Retargeting R-Type Pyocins To Generate Novel Bactericidal Protein Complexes

Abstract

R-type pyocins are high-molecular-weight bacteriocins that resemble bacteriophage tail structures and are produced by some Pseudomonas aeruginosa strains. R-type pyocins kill by dissipating the bacterial membrane potential after binding. The high-potency, single-hit bactericidal kinetics of R-type pyocins suggest that they could be effective antimicrobials. However, the limited antibacterial spectra of natural R-type pyocins would ultimately compromise their clinical utility. The spectra of these protein complexes are determined in large part by their tail fibers. By replacing the pyocin tail fibers with tail fibers of Pseudomonas phage PS17, we changed the bactericidal specificity of R2 pyocin particles to a different subset of P. aeruginosa strains, including some resistant to PS17 phage. We further extended this idea by fusing parts of R2 tail fibers with parts of tail fibers from phages that infect other bacteria, including Escherichia coli and Yersinia pestis, changing the killing spectrum of pyocins from P. aeruginosa to the bacterial genus, species, or strain that serves as a host for the donor phage. The assembly of active R-type pyocins requires chaperones specific for the C-terminal portion of the tail fiber. Natural and retargeted R-type pyocins exhibit narrow bactericidal spectra and thus can be expected to cause little collateral damage to the healthy microbiotae and not to promote the horizontal spread of multidrug resistance among bacteria. Engineered R-type pyocins may offer a novel alternative to traditional antibiotics in some infections.

The discovery and development of broad-spectrum antibiotics about 60 years ago added to the armamentarium of the physician and surgeon one of the most effective tools of their trade. However, the recent appearance and dissemination of multidrug resistance among bacteria have created a new threat for which traditional antibiotics offer little protection (18, 33, 41).

R-type pyocins are high-molecular-weight bacteriocins produced by certain Pseudomonas aeruginosa strains apparently to defend against other strains of the same species (24). R-type pyocins resemble the nonflexible, contractile tail structures of bacteriophages of the Myoviridae family (Fig. 1) (23) and are encoded in a single cluster in the P. aeruginosa genome (38). Five naturally occurring R-type pyocins have been identified to date and, based on their bactericidal spectra, are termed R1 through R5 (22). The bactericidal specificity of R-type pyocins is conferred by their tail fibers encoded by prf15. Prf15 protein is closely related to the tail fibers of phages of the Myoviridae family, particularly P2-like phages (32). A predicted chaperone (based on homology to phage P2) is encoded by prf16, immediately downstream of prf15, and is thought to be required for tail fiber assembly (32). P. aeruginosa strain PAO1 produces R2 pyocin, which is encoded in a gene cluster consisting of 16 open reading frames analogous to proteins of known function, 12 of which show significant sequence similarity to open reading frames of bacteriophages P2, ΦCTX, and other P2-like phages (32). R-type pyocin production is induced by DNA damage and regulated by RecA, which degrades PrtR, the repressor of PrtN, a positive transcription regulator of the cluster (25). Induction of R-type pyocin genes results in synthesis of approximately 200 pyocin particles per bacterial cell followed by lysis of the cell from within by mechanisms similar to those of bacteriophage-mediated lysis. R-type pyocins rapidly and specifically kill target cells by first binding to the bacterial cell surface via their tail fibers. Once bound, pyocins are triggered to contract their sheath, leading to penetration of the core through the bacterial outer membrane, cell wall, and cytoplasmic membrane. The last penetration results in the escape of concentrated intracellular ions through the patent core, depolarization of the cytoplasmic membrane, and consequential death (40, 42). In many respects, R-type pyocins can be viewed as defective prophages that produce noninfectious particles consisting only of the tail apparatus, that is, without capsids or DNA, and that have been adapted by the host as defensive antibacterial agents. The producing bacteria and their offspring are resistant to the produced pyocins (19). Since the producing bacterium is lysed in the process of releasing the pyocins, this defense system might be viewed as altruistic—self-sacrifice to protect kin from competitive unrelated strains.

FIG. 1.

R-type pyocin electron microscopic image (A) and diagram (B). Shown is a scanning electron micrograph of R2 pyocin from PAO1 negatively stained with phosphotungstic acid at ×115,000 (courtesy of Jeff Miller and Mari Gingery, UCLA). The diagram depicts the major structural features as labeled. The length of the sheath is approximately 80 μm.

