Genomic comparison between different P. profundum strains
The first question that arises from the genomic comparison between SS9 and the other two strains is: "how many SS9 genes are missing or highly divergent in the 3TCK and DSJ4 genomes?". 544 ORFs were determined to be absent in 3TCK strain genome, 313 (9.1% of the ORFs located on chr1) belong to the SS9 chr1 and 231 (11.5% of the ORFs located on chr2) to chr2. 562 ORFs are absent in the DSJ4 genome, 292 (8.5%) are located on SS9 chr1 and 270 (13.5%) on chr2.
An interesting aspect of these data is that for both strains the percentage of missing/divergent regions is higher on chr2 than chr1. This indicates that chr2 (Figure ) contains a proportionally larger flexible gene pool and that it has been the target of more gene transfer events for its size (~2.2 Mbp) than chr1 (~4.1 Mbp) that contains the most "established" genes. This is also true for other Vibrionaceae
Figure 2 Genomic organization of the three P. profundum strains compared with expression level and differentially expressed genes obtained from microarray experiments. Form the outside inward circles represent: 1, 2) predicted protein-coding ORFs on the plus and (more ...)
In order to define if the regions absent in the 3TCK-DSJ4 strains could be considered to have been acquired by horizontal gene transfer we performed three different analyses: GC content variation (Figure , 9th
circle), tetranucleotide composition (genomic signature) (Figure , 10th
] and codon bias relative to the average gene versus S3 percentage (G+C content of codon site 3) (Figure and Additional file 1
Figure 3 Evidence for lateral gene transfer in SS9. Each P. profundum gene of ≥ 200 codons is represented in this graph by a point with co-ordinates corresponding to its codon bias relative to average gene and G+C frequency of codon site three. Genes having (more ...)
Taken together these three different analyses were able to identify a large number of potentially laterally transferred regions. For example, a region named Chr1.11 (named for its chromosome location and clockwise order of position, Figure ) has an altered tetranucleotide composition but its GC content is similar to the surrounding regions and it has a "normal" codon bias relative to the average gene versus S3 percentage (data not shown). Conversely, region Chr2.1 is only characterized by a slight GC content variation. The results obtained from these analyses are discussed in-depth below.
A BLASTP similarity search of SS9 proteins identified various phage-related proteins, mostly encoded in three regions named Chr1.8, Chr2.3 and Chr2.5. Microarray data obtained by comparing the SS9, 3TCK and DSJ4 genomes (Figure ) confirmed that these genomic portions are absent in both the 3TCK and DSJ4. These regions present characteristics typical of a genomic island (GI): (1) GC content anomalies, (2) altered codon bias, (3) insertion at the 3'-end of a tRNA gene (tRNA-N) and (4) presence of a gene encoding an integrase at one end (Table ) [10
General characteristics of putative horizontally transferred elements found in the genome of P. profundum SS9 strain.
Chr1.8 (46.8 kbp) has an altered tetranucleotide composition and presents twelve phage-related proteins, one of these (PBPRA1336) encodes a putative integrase protein. Nevertheless this region lacks the other two characteristics of a typical GI: GC content anomalies and the presence of a tRNA gene at one end. This region also contains two genes, encoding tryptophanase (TnaA) (PBPRA1344) and a hypothetical tryptophan-specific transport protein (PBPRA1345), involved in tryptophan transport and metabolism, and 27 hypothetical or conserved hypothetical proteins.
The region Chr2.3 is located on chr2, spans approximately 42.4 kbp, has no GC content anomalies (41.1%) but has a slightly altered tetranucleotide composition. Chr2.3 contains twelve ORFs that have similarity with phage proteins, one of these being a hypothetical integrase (PBPRB0551). Moreover it encodes a putative NAD(P)H oxidoreductase (PBPRB0548) and a putative TrkA family protein (glutathione-regulated potassium efflux system protein) (PBPRB0550). Furthermore PBPRB0559 gene has similarities with enterohemolysin 1, a gene also present in the Gifsy-1 prophage, and PBPRB0560 has similarities with exodeoxyribonuclease of the Gifsy-2 prophage of Salmonella typhimurium
Chr2.5 region appears to be completely absent in DSJ4 whereas only the first part is absent in 3TCK. Chr2.5 contains a hypothetical integrase gene (PBPRB1271), twelve phage related proteins and various hypothetical proteins. The higher GC content (46.2%) of Chr2.5 suggests that it has been acquired more recently than Chr1.8 and Chr2.3.
