INTRODUCTION

For more than 60 years, it has been known that some bacteria produce phage tail–like bacteriocins (PTLBs): high-molecular-weight bactericidal protein particles that resemble, and are evolutionarily related to, the tail structures of various bacteriophages. They have since been found to be widespread among the eubacteria. However, most laboratories have not devoted their efforts to study PTLBs beyond a handful of publications each, and there has yet to be a review focusing just on this class of bacteriocin. This review covers the diversity and distribution of these structures, their biology, and their potential applications. It discusses only bacteriocins—proteins produced by bacteria that kill other types of bacteria. It does not cover other biological entities that share an evolutionary relationship with bacteriophage tail structures, such as type VI secretion systems (1–3), phage tail–like structures produced by bacteria that target insect cells (4–6), or phage tail–like “arrays” involved in interactions between bacteria and eukaryote hosts (7).

A few words about nomenclature. Many laboratories that have studied these bacteriocins simply use the term phage tail–like bacteriocins (8–16). We choose to continue that trend. Occasionally they are casually referred to as defective prophages, phage remnants, etc., but many, if not most, bacterial genomes contain phage-related sequences that do not function as bacteriocins. The term tailocin has recently arisen (6, 17, 18); however, this has caused confusion in that it has been used both for bacteriocins and for phage tail–like insecticidal particles, which have quite a different biological function (6). Indeed, Sarris et al. (19) pointed out that these entities should be classified separately from bacteriocins and coined the term phage-like protein translocation structure for these distinct biological entities.

Although investigators have appreciated that PTLBs are a distinct class of bacteriocins, they have almost all named them in the tradition of other, smaller, non–phage tail–like bacteriocins, which are given names corresponding to the bacteria that produce them. Examples include maltocin, of Stenotrophomonas maltophilia; carotovoricin, of Erwinia carotovora; diffocin, of Clostridium difficile; and pyocin, of Pseudomonas aeruginosa (pyocyanea). Unfortunately, this tradition creates additional confusion, because all of the other bacteriocins are named similarly. In the case of P. aeruginosa, this confusion has historically been sorted out by designating three types of pyocins. The R-type pyocins are PTLBs related to Myoviridae tail structures, and the F-type pyocins are PTLBs related to Siphoviridae tail structures. The S-type pyocins are lower-molecular-weight non–phage tail–like structures (20). We have retained the R- and F-type designations for other PTLBs in other bacteria; for example, F-type monocins are produced in Listeria monocytogenes and are related to Siphoviridae, and R-type diffocins are produced by C. difficile and are related to Myoviridae.

GENERAL BIOLOGY OF PHAGE TAIL–LIKE BACTERIOCINS

PTLBs are very large (2×10⁶–1×10⁷ Da) protein structures that consist of approximately eight to fourteen different polypeptide subunits and are structurally related to various phage tails (Figure 1). PTLBs are encoded in the genomes of bacteria in genetic clusters that are similar to bacteriophage tail structural modules (see Figure 2). Typically present in these loci are genes that encode the proteins that compose the particle structure as well as assembly proteins or chaperones that help catalyze the formation of the structure. Also usually present are regulatory genes that encode proteins responsible for transcriptional control of the genetic locus. A lysis cassette is often found, encoding proteins that function to release the particles into the surrounding medium (21).

The conditions in which PTLB production is triggered naturally have not been studied in any detail. In laboratory conditions, they can often be induced by DNA damage. In the case of the R- and F-type pyocins of P. aeruginosa, induction is linked to the SOS response; this is likely also true for the other PTLBs studied thus far.

PTLBs can be grouped into two types. R-type bacteriocins are contractile particles related to Myoviridae phage tails and consist of a long tube surrounded by a sheath; at one end is a complex baseplate structure in which receptor-binding proteins (RBPs), usually tail fibers, are attached. F-type bacteriocins are related to Siphoviridae phage tails, are noncontractile, and have a simpler structure consisting of just a tube with no sheath. F-type bacteriocins also possess RBPs that recognize target strains. Some bacteria, particularly P. aeruginosa, produce both R- and F-type bacteriocins.

PTLBs first bind to a target cell via RBPs. Once bound to the target, PTLBs cause rapid death. In the case of R-type bacteriocins, the mechanism likely involves contraction of the sheath and penetration of the core through the cell envelope, resulting in a channel or pore that decouples the membrane potential (see below). For F-type bacteriocins, the mechanism of killing has not been well studied but may also involve disruption of membrane gradients.