Although P. aeruginosa R-type pyocins kill mainly strains of P. aeruginosa, they have also been shown to kill some strains of Haemophilus, Neisseria, and Campylobacter species (3-5, 10, 30, 31). Similar high-molecular-weight bacteriocins have been described in other gram-negative as well as gram-positive bacteria (2, 9, 20, 40, 46).

The high potency of R-type pyocins suggests that they could be developed as antimicrobial agents (27), and a few studies have shown their antibacterial efficacy in mice (12, 26, 35). However, at the time of the early studies, broad-spectrum antibiotics were clinically effective, and there was little interest in pursuing this avenue, particularly since the limited spectra of R-type pyocins were viewed as problematic. Now that antibiotic resistance has become rampant, especially among gram-negative bacteria (43), there is incentive to reexamine R-type pyocins as antimicrobials. For pyocins to be clinically useful as antibacterial agents, however, their limited bactericidal spectra must be addressed. In this work, we show that it is possible to retarget the spectrum of R2 pyocin by swapping tail fibers or fusing the N-terminal portion of its tail fiber to the C-terminal portion of the tail fiber of bacteriophages that infect hosts other than R2-sensitive P. aeruginosa. By fusing the C terminus of the P2 tail fiber to the R2 Prf15, we have changed the spectrum of R2 pyocin to kill Escherichia coli strain C and multiple uropathogenic E. coli (UPEC) strains. By a similar approach, we have created pyocins that kill Yersinia pestis.

MATERIALS AND METHODS

Bacterial strains, phages, plasmids, and media.

Table 1 lists the bacterial strains, phages, and plasmids used in this study and their sources. P. aeruginosa strains were grown on tryptic soy broth and tryptic soy agar. E. coli and Y. pestis strains were grown on standard Luria-Bertani (LB) broth and LB agar. For maintenance of plasmids, gentamicin was added at 15 μg/ml for E. coli or 100 μg/ml for P. aeruginosa. LBS agar is LB agar lacking NaCl and containing 50 mg/ml sucrose. Pyocin inductions were performed in G medium (17) with 50 μg/ml gentamicin where appropriate.

TABLE 1.

Characteristics of bacterial strains, bacteriophages, and plasmids used in this study

Strain, phage, or plasmid	Relevant characteristics	Source or referencea
Strains
P. aeruginosa
PAO1	Clinical wound isolate; R2 pyocin producer	Jeff F. Miller, UCLA
PAO1 Δprf15	Deletion of codons 11-301 in prf15	This work
PAO1 Δprf15-Pscr	PAO1 Δprf15 resistant to phage PS17c5	This work
NIH-K	R1 pyocin producer	ATCC 25350
NIH-I	R3 pyocin producer	ATCC 25348
NIH-H	R4 pyocin producer	ATCC 25347
NIH-1	R5 pyocin producer	ATCC 25313
PML14	Host strain for PS17	Jeff F. Miller, UCLA
28-Strain group	Clinical isolates labeled 1 to 10 and 1s to 18s	Stanford CML
13s	Indicator strain for R1 to R5 pyocins	Stanford CML
E. coli
C-1a	Host strain for phage P2	Richard Calendar, UCB
MG1655	Wild-type K-12 strain	Sankar Adhya, NIH
8-Strain group	Uropathogenic clinical isolates	Harry Mobley, University of Michigan
Y. pestis KIM D27	pgm mutant avirulent strain	Sankar Adhya, NIH
Bacteriophages
P2	P2 phage	Richard Calendar, UCB
PS17	Temperate P. aeruginosa phage	Université Laval
PS17c5	Spontaneous clear-plaque mutant of PS17	This work
Plasmids
pEXGm18	Allelic exchange suicide vector	Herbert Schweizer, CSU
pEXGmΔprf15	pEXGm18-based Δprf15 vector	This work
pUCP30T	Gmr broad-host-range plasmid	Herbert Schweizer, CSU
pUCPtac	laclqtac MCSb in pUCP30T	This work

^aUCLA, University of California, Los Angeles; ATCC, American Type Culture Collection; Stanford CML, Stanford Clinical Microbiology Lab; UCB, University of California, Berkeley; NIH, National Institutes of Health; CSU, Colorado State University.

^bMCS, multiple cloning site.

PCR conditions.

Amplification of fragments to be used for cloning was performed using Pfx50 polymerase (Invitrogen) according to the supplier's recommendations. Analytical PCRs were performed using Taq 2X master mix (New England BioLabs).

Construction of Δprf15 mutant in PAO1 by allelic exchange.