A large part of the genes located in the Chr1.8, Chr2.3 and Chr2.5 regions clearly lacks orthologous genes in others bacteria [12
] such as V. cholerae
, V. vulnificus
(strains CMCP6 and YJ016), V. parahaemolyticus
, V. fisheri
, E. coli
, and B. subtilis
. The high number of hypothetical proteins encoded in these regions suggests that these loci could have been acquired from bacteria still unknown. Consistent with its recent acquisition Chr2.5 presents an altered codon bias and a large percentage of its ORFs are located on the right horn of a graph of the codon bias versus the GC content frequency in third position (Additional file 1
A large 69.1 kbp element (Chr2.7-PI for Plasmid Integration) is present only in strain SS9 and seems to be the result of plasmid integration into the chromosome. This region has a high GC content and altered codon bias and genomic signature. Various bacterial conjugation factors (TrbCDEJBFGI and TraFLGIKL) are present in this element. These genes are typically found in widespread conjugationally transmitted plasmids [13
]. This element carries a large number of genes, but of particular interest is a multidrug efflux system (PBPRB1635, PBPRB1637, PBPRB1638).
SS9 also contains a 80 kbp plasmid (named "plasmid" in Table ) that shows similarity with V. vulnificus
plasmid YJ016 at least in the region spanning the genes related to conjugation. This plasmid is absent in both DSJ4 and 3TCK and presents characteristics of horizontally transferred DNA, in fact various ORFs belonging to this element are localized to the right horn of Figure (see also Additional file 1
PCR examination for the presence of this plasmid in various laboratory derivatives of P. profundum
SS9 has revealed that it can be lost: strain TW30 [14
] is a toxR-
derivative of DB110 [15
] which lacks the plasmid (Additional file 2
), yet TW30 exhibits no pressure or temperature growth defects. This raises the question of the function of the genes located on the plasmid which must be playing some role in environmental adaptation for the plasmid to be retained [16
One of the most interesting results obtained from the annotation process was the finding that in the SS9 genome there are two flagellar clusters. One of these (tentatively identified as the polar flagella cluster, PF) is located on chr1 between 993676 and 1046789 bp and the second cluster (tentatively identified as the lateral flagella cluster, Chr1.1-LF) is localized on chr1 near the origin between 13496 and 49254 bp. Similarity searches revealed that the PF region contains most of the genes involved in polar flagellum assembly, with a gene organization typical of a Vibrionaceae
polar flagellar cluster [17
]. Chr1.1-LF contains all the genes involved in lateral flagellar synthesis and most of the genes localized in this region have similarity with the flagellar cluster of V. parahaemolyticus
that codes for lateral flagella [18
]. This region is absent only in 3TCK strain (Figure ).
Finally, the data on variable regions was compared with the transcriptome results to discern if any of the horizontally acquired genes might be indicated to perform a role in pressure or temperature adaptation. The absolute fluorescence value from microarray analysis indicates that the expression levels of most of these genes was quite low. Indeed, only three regions in chr1 (Chr1.13, Chr1.14, Chr1.15) and two in chr2 (Chr2.8, Chr2.9) had fluorescence values higher than the mean fluorescence level of the entire chromosome (Table and Figure , 11th circle).
Region Chr1.12 is absent in DSJ4 (and in 3TCK this region is smaller) and has three genes, PBPRA2087 (hypothetical protein), PBPRA2084 (putatively evolved beta-D-galactosidase, alpha subunit) and PBPRA2086 (putative oxidoreductase), differentially expressed in pressure experiments.
Region Chr1.13 is absent in both strains and contains a large gene cluster involved in the tricarboxylic acid fermentation and in the cleavage of citrate to oxaloacetate and acetate. This region contains six genes up-regulated at 16°C and/or 0.1 MPa (PBPRA2289, PBPRA2292, PBPRA2295, PBPRA2298, PBPRA2300, PBPRA2303) (Additional file 3
Region Chr1.14 is absent in DSJ4 and contains genes involved in pilus assembly, some of these are up-regulated at 28 MPa (vs. 0.1 MPa) and down-regulated at 45 MPa (PBPRA2498, PBPRA2499, PBPRA2505).
Finally, region Chr1.15 is lacking in both of the comparison strains and contains genes having a high expression level at 28 MPa, some of which are differentially expressed at 28 MPa and 4°C (PBPRA2692, PBPRA2701, PBPRA2710). This region contains flm
genes that in other bacteria are involved in LPS O-Ag biosynthesis and flagellar filament assembly [19
]. Interestingly changes in LPS O-antigen structure have been observed in Yersinia pestis
KM218 grown at low temperatures [20
]. This element has very low GC content (29%–37%), altered genomic signature and most part of its genes are localized to the left horn of the graph reported in supporting online material (Table , Figure and Additional file 1
), thus supporting the idea that it could have been laterally acquired.
It is curious that some of the variable regions differentially expressed in pressure experiments are lacking or are highly divergent in both 3TCK and DSJ4. This indicates that although genes located in these regions could be involved in the high pressure response of SS9, they are not essential to it and other P. profundum strains will achieve piezophily with different strategies. Moreover, genes differentially expressed at 28 MPa or 45 MPa, but present in both DSJ4 and 3TCK, could be beneficial but not sufficient for high pressure adaptation.
