Introduction

Photobacterium are Gram-negative bacteria and represent one of the major genera of the family Vibrionaceae. This genus consists of about two-dozen validated species (including subspecies) and more than 450 undefined species based on the 16S rRNA gene sequence analysis (see NCBI taxonomy website). They are facultative aerobes and derive energy from oxido-reduction of organic compounds. Photobacterium species are widespread in coastal, open-ocean and deep-sea environments, and occur in free-living form in seawater and sediments, or associated with marine animals. They function as decomposers of dead fish, pathogens for marine animals and human, or symbionts of light organs of fish and squid in the marine ecosystems. Seven Photobacterium species are luminous (Ast and Dunlap 2005). Three species, P. kishitanii, P. leiognathi and P. mandapamensis form bioluminescent symbioses with marine animals (Urbanczyk et al. 2011). Genetic attributes underpinning the eco-physiological adaptation remain poorly understood. Twelve Photobacterium genomes are available till now (Table 1), including piezophilic and piezosensitive P. profundum strains, which have been studied thoroughly for understanding the mechanism of bacteria adaptation to the high hydrostatic pressure. Eloe et al. have compared the two P. profundum genomes and found two sets of flagellar genes coding for lateral and polar flagella in the piezophilic strain SS9 but only the polar flagellar system in the piezosensitive strain 3TCK (Eloe et al. 2008). Similar dual flagellar systems have been reported for the deep-sea P. profundum DSJ4 (Campanaro et al. 2005). Moreover, synthesis of several terminal oxidases of anaerobic respiration in the piezophilic strain SS9 is up-regulated at high hydrostatic pressure (Campanaro et al. 2005; El-Hajj et al. 2010; Vezzi et al. 2005). The occurrence of dual flagellar systems and up-regulation of enzymes involved in energy metabolism have been proposed to be developed by deep-sea microbes to adapt to the high-pressure environments.

Table 1 Overview of Photobacterium genomes
Full size table

The Photobacterium phosphoreum strain ANT-2200 (hereafter called ANT-2200) was isolated at 2200 m depth from the Mediterranean Sea and capable of emitting bioluminescence (AlAli et al. 2010; Martini et al. 2013). Incubation at 22 MPa increases both its growth rate and light emission compared to cultures at atmosphere condition, indicating a moderately piezophilic feature (AlAli et al. 2010; Martini et al. 2013). In order to understand the evolutionary strategy developed by this strain to adapt to the deep marine environment, we sequenced its genome and carried out physiological analyses. Metabolism pathway analysis revealed a versatile growth capacity. The strain ANT-2200 lives in a place with saturated dissolved oxygen (Tamburini et al. 2013). Consistently, we identified 10 genes required for the aerobic respiration with oxygen as terminal electron accepter. Interestingly we also found genes coding for the typical anaerobic respiration enzymes and confirmed their functions by growth analysis under various conditions. Moreover, we identified four gene clusters encoding Trimethylamine N-oxide (TMAO) reductases. TMAO is an efficient organic osmolyte that counteracts the effects of pressure on proteins (Yancey et al. 1982). TMAO content in tissues of deep-sea animals increases with depth (Yancey et al. 2002) and its accumulation increases internal osmolarity of fish. The up-limit of tissue TMAO concentration seems to constrain the marine animals from inhabiting the deepest ocean environment (Yancey et al. 2014). TMAO also serves as an electron acceptor in bacteria for generating energy via respiration. It is plausible that TMAO released from dead fish increases its instant availability in the deep-sea habitats. We observed that addition of TMAO significantly improved the anaerobic growth of ANT-2200 and that its TMAO reductase activity is enhanced by high hydrostatic pressure, implying an important role of this enzyme in adaptation of bacteria to deep-sea habitats.

Materials and methods

Growth media and cultures

P. phosphoreum strain ANT-2200 was grown in YPG rich medium (Martini et al. 2013) or ANT-minimal medium that consists of artificial seawater supplemented with vitamins and trace elements [as described in (Frankel et al. 1997)], and HEPES (0.3 % final w/v concentration), NH4Cl (0.2 %), K2HPO4 (1.86 %), Na2MoO4 (0.0024 %), and Na2SeO3 (0.0017 %). When indicated, sodium formate (0.2 %, final concentration), sodium nitrate (0.1 %), sodium fumarate (1.5 %), TMAO (0.1 %), glucose (0.2 %) or maltose (0.2 %) was added. The cultures of this mesophile, deep-sea strain were incubated anaerobically at room temperature (22–25 °C) in dark.