An in-frame deletion of codons 11 to 301 of prf15 was made in P. aeruginosa PAO1 as follows. An approximately 1,140-bp fragment upstream of the proposed deletion was amplified from PAO1 genomic DNA using primers AV85 (5′-GCTTCAATGTGCAGCGTTTGC) and AV88 (5′- GCCACACCGGTAGCGGAAAGGCCACCGTATTTCGGAGTAT) and then digested with KpnI and AgeI. An approximately 2,195-bp fragment downstream of the proposed deletion was similarly amplified using primers AV87 (5′-ATACTCCGAAATACGGTGGCCTTTCCGCTACCGGTGTGGC) and AV86 (5′-TCCTTGAATTCCGCTTGCTGCCGAAGTTCTT) and then digested with AgeI and EcoRI. These two fragments were ligated into KpnI- and EcoRI-digested pEX18Gm (15) (GenBank accession no. AF047518) to make pEX18Δprf15. The allelic exchange vector was introduced into PAO1 via electroporation (7). Integrants (single-crossover events) were selected with gentamicin, and then an additional crossover event was selected for by growing the cells in nonselective LB broth for 3 h before plating on LBS agar plates at 30°C. Sucrose-resistant gentamicin-sensitive clones were analyzed by PCR. The desired deletion was confirmed by DNA sequencing. Several of the prf15-deleted clones were verified not to produce active pyocin when induced by mitomycin C, and one clone was designated PAO1 Δprf15.

Cloning and sequencing of R-type pyocin tail fiber genes.

PCR fragments were amplified from genomic DNA isolated from P. aeruginosa strains NIH-I (R3), NIH-H (R4), and NIH-1 (R5) using primers PRF13-F (5′-TATCGAGAACTGCTGCTGCGGG) and AV86 and then cloned using a TOPO cloning kit (Invitrogen). The inserts from representative clones were sequenced.

Construction of prf15 expression plasmids.

The broad-host-range vector pUCP30T (GenBank accession no. U33752) was modified by filling in the unique BspHI site, and a lacI^q gene amplified by PCR from pMAL-2E (New England BioLabs), a tac promoter, and multiple-cloning-site segment amplified from pGEX-2T (GE Healthcare) were all ligated into the resulting plasmid digested with SapI and EcoRI to generate pUCPtac. The R1, R2, R3, R4, and R5 prf15 or prf15/16 coding regions were amplified by PCR from the sequenced clones, and the amplified fragments were cloned into pUCPtac previously digested with BspHI and EcoRI, to yield the series of pUCPtac-prf15 and pUCPtac-prf15/16 plasmids.

Construction of P2 and L-413C tail fiber fusions and expression vectors.

To construct the R2-P2 fusions, the portion of prf15 encoding amino acids 1 to 164 was amplified, changing 164V to 164L to create a HindIII site, and was digested with BspHI and HindIII. P2 gene H encoding amino acids 158 to 669, with or without the downstream gene G, was amplified from a P2 phage stock. Each of the PCR products from P2 was digested with HindIII and EcoRI. pUCPtac-R2-P2H was created by cloning the prf15 1-164 fragment together with the P2 gene H 158-669 fragment into pUCPtac previously digested with BspHI and EcoRI. pUCPtac-R2-P2HG was generated by cloning the prf15 1-164 fragment together with the P2 gene H 158-669 fragment plus the gene G fragment into pUCP-tac previously digested with BspHI and EcoRI. The series of R2 prf15/P2 gene H fusions were similarly constructed by PCR amplification of the desired fragment size with added appropriate restriction sites. In all cases, gene G was amplified along with the C terminus of gene H and included in the constructs. The R2-L-413C fusion was made by synthesizing the entire gene H and gene G of phage L-413C based on the published sequence (GenBank accession no. NC004745). The synthesized genes were cloned into pUC19 and were confirmed by sequence analysis. The C-terminal portion of L-413C gene H along with gene G was amplified and cloned as described for the P2 fusions.

Cloning and sequencing of PS17 tail fiber gene.

PS17 plaques were produced by infection of P. aeruginosa strain PML14. PML14 cells lysogenic for phage PS17 were isolated by streaking cells from a hazy plaque onto a fresh plate. Mitomycin C at a final concentration of 3 μg/ml was added to a logarithmic culture at an A₆₀₀ of 0.2 to induce the phage. After lysis, phage DNA was purified using a Qiagen Lambda miniprep kit. Based on the published restriction map (37), we isolated an approximately 4.2-kb BglII fragment on an agarose gel, ligated the fragment into BamHI-digested pUC19, and sequenced it by primer walking. Two overlapping open reading frames were found with significant homology to R2 prf15 and prf16.

