase large subunit (YP_001089830)449 (63%, 79%) Clostridium difficile
30MobAMolybdopterin-guanine dinucleotide biosynthesis protein A (YP_357250)127 (36%, 51%) Pelobacter carbinolicus
31HgdCHydroxyglutaryl-CoA dehydratase activator (NP_070783)254 (43%, 61%) Archaeoglobus fulgidus
32HgdB2-Hydroxyglutaryl-CoA dehydratase (YP_518979)379 (61%, 77%) Desulfitobacterium hafniense
33ABC transporterABC HisP/GlnQ permease (ZP_01230806)246 (67%, 85%) Clostridium difficile
34ABC transporterAmino acid ABC transporter, substrate-binding protein (CAJ69058)268 (36%, 54%) Clostridium difficile
35ABC transporterAmino acid ABC transporter, permease protein (CAJ69059)180 (52%, 77%) Clostridium difficile
36SelAl-seryl-tRNA selenium transferase (ZP_00962173)460 (45%, 61%) Sulfitobacter sp.
37SelBSelenocysteine-specific elongation factor (YP_074902)650 (36%, 55%) Symbiobacterium thermophilum
38CsdBSelenocysteine lyase (NP_816209)395 (44%, 60%) Enterococcus faecalis
39SirA-likeN-terminal SirA-like protein (YP_520926)209 (44%, 61%) Desulfitobacterium hafniense
40SelDSelenide, water dikinase (ZP_01372141)185 (54%, 67%) Desulfitobacterium hafniense
41Sec-tRNATransfer RNA, selenocysteinyl-tRNANA
42CodBHydroxymethylpyrimidine transporter GytX (ZP_01455232)386 (35%, 59%) Thermoanaerobacter ethanolicus
43FDHFormate dehydrogenase H (CAJ70215)715 (67%, 79%) Clostridium difficile

The region of the T. primitia genome examined in this study contains a four-gene cluster encoding methylene-THF reductase (MTHFR), FTHFS, MTHFC, and the complete, partially redundant FolD-type MTHFD/C (Fig. 1A, grey arrows). Together, these methyl-branch genes encode the enzymes responsible for the tetrahydrofolate-dependent functions involved in the reduction of carbon from the level of formate to methanol. When compared with orthologous genes found in other bacteria, each of the genes in this gene cluster affiliates phylogenetically with clostridial members of the phylum Firmicutes (data not shown).

Downstream from this four-gene cluster is a putative foreign DNA element that exhibited a marked decrease in G+C content: 36% over an 18 kb region (Fig. 1B, bracket), versus the c. 50% T. primitia genome average (Graber et al., 2004). The ORFs within this low G+C region (Fig. 1A, stippled arrows) generally have no homologues of known function in the available databases, with the exception of a putative ATPase, helicase and repressor protein.

Genes for FDH and selenocysteine metabolism and insertion

Two genes were identified that encode putative FDH enzymes (Fig. 1, bold open arrows labelled Cys and Sec). FDH enzymes have reversible catalytic activity and are diverse with respect to their roles in cell physiology as well as their associated physiological electron donors (Ferry, 1990b). In acetogenesis, these enzymes catalyse the first step in the reduction of CO2 through the methyl-branch of the Wood-Ljungdahl pathway (Ragsdale and Pierce, 2008). The T. primitia FDH enzymes cluster phylogenetically within a well-supported clade comprised of predicted hydrogenase-linked (FDHH-type) FDH enzymes (Fig. 2A, grey box). These hydrogenase-linked enzymes are predominantly found in non-acetogenic enteric γ-Proteobacteria that typically operate this enzyme in the reverse direction, i.e. during the oxidation of formate generated from pyruvate-formate lyase in anaerobic sugar metabolism (Andrews et al., 1997). A recently identified FDH enzyme in Moorella thermoacetica that groups at the base of this clade is also predicted to be hydrogenase-linked (Pierce et al., 2008).

Figure 2.

Phylogenetic position of T. primitia formate dehydrogenase (FDH) enzymes.
A. Protein phylogeny determined by Phylip ProML maximum-likelihood analysis of 424 unambiguously aligned amino acid positions corresponding to molybdopterin oxidoreductase domain (pfam00384). Per cent support (when greater than 50%) for the nodes of the tree were generated by 1000 bootstrap analysis of the data set using protein maximum-likelihood (reported above nodes) and maximum-parsimony (reported below nodes). Bar indicates distance given as 0.1 changes per amino acid position. Grey box highlights predicted hydrogenase-linked FDH enzymes similar to E. coli FDHH. Species indicated in bold contain FDH enzymes discussed in the text. Dots behind species indicate FDH sequences predicted to contain a catalytic selenocysteine. Numerals in the upper right corner of each box refer to domain feature types.
B. Domain structure of FDH enzymes as determined by comparing protein sequences against the InterPro database. FDH sequences were classified as ‘Type I, II, III or IV’ according to domain features and organization. Bar length is equal to 200 amino acids.