TMAO reductase analysis

The periplasmic fractions were prepared by osmotic-shock treatment and TMAO reductase activity was measured by enzymatic assay using benzyl viologen as an electron donor in anaerobic cuvette or visualized by activity staining using methyl viologen as an electron donor after resolving periplasmic proteins on native polyacrylamide gels as previously reported (Santini et al. 1998). The bands exhibiting activity were excised from the gels and their protein content was identified by nanoLC-ESI–MS/MS spectrometry analysis using the Ultimate 3000/LTQ Orbitrap XL instrument as described by Christie-Oleza et al. (Christie-Oleza et al. 2012).

Genome sequencing and analysis

The P. phosphoreum ANT-2200 whole genome was sequenced at Genoscope as previously described (Zhang et al. 2014a). The annotation was performed using the Microscope platform (https://www.genoscope.cns.fr/agc/microscope/home/index.php) (Vallenet et al. 2013) and promoter prediction was performed using Virtual Footprint version 3.0 (http://www.prodoric.de/vfp/vfp_promoter.php). Ortholog analysis has been carried out at MBGD (http://mbgd.genome.ad.jp/). The updated genome sequence has been deposited at EMBL with the accession numbers WGS: CCAR020000001-CCAR020000010.

Results and discussion

Genome overview

Using optical mapping we have updated and reassembled the draft genome of P. phosphoreum ANT-2200. The genome consists of chromosome I (3,293,680 base pairs), chromosome II (1,661,466 base pairs), a plasmid (113,194 base pairs) and an un-positioned scaffold (32,257 base pairs), with a total size of 5,100,597 base pairs (Fig. 1; Table 2). The parAB genes (PPBDW_v2_p0098/97) encoding plasmid partition proteins were found on the plasmid. The chromosome II contains a large region of 90,335 bps with predicted CDSs only on the clockwise-transcription strand (Fig. 1, at 9 o’clock direction). In this region, several exported proteins of unknown function are predicted, which are conserved among Photobacterium genomes. The two-chromosome composition has been found in P. profundum SS9, P. mandapamensis strain svers. 1.1, P. leiognathi, P. angustum and P. gaetbulicola Gung47 (Okada et al. 2005a; Urbanczyk et al. 2011; Vezzi et al. 2005) (Table 1). We have compared the chromosomes of ANT-2200 with those of P. profundum piezophilic strain SS9 and piezosensitive strain 3TCK using Blastn program. As shown in the Dot Plot View, the alignment of two chromosomes I from ANT-2200 and SS9 showed a main diagonal line with two inverse sections as an opposite diagonal line (Supplementary information, Figure S1, A1). The homologous regions cover 62 % of ANT-2200 chromosome I with 90 % sequence identity. The chromosomes II display only sporadic short homologous fragments (Figure S1, A2). When comparing with the draft genome of P. profundum piezosensitive strain 3TCK, chromosome I of ANT-2200 and the region from 2-Mbp to the end of the 3TCK genome shared sequence identity of 90 % (Figure S1, B1), while limited alignment has been observed between the ANT-2200 chromosome II and the first 2-Mbp region of the 3TCK draft genome (Figure S1, B2). Similar results were obtained between the chromosomes of P. profundum piezophilic strain SS9 and piezosensitive strain 3TCK (Figure S1, C1 and C2). Thus, it is possible that 3TCK genome also consists of two chromosomes. The chromosomes I and II of ANT-2200 possess 8.25 and 12.9 % coding DNA sequence (CDS) showing no homology to any previously reported sequences. Taken together, the results indicate that the chromosome I of ANT-2200 is conserved while the chromosome II is more diverse compared to those from other Photobacterium spp., which is consistent with the report that the chromosome I is more stable and contains the most established genes in P. profundum SS9 (Vezzi et al. 2005) and Vibrionaceae in general (Dryselius et al. 2007; Okada et al. 2005b).

