Study of 32 new phage tail-like bacteriocins (pyocins) from a clinical collection of Pseudomonas aeruginosa and of their potential use as typing markers and antimicrobial agents

Phage tail-like bacteriocins (PTLBs) are large proteomic structures similar to the tail phages. These structures function in bacterial competition by making pores in the membrane of their competitors. The PTLBs identified in Pseudomonas aeruginosa are known as R-type and F-type pyocins, which have a narrow spectrum of action. Their specificity is determined by the tail fiber and is closely related to the lipopolysaccharide type of the target competitor strain. In this study, the genome sequences of 32 clinical of P. aeruginosa clinical isolates were analysed to investigate the presence of R-type and F-type pyocins, and one was detected in all strains tested. The pyocins were classified into 4 groups on the basis of the tail fiber and also the homology, phylogeny and structure of the cluster components. A relationship was established between these groups and the sequence type and serotype of the strain of origin and finally the killing spectrum of the representative pyocins was determined showing a variable range of activity between 0 and 37.5%. The findings showed that these pyocins could potentially be used for typing of P. aeruginosa clinical isolates, on the basis of their genomic sequence and cluster structure, and also as antimicrobial agents.


Results
Identification and characterization of the phage tail-like bacteriocins. Thirty-two genomic sequences of P. aeruginosa were analysed to search for PTLBs (Table 1). In all of the genomes analysed, at least one cluster corresponding to a pyocin was found between tryptophan operon genes, trpE and trpG ( Table 2;  Table 1S. Supplementary material). Thus, 21 of the strains contained a cluster that corresponded to a unique pyocin corresponding to an R-type pyocin. Dual clusters were identified in the 11 remaining strains. In the P. aeruginosa PAO1 reference strain 5 , one of these clusters corresponded to a R-type pyocin, and the contiguous www.nature.com/scientificreports/ cluster corresponded to an F-type pyocin, both sharing the regulatory and lytic genes, as previously described for P. aeruginosa PAO1 12 .
In order to identify the PTLBs as R or F subtypes, homology analysis of the tail fiber was conducted. In the R-type pyocins, the tail fiber proteins were compared against the reference sequence for each R subtype (R1, R2, R3, R4 and R5); the results revealed that 21 of the proteins belonged to the R5 subtype, while 11 belonged to the R2 subtype, which corresponded to those that were followed by a F-type pyocin (Fig. 1A,B). For the F-type group, the results showed a group of 5 pyocins belonging to the F2 subtype, comprising a R2-F2 pyocin, similar to the PAO1 R2-F2 pyocin 5 , and a group of 6 that were similar to the P. aeruginosa PA14 F-type pyocin 12 , thus giving rise to a pyocin cluster R2-F(PA14) (Fig. 1A,B).
The genomic analysis of the pyocin clusters identified showed a % GC content very similar, between 63.9 and 65.4 ( Table 2). Analysis of the protein sequence of the pyocins revealed some differences in the protein number between the R5-type pyocins (Table 2). Thus, these pyocins were classified in two groups, including a group of 12 pyocins (group A) constituted by an R-type cluster of 14 genes, 4 lytic genes and preceded by 5 regulator genes. The second group (group B) of 8 pyocins differed from group A in the absence of the latter protein belonging to the lytic cassette. In both groups, the cluster was preceded by the regulatory region composed by 5 genes, while in the reference P. aeruginosa PAO1 strain it is composed by 4 genes (Fig. 1B; Table 2) 7 . The R2-F2 pyocins (group C) were composed by 38 genes, which corresponded to an R-type cluster of 14 proteins and a F-type cluster composed by 16 genes, while in P. aeruginosa PAO1 the F-type cluster is formed by 17 genes. In addition, the two clusters share a regulatory region of five genes and a lytic cassette composed by four proteins (Fig. 1B). The pyocins R2-F(PA14) (group D) comprised two consecutive R and F clusters: the R-type comprised 14 proteins and the F-type comprised 13 proteins, unlike P. aeruginosa PAO1 and the group C pyocins, in which the last three proteins are duplicated (Fig. 1B) 7 . www.nature.com/scientificreports/ Homology and phylogenetic analysis of pyocins. The results obtained by the homology studies of the tail fiber specificity genes were confirmed by the homology and phylogenetic analysis of the complete pyocin genomes. The homology analysis showed that the R5-type pyocins were very similar and can be grouped in two blocks corresponding to the established groups A and B, sharing a query cover value of 97-98% and an identity value of 99.35%. The pyocin H52-R5 homology BRIG differed slightly from the two blocks but had similar homology values (Fig. 2). In the case of the R-F pyocin clusters, the homology results showed two blocks of homology, one corresponding to the R2-F2 pyocins (group C) and another corresponding to the R2-F(PA14) pyocin (group D) (Fig. 2). Despite the presence of two groups, the pyocin clusters represented by them were also similar, with a query cover value between 88 and 89% and an identity value of 98%. The phylogenetic study of all the pyocin clusters revealed, as previously observed, that the pyocins identified are divided into four phylogenetic groups. Two closely related clades of the phylogenetic tree were represented by two blocks, one corresponding to the R5-type pyocins included in group A and another also corresponding to an R5-type pyocin grouped in B. Another two closely related clades included one corresponding to group C, which was constituted by the R2-F2 pyocin and group D, constituted by the R2-F(PA14) (Fig. 3).