The organization of protein domains found within T. primitia FDH enzymes provide independent support of their phylogenetic position because only the molybdopterin oxidoreductase domain, the largest of the domains common to all FDH enzymes analysed (Fig. 2B, diagonal striped box), was used in phylogenetic tree reconstruction. All of the FDH enzymes contained within the FDHH clade included a configuration of protein domains similar to a ‘Type I’ organization (Fig. 2B). These enzymes contain three domains (molybdopterin oxidoreductase 4Fe–4S, molybdopterin oxidoreductase (Sec/Cys) and molybdopterin guanine dinucleotide binding) that together constitute the majority of the protein. While these molybdopterin domains were found in all of the FDH enzymes analysed, the hydrogenase-linked FDH enzymes notably lack a predicted NADH-ubiquinone oxidoreductase iron–sulfur binding domain common to NAD(P)H-linked enzymes such as the well-characterized NADPH-linked FDH of M. thermoacetica (Li et al., 1966; Thauer, 1972; Ragsdale and Pierce, 2008). A Type I domain configuration is also found in F420-linked FDH enzymes of methanogenic archaea (Schauer and Ferry, 1982); however, these enzymes are phylogenetically distinct from the FDH enzymes of T. primitia. The domain structure of T. primitia FDH enzymes together with the robust phylogenetic support for the FDHH clade suggests that T. primitia FDH enzymes may also be hydrogenase-linked. A gene for an [FeFe] hydrogenase large-subunit enzyme (hydA), a type of [FeFe] hydrogenase that was found to be abundant in a termite gut metagenome data set (Warnecke et al., 2007), along with two genes predicted to encode hydrogenase-associated electron transport proteins (hycB) was identified immediately downstream from one of the FDH genes in T. primitia (Fig. 1, black arrows).

Analysis of the nucleotide sequence of the two T. primitia FDH genes revealed a key difference between them. One FDH gene encodes a putative selenoprotein with an in-frame TGA stop codon at amino acid position 145 corresponding to the catalytic selenocysteine (Sec140) encoded by TGA in the E. coli FDH gene, fdhF (Axley et al., 1991; Gladyshev et al., 1996), while the other encodes a cysteine at the corresponding position (Cys149). Incorporation of the non-canonical amino acid selenocysteine during translation in bacteria is known to be facilitated by Seleno Cysteine Incorporation Sequence (SECIS) elements, which are nucleotide sequences that form mRNA stem–loop structures located immediately downstream from TGA codons for Sec (Fourmy et al., 2002; Krol, 2002; Thanbichler and Böck, 2002). Using the search algorithm bSECIS (Zhang and Gladyshev, 2005) and mFold (Zuker, 2003), SECIS-like stem–loop structures were identified in both T. primitia FDH genes (Fig. 3A). The location of the Sec codon and the SECIS element strongly suggests that the FDH gene containing the in-frame TGA codon in T. primitia (hereafter referred to as fdhFSec) encodes a selenoprotein. Phylogenetic analysis (Fig. 2) revealed that the two FDH proteins of T. primitia are each other's closest relatives, grouping to the exclusion of all others currently in public databases. This relationship and the presence of a predicted SECIS-like mRNA secondary structure within the Cys-containing FDH (fdhFCys) suggestthat fdhFCys might have arisen via a gene duplication event with subsequent mutational modification, meaning that the relationship between fdhFSec and fdhFCys genes of T. primitia is likely one of paralogy (Sonnhammer and Koonin, 2002).

Figure 3.

Predicted SECIS-like elements in T. primitia formate dehydrogenases (FDHs).
A. mRNA secondary structures located downstream from the selenocysteine (fdhFSec) and cysteine (fdhFCys) codons in the FDH variants. Stem–loop structures were predicted using bSECIS and mFold algorithms (Zuker, 2003; Zhang and Gladyshev, 2005). A guanine nucleotide located near the apex of stem loop (marked in bold) has been shown in other SECIS elements to be involved in SelB binding.
B. Nucleotide and translated amino acid sequences of a 50 bp region spanning the selenocysteine or cysteine codon and stem–loop structures in fdhFSec (top) and fdhFCys (bottom). Nucleotides comprising the stem–loop structures are underlined. Mutations resulting in compensatory changes that preserve base-pairing in the fdhFCys stem structure are marked with asterisks.

