Prophages integrating into prophages: A mechanism to accumulate type III secretion effector genes and duplicate Shiga toxin-encoding prophages in Escherichia coli

Abstract

Bacteriophages (or phages) play major roles in the evolution of bacterial pathogens via horizontal gene transfer. Multiple phages are often integrated in a host chromosome as prophages, not only carrying various novel virulence-related genetic determinants into host bacteria but also providing various possibilities for prophage-prophage interactions in bacterial cells. In particular, Escherichia coli strains such as Shiga toxin (Stx)-producing E. coli (STEC) and enteropathogenic E. coli (EPEC) strains have acquired more than 10 prophages (up to 21 prophages), many of which encode type III secretion system (T3SS) effector gene clusters. In these strains, some prophages are present at a single locus in tandem, which is usually interpreted as the integration of phages that use the same attachment (att) sequence. Here, we present phages integrating into T3SS effector gene cluster-associated loci in prophages, which are widely distributed in STEC and EPEC. Some of the phages integrated into prophages are Stx-encoding phages (Stx phages) and have induced the duplication of Stx phages in a single cell. The identified attB sequences in prophage genomes are apparently derived from host chromosomes. In addition, two or three different attB sequences are present in some prophages, which results in the generation of prophage clusters in various complex configurations. These phages integrating into prophages represent a medically and biologically important type of inter-phage interaction that promotes the accumulation of T3SS effector genes in STEC and EPEC, the duplication of Stx phages in STEC, and the conversion of EPEC to STEC and that may be distributed in other types of E. coli strains as well as other prophage-rich bacterial species.

Fig 1. Integration sites of the inducible and packageable duplicated Stx2a phages in two STEC O145:H28 strains.

(A) The duplicated Stx2a phages and their att sequences in strain 112648. The genome structures of three prophages (P08, P09, and P12) are drawn to scale. The att sites of each prophage are indicated by open (attL) or filled (attR) symbols (P08, rhombus; P09, circle; P12, square). The att sequences of the Stx2a phages (P09 and P12) are shown in the inset. (B) The genome structures of two Stx2a phages and a lambda-like phage integrated into ompW (PPompW) in strain 12E129. Sequence homology between the two Stx2a phages is also shown, with their integration sites indicated in parentheses. Homologous regions are indicated by shading with different colors according to sequence identity. The integrase gene of PPompW and the cI gene on each Stx2a phage were targeted by the PCR primers used in Fig 1C. (C) Detection of packaged DNA of the three prophages in the DNase-treated lysates of strain 12E129 with (+) or without (-) MMC treatment. The chromosome backbone (CB) region was amplified as a negative control.

Fig 2. Variation in the prophage content at the ompW, attB in PPompW, and yecE loci in STEC O145:H28.

In the left panel, an ML tree of 64 O145:H28 strains is shown. Completely sequenced strains are indicated in bold (plasmids were not finished for strain 2015C-3125). The tree was constructed based on the recombination-free SNPs (3,277 sites) identified on the conserved chromosome backbone (3,961,936 bp in total length) by RAxML using the GTR gamma substitution model [43]. The reliabilities of the tree’ s internal branches were assessed using bootstrapping with 1,000 pseudoreplicates. Along with the tree, the geographic and ST/clade information of strains, the presence or absence of prophages at three loci (ompW, attB in PPompW, and yecE) and the types of prophages at the attB in PPompW and yecE loci are shown. Prophages sequenced in this study and those in the finished genomes are indicated by asterisks and daggers, respectively. Note that the attB in PPompW sequence is missing from the PPompWs of strains EH2246 and 12E109; a deletion in the latter stain was detected in its draft genome assembly. The bar indicates the mean number of nucleotide substitutions per site. In the right panel, the patterns of the prophage content at the three loci are schematically presented. Strains showing each pattern are also indicated in the left panel by diagrams. For more details about the T3SS effector set, see Fig 3 and main text. Note that we detected recombination between the Stx2a phage at attB in PPompW and a prophage present at the ydfJ locus that induced a large chromosome inversion in strains 10942 and 499. In strain EH1910, an inversion appeared to have occurred by the recombination between PPompW and a prophage at ydfJ.

Fig 3. Phylogenetic positions of E. coli strains carrying PPompW and the genome structures of their EELs associated with 21-bp attB sequence.

In the upper panel, an ML tree of 92 complete genomes of E. coli strains that carry PPompW is shown. The tree was constructed based on 109,927 SNP sites in 2,642 core genes and rooted by cryptic Escherichia clade I strain TW15838 (No. AEKA01000000) used as an outlier. Along with the tree, strain IDs used in this paper (see S5 Table for more details), phylogroups, and the presence (colored) or absence (open) of 21-bp attB sequence in each strain are indicated. The bar indicates the mean number of nucleotide substitutions per site. In the lower panel, the repertoires of T3SS effector genes that were encoded by the effector exchangeable loci (EELs) in the PPompWs containing the 21-bp attB sequence are shown. The genomic structures of EELs are drawn to scale. All effector genes were aligned using BLASTN, and orthologous genes (sequence identity; >90%, coverage; >90%) are indicated by the same color. Genes with over 90% identity but less than 90% coverage and those containing indels and nonsense mutations in the sequence alignment to intact genes are indicated by asterisks.

Fig 4. Prophage clusters that contained prophage carrying potential att sequences in O157:H7 and O177:H25 strains.

The genomic structures of three representative prophage clusters of the 33 clusters found in O157:H7 and that of O177:H25 strains are shown (A, strain FRIK2069; B, strain FRIK944; C, atypical O157:H7 strain PV15-279; D, O177:H25 strain SMN152S1). The identified attB sequences, coding sequences (CDSs) (including pseudogenes), and ISs in each prophage are indicated. T3SS effector genes found in the PPompW EELs (Fig 3) and other effector genes (nleG variants) are distinguished by different colors. In panel C, the attB sequence indicated by an asterisk is truncated by an IS insertion, and integration of an Stx2a phage into the attB-in-PP_1 site is schematically presented. The integrase (int) genes and the nleH genes that have been degraded are indicated by (d). The genome structures of all prophage clusters identified in this analysis are illustrated in S5 Fig.

Fig 5. Locations of the attB-in-PP sequences in prophages and the prophage genome regions homologous to E. coli chromosome regions.

Three loci in the E. coli chromosome showing sequence homology to three identified attB-in-PP sequences and their flanking sequences are shown at the top. The left- and right-end regions of representative prophages that contained the att-in-PP sequences are shown below. Homologous sequences are indicated by the same color. The color used for each attB-in-PP sequence is the same as that used in Figs 4 and S5. See these figures for the details of “PPompWs” and “prophages in PPmlrA/PPydfJ” and S9 Fig for information on the prophages in mlrA, ydfJ, ssrA, and yecE and PPompW. Alignments of the attB-in-PP_1 and attB-in-PP_2 sequences and their flanking sequences with corresponding chromosome sequences are shown in S8 and S9 Figs, respectively.

Fig 6. Summary of the variable phage integration patterns found in this study.