Isolation of PAO1 Δprf15-Psc^r.

To enable the harvest of engineered R2-PS17 pyocins from PAO1 Δprf15, which is sensitive to PS17 phage, we wished to isolate a PS17-resistant PAO1 Δprf15 variant as an expression host. Since PS17 lysogenizes frequently, it was necessary to first isolate a virulent form. Clear-plaque mutants of PS17 were isolated by plating phages at different densities on strain PML14 and picking clear plaques for further use. Five clear-plaque versions of PS17 were isolated and confirmed by PCR to contain PS17 sequences, and the phage designated PS17c5 was chosen for further work. A 50-ml culture of PAO1 Δprf15 at an A₆₀₀ of 0.2 was infected by adding a single plaque of PS17c5. The culture lysed after approximately 3 h and was left shaking overnight to allow phage-resistant cells to grow. Single colonies were isolated after streaking out cells from the overnight culture. The clone designated PAO1 Δprf15-Psc^r was resistant to PS17 and PS17c5 and was shown to be derived from PAO1 Δprf15 because it contained the prf15 deletion as confirmed by analytical PCR.

Pyocin purification and assays.

Saturated cultures of P. aeruginosa were diluted 1:100 in G medium supplemented with 50 μg/ml gentamicin when needed to maintain plasmids and shaken at 225 rpm at 37°C. When the cultures reached an optical density at 600 nm of 0.250, mitomycin C was added to a final concentration of 3 μg/ml. In those cultures containing a plasmid encoding a recombinant tail fiber gene under control of a tac promoter, a final concentration of 250 μM isopropyl β-d-1-thiogalactopyranoside (IPTG) was added at the same time as the mitomycin C. Cultures were incubated for an additional 2.5 h or until lysis occurred. DNase I (Invitrogen) to a final concentration of 2 U/ml was added to reduce viscosity, and the culture was incubated for an additional 30 min. Debris was removed by centrifugation at 17,000 × g for 1 h. Saturated ammonium sulfate (4 M) was added to the supernatant at a rate of 1 ml/min while stirring on ice to a final concentration of 1.6 M, and the suspension was stored at 4°C overnight. The ammonium sulfate precipitate was sedimented at 17,000 × g for 1 h at 4°C, and the pellet was resuspended in 1/10 the original culture volume of cold TN50 buffer (50 mM NaCl, 10 mM Tris-HCl [pH 7.5]). Pyocin particles were then sedimented at 58,000 × g for 1 h at 4°C and resuspended in 1/20 the original culture volume of TN50 buffer. For small-scale preparations (6 ml culture volume), the ammonium sulfate precipitation was omitted and pyocins were pelleted directly from the cleared lysates. Pyocin preparations were estimated to be 70 to 90% pure by electrophoretic analysis on sodium dodecyl sulfate-polyacrylamide gels with Coomassie blue or silver staining (35).

Quantitative pyocin assays were performed by counting bacterial survival using a slight modification of the method described by Kageyama et al. (24). Pyocin samples were incubated in liquid with target bacteria (approximately 1 × 10⁹ CFU/ml) for 40 min at 37°C. The samples were then serially diluted and plated to count survivors. The number of pyocin particles is related to the fraction of bacterial survivors in a Poisson distribution, m = −ln S, where m equals the average number of lethal events per cell and S is the fraction of survivors. The total number of active pyocin particles/ml = m × cells/ml. Pseudomonas aeruginosa 13s is a clinical isolate sensitive to all natural R-type pyocins and was used for the target indicator strain for wild type R2. C1a was used as the indicator for the R-type pyocin constructs predicted to kill E. coli, and KIM D27 was the Y. pestis indicator.

Semiquantitative assays were also performed by a spot method wherein pyocin samples were fivefold serially diluted in TN50 buffer and 5-μl aliquots were spotted onto lawns of target bacteria. After overnight incubation at 37°C (30°C for Y. pestis cultures), pyocin activity could be observed as a distinct clear zone of killing on the lawn.

Nucleotide sequence accession numbers.

The sequences of the prf15 and prf16 genes from representative clones have been deposited into the EMBL/GenBank database under accession no. EF533708, EF533709, and EF533710 for R3, R4, and R5, respectively.

