Complete genome sequence of ‘Halanaeroarchaeum sulfurireducens’ M27-SA2, a sulfur-reducing and acetate-oxidizing haloarchaeon from the deep-sea hypersaline anoxic lake Medee

Strain M27-SA2 was isolated from the deep-sea salt-saturated anoxic lake Medee, which represents one of the most hostile extreme environments on our planet. On the basis of physiological studies and phylogenetic positioning this extremely halophilic euryarchaeon belongs to a novel genus ‘Halanaeroarchaeum’ within the family Halobacteriaceae. All members of this genus cultivated so far are strict anaerobes using acetate as the sole carbon and energy source and elemental sulfur as electron acceptor. Here we report the complete genome sequence of the strain M27-SA2 which is composed of a 2,129,244-bp chromosome and a 124,256-bp plasmid. This is the second complete genome sequence within the genus Halanaeroarchaeum. We demonstrate that genome of ‘Halanaeroarchaeum sulfurireducens’ M27-SA2 harbors complete metabolic pathways for acetate and sulfur catabolism and for de novo biosynthesis of 19 amino acids. The genomic analysis also reveals that ‘Halanaeroarchaeum sulfurireducens’ M27-SA2 harbors two prophage loci and one CRISPR locus, highly similar to that of Kulunda Steppe (Altai, Russia) isolate ‘H. sulfurireducens’ HSR2T. The discovery of sulfur-respiring acetate-utilizing haloarchaeon in deep-sea hypersaline anoxic lakes has certain significance for understanding the biogeochemical functioning of these harsh ecosystems, which are incompatible with life for common organisms. Moreover, isolations of Halanaeroarchaeum members from geographically distant salt-saturated sites of different origin suggest a high degree of evolutionary success in their adaptation to this type of extreme biotopes around the world.


Introduction
'Halanaeroarchaeum sulfurireducens' M27-SA2 was isolated from the deep-sea hypersaline anoxic lake Medee (Ionian Sea, Eastern Mediterranean, water depth 3105 m). Together with other five strains, previously isolated from shallow and terrestrial athalassic hypersaline sites of Russia and Spain [1], this haloarchaeon possesses maximum of 91-93 % 16S rDNA sequence similarity to the nearest cultured members of Halobacteriaceae. All Halanaeroarchaeum isolates represent a novel type of strictly anaerobic haloarchaea that grow best in NaCl brines close to saturation and use acetate as sole electron donor and carbon source with elemental sulfur as the only electron acceptor. Little is known about anaerobic sulfur metabolism at saturated salt conditions [2]. There is some evidence suggesting that bacterial sulfate reduction is possible under salt-saturated conditions [3], but sulfur respiration under such conditions has so far been very poorly investigated, except for the 'Halanaeroarchaeum' strain HSR2 T and two extremely haloalkaliphilic bacteria of the order Halanaerobiales, Halarsenatibacter silvermanii and Natroniella sulfidigena [1,4,5]. Following the fact, that we were able to isolate these haloarchaea from various geographically and physico-chemically distinct hypersaline sites [1], the sulfidogenic anaerobic oxidation of acetate is likely a common feature in anoxic salt-saturated habitats, overlooked so far.
In this paper we describe the genome properties of 'Halanaeroarchaeum sulfurireducens' M27-SA2 providing details on carbon and sulfur metabolism, on clustered regularly interspaced short palindromic repeats (CRISPR) and on presence of prophage loci and genomic islands.

