Introduction

Although key microbial players in various biogeochemical processes have been identified, they are often not available in pure culture. One major group of such organisms is large sulfur-oxidizing bacteria that have the capacity to accumulate a large amount of intracellular nitrate. In habitats where a sufficient supply of sulfide is available, they can form dense and widespread bacterial mats. Because of their large biomass and ecophysiological characteristics, they have been regarded as important constituents of aquatic ecosystems. In fact, they have been shown to have a considerable impact on nitrogen and phosphorus dynamics in marine sediments (Zopfi et al., 2001; Schulz and Schulz, 2005; Prokopenko et al., 2013).

Until recently, these organisms were classified into three genera, Beggiatoa, Thioploca and Thiomargarita, on the basis of their morphological features (Salman et al., 2011). Although recent studies have revealed that these microorganisms are much more diverse than previously thought (Salman et al., 2013), all known nitrate-storing sulfur oxidizers belong to a particular lineage within the class Gammaproteobacteria. Using a large set of 16S ribosomal RNA (rRNA) gene sequences, the bacteria in this lineage were reclassified in 2011, and the revised family Beggiatoaceae was proposed to encompass all of these bacteria (Salman et al., 2011). The extended family contains three genera along with nine candidate genera, two of which were proposed after the reclassification (Salman et al., 2013). From this family, only a few strains of the genus Beggiatoa have been isolated in pure culture (Salman et al., 2013).

As a cultured representative of this family, Beggiatoa alba B18LD has been subjected to whole-genome sequencing, and its draft genome sequence is now available in public databases; however, this strain cannot accumulate nitrate. The draft genome sequences of nitrate-storing sulfur oxidizers have been obtained for Candidatus Isobeggiatoa and Candidatus Parabeggiatoa, both of which are from coastal marine sediments (Mußmann et al., 2007), and for Candidatus Maribeggiatoa, which is from a deep-sea sediment that is influenced by hydrothermal fluid (MacGregor et al., 2013a, 2013b). These genome sequencing projects all employed whole-genome multiple displacement amplification to obtain sufficient amounts of DNA for sequencing from single bacterial filaments that are expected to consist of clonal cells. The single-filament approach may be effective for coping with genetic diversities among the morphologically indistinguishable organisms inhabiting the same sediment, but risks generating chimeric sequences during the process of amplification. Presumably because of the presence of such chimeric sequences and/or other difficulties (for example, short reads and repeat regions in the genomes), sequence assembly in these studies was not fully successful, as illustrated by large numbers of contigs (>800 contigs). Although these draft genome sequences have provided valuable insights into the physiology and evolution of this group of microorganisms, the availability of complete genome sequences of large sulfur-oxidizing bacteria is highly desirable.

Members of the genus Thioploca are gliding filamentous bacteria that have a common sheath surrounding the trichomes. The first description of the genus was made in the early twentieth century, with the type species of Thioploca schmidlei obtained from freshwater lake sediment (Lauterborn, 1907). Marine species with much larger cell sizes were once included in this genus, but they have been reclassified within a candidate genus, Candidatus Marithioploca (Salman et al., 2011). Thioploca ingrica was described as the second species of the genus (Wislouch, 1912) and was listed in the Approved Lists of Bacterial Names after a temporary loss of status as valid name (Maier, 1984). It has been retained in this genus, after the reclassification of the family Beggiatoaceae (Salman et al., 2011), as the sole species whose 16S rRNA gene sequences are available. Morphologically, T. ingrica was defined as a Thioploca species having a trichome 2.0–4.5 μm in diameter (Wislouch, 1912; Maier, 1984). Organisms that fit this description have been found in freshwater and brackish sediments of various localities, and the placement of these organisms in the same species has been generally supported by their 16S rRNA gene sequences (Maier and Murray, 1965; Nishino et al., 1998; Zemskaya et al., 2001, 2009; Dermott and Legner, 2002; Kojima et al., 2003, 2006; Høgslund et al., 2010; Nemoto et al., 2011, 2012; Salman et al., 2011). Nitrate accumulation by T. ingrica was reported in previous studies, although the intracellular concentrations were much lower than in relatives with large vacuoles (Zemskaya et al., 2001; Kojima et al., 2007; Høgslund et al., 2010).