RESULTS

Complementation of an R2 prf15 deletion.

In order to determine whether the specificities of R-type pyocins could be altered, we first made an in-frame deletion in the tail fiber gene prf15 of the P. aeruginosa PAO1 pyocin gene cluster (Fig. 2). This created a pyocin production host (PAO1 Δprf15) in which to express in trans complementing tail fiber genes. The coding region for the R2 tail fiber chaperone, prf16, overlaps the 3′ end of the R2 prf15 gene by 8 nucleotides, and thus the prf16 ribosome binding site resides within the prf15 coding region (Fig. 2) (14). Both the transcription start site for prf16 and its ribosome binding site were left intact in PAO1 Δprf15 such that the chaperone could be expressed upon induction of the pyocin cluster by mitomycin C.

FIG. 2.

R2 pyocin gene clusters of Pseudomonas aeruginosa PAO1 and deletion strain PAO1 Δprf15. (A) Wild-type gene cluster. The designations along the top are the presumed functions of the genes. The numbers along the bottom are the prf designations of the genes. The transcription directions are indicated by the arrows. (B) Gene cluster showing the region of prf15 that was deleted to generate the production host for the trans-complementation experiments. The ribosome-binding site (RBS) and the translation start site of prf16 are indicated, as is the translation stop site of prf15.

R2-pyocin prf15 was expressed in trans with or without prf16 by cloning the gene(s) into pUCPtac and transforming PAO1 Δprf15. Transformed bacteria in log-phase suspension growth were exposed to mitomycin C to induce pyocin production. Simultaneously, the plasmid-encoded tail fiber with or without its cognate chaperone was induced with IPTG. Active pyocin particles were produced upon induction. When R2 prf15 was expressed from the plasmid in PAO1 Δprf15, the pyocins had comparable yield (about 5 × 10¹¹ pyocin particles per ml in the lysate) and the same spectrum as R2 pyocins produced from wild-type PAO1 (data not shown). When both R2 prf15 and R2 prf16 were expressed in trans from the plasmid, the yield of R2 pyocin was consistently greater (2× to 5×) than when the only source of Prf16 was the genome of PAO1 or the production host, PAO1 Δprf15. This observation suggests that in PAO1 the availability of the chaperone, Prf16, limits the production of natural R2 pyocin.

R1, R3, R4, and R5 tail fibers can be incorporated into the R2 pyocin particle.

Of the five known R-type pyocins, the gene sequences for the tail fibers were known only for R1 and R2. From their producer strains (ATCC [see Materials and Methods]), we isolated by PCR and sequenced R3, R4, and R5 pyocin tail fiber (prf15) genes along with their cognate chaperone genes (prf16) (GenBank accession no. EF533708, EF533709, and EF533710 for R3, R4, and R5, respectively). Primers were designed from flanking sequences based on PAO1, all of which resulted in a PCR product, indicating that the flanking sequences in all of these strains are very similar. R3 and R4 Prf15 tail fiber proteins are nearly identical to that of R2 (see the supplemental material) (6, 8), as are their cognate chaperones (see the supplemental material). R1 and R5 Prf15 tail fibers, in contrast, have diverged considerably from R2 Prf15, mainly in the C-terminal portion of the gene product. R1 and R5 Prf16 chaperones have also diverged considerably from their R2 counterpart (see the supplemental material).

To determine whether the R1, R3, R4, and R5 tail fiber genes could complement the R2 prf15-deleted pyocin, we cloned and expressed in trans their respective tail fiber genes, with or without their cognate chaperones, in PAO1 Δprf15, as described above for R2 prf15. Each of these constructed strains was then induced with mitomycin C and IPTG to produce pyocins, and the relative yield of bactericidal activity of each harvested pyocin preparation was determined by spot assay using P. aeruginosa strain 13s (a clinical isolate sensitive to all five natural pyocins) as an initial indicator (see Fig. 4). R3 prf15 and R4 prf15, with or without coexpression of their cognate chaperones, complemented pyocin production by PAO1 Δprf15—not surprisingly since their chaperone genes are nearly identical to that of R2; R2 PAO1 Δprf15 was designed such that active PRF16 should still be produced. In contrast, optimal activity from the R1 prf15-complementing tail fiber required coexpression of its cognate chaperone, indicative of chaperone specificity (Fig. 3). In fact, the tail fiber chaperones of R1 and R2 pyocins have diverged considerably, presumably to support the folding or assembly of their respective tail fibers that are quite divergent at their C termini, the portion likely responsible for their binding to different targets. Likewise, the C-terminal portion of the R5 tail fiber and its cognate chaperone have both diverged considerably from those of R2 (see the supplemental material). In the case of expressing R5 prf15, practically no active pyocin particles were produced (based on spot tests and sodium dodecyl sulfate gels) unless the R5 chaperone gene was coexpressed with the R5 prf15 gene.

