A widespread family of polymorphic toxins encoded by temperate phages

Abstract

Background: Polymorphic toxins (PTs) are multi-domain bacterial exotoxins belonging to distinct families that share common features in terms of domain organization. PTs are found in all major bacterial clades, including many toxic effectors of type V and type VI secretion systems. PTs modulate the dynamics of microbial communities by killing or inhibiting the growth of bacterial competitors lacking protective immunity proteins.

Results: In this work, we identified a novel widespread family of PTs, named MuF toxins, which were exclusively encoded within temperate phages and their prophages. By analyzing the predicted proteomes of 1845 bacteriophages and 2464 bacterial genomes, we found that MuF-containing proteins were frequently part of the DNA packaging module of tailed phages. Interestingly, MuF toxins were abundant in the human gut microbiome.

Conclusions: Our results uncovered the presence of the MuF toxin family in the temperate phages of Firmicutes. The MuF toxin family is likely to play an important role in the ecology of the human microbiota where pathogens and commensal species belonging to the Firmicutes are abundant. We propose that MuF toxins could be delivered by phages into host bacteria and either influence the lysogeny decision or serve as bacterial weapons by inhibiting the growth of competing bacteria.

Keywords: Bacterial genomics; Bacterial toxins; Phages.

Fig. 1

Genetic organization of the head morphogenesis and DNA packaging modules of four phages encoding a MuF protein. SPP1 is a virulent Bacillus subtilis phage of the Siphoviridae family [31]. SPP1 encodes a short MuF1 protein. SF370.1 is a mitomycin C inducible prophage of M1 serotype Streptococcus pyogenes isolate SF370 and belongs to the Siphoviridae family [38]. SF370.1 encodes a MuF toxin of the MuF2 family with a putative nuclease activity. SM1 is a mitomycin C inducible prophage of Streptococcus mitis belonging to the Siphoviridae family [35]. SM1 encodes a MuF toxin of the MuF3 family with a putative nuclease activity. Mycobacteriophage Angel is a temperate phage of Mycobacterium smegmatis encoding a MuF4 protein with a C-terminal extension without predicted domain [69]

Fig. 2

General description of the 1515 MuF proteins detected in this study. a Schematic representation of the main domain architectures of MuF proteins. MuF proteins containing a MuF domain (blue box) without a C-terminal extension (Ct_ext) are called short MuFs. MuF proteins with a Ct_ext either harbor a known toxin domain (red box) or an unknown domain (gray box). b The inner circle represents the proportion of MuF proteins without Ct_ext (in blue), with a toxin domain (in red), or with an unknown domain (in gray). The outer circle represents their distribution within bacteriophages (in light gray) and bacterial genomes (in black). c Taxonomic distribution of bacteria and of hosts of the phages encoding a MuF protein according to the aforementioned categories. There is a significant association of muf genes encoding toxin proteins with Firmicutes compared to muf genes encoding short proteins (p < 0.0001, two-tailed Fisher's exact test), and there is a significant association of muf genes encoding proteins with a C-terminal extension without known domain with Proteobacteria compared to muf genes encoding short proteins (p < 0.0001, two-tailed Fisher's exact test). d Association of known toxin domains (red nodes) with MuF domain families (orange nodes) and with other polymorphic toxin families (blue nodes). Only known toxin domains harbored by at least five MuF proteins were reported in this network that includes 172 MuF toxins. The thickness of the edges is proportional to the abundance of the toxin and MuF domain combinations. The size of the orange (MuF families) and red (toxin domains) nodes is proportional to the number of MuF proteins. Toxin domains are described in Additional file 2: Table S7

Fig. 3

Proportion of genomes encoding MuF per clade. Proportion of genomes encoding at least one MuF protein (MuF+, in red) or without MuF protein (MuF–, in blue) according to the taxonomy of the host’s Caudovirales bacteriophage (left) and of the bacterial genome (right). The total number of genomes in each clade was indicated for both datasets. Only bacterial clades with at least four sequenced genomes were reported (for simplicity). Around 35% of Caudovirales and 25% of bacterial genomes contained at least one MuF protein (MuF+)

Fig. 4

Taxonomic distribution of the 1515 MuF proteins according to their MuF families. a Proportion of MuF protein families and their repartitions within bacteriophage (in light gray) and bacterial (in dark gray) genomes. b Taxonomic distribution of bacteria and of phages’ hosts encoding MuF proteins. Only bacterial clades with at least eight sequenced genomes were reported (for simplicity). The size of the circles is proportional to the number of MuF proteins

Fig. 5

Proportion of bacteriophage genomes encoding MuF per phage family and lifestyle. Proportion of tailed-phage genomes encoding a MuF protein (MuF+, in orange) or none (MuF–, in black) according to the phage family (left) and the phage lifestyle (center). Proportion of short (in blue), with toxin domain (in red), or with unknown domain (in gray) MuF proteins and their repartition within virulent and temperate tailed phages (right). There is a significant association of muf genes with Siphoviridae compared to other families of Caudoviridae (p < 0.0001, two-tailed Fisher's exact test), and there is a significant association of muf genes with temperate compared to virulent phages (p < 0.0001, two-tailed Fisher's exact test). The association of muf toxin genes with temperate phages is not significant (p = 0.055, two-tailed Fisher's exact test) due to the small number of muf toxin genes in the phage dataset

Fig. 6

Proposed model for the role of phage-delivered MuF toxins exhibiting a low (a) or high (b) toxic effect in host bacteria. a MuF toxin with a low toxic effect such as proteases or ADP-ribosyl transferases could influence the lysis-lysogeny decision of the phage with the aim of making the optimal decision at the time of infection. b MuF toxin with a high toxic effect could be involved in inter-bacterial competition. When population A of lysogens is mixed with population B of non-lysogens, phages carrying a MuF toxin could deliver their toxins to susceptible bacteria of population B. The delivery of a toxin with a highly toxic effect, such as a nuclease, into bacteria of population B leads to either a direct inhibition of their growth (dormancy) or killing by induction of the lytic cycle. In both cases, population A of lysogens will outcompete population B of non-lysogens. Thus, bacteria harboring a phage encoding a MuF toxin will have a competitive advantage