Fig. 1
figure 1

Overview of P. phosphoreum ANT-2200 genome. Two chromosomes, the plasmid and the un-positioned scaffold are presented in proportion with to their sizes. Circles display (from the outside): (1) G + C percent deviation (GC window—mean GC) in a 1000-bp window. (2) Predicted CDSs transcribed in the clockwise direction. (3) Predicted CDSs transcribed in the counterclockwise direction. Genes displayed in (2) and (3) are color-coded according different categories: red and blue manually curated gene functions; purple primary/Automatic annotations. (4) GC skew (G + C/G-C) in a 1000-bp window. (5) rRNA (blue), tRNA (green), misc_RNA (orange), Transposable elements (pink) and pseudogenes (grey)

Full size image
Table 2 Summary of genomic features of Photobacterium phospheorus ANT-2200
Full size table

Among the thirteen available genomes of Photobacterium spp., ANT-2200 has an average genomic size but the lowest G + C % content (Table 1). Currently only the genomes of P. profundum SS9 and P. gaetbulicola Gung47 have been completely sequenced and assembled at chromosome level, whereas those of ANT-2200 and other Photobacterium spp. are at either scaffold or contig levels with gaps (Table 2). Therefore, caution must be taken in functional genomic comparisons and analyses. The ANT-2200 genome encodes at least 169 tRNA genes that cover all the 20 common amino acids in addition to 1 SeC(p) tRNA for Selenocysteine (Sec, U, or Se-Cys). Twenty-four rRNA genes have been identified, all located in chromosome I, including 11 encoding 5S rRNA, 7 encoding 23S rRNA and 6 encoding 16S rRNA, while at least three more 16S rRNA genes are expected as 3 incomplete sequenced rDNA clusters are present at the end of contigs. The 16S rRNA genes of ANT-2200 shared 99.93 % identity, which is in contrast to the fifteen 16S rRNA genes from SS9 displaying 5.13 % sequence divergence (Vezzi et al. 2005).

Ortholog analysis has been performed between the genome of P. phosphoreum ANT-2200 and the two completely sequenced genomes of P. profundum SS9 and P. gaetbulicola Gung47. Among total 7576 ortholog clusters identified, 2686 (35.5 %) are common to all three genomes (Supplementary Information Table S1). In addition, 831 (11.0 %) and 259 (3.4 %) ortholog clusters are shared by P. phosphoreum ANT-2200 with P. gaetbulicola Gung47, or with P. profundum SS9, respectively. Therefore, P. phosphoreum ANT-2200 seems more closely related with P. gaetbulicola Gung47 than with P. profundum SS9. Among the 259 orgholog clusters shared by P. phosphoreum ANT-2200 and P. profundum SS9 but absent from P. gaetbulicola Gung47 about 52.9 % has been annotated as Hypothetical proteins for the genome of the strain SS9 at MBGD. Another high functional category (8.9 %) is involved in cross-membrane transport. Therefore, it is difficult to provide a clear explanation of how the two species adapt to deep-sea habitats from comparative genome analysis.

P. phosphoreum ANT-2200 is moderately piezophilic (AlAli et al. 2010; Martini et al. 2013). Adaptation to high hydrostatic pressure of deep marine biosphere has been extensively studied in the moderately piezophilic strain SS9. The toxR/S and ompL/ompH gene clusters are widely distributed in members of the family Vibrionaceae. Welch and Bartlett have identified mutants of the toxR gene and demonstrated its involvement in pressure-responsive expression of OmpL and OmpH in P. profundum SS9 (Welch and Bartlett 1998). We have identified the toxR/S (PPBDW_v2_I21883/84) and ompH (PPBDW_v2_I21892) within a gene cluster, and a separated ompL (PPBDW_v2_I30078) on the chromosome I of ANT-2200 strain. A potential pressure sensing function of the ToxR in ANT-2200 needs to be confirmed.