Identification of a SECIS-like secondary structure in fdhFCys was unexpected. The predicted architecture of the stem–loop structure in fdhFCys is consistent with a SECIS element, conserving the spacing from the cysteine codon, apical stem length, and the location of a guanine nucleotide in the apical loop required for SelB binding (Heider and Böck, 1992; Liu et al., 1998; Yoshizawa et al., 2005; Zhang and Gladyshev, 2005). However, numerous mutations have contributed to divergence (0.39 changes per nucleotide position; compared with 0.32 changes per nucleotide position for the entire gene) between the sequence of the fdhFSec stem–loop structure and the corresponding nucleotide positions in fdhFCys (Fig. 3B). Interestingly, the majority of these mutations (10 out of 17) have resulted in compensatory changes that preserve a stem–loop structure. In addition, the region spanning the fdhFCys stem–loop exhibits an elevated G+C content over its fdhFSec counterpart as well as the fdhFCys gene on the whole (68.4% versus 54.5% and 54.9% respectively). This has contributed to greater predicted stability in the fdhFCys stem–loop structure for which the free energy is −14.6 kcal mol−1 compared with −7.8 kcal mol−1 for the fdhFSec structure as calculated by mFold (Zuker, 2003).

The synthesis of selenocysteine and its incorporation into nascent polypeptides requires additional enzymes and other factors. All anticipated genes for such are located in a gene cluster near fdhFSec. A gene for the SECIS element-binding, selenocysteine translation elongation factor (selB) is located downstream from a gene for l-seryl-tRNA selenium transferase (selA) in a putative two-gene operon (Fig. 1A, diagonal striped arrows). Adjacent to this gene cluster and oriented in the opposite transcriptional direction are genes encoding selenocysteine lyase (csdB), a response regulator-like protein (sirA), and a selenide, water dikinase (selD). Selenocysteine lyase is involved in degradation of environmental or protein turnover-derived selenocysteine in the generation of alanine and elemental selenium; selenide, water dikinase synthesizes monoselenophosphate (required by SelA to modify seryl-tRNA into selenocysteinyl-tRNA) from a reduced selenium species and ATP (Veres et al., 1994; Lacourciere et al., 2000).

No other characterized spirochetes are known to possess FDH genes or demonstrate FDH enzyme activity. However, an oral spirochete closely related to T. primitia, Treponema denticola, contains a putative selenium-containing glycine reductase and SelB (Rother et al., 2001). We examined the phylogenetic position of the SelB homologues from these two treponemes (Fig. 4). The analysis indicates that the SelB proteins of these related spirochete species have different protein phylogenies. SelB from T. primitia affiliates with SelB of Dethiosulfovibrio peptidovorans, a species belonging to the phylum δ-Proteobacteria, within a loosely associated cluster of homologues from anaerobic Firmicutes. In contrast, the SelB of T. denticola grouped with SelB of two distantly related spirochetes Brachyspira hyodysenteriae and B. murdochii within a well-supported clade that included SelB proteins from several members of the phylum γ-Proteobacteria.

Figure 4.

Phylogenetic position of T. primitia selenocysteine elongation factor (SelB). Protein phylogeny was determined by Phylip ProML maximum-likelihood analysis of 312 unambiguously aligned amino acid residue positions. Per cent support (when greater than 50%) for the nodes of the tree were generated by 1000 bootstrap analysis of the data set using protein maximum-likelihood (reported above nodes) and maximum-parsimony (reported below nodes). Bar indicates distance as 0.1 changes per amino acid position. Species in bold type belong to phylum Spirochaetes.

Influence of selenium on growth

The influence of different forms of selenium on the growth of T. primitia was examined, as selenoproteins are often superior with respect to catalytic efficiency to their cysteine-containing analogues (Jones and Stadtman, 1981; Axley et al., 1991; Berry et al., 1992; Lee et al., 2000; Gromer et al., 2003; Metanis et al., 2006), and because FDH is a central enzyme in the carbon and energy metabolism of this CO2-reducing, acetogenic spirochete. No reproducible growth rate or yield difference was observed over a range of sodium selenite quantities added to T. primitia cultures (0.05–100 nM) compared with control cultures; higher concentrations (500 nM) resulted in a growth defect (data not shown). Several additional selenium compounds (sodium selenate, selenium dioxide and selenocystine) added at 50 nM concentrations similarly did not influence growth rate. As molybdenum is an essential cofactor in FDH enzymes (Ferry, 1990a), we tested whether or not Mo was growth limiting at the concentration routinely used to grow T. primitia (36 µg l−1). Higher sodium molybdate concentrations (360 µg l−1, 1.8 mg l−1 or 18 mg l−1) provided no growth benefit. Finally, measurements of end-point acetate production and culture headspace gas consumptions did not reveal significant reproducible differences between the different selenium treatment conditions tested and control cultures.

