Genome-wide analyses of chitin synthases identify horizontal gene transfers towards bacteria and allow a robust and unifying classification into fungi
Chitin, the second most abundant biopolymer on earth after cellulose, is found in probably all fungi, many animals (mainly invertebrates), several protists and a few algae, playing an essential role in the development of many of them. This polysaccharide is produced by type 2 glycosyltransferases, called chitin synthases (CHS). There are several contradictory classifications of CHS isoenzymes and, as regards their evolutionary history, their origin and diversity is still a matter of debate.
A genome-wide analysis resulted in the detection of more than eight hundred putative chitin synthases in proteomes associated with about 130 genomes. Phylogenetic analyses were performed with special care to avoid any pitfalls associated with the peculiarities of these sequences (e.g. highly variable regions, truncated or recombined sequences, long-branch attraction). This allowed us to revise and unify the fungal CHS classification and to study the evolutionary history of the CHS multigenic family. This update has the advantage of being user-friendly due to the development of a dedicated website (http://wwwabi.snv.jussieu.fr/public/CHSdb), and it includes any correspondences with previously published classifications and mutants. Concerning the evolutionary history of CHS, this family has mainly evolved via duplications and losses. However, it is likely that several horizontal gene transfers (HGT) also occurred in eukaryotic microorganisms and, even more surprisingly, in bacteria.
This comprehensive multi-species analysis contributes to the classification of fungal CHS, in particular by optimizing its robustness, consensuality and accessibility. It also highlights the importance of HGT in the evolutionary history of CHS and describes bacterial chs genes for the first time. Many of the bacteria that have acquired a chitin synthase are plant pathogens (e.g. Dickeya spp; Pectobacterium spp; Brenneria spp; Agrobacterium vitis and Pseudomonas cichorii). Whether they are able to produce a chitin exopolysaccharide or secrete chitooligosaccharides requires further investigation.
KeywordsChitin synthase Bacteria Evolution Classification Horizontal gene transfer Fungi
Chlorovirus-like CHS class
Ectocarpus siliculosus-like CHS class
Horizontal gene transfer(s)
Long branch attraction
Last Opistokonta common ancestor
Recombined chitin synthase
UDP-N-acetylglucosamine 6-dehydrogenase gene
Chitin is a biological polymer consisting of carbohydrate molecules bonded together to form long chains of polysaccharides. Unlike starch and glycogen that are storage polysaccharides in plants and animals respectively, chitin is a structural polysaccharide organised as crystalline microfibrils and with enormous tensile strength. It contributes to the rigidity and integrity of cells, tissues or body surfaces in a wide range of organisms, protecting and giving them shape, as seen with cellulose or pectin in plant and algal cell walls. So far, the presence of this structural polysaccharide has been mainly demonstrated in the cell walls of mycota, the exoskeleton of hexapoda or crustacea, and in the radula or beak of mollusca , where it plays a major role in development and growth. In addition, the presence and subcellular location of chitin in invertebrate hemocytes suggests another role for this polysaccharide in the immune system of diverse animals . It was generally thought that there was no chitin in vertebrates but this polymer has been described in several ray-finned fishes (Actinopterygi) [3, 4] and in some amphibians . The role of endogenous chitin in the biology of these vertebrates remains elusive. Chitin has also been sporadically found in structures from a diverse range of eukaryotic microorganisms, such as the cell wall of a few chlorophyta (green microalgae), the cyst wall or lorica, of ciliophora (ciliated protozoans), the theca of choanoflagellida (flagellated protozoans), and the test or cyst wall, of amoebozoa (amoeboid protozoans) . It is also present in the large family of heterokonta protists, for example in the cell wall of oomycota, the spines of diatomae and the stalk of chrysophyta .
