Pseudomonas aeruginosa produces three types of bacteriocins: R-, F- and S-type pyocins. The S-type pyocin is a colicin-like protein, whereas the R-type pyocin resembles a contractile but non-flexible tail structure of bacteriophage, and the F-type a flexible but non-contractile one. As genetically related phages exist for each type, these pyocins have been thought to be variations of defective phage. In the present study, the nucleotide sequence of R2 pyocin genes, along with those for F2 pyocin, which are located downstream of the R2 gene cluster on the chromosome of P. aeruginosa PAO1, was analysed in order to elucidate the relationship between the pyocins and bacteriophages. The results clearly demonstrated that the R-type pyocin is derived from a common ancestral origin with P2 phage and the F-type from λ phage. This notion was supported by identification of a lysis gene cassette similar to those for bacteriophages. The gene organization of the R2 and F2 pyocin gene cluster, however, suggested that both pyocins are not simple defective phages, but are phage tails that have been evolutionarily specialized as bacteriocins. A systematic polymerase chain reaction (PCR) analysis of P. aeruginosa strains that produce various subtypes of R and F pyocins revealed that the genes for every subtype are located between trpE and trpG in the same or very similar gene organization as for R2 and F2 pyocins, but with alterations in genes that determine the receptor specificity.
Many bacteria produce bactericidal compounds referred to as bacteriocins, which are generally effective only against the same or closely related species. Bacteriocins produced by Pseudomonas aeruginosa are called pyocins. Three different types of pyocins have been identified: R-type, F-type and S-type. Although bacteriocins are often encoded on plasmids, pyocin genes are located on the chromosome of P. aeruginosa (Kageyama, 1975).
S-type pyocins are colicin-like and protease-sensitive simple proteins, which are further separated into four types: S1, S2, S3 and AP41 (Sano et al., 1993; Sano and Kageyama, 1993; Duport et al., 1995). Although they are distinguished by their different killing spectra, all possess DNase activity. Their C-terminal halves responsible for the activity are highly homologous to the DNase domain of colicin E2. Their immunity proteins are also highly homologous to that of E2 colicin. Thus, a close evolutionary relationship has been suggested between S-type pyocins and E2 group colicins (Sano et al., 1993; Kageyama et al., 1996; Riley, 1998). Recently, genes for two additional S-type pyocins, S4 and S5, have been identified on the chromosome of P. aeruginosa strain PAO1 (Parret and De Mot, 2000).
R-type pyocins resemble inflexible and contractile tails of bacteriophages and are further classified into five groups: R1, R2, R3, R4 and R5 (Kageyama, 1975). They are similar to each other in their structural and serological properties, but different in receptor specificity. The tail fibre protein, an apparatus for binding to the receptor of a sensitive bacterial strain, has been proposed to account for the main difference (Ohsumi et al., 1980). R-type pyocins, when used to challenge sensitive cells, cause depolarization of the cytoplasmic membrane and inhibit active transport (Uratani and Hoshino, 1984). Contraction of the tail-like structure is necessary for this bactericidal action (Shinomiya et al., 1975).
F-type pyocins also resemble phage tails, but flexible and non-contractile ones. Pyocin 28 (Takeya et al., 1969) was the first to be reported as a variation of F-type pyocin, and then 430F particle (Govan, 1974), F1 and F2 (Kuroda and Kageyama, 1979) and F3 (Kuroda and Kageyama, 1981) were reported. Among them, F1, F2 and F3 have been analysed in detail. They are similar in structure and serological properties, but are again different in receptor specificities.
As the structure of R-type and F-type pyocins resembled phage tail structures, we searched for phages that could be the origins of these pyocins. Several bacteriophages called R or F pyocin-related phages were isolated: those neutralized by anti-R sera, PS3 (Ito and Kageyama, 1970) and PS17 (Kageyama et al., 1979), and that neutralized by anti-F sera, KF1 (Kuroda et al., 1983). Cytotoxin-converting phages of P. aeruginosa were also found to be R pyocin-related phages (Hayashi et al., 1990; 1994). Genetic evidence has also indicated a relatedness of these phages to R pyocins (Shinomiya and Ina, 1989; Hayashi et al., 1994). Headless mutants of phage PS17 have been demonstrated to kill sensitive cells as bacteriocins do, although the bactericidal activities were considerably lower than that of R-type pyocin (Shinomiya and Shiga, 1979). Moreover, it has been shown that component exchanges, known as phenotypic mixing, occurred between R2 pyocin and PS17 phage (Shinomiya, 1984). It has been suggested that R-type and F-type pyocins are defective phages derived from different phage types (Kageyama, 1985).
R-type pyocin genes (at least R1, R2 and R3) are located between trpE and trpGCD on the chromosome (Shinomiya et al., 1983a,b; Kageyama, 1985). In strain PAO1 (R2+, F2+), F2 pyocin genes are located between the R2 pyocin gene cluster and trpGCD (Shinomiya et al., 1983b; Matsui et al., 1993). As for R2 pyocin genes of PAO1, 14 genes (prtA–N, prtR) were identified on a 13 kb segment between trpE and trpGCD (Shinomiya et al., 1983a,b; Matsui et al., 1993). Among them, prtR and prtN were shown to regulate the expression of pyocin genes (Matsui et al., 1993). The prtN gene encodes a transcriptional activator for all these pyocin types: R, F and S. A sequence motif called the P-box has been proposed to be the binding site for this protein. The PrtR protein is a repressor for the prtN gene and is inactivated in the presence of activated RecA protein. This regulatory system resembles those of temperate bacteriophages, and UV irradiation or mitomycin treatment induces the expression of pyocin genes. An actual sequence similarity was found between PrtR and phage repressors, although none between the PrtN protein and phage regulators.