Influence of selenium on FDH transcription

To examine the possible influences of selenium on the expression of select genes, total RNA was extracted from cells harvested during exponential growth of T. primitia cultures grown under several selenium treatments. Transcript levels for clpX, fdhFCys, fdhFSec and selB were determined using quantitative RT-PCR. Transcript levels of a putative housekeeping gene (clpX) did not change appreciably (< 2-fold, on average) for any condition tested. Therefore, clpX transcript levels were employed as endogenous internal standards to normalize for sample handling.

The expression of fdhFCys was negatively controlled by each of the selenium forms tested, with up to c. 40-fold decreases (normalized to clpX) in the levels of fdhFCys transcript observed (Table 2). When RT-PCR primers targeting the portion of fdhFSecupstream from the SECIS element were employed, the addition of selenium compounds was not observed to affect transcript levels (Table 2). In contrast, when primers targeting a region downstream from the SECIS element in fdhFSec were employed, transcript levels increased up to 10-fold (Table 2) in cultures amended with reduced selenium forms such as sodium selenite, selenium dioxide or selenocystine compared with unamended cultures, but not with the fully oxidized form, selenate.

Table 2.  Effect of selenium compounds on T. primitia formate dehydrogenase transcript levels.
TreatmentbFold change in transcript levelsa
  • a. 

    Given in fold increase or decrease relative to selenium-free controls. Values (± the standard error of the mean) represent the average of three replicate experiments measured in duplicate and corrected for variations in clpX transcript abundance. NC, no change, meaning experimental values varied less than ± 2-fold from controls.

  • b. 

    Cultures amended with selenium compounds (50 nM final concentration) at the time of inoculation.

Sodium selenite−39 ± 4NC11 ± 2
Sodium selenate−31 ± 5NCNC
Selenium dioxide−38 ± 3NC11 ± 3
Seleno-dl-cystine−40 ± 3NC11 ± 2

As a means to assess the sensitivity of the T. primitia selenium response, transcript levels of fdhFCys and fdhFSec were measured in cell cultures amended with a wide range (0.05–100 nM) of sodium selenite concentrations. Transcription of fdhFCys was markedly more sensitive to sodium selenite than was fdhFSec (Fig. 5). The amount of sodium selenite associated with a half-maximal decrease in transcript level of fdhFCys was less than 50 pM, whereas the amount required for half-maximal increase in transcript level of full-length fdhFSec was approximately 1.5 nM. The transcription of selB was constitutive and not influenced by selenite concentration.

Figure 5.

Sodium selenite concentration dependence of formate dehydrogenase transcription.
A. Schematic of the location of RT-PCR amplicons relative to the Sec codon and SECIS element in a hypothetical transcript for fdhFSec (not drawn to scale).
B. Quantitative RT-PCR amplification signal intensities of formate dehydrogenase and selB transcripts measured in T. primitia cultures grown with different selenium concentrations. Symbols: closed circles, fdhFSec (5′); open circles, fdhFSec (3′); closed triangles, fdhFCys (5′); +, selB. Values of all data points represent the average of two replicate experiments measured in duplicate. Lines are intended to guide the eye through the data for fdhFCys (5′) (solid line) and fdhFSec (3′) (dashed line).


In this study, the T. primitia gene for MTHFC was used to identify other key methyl-branch genes of the Wood-Ljungdahl pathway for CO2-reductive acetogenesis. A genome locus containing 18 ORFs associated with Wood-Ljungdahl pathway functions was identified. Among these ORFs were two genes encoding two FDH variants, one containing a catalytic selenocysteine amino acid residue and the other containing a cysteine substitute at the comparable amino acid position.

Although T. primitia belongs to the monophyletic group of bacteria known as spirochetes, several of the Wood-Ljungdahl pathway enzymes associate phylogenetically with homologues from members of the bacterial phylum Firmicutes. This had led Salmassi and Leadbetter to propose that lateral gene transfer from one or more Firmicutes in the distant past may have contributed to the capacity for actogenesis in T. primitia (Salmassi and Leadbetter, 2003). In support of this hypothesis, we found that the FDH enzymes of T. primitia associate with FDH enzymes of several Firmicutes belonging to the genus Clostridium (Fig. 2A). We explored whether or not the T. primitia SelB enzyme required for the synthesis of the Sec-variant FDH was similar to known SelB enzymes in other spirochetes. The protein phylogeny shows that SelB from Treponema pallidum, a spirochete species belonging to same genus as T. primitia, affiliates with SelB of other, more distantly related spirochete species belonging to the genus Brachyspira (Fig. 4). From this phylogenetic relationship, it appears that SelB arose in T. primitia independently from other spirochetes and thus, like acetogenesis, the capacity for the synthesis of selenoproteins in T. primitia may be an acquired trait.