Organism information
Classification and features 'Halanaeroarchaeum sulfurireducens' M27-SA2 has typical haloarchaeal pleomorphic cell morphology, ranging from flattened rods to coccoid or irregular forms (Fig. 1). The pleomorphism of M27-SA2 strain increased with the cultivation time, as is often observed for members of the family Halobacteriaceae. The 16S rRNA gene of M27-SA2 exhibited 99.58 % sequence similarity with H. sulfurireducens strain HSR2 T and 97-98 % sequence similarity with clones of uncultured haloarchaea obtained from  Tree was inferred from a 16S rRNA gene sequence alignment with PAUP*4.b10 [59] using a LogDet/paralinear distance method. Support for nodes in the tree corresponds to bootstrap values for 1000 pseudo-replicates. Only bootstrap values greater than 75 % are displayed as solid circles. The tree has been arbitrarily rooted on sequence of Natronomonas pharaonis (D87971) and Halomarina oriensis (AB519798). The 16S rRNA gene sequence of Methanohalophilus halophilus (FN870068) was used as the outgroup. The scale bars represent a 5 % nucleotide sequence divergence hypersaline anoxic soils, brines and sediments around the world [1] (Fig. 2).
Together with other Halanaeroarchaeum isolates, M27-SA2 represents the only type of obligate and strictly anaerobic haloarchaea. Most of the known cultivated extremely halophilic euryarchaeota are aerobic heterotrophs except for a few examples of facultatively anaerobic species capable of growth by fermentation [6], denitrification [7], fumarate, DMSO and TMAO reduction [8,9]. Strain M27-SA2 was isolated from the brine (320 g l −1 of total salt content) of deep-sea Lake Medee (Eastern Mediterranean) collected in September 2012 at depth of 3,010 m. The collected Medee brine was transferred into the serum vials (120 ml) prefilled with the artificial brine to attain 230 g l −1 of final salinity. The artificial brine has the following composition: NaCl 200 g l −1 ; KH 2 PO 4 0.33 g l −1 ; yeast extract 50 mg l −1 ; Na 2 S 0.5 g l −1 ; acetate 15 mmol l −1 ; S°2.5 g l −1 , 10 ml l −1 trace elements solution (DSMZ medium 320); and 10 ml l −1 vitamin solution (DSMZ medium 141); pH values were adjusted to 6.7 corresponding to in situ values of the brine. Similar to all known 'Halanaeroarchaeum' isolates, strain M27-SA2 grew between pH 6.7 and 8.0 (with the optimum at pH 7.2-7.5), 3.0 and 5.0 M of NaCl with the optimum growth observed at total salinity of 250 g l −1 . Notwithstanding the isolation from the environment with permanent temperature of 15°C [10], strain M27-SA2 has the optimal temperature of growth at 40°C (Table 1). The isolate has a very limited metabolic profile restricted to acetate and pyruvate as the only available sources of carbon and energy and elemental sulfur as a electron acceptor [1]. Nevertheless, yeast extract should be added to the medium in concentrations of , not directly observed for the living, isolated sample, but based on a generally accepted property for the species, or anecdotal evidence). These evidence codes are from the Gene Ontology project [58]. If the evidence is IDA, then the property was directly observed for a live isolate by one of the authors or an expert mentioned in the acknowledgements

Genome sequencing information
Genome project history 'Halanaeroarchaeum sulfurireducens' strain M27-SA2 was selected for sequencing on the basis of its phylogenetic positions, its particular feature as a novel strictly anaerobic haloarchaeon from the deep-sea anoxic salt-saturated lake and the interest of studying this unique mechanism of anaerobic respiration, recently discovered for the first time among entire Archaea domain [1]. The respective genome project is deposited on the NCBI BioProject PRJNA284332 and the complete genome sequence in GenBank CP011564 and CP011565 (chromosome and plasmid) is available since 30 of September 2015. The main project information is summarized in Table 2.

Growth conditions and genomic DNA preparation
Strain M27-SA2 was routinely grown anaerobically to early stationary phase in 120-ml flasks using a protocol described elsewhere [1]. Genomic DNA was isolated from the cell paste according to extraction method from Urakawa et al. [11]. DNA quality and quantity were determined with a Nanodrop spectrometer (Thermo Scientific, Wilmington, USA).

Genome sequencing and assembly
The  . From outside to center: genes on forward strand (COG color-coded), genes on reverse strand (COG color-coded), RNA genes and other (only for chromosome: tRNAs green, rRNAs red, CRISPR azure, Prophages dark blue), GC content, GC skew [18,60] of 124,256 bp. Together, all matching sequences provided 634× coverage for chromosome and 691× for plasmid.

Genome annotation
Protein-coding genes were predicted by Glimmer 3.02 [13]; rRNA genes by RNAmmer 1.2 Server online tool [14]; tRNA-coding sequences by tRNAscan-SE 1.21 online tool [15]; while operon prediction was performed by the FgenesB online tool [16]. Some of structural and functional annotations were performed as it was described by Toshchakov et al. [17]. For each predicted gene similarity search was performed by Geneious 7.1 BLAST embedded tool against public amino acid sequence databases (nr, SwissProt), conserved domains families databases (Pfam, COG). Finally, annotations were manually curated using the Artemis 16.0 program [18] and refined for each gene with NCBI blastx against nr database (only for control) [19].