Chitin is an hexosamine polymer composed of beta-(1,4)-linked linear chains of more than 5,000 N-acetylglucosamine residues that are highly cross-linked with hydrogen bonds. In insects, chitin is deposited exclusively on the apical sides of epithelial cells, facing the external environment (body surface, gut and tracheal lumen) . In fungal cell walls, there is a common fibrillar core composed of branched beta-(1,3)-glucan to which chitin and other polysaccharides are covalently bound. Chitin accounts for 1-2% of the cell wall mass in yeasts and up to 30% in molds [9, 10]. Elongation of the chitin polymer is catalyzed by a highly conserved enzyme called chitin synthase, CHS (UDP-N-acetyl-D-glucosamine: chitin 4-beta-N-acetylglucosaminyl-transferase; EC 188.8.131.52). The CHS enzyme belongs to the GT2 family of processive polymerizing glycosyltransferases which includes the synthases for cellulose, callose, curdlan, mannan, hyaluronate and alginate polymers . In fungi, chitin biosynthesis requires a set of multiple CHS isoenzymes that are encoded by a multigenic family. Although they share a common central catalytic domain, CHS isoenzymes from fungi and other species (metazoa, protists) can differ greatly in their N- and C-termini parts. All CHS utilize UDP-N-acetylglucosamine (UDP-GlcNAc) from the cytoplasm as a substrate and catalyze multiple transfers of the activated sugar donor to the non-reducing end of the growing chain. Multiple transmembrane domains are found in every CHS protein and these probably form a channel in the cell membrane through which linear chitin is extruded into the extracellular space as it is described for cellulose biosynthesis . Nevertheless, the exact mechanism by which chitin is assembled into microfibrils and cross-linked with other components of the cell wall is still poorly understood.
Summarized history of chitin synthase protein classifications in fungi
In order to update and standardize CHS classifications, we performed a comparative multi-species analysis across many sequenced and annotated genomes. A databank of CHS proteins was generated from a similarity search of CHS Pfam domains in the complete proteomes. Maximum likelihood (ML) and Bayesian phylogenetic trees were constructed to provide a global view of the whole CHS family. Applying a rigorous method of phylogenetic analysis, we organised the fungal sequences and obtained a more robust classification. We extended the study to chitin synthases from other species in order to elucidate chs gene evolution. In particular, this genome-wide phylogenetic analysis confirms the occurrence of multiple gene losses, duplications and horizontal gene transfers within this family of glycosyltransferases. Surprisingly, this work provides, for the first time, evidence of a chitin synthase horizontal gene transfer from eukaryota to bacterial genomes.
Results and discussion
Update of CHS fungi classification
We observed that in Dikarya fungi, Ascomycota and Basidiomycota, up to 15 CHS-encoding genes were found in the same species (e.g. Postia placenta; Additional file 3: Table S3). In Ascomycota yeasts, the number of CHS is usually lower than in filamentous fungi and ranges from one (e.g. Schizosaccharomyces pombe) to seven (Yarrowia lipolytica) . In Hemiascomycota yeasts, the presence of seven CHS in Yarrowia lipolytica and only three in Saccharomyces cerevisiae can only be explained by several losses in this clade (Additional file 4: Figure S1). In Mucoromycotina, Chytridiomycota and Blastocladiomycota, 15 to 38 chs genes were found depending on the species suggesting an even larger expansion of the CHS family in these early-branching fungal lineages (Additional file 3: Table S3). Finally, none or only one, chs gene was found in four Microsporidia fungal genomes (Encephalitozoon cuniculi, Encephalitozoon intestinalis, Nematocida parisii and Nosema ceranae). This characteristic seems consistent with the extreme reduction and compaction that was observed in these particular genomes . The large number of detected CHS, and their diversity, provided the opportunity to establish a more robust fungal classification for this protein family. In our definition, a class means a set of CHS sequences, from different Ascomycota, Basidiomycota and/or Mucoromycotina species, which forms a well-supported group in the trees constructed with two phylogenetic approaches (ML and Bayesian). For now, we consider the genome quality and number from Chytridiomycota and Blastocladiomycota species insufficient to accurately classify these sequences. Ascomycota yeast CHS sequences were excluded from most phylogenetic analyses in order to reduce the risk of long-branch attraction (LBA) as substantial artefact was described in the phylogeny of Saccharomyces species . LBA could have contributed to erroneously cluster together long branches, irrespective of the true relationships of sequences . Such exclusion is not a problem as classification of CHS from yeasts is easy (Additional file 4: Figure S1).