The finding that T. primitia FDH enzymes identified in this study cluster phylogenetically with hydrogenase-linked rather than NAD(P)H-linked FDH enzymes such as the one used for acetogenesis in M. thermoacetica (Li et al., 1966) has biochemical implications. In the well-characterized Enterobacteriaceae FDHH enzyme from E. coli, formate generated by pyruvate-formate lyase during hexose fermentations is oxidized to yield H2 and CO2 (Zinoni et al., 1986; Axley et al., 1991; Andrews et al., 1997; Hakobyan et al., 2005). Although, to our knowledge, there has not yet been a biochemical demonstration of any hydrogenase-linked FDH enzyme operating in the direction of CO2 reduction (as most FDH enzyme assays are performed in the direction of formate oxidation) and [FeFe] hydrogenases, such as the one located near by fdhFCys in T. primitia, often function in the direction of H2 production (Frey, 2002), the unambiguous H2-dependent generation of 14C-formate from 14C-CO2 has previously been observed to occur during in vitro and in situ analyses of termite gut contents (Breznak and Switzer, 1986; Brauman et al., 1992; Tholen and Brune, 1999) and in T. primitia cultures (Leadbetter et al., 1999). These observations are consistent with the concept that termite gut acetogens like T. primitia are operating FDH enzymes in the direction of CO2 reduction with electrons derived from H2. During the preparation of this report, the genome of T. primitia was closed (S. Tetu, X. Zhang, A. Rosenthal, Q. Ren, R. Seshadri, E. Matson, et al., unpublished) and an analysis of the draft genome revealed no additional recognizable FDH enzymes of any type in T. primitia.

The initial step of acetogenesis is calculated to be an energy-requiring reaction. Under standard conditions ΔG′o of CO2(g) + H2 ⇔ HCO2- + H+ is +3.4 kJ reaction–1 (Fuchs, 1986). In termite gut environments such as the one T. primitia inhabits, H2 generated during polysaccharide and sugar fermentations by protozoa (Yamin, 1981; Odelson and Breznak, 1985a,b) and other spirochetes (Graber et al., 2004) reaches high standing concentrations of up to 70 kPa (Pester and Brune, 2007). At 70 kPa H2 the thermodynamic value of CO2 reduction to formate is similar to ΔG′0 (i.e. +4 kJ mol−1 assuming physiological concentrations of CO2(aq), 3.2 mM and formate, 3.5 mM) (Thauer et al., 1977; Ebert and Brune, 1997; Tholen and Brune, 2000). In E. coli, the FDHH operates in the direction of formate oxidation and is associated with either hydrogenase 3 (Hyc) or hydrogenase 4 (Hyf) (Andrews et al., 1997). Hyc and Hyf are both membrane-associated [NiFe] hydrogenases. When FDH is linked with Hyf the complex operates in an energy-conserving process that couples formate oxidation to proton translocation (Andrews et al., 1997). If T. primitia uses a similar proton-traslocating formate–hydrogen lyase complex to couple hydrogen uptake and CO2 reduction, the dissipation of proton motive force could supply the small amount of energy calculated to be required for the first step in acetogenesis in the termite gut environment; however, whether or not the [FeFe] hydrogenase identified in this study is membrane associated or is ion-translocating remains to be investigated.

The use of hydrogenase-linked FDH enzymes may reflect a specialization whereby T. primitia has adapted to its high H2 environment by using H2 directly instead of NAD(P)H as an electron donor for CO2-reductive acetogenesis. Treponema primitia may benefit from a gain in efficiency in the reduction of CO2 to formate by using a formate–hydrogen lyase-like complex. Acetogens that use NADPH-linked FDH enzymes like M. thermoacetica can generate reducing equivalents for CO2-reductive acetogenesis from a wide variety of fermentable substrates. Doing so requires additional enzymatic steps to generate a sufficiently high NADPH/NADP+ ratio for the reduction of CO2 to formate. The flexibility that T. primitia loses in using hydrogenase-linked FDH enzymes may be offset by the abundance of hydrogen in its environment combined with possible efficiency gains over other, physiological electron donors in the reduction of CO2 by using electrons derived directly from H2.

The phylogenetic relationship between fdhFSec and fdhFCys supports the conclusion that these genes are paralogues (Fig. 2). We speculate that the latter arose after a gene duplication event and has been subsequently modified and maintained to meet the challenge of periods of selenium limitation that we infer this treponeme may encounter in the termite gut. Although selenocysteine-containing enzymes often provide greater catalytic efficiency than cysteine-containing enzymes (Jones and Stadtman, 1981; Axley et al., 1991; Hazebrouck et al., 2000; Kim et al., 2006), conditions of selenium limitation would allow an otherwise inferior cysteine version of the enzyme to be useful, if not essential. Organisms previously identified to have both Sec and Cys (though not necessarily paralogous) variants of key enzymes, and to regulate the expression of these genes as a function of selenium availability, include: the two FDHs in Methanococcus vannielii (Jones and Stadtman, 1981), [NiFe] and [NiFeSe] hydrogenases in Desulfovibrio vulgaris Hildenborough (Valente et al., 2006), the Frc and Vhc hydrogenases of Methanococcus voltae (Halboth and Klein, 1992; Berghöfer et al., 1994; Noll et al., 1999; Sun and Klein, 2004), and the formylmethanofuran dehydrogenases of Methanopyrus kandleri (Vorholt et al., 1997).