Genome properties
The genome of strain M27-SA2 comprises two circular replicons: a 2,129,244-bp chromosome and a 124,256-bp plasmid ( Fig. 3 and Table 3). The chromosome has a 63.19 % GC content. Of the 2,200 predicted genes     Table 4). The majority of the protein-coding genes (60.72 %) were assigned with a putative function, while remaining sequences were annotated as hypothetical proteins. An assignment of genes by COGs functional categories is presented in Table 5. The plasmid has 55.39 % GC content and contains 119 protein-coding genes. Only 24 of them (20.16 %) were assigned to COGs (Table 6 and Table 7) while the remaining genes were annotated as hypothetical proteins.

Insights from the genome sequence
Genome comparisons: M27-SA2 vs HSR2 T As a demonstration of their extreme similarity, the genomes of 'Halanaeroarchaeum sulfurireducens' M27-SA2 and 'Halanaeroarchaeum sulfurireducens' HSR2 T , were compared with three different methods: the Artemis Comparison Tool program [20], the LAST web service [21], and the Multiple Genome Alignment system software (Mauve) [22]. Additionally, Double ACT web service was used to generate the required ACT comparison file. The results of these tools are shown in Fig. 4. It follows that the chromosome of M27-SA2 has high average nucleotide identities, over 97 %, to the corresponding replicon of HSR2 T , whereas the plasmids of both isolates possess even higher average nucleotide identities values, over 99 %. The only difference with respect to the HSR2 T chromosome (the gap visible for all methods used) was due to the   . Similarly to what was found in HSR2 T , the genome analysis of M27-SA2 identified two blocks of genes responsible for the oxidation of acetate to CO 2 with elemental sulfur as the electron acceptor. The acetate oxidation pathway occurred by means of an ATPdependent acetyl-CoA synthase and TCA cycle, while sulfur dissimilation could be accomplished by four different operons, coding for molybdopterin oxidoreductases (HLASA_0051-0056; HLASA_0525-0529; HLASA_0688-0694; HLASA_1275-1271). Notwithstanding, these two strains were isolated apparently from very different and geographically very distant habitats, e.g. top 10 cm-layer sediments of Kulunda Steppe (Central Russia) terrestrial hypersaline lakes (HSR2 T ), and from hypersaline brine at 3105 m depth of Lake Medee (Eastern Mediterranean) M27-SA2, we failed to find any genetic determinants reflecting such significant difference in environmental settings of these two habitats.

Amino acid biosynthesis pathways
The addition of yeast extract in amounts 10-20 mg l −1 is necessary for growth of strain M27-SA2, which is likely indicating that some amino acids are not synthesized or their biosynthesis could be arduous and they should be imported from the environment. We reconstructed the amino acid biosynthetic pathways of M27-SA2 using both the SEED subsystem [23] and KEGG orthology [24] assignments (Fig. 5). Similarly to what found on strain HSR2 T (genomes were nearly identical) this analysis indicated that the genome of strain M27-SA2 harbors all the genes required for complete synthesis of at least 19 amino acids. Seven of the eight genes involved in conversion of aspartate to lysine via tetrahydrodipicolinate, which should involve succinylated intermediates, were found. The gene dapC encoding N-succinyldiaminopimelateaminotransferase was not identified in M27-SA2 genome.
The pathway for the biosynthesis of isoleucine, valine, and leucine from pyruvate seems to be fully present and all genes were detected in the analyzed genome. Interestingly, a branched-chain amino acid transport system related to the permease protein Liv (HLASA_0776-0780), was detected in the M27-SA2 genome, suggesting that nonsecreted amino acids could be imported via various transporters.