Fungal division 1 : five new classes A to E in addition to classes I, II and III
Fungal division 2 : two superclasses IV and V
Fungal division 2: two virus-like classes
In fungal division 2, in addition to the two superclasses IV and V, two others classes were found (Fig. 2). These fungal classes, that we have called ESV (Ectocarpus siliculosus Virus-like CHS) and CV (Chloroviruses-like CHS), were found in proteomes of some Ascomycota (Sordariomycetes, Dothideomycetes and Eurotiomycetes), some Basidiomycota (Ustilaginomycetes only possessing a ESV CHS), a Chytridiomycota (only possessing a CV CHS) and some Phycodnaviridae giant algal-viruses (Additional file 6: Table S4). The CV class was recently proposed as a new class VIII  but we do not recommend this denomination as class VIII is used for other distinct groups of putative CHS [10, 23] (Additional file 2: Table S2). Independent horizontal transfers of these chs genes with their neighboring genes, in fungal genomes from chloroviruses and phaeoviruses respectively, have already been suggested . In addition to the strong similarity between CV and ESV viral and fungal sequences, the corresponding genes are clustered on genomes with an UDP-N-acetylglucosamine 6-dehydrogenase gene (UNGD). This two-gene cluster, conserved in some fungal species and in viruses of distantly related algae, suggests that these two genes were probably transmitted together and that this gene pair might work in concert : UNGD family enzymes provide precursors for glycosyltransferase enzymes . Moreover, the fungal and viral CV genes are also colocated with a polysaccharide deacetylase. These genetic material exchanges between fungi and algal-viruses are perplexing. However, as exemplified by the 600 million year old lichen-fossils, interactions betwen fungi and algae have long existed . Furthermore, gene transfers from the ancestor of Dothideo/Sordariomycetes to the ancestor of the terrestrial alga Trebouxia decolorans have also been proposed . There is currently no evidence for the functionality of the fungal proteins encoded by the genes from the ESV class. However, the CV class genes of Glomerella graminicola and Gibberella zeae were found to be differentially expressed during plant infection  and a deletion mutant in a CV class chs gene was recently described in Fusarium graminearum (called Fgchs8 gene) . Disruption of this gene resulted in a reduced accumulation of chitin, decreased CHS activity, sensitivity to SDS and reduced pathogenicity. It has been suggested that the Fgchs8 gene is required for cell wall development in F. graminearum. The chs genes from the CV and ESV classes were duplicated in some Ascomycota lineages and they were lost in others  (Additional file 6: Table S4). They also show higher evolutionary rates than other fungal CHS (Fig. 2; Additional file 5: Figure S2). The virus containing a chs gene from the ESV class is, more precisely, a lysogenic phaeovirus. DNA from this virus, including the ESV chs gene, is integrated into the genome of the pluricellular brown alga Ectocarpus Siliculosus . This integrated viral ESV chs gene is transcriptionally silent in the algae and is probably not functional . Viruses with at least one chs gene from the CV class include some chloroviruses infecting the unicellular green alga Chlorella with a lytic infection style (e.g. Paramecium Bursaria Chlorella viruses). Some chloroviruses form chitin on the surface of infected cells which might protect virus-infected algae from uptake by other organisms . Indeed, the heterologous expression of the chlorovirus CVK2 chs gene was performed in E. coli leading to the production of fibers by the bacterium .
Outside fungal divisions : a class of CHS probably emerging from a recombination event
In our analysis, a complete CHS class, previously called VI, VII (Table 1; Additional file 2: Table S2) or sometimes division 3 , was treated separately. We suspect a recombination event of being at the origin of this class (Fig. 3). Associated with a reduced taxonomic distribution in fungi, including some Ascomycota groups (Sordariomycetes, Dothideomycetes, Eurotiomycetes and Leotiomycetes) and a Mucoromycotina species (Additional file 6: Table S4), this class was also recently detected in a chromalveolate (Rhizaria), Plasmodiophora brassicae, an obligate biotrophic pathogen of crucifers . In this study, we suggest that the corresponding protein sequences have characteristics associated with a rearrangement in their ancestor: a duplication of the QXXXY motif (motif d in Fig. 3 and Additional file 7: Figure S3) and a phylogenetic signal which seems to differ in the N- and C-termini sequence fragments located on either side of this duplication (Additional file 8: Figure S4). Indeed, while the C-terminal region appears to be similar to that in other CHS (Additional file 8: Figure S4A), the N-terminal region is closer to hyaluronan synthase proteins in the phylogenetic tree (Additional file 8: Figure S4B) and shares the same organization of transmembrane domains (Fig. 3). This observation could be explained by an ancient recombination between the ancestor’s sequences of two glycosytransferase family 2 proteins. An alternative explanation, that we can not completely exclude, is that the ancestor of this class underwent a period of an accelerated rate of evolution which blurred the phylogenetic signal. In both cases (recombination or transient high evolutionary rate), we excluded them from phylogenetic analyses as they might have provoked artefactual groups due to long-branch attractions. HGTs are probably at the origin of this CHS class in one Mucoromycotina (Mortierella verticillata), one Chytridiomycota (Batrachochytrium dendrobatidis) and one chromalveolata (Plasmodiophora brassicae) (Additional file 6: Table S4). In Ascomycota groups, where this class is present, the corresponding genes are probably functional as they are strictly fixed in one well-conserved copy (Additional file 3: Table S3; Additional file 5: Figure S2). However, their chitin synthase activity has not yet been proved so we recommend using the term recCHS (“recombined” CHS) for these sequences. Few fungal deletion mutants were obtained for this class and their phenotypes are divergent. RecCHS orthologs have probably evolved with different roles in these fungi during growth and development [41, 42].