FIG. 4.

Fusions between R2 pyocin tail fiber and P2 phage tail fiber to create a pyocin bactericidal to E. coli C. Fusions between R2 prf15 (blue) and P2 gene H (red) were made at the indicated positions of the corresponding amino acids. Mutations were made at some junction points to facilitate cloning (R2 prf15, V164L; and P2 gene H, D202K, A321K, I422L N557K, Q620K, N557K, and Q620K). The full length of the R2 tail fiber is 691 amino acids, and that of P2 is 669 amino acids. All fusion genes were expressed in trans along with the P2 chaperone gene, gene G, in PAO1 Δprf15. After induction, preparations were assayed for bactericidal activity by spot tests on E. coli C1a. The assay results are scored as active (+) or inactive (−) based on whether a clearing at a dilution of >1:25 was observed on the bacterial lawn. The Coomassie blue-stained sodium dodecyl sulfate gel analysis showed that all constructs scored as active produced pyocin particles with a detectable tail fiber fusion. Those scored as inactive produced no intact particles.

FIG. 3.

Relative yields of active pyocin particles produced after trans complementation of PAO1 Δprf15. The five different natural pyocin tail fiber genes were expressed from the plasmid pUCPtac with or without their cognate chaperone genes. Activities were determined by spot tests using P. aeruginosa strain 13s (sensitive to all five natural pyocin types) as an indicator. The bar marked PAO1 is R2 pyocin produced from wild-type (WT) PAO1 for comparison. PAO1 Δprf15 itself produced no detectable pyocin activity.

These data indicate that the five different natural pyocin tail fiber genes could complement the prf15 deletion. All of these natural tail fibers are conserved at their N-terminal portions, presumably allowing them to attach to the R2 baseplate (see the supplemental material). In the case of R1 and R5 pyocins, only their tail fiber C termini have diverged significantly from that of R2, and yet assembly of active pyocin particles was not supported by the endogenous R2 chaperone. Thus, the pyocin chaperones probably acted near the C terminus of their respective tail fibers.

By assaying the bactericidal activities of the above complemented pyocins against a collection of 27 P. aeruginosa clinical isolates and comparing their activities to those of the five natural R-type pyocins (data not shown), we confirmed that each acquired the specific bactericidal spectrum of its tail fiber gene donor. These data indicate that the C-terminal portion of the tail fiber is the primary determinant of the bactericidal spectrum.

Novel spectrum of a pyocin with a PS17 tail fiber.

PS17 is a bacteriophage structurally similar to P2 and appears to share recent ancestry with R-type pyocins (32, 36). The tail fiber of PS17 has 53% amino acid sequence identity to the R2 tail fiber (32) (see the supplemental material).

PS17 also encodes a chaperone analogous to the R2 pyocin Prf16. The host range of PS17 encompasses some P. aeruginosa strains that are resistant to all five known natural pyocins, suggesting that if the PS17 tail fiber were swapped with that of R2, one could generate a pyocin with a unique specificity.

The PS17 tail fiber gene with and without its cognate chaperone gene was cloned and expressed in trans in PAO1 Δprf15 as described above. We also expressed the construct in PAO1 Δprf15-Psc^r. This is a mutant strain resistant to PS17. Active pyocin particles were produced only by constructs that included the PS17 chaperone gene, and only in PAO1 Δprf15-Psc^r. We believe that, because the PS17 tail fiber can bind to wild-type PAO1, production in a PS17-sensitive strain leads to binding of the pyocin particles to cellular debris. This pyocin, termed pyR2-PS17, has a bactericidal spectrum distinct from that of each of the five natural pyocins. Quantitative assays show that yields of pyR2-PS17 are comparable to those of wild-type R2: 2 × 10¹¹ to 5 × 10¹¹ pyocin particles per ml lysate. The spectrum of pyR2-PS17, while incompletely overlapping that of R5 pyocin (the broadest of the natural pyocins), included P. aeruginosa isolates not sensitive to any of the natural R-type pyocins (data not shown). We thus have expanded the number of P. aeruginosa isolates susceptible to at least one pyocin species, either natural or engineered. The spectrum of pyR2-PS17 was also broader than the spectrum of the PS17 phage itself, a phenomenon described below for another pyocin, pyR2-P2, with P2 phage-derived tail fibers (Table 2). We did not observe any P. aeruginosa isolate (out of 38) that was sensitive to the PS17 phage but not sensitive to pyR2-PS17.