In this study, the complete genome sequence of T. ingrica was reconstituted from metagenomic sequences, uncovering the metabolic and genetic characteristics of this organism. In addition, proteomic analysis was performed to investigate the physiology of Thioploca in lake sediments.

Materials and methods

Sampling

Thioploca samples were obtained at a site near the north shore of Lake Okotanpe, ∼200 m from the site of our previous study (Nemoto et al., 2011). Sediment samples were obtained with an Ekman–Birge grab sampler and were immediately sieved at the site with a 0.25-mm mesh in lake water. The materials retained on the mesh were transferred to lake water and kept at 4 °C in the dark until processing in the laboratory. Upon returning to the laboratory, Thioploca filaments were individually removed with forceps from the materials collected by sieving and were then repeatedly washed with filter-sterilized lake water. The washed filaments were stored at −30 °C for DNA extraction (the samples obtained in 2013) or at −80 °C for protein extraction (in 2012) or were immediately subjected to protein extraction (in 2011).

Sequencing and genome assembly

Genomic DNA was prepared from the washed Thioploca filaments using the Wizard Genomic DNA Purification Kit (Promega, Madison, WI, USA). A PCR-free paired-end library was constructed using the TruSeq DNA PCR-free Sample Prep Kit (Illumina, San Diego, CA, USA) and sequenced on the Illumina MiSeq platform (2 × 300 bp). After trimming low-quality bases (quality score <25) and adapter sequences, sequence assembly using varying numbers of reads (5 000 000, 4 000 000, 3 000 000, 2 000 000 or 1 000 000 reads) was performed using the Platanus assembler (http://platanus.bio.titech.ac.jp; Kajitani et al., 2014) with the default setting to determine the optimal number of input sequences for assembly. Based on the results of this analysis, we used 3 000 000 reads (1 500 000 sequence pairs) for assembly and obtained 38 scaffolds (>1 kb). Of these, 24 were larger than 5 kb. As 23 out of the 24 scaffolds showed similar ranges of GC content and sequence coverage, a single circular superscaffold was manually constructed from the 23 scaffolds by PCR. All gaps within and between scaffolds were closed by sequencing gap-spanning PCR products using an ABI3130xl DNA sequencer (Applied Biosystems, Foster City, CA, USA). Because one scaffold (28.75 kb in size) was derived from a plasmid-like circular DNA molecule, gaps in this scaffold were also closed in the same way. The BWA program (http://bio-bwa.sourceforge.net; Li and Durbin, 2009) was used for mapping analysis of the Illumina reads to the reconstituted genome. The reconstituted sequences have been deposited in the DDBJ/NCBI/EMBL databases (accession no. AP014633).

Annotation

Protein-coding sequences (CDSs) were predicted using MetaGeneAnnotator (Noguchi et al., 2008). The transfer RNA (tRNA) and rRNA genes were identified using tRNAScan-SE (Lowe and Eddy, 1997) and RNAmmer (Lagesen et al., 2007), respectively. The open reading frame extraction function in the In Silico Molecular Cloning Genomic Edition version 5.2.65 software (In Silico Biology, Inc., Yokohama, Japan) was used for additional gene prediction. Functional annotation of CDSs was performed on the basis of the results of BLASTP searches (Altschul et al., 1997) against the NCBI (National Center for Biotechnology Information) nonredundant database, the KEGG (Kyoto Encyclopedia of Genes and Genomes) database (Kanehisa et al., 2014) and the NCBI Clusters of Orthologous Groups (COG). CDSs were annotated as hypothetical protein-coding genes when the CDSs met any of the following three criteria in the top hit of the BLASTP analysis: (1) E-value >1e−8, (2) length coverage <60% against query sequence or (3) sequence identity <30%.