Identification of pyocins by transmission electron microscopy (TEM).
The purified pyocins were examined by TEM, and the images obtained ( Fig. 4) revealed the presence of two different type of pyocins. One type had a structure similar to a tail of the viral family Myoviridae, corresponding to the R-type pyocins, observed in three conformations: a complete form, a contracted form and an empty sheath 27 . The other type was the F-type, observed as a flexible structure similar to a phage tail of the viral family Syphoviridae. Table 2. Genomic characteristics and subtype of the 32 pyocins (PTBLs) identified in this study. Each pyocin name corresponds to the producer strain. It is indicated the Genbank code, the genome size, % GC, number of ORF and the group to which each was assigned.

Pyocin name
Pyocin subtype Genbank Genomic lenght (pb) % GC ORF Group Killing spectrum of the pyocins. The target range of the pyocins was studied by the spot test technique (Fig. 5). The pyocins included in this analysis were selected by according to ST and serotype of the strain from which they were isolated. The results revealed a great variability in the susceptibility of the strains to the pyocins of the same subtype. In addition, no spots occurred when the target strain belonged to the same ST and serotype as the source strain of the pyocin, except for pyocin 10-58_R2-F(PA14), which produced a spot in strain 3-5,

Discussion
Pyocins are PTLBs produced by P. aeruginosa. Like all PTBLs they are protein complexes with the same structure as phage tails. Like other phage tail particles, such as the type VI secretion systems (T6SS), they also play a role in defence and in interbacterial competition 28 . The genetic and structural similarities between the phage tails and PTBLs initially suggested that the PTBLs evolved as defective phages; however, structural comparison between the T6SS, R-type pyocins and the contractile tail phages suggests evolution from a common ancestor 5,28 . The presence of pyocins in clinical strains of P. aeruginosa seems to be variable. Thus, in a study conducted by Mei et al. 3 , from an analysis of 852 clinical isolates of P. aeruginosa they found that 448 belonged to the R-type pyocin and 300 contained genes of the F-type pyocins. From those included in the R-type pyocins 144 belonged   14 analysed the tracheal aspirates of 61 patients, in search of R-type pyocins, detecting different R subtype pyocins in 77% of the isolates. In another study of 24 isolates from the lungs of CF patients, all isolates were found to contain R-type pyocins, also mainly of the R1 subtype, and it was concluded that this subtype confers a competitive advantage in biofilms, explaining why certain strains displace others in the CF lungs 6 . In the present study, the genome of 32 clinical isolates of P. aeruginosa from several origins, including urinary tract infections, lower respiratory tract infections and intra-abdominal infections, were analysed in search of pyocins, and at least one cluster was found in all isolates. In contrast to the previously mentioned studies, both R-type clusters and F-type clusters were found, resulting in R pyocins and dual R-F pyocins. Analysis of the tail fibers, which determines the specificity (Fig. 1), showed that only R2, R5, F2 and F(PA14) were present in these isolates, with an equal representation of the R5 and R2 subtypes, and a lower representation of F2 than F(PA14) (Fig. 2). In other studies of CF isolates, the R1 subtype was the most commonly identified and in contrast to the results of the present study, R5 was the least well represented subtype 3,6 . In a study conducted by Köhler et al. 14 , pyocins from R1, R2 and R5 were found in the same proportions. Homology modelling and phylogenetic studies of the pyocins identified in these clinical isolates confirmed the grouping of the pyocins and also revealed the great similarity between them. The main difference observed between the R2-type and the R5-type pyocins was in the genomic region corresponding to the tail fiber, and the same was observed between the F2-type and the F(PA14)-type pyocins (Fig. 3). The tail fiber region has been described as being responsible for the specificity of the pyocins, as it acts as an RBP recognizing the target in the LPS of the target bacteria 5,8 .  