TABLE 2.

Sensitivity of P. aeruginosa and E. coli strains to phage PS17 or P2 or pyocin pyR2-PS17 or pyR2-P2

Strain	Sensitivity to:
	Phage	Pyocin
P. aeruginosa	PS17	pyR2-PS17
PAO1	+	+
PA14	−	−
PML14	+	+
NIH-I	−	+
2	−	+
3	−	+
6	−	+
7	−	+
9	−	+
2s	−	+
4s	−	+
5s	−	+
11s	−	+
17s	−	+
18s	−	+
E. coli (urinary tract infection)	P2	pyR2-P2
F3	−	+
F11	−	+
F24	−	+
F54	−	−
CFT073	−	+
CFT204	−	−
CFT269	−	+
CFT325	−	+

Retargeting R2 pyocin to kill E. coli.

Bacteriophage P2, which infects many E. coli strains, has a tail fiber encoded by gene H as well as a chaperone gene, G, analogous to the pyocin chaperone (14). Although P2 phage is closely related structurally to PS17, the P2 tail fiber gene (14) has diverged considerably from tail fiber genes of both PS17 and R2 pyocin. We attempted to replace the R2 tail fiber with the P2 tail fiber as described above with PS17 to generate pyocins that specifically kill E. coli. However, we did not generate functional pyocin particles, even when the P2 chaperone gene was coexpressed (Fig. 4).

We then made a series of chimeric tail fiber genes by fusing different N-terminal region-encoding portions of R2 prf15 to different portions of gene H encoding C-terminal regions of the P2 tail fiber (Fig. 4). When the various fusion genes were expressed in trans in PAO1 Δprf15 simultaneously with the intrinsic pyocin genes, we were able to detect, using E. coli C1a as an indicator, active pyocin particles with some of the chimeric tail fiber genes, but even then only when the P2 chaperone gene G was coexpressed. We further characterized the pyocin particles, termed pyR2-P2, formed with the chimeric tail fiber comprised of amino acid residues 1 to 164 of R2 Prf15 and the C-terminal 512 amino acid residues of the P2 tail fiber. pyR2-P2 had bactericidal activity against E. coli C1a and E. coli MG1655 (both hosts for P2 phage), as well as six of eight UPEC isolates, even though P2 phage from which the tail fiber gene was isolated did not form plaques on any of the eight UPEC strains (Table 2). pyR2-P2 had no bactericidal activity against any P. aeruginosa strain tested, indicating that its bactericidal spectrum was distinct from that of R2 pyocin (Fig. 5). Like pyR2-PS17, pyocin yields of pyR2-P2 are comparable to those of wild-type R2: 2 × 10¹¹ to 5 × 10¹¹ pyocin particles per ml of lysate. Further purification of pyR2-P2 could be accomplished exactly as that developed for wild-type R2 (Materials and Methods).

FIG. 5.

Assays of the bactericidal activities of pyocins R2 and pyR2-P2 on P. aeruginosa and E. coli strain C. Activities were determined by semiquantitative assays with spotting of serial fivefold dilutions of the indicated pyocin preparations on P. aeruginosa or E. coli bacteria and observation after overnight incubation at 37°C. The row marked pyR2Δprf15 was a preparation from the production strain, PAO1 Δprf15, which lacks the expression plasmid.

Engineering a Y. pestis-specific pyocin.

Phage L-413c is a P2-like yersiniophage that infects most strains of Y. pestis as well as E. coli C (11) (GenBank accession no. AY251033). Most of the L-413c genome is very similar to P2, with the notable exception of the tail fiber gene H, which has diverged. While the N terminus of L-413c gene H does have considerable sequence similarity to its P2 counterpart, the central region of the L-413c tail fiber appears to be unrelated to the P2 tail fiber and is more closely related to portions of the phage T6 tail fiber (11). A gene fusion was constructed with the R2 prf15 portion encoding N-terminal amino acids 1 to 164 and the portion of L-413c gene H encoding the C-terminal portion (amino acids 158 to 913) of its tail fiber. The fusion gene was expressed in PAO1 Δprf15 as described above, with and without the putative chaperone gene G of L-413c. The purified pyocin particles, termed pyR2-413c, killed Y. pestis KIM D27 as well as E. coli C (data not shown) and thus exhibited a killing spectrum analogous to the host range of bacteriophage L-413c (11). The pyR2-413c pyocin did not kill any of the P. aeruginosa strains sensitive to R2, nor did it kill Yersina pseudotuberculosis, Yersinia frederiksenii, or Yersinia enterocolitica. Again, coexpression of the putative chaperone, encoded by the L-413c gene G, was required to produce a functional pyocin. Like the other engineered pyocins, yields of pyR2-413c are comparable to those of wild-type R2.