The addition of defined concentrations of an assortment of selenium compounds caused a reduction in the abundance of fdhFCys mRNA and an increase in the abundance of fdhFSec mRNA in comparison with mRNA levels in T. primitia cells grown in control cultures. Although part of the fdhFSec gene upstream from Sec145 was constitutively transcribed, production of full-length mRNA was dependent on the addition of a selenium source. The details underlying the mechanism by which this occurs remain to be studied, although one possibility that is consistent with the data is that fdhFSec transcription might be linked to protein translation in T. primitia. The SECIS element along with its interaction with Sec-tRNA and transcription elongation factor SelB, which was found to be transcribed regardless of the selenium treatment, directs selenocysteine incorporation into nascent peptides. We hypothesize that the increasing abundance of full-length fdhFSec transcript measured at incrementally higher selenium concentrations in the growth medium could depend on the frequency of Sec translation driven by increasing amounts of Sec-tRNA formed within T. primitia cells. A SECIS-like stem–loop structure was also identified downstream from the cysteine codon in fdhFCys and it appears to have been maintained over evolutionary time; however, it is not clear what function, if any, it plays in transcription or translation and these roles have yet to be explored experimentally in any system.

The observation that cultures amended with 50 pM selenite exhibited low levels of transcript for both FDH genes (Fig. 5) was surprising. Lower FDH transcript levels on the whole might be expected to ultimately restrict the growth ofT. primitia by causing a bottleneck in CO2 reduction into acetate, the energy-yielding pathway driving the metabolism of this organism. Curiously, neither growth rate nor yield was impacted by changes in selenium concentration or form. It may be that FDH is not the rate-limiting step in acetogenesis under the conditions used in these studies or that the growth of T. primitia or production of active FDH enzyme in laboratory media is not unrestricted with respect to other nutrients, thus remaining so suboptimal that low FDH transcript levels have no measurable negative impact on growth. In that regard, T. primitia grows, at best, with an approximately 29 h doubling time (Graber and Breznak, 2004), whereas several closely related free-living and oral Treponema species grow with well under 10 h doubling times (Hespell and Canale-Parola, 1970; Pöhlschroeder et al., 1994; Mikx, 1997). The impact of selenium availability on growth could be re-evaluated in the future should there be any marked improvement in the formulation of the cultivation medium for this species.

The prevailing environmental conditions, challenges and opportunities T. primitia and other symbiotic acetogenic spirochetes like it may face are undoubtedly different from those encountered by free-living acetogens. The results described in this study highlight several potential adaptations of T. primitia towards being an obligate termite gut symbiont and CO2-reducing acetogen. First, lateral gene transfer has contributed to the acquisition of Wood-Ljungdahl pathway genes in T. primitia and may have been, or still be (Graber and Breznak, 2005), an influential process in shaping the genomes of other members in the hindgut community. Second, the hydrogenase-linked FDH enzymes possessed by T. primitia are predicted to be favoured over FDH enzymes using other, physiological electron donors for acetogenesis in an environment with such an ample standing supply of H2. Finally, T. primitia has acquired and maintained both a selenium-dependent and a selenium-independent FDH, along with mechanisms for their regulation, which would allow this organism to respond to spatial or temporal fluctuations of selenium availability in its environment.