www.nature.com/scientificreports/ Analysis of the pyocins identified in the genomic sequence of the P. aeruginosa clinical isolates and the clinical ST, serotype and clinical origin established a relationship between the pyocin type and the ST and serotype, but not the clinical origin of the isolate. Although a relationship with both ST and serotype was identified, it seems that there is more variability between the serotype and the presence of pyocins than with the ST. Thus, the association between the pyocin subtype and the ST and serotype can be extended to the groups established in this work beyond the pyocin subtype, as the A and B groups, which both correspond to an R5-type pyocin, were associated with a different ST and serotype. Although the association between serotype and pyocin subtype has long been recognised 14 , to our knowledge the present is the first report of the relationship with ST. The relationship between the serotype and the pyocin subtype 14 has always been established with the R-type cluster, but in the present study a relationship between the serotype and the F-type pyocin was demonstrated, as the group C and D pyocins only differ in the F-type cluster. Both groups were related to two different serotypes, but both share O12 and differ in O5 and O11, possibly as a consequence of the presence of the R-pyocin and the F-pyocin, which in this case would confer a competitive advantage to the strain as they are protected by different pyocins 28 . The observed relationship between the serotype, ST and pyocin from the clinical isolates of P. aeruginosa tested in this study suggests that the pyocins could potentially be used via analysis of the cluster genomic sequences. The pyocin-serotype association has been used for bacterial typing, but unlike in the present study, the typing was based on the killing activity of a pyocin from an unknown isolate over a collection of indicator strains; this method was abandoned, as it is slow and laborious, and was substituted by molecular methods [17][18][19] . Currently, thanks to the genome analysis, the study of the PTBLs can complement the traditional typing methods employed in the clinical laboratories.
The serotype in P. aeruginosa is determined by the O-antigen, which is a B-band repeating unit of LPS considered a virulence factor 14 . It is known that the pyocin tail fiber proteins recognize the LPS of the competitor strains but do not recognize their own LPS as a target, so they cannot lyse those strains with the same serotype. In a study carried out in 2010, Köhler et al. 14 deleted different genes responsible for the synthesis of the O-antigen and determined which LPS residues act as receptors for R1-type, R2-type and R5-type pyocins. When the 32 clinical isolates were tested against the 32 isolated pyocins, none were able to lyse the source strain or those strains with the same serotype or ST. As previously described, variability in the susceptibility to the pyocins was observed between those isolates that shared ST and/or serotype and the same type of pyocin, probably due to the frequently observed mutations in the LPS genes in CF, affecting recognition by the RBP 3,29,30 . Pyocins have www.nature.com/scientificreports/ been considered potential alternatives to antibiotics, and several studies have demonstrated antimicrobial activity of pyocins alone or in combination with other antimicrobials and those sharing the LPS as target 13,22,23,31,32 .
In this study, 32 pyocins belonging to the R5-type, R2-F2 type and R2-F(PA14) type were identified. Homology and phylogenetic analysis established four groups of pyocins (A, B, C and D), each of which was found to be related to the serotype and ST of the source strain. Pyocins are therefore good candidate markers for typing strains of P. aeruginosa by analysis of the tail fiber protein of the pyocin. We also observed that they could be used to type R5-type pyocins, by the number and distribution of the genes comprising the pyocin cluster. These pyocins also displayed potential antimicrobial activity as they were able to lyse some of the clinical isolates tested, particularly the 9-86_pyoR2-F(PA14) pyocin, which exhibited the highest range of activity.