Genomic comparison and phylogenetic analyses

The genome sequences of Thioploca relatives were obtained from the following sites. Beggiatoa sp. SS (Candidatus Parabeggiatoa, henceforth ‘SS’) and Beggiatoa sp. PS (Candidatus Isobeggiatoa, ‘PS’) were taken from the NCBI Bacteria draft genome FTP site (ftp://ftp.ncbi.nlm.nih.gov/genomes/Bacteria_DRAFT); Beggiatoa sp. Orange Guaymas (Candidatus Maribeggiatoa, ‘Orange Guaymas’) and B. alba B18LD were taken from DOE-JGI IMG (https://img.jgi.doe.gov/w); and Thioploca araucae Tha-CCL (Candidatus Marithioploca, ‘Tha-CCL’) was taken from JCVI (https://moore.jcvi.org/moore/SingleOrganism.do?speciesTag=THACCL). Functional annotation of CDSs was performed as described above. To search for homologous sequences, all-against-all BLASTP searches (E-value<0.001) were performed for all of the CDSs from the six strains (T. ingrica and relatives). On the basis of the BLASTP results, the CDSs were clustered by applying OrthoMCL (Li et al., 2003) with default parameters.

Homologous gene clusters consisting of members from two or more strains were used to estimate the gene content phylogeny. The Euclidean distances among strains were calculated from the homolog content matrix among these strains, and complete-linkage hierarchical clustering was then conducted using the hclust function in the R software (http://www.r-project.org/).

A maximum-likelihood phylogenetic tree for T. ingrica and its relatives was built upon a concatenated protein sequence alignment of the universal single-copy genes (USCGs) and the gyrB gene. The genome of SS was excluded from the analysis because the quality of the draft genome was too low, whereas the sequence of Thioalkalivibrio nitratireducens DSM 14787 was included in the analysis. Among the 35 USCGs originally proposed (Raes et al., 2007), 4 are duplicated in the genomes of Tha-CCL and T. ingrica (rpsG and rpsS in Tha-CCL, rplM and rpsI in T. ingrica). Therefore, these four genes were excluded from the analysis. Because the valS and gyrB genes of Tha-CCL were both fragmented into two different contigs, they were manually merged. Each homolog cluster was separately aligned using MAFFT software version 7.130b (Katoh and Standley, 2013) with default parameters, and then the 32 alignments (31 USCGs and gyrB) were concatenated. Because some homologs were not found in the draft genomes of PS and Tha-CCL (gyrB and ychF in PS and pheS and rplK in Tha-CCL), the sequences of these genes were replaced with gaps in all positions. From the alignment of the concatenated sequences, a maximum likelihood tree was constructed using MEGA5 (Tamura et al., 2011). The Whelan and Goldman model (Whelan and Goldman, 2001) was selected based on the result of a likelihood ratio test. An initial tree for the heuristic search was obtained by applying the Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using the JTT (Jones–Thornton–Taylor) model and then selecting the topology with a superior log likelihood value. A discrete gamma distribution was used to model evolutionary rate differences among sites (five categories (+G, parameter =0.6269)). All positions with >50% alignment gaps were eliminated.

Protein analysis

Proteomic analyses were conducted using fresh or stored (at −80 °C for 120 days) samples obtained in 2011 and 2012, respectively. Protein extraction, sodium dodecyl sulfate polyacrylamide gel electrophoresis, in-gel trypsin digestion and nano-liquid chromatography tandem mass spectrometry were performed as previously described (Watanabe et al., 2012). The reconstituted chromosome sequence was used to generate the database for protein identification with the Mascot search program version 2.4.0 (MS/MS Ion Search; Matrix Science, Boston, MA, USA). Search parameters were set as follows: tryptic digest with a maximum of two missed cleavages; fixed modifications, carbamidomethyl cysteine; variable modifications, methionine oxidation; peptide masses, monoisotopic; positive charge (+1, +2, +3) of peptide; and mass tolerance of 1.2 Da for precursor ions and 0.8 Da for product ions. The false discovery rate was estimated using an automatic decoy search against a randomized database with a significance threshold of P<0.05. Protein detection was judged as positive when two or more different peptides were detected, and the exponentially modified protein abundance index was calculated as previously described (Ishihama et al., 2005). The normalized protein content value was calculated as a percentage of each exponentially modified protein abundance index in the summation of all identified proteins.