Besides the similarity we also observed interesting differences between the two moderately piezophilic Photobacterium strains, i.e. SS9 and ANT-2200. It has been reported that photo-activated photolyase genes are absent from the genome of SS9, as expected for deep-sea bacteria living in habitats without sunlight (Vezzi et al. 2005). However, we found a phr gene (PPBDW_v2_II0175) encoding deoxyribodipyrimidine photolyase in a synteny group conserved among P. mandapamensis, P. damselae subsp. damselae CIP, Photobacterium sp. AK15, P. angustum S14, Photobacterium sp. SKA34 and piezosensitive P. profundum strain 3TCK, but absent from the piezophilic strain SS9. Deletion and insertion events occurred in the region downstream of phr in the synteny in the compared Photobacterium genomes, suggesting a genomic plasticity of this cluster. It’s possible that P. phosphoreum ANT-2200 might be in an early stage of its adaptation to deep biosphere and the vestigial phr gene functioning in euphotic bacteria has not been lost yet.

Flagellar apparatus

Bacterial motility and flagellar apparatus are evolved and regulated during their adaptation to different environment. Recently we reported a robust flagellar apparatus containing 7 flagella and 24 fibrils arranged into seven intertwined hexagonal arrays within a sheath (Ruan et al. 2012). Bacteria possessing such exquisite architecture are capable of circumventing and squeezing through obstacles (Ruan et al. 2012; Zhang et al. 2014b; Zhang et al. 2012), implying an advantage in their searching for nutrients at different layers of marine sediments. Flagellar composition has been proposed as a trait of adaptation to deep-sea habitats, i.e. the genomes of deep-sea piezotolerant bacterium Shewanella piezotolerans WP3 (Wang et al. 2008), piezophilic P. profundum SS9 and its deep-sea relative P. profundum DSJ4 (Campanaro et al. 2005; Eloe et al. 2008) possess both polar flagellar (PF) and lateral flagellar (LF) systems. In contrast, the piezosensitive strain P. profundum 3TCK lacks the lateral flagella (Campanaro et al. 2005). Interestingly P. phosphoreum ANT-2200 bears two major flagellar gene clusters, in addition to four loci encoding sodium ion-driven flagellar motor component (pomBA, motY, motX) and a separate fliL2 gene (Fig. 2). Cluster I on the Chromosome I (PPBDW_v2_I21979 to PPBDW_v2_I22023) has about 45 genes and is perfectly conserved compared to the polar flagellar gene cluster of P. profundum SS9 and 3TCK strains (Fig. 2). A locus composed of cheYZABW genes upstream of fliA is also conserved in the three genomes (data not shown). The major difference between them is the copy number of flagellin fliC gene, which has a single copy in ANT-2200 PF cluster but two copies in P. profundum. The gene cluster II consisting of 26 flagellar genes (PPBDW_v2_p0015 to PPBDW_v2_p0050) is located on the plasmid, suggesting a possible transfer among closely related bacteria. Five loci encompassing more than two genes (loci i to v in Fig. 2) are located in different order compared to those in gene cluster I, indicating a rearrangement during the evolution of flagellar gene clusters. As in P. profundum SS9, motA and motB encoding a proton-driven flagellar motor reside within the LF gene cluster, pomAB, moxY and motX encoding sodium ion-driven motors are present at three separated loci on chromosome I of ANT-2200 (Fig. 2).

Fig. 2
figure 2

Gene clusters coding flagellar systems in P. phosphoreum ANT-2200. Gene clusters encoding polar flagella (from PPBDW_v2_I21979 to PPBDW_v2_I22023) or lateral flagella (from PPBDW_v2_p0015 to PPBDW_v2_p0050) or loci of sodium-driven flagellar motor genes pomAB (PPBDW_v2_I22099 and PPBDW_v2_I22098), motY (PPBDW_v2_I20639) and motX (PPBDW_v2_I40044), or fliL2 (PPBDW_v2_I10172, putative basal-body associated protein) are presented in order and proportionally as in the genome with the functional colors indicated at the bottom-right. The cluster I is perfectly conserved in the genome of P. profundum SS9 except an inserted ‘::PBPRA0919’ gene, duplicated flagellin genes ‘flaA-flaC’ and 2 genes absent from SS9 genome as indicated with solid lines under the corresponding genes found in the ANT-2200 genome. The gene loci conserved in cluster I and cluster II are indicated with i, ii, iii, iv and v

