Phylogenetic relationship of the stringent response-related genes of marine bacteria

Bacteria living in marine environment encounter various challenges and limitations, thus in order to survive, they need to employ efficient stress-response mechanisms. One of these mechanisms is the stringent response, where unusual nucleotides, guanosine tetraand pentaphosphates, herald starvation and physico-chemical stresses. All so far sequenced free-living bacteria contain the gene(s) responsible for (p)ppGpp synthesis — rsh (named after Escherichia coli genes, relA and spoT). Two similar genes were identified mostly in βand γ-proteobacteria while other bacteria have only one gene coding the dual function of (p)ppGpp synthesis and degradation. Although the presence of (p)ppGpp-mediated response to the stress conditions has been shown for a few, and predicted for some other marine microorganisms, the (p)ppGpp effects may vary among different organisms. Thus, in this work we asked whether marine bacteria could have evolved a genetic adaptation specifically suited to adapt to environment with limited resources. The phylogenetic analyses of SpoT, RelA and RSH proteins from organisms associated with marine environment showed, however, that the evolutionary correlations obtained for these proteins are congruent with those constructed for 16S rRNA sequences and reflect taxonomical relationships of these organisms. Likewise, the similarity of specific amino acid residues indispensable for catalytic activity of these enzymes is very high, and any observed changes parallel with the taxonomical and evolutionary relationships. However, potential homologs of Mesh1 enzyme (metazoan SpoT homologs) that occur in both eukaryotic and prokaryotic organisms and contain the hydrolytic domain orthologous to SpoT were identified in Cellulophaga, Erythrobacter and Flavobacterium genera for the first time, as well as in soil bacterium Cytophaga hutchinsonii and freshwater Rhodothermus marinus.


INTRODUCTION
Marine environment is one of the most challenging habitats, because of recurrent changes in salinity, nutrient availability, temperature and many other factors such as pollution and UV radiation.Marine microorganisms, one of the most abundant groups in this habitat, are responsible for most of biomass turnover and food and energy cycles (Sogin et al., 2006).Unicellular organisms, including bacteria, are particularly sensitive to environmental alterations and challenges.Thus, a key role in their survival plays a prompt and effective response to these changes at the biochemical and metabolic level.In fact, marine bacteria are particularly well-adapted to an environment with limited resources; it has been documented that they can stop and resume their biological activities faster than bacteria that thrive in less restrictive environments (Amy et al., 1983;Kurath & Morita, 1983).However, the knowledge about the specific adaptation mechanisms in marine environment is limited.For example, one of the global regulatory mechanisms ensuring the survival under the stress condition, the stringent response, is studied mostly in Gram-negative models of Escherichia coli, soil bacterium Pseudomonas putida or Grampositive model bacterium Bacillus subtilis.
During the stringent response, unusual nucleotides, guanosine tetra-and pentaphosphate, ppGpp and pp-pGpp, referred to as (p)ppGpp, are synthesized promptly after starvation and physico-chemical stress, directly and indirectly affecting all major cellular processes such as sporulation, biofilm formation, quorum sensing, adaptation to adverse conditions, bacterial virulence (Potrykus &Cashel, 2008 andrefs therein, Dalebroux et al., 2010).However, the effects vary among different organisms and may depend on the type of stress, (p)ppGpp levels, the mechanism of (p)ppGpp action and the inducing conditions.(p)ppGpp has been identified in all free living eubacteria tested (Potrykus & Cashel, 2008) and chloroplast bearing plants (Braeken et al., 2006) but the enzymes responsible for its metabolism differ.
Escherichia coli and some of β-and γ-proteobacteria have two similar 74 kDa RSH (Rel Spo homolog) proteins: synthetase I, encoded by the relA gene, responsible for ribosome-dependent production of ppGpp upon amino acid starvation, and bifunctional synthetase/hydrolase, product of the spoT gene.SpoT-mediated production of ppGpp is induced by limitation of other nutrients (carbon, iron, nitrogen, phosphate, fatty acids) or by stresses (membrane, osmotic).Both enzymes bear high similarity to each other, however the strong hydrolase activity, localized in the N-terminal part of the protein (HD domain), is present only in SpoT.The synthesis activity is dependent on a neighboring domain that is similar in both proteins.The C-terminal domain is responsible for regulation of the enzyme's activity, and, for RelA, interaction with ribosomes.
A functional and structural study was performed on the RelSeq protein from Streptococcus equisimilis, including the crystal structure and mutational analysis of domains and importance of particular amino acid residues (Hogg et al., 2004).This protein, named RelSeq is an example of a single RSH enzyme with bifunctional synthesis and hydrolytic activities, present in many bacterial groups.The variety of (p)ppGpp metabolism-related enzymes has been evolutionarily classified by Mittenhuber (2001).Later, the thorough analysis including the class of short enzymes with only synthesis domains (for e.g.present in Gram-positive bacteria) was presented by Atkinson and collaborators (2011).An ortholog of the functional ppGpp hydrolase domain was also discovered in animal cells (Sun et al., 2010).This suggests a possible general role for ppGpp in all living organisms, not just bacteria and plants.
The presence of (p)ppGpp-mediated regulation in marine bacteria is expected from several lines of evidence: i) evolutionary benefits for their survival under conditions of nutrient and stress challenges, ii) impaired survival of strains with defective (p)ppGpp synthetase genes (Ostling et al., 1995;1996), iii) vast majority of bacteria analyzed to date have genes coding for (p)ppGpp-synthetizing enzymes.However, the information on the stringent response in marine microorganisms is very limited with only a handful of publications describing the stringent response of a single species, Vibrio sp.S14 identified later as V. angustum which can synthesize ppGpp during amino acid and carbon starvation (Flardh et al., 1992;1994;Ostling et al., 1996).
It was also hypothesized that the stringent response in marine bacteria may differ from the E. coli model: marine microorganisms retain a considerably higher residual rate of ribosomal synthesis during starvation (Flardh et al., 1992) and cell division occurs at a notably lower critical cell mass (Amy et al., 1983).Thus, we asked in this work whether marine bacteria could have evolved a specific genetic adaptation mechanism in terms of the stringent response to ensure optimal survival and efficient usage of the limited resources in this environment.