PTLBs function to kill competing bacteria. The bactericidal spectrum is often very narrow—usually a different subset of strains within the same species as the bacteria producing that particular PTLB. However, there are examples where a PTLB has bactericidal activity against a different species (12, 22–27). Sister cells of the strains that produce these particles are almost always resistant to the bactericidal activity of their own PTLB. The competitive advantage of producing PTLBs is likely very subtle; production provides kin a selective advantage over more distantly related competitors within the species. Killing via PTLBs is often thought of as microbial altruism, where cells must themselves die to release the particles that provide an advantage to sister cells.

THE CANONICAL PHAGE TAIL–LIKE BACTERIOCINS: R-TYPE AND F-TYPE PYOCINS OF PSEUDOMONAS AERUGINOSA

The first PTLBs discovered were probably the R-type pyocins (sometimes termed pyocines) from P. aeruginosa in 1952 (28), but they were not well studied until Kageyama & Egami began investigating them in 1962 (29). The first work began with biochemical studies that involved purifying PTLBs and characterizing their physical properties. Over the next two years Kageyama's group went on to conduct extensive characterization of R-type pyocins, which was published in a set of three papers (30–32). This fundamental work included good electron micrographs confirming that PTLBs are phage tail–like structures, outlined purification procedures, and, importantly, laid the groundwork for bactericidal assays including the survival assay showing that contact of a cell with a single pyocin particle can result in death.

Five different types of R-type pyocins, R1–R5, have been characterized based on killing spectra (33, 34). Observations of the bactericidal spectrum and analysis of pyocin-resistant mutants revealed an activity relationship among the different types. R5 has the broadest spectrum; it encompasses the spectrum of all the others plus additional strains. R2 has a narrower spectrum, but encompasses that of R4 and R3. R4 encompasses the range of R3. Although R5 encompasses the spectrum of R1, the spectrum of R1 is not related to the spectra of the others and therefore forms a branch. Kageyama (34) speculated that this relationship likely was related to different moieties of the cellular receptor, which is most likely lipopolysaccharide core (34–37).

F-type pyocins were first reported by Kuroda & Kageyama (38) as flexuous particles produced by some P. aeruginosa strains, sometimes along with R-type pyocins. In these studies they were able to separate the particles from R-type pyocins and show distinct bactericidal activity. They later went on to identify three types based on strain killing spectrum—F1, F2, and F3—and showed that they have immunogenic cross-reactivity with a bacteriophage KF1 (39–42).

The R2 pyocin genetic locus was mapped to trp on the P. aeruginosa PAO1 chromosome, specifically between trpE and trpGCD (43–45). The F2 pyocin locus mapped between the R2 locus and trpGCD. In 2000, the R- and F-type gene cluster of P. aeruginosa PAO1 was sequenced (46). This study demonstrated through sequence similarity that F-type pyocins are indeed related to Siphoviridae phage tails (phage λ) and R-type pyocins to Myoviridae phage tails (phage P2), with each possessing many gene homologs to their corresponding phages. This work refined the location of the gene cluster to between trpE and trpG, and polymerase chain reaction studies showed that genetic loci of other high-molecular-weight pyocins are located in this position as well. It is notable that in PAO1, R- and F-type pyocins are encoded next to each other and share common regulatory genes and common lysis genes. In fact, the arrangement can be viewed as the R-type gene cluster being inserted between the holin and lysin genes of the F-type gene cluster (Figure 2).

R- and F-type pyocins, as well as most other PTLBs, are induced in the laboratory by agents that cause DNA damage, such as mitomycin C, and they have always been assumed to be under control of the SOS response. Matsui et al. (45) confirmed this in 1993 and proposed a model for gene regulation. Two pyocin-specific regulators, PrtR (repressor) and PrtN (activator), are involved, along with RecA. PrtN is a positive regulator that activates transcription by binding to P box sequences upstream of the pyocin genes. Under normal growth conditions prtN is repressed by PrtR. When DNA is damaged, activated RecA can cleave PrtR, resulting in production of PrtN, which activates the genes. No other PTLB gene cluster has been studied experimentally to determine the mechanism of gene regulation.

PHAGE TAIL–LIKE BACTERIOCINS DESCRIBED IN OTHER BACTERIA

Numerous examples of PTLBs have been described throughout the eubacteria, although none have been studied in as much detail as the R- and F-type pyocins. One of the earliest types of PTLBs to be characterized after pyocins was the vibriocins, from Vibrio cholerae (47). This work produced electron micrographs that clearly showed R-type contractile bacteriocins. High-molecular-weight boticins were described in 1970 (48). These investigators did not conduct electron microscopy studies, but they separated two bacteriocin activities, one of which had a very high estimated molecular weight and was likely a PTLB. An excellent study of an R-type PTLB from Rhizobium lupine was done by Lotz & Mayer in 1972 (8). This study included electron micrographs of contracted and uncontracted structures as well as bacteriocin particles attaching to target cells.