Origin of CHS in eukaryotes
The phylogeny was obtained with a protein sequences sample representative of the entire set of chitin synthases detected in this study (Fig. 2). At least three chitin synthase genes were present in the ancestor genome of Opistokonta: one ancestral chs for the Metazoa division, one for the division 2 and one for the division 1 (see red triangles in Fig. 2). A previous study suggested four ancestral chs genes in the Last Opisthokonta Common Ancestor (LOCA)  but we propose that two of them belong to division 1 and they may have diverged from a common ancestral sequence in LOCA (see black triangles in Fig. 2). The chitin synthase genes probably appeared earlier, given the basal positions in the Metazoa division of one sequence of the Amoebozoa Entamoeba histolytica and the CHS of the apusomonad Thecamonas trahens. The distribution of taxa possessing a chs gene in Metazoa indicates that chs genes were lost several times independently in this phylum (Fig. 1; in Platyhelminthes, in Hemichordata, in some Actinopterygii etc.). On the other hand, family expansion occurred in some species, such as in the gastropod Lottia gigantea  and the amphioxus Branchiostoma floridae . It is noteworthy that a second CHS sequence from Entamoeba histolytica did not significantly group with any of the chitin synthase divisions, which raises the question of its origin. However, heterologous expression of this protein in the yeast Saccharomyces cerevisiae confirmed its CHS activity . The presence of chitin synthases belonging to different divisions in diverse chromalveolates, such as ciliates, diatoms, oomycetes and other protists, might be the result of a deeper evolutionary origin of CHS in the Last Common Eukaryotic Ancestor [40, 47] (Figs. 1 and 2; Additional file 3: Table S3). However, an alternative plausible model for the origin of chromalveolate CHS could imply several independent horizontal gene transfers at different times during the chromalveolate evolutionary history , as we suggested for the chs of the rhizaria Plasmodiophora brassicae (see above in Update of CHS fungi classification). Diatom chs genes from Thalassiosira pseudonana (division 2) and Phaeodactylum tricornutum (division 1), were probably acquired by an HGT from Opistokonta (fungi or metazoa; Fig. 2).
Other HGT must have occurred to explain the actual taxonomic distribution of chs genes and their sequence diversity. If the division 1 chs of the green algae (Trebouxiophyceae) Chlorella  (Fig. 2) and Picochlorum  were vertically inherited from the Plantae ancestor, it would imply that these genes were independently lost in many lineages of the plantae. As mentioned for the CV and ESV classes of division 2 (Update of CHS fungi classification), chs gene exchanges might have occurred between algae-related viruses, including chloroviruses and fungi. However, the Chlorella CHS belongs to division 1 and seems to resemble oomycete and ciliate CHS more than fungal ones (Fig. 2). Therefore, a possible source of the transfer could be a ciliate living in symbiosis with a green alga, such as Paramecium bursaria.
Hence, the evolutionary history of chitin synthase genes suggests different independent horizontal gene transfers among diverse eukaryotic microorganisms (Fig. 2). While viruses are known to be gene transfer agents, membrane vesicles, which are not impaired by receptor recognition in the way that viruses are, could be large spectrum transducing agents [51, 52, 53].
Bacterial chitin synthase genes
In a large variety of bacteria, type 2 glycosyltransferases (GT2) homologous to chitin synthases have been described. These transmembrane enzymes are localized at the inner membrane of Gram negative bacteria and synthesize different exopolysaccharides (EPS) into the periplasm. These EPS are made of cellulose, curdlan, alginate or hyaluronic acid . In Pseudomonas aeruginosa, secretion of alginate through the peptidoglycan and the outer membrane is ensured by an envelope-spanning multiprotein complex including channel proteins . We found that some bacterial chs genes are colocated with porin or efflux protein-encoding genes (Fig. 6). It is possible that these bacterial chs genes, and their neighbouring transporter encoding-genes, are involved in the production and secretion of an EPS made of chitin.