MATERIALS AND METHODS
All sequences used in this study were obtained from GenBank (http://www.ncbi.nlm.nih.gov/) or UniProt (http://www.uniprot.org/)databases and are presented in Table 1, except of Flavobacterium sp. and Paracoccus sp. that were generated in our lab (Joanna Karczewska-Golec, Maja Kochanowska-Łyżen, Paweł Olszewski, Marta Moskot, Magdalena Bałut, Arkadiusz Piotrowski, Piotr Golec and Agnieszka Szalewska-Palasz, to be published elsewhere) and deposited as a Whole Genome Shotgun project at DDBJ/EMBL/GenBank under the accession number JYGZ00000000 for Flavobacterium sp. and JYGY00000000 for Paracoccus sp.
We selected marine bacteria for which SpoT, RelA or RSH homolog protein sequences were available.Moreover, 16S rRNA sequences from the same taxa were downloaded.Sequences of Anabaena cylindrica and A. variabilis were also used in further analysis.The similarity searches for sequences were carried out by BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and alignments were done using MAFFT (http://www.ebi.ac.uk/Tools/ msa/mafft/).Next, the alignments were adjusted manually using MEGA5 (Tamura et al., 2011).
The trees were calculated using RaxML v.8 on CIP-RES Science Gateway V 3.3 (https://www.phylo.org/portal2/home.action)(Miller et al., 2010).For SpoT, RelA or Rsh trees PROTGAMMA model was employed and 100 bootstrap replicates were performed.For 16S rRNA tree, a GTR model was used and 100 bootstrap replicates were performed.The 16S rRNA tree was visualized using FigTree v 1.4.2.The branches representing multiple species belonging to the same genus are shown as collapsed.Other trees were visualized using TreeView (Page, 1996).Bootstrap supports ≥ 70 are shown above branches.

