Fig 1.
The flowchart of the workflow employed in this work for the analysis of sequence diversity and evolution of the GH19 family.
Initially, CAZy and literature screenings were employed to identify the characterized seed sequences. BLAST searches of all seed sequences were conducted and sequences shorter than 120 amino acids were removed. The obtained matches were used to create the GH19ED database, in which the GH19 catalytic domain was annotated with an available profile hidden Markov model (HMM) from the Pfam database (1). Then, catalytic domain sequence networks were obtained by all-vs.-all pairwise aligned sequences and a threshold of 40% identity (2) that permitted to identify subfamilies containing enzymes specialized in one type of activity. Subfamilies were annotated in the database. The properties of the GH19 domain sequence space were also investigated by the analysis of network properties obtained at varying identity thresholds (3). Then, a representative sequence was defined for each subfamily, and an alignment with other characterized members (4) was used for new profile HMMs, to define a standard numbering scheme (SNS) to identify homologous sites within each subfamily. An independent evolutionary conservation analysis with Rate4Site was done for each subfamily (5); by aligning the sequences and structures of the most conserved sites between subfamilies (6), sequence patters specific for each subfamily were identified. Each subfamily was further split into groups from catalytic domain sequence networks, by choosing a 60% identity threshold (7) and these groups were annotated in the database. By functional and structural motifs defined in literature and profile HMMs available for accessory binding modules (8), other annotations were inserted into the database In the final step, GH19 catalytic domain sequences from each group were clustered to select representative centroids (9) to build a large-scale phylogeny, in order to investigate the evolution of structural features, previously annotated and extracted from the database (10). The panels with a dashed outline represent results generated in this study. *Structural information in this study refers to chitinase loops, the endolysin 3-helix peptidoglycan binding bundle and accessory binding modules.
Fig 2.
Protein sequence networks of representative domains of the two bigger clusters containing seed sequences (5067 nodes, 2329 nodes on the left for CHITs, chitinases, and 2738 nodes on the right for ELYSs, endolysins) connected by edges with a sequence identity threshold of 40%.
The prefuse force-directed OpenCL layout with respect to the edge weights was used for network visualization. The domains were extracted by scanning the sequences collected through BLAST searches (using the seed sequences reported in S1 Table as queries) with the GH19 profile HMM PF00182 from Pfam. Nodes are colored according to their annotated taxonomic source. The remaining smaller network clusters are visualized in S5 Fig.
Fig 3.
Protein sequence networks of representative domains of CHITs (A) and ELYSs (B) (1860 nodes for CHITs and 1521 nodes for ELYSs, respectively), connected by edges with a sequence identity threshold of 60%. The prefuse force-directed OpenCL layout with respect to the edge weights was used for network visualization. The domains were extracted by using profile HMMs of CHITs and ELYSs (generated in this study) to scan the sequences in the GH19ED database. Nodes are colored according to their annotated taxonomic source. Seed sequences are highlighted, with a different border if a structure is available in the PDB. Nodes representing characterized “chitinase-like” proteins (CLPs) are also highlighted and presented in S3 Table.
Fig 4.
The most conserved and structurally aligned positions between CHITs and ELYSs (reported in Table 1).
The solvent accessible surface of these positions is plotted onto the reference models of CHIT (A) and ELYS (B) subfamilies (PDB accessions 4j0l of “loopful” plant chitinase from rye seed Secale cereale, and 4ok7 of bacteriophage SPN1S endolysin from Salmonella typhimurium, respectively), represented in cartoon style. In (C) and (D), the same models are rotated by 90° around the vertical axis.
Table 1.
Conserved core shared in CHIT and ELYS subfamilies.
Structurally aligned positions are listed in each row, numbered according to each subfamily-specific standard numbering scheme (see Methods section). Information is provided about the percentage of conserved residues if higher than 5%.
Table 2.
Frequency distributions of amino acids at standard positions used to define sequence patterns specific for CHIT and ELYS subfamilies.
Information is provided about the percentage of conserved residues at each subfamily specific standard numbering scheme position if higher than 5%.
Table 3.
Frequency distributions of loop annotations among CHIT groups.
Names are defined according to Fig 3A (occurrences not displayed if below 5%). h-fam = homologous family (group) name in the GH19ED; ID = group identifier. Binary loop code: ‘0’ = absent; ‘1’ = present; ‘-‘ = undefined.
Fig 5.
Phylogeny of centroids representative of GH19 sequence space, plotted on structural patterns analyzed in this study.
Sequences are indicated with the respective subfamily name (ELYS or CHIT) followed by the group identifier (homologous family in GH19ED database) / number of represented sequences, followed by sequence length of the centroid in parentheses. Sub-clusters according to S15 and S16 Figs are reported as Roman numerals. Sequences representing clusters that contain characterized seed sequences are depicted in bold. *This centroid sequence is a fragment. **A fraction of sequences from group CHIT 6 has longer loops (see Table 3). ***All sequences from group CHIT 7 have a different N-terminal portion in their catalytic domain. HGT = horizontal gene transfer; ABM = accessory binding module; PBM: 3-helix peptidoglycan binding bundle; PG_b_1 = PG_binding_1; CBM = carbohydrate binding module; ND = not defined because too variable within the group (homologous family). The numbers at internal nodes indicate posterior probabilities only if < 1; internal nodes are collapsed if posterior probability is less than 0.5. ELYS 1 = Pseudomonas prophage like; ELYS 2 = Salmonella typhimurium like; ELYS 3 = Salmonella phage PVP-SE1 like; ELYS 4 = Ralstonia phage like; ELYS 5 = Pseudomonas phage OBP like; ELYS 6 = Acinetobacter phage like; ELYS 7 = Mycrocystis phage like; ELYS 8 = Mycobacterium phage like; ELYS 9 to 34 = other putative endolysins from phages and prophages; CHIT 1 = plant “loopful”; 2a-b = plant “loopless”; CHIT 3 = plant CLP with regulatory function; CHIT 4 = Urtica dioica like CLP lectins; CHIT 5 = bacterial “loopless”; CHIT 6–7 = Proteobacteria; CHIT 8 to 12 = other putative bacterial chitinases; CHIT 13–14 = Fungi; CHIT 15 = Metazoa; CHIT 16–17 = Oomycota.
Fig 6.
The positions of residues corresponding to subfamily-specific patterns.
The residues of CHIT (A) and ELYS (B) subfamily-specific sequence patterns identified in this work are labelled and depicted as blue solvent accessible surfaces onto the reference models (PDB accessions 4j0l and 4ok7 for CHIT and ELYS subfamily, respectively), displayed in cartoon style. In (C) and (D), the same models are rotated by 90° around the vertical axis.