CRISPR analysis
Pilercr v1.02 [25] with default parameters was used to identify Clustered Regularly Interspaced Short Palindromic Repeats array in M27-SA2 genome. The CRISPRfinder tool was used for CRISPR search as a control [26]. Cas genes were identified with the NCBI Blastn online tool [19]. Spacer sequences detected in CRISPR array were analyzed in order to find similarities with plasmids, phages or haloarchaeal chromosomes. Spacer sequences were blasted against nt, env_nt and wgs databases using NCBI BLAST+ blastn tool installed into web-based Galaxy platform. Additionally, spacer sequences were blasted against a local database made of several hypersaline metagenomes, including those from the anoxic hypersaline lakes Kryos (M. Yakimov, unpublished results) and Thetis [27], the hypersaline Australian lake Tyrrell [28], and solar salterns of Santa Pola [29] and South Bay Salt [30]. Spacers with ≤7 SNPs (80 % match or 30/37 nucleotides) were considered as positive hits. Obtained matches of at least 100 bp-long were compared to the nr and nt NCBI databases using NCBI Blast + blastx and blastn, respectively. Most sequenced so far archaeal genomes contain at least one CRISPR-Cas system [26]. DNA fragment of 13.1 kbp that included CRISPR array and associated cas genes was detected in the M27-SA2 genome. The CRISPR array found in M27-SA2 was practically identical to that found in 'Halanaeroarchaeum sulfurireducens' HSR2 T , contained the same 30-bp direct Similarly, eight cas vgenes were detected in vicinity of the CRISPR array: cas6, cas8b/ csh1, cas7/csh2, cas5, cas3, cas4, cas1, cas2 (Fig. 6). All cas genes had high level of similarity to cas genes of Halorhabdus tiamatea and Haloarcula argentinensis (with e-value ranged from 1e −37 to 0.0), and thus were highly conserved between closely relative haloarchaeal genera. According to the current classification, this system was affiliated to I-B subtype or CASS7 [31].
When we compared 57 spacers extracted from 'Halanaeroarchaeum sulfurireducens' M27-SA2 to nt, env_nt and wgs Genbank databases, no matches were obtained. We blasted the spacers against metagenomic sequences of samples obtained from aforementioned hypersaline lake environments. This analysis identified six spacers that matched the metagenomic sequences (protospacers) obtained from the salt-saturated lakes  [38]; Hmuc, Haloferax mucosum contig AOLN01000011 [42]; Hvol, Haloferax volcanii chrom1 [42]  Kryos and Tyrrel (Table 8). Spacer #37 matched to DNA fragment from lake Kryos metagenome that contained CRISPR array with repeat type of H. marismortui ATCC 43049 plasmid pNG400. These results suggest the occurrence of bacteria or/and viruses transfer between hypersaline biotopes. The spacer #7 matched to metagenomic sequences from viral fraction of lake Tyrrell. When those reads were compared to nt database, hits to viral contigs eHP-1, eHP-4, eHP-15 and eHP-19 from Santa-Pola solar saltern [29], and to the sequence of uncultured virus clone from Tunisian solar salterns [32] were found. Four other spacers (#24, #34, #52 and #54) matched to sequences from viral fraction of metagenome from lake Tyrrell. No homology in the GenBank was found. These results demonstrate that 'Halanaeroarchaeum sulfurireducens' M27-SA2 CRISPR spacers, likewise HSR2 T CRISPR spacers, target mobile genetic elements that have been identified in distant solar and deep-sea hypersaline lakes, suggesting that this haloarchaeon could have been adapted to yet unexplored haloviruses. The presence of a short protospacer adjacent motif located upstream of the protospacer is required for immunity of type I CRISPR-Cas system [33]. The PAM sequence varies in different CRISPR subsystems. It has been shown that archaeal I-B CRISPR-Cas system has several different PAM sequences upstream of the protospacer: TTC, ACT, TAA, TAT, TAG and CAC for Haloferax volcani [34], TTC for Haloquadratum walsbyi [34], TTT, TTC, TTG, and CCC for H. hispanica [35]. Our analysis detected several variants of trinucleotide sequence upstream of different protospacers (Table 8). Interestingly, most of them had TTC/TTT sequence, which are highly conserved among archaeal I-B subsystems reported so far [36]. This fact suggests that CRISPR-Cas system is likely active in 'Halanaeroarchaeum sulfurireducens' M27-SA2.