RESULTS AND DISCUSSION
We performed independent phylogenetic analyses of SpoT, RelA and RSH proteins from organisms associated with marine environment (Figs. 1, 2 and 3, respectively).We selected bacteria that are reportedly present in the Baltic Sea or other marine environment and whose SpoT, RelA or RSH sequences were available (Table 1).
The selection was based on information from publication records (Mudryk & Podgórska, 2005;Cabaj et al., 2006;Riemann et al., 2008;Stolle et al., 2011;Sjöstedt et al., 2012) and we also added strains that were isolated from the Baltic Sea in our laboratory.However, only some of them are strictly marine bacteria, while others are more flexible with respect to their habitats as they may occur in soil, rivers, as pathogens of different organisms etc.The information on their habitats is provided in Table 1.The tree was generated using RaxML and bootstrap supports are provided above branches.Names of species are followed with GenBank accession numbers of their SpoT amino acid sequences.The tree was generated using RaxML and bootstrap supports are provided above branches.Names of species are followed with GenBank accession numbers of their RelA amino acid sequences.

Figure 3. Maximum likelihood phylogeny of RSH homolog proteins from selected bacteria associated with marine environments.
The tree was generated using RaxML and bootstrap supports are provided above branches.Names of species are followed with GenBank accession numbers of their RSH amino acid sequences.The tree was generated using RaxML and bootstrap supports are provided above branches.Names of species are followed with GenBank accession numbers of their 16S rRNA nucleotide sequences.Some branches are presented as collapsed for multiple species of the same genus and only names of genera are provided.
In case of SpoT homologs, we used 71 sequences from different taxa and the final alignment had 734 positions.For RelA homolog analysis, we downloaded 67 sequences and the final alignment had 800 positions.In case of RSH we included 66 sequences and the alignment had 795 positions including Anabaena spp. that was used as an outgroup.
Simultaneously, we used the 16S rRNA gene to construct a phylogenetic tree (Fig. 4) for all organisms that were used in our analyses of proteins.In total, 128 taxa were selected and the final alignment of 16S rRNA used for phylogenetic analysis had 1437 positions.Anabaena cylindrica and A. variabilis were used as an outgroup.To simplify the tree, some branches representing species belonging to the same genus were collapsed.In all trees (Figs. 1, 2, 3, and 4) numbers above branches indicate bootstrap supports based on 100 replicates.
In the 16S rRNA tree (Fig. 4) bacteria that belong to β and γ-proteobacteria form highly supported clades with bootstrap supports of 100.These organisms are Gramnegative bacteria and occur in different environments.Alteromonas, Marinomonas, Photobacterium, Rheinheimera, Shewanella and Vibrio represent aquatic species (mainly marine bacteria) while others such as Acinetobacter, Pseudomonas or Serratia are not strictly associated with aquatic habitats and are often causative agents of diseases.Bacteria belonging to β and γ-proteobacteria encode two  Hogg et al., 2004).Blue lines underneath the sequences indicate the hydrolytic domain, red -synthesis domain, light blue -TGS domain, magenta -ACT domain.The nucleotide binding pocket is indicated by blue and red boxes, for hydrolytic and synthesis domains, respectively (based on Atkinson et al., 2011).Genetics of stringent response in marine bacteria paralogue enzymes in a single genome.SpoT and RelA homologs probably evolved after gene duplication or gene transfer, thus β and γ-proteobacteria gained an additional protein involved in the (p)ppGpp metabolism.Atkinson et al. (2011) proposed a hypothetical evolutionary history of RSH, RelA and SpoT and their functions in different lineages of bacteria suggesting gene duplication and then loss of the synthetase function of SpoT in Moraxellaceae.In our study, SpoT homologs from Acinetobacter and Psychrobacter spp.are also very divergent from those in other γ-proteobacteria.They form a highly supported clade in the SpoT tree, but with particularly long branches (Fig. 1) that reflects their individuality and perhaps a separate evolutionary history.In contrast, RelA proteins from Acinetobacter and Psychrobacter spp.do not differ significantly from homologs of other γ-proteobacteria (Fig. 2) and the RelA phylogenetic tree is congruent with the 16S rRNA tree (Fig. 4).