EPS are matrix components of bacterial biofilms which are known to play a major role in pathogenic or symbiotic interactions between bacteria and animals or plants . It is noticeable that, among bacteria in which a chs gene was found, Pseudomonas cichorii and Agrobacterium vitis are plant-pathogens and secrete an EPS made of alginate and curdlan, respectively [69, 70]. Indeed, A. vitis is able to build biofilms on abiotic as well as on plant root surfaces, probably due to its EPS . Other bacterial species displaying a chs gene are also plant pathogens: a Brenneria spp., three Pectobacterium species and two Dickeya dadantii isolates. The later two bacterial isolates, D. dadantii Ech586 and D. dadantii Ech703, should be reclassified as two different species, Dickeya zeae Ech586 and Dickeya paradisiaca Ech703, respectively . It may be that the bacterial chitin synthase plays a role in the parasitic interaction between these bacteria and their plant hosts.
CHS activities in fungi or metazoa are usually described as producing long crystalline chains of chitin (fibers) but it is possible that a bacterial CHS secretes small soluble chitosaccharides, or eventually chitooligosaccharides (COS), instead. Indeed, the phytopathogenic oomycete Aphanomyces euteiches does not secrete chitin fibers although two chs genes are present in this species . However, the encoded CHS activities of A. euteiches are thought to be active because a small soluble and noncrystalline glucan-chitosaccharide was detected in its cell wall . Interestingly, one of the two chs genes in D. dadantii Ech703 and the chs gene of Pseudomonas cichorii are co-located and have the same orientation as a putative chitoporin-encoding gene (Fig. 6). In the marine bacterium Vibrio harveyi, chitoporin is a transporter of COS through the outer membrane . COS are known to play a key role as signal molecules in multiple plant-microbe interactions. In parasitic interactions between plants and fungi, COS are released from the digestion of the fungal cell wall by secreted plant chitinases. The resulting COS are then recognized by plant receptors as Microbial Associated Molecular Patterns (MAMP) and they induce plant immunity . In symbiotic interactions, modified COS, called lipochitooligosaccharides (LCOs), which include Myc and Nod factors, modulate plant host immunity. Myc factors are produced and secreted by arbuscular mycorrhizal fungi but the enzymes responsible for their biosynthesis have not yet been identified . Nod factors are synthesized by NodC, a bacterial GT2 homologous to CHS, and they induce nodule formation during a symbiotic interaction between Rhizobiacea (nitrogen-fixing bacteria) and leguminous host plants . Whether bacterial chs genes are able to synthesize COS implicated in plant interactions is still uncertain.
Several bacterial chs genes are fused with a putative aminotransferase domain and they are colocated with a putative sugar epimerase-encoding gene (Fig. 6). These two additional functions could be associated with modifications of chitin or COS. A gene of unknown functions, and composed of the DUF1501 domain, was also found colocated with several bacterial chs. The encoded DUF1501-containing protein carries a twin-arginine motif which would imply that it is exported by the twin-arginine translocation (Tat) pathway (http://www.compgen.org/tools/PRED-TAT/ ). The bacterial Tat system allows folded proteins to be moved across membranes without significant ion leakage . The DUF1501 domain has also been described as forming part of a conserved machinery in compartimentalized species from the Planctomycetes, Verrucomicrobia and Chlamydiae (PVC) super-phylum . These informations about DUF1501 proteins are hardly connectable with the biological role of known CHS. However, our hypothesis is that DUF1501, DUF1800 and CHS proteins might have a functional link in bacteria as they share a common conserved operon organization. It would be very interesting to study the function of bacterial CHS and also of the proteins that seem associated to them. Finally, in the particular cases of Teredinibacter turnerae and Cellvibrio mixtus subsp. mixtus, the chs genes are localized in the middle of secondary metabolism gene clusters (80 Kb and 35 Kb, respectively) and they could be involved in the production of a bioactive molecule .