Results and discussion

Reconstitution of the T. ingrica genome

To reconstitute the T. ingrica genome, we used the washed T. ingrica filaments to generate >100 000 000 paired-end metagenomic sequences using the Illumina MiSeq platform. Because low-redundancy sequences derived from minor contaminating bacteria disturbed sequence assembly, 3 000 000 reads were used as input sequences, in accordance with the results of an optimization procedure (see Materials and methods). As a result, 24 scaffolds of >5 kb were constructed, all of which showed a similar range of GC content (from 39.6 to 44.3%, mostly ∼41%). All but one of the scaffolds showed a similar sequence coverage (from 104 × to 116 × ), suggesting that these scaffolds were derived from a single dominant species. Furthermore, all 16S rRNA gene reads belonging to these scaffolds were identical to the published 16S rRNA sequences of T. ingrica samples from Lake Ogawara and Lake Okotanpe (Kojima et al., 2006; Nemoto et al., 2011). We manually constructed a single circular super-scaffold from the 23 scaffolds with similar sequence coverage and, finally, reconstituted a complete circular chromosome by closing all of the gaps within and between the scaffolds. We searched the reconstituted chromosome for the USCGs (Raes et al., 2007) and identified a single homolog for each of the USCGs, except for rplM and rpsI that had been duplicated. This finding confirmed successful reconstitution of the T. ingrica genome.

A scaffold that was not included in the reconstituted chromosome showed a GC content (40.3%) similar to the chromosome, but its sequence coverage (381 × ) was much higher. Sequence analysis revealed that this scaffold encoded a putative replication protein, an integrase, a lytic transglycosidase and a response regulator (all other putative genes encoded hypothetical proteins), suggesting that the scaffold represented a plasmid-like DNA molecule. In fact, using PCR examination and gap filling, we reconstituted a single circular sequence (28 669 bp) from this scaffold. Although this plasmid-like element encoded an integrase, we did not find any evidence that it had been integrated into the chromosome of T. ingrica. The similarity in GC content suggested that the plasmid-like element may be associated with T. ingrica, but this idea needs to be experimentally confirmed.

In a mapping analysis of Illumina reads to the reconstituted sequences, 77.1% and 1.7% of the reads used for assembly mapped to the T. ingrica chromosome and the plasmid-like element, respectively. This finding indicated that the DNA from the washed T. ingrica filaments represented ca. 79% of the DNA preparation used for genome reconstitution.

General features of the T. ingrica genome

The general features of the T. ingrica chromosome are shown in Table 1 and Figure 1. Nucleotide position 1 of the chromosome was arbitrarily determined, as the replication origin could not be identified. The genome lacks an identifiable dnaA gene, and the GC skew fluctuates irregularly throughout the chromosome. The lack of dnaA may be a characteristic shared by nitrate-storing sulfur oxidizers, whereas dnaA was identified in B. alba B18LD. In all available draft genomes of nitrate-storing sulfur oxidizers, no dnaA gene was found by BLASTP or BLASTN searches when using the sequence of B. alba as a query.

Table 1 General features of the Thioploca ingrica genome
Full size table
Figure 1
figure 1

Circular view of the reconstituted T. ingrica chromosome. The outermost two circles show predicted CDSs on the plus and minus strands. The CDSs are color coded by COG categories (unclassified genes are shown in gray). The third and fourth circles indicate GC content (mean value =41%) and GC skew, respectively. Genes encoding rRNA (red) and tRNA (gray) are shown in the two innermost circles.

Full size image

Approximately 48% of the predicted genes of T. ingrica were classified into COG functional categories (Figure 1, Supplementary Table S1). No significant differences were identified among T. ingrica, Orange Guaymas and B. alba B18LD in their distributions of genes into COG functional categories (Supplementary Table S1).

Genomic comparison and phylogenetic analyses

Homolog distributions among the three strains (T. ingrica, Orange Guaymas and B. alba B18LD) are summarized in Supplementary Figure S1 and Supplementary Table S2. The number of genes shared by the three strains suggests that the number of core genes in the family Beggiatoaceae is <1500.

The phylogenetic trees constructed based on the sequences of USCGs or on gene content are shown in Figure 2. Both trees located T. ingrica at a position between its marine relatives and B. alba. This result is consistent with the phylogenetic position of T. ingrica in the family Beggiatoaceae that was inferred by 16S rRNA sequence analysis (Salman et al., 2011).

Figure 2
figure 2

Phylogenetic trees of T. ingrica and its relatives. (a) A tree constructed using the concatenated sequences of USCGs. The numbers at each node represent the percentage values from 1000 bootstrap resamplings. (b) A tree constructed using the gene contents of each strain.