Recently, we have determined the whole nucleotide sequence of a cytotoxin-converting phage, φCTX, and found that the genome of the phage was highly homologous to that of the temperate coliphage P2 (Nakayama et al., 1999). This indicated that φCTX belongs to the P2 phage family and suggested that R-type pyocins are genetically related to P2 phage. In the present study, to elucidate the relationship of pyocin and bacteriophage, the nucleotide sequence of the region encoding the R2 pyocin of P. aeruginosa PAO1, along with that for the F2 pyocin genes located just downstream of the R2 genes, were analysed. The results clearly indicated that the R-type pyocin is derived from a common ancestral origin with the P2 phage family, and the F-type pyocin from the λ phage family. The presence of a lysis gene cassette similar to those for bacteriophages also supported this notion. Both pyocins, however, appear to be phage tails especially evolved as bacteriocins rather than simple defective phages. Furthermore, we performed a systematic polymerase chain reaction (PCR) analysis (designated ‘PCR scanning’) of the trpE–trpG regions of P. aeruginosa strains that produce various subtypes of R and F pyocins. The results suggested that the genes for every subtype are located between trpE and trpG on the P. aeruginosa chromosome in the same or very similar gene organization as those for R2 and F2 pyocins, but that the genes determining the receptor specificity have been altered.
Results and discussion
Nucleotide sequence of the R2 and F2 pyocin locus on the chromosome of strain PAO1
The R2 and F2 pyocin gene cluster of PAO1 was cloned using R-prime plasmids, and the R2 pyocin gene cluster and a part of the F2 pyocin gene cluster were localized on two contiguous HindIII fragments of 8 kb and 16 kb (Shinomiya et al., 1983b; Fig. 1). Both fragments were subcloned into a broad-host-range cloning vector, pKT230 (Bagdasarian et al., 1983). The fragments were purified from the agarose gel and subjected to random shotgun sequencing. The precise length of the 8 kb fragment was 7859 bp and that of the 16 kb fragment 15 500 bp, giving a total of 23 355 bp (Fig. 1; DDBJ accession no. AB030825).
By the time of our determination, a large part of the genome sequence of PAO1 was available through the P. aeruginosa genome project. Our sequence was found to coincide completely with the project data (most recently updated on 15 December, 1999). As it was known that the rest of the F2 pyocin genes were located downstream of the 16 kb HindIII fragment but upstream of trpGCD (Shinomiya et al., 1983b; K. Kuroda et al., unpublished), we searched the PAO1 genome sequence for trpGCD (Essar et al., 1990) and found the genes ≈ 10 kb downstream of the 16 kb HindIII fragment. We thus decided to include the genome project data for this region in our analysis (Fig. 1).
In the 34.4 kb segment analysed, 47 open reading frames (ORFs) were identified (tentatively named PRF0 to PRF44; Fig. 1 and Table 1). Among these, PRF2 and PRF44 correspond to trpE and trpG, respectively, and PRF3 and PRF4 represent the genes previously identified as repressor and activator genes for pyocin production, prtN and prtR.
Table 1. The list of ORFs of known, suggested or hypothetical genes in the R2/F2 pyocin region of PAO1.
Y. pestis pMT1 (Y1051), phage N15 (gp25), phage HK97 (Stf)
Anthranilate synthetase component II
Nucleotide sequence of the trpE–F2 pyocin region of strain PML14
PML14 is a P. aeruginosa strain that produces only F2 pyocin. An 11 kb HindIII fragment containing the trpE gene and part of the F2 gene cluster of the strain was cloned previously, and it was suggested from heteroduplex formation analysis of this 11 kb fragment with the above-mentioned 8 kb and 16 kb HindIII fragments of PAO1 that an R2 pyocin-specific region of ≈ 13 kb existed between trpE and the F2 pyocin gene cluster in PAO1 (Shinomiya et al., 1983b). To confirm this, we prepared a pair of PCR primers based on the determined PAO1 sequence and performed PCR analysis of the PML14 chromosome. Primers were synthesized according to the sequences of trpE and PRF31, the latter of which was assumed to be part of the F2 pyocin gene cluster from the homology to a tail gene (gene H) of λ phage (see below). Using these primers, an ≈ 6.5 kb DNA fragment was amplified from both the cloned 11 kb HindIII fragment and the genomic DNA of PML14. As the distance between the primers on the PAO1 genome was 18 876 bp, this result was roughly consistent with the previously estimated size of the R2 pyocin-specific region. We therefore determined the nucleotide sequence of this 6.5 kb fragment to identify the boundary of the R2 pyocin-specific region. The entire sequence determined was 6744 bp (Fig. 1; DDBJ accession no. AB030826).
Comparison of PML14 and PAO1 sequences revealed the presence of an R2 pyocin-specific region of 12 156 bp that was found only in PAO1 (Fig. 1). The region encompassed sequence from the 3′ end of PRF9 to PRF23. In PML14, a novel 26 bp sequence was found in the relevant region, which appeared to be replaced by the R2 pyocin-specific region on the PAO1 genome. Replacement by the R2-specific region resulted in different amino acid sequences of the C-terminal region of PRF9 between the two strains (Fig. 1). In the region common to both strains, 56 base changes and one single base deletion were found. Although most of them were either located in non-coding regions or were silent changes, a base deletion and a base change that took place in the coding regions for PRF8 and PRF27 introduced a frameshift and a premature termination codon in each ORF of PML14. Considering that, in general, genes encoding functional proteins are evolutionarily conserved, it is more likely that PRF8 and PRF27 of PAO1 do not encode functional proteins. Rather, the regions corresponding to the two ORFs may serve as regulatory regions for pyocin gene expression, as P-box-like sequences, possible binding sites for the transcriptional activator PrtN, were present in both regions (data not shown).
