In silico Identification of Novel Toxin Homologs and Associated Mobile Genetic Elements in Clostridium perfringens

Abstract

Clostridium perfringens causes a wide range of diseases in a variety of hosts, due to the production of a diverse set of toxins and extracellular enzymes. The C. perfringens toxins play an important role in pathogenesis, such that the presence and absence of the toxins is used as a typing scheme for the species. In recent years, several new toxins have been discovered that have been shown to be essential or highly correlated to diseases; these include binary enterotoxin (BecAB), NetB and NetF. In the current study, genome sequence analysis of C. perfringens isolates from diverse sources revealed several putative novel toxin homologs, some of which appeared to be associated with potential mobile genetic elements, including transposons and plasmids. Four novel toxin homologs encoding proteins related to the pore-forming Leukocidin/Hemolysin family were found in type A and G isolates. Two novel toxin homologs encoding proteins related to the epsilon aerolysin-like toxin family were identified in Type A and F isolates from humans, contaminated food and turkeys. A novel set of proteins related to clostridial binary toxins was also identified. While phenotypic characterisation is required before any of these homologs can be established as functional toxins, the in silico identification of these novel homologs on mobile genetic elements suggests the potential toxin reservoir of C. perfringens may be much larger than previously thought.

Keywords: Clostridium perfringens; binary toxin; epsilon; hemolysin; leukotoxin; pCP13; pCW3; plasmid; toxin.

Figure 1

(A) Schematic showing key features of the C. perfringens leukocidin domain containing proteins. The purple region represents the signal peptide and the blue region the PF07968 leukocidin/hemolysin domain. Numbers marked correspond to amino acid positions of the start and end of the features with the range showing the variation between the different protein sequences (n = 13). (B) A maximum likelihood tree based on alignment of novel toxin homologs (marked with *) against protein sequences of the C. perfringens leukocidin domain containing proteins and representative sequences of CctA from Clostridium chauvei the Staphylococcus aureus hemolysin, leukotoxin components F and D. Protein sequences were aligned using clustal omega, and maximum likelihood was implemented in IQtree. The tree was inferred using the LG+F+G4 model and rapid bootstrapping -bb 2000; bootstrap support is shown at the nodes. Scale bar indicates the number of changes per site. Heatmap shows percent identity matrix of protein alignments, colours correspond to the following percent identity: dark red, 80–100%; light red, 60–79%; orange, 40–59%; bright yellow, 30–39%; pale yellow, 20–29% and white, <19%.

Figure 2

Schematic representation of the genomic location of delta toxin (cpd) in NCTC3182 and leukotoxin domain protein A (ldpA) from strain 16SBCL572 and delta-like protein (dlpA) from strain T84. Chromosomal regions are coloured blue, the unique regions grey, and toxin genes are coloured in red. Genbank accession numbers for sequences are as follows: T84 dlpA region (MK285064), NCTC3182 cpd region (MK285058), and 16SBCL571 ldpA region (MK285056).

Figure 3

Schematic representation showing the comparative alignment of sequenced C. perfringens plasmids compared to plasmid contigs containing toxin homologs, made in EasyFig v2.2.2. (A) Shown are a sequence alignment of pCP13 (cpb2-con) compared to the epsilon domain protein containing plasmids pCPNY83906550-1 (edpA) and pCPT1 (edpB). (B) Shown are sequence alignments of pCW3 (Tet), pCP16SBCL1142-1 (LdpC), pCPT6-1 (ldpB, cpb2-atyp), pCP16SBCL648-1 (ldpB, Tet) and pCPT84-1 (ilpA/ilpB). The ORFs in the conserved backbone for pCP13-like plasmid are depicted as light blue arrows. The ORFs in the conserved backbone for pCW3-lke plasmids are depicted as dark blue arrows. Virulence factors and toxin homologs are labelled and other open reading frames are shown as red arrows. Light grey arrows represent open reading frames that are unique to that plasmid, * denotes plasmids containing a toxin homolog. Genbank accession numbers for plasmid sequences are DQ366035 for pCW3, AP003515 for pCP13, MK285071 for pCPNY83906550-1, MK285059 for pCPT1, MK285071 for pCP16SBCL1142-1, MK285060 for pCPT6-1, MK285061 for pCP16SBCL648-1 and MK285057 for pCPT84-1.

Figure 4

(A) Schematic showing key features of the C. perfringens epsilon aerolysin domain family proteins, as characterised in Pfam. The PF03318 epsilon toxin ETX/Bacillus mosquitocidal toxin MTX2 domain is coloured green, the signal peptide is coloured purple and amino acid positions of the domains are marked. Numbers marked correspond to amino acid positions of the start and end of the features with the range showing the variation between the different protein sequences. (B) A maximum likelihood tree based on alignment of novel toxin homologs (marked with *) against protein sequences of the C. perfringens epsilon domain containing proteins and representative sequences from B. lacterosporus (HOU5B4, HOUDJ3) and epsilon toxin from C. botulinum. Protein sequences were aligned using clustal omega, and maximum likelihood was implemented in IQtree. The tree was inferred using the LG+F+G4 model and rapid bootstrapping -bb 2000; bootstrap support is shown at the nodes. Scale bar indicates the number of changes per site. A heatmap showing percent identity matrix of protein alignments is also shown, and colours correspond to the following percent identity: dark red, 80–100%; light red, 60–79%; orange, 40–59%; bright yellow, 30–39%; pale yellow, 20–29% and white, <19%.

Figure 5

(A) Schematic showing key features of the C. perfringens functional domains of the iota binary toxins family of proteins as characterised in Pfam. The PF03496 regions are ADP-ribosytransferase toxin A domains and are coloured green; PA14 is coloured blue; PF03295, PF17475 and PF17476, the toxB domains, are coloured red and the signal peptide is coloured purple. Numbers marked correspond to amino acid positions of the start and end of the features with the range showing the variation between the different protein sequences. (B,(C)) Maximum likelihood trees based on alignment of novel toxin homologs of each toxin component (marked with *) against protein sequences of the C. perfringens binary toxin proteins iota and Bec and representative sequences from C. botunilim C2, C. difficile binary toxin and C. spiroforme binary toxin. Protein sequences were aligned using clustal omega, and maximum likelihood was implemented in IQtree. The tree was inferred using the LG+F+G4 model and rapid bootstrapping -bb 2000; bootstrap support is shown at the nodes. Scale bar indicates the number of changes per site. Heatmap showing percent identity matrix of protein alignments, colours correspond to the following percent identity: dark red, 80–100%; light red, 60–79%; orange, 40–59%; bright yellow, 30–39%; pale yellow, 20–29% and white, <19%.