Full size image

Proteomic analysis

Metagenomic analysis indicated that 79% of the DNA in the samples originated from T. ingrica. Considering that T. ingrica has a very large cell size and a normal genome size, T. ingrica cells may have a large protein/DNA ratio. If so, protein extracts obtained from the washed T. ingrica filaments may contain only a trace amount of proteins from microorganisms contaminating the filament sample. From two samples obtained in different years, we detected 864 and 826 proteins; among these, 563 were detected in both samples (Supplementary Tables S3 and S4). Among the 54 ribosomal proteins encoded in the genome, only 21 were detected in both samples, and 20 were not detected in either sample. This result indicated that only a limited portion of the T. ingrica proteome was detected and that a considerable portion of the proteins might have been missed by this analysis. However, the result also suggested that the proteins detected in this analysis were contained in the samples at relatively high concentrations. Therefore, the detected proteins are most likely T. ingrica proteins, especially those with high relative abundances. Protein abundance was evaluated based on exponentially modified protein abundance index, and the relative abundances of major proteins are shown in Figure 3. It should be noted that the samples subjected to the protein analysis were mixtures of cells positioned in different layers of sediment. Therefore, the detected proteins might derive from cells experiencing different environmental conditions. For instance, key enzymes for respiration with each of three electron acceptors (oxygen, nitrate and dimethyl sulfoxide) were all detected in both samples, but there is no way to know whether these proteins originated from the same cell or from cells at different positions in the redox gradient. Nevertheless, the detected proteins provide some insight into the ecophysiology of T. ingrica, as described below.

Figure 3
figure 3

Overviews of the proteomic analysis results from washed Thioploca filaments obtained in 2011 (a) and 2012 (b). The proteins are sorted according to their relative abundances in each sample, and the 150 most abundant proteins are shown. The data for the 500 most abundant proteins are outlined in the small nested boxes. Proteins involved in sulfur oxidation and nitrate respiration are indicated in red and green, respectively. The RuBisCO proteins are indicated by blue bars, and ribosomal proteins are shown in black for comparison.

Full size image

Sulfur oxidation

Although an array of circumstantial evidence has suggested sulfur oxidation by Thioploca, there has been no solid evidence to indicate that T. ingrica is truly a sulfur oxidizer. In the T. ingrica genome, genes involved in the proposed sulfur oxidation pathway were identified, and their gene products were detected in the protein analysis (Figures 3 and 4). The sulfur oxidation pathway of T. ingrica is basically the same as that proposed for its closest relatives, but it is notable for being the first case to confirm the presence of a full set of genes. Furthermore, the sequences of all of the genes are now fully available.

Figure 4
figure 4

Pathways for sulfur oxidation and dissimilatory nitrate reduction deduced from the T. ingrica genome. Proteins in red were detected in both samples subjected to protein analysis, and those in green were detected in one of the samples. Proteins encoded by the genome but not detected in the protein analysis are shown in blue. Gray dashed lines indicate alternative pathways mediated by the enzymes (shown in gray) that are not encoded by the T. ingrica genome but are found in its close relatives.

Full size image

As shown in Figures 3 and 4, it appears that sulfide oxidation by T. ingrica is mediated by the flavocytochrome c/sulfide dehydrogenase encoded by the fccAB (soxEF) genes (THII_1691-1692). The gene for another enzyme mediating sulfide oxidation, sqr, which encodes a sulfide:quinone oxidoreductase, was also identified in the T. ingrica genome (THII_0779).

It appears that T. ingrica oxidizes elemental sulfur to sulfite via the dissimilatory sulfite reductase (DSR) system. T. ingrica possesses a full set of genes for a DSR system, dsrABEFHCMKLJOP (THII_0611-0622), dsrNR (THII_3808-3809) and dsrS (THII_1535). In general, genes for the DSR system are mutually exclusive with the soxCD genes, with an exception in an environmental fosmid sequence (Lenk et al., 2012). Accordingly, genes corresponding to soxCD were not found in the T. ingrica genome.