DISCUSSION

Redirection of the bactericidal activity of R-type pyocins to different target bacteria was achieved by swapping tail fibers or engineering chimeric tail fibers that bind to presumably different bacterial surface molecules. The described chimeric tail fibers that formed active pyocins resulted from many trials of empirically chosen fusion sites between R2 prf15 and portions of phage tail fibers. We were not able to predict, using secondary or tertiary structural analysis programs such as the protein homology/analogue recognition engine, Phyre (Imperial College, London, United Kingdom; http://www.sbg.bio.ic.ac.uk/phyre/), any particular junctions that might be prone to forming active chimeric tail fibers. However, once the analyses of a large series of fusions between R2 prf15 and gene H of P2 were completed, it became apparent that there were junctional sites of R2 Prf15 that were more likely than others to form functioning fusions. One such site of R2 Prf15 was at amino acid 164, and subsequently the gene fragment encoding amino acids 1 to 164 was tried early in each process of creating and evaluating novel chimeric tail structures (i.e., L-413c). In each example of creating novel pyocins, the formation of active particles consistently required coexpression of the cognate or a nearly identical chaperone specific for the C-terminal, receptor-binding portion of the tail fiber. Interestingly, for R2 pyocin the data suggested that R2 Prf16 was rate limiting for the formation of active R2 pyocin particles by even the wild-type PAO1 strain.

Although many R-type pyocin genes share common ancestry with bacteriophage genes, pyocins are not bacteriophages; they do not contain nucleic acid and cannot replicate. As protein particles, they do not have a genome. They kill by a mechanism unlike phage-mediated lysis, even though the initial event toward killing by both R-type pyocins and phages is binding to an accessible receptor on the target bacterium. In fact, as observed in these studies, R-type pyocins with tail fibers identical to those of a phage exhibited a killing spectrum broader than the host range of the phage from which their specificity-determining tail fibers had been derived. This phenomenon is likely due to DNA restriction or immunity systems of the host that can prevent phage infection or to other host-encoded functions required for phage replication and lysis of the host but not required for pyocin-mediated killing. The killing of target bacteria by R-type pyocins also does not result in the immediate release of endotoxins, as does lytic killing by phages and by many bactericidal antibiotics (13, 16, 44).

Tail fibers from the enormous number and diversity of bacteriophages in the environment provide a rich source of binding specificities for engineering pyocins. Thus, we anticipate being able to engineer R-type pyocins to target a broad range of bacterial pathogens, targeting many diverse surface structures. The application of phage display techniques may provide even greater diversity of binding specificities than exists naturally on phages. Given this dependency on specific surface receptors, R-type pyocins probably cannot be engineered to provide broad-spectrum antibacterial activity. However, these approaches based on fusing phage tail fibers to pyocin tail fibers could lead to an almost unlimited number of narrow-spectrum bactericidal agents. As for nearly any antibacterial agent, pyocin-resistant bacteria are likely to arise. In the case of R-type pyocins, this resistance most likely would be due to loss of a surface receptor (5, 10, 29, 34), a trait not known to be transmitted horizontally by the acquisition of a pathogenicity island or plasmid, a phenomenon that has proven to be troublesome with traditional antibiotics (1, 21). In addition, if the surface receptor molecule targeted by the agent is a virulence factor or fitness factor (lipopolysaccharide often is), then the inevitable emergence of resistance might often genetically compromise the pathogen's virulence (39).

From the time of their early use, broad-spectrum antibiotics were also recognized to cause unintended collateral damage to the normal flora, or microbiotae, resulting in secondary infections by bacterial and fungal pathogens resistant to the administered antibiotic (28, 45). This problem should prove to be minimal with a narrowly focused agent. R-type pyocins seem to be an attractive candidate with many of the above properties.