Other organisms have only a single RSH protein that is considered as an ancestral state.In 16S rRNA tree (Fig. 4) subclades representing each group of bacteria belong to Actinobacteria, Bacterioidetes and α-proteobacteria, and are highly supported with bootstrap values of 100.Among them Hyphomonas, Hirschia, Maricaulis, Erythrobacter, Thalassobaculum, Aurantimonas, Pelagibacter, Jannaschia, Flavobacterium, Cellulophaga, Formosa, Owenweeksia, Prolixibacter, Rhodothermus and Gracilimonas are associated with aquatic habitats.Others represent various lifestyles (Ta- ble 1).Comparison of 16S rRNA tree (Fig. 4) and RSH tree (Fig. 3) shows that each subgroup i.e. α-proteobacteria, Bacterioidetes and Actinobacteria form a highly supported clade with bootstrap supports of 100 in each tree.The evolutionary relationships obtained for the RSH protein are congruent with those determined on the basis of 16S rRNA data and reflect taxonomical resolution of these organisms.
We also generated and presented consensus sequence alignments of SpoT, RelA and RSH homologs from selected marine bacteria analysed in this study (Figs. 5,6 and 7 respectively).Escherichia coli homologs were used as reference for SpoT and RelA alignment (Fig. 5 and  6), while RelSeq from Streptococcus dysgalactiae subsp.equisimilis served as reference for bifunctional RSH proteins (Fig. 7).According to Hogg and collaborators (2004), we indicated sites that are indispensable for catalytic activities in RSH proteins with triangles.Analysis of consensus sequence alignments showed that all important positions in hydrolytic and synthesis domains are conserved in all SpoT homologs (Fig. 5).In case of RelA homologs from marine bacteria, the hydrolytic domain is highly mutated and that leads to loss of its activity as was also reported for other bacterial species (e.g.Atkinson et al., 2011), but the synthesis domain of RelA is conserved (Fig. 6).Bifunctional RSH proteins present as a single enzyme in these organisms exhibit high similarity in amino acid residues responsible for hydrolytic and synthesis activity of proteins (Fig. 7).Carboxyterminal region of RelA, SpoT and RSH also contains two domains: TGS and ACT.The conserved TGS region in SpoT and RSH plays a role in the regulation of catalytic activity of the enzyme, e.g.sensing the fatty acid starvation by binding the acyl carrier protein (Battesti &Bouveret, 2006;Potrykus & Cashel, 2008).This region is also conserved in analysed marine bacteria.The presence of ACT domain in CTD region was reported for typical RSH, RelA and SpoT enzymes, and it is also present in the sequences of marine bacteria chosen for these studies.The level of conservation of this domain is higher for RelA than for RSH.The ACT domain was suggested to play a role in modulating the intramolecular interactions and regulation of the enzyme activity.Thus, the differences in the amino acid sequences of these domains may indicate specific adaptations to environmental stresses.
The presence of an enzyme containing ppGpp hydrolysis domain (Mesh1) has been reported for metazoa (Sun et al., 2010).Some of bacterial genera also harbor Mesh1 homologs (Atkinson et al., 2011), thus we performed the search for Mesh1 in the collection of microorganisms analyzed in this study.In the analyses based on Drosophila melanogaster and bacterial (Methylobacterium extorquens DM4) Mesh1 sequences (Atkinson et al., 2011) we found Mesh1 homologs for e.g. in Burkholderia spp., Cellulophaga spp., Cytophaga spp., Erythrobacter spp., Flavobacterium spp., Methylobacterium aquaticum, Methylobacterium populi, Pelagibacter ubique, Pseudomonas spp., Rhodobacter sphaeroides and Rhodothermus marinus.These bacterial species belong to α-, βand γ-proteobacteria.The presence of Mesh1 in these classes of bacteria has been reported by Atkinson et al. (2011), including genera such as: Pseudomonas, Methylobacterium, Burkholderia and Rhodobacter.In some genera, such as Cellulophaga, Cytophaga, Erythrobacter or Flavobacterium, Pelagibacter, Rhodothermus, the presence of Mesh1 has not been reported previously.Although the role of Mesh1 in bacteria is unknown, the presence of Mesh1 homologs in the genomes of marine bacteria confirms that their genetic background regarding (p) ppGpp metabolism follows the pattern described for other microorganisms.Hogg et al., 2004).Blue lines underneath the sequences indicate the hydrolytic domain, red -synthesis domain, light blue -TGS domain, magenta -ACT domain.The nucleotide binding pocket is indicated by blue and red boxes, for hydrolytic and synthesis domains, respectively (based on Atkinson et al., 2011).