For sulfite oxidation to sulfate, T. ingrica contains the aprBA genes, encoding an adenosine-5′-phosphosulfate reductase (THII_3105-3106); the sat gene, encoding a sulfate adenylyltransferase (THII_1057); and the hdrAACB genes (THII_2277-2280). The proteins encoded by the hdr genes are thought to be components of the Qmo complex that interacts with the adenosine-5′-phosphosulfate reductase (Dahl et al., 2013). Sulfite oxidation by these enzymes occurs in the cytoplasm, where sulfite is generated by DsrAB (Pott and Dahl, 1998). The T. ingrica genome also encodes an enzyme that mediates direct sulfite oxidation to sulfate (the sorAB genes; THII_2320-2321), but this sulfite oxidase usually works outside of the cytoplasm (Kappler et al., 2000; Kappler and Bailey, 2005).

The T. ingrica genome sequence further suggested that this microorganism is capable of thiosulfate oxidation by a system referred to as the Sox system (Friedrich et al., 2001). Of the three core components (SoxAX, SoxYZ and SoxB) that are thought to be indispensable for thiosulfate oxidation by this system (Hansen et al., 2006), two are encoded in the soxYZB gene cluster (THII_1577–1579) in T. ingrica genome. However, genes encoding SoxA and SoxX were not identified. These genes are usually located in a soxXYZAB gene cluster (Kappler and Maher, 2013). In T. ingrica, the function of the SoxAX complex might be substituted by a single protein, encoded by a gene located adjacent to the soxYZB cluster (THII_1580). The protein encoded by this gene is closely related to a protein designated as ‘SoxXA’ in the PS genome. Although these proteins are larger than conventional SoxA proteins, their sequences exhibit a partial similarity to SoxA.

The lifestyle of T. ingrica as a sulfur oxidizer, which was deduced from the genome sequence (Figure 4), is supported by the fact that most of the enzymes described above were among the most abundant proteins identified in the proteomic analysis of the samples directly collected from lake sediment (Figure 3).

Nitrate respiration

A full set of genes required for respiratory reduction of nitrate to N2 was identified in the T. ingrica genome (Figure 4). T. ingrica contains a narGHJI gene cluster (THII_1673–1676), encoding a membrane-bound nitrate reductase that mediates nitrate reduction to nitrite. In contrast to its closest relatives, the napAB genes that encode a periplasmic nitrate reductase were not detected in the T. ingrica genome. T. ingrica appears to generate N2 as the end-product of respiration, as it possesses nirS, encoding a nitrite reductase (THII_2875); norCBQ, encoding a nitric oxide reductase (THII_0334-0336); and nosZ, encoding a nitrous oxide reductase (THII_2884). The norD gene (THII_0363), which encodes a nitric oxide reductase activation protein, was also identified in the genome. Among these, the nosZ gene, which is directly responsible for the generation of N2, has never been found in the genomes of close relatives of T. ingrica. The nrfA gene, which encodes an enzyme for dissimilatory nitrite reduction to ammonium, was not found in T. ingrica.

The active reduction of nitrate to N2 by Thioploca in sediments has gained strong support from proteomic analysis; enzymes involved in all steps of the successive reduction (Figure 4) were detected, and many of these were major components in the samples (Figure 3). Recently, a large contribution by Candidatus Marithioploca to nitrogen loss in anoxic marine sediments was suggested by Prokopenko et al. (2013). This hypothesis was premised on nitrate reduction to ammonium by Candidatus Marithioploca and on the cooperative involvement of anaerobic ammonia-oxidizing bacteria. Interaction with ammonia oxidizers was also suggested for sulfur oxidizers inhabiting hydrothermal sediments, but the retention of nitrogen is assumed to take place in this case (Winkel et al., 2013). T. ingrica may contribute to nitrogen loss in freshwater lake environments in a different way from that of its marine counterparts.

Nitrogen assimilation

Nitrogen (N2) fixation by Beggiatoa species has been previously reported (Nelson et al., 1982), and genes involved in diazotrophy were identified in the Beggiatoa alba B18LD genome. In the T. ingrica genome, nifHDKT (THII_1806-1809), nifY (THII_1811), nifENX (THII_1813–1815), nifZW (THII_3085–3086), nifV (THII_3088), nifQ (THII_3125), nifB (THII_3130), nifM (THII_3738) and two copies of nifA (THII_2556 and THII_2572) were identified as genes putatively involved in N2 fixation and its regulation. Among the products of these nif genes, the regulatory protein NifA encoded by THII_2556 was detected in the proteomic analysis of both of the samples obtained in 2011 and 2012 (Supplementary Table S4). The others were not detected in the protein analysis, except for NifY that was detected only in the sample from 2011.