Material and methods
Strains and culture conditions. Thirty-two clinical strains of P. aeruginosa isolated in Portuguese and Spanish hospitals within the framework of two multicentre studies, STEP in Portugal and SUPERIOR in Spain 33 , were used in the study. The software mlst (v2.16.1) (https:// github. com/ tseem ann/ mlst) was used for the in silico MLST assignment 33 . Serogroups based on the O-specific antigen (OSA) gene cluster sequences were determined using Blastn tool (v2.9.0+) (http:// blast. ncbi. nlm. nih. gov/ Blast. cgi) and the OSA database 33 . The origin, ST and serotype of the isolates are shown in Table 1. All strains were grown in Luria-Bertani broth (LB) medium (0.5% NaCl, 0.5% yeast extract, 1% tryptone) at 37 °C and 180 rpm. LB was supplemented with agar (1.5%) when necessary.
Identification of pyocin clusters in silico. The genomes of 32 clinical isolates P. aeruginosa (NCBI Bio-Project: PRJNA629475) were analysed to search for pyocins. For this purpose, the anthranilate flanking genes, trpD and trpE, were identified in the search for the genes that typically compose the pyocin clusters, which have been reported to be included between these genes 5 .
When the pyocin cluster sequences were distributed in several contigs, they were then compared by homology and assembled using BLASTn and Vector NTI Advance™ 11 (Invitrogen) programs. The complete sequences of the pyocin clusters were annotated by RAST 34 , HMMER (http:// hmmer. org) and HHPRED 35 . The pyocin cluster R2-F2 of P. aeruginosa PAO1 was used as a reference sequence, which corresponds to the region between the genes PAO610-PAO648 (Genbank: AE004091.2-AE004091.2) from the Pseudomonas aeruginosa database 36 .
All the nucleotide sequence data reported are available in the Third Party Annotation Section of the DDBJ/ ENA/GenBank databases under the accession numbers TPA: BK062614-BK062645 (Table 2).
Pyocin cluster type identification. The pyocin cluster types were assigned by homology with the tail fiber corresponding to the protein of the reference strain P. aeruginosa PAO1: PAO620 for the R-type pyocins and by PAO646 for the F-type pyocins 12 .
Homology and phylogenetic analysis. The sequences obtained for the different pyocin clusters were analysed in order to study their homology. The studies were carried out using the Easyfig 2.2.5 37 and BRIG 0.95 38 tools with the tBLASTx option. The Query Cover and Identity values were analysed with BLASTn.
The sequences were aligned and a phylogenetic tree was constructed with the bioinformatic software Geneious Prime (Dotmatics).
Extraction and concentration of phage tail-like bacteriocins. The selected strains used as sources of pyocins were cultured overnight in LB broth at 37 °C. The culture was then diluted 1:100 in LB and incubated at 37 °C and 180 rpm. Once the optical density measured at a wavelength of 600 nm (OD 600nm ) of 0.4 was reached, 10 µg/ml of mitomycin C (Sigma-Aldrich) was added and the culture was incubated until it turned clear. The lysed cultures were centrifuged at 4000 rpm for 10 min, and the supernatant with the pyocins was recovered and incubated with 1% chloroform for 30 min. Finally, the supernatant with the pyocins was filtered through a 45 µm filter (FILTER-LAB®PES Syringe filter).
For concentration, the pyocins were precipitated with polyethylene glycol (PEG). The pyocin solution was precipitated overnight at 4 °C with 10% PEG and 0.5 M NaCl. The pyocins were collected by centrifugation for 15 min at 11,000 rpm and 4 °C. The supernatant was discarded and the pellet suspended in SM buffer (0.1 M NaCl, 1 mM MgSO4, 0.2 M Tris-HCl, pH 7.5), to obtain a tenfold concentration. Finally, a 1:1 volume of chloroform was added and incubated with gentle shaking for 20 min, and the phases were then separated by centrifugation for 10 min at 4000 rpm. The aqueous phase with the pyocin suspension was recovered and stored at 4 °C until use.
Pyocin transmission electron microscopy (TEM). The pyocin solutions were fixed in a grid and negatively stained in 1% aqueous uranil acetate for 5 min and examined in a transmission electron microscope JEOL JEM-1011.
Pyocin killing spectrum. In order to determine the target range of each isolated pyocin, each was tested against the 32 strains from which they were isolated ( Table 1). The killing activity was assayed by spot test 39  www.nature.com/scientificreports/ Briefly, double agar layer plates were prepared with the putative host strain mixed with the soft upper agar layer (0.4% agar). Once solidified, a drop (10 μl) of the pyocin solution was deposited on top of the agar layer, and the plates were incubated at 37 °C for 20 h.
To differentiate the R-type and F-type pyocins from the S-type pyocins, proteinase K was added to the plates, as R and F-type pyocins are protease resistant and S-type is protease sensitive 12 . In order to differentiate the pyocins from prophages induced with mitomycin C, serial dilutions were spotted on agar plates. When no individual plaques were observed at the higher dilutions, the presence of a spot was considered to be the result of the killing activity of the pyocin.

Data availability
All data generated or analysed during this study are included in this published article [and its supplementary information files].