B. Guzow-Krzemińska and others
Table 1.List of organisms analysed in this study and GenBank Accession Numbers of their 16S rRNA gene sequences and RelA, SpoT or RSH protein sequences.Organism from which amino acid sequences were used for consensus alignment are indicated in bold.Information on habitat of the organisms is also provided.*"widespread" indicates that bacteria can inhabit various environments: soil, freshwater, marine water etc. **Accession number for whole genome sequencing, 16S DNA coordinates are 700320-701869.Marine microorganisms need to cope with changes in their environment and rely on signalling molecules such as (p)ppGpp to adapt to challenging conditions.Their lifestyles might be the reason for the evolution of two genes belonging to the RelA/SpoT family.However, we did not find any specific adaptation of marine bacteria in these terms as there are no obvious correlations with the presence of single RSH enzyme or both RelA and SpoT proteins and the bacterial lifestyles.Moreover, the similarity of amino acid sequences, and in particularly, specific amino acid residues indispensable for catalytic activity of enzymes is very high, and any observed changes are parallel with the taxonomical and evolutionary correlations.

Figure 1 .
Figure 1.Maximum likelihood phylogeny of SpoT homolog proteins from selected bacteria associated with marine environments.The tree was generated using RaxML and bootstrap supports are provided above branches.Names of species are followed with GenBank accession numbers of their SpoT amino acid sequences.

Figure 2 .
Figure 2. Maximum likelihood phylogeny of RelA homolog proteins from selected bacteria associated with marine environments.The tree was generated using RaxML and bootstrap supports are provided above branches.Names of species are followed with GenBank accession numbers of their RelA amino acid sequences.

Figure 4 .
Figure 4. Maximum likelihood phylogeny based on 16S rRNA gene sequences from selected bacteria associated with marine environments.The tree was generated using RaxML and bootstrap supports are provided above branches.Names of species are followed with GenBank accession numbers of their 16S rRNA nucleotide sequences.Some branches are presented as collapsed for multiple species of the same genus and only names of genera are provided.

Figure 5 .
Figure 5. Consensus alignment of SpoT homologs from selected marine bacteria.The E. coli SpoT was added for comparison.Positions that are indispensable for catalytic activity are indicated with triangles (blueamino acid residues conserved in SpoT, RelA and RSH, red -amino acid residues conserved only in SpoT and bifunctional RSH enzymes) (based on Hogg et al., 2004).Blue lines underneath the sequences indicate the hydrolytic domain, red -synthesis domain, light blue -TGS domain, magenta -ACT domain.The nucleotide binding pocket is indicated by blue and red boxes, for hydrolytic and synthesis domains, respectively (based on Atkinson et al., 2011).

Figure 6 .
Figure 6.Consensus alignment of RelA homologs from selected marine bacteria.The E. coli RelA was added for comparison.Positions that are indispensable for catalytic activity are indicated with blue triangles as amino acid residues conserved in SpoT, RelA and RSH (based onHogg et al., 2004).Blue lines underneath the sequences indicate the hydrolytic domain, red -synthesis domain, light blue -TGS domain, magenta -ACT domain.The nucleotide binding pocket is indicated by blue and red boxes, for hydrolytic and synthesis domains, respectively (based onAtkinson et al., 2011).

Figure 7 .
Figure 7. Consensus alignment of Rsh homologs from selected marine bacteria.The Streptococcus equisimilis RelSeq was added for comparison.Positions that are indispensable for catalytic activity are indicated with triangles (blue -amino acid residues conserved in SpoT, RelA and RSH, red -amino acid residues conserved only in SpoT and bifunctional RSH enzymes) (based on Hogg et al., 2004).Blue lines underneath the sequences indicate the hydrolytic domain, red -synthesis domain, light blue -TGS domain, magenta -ACT domain.The nucleotide binding pocket is indicated by blue and red boxes, for hydrolytic and synthesis domains, respectively (based on Atkinson et al., 2011).