Full size image

Versatile growth capacity of ANT-2200

Analysis of the genome of P. phosphoreum strain ANT-2200 reveals its versatile metabolic capacities. We have performed comparative analysis of MicroCyc metabolic pathways between P. phosphoreum ANT-2200 and Photobacterium profundum piezophilic strain SS9 and piezosensitive strain 3TCK at MaGe plateform. Among the 446 pathways analyzed, 249 are fully detected and 139 partially occur in ANT-2200 genome, while 239 are fully and 189 partially found in the genome of the piezophilic strain SS9 (Supplementary Information Table S2). In contrast, only 5 and 48 pathways are fully and partially detected in the partially sequenced genome of the piezosensitive strain 3TCK. Apparently more metabolic pathways were found in the partially sequenced genome of P. phosphoreum ANT-2200 than in the completely sequenced genome of P. profundum SS9. To corroborate the growth potential we analyzed its growth in minimal (oligotroph) and rich media (copiotroph) based on several metabolic pathways. The cultures were performed in dark to mimic conditions at 2200 m and at room temperature (22–25 °C) as it is a mesophile. Under anaerobic conditions bacteria derive the energy via fermentation on sugars or anaerobic respiration with organic or inorganic electron donors and acceptors. Genes encoding typical glucose phosphotransferase (PTS) systems and periplasmic binding protein dependent maltose uptake system were found in the genome of ANT-2200 (Table 3). When inoculated in the ANT-minimal media supplemented with either glucose or maltose (see “Materials and methods”), ANT-2200 grew well with maximal yield of about 0.3 absorbance at 600 nm (Fig. 3). These results confirmed the fermentation capacity of ANT-2200 on these sugars. Bacteria often use formate as an electron donor and nitrate, trimethylamine N-oxide (TMAO) or fumarate as electron acceptors to establish respiration chains. We found genes coding for formate dehydrogenases, nitrate reductase, TMAO reductases and fumarate reductase (Table 3). Supply of formate together with one of the three electron acceptors to minimal media sustained the growth of ANT-2200 whereas no growth was observed without formate or either electron acceptor. The low yield of biomass is probably due to poor carbon supply, and addition of fumarate increased the yield by about 50 % in comparison with those of TMAO and nitrate (Fig. 3). Both the growth rate and the maximal yield in rich media are largely increased comparing to the growth in minimal media, and addition of either the three electron acceptors in rich media could further improve the growth. Together these results show the versatile growth capacity of the strain P. phosphoreum ANT-2200 living at deep-sea habitat with sporadic inputs of various organic nutrients.

Table 3 P. phosphoreum ANT-2200 genes supporting fermentation, respiration and chemoorganotrophic growth analyzed in this study
Full size table
Fig. 3
figure 3

Growth of P. phosphoreum ANT-2200 under fermentation, respiration or chemoorganotrophic conditions. P. phosphoreum ANT-2200 was inoculated anaerobically in minimal marine media or rich media without glycerol (Rich) supplemented with formate (F), TMAO, nitrate, glucose or maltose. There was no growth in the minimal media with formate but without electron acceptor, that was used as blank reference in the measurement of absorbance

Full size image

It is noticed that while addition of nitrate and fumarate to the rich media raise the maximal yield slightly (A 600 increased from 0.5 to 0.6), addition of TMAO augmented it by around twofolds (A 600 increased from 0.5 to 1.1), indicating that TMAO might be the most suitable electron acceptor of those tested in supporting the growth of ANT-2200.