A homology search was performed for the 26 bp sequence specific to PML14, but no sequence with significant similarity was retrieved. Two kinds of repeated sequences, AGGAGGA and TGAAACtgCtCGaag, were found around the boundaries of the R2 pyocin-specific region, although their roles in the formation of the genetic structures observed are yet to be elucidated.
Genes in the R2 and F2 pyocin gene clusters
Every identified ORF was examined for homology to known genes using the blast and fasta programs (Pearson and Lipman, 1988; Altschul et al., 1997). As expected, genes within the R2 pyocin-specific region showed extensive homology to the tail genes of the P2 phage family, such as P2, 186 and φCTX. In addition, genes within the F2 pyocin region were found to be highly homologous to the tail genes of λ and λ-related phages (Table 1, Fig. 2).
R2 pyocin genes. Among 16 ORFs identified in the R2 pyocin-specific region, 12 ORFs showed significant similarities to the tail genes of P2 and φCTX, as well as to those of 186, a close relative of P2 (Portelli et al., 1998). The arrangement of the genes was the same as that of these phages, except for PRF22. X genes in P2 and φCTX, homologues of PRF22, are located on the other side of the lysis gene cluster, apart from the other tail genes. In the R2 gene cluster, however, PRF22 was located at a site corresponding to the U-D intergenic region in P2.
Although PRF16 showed no homology to P2 genes, it showed marked homology to φCTX ORF21 (43.1% identity throughout the entire length), which has been suggested to be functionally equivalent to the P2 G (tail fibre assembly) gene (Nakayama et al., 1999). As for PRF15, which is a homologue of the P2 H (tail fibre) gene, homology to P2 gpH was detected only at the N-terminal portion (≈170 amino acid residues), whereas it was similar to gpHφCTX for almost its full length (Fig. 3). In many double-stranded DNA phages, including the P2 phage family, N-terminal parts of the tail fibre proteins are considered to be responsible for connection to base plates, and the C-terminal parts for binding to host receptors. It has also been demonstrated that the latter consist of several modules (Haggård-Ljungquist et al., 1992; Sandmeier, 1994). Comparison of PRF16 with gpHφCTX revealed that PRF16 lacked one of the modules that repeated 2.5 times in gpHφCTX and that no similarity was detected at the C-terminal region (Fig. 3). These features are consistent with the following findings: (i) the baseplate proteins of P2, φCTX and R2 pyocin (gpJ, gpJφCTX, PRF13) closely resemble each other; (ii) although the receptors for the phages and R2 pyocin are core oligosaccharides of lipopolysaccharide (LPS) (Bertani and Bertani, 1971; Meadow and Wells, 1978; Yokota et al., 1994), P. aeruginosa is insensitive to P2, and E. coli to φCTX or R2 pyocin (Nakayama et al., 1999); and (iii) the binding sites in the LPS core of P. aeruginosa differ between φCTX and R2 pyocin (Yokota et al., 1994).
No similarity to P2 or φCTX genes was detected for PRFs 19, 19.5 and 20. However, they were presumed to correspond to the E, E′ and T genes of P2, or to the EφCTX, E′φCTX and TφCTX genes of φCTX, as the locations and sizes resemble each other. Both gpT and gpTφCTX are proposed to be the tail length determinator proteins and contain extensive regions of α-helix (Nakayama et al., 1999; G. Christie, personal communication). Prediction of secondary structure also indicated PRF 20 to be rich in α-helix (Chou and Fasman, 1978). A −1 translational frameshift takes place just distal to the poly(T) track in the 3′-terminal region of the P2 E gene, resulting in the synthesis of a fusion protein, gpE–E′, in P2 (G. Christie, personal communication). EφCTX also contains a poly(T) track in the relevant region, suggesting a similar translational frameshift in the EφCTX gene (Nakayama et al., 1999). However, no candidate ‘slippery’ sequence such as the poly(T) track was found in the 3′-terminal region of PRF 19. Furthermore, +1 frameshift is necessary to fuse PRF 19 to PRF 19.5. Thus, some mechanism different from those for the E and E′ genes, such as ‘hopping’, would be required for the R2 genes to synthesize a fusion protein.
No ORFs homologous to the P2 R or S genes were present in the R2 pyocin gene cluster. These P2 genes are responsible for tail completion and connection of the head and tail structures (Linderoth et al., 1994). At the region corresponding to the R and S genes (upstream of the V gene in P2), PRF 10 was found instead. Provided that tail assembly in the headless R2 pyocin is completed in a different manner from the normal phages, P2 and φCTX, PRF 10 could be the gene responsible for tail completion in R2 pyocin.
In P2 and φCTX, termination and initiation codons overlap between genes V and W, and between the VφCTX and WφCTX genes. In contrast, a gap of 152 bp was present between PRF 11 and PRF 12, the V and W homologues of R2 pyocin. The sequence of this intergenic region was also analysed, and a small ORF (PRF 11.5) that encodes a putative gene product of 23 amino acid residues was identified. The N-terminal half of PRF 11.5 showed significant similarity to the C-terminal region of gpV and gpVφCTX. Moreover, the termination codon for PRF 11.5 and the initiation codon for PRF 12 were found to overlap as in V and W. The V gene homologue of R2 pyocin was therefore assumed to have split into two genes. At present, it is not known whether PRF 11.5 encodes a functional protein or if it is required only for the translation of PRF 12.