Considering only pressure regulated genes belonging to the group absent in 3TCK and/or DSJ4, 29 genes are absent in both strains but only 9 genes are absent in 3TCK strain alone (Additional file 3
and Additional file 4
). These data were obtained using a profile search with JExpress software [21
]. Of these 9 genes, 6 are up-regulated at 28 MPa (vs. 0.1 MPa) and/or 45 MPa (PBPRB0026 hypothetical sensor protein TorS; PBPRA0776 hypothetical protein; PBPRA1912 hypothetical protein; PBPRA2251 hypothetical ABC transporter, periplasmic solute-binding protein, family 5; PBPRA2252 hypothetical ABC transporter, permease protein; PBPRA2573 putative long-chain fatty acid transport protein).
In the SS9 genome we found two genes for TorS proteins (PBPRA1232 and PBPRB0026). Only one of these (named PBPRB0026) is differentially expressed at 28 MPa and this gene is also absent in 3TCK strain. TorS is able to regulate various genes in response to trimethylamine N-oxide (TMAO) [22
], in particular it regulates TMAO reductase (PBPRA1467) that is also up-regulated at 28 MPa. It is conceivable that trimethylamine reduction increases the pH of the cytoplasm and, for this reason, other genes identified with microarray experiments, such as tryptophanase (PBPRB0382 and PBPRA2532) increase their expression in order to counter this alcalinization. Alternatively, since no TMAO was added to the SS9 cultures used for the microarray experiments, the second TorS could also be responding to an as yet undiscovered signal.
In addition to these six genes, others could perform an important role in high pressure adaptation. We expect that a high number of genes and proteins are regulated at the post-transcriptional level and have an important role in high pressure adaptation but these studies are beyond the scope of this paper. Moreover protein structural adaptations, that were not considered in this analysis, could also have great importance for SS9 piezophily.
Analysis of co-regulated genes in 28 MPa and 45 MPa experiments
The expression profile of SS9 at 28 MPa (the optimal growth pressure) was also compared with that at 45 MPa.
Interestingly the 45 MPa vs. 28 MPa expression profile comparison revealed only 68 differentially expressed genes (33 up-regulated and 35 down-regulated), in contrast to the high number of differentially expressed genes between 28 MPa and 0.1 MPa (101 up-regulated and 108 down-regulated). Of these 68 differentially expressed genes, only 31 were specific for very high pressure adaptation, the remaining 37 also being expressed under other environmental conditions tested. This result indicates that SS9 undergoes a heavy reorganization in gene expression between atmospheric pressure and 28 MPa, while this is not seen moving from 28 MPa to 45 MPa.
A Gene Ontology search with GoMiner software [24
] indicated that among the 31 genes specific for very high pressure adaptation there is an enrichment of genes involved in arginine metabolism (GO: 0006525), catabolism (GO: 0006527) (Additional file 5
) and transport (PBPRA2073-PBPRA2076). Experiments at 45 MPa were also useful in identifying genes whose expression follows the direction of pressure variation, being up-regulated or down-regulated both at 28 MPa (compared to 0.1 MPa) and at 45 MPa (compared to 28 MPa). Twenty one genes matched this expression profile (Additional file 3
One of these genes encodes a putative delta-9 fatty acid desaturase (PBPRB0742). High pressure increases the rigidity of membranes [27
], and for this reason genes such as the putative delta-9 fatty acid desaturase presumably were up-regulated in order to increase the membrane unsaturation and thus membrane fluidity. Despite the fact that much is known about membrane modification in response to pressure variation in SS9 [29
], these experiments reveal the possible involvement of a previously unrecognized gene in fatty acid unsaturation. This is particularly noteworthy because fatty acid unsaturation is critical to high pressure growth of SS9.
A search performed using GoMiner software [24
] on differentially expressed genes obtained in the 28 MPa vs. 0.1 MPa experiments and in the 28 MPa vs. 45 MPa experiments indicated that transport is one of the main biological processes involved (Additional file 5
). Similar result was obtained using FatiGO software [25
] (data not shown).
Moreover six of the ORFs that are up-regulated both at 28 MPa (vs. 0.1 MPa) and 45 MPa (vs. 28 MPa) are involved in transport processes (GO:0006810) (PBPRB1789, PBPRB1788, PBPRA2251, PBPRA1366, PBPRA0555, PBPRA1297). As described in the previously published 28 MPa versus 0.1 MPa experiments [4
], transport is strongly influenced by pressure, probably due to the effect of pressure on membrane modification and because of the pressure influence on the activation volume ΔV#
(the difference between the transition state volume and the initial volume in the system at equilibrium) of the transport process.