Bacterial chitin synthase genes constitute a new example of genes acquired by a bacterium via horizontal transfer from a eukaryotic donor. The chitin synthase activity of the bacterial CHS and the possible selective advantage for the corresponding bacteria, often implied in plant interactions, needs further investigation. CHS-encoding genes have also been transferred between eukaryotic microorganisms. We recommend avoiding the use of this multigenic family to elucidate the phylogenetic relationships between the different eukaryotic species, especially since many duplications and losses are also observed in different lineages.
Fungal CHS are, unfortunately, not an exception and the difficulty of classifying these sequences has led to discordant classifications. We took advantage of the current study to determine which can be robustly classified and then to construct the most consensual classification possible. To facilitate the use of this new classification, any information that corresponds with the previously published versions is provided (Additional file 2: Table S2), together with the databank, in fasta format, of all the classified CHS sequences (Additional file 12: File S1). A website also permits blast queries to the databank (http://wwwabi.snv.jussieu.fr/public/CHSdb). This facilitates a search for the class of a previously identified CHS, even if the corresponding accession number has changed (as is often the case for fungal sequences). It is also an aid for the classification of CHS from species closely related to those analyzed.
The 208 eukaryotic complete proteomes were downloaded from different databases (Additional file 13: Table S6). The main sources were: Ensembl (56 species; http://www.ensembl.org/), the DOE Joint Genome Institute (53 species; http://www.jgi.doe.gov/), the National Center for Biotechnology Information (31 species; http://www.ncbi.nlm.nih.gov/), and the Broad Institute of MIT and Harvard (29 species; http://www.broadinstitute.org/). The tested proteomes of 1218 bacteria, 97 archaea and 2398 viruses were also downloaded from the NCBI (Additional file 14: Table S7). The hyaluronan synthase and nodulation NodC protein sequences employed in this study were those used in  and .
Identification of chitin synthases
CHS protein sequences share several motifs (Fig. 3; Additional file 7: Figure S3), spanning the catalytic site and the C-terminal part of these enzymes, and two major related chitin synthase Pfam domains , Chitin_synth_1 (PF01644) and Chitin_synth_2 (PF03142). To identify the CHS sequences, similarity searches were first performed with these two domains in complete proteomes (see Methods – Data collection) using the RPS-BLAST program , with an E-value cutoff of 1e-05. This cutoff was judged sufficiently high to detect all chitin synthase proteins as some hits corresponded to other glycosyltransferase proteins of the GT2 family. Indeed, unlike the Chitin_synth_1 domain which is very specific to division 1 CHS sequences, the Chitin_synth_2 domain could be found in dozens of glycosyl transferase sequences that were not chitin synthases (e.g. hyaluronan synthases, cellulose synthases etc.). To eliminate these non-CHS sequences, phylogenetic trees were performed (see Methods – Sequences analysis) and only proteins forming a clear monophyletic group with known chitin synthase divisions, using hyaluronan synthase and nodulation NodC protein sequences as the outgroups, were considered as chitin synthase proteins. The presence of conserved motifs essential for enzymatic activity (D, D, D, QXXRW) was also checked as it guarantees that the proteins are potentially functional and could endow a chitin synthase activity. Proteins that do not possess these motifs were, thus, considered as dubious and disqualified from our study. For example, a recently proposed class VIII  was not retained due to the absence of the essential motif QXXRW.
Some CHS classes showed reduced taxonomic distribution (recCHS, ESV, CV, Bacterial CHS etc.). To gain a better understanding of the origin of these sequences, it was important to have a more precise idea of the species that possess the corresponding genes. In these cases, the analysis was completed by a BLASTP search, carried out at the NCBI, using the non-redundant protein sequences (nr) database.
All phylogenetic analyses were performed using the following procedure. First, amino acid sequences were aligned using MAFFT with the E-INSI algorithm and default settings . Next, regions in the resulting multiple sequence alignments that were suitable for phylogenetic inference were selected using BMGE . The BLOSUM60 and BLOSUM30 matrices were used, respectively, for alignments with sequences from a single division and alignments containing sequences from different divisions. This step removed any ambiguously aligned or highly variable regions in order to improve the overall performance of the phylogenetic reconstructions. Thirdly, phylogenetic inferences were obtained with two approaches. ML trees were constructed with PhyML 3.0  using the following parameters: the LG model, with empirical amino acid frequencies, an estimated proportion of invariable sites, subtree pruning and regrafting (SPR) and five random starting trees added to the standard BioNJ starting tree. The support of the data for each internal branch of the phylogenies was estimated using non-parametric bootstraps, with 100 replicates. Bayesian inferences were obtained with MrBayes v3.2.6  with a fixed WAG model of amino acid substitution and a gamma correction (four discrete categories plus a proportion of invariant sites). The program was run with four chains for 100 000 generations and trees were sampled every 100 generations. To construct the consensus tree, the first 250 trees were discarded “burnin” and posterior probabilities were used as support for internal branches. NJplot  and FigTree  were used for outputting ML and Bayesian trees respectively.