When purified R pyocin particles were analysed by SDS–PAGE, some 20 different protein bands were detected as structural proteins (Shinomiya, 1972). Band 6, of approximately 36 kDa, and band 16, of 19 kDa, were the tail sheath and tail tube of R2 pyocin respectively. These molecular weights coincide roughly with those of PRFs 17 and 18 (41.2 kDa and 18.0 kDa), which were predicted, from homology with the P2 genes, to be the tail sheath and tail tube proteins of R2 pyocin. In R1 pyocin, the protein profile of which was mostly the same as R2, band 2 (71 kDa) and band 9 (31 kDa) were demonstrated to be contained in the isolated tail fibre (Kumazaki et al., 1982). Based on the homology with P2 genes, the gene products of PRFs 15 and 13 are predicted to be the tail fibre and base plate respectively. The predicted molecular weights of these gene products were 72.3 kDa and 31.9 kDa respectively. The molecule isolated by Kumazaki et al. (1982) was therefore concluded to be a complex of the tail fibre protein (PRF 15 homologue of R1) and the base plate protein (PRF 13 homologue of R1). N-terminal amino acid sequences of the protein in band 2 from pyocins R1 and R2 have actually been found to be the same as that predicted for PRF 15 (T. Kumazaki, personal communication).
F2 pyocin genes. F2 pyocin genes were previously localized between R2 pyocin genes and trpGCD (Shinomiya et al., 1983b; Matsui et al., 1993). As three ORFs (PRFs 24, 25 and 26) just downstream of the R2 pyocin-specific region were considered to be related to lysis function (see below), it is most likely that PRF 28 and those thereafter are responsible for F2 pyocin synthesis. Eight ORFs (PRF 28 and PRFs 30–36) within this region indeed showed marked similarities to the tail genes of λ phage or its related phages, such as HK022, HK97 and N15 (Table 1). Moreover, the order of these genes was completely conserved (Fig. 2). This strongly suggested that the F2 pyocin is genetically related to the λ phage family, the tail structures of which are also similar to that of F-type pyocins.
In the previous SDS–PAGE analysis of purified F2 pyocin, 10 protein bands were detected as the structural proteins (Kuroda and Kageyama, 1981). The molecular weights were estimated to be 135 kDa (band 1), 63 kDa (band 2), 52 kDa (band 3), 36 and 36.5 kDa (bands 4 and 4*), 25.5 kDa (band 5), 19.5 kDa (band 6), 18 kDa (band 6*), 13.5 kDa (band 7*) and 12.5 kDa (band 8*). Of these, band 6 is the major component of F pyocin. The calculated molecular weight of PRF 28, a homologue of major tail protein of the λ phage family, coincided roughly with that of band 6 (19.5 kDa).
No significant homology was found between PRF 29 and genes of the λ phage family. However, the location and size suggested that PRF 29 corresponds to gene G and its homologues from the λ phage family (Fig. 2). The λG gene encodes not only a 16 kDa protein (gpG) but also, jointly with the overlapping downstream T gene, a 31 kDa protein (gpG-T). gpG-T is synthesized as the result of a programmed translational frameshift, which occurs at the sequence GGGAAAG in the 3′ region of gene G (Levin et al., 1993). Interestingly, the sequence of the 3′ end of PRF 29 was found to be AAAAAACTGA, which was followed by a stable stem–loop structure. The presence of such a ‘slippery’ sequence and a downstream ‘stimulator’-like structure (Atkins et al., 1990) suggests that −1 frameshifting occurs at the 3′ end of PRF 29 as well, which results in the synthesis of a PRF 29–30 fusion protein. Although PRF 30 was preceded by a recognizable putative Shine–Dalgarno (SD) sequence, the distance between the SD and the start codon (GTG) was atypical (14 bp). F2 pyocin appeared to lack the genes corresponding to the λZ and U genes, which are responsible for tail completion and connection of the tail and head structures.
In the region corresponding to the cryptic genes for side tail fibre synthesis on the λ phage genome, seven ORFs were identified (Fig. 2). Six of these appear to consist of a duplicated gene cassette containing three ORFs. This conclusion is based on similarities in size and marked sequence homology between PRFs 38 and 41, PRFs 39 and 42, as well as PRFs 40 and 43 (Fig. 4). Band 4 of F-type pyocin was identified as a component of the tail fibre that was responsible for binding to sensitive cells (Kuroda and Kageyama, 1981; Kuroda and Kagiyama, 1983). Unlike F1 and F3 pyocins, band 4 of F2 pyocin was found to be accompanied by an additional protein band (band 4*), which was of equal density but slightly different in size (Kuroda and Kageyama, 1981). According to the predicted molecular weights, bands 4 and 4* are likely to be the products of PRF 38 (predicted molecular weight is 38.7 kDa) and PRF 41 (36.9 kDa). It appears therefore that both gene cassettes are involved in tail fibre formation in F2 pyocin. It should be noted that the N-terminal region of PRF 38 (and PRF 41) showed a similarity to that of a putative tail fibre protein of pahe N15 (gp25) and to that of the side tail fibre protein of phage HK97, as well as to the N-terminal part of the Y1051 protein encoded by a λ-like gene element on the virulence plasmid of Yersinia pestis, pMT1 (Lindler et al., 1998). The Y1051 gene of pMT1 has been annotated as a homologue of the side tail fibre gene of λ (accession no. AF074611).