Experimental procedures

Bacterial cultivation

For routine cultivation, T. primitia str. ZAS-2 cells were grown in anaerobic YACo medium with 20 mM maltose and 4% yeast autolysate under an atmosphere of 80% H2 + 20% CO2 as previously described (Leadbetter et al., 1999). Growth of T. primitia cultures was assessed spectrophotometrically (OD600) and by measuring acetate production by HPLC (Graber and Breznak, 2004) and headspace gas consumption. For studies on selenium metabolism, all glassware was washed with HCl (10% v/v) and cells were grown on H2 + CO2, in a medium employing a specially prepared yeast autolysate. For preparation of selenium-free yeast autolysate, Red Star baker's yeast (Lesaffre Yeast Corp., Milwaukee, WI) pure cultures were grown in defined, selenium-free, minimal medium MV containing 2% dextrose (Miozzari et al., 1978). After growth for 36 h, the yeast cells were pelleted (yielding biomass equivalent to 7 g dry weight from 6 l of 36-h-old cultures), washed twice with distilled, deionized H2O, and suspended in 25 ml of the same. The suspension was incubated at 56°C for 48 h and then pelleted at 3700 g for 30 min to remove the autolysed cells. Se-free yeast autolysate protein concentrations were comparable to autolysates prepared from commercially available yeast as described previously (Leadbetter et al., 1999), and supported the growth of T. primitia at similar rates when added at the same concentration (4% v/v). Treponema primitia cultures were serially passed in selenium-free medium (minimum of two transfers) prior to their use as inocula for selenium experiments. Sodium selenite (Cat. No. S5261), sodium selenate (Cat. No. S8295), selenium dioxide (Cat. No. 325473) and seleno-dl-cystine (Cat. No. S1650) were purchased from Sigma-Aldrich Corp. (St. Louis, MO) and prepared as 50× concentrated stock solutions in water. The stock solutions were filter-sterilized (0.2 µm) into sealed anaerobic bottles and vacuum-purged with N2 to make them anoxic. Approximately 100 µl volumes of the selenium stock solutions (or a similarly treated water control) were used to amend growth medium (5 ml) prior to inoculation.

Genome insert library screening

A fourfold-coverageT. primitia str. ZAS-2 genome insert fosmid library, consisting of 576 E. coli clones with an average insert size of 30 kb, was screened by PCR using primers MTCF (5′-TGG TGG GCT CCC TTA CGG-3′) and MTCR (5′-CGG TCA TGG CAG CGG TAT-3′) targeting the methenyl-THF cyclohydrolase gene [GenBank Accession No. AY254548 (Salmassi and Leadbetter, 2003)]. These primers identified a single clone in the library, pCC1.57A, the 5′-terminus of which contained the 3′-flank of the partial FTHFS gene, and the entire MTHFC gene (Fig. 1C). To span a local gap in the genome insert library, which included a portion of FTHFS, a 3.3 kb inverse PCR product (Ochman et al., 1990) generated from T. primitia genomic DNA was used to sequence the remainder of that gene and to identify clone pCC1.25D located upstream from FTHFS. To generate the inverse PCR product, T. primitia ZAS-2 DNA was restriction digested with AgeI at 37°C for 1 h. The digested DNAs (200 ng) were ligated overnight at 15°C using T4 DNA ligase enzyme (1 U µl−1). Ligated products (1 ng) were PCR amplified using outwardly facing primers. The PCR amplicon (3.3 kb) was purified and used directly as template for sequencing. A third clone, pCC1.34E, that extended downstream from pCC1.57A was identified using primers targeting the end of pCC1.57A opposite the MTHFC gene.

Nucleotide sequencing and gene identification

Fosmids were extracted using HiSpeed Midi kit columns (QIAGEN Sciences, Valencia, CA) from 50 ml volumes of E. coli clones grown overnight at 37°C in LB broth with chloramphenicol selection (12 µg ml−1). The fosmids were further purified and concentrated using Microcon YM-100 (100 kDa cut-off) centrifugal membrane filter devices (Millipore Corp., Bedford, MA). Cloned inserts were sequenced inward from the vector priming sequences. Sequencing reactions were performed with a capillary-based, Applied Biosystems 3730 DNA analyser at the California Institute of Technology Sequencing Facility.

Sequencing reads were trimmed, assembled and edited using the SeqMan module of the Lasergene v. 7.0 sequence analysis software (DNASTAR, Madison, WI). Putative ORFs were predicted and G+C composition was determined using the GeneQuest module of the same software package. Nucleotide sequences were analysed using the tRNAscan-SE (v.1.21) algorithm to identify tRNA genes (Lowe and Eddy, 1997). Translated nucleotide sequences of ORFs identified in this study were compared with sequences in public databases at the National Center for Biotechnology (NCBI) using blast algorithms (Altschul et al., 1997). Phylogenetic analysis of several genes was performed using the ARB software package v. 05.05.26 (Ludwig et al., 2004). Phylogenetic trees were reconstructed from filtered amino acid sequence alignments created using muscle (Edgar, 2004). Figure legends of phylograms provide details regarding the treeing methods and number of residues used in the filtered analysis of unambiguous alignments.

RNA extraction and reverse transcription

Upon harvesting cells from 5 ml of cultures of T. primitia cells, RNA was immediately stabilized by the addition of 10 ml of RNAprotect Bacteria Reagent (QIAGEN). Cells were pelleted via centrifugation (5000 g for 10 min), and total RNA was extracted using RNeasy Mini columns with on-column DNase I treatment (QIAGEN). RNA was then subjected to a second, 30 min, 37°C off-column DNA digestion using RQ1 DNase enzyme (0.1 U µl−1) in 1× DNase buffer (Promega Corp., Madison, WI). Following the second digest, RNA samples were again purified with RNeasy columns to remove the DNase enzyme. This RNA purification regimen consistently yielded RNA samples of approximately 100 ng µl−1. The RNA samples (500 ng total) were immediately converted to cDNA by randomly primed reverse transcription using iScript reverse transcriptase and cDNA synthesis premix (Bio-Rad Laboratories, Hercules, CA). Duplicate samples lacking reverse transcriptase were prepared for each RNA sample as a negative control to assess residual DNA contamination.