In the sediment of Lake Okotanpe, both nitrate and ammonium are available (Nemoto et al., 2011); thus, N2 fixation would not be advantageous because of the energetically high cost required for the process. The T. ingrica genome encodes both an ammonium transporter (THII_2433) and an assimilatory nitrate reductase, although these proteins were not detected in the protein analysis. T. ingrica may also use organic compounds as nitrogen sources, as several subunits for amino acid transporters were detected in the proteomic analysis. Taken together, multiple and flexible nitrogen assimilation pathways are deduced for T. ingrica.

Nitrate storage

The capacity to store a large amount of nitrate within cells is a unique property of Thioploca and closely related sulfur oxidizers. Based on genome sequence information for large marine sulfur oxidizers, a hypothetical model for nitrate accumulation in the vacuole was proposed in analogy to that in plants (Mußmann et al., 2007). The model includes two processes: the electrochemical gradient formed by proton pumping and the NO3−/H+ antiporter that depends on the gradient. When the model was proposed, three enzymes were taken into consideration as candidate enzymes responsible for the electrochemical gradient. All three enzymes, namely, a vacuolar-type ATPase, an H+-pyrophosphatase (THII_0754) and a Ca2+-translocating ATPase (THII_0386), are encoded in the T. ingrica genome. The latter two are also present in the B. alba genome that lacks nitrate-storing capacity. In the originally proposed model, the vacuolar-type ATPase and H+-pyrophosphatase were assumed to generate a proton motive force, but experiments using Candidatus Allobeggiatoa halophila revealed that these enzymes work in the reverse direction by consuming the proton gradient to generate adenosine triphosphate (ATP) and pyrophosphate (Beutler et al., 2012). The originally proposed model also predicted that the NO3-/H+ antiporter is another key protein directly responsible for nitrate accumulation. Whereas BgP0076 and BgP4800 in the genome of PS were assumed to encode NO3-/H+ antiporters, corresponding genes were not identified in T. ingrica. This finding indicates that the previously proposed model cannot be fully applicable to T. ingrica. Among the proteins mentioned above, an H+-pyrophophatase and a Ca2+-translocating ATPase were both detected in the proteomic analyses of filament samples collected in both 2011 and 2012 (Supplementary Table S4).

The molecular mechanism of nitrate accumulation in T. ingrica and relatives is still not fully understood, and further studies are necessary. The complete genome sequence information of T. ingrica obtained in this work will facilitate such studies.

Carbon metabolism

Many genes for the enzymes constituting the tricarboxylic acid cycle have been identified in other members of the family Beggiatoaceae, but some of the genes are missing from their draft genomes (MacGregor et al., 2013a). In the genome of T. ingrica, a full set of genes for tricarboxylic acid cycle enzymes was identified. Most of these enzymes were also detected in the protein analysis, confirming that the cycle was actually operating in T. ingrica cells. Similarly, genes encoding enzymes for the glycolytic pathway were also identified in the reconstituted genome, and the enzymes mediating all pathway steps were detected in the protein analysis (Supplementary Table S4).

In previous studies, acetate assimilation by T. ingrica was demonstrated by microautoradiography (Kojima et al., 2007; Høgslund et al., 2010). The gene for acetate uptake, actP, which encodes a cation/acetate symporter (THII_3816), was identified in the genome, as was the acs gene, encoding an acetyl-CoA synthetase (THII_1554) that converts acetate into acetyl-CoA. In addition to the tricarboxylic acid cycle, enzymes in the glyoxylate cycle (isocitrate lyase and malate synthase, encoded by aceA and aceB, respectively; THII_0533–0534) are also encoded by the genome; thus, acetyl-CoA can probably be used for both dissimilatory and assimilatory carbon metabolism.

Inorganic carbon fixation by T. ingrica has also been demonstrated in a previous study (Høgslund et al., 2010), and the key enzymatic activities of the reductive pentose phosphate cycle were detected in some strains of Beggiatoaceae (McHatton et al., 1996). In the T. ingrica genome, the rbcLS genes (THII_3311–3312), which encode the large and small chains of form I ribulose bisphosphate carboxylase (RuBisCO), were identified, in addition to eight other genes for the enzymes of Calvin–Benson–Bassham cycle. The gene for form II RuBisCO was not found in the genome.