Genes related to lysis function. No gene related to lysis function has been identified for any type of pyocin so far. However, some kind of lytic system is necessary for a large molecule, such as R- or F-type pyocin, to be released from the cells. Among the pyocin genes, PRF 24 showed a weak but significant similarity to the active domains of various chitinases (Table 1). Lytic enzymes of bacteriophages hydrolyse bacterial peptidoglycan consisting of N-acetyl-d-glucosamine and N-acetyl-d-muramic acid, and chitinases hydrolyse homopolysaccharide of N-acetyl-d-glucosamine, raising the possibility that PRF 24 could be the lytic enzyme for the pyocins. Phage lysis genes (genes for a lytic enzyme, a holin protein that participates in translocation of the lytic enzyme, and their modulators) are generally present as a gene cassette (Young, 1992). When we examined the hydrophobicity profile of ORFs located in the vicinity of PRF 24, PRF 9 was found to exhibit a hydrophobicity profile similar to those of the holins of P2, φCTX as well as λ phage. Moreover, the hydrophobicity profiles of PRF 25 and PRF 26 were strikingly similar to those of the products of lysB and lysC, the lysis-controlling genes of P2 (Ziermann et al., 1994; G. Christie, personal communication), and their homologues in φCTX (data not shown). These data suggest that PRFs 9, 24, 25 and 26 encode a lytic system similar to those of bacteriophages.
To confirm this, we cloned PRF 9 and PRF 24 of PAO1 into a broad-host-range expression vector, pMMB24 (Bagdasarian et al., 1983), and examined the effect of expressing these genes in Escherichia coli and P. aeruginosa cells. PRF 9 of PML14 was also examined as its C-terminal amino acid sequence was different from that of PAO1. As shown in Fig. 5A and B, PRF 9 of both strains exhibited strong lethal activity in P. aeruginosa upon induction by the addition of IPTG. In contrast, expression of PRF 24 did not affect cell growth (Fig. 5C). However, when a small amount of chloroform was added to permealyse the cell membrane to allow the lytic enzyme produced in the cytoplasm to be accessible to the peptidogycan molecules in the periplasmic space, quick lysis of the host cells was observed in both P. aeruginosa and E. coli. Furthermore, when PRF 24 was cloned into another expression vector, pME6012tlq, and co-expressed with PRF 9 in E. coli, clear cell lysis occurred in the absence of chloroform treatment (Fig. 5D). These results clearly indicate that PRFs 9 and 24 encode a holin and a lytic enzyme respectively. R2 and F2 pyocins appear to share this lytic system, which probably includes a lysis control system encoded by PRFs 25 and 26. It is also likely that the system is shared by the S-type pyocin, as the S pyocin genes examined so far were not accompanied by any lysis gene (Sano et al., 1993; Kageyama et al., 1996).
In contrast to PRF 9 of PAO1, PRF 9 of PML14 showed no toxic effect in E. coli (Fig. 5A and B). In addition, manifestation of lethal activity in P. aeruginosa was delayed significantly for PRF 9 of PML14 compared with that of PAO1. As the nucleotide sequences of both genes were almost identical except for the C-terminal regions, the expression levels of these genes were not expected to differ significantly. These phenomena were thus attributed to the structural difference in the C-terminal hydrophilic tails. Considering that holins from Gram-positive bacteria expressed lethal activity to E. coli (Steiner et al., 1993; Díaz et al., 1996; Chandry et al., 1997) and that λ holin was toxic even in yeast (Garrett et al., 1990), this apparent ‘host specificity’ of the PML14 holin was surprising. The same phenomenon has been observed for the holin of φCTX as well (K. Nakayama and T. Hayashi, unpublished).
Other genes. Two known genes, prtR and prtN, are located farthest upstream in the pyocin gene cluster, and their transcription is in the opposite direction from the other genes (Fig. 1). These two genes, together with the recA gene, regulate the induction of all three types of pyocins, as mentioned above.
Between prtR and PRF 9 (holin gene), three ORFs (PRFs 5, 6 and 7) of unknown function were found (Fig. 1). The gene configuration suggested that they may form a single transcription unit. Among these, PRF 6 encoded a small protein belonging to the TraR protein family, which are prokaryotic C4-type zinc finger proteins. The family is characterized by the sequence -C-X2-C-X17-C-X2-C-, which is located in the C-terminal half of the molecule. As shown in Fig. 6, homologues of PRF 6 were found on a variety of genetic elements, including the tra operons of F and R100 plasmids, members of the P2 phage family, λ phage (the homologue of λ phage was not identified before) and the chromosomes of E. coli, Haemophilus influenzae, Y. pestis and Yersinia pseudotuberculosis. The highest homology was found with ORF39 of φCTX (50% identity), a constituent of the φCTX early gene cluster. Although no function has been assigned to any members of the TraR family, a similar C4-type zinc finger motif is present in the C-terminal region of the DnaK suppressor protein family.
Overall gene organization. Based on the gene arrangement and predicted functions, the ORFs in the region between trpE and trpGCD have been assigned to five gene clusters: (i) genes regulating the gene expression; (ii) genes of unknown function; (iii) genes related to lysis function; (iv) R2 pyocin structural genes; and (v) F2 pyocin structural genes (Fig. 2). Among these, the regulatory genes and lysis genes are shared by both pyocins, and probably by S-type pyocins as well. The presence of P-box-like sequences upstream of PRFs 9 and 28 suggested that, in PML14, the lysis genes and F2 structural genes are transcribed as two independent transcription units. In PAO1, the R2 genes appear to be successfully integrated in the lysis gene operon of F2 pyocin.