Hyaluronan synthase (HAS) and NodC sequences were used as the outgroup for the global CHS tree (Fig. 2) as they are distinct type 2 glycosyltransferases. All CHS sequences from the databank were classified using phylogenies. However, only a part of them was used in the phylogenetic trees presented here, so that the trees remain readable. They correspond to the CHS of a sample of species chosen to maximised the preservation of the global CHS tree topology observed with all the sequences.
Genomic context analysis
The genomic context of bacterial chitin synthases was analyzed with multiple genome alignments using Mauve 2.3.1 . Gene organization around the chitin synthase encoding-genes was compared with that of ortholog regions in genome(s) from proximate species lacking the chitin synthase gene. Hence, four groups of genomes were compared. First, the Pectobacterium group comprised Pectobacterium carotovorum subsp. carotovorum WPP14, P. carotovorum subsp. brasiliensis PBR1692, Brenneria sp. EniD312, P. wasabiae WPP163, P. carotovorum subsp. carotovorum PC1 and P. atrosepticum SCRI1043. Second, a Dickeya group comprised Dickeya dadantii 3937, D. dadantii Ech586, D. dadantii Ech703 and D. zeae Ech1591. Third, an Agrobacterium group was composed of Agrobacterium vitis S4 and Agrobacterium radiobacter K84. Finally, a Pseudomonas group comprised P. cichorii JBC1, P. syringae pv. tomato DC300, P. syringae pv. syringae B728a and P. syringae pv. phaseolicola 1448A. In each case, this allowed us to localize the limits between the variable region containing the chitin synthase coding gene and the core surrounding regions shared by all proximate species (syntenic shared blocks). Predictions concerning genomic islands were retrieved from IslandViewer  and predicted prophages were sourced from Prophinder . A G + C content analysis was performed with overlapping sliding windows of 1000 bp at a step of 30 bp using JaDis .
We are grateful to the D.S.I Pôle Calcul (University of Paris VI) for providing computational resources. We also thank Florence Hommais and Joel Pothier for helpful discussions.
This work was supported by the University Claude Bernard Lyon 1, the CNRS and Bayer SAS.
Availability of data and materials
IRG and MC conceived and coordinated the study. IRG, SB, CS and NC acquired the sequence data and made the bioinformatic and phylogenetic analyses. MC and MCS acquired and compared the data on mutants. IRG and SB developped the website. MC, IRG, SG and MB interpreted the data and drafted the paper. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Consent for publication
Ethics approval and consent to participate
- 4.Zaku SG, Emmanuel SA, Aguzue OC, Thomas SA. Extraction and characterization of chitin; a functional biopolymer obtained from scales of common carp fish (Cyprinus carpio L.): A lesser known source. Afr J Food Sci. 2011;5:478–83.Google Scholar
- 55.Anderson MT, Seifert HS. Opportunity and means: horizontal gene transfer from the human host to a bacterial pathogen. mBio. 2011;2:e00005-11-e00005-11.Google Scholar
- 59.Miyashita H, Kuroki Y, Kretsinger RH, Matsushima N. Horizontal gene transfer of plant-specific leucine-rich repeats between plants and bacteria. Nat Sci. 2013;5:580–98.Google Scholar
- 74.Nars A, Lafitte C, Chabaud M, Drouillard S, Mélida H, Danoun S, et al. Aphanomyces euteiches cell wall fractions containing novel glucan-chitosaccharides induce defense genes and nuclear calcium oscillations in the plant host Medicago truncatula. PLoS One. 2013;8:e75039.PubMedPubMedCentralCrossRefGoogle Scholar
- 82.Yang JC, Madupu R, Durkin AS, Ekborg NA, Pedamallu CS, Hostetler JB, et al. The complete genome of Teredinibacter turnerae T7901: an intracellular Endosymbiont of marine wood-boring bivalves (Shipworms). Ahmed N, editor. PLoS ONE. 2009;4:e6085.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.