Quantitative PCR

Primers were designed using Primer3 release 1.0.1 (Rozen and Skaletsky, 2000) to amplify regions of the T. primitia FDH genes fdhFSec and fdhFCys located upstream and downstream of the SECIS-like stem–loop structures (Fig. 5A). Primer set QfdhFSecF5′ (5′-GTA CGG CTG GGA TTA CCT CA-3′), QfdhFSecR5′ (5′-TTC CAG AGG GCT ATT TTT GC-3′), and set QfdhFCysF5′ (5′-GTT GTT ACT GCG GGA CAG GT-3′), QfdhFCysR5′ (5′-GTA CCA GCC CTT CAA ACA GG-3′) generated amplicons of 94 bp and 125 bp for the upstream portions of fdhFSec and fdhFCys respectively. Primer set QfdhFSecF3′ (5′-TGA CTC GGT CTT TTC CTG CT-3′), QfdhFSecR3′ (5′-GCT GAT GAT TTC CCA GTC GT-3′), and set QfdhFCysF3′ (5′-CAT CCC GGA AAT TGA GAA TG-3′), QfdhFCysR3′ (5′-TTC CTT GGC CTT GAC GAT AC-3′) generated amplicons of 97 bp and 100 bp for the downstream portions fdhFSec and fdhFCys respectively. The fdhFSec and fdhFCys primer sets did not amplify their non-target fdhF variants, as verified using copies of fdhFSec and fdhFCys cloned separately into plasmid vectors as template in PCR control reactions. For the amplification of selB, the primer set QselBF (5′-GCG GTG ATA TTA CGG TTG CT-3′), QselBR (5′-ATC CCT GTG CAT CCA GAA AG-3′) generated an amplicon of 93 bp for the selB gene.

Transcript levels for the T. primitia ZAS-2 gene clpX were monitored as an endogenous RT-PCR control. As this gene had not yet been identified in T. primitia, degenerate primers X311f (5′-GCI GTK TAC AAC CAY TAY AAR YG-3′) and X1066r (5′-AAG CWA DIG AIG CCT GRS ACT G-3′) were designed using an alignment of T. pallidum (Fraser et al., 1998), T. denticola (Seshadri et al., 2004), and putative clpX sequences from a termite hindgut bacterial community metagenomic data set (Warnecke et al., 2007), identified as likely being treponemal via PhyloPythia analysis (Mchardy et al., 2007). PCR with these degenerate primers (10 pmol µl−1, each) was performed using T. primitia ZAS-2 DNA (10 ng) and Expand High Fidelity Polymerase (Roche Applied Science, Indianapolis, IN) in FailSafe PCR PreMix D (Epicentre Biotechnologies, Madison, WI). The PCR amplicon was cloned and sequenced. Specific primers for amplifying clpX by real-time PCR were designed using Primer3. The primer set QclpXF (5′-CTC CCG TTT CAT TTC TTC CA-3′), QclpXR (5′-GAA ATG TTA GAC GCC CTC CA-3′) generated an amplicon of 108 bp. The PCR product was sequenced to verify the identity as clpX.

Quantitative PCR was performed in 15 µl reaction volumes of iQ SYBR Green Supermix (Bio-Rad Laboratories) using 25 ng of cDNA per reaction and forward and reverse primers (10 pmol each) for fdhFSec, fdhFCys, selB and clpX in separate reactions. A parallel set of reactions was prepared for each primer set using a 10-fold dilution series of T. primitia genomic DNA as template. Thermocycling and amplification detection was performed using a Bio-Rad DNAEngine thermocycler outfitted with a Chromo4 real-time detector. Thermocycling conditions for all quantitative PCRs were: initial denaturation at 95°C for 3 min followed by 44 cycles of 95°C for 15 s and 60°C for 30 s.

The nucleotide and deduced amino acid sequences for genes identified in this study have been deposited in GenBank under Accession No. FJ479767 (fragment containing clpX) and FJ479768 (large genome fragment).


This research was supported by NSF Grant DEB-0321753 (J.R.L.) and a NSF predoctoral fellowship (X.Z.). We thank John Breznak and Kwi Suk Kim for providing the T. primitia ZAS-2 genomic library, construction of which was supported by NSF Grant IBN-0114505. We thank Elizabeth Ottesen and our other laboratory colleagues for their helpful discussions and comments.