Evolution of R2 and F2 pyocin genes
Based on previous genetic, biochemical and morphological studies on R- and F-type pyocins, it has been proposed that both pyocins were variations of defective phages (Kageyama, 1985). Extensive homology of R2 and F2 pyocin genes to the tail genes of P2 and λ phage families, along with the highly conserved gene orders, clearly demonstrated that each pyocin is derived from a common ancestor with the P2 or λ phage families. The presence of a lysis gene cassette similar to those of bacteriophages also supports this notion. However, no other phage-related genes, such as genes for head formation, replication and integration, are found in the R2/F2 pyocin gene cluster. Moreover, gene expression of both R2/F2 pyocins, together with that for S-type pyocins, is integrated under a single regulatory system, and these genes are co-ordinately regulated (Matsui et al., 1993). Thus, both types of pyocins are regarded as phage tails that have been evolutionarily specialized as bacteriocins rather than as simple defective phages. Considering the broad distribution of P2 and λ phage families among bacterial species and genera, it seems likely that bacteriocins similar to R and F pyocins would also be distributed over a wide range of bacterial species. In this regard, the genetic structures of the other defective phage-like bacteriocins, such as monocins and carotovoricins of Listera and Erwinia species, are of interest (Itoh et al., 1978; Zink et al., 1995). Some genetic relatedness has been found recently between carotovoricin and the P2 phage family (Nguyen et al., 1999).
To elucidate the phylogenetic relationship of R pyocin with the P2 phage family, we compared the amino acid sequences of gpFI and gpFII (or their homologues) from members of the P2 phage family (P2, 186 and φCTX) and R2 pyocin (Fig. 7). As the DNA sequence of the FI and FII genes of PS17 phage, a prototype of R-type pyocin-related phage, has been determined (Sasaki et al., 1997), this was also included in the analysis. The comparison indicated that PS17 and R2 pyocin were evolutionarily very closely related, which explains the phenotypic mixing observed between them. In contrast, φCTX was more closely related to P2 and 186, rather than R2 and PS17. It implies a complicated evolution of the P2 phage family in P. aeruginosa.
Genes for other subtypes of R and F pyocins
The base composition of a large part of the R2 and F2 pyocin locus was similar to that of the P. aeruginosa chromosome (66 G+C mol%). Most of the R2 and F2 pyocin genes are therefore regarded as having adapted to the P. aeruginosa chromosome. This notion was confirmed by codon adaptation analysis of the genes (Kanaya et al., 1996; Nakayama et al., 1999); most genes exhibited a Z1-value > −1 (Table 1). However, several AT-rich segments were evident in the R2–F2 region as shown in Fig. 1; in particular, a region ranging from the middle of PRF 15 to PRF 16 and that from the 3′-terminus of PRF 36 to PRF 43. They were assumed to be regions introduced from other genomes not so far in the past. Interestingly, both were regions responsible for the formation of tail fibres that are involved in binding to sensitive cells. As already mentioned, genes for tail fibre formation of double-stranded DNA phages have been proposed to consist of several modules and to have undergone complicated recombination. Modules are regarded as the basic unit of the recombination and, in some cases, these modules are of foreign origin. Genes responsible for tail fibre formation of the pyocins are also suggested to have undergone the same genetic exchanges as in double-stranded DNA phage.
In this context, it is of interest that several kinds of subtypes have been found for both R-type and F-type pyocins, which exhibit different killing spectra. These subtypes resemble each other very closely, not only in morphology but also in protein composition, except for the tail fibre components, which determine the specific range of sensitive cells for each subtype. It may therefore be possible to assume that every subtype of R-type or F-type pyocin is principally encoded by the same set of genes as those for R2 or F2 (thus, have the same structures), but that only the genes for tail fibre formation have been altered. Gene sets for other subtypes of R- and F-type pyocins may also exist between trpE and trpGCD, as has been suggested for R1 and R3 pyocins (Kageyama, 1975)
To test this presumption, we prepared 12 pairs of PCR primers that covered the entire trpE–trpG region of PAO1 in 1.3–5.8 kb intervals and performed a systematic PCR analysis of the trpE–trpG regions in P. aeruginosa strains that produced various subtypes of R and F pyocins (Figs 8 and 9; we call this method ‘PCR scanning’). In this analysis, each pair of primers was designed to amplify the coding region of the selected gene located in the trpE–trpG region (Fig. 9). We can thus easily know whether these target genes are present in each tested strain. Furthermore, using various combinations of primers, we can examine whether the target genes are arranged in the same order as in PAO1. In addition, by comparing the sizes of amplified fragments with those from PAO1, we can predict whether the regions between the target genes have undergone any structural changes or not.
As shown in Fig. 8, all the primers worked well, and the obtained data are summarized in Fig. 9. The results suggested that genes for other subtypes of R and F pyocins also exist between trpE and trpG, and that the gene organizations for each subtype are principally the same as those for R2/F2 pyocins. A complete set of R pyocin genes was missing in strain PAC1, as in PML14. Strain SLM6 seemed to lack the complete set of F pyocin genes, although P. aeruginosa strains with only R-type pyocin genes are not known. To confirm this and to determine the boundary of the F pyocin-specific region, we determined the nucleotide sequence of the PRF 24–trpG region of SLM6 (Fig. 1, DDBJ accession no. AB045308). The results clearly indicated that all genes that were predicted to be responsible for F pyocin production were absent in SLM6 and, instead, a 28 -bp sequence specific to this strain was found between PRF 26 and trpG.