As described above, the protein analysis suggested that T. ingrica cells gain energy from sulfur oxidation. The detection of other key enzymes, namely, the gene products of rbcLS, actP and acs, in both protein samples further supported the notion that acetate and bicarbonate are serving as carbon sources for T. ingrica in lake sediment. Methylotrophy in Beggiatoa alba has been demonstrated, and genes for key enzymes have been identified (Jewell et al., 2008), but such genes were not found in the T. ingrica genome.

Oxygen and dimethyl sulfoxide respiration

T. ingrica most likely has the capacity to use oxygen as a terminal electron acceptor for respiration, as it contains the ccoNOQP gene cluster that encodes a high-affinity cytochrome c bb3-oxidase (THII_1615–1617), and as the proteomic analysis detected the CcoN, CcoO and CcoP proteins (Supplementary Table S4). In the PS genome, the genes for the low-affinity cytochrome c aa3-oxidase were also found, but their counterparts were not identified in the T. ingrica genome. Notably, all of the subunits of dimethyl sulfoxide reductase (THII_3261–3263) were detected in both samples subjected to protein analysis, suggesting that dimethyl sulfoxide is also utilized by T. ingrica for respiration.

Phosphorus metabolism

Accumulation of phosphorus in the form of polyphosphate has been reported for some bacteria of the family Beggiatoaceae (Schulz and Schulz, 2005; Brock and Schulz-Vogt, 2011; Brock et al., 2012). In previous studies, T. ingrica samples were subjected to Toluidine blue staining to visualize intracellular polyphosphate granules, but only negative results were obtained (Høgslund et al., 2010; Nemoto et al., 2011). In the T. ingrica genome, the gene for polyphosphate kinase (ppk) was identified (THII_2967). Polyphosphate is a multifunctional molecule, and the presence of this gene may not yield the potential to form polyphosphate granules; however, the ppgK gene, encoding a polyphosphate glucokinase, was also identified in the genome (THII_0714). Previously, the phytase-encoding gene found in PS was discussed in relation to the acquisition of inorganic phosphate (Mußmann et al., 2007). This gene is also present in the T. ingrica (THII_2499) and B. alba genomes. In the protein analysis, polyphosphate kinase, polyphosphate glucokinase and phytase were all repeatedly detected.

Osmoregulation by the glycine betaine/proline transporter

Thioploca is unique in habitat preference among nitrate-storing sulfur oxidizers. In contrast to its relatives living in marine sediments, T. ingrica inhabits freshwater and brackish environments. To adapt to habitats of differing salinity, osmoregulation systems should play important roles. T. ingrica has a full set of genes for the ProU glycine betaine/proline transport system. The system consists of the ATP-binding protein ProV (THII_1898), the permease protein ProW (THII_1897) and the substrate-binding protein ProX (THII_1896). The ProU system is the transport system for various osmolytes, such as glycine betaine, proline and proline betaine, in Gram-negative bacteria (Sleator and Colin, 2001). Some bacteria harboring the ProU system can cope with increased environmental osmolarity by accumulating glycine betaine (Lucht and Bremer, 1994). A full set of genes for this system is also present in the genome of Orange Guaymas, obtained from marine sediment under the influence of hydrothermal fluid in the Guaymas Basin (MacGregor et al., 2013a). It was suggested that the habitat of Orange Guaymas is exposed to large fluctuation of temperature (McKay et al., 2012), presumably because of changes in supply of the hot water. The osmoregulation systems may also be effective to adapt to frequent changes in the composition of surrounding water, brought about by fluctuating mixing ratio of hydrothermal fluid.

Conclusion

The complete genome sequence of T. ingrica was successfully reconstituted from metagenomic sequences, thus representing the first complete genome sequence of a nitrate-storing sulfur oxidizer. We found that T. ingrica possesses all of the genes required for complete denitrification. The presence of the denitrification system in T. ingrica was further confirmed by protein analysis, as were several other physiologically important functions, such as sulfur oxidation and inorganic carbon fixation. The complete genome sequence of T. ingrica will be a valuable genetic basis for a wide range of future studies on nitrate-storing sulfur oxidizers that are important constituents of aquatic ecosystems.