TMAO reductase in P. phosphoreum ANT-2200

Anaerobic respiration of TMAO involves the reduction of TMAO into TMA (trimethylamine), which is mainly catalyzed by TMAO reductase. In E. coli genes responsible for the TMAO reductase synthesis are located at two loci: torS-torT-torR-torCAD, and torYZ [(Ansaldi et al. 2000; Bordi et al. 2004; Simon et al. 1995), Fig. 4]. TMAO reductase consists of a periplasmic catalytic subunit (TorA or TorZ) containing molybdo-cofactor and a membrane-anchored cytochrome C (TorC or TorY). A dedicated chaperone TorD is involved in maturation of TorA (Genest et al. 2005). The expression of torCAD operon is induced by TMAO via the TorS-TorR two-component system and a periplasmic TMAO-binding protein TorT in E. coli (Ansaldi et al. 2001). We found four loci coding for TMAO reductases on the chromosome I of P. phosphoreum ANT-2200 (Fig. 4). The gene composition and cistronic structure are different from those found in E. coli. The first locus encompasses the regulatory genes torS-torT (PPBDW_v2_I20759/58) and TMAO reductase torC 1 -torA 1 genes (PPBDW_v2_I20753/52). The two tor clusters are separated by four CDS unrelated to TMAO metabolism. Intriguingly the torC 1 and torA 1 are encoded by the complementary DNA strands and unlike all current known torCA genes that form single transcription units (Fig. 4). The second (PPBDW_v2_I20849 to PPBDW_v2_I20847) and the fourth (PPBDW_v2_I21631 to PPBDW_v2_I21629) tor loci on the ANT-2200 genome are composed of torECA gene clusters (Fig. 4). The third torR-torC 3 A 3 D 1 locus (PPBDW_v2_I20986 to PPBDW_v2_I20983) corresponds to the canonical torR-torCAD gene cluster. The four TorA proteins in ANT-2200 shared identity ranging from 36 to 72 %, suggesting differences in their function. The piezotolerant strain SS9 also carries more than one copy of TMAO reductases in its genome, including a torR-torCAD locus (PBPRA1497 to PBPRA1494), a torECA locus (PBPRA1469 to PBPRA1467), a torCA locus (PBPRA2364 to PBPRA2363), a torT-torS1 cluster, (PBPRA1231 to PBPRA1232) and a torR 2 S 2 locus (PBPRA0025 to PBPRA0026).

Fig. 4
figure 4

Structure and composition of tor gene clusters for TMAO reductases. Arrows indicate the transcription and translation directions of the tor genes and arrow lengths are proportional to the gene sizes. Paralogous genes are indicated by the same color and grey color shows the genes unrelated to the TMAO metabolism

Full size image

Enhancement of ANT-2200 TMAO reductases by elevated pressure

To assess the effect of high hydrostatic pressure on TMAO reductases in ANT-2200 we measured the TMAO reductase activity of cells grown under different conditions. The four TorA proteins of ANT-2200 possess the twin-arginine translocation (TAT) export signal peptides, and thus, should be exported into the periplasm via the TAT pathway as previously reported for TorA in E. coli (Santini et al. 1998). We prepared the periplasmic fractions from ANT-2200 and analyzed the TMAO reductase activities by both activity staining after resolving the proteins on native polyacrylamide gels or enzyme assay of the periplasmic fractions. TorA alone exhibits the TMAO reductase activity when benzyl viologen or methyl viologen is used as an artificial electron donor to reduce TMAO. Two bands displaying the basic activity of TMAO reductase were observed in the periplasmic fractions of ANT-2200 cells incubated at atmosphere pressure without inducer (TMAO) (Fig. 5a, lane 1). When ANT-2200 cells were incubated at 22 MPa, equivalent to the pressure at depth of 2200 m, the activity of the upper-band was enhanced while the lower band remained unchanged (Fig. 5a, lane 2). The total TMAO reductase activity increased about three-fold (Fig. 5a). When TMAO as an inducer was added in the growth media, the total activity of TMAO reductase almost doubled. Meanwhile, the intensities of both bands augmented, and a barely visible band appeared above the upper-band (Fig. 5a, lane 3). However, no additional increase of the activity was observed when both TMAO and high pressure was applied (Fig. 5a, lane 4).

Fig. 5
figure 5

TMAO reductase activity of P. phosphoreum ANT-2200. ANT-2200 cells are incubated in rich media at atmosphere pressure (−) or 22 MPa (+), without (−) or with (+) TMAO. Periplasmic fractions were resolved on 10 % native polyacrylamide gels and TMAO reductase activity was visualized by activity staining (a). In parallel the TMAO reductase activity was measured spectrophotometrically by following the oxidation of benzyl viologen and specific activities (µmol TMAO reduced per min per mg of proteins) are presented (a). The normalized spectral counts for each TorA isoenzyme identified by tandem mass spectrometry are indicated in Panel b. Notably, only the values higher than 3 are meaningful and the spectral count 1 (in italic) for TorA3 in the upper-band might be due to a carry-over from other samples, which can be considered as an artifact