In several strains, the gene organizations of PRF 36–trpG regions appeared to be different from that of F2. To investigate the structural difference, we determined the nucleotide sequence of the PRF 36–trpG region of strain PML15, which produced R1 and F1 pyocins (Fig. 4, DDBJ accession no. AB046379). The nucleotide sequence of the region ranging from the 3′ end of PRF 36 to 311 bp upstream of trpG was different from that of PAO1 and, as shown in Fig. 4, six ORFs were identified in the PML15 region. Three of them were homologous to the three genes constituting a duplicated gene cassette of F2 pyocin (PRFs 38–40 and PRFs 41–43), but with significant sequence diversity. These data suggested that the region responsible for tail fibre formation had been altered in PML15(F1 pyocin). This is in a good agreement with the previous finding that different killing spectra of F-type pyocins are associated with the difference in tail fibre of each subtype. The occurrence of one gene cassette is also consistent with the protein profile of F1 pyocin; only one tail fibre protein was detected in F1, but two in F2 (Kuroda and Kageyama, 1981).
PRF 15 encodes the tail fibre protein of R-type pyocin, which is also suspected to be responsible for the different killing spectra of each subtype of R pyocin. Thus, some sequence diversity could be expected among the R pyocin subtypes. PRFs 13–17 regions of the strains tested, however, did not show apparent structural diversity in the PCR scanning analysis. To confirm this, the nucleotide sequence of the PRFs 13–17 region of R1 pyocin (PML15) was determined (DDBJ accession no. AB046380) and compared with that of R2. As expected, the nucleotide sequence for the region was highly conserved with those for R2, except for a short segment encoding the C-terminal one-third of PRF 15 and PRF 16. Despite the sequence diversity, the gene organization of PRFs 15 and 16 was conserved, and they showed overall amino acid sequence similarities to their counterparts in R2(Fig. 3). However, the C-terminal region of PRF 15, a putative receptor binding region of tail fibre, was completely different from that of R2 pyocin.
All these data obtained from the PCR scanning analysis and subsequent sequence analyses of amplified fragments supported our presumptions that: (i) gene sets for every subtype of R- and F-type pyocins exist between trpE and trpG; (ii) every subtype of R or F pyocin is principally encoded by the same set of genes as those for R2 and F2; and (iii) only the genes for tail fibre formation have been altered, which correspond to the difference in killing spectra. Nucleotide sequence determination of the genes related to tail fibre formation in other subtypes will further confirm our presumptions and provide more detailed information on the structural features and evolution of R-type and F-type pyocins in P. aeruginosa.
The terms for the R and F pyocin genes
Because of the extensive similarity between R2 pyocin genes and P2 tail genes, it is reasonable to consider the ORFs with marked homologies to the P2 phage genes as having functions equivalent or very similar to those of their P2 counterparts, and it is more convenient and practical to designate these R2 pyocin genes according to the P2 system that we used for the genes of φCTX phage (Nakayama et al., 1999). Thus, we will call these pyocin genes VR2 for PRF 11, WR2 for PRF 12, etc. (Table 1). Using the same rationale, we propose to name F2 pyocin genes homologous to known λ genes according to the λ system: VF2 for PRF 28 and HF2 for PRF 31, etc. (Table 1). It should be noted here, however, that we have not confirmed the correlation between the ORFs identified in the present study and the genes previously identified by a classical genetic approach and, thus, the gene names proposed here do not correspond to those used in the previous study (prtA–M;Shinomiya et al., 1983a).
Bacterial strains, culture medium and plasmids
P. aeruginosa PAO1 is a strain that produces R2 and F2 pyocins, and PML14 only F2 pyocin (Holloway, 1969; Kageyama, 1975). P. aeruginosa PAO4260 (met-9092, pro-9024, blaP-9208, rec-102) has been described previously (Hayashi et al., 1989). As a cloning vector and a host for preparing random shotgun libraries, pUC18 and E. coli DH5α MCR (Gibco BRL) were used. For the routine cultivation of clones in the library, Luria–Bertani (LB) broth containing 100 µg ml−1 ampicillin was used as liquid medium, and LB agar containing 20 µg ml−1 ampicillin and 80 µg ml−1 methicillin as solid medium.
Determination of nucleotide sequences
Two contiguous HindIII fragments of 8 kb and 11 kb from PAO1 (Shinomiya et al., 1983b) were individually subcloned into the HindIII site of pKT230, resulting in pHKM2 and pIM1. Each plasmid was digested with HindIII (TaKaRa), and HindIII fragments were purified from the gel using Prep-A-Gene DNA purification kits (Bio-Rad). Each purified fragment was fragmented by sonication, and DNA fragments of 1–2 kb in size were fractionated by electrophoresis on a 1% agarose gel. After treatment with BAL 31 exonuclease (TaKaRa) and T4 DNA polymerase (TaKaRa), DNA fragments of 1–2 kb were again purified by agarose gel electrophoresis. Purified fragments were then ligated with pUC18 that had been digested with SmaI and treated with bacterial alkaline phosphatase, using ligation kit version 2 (TaKaRa). The recombinant DNAs were used to transform E. coli cells to prepare random shotgun libraries. Transformants were cultivated at 28°C overnight in 96-well microplates, and the inserted DNA fragments were amplified using an Ex PCR kit (TaKaRa) and a pair of primers (LF, 5′-GTGCTGCAAGGC GATTAAGTTGG-3′; and LR, 5′-TCCGGCTCGTATGT TGTGTGG-3′) with 30 cycles at 96°C for 15 s/68°C for 2 min 40 s. After treatment with exonuclease I and shrimp alkaline phosphatase (Amersham), each PCR product was used directly as a template for DNA sequencing. The sequencing reactions were performed using Prism Dye Terminator cycle sequencing kit FF or FS (Applied Biosystems), and sequences were determined by ABI Prism 377 or 373A autosequencers (Applied Biosystems). As a sequencing primer, −21M13 universal forward primer was used regularly. Sequences obtained were assembled by seqencher DNA sequencing software (version 3.0, Gene codes). The sequences of gap regions and regions where only one strand had been obtained were determined by the primer walking method using custom primers or by sequencing selected clones using a reverse primer (5′-GTGTGGAATTGT GAGCGG-3′). As a result, sequences of both strands were determined at least once for the entire region.