Phage-like elements
It has been estimated that 60-70 % of prokaryotic genomes deposited to GenBank contain prophage sequences [37]. We analyzed the genome of 'Halanaeroarchaeum sulfurireducens' M27-SA2 in terms of presence of prophages. Apparently, there were two fragments of 28.5 kbp (prophage 1) and 49.5 kbp (prophage 2) that contained clusters of genes of viral origin. Manual annotation of prophage 1 gave best matches to putative ORFs of haloarchaeal pleomorphic phages HRPV1, HRPV2, HRPV3, HRPV6, HHPV1 and HHPV2 of Haloferax lucentense, pHK2 plasmid, and prophages in the genomes of Halomicrobium mukohataei and Haloferax volcanii. The genome of prophage 1, presents in both M27-SA2 and HSR2 T strains, was located near the tRNA gene, a common site for prophage insertions [38], and contained a putative XerC/D integrase/recombinase gene on the opposite to tRNA gene flank. Presence of an integrase and a tRNA insertion site could be interpreted as indicator of an active prophage. The alignment of the genome of prophage 1 and its closest relatives shows that prophage 1 contains ORFs homologous to the whole set of core genes in the genomes of lytic pleoviruses HRPV-1, HRPV-2, HRPV-3 and HHPV-1 ( Fig. 7 and Tables 9 and 10). However, we have found a single CRISPR array and eight associated cas genes of I-B subtype inside the genome of prophage 1. The insert is~14 kbp long and occupies half of the prophage 1 genome (~28.5 kbp). Earlier, the entire CRISPR-Cas system including cas genes of I-F type and a CRISPR array was found in the genome of myovirus ICP1 of Vibrio cholera serogroup O1 [39]. Most of the spacers from ICP1 CRISPR array targeted PICI-like element from the genome of V. cholera, an excised circular DNA fragment, which becomes induced and interferes with phage reproduction during infection. Therefore, bacteriophages can acquire CRISPR-Cas systems from the host genome or from the environment through natural transformation of the host cell and use it to abolish anti-phage cellular mechanisms. Two opposite suggestions could be made based on the presence of the CRISPR-Cas system in the prophage 1. On one hand, the insert of DNA fragment containing CRISPR-Cas system equal to the length of the "viral" fragment of the prophage 1 as well as presence of a number of small ORFs of host origin interspaced by long non-coding regions with numerous stop codons on the right wings of its genome (see Fig. 7) would compromise release of viral particles. On the other hand, the pleomorphic nature of the prophage 1 could allow of formation viable particles with extended genomes, as viral packaging and release are driven by a budding vesicle from the plasma membrane, and the size of the genome therefore dictates the size of the vesicle [40]. According to this logic, CRISPR-Cas system of prophage 1 would not be active during lysogenic stage of infection as the expression of most of the lytic genes of a prophage are usually shut off [41], but could become active during lytic stage of infection and be used to overcome bacterial defense. Additional experiments can be designed and performed to distinguish between these two scenarios.
Another prophage 2 is~43.5 kbp in length and containes 49 ORFs. One of the ORFs encodes a putative tape tail measure protein, which is a key feature of Siphoviridae morphological family. The closest homologues of ORFs of prophage 2 were related to several haloviruses: Bj1 (siphovirus that infects Halorubrum), phiCh1 (myovirus of Natrialba), HHTV1 (siphovirus of H. hispanica), HGTV1 (myovirus of Halogranum sp.), prophages in the Haloferax mucosum and Haloferax elongans genomes [42] and to environmental viral contigs (environmental halophages eHP-2, eHP-14, eHP-32, eHP-34, eHP-36). Interestingly, prophage 2 encodes an adenine-specific DNA methylase, which could be responsible for protection from host restriction endonucleases through methylation of the phage genome [43]. Prophage 2 is located next to a tRNA integration site and has a XerC/D integrase/recombinase gene on the opposite side of the genome: both features are always present as a part of a viable prophage. Archaeal tailed viral genomes integrated into cellular chromosomes have been identified before [44]. The fact that the closest host homologue in the database (H. sulfurireducens strain HSR2 T , accession number CP008874) doesn't contain a sequence of the prophage 2 suggests recent acquisition of this prophage. Viruses of Euryarchaeota encompass different morphologies, including spindle-shaped viruses (Salteproviridae), pleomorphic viruses (Pleolipoviridae), head-tailed viruses (Caudovirales), spherical viruses (Plasmaviridae), and unclassified icosahedral dsDNA viruses with an inner lipid layer [45]. Therefore, the presence of two different prophages in the same genome of H. sulfurireducens M27-SA2 suggests that archaea in hypersaline environemnts located at the bottom of the Mediterranean Sea are exposed to a constant threat of phage predation.

Genomic islands
Horizontally transferred genomic islands (GIs) in 'Halanaeroarchaeum sulfurireducens' M27-SA2 genome were determined by the SeqWord Gene Island Sniffer program [46] and by the GOHTAM online tool [47]. The results of both GI identification methods are shown in Fig. 8. Six putative GIs characterized by alternative oligonucleotide usage patterns were detected by SWGIS, while GOHTAM search returned many short region (overall 52, see Fig. 8b) in addition to the six previously identified, probably due to a lower default sensitivity threshold of the latter method. Predicted GIs harbored mainly hypothetical proteins, transposases, glycosyltransferases (many in the third GI), and other enzyme-coding genes (transport and metabolism), a tRNA in the first GI, while the sixth GI covers the prophage