Full size image

We then identified TMAO reductase isoenzymes present in the upper and lower bands using a label-free shotgun proteomic procedure (Armengaud et al. 2014). Identification of a given isoenzyme was validated only when at least two specific and distinct proteotypic peptides had been detected. As shown in Fig. 5b, the upper-band contains TorA1 and TorA2 while the lower band has only TorA3. This identification was unambiguous as 34, 43 and 38 different peptide sequences were certified by tandem mass spectrometry, respectively. TorA4 has not been identified in this experiment. The normalized spectral counts allow the comparison of relative abundances of proteins between samples (Armengaud et al. 2014). When incubated at high pressure (22 MPa) TorA1 protein content increased by 3.8 folds whereas TorA2 and TorA3 only slightly increased (about 1.3 fold) compared to the cultures at atmospheric pressure. Addition of TMAO increases the TorA1 and TorA3 quantities by about 2.5 folds. TorA2 is less enhanced by elevated pressure or TMAO induction, but it is the most abundant TMAO reductases among the three isoenzymes. Therefore, TorA2 was constantly produced regardless the growth conditions.

TMAO regulates tor gene expression via TorR regulator that binds at TTCATA motif in the regulation regions of tor gene clusters (Ansaldi et al. 2000; Bordi et al. 2004; Simon et al. 1995). We found one binding site upstream of the torC 3 A 3 D 1 operon, one in the −70 to −80 region from the translation start codon of torC 1 and two in front of the torE 1 C 2 A 2 operon (Figure S2). Notably, transcription of torA1 is expected in opposite direction with that of torC1. The presence of these putative TorR binding sites is consistent with the increase in TMAO-stimulated reductase activities. The contribution of each TorA in supporting the ANT-2200 growth and mechanism of high pressure effect on torA gene expression could be analyzed once genetic tool has been established for this strain.

TMAO is highly abundant in fish tissues and the content increases proportionally with the depth where fish lives in ocean (Yancey et al. 2014). It serves as an efficient osmolyte to stabilize proteins against high hydrostatic pressure. The up-limit of the predicated isoosmotic state at 8,200 m has been considered as a biochemistry restriction that accounts for the absence of fish in the deepest 25 % of the ocean (8400–11,000 m) (Yancey et al. 2014). In parallel, transcriptome analysis showed that high pressure up-regulates the transcription of several genes involved in TMAO metabolism in P. profundum piezophilic strain SS9, including a TMAO sensor gene torS (Campanaro et al. 2005) and structural genes of TMAO reductase (subunit TorA) (Le Bihan et al. 2013; Vezzi et al. 2005). In addition, high pressure up-regulates the expression of protein TnaA tryptophanase that probably plays a role in counter-balancing the putative alkalinization due to trimethylamine reduction (Le Bihan et al. 2013). We identified the counterparts of these pressure up-regulated genes, except tnaA, in the genome of P. phosphoreum ANT-2200. Interestingly, among the three TMAO reductase isoenzymes of ANT-2200 we found that only TorA1 was significantly enhanced by elevated pressure at protein and enzymatic activity levels. In addition, simultaneous application of high hydrostatic pressure and TMAO did not produce an accumulative effect on the total TMAO reductase activity. Instead, it was even slightly reduced in comparison to sole application of pressure (Fig. 5). The pressure enhancement of TorA1 isoenzyme is thus dissociated from the induction and utilization of TMAO, suggesting it might be synthesized constantly in deep-sea piezosphere to quickly react to release of TMAO from fish in pervasive and changing gradients of nutrients.

Microorganisms adapt to the high hydrostatic pressure by various strategies, e.g. increasing polyunsaturation of membrane fatty acids, switching respiratory chains in energy metabolism and controlling gene expression via multiple regulatory systems (ToxS/R, OmpH/L, RecD) (Abe et al. 1999). Our genomic analysis confirmed the occurrence of these regulators in the deep-sea luminous strain ANT-2200 and revealed the highest copy number of gene clusters encoding TMAO reductases in bacteria. Biochemistry study showed the enhanced TMAO reductase activity under elevated hydrostatic pressure. Proteomic analysis corroborated the enhancement and further pointed out the increase of isoenzyme TorA1 as the major contribution to the activity augmentation. Together these results suggest that increasing TMAO reductase-coding gene copies and the enzymatic activity might be a strategy developed by bacteria to adapt to the deep-sea habitats where TMAO content increases with depth. Moreover, TMAO metabolism represents a novel model to shed light on molecular mechanism that governs the adaptation of microorganisms to the piezosphere.