For the sequence determination of the trpE–PRF31 region of PML14, the PRF13–PRF17 and PRF36–trpG regions of PML15, each region was amplified by PCR using an LA PCR kit (TaKaRa) and the primers presented in Table 2. Sequencing of the amplified fragments was performed by random shotgun strategy as described above. To rule out PCR errors, at least three independent clones were sequenced for the entire region. In addition, the sequences with any discrepancy were confirmed by direct sequencing of the PCR product.
Table 2. Primers used in this study.
Lowercases indicate nucleotide sequences that were introduced to create restriction sites (underlined).
For lysis gene analysis
For PCR scanning
Prediction of protein coding regions (ORFs)
ORFs of more than 150 bp, with ATG or GTG as the initiation codon, were predicted for all six frames by Gene Mark (version 2.0) using the matrix for P. aeruginosa. Intergenic regions longer than 100 bp were searched manually for ORFs with ribosome-binding sequences or homology to known genes.
Total genomic DNAs used as templates were prepared as described previously (Hayashi et al., 1989). All the P. aeruginosa strains tested were from our laboratory stock. Known subtypes of R and F pyocins produced by these strains are indicated in Fig. 9. All primers were designed based on the nucleotide sequence of PAO1 (Table 2). PCR reaction was performed on a T3 thermocycler (Biometra) using an LA or Ex PCR kit with 30 cycles at 96°C for 15 s/68°C for 2–7 min. PCR products were analysed on 2% or 0.5% agarose gels.
Homology searching was performed using the blast (version 1.49) and fasta (version 3.0) programs implemented at DDBJ. Multiple alignment and phylogenetic analysis of protein sequences were performed by clustal w through DDBJ. Adaptiveness of ORFs to the P. aeruginosa genome was analysed as described previously (Nakayama et al., 1999), and the adaptiveness was expressed by a Z1-value (Kanaya et al., 1996). Other analyses were performed using the programs contained in the genetyxmac software package (version 10.0, Software Development).
Functional analysis of PRFs 9 and 24
PRF 9 of PAO1 and PML14 and PRF 24 of PAO1 were obtained by PCR amplification using three sets of primers; R2-H9 and R2-9B for PRF 9 of PAO1, R2-H9 and R2-9B14 for PRF 9 of PML14, R2-H24 and R2-24B for PRF24 of PAO1 (Table 2). For cloning the amplified fragments, HindIII or BamHI restriction sites were created in the primers. PCR reactions were performed using the LA PCR kit with 25 cycles of 96°C for 20 s/58°C for 60 s/72°C for 90 s. After digestion with HindIII and BamHI, each PCR product was ligated with HindIII–BamHI double-digested pMMB24 so that the genes were placed under the control of the tac promoter, yielding pPAO1-9, pPML14-9 or pPAO1-24A. These recombinant plasmids were introduced to DH5α, and the absence of mutation was confirmed by sequencing. The plasmids were then introduced to P. aeruginosa PAO4260. DH5αF′Iq was used as the E. coli host strain for pPAO1-24A.
DH5α and PAO4260 containing pPAO1-9, pPML14-9 or pMMB24 were precultured at 30°C in LB broth supplemented with carbenicillin (150 µg ml−1) and glucose (0.2% w/v). The overnight cultures were diluted into fresh LB broth supplemented with carbenicillin (150 µg ml−1) to an OD600 of ≈ 0.1 and grown at 37°C with vigorous shaking. At an OD600 of 0.2–0.3, half the culture was induced with IPTG at a final concentration of 1 mM. The OD600 of the culture was measured at various time points. To count the viable cells, aliquots of each culture were recovered at various time points, serially diluted with LB broth and plated on LB agar plates supplemented with carbenicillin (150 µg ml−1) and glucose (0.2% w/v). Plates were incubated at 30°C for 24 h (DH5α) or 42 h (PAO4260), and the number of colonies was counted. The effect of expression of PRF 24 from PAO1 on the growth of DH5αF′Iq and PAO4260 was examined in a similar manner, except that, at an OD600 of ≈ 1, chloroform was added to the cultures at a final concentration of 1.0% (v/v) for DH5αF′Iq and 0.5% (v/v) for PAO4260.
For co-expression of PRFs 9 and 24, we first constructed a broad-host-range expression vector pME6012tIq by inserting a 260 bp EcoRI–HindIII fragment and a 1.1 kb HindIII fragment of pMMB22 (Bagdasarian et al., 1983), which carried the tac promoter and the lacIq gene, respectively, into the multicloning site of pME6012 (Heeb et al., 2000). PRF 24 was then cloned into pME6012tIq so that the gene was placed under the control of the tac promoter, yielding pPAO1-24B. As pME6012 is compatible with pMMB24, pPAO1-24B could be introduced to DH5αF′Iq containing pPAO1-9, and PRFs 9 and 24 could be co-induced with IPTG. The effect of co-expression of the two genes was examined in a similar manner to that described above, except that tetracycline (10 µg ml−1) was also supplemented in the culture medium.
We thank Gail E. Christie for critical reading of the manuscript, and Gail E. Christie and Takashi Kumazaki for communication of unpublished data. We are grateful to Yoshiro Terawaki and Tatsuo Sakai for encouragement, to Mika Takahashi, Shuko Setsu and Kaori Satoh for technical assistance, and to Yumiko Hayashi for editorial assistance. This work was supported by a grant-in-aid for scientific research from the Ministry of Education, Sports and Culture of Japan.