In more recent years, several PTLBs were isolated from various gram-negative bacteria, including Xenorhabdus nematophilus (9), Xenorhabdus bovienii (49), E. carotovora (10), Yersinia enterocolitica (11), Serratia plymithicum (12), Budvicia aquatica (13), Pragia fontium (13), Pseudomonas fluorescens (14), S. maltophilia (15), Pseudomonas syringae (50), Proteus vulgaris (22), and Proteus mirabilis (51). These are all contractile R-type bacteriocins. Carotovoricin Er of E. carotovora is notable in that downstream of its tail fiber gene is a gene encoding an invertase, which inverts the DNA encoding the distal portion of the tail fiber, resulting in a shift in bactericidal spectrum and production of two different particle specificities (10).

Besides F-type pyocins, the only other well-studied F-type PTLBs are the monocins of L. monocytogenes. These were first used for typing Listeria strains (52–54). Very recently, Lee et al. (16) identified the monocin genetic locus, cloned and expressed the monocins in Bacillus subtilis, identified the RBP, and engineered the target specificity. Unlike the F-type pyocins of P. aeruginosa, which are related to λ tail structures, monocins are more closely related to the tail structure of TP901-1-like phages, which, although they are also Siphoviridae phages, have a much different baseplate structure than λ-like phage tails (55). Little is known about monocin gene regulation, although the monocin gene cluster encodes several putative regulatory proteins and is probably more complex than that of pyocins. The entire cluster, however, could be placed under inducible regulation with an exogenous promoter. The monocin gene cluster is relatively small, and although monocins are related to the tail structures of Listeria phages, particularly A118, they have diverged considerably (Figure 2), suggesting a more ancient relationship with phages.

The diffocins of C. difficile are the only gram-positive contractile R-type PTLBs that have been studied in any detail (56, 57). The biology and mechanism of action of these structures are likely to be similar to those of R-type pyocins. The gene cluster is shown in Figure 2 and is related to the tail structural modules of C. difficile phages such as φ119. Regulation is likely also controlled by SOS, but LexA may serve as the repressor because there are putative binding sites. Diffocins are interesting in that the RBP is a very large protein, of ∼200 kDa, giving the structure a flower-like appearance.

STRUCTURE AND MECHANISM OF ACTION

Structural studies beyond basic electron microscopy have been done in detail only with the R-type pyocins; however, much can be inferred from comparison of the components of the various PTLBs with their related phage tails.

R-type pyocins consist of a core that is a tube of a polymer of a single polypeptide (Figure 3). Attached to the core is a pointed trimeric tailspike protein that has an iron moiety at the tip (58). The core is surrounded by a sheath, which is also a polymer of a single polypeptide. The sheath/core assembly in uncontracted form is ∼120 nm in length (59). At one end is the complex baseplate structure, which consists of seven separate polypeptides (P. Ge, J. Avaylon, D. Scholl, J. Miller & H. Zhou, manuscript in preparation). Attached to the baseplate are tail fibers, which are composed of a single polypeptide and serve as the RBP. Based on similarity with phages and preliminary crystallography data, the tail fibers are inferred to be a homotrimer of which six copies are present on the particle (P. Leiman, personal communication). Sequence comparison of the tail fibers between the five different R-type pyocins showed that the C terminus has diverged. Based on this observation it was hypothesized that this divergence is responsible for different target receptor specificities and that this variable region is the part of the RBP that interacts with different moieties on the cell lipopolysaccharide receptor. This hypothesis has been confirmed by complementation experiments that involved swapping the tail fiber types (60). It was also shown that a small chaperone protein is required, most likely for tail fiber assembly, and is not part of the structure. Each pyocin type encodes a different chaperone corresponding to the divergent RBPs.

Cryoelectron microscopy structures of the sheath/core assembly in the contracted and uncontracted states have revealed the mechanism of contraction and interactions of the sheath and core (59). During contraction, the arrangement of sheath subunits undergoes a gross conformational change, with a rotation of 85° and expansion of the sheath diameter, which drives the core downward after the particle binds a cell. This study also revealed a significant functional adaptation. The inner core of the tube has a net negative charge that would not be well suited for DNA translocation but would be better adapted for translocation of positively charged ions. Its most closely related phage, PS17, has a positive inner charge.

The mechanism of killing (bactericidal activity) has only been studied in any detail for the contractile R-type pyocins. It was shown early on that contact with a single pyocin particle is sufficient to kill a bacterial cell, suggesting that the killing mechanism is quite potent (32). Uratani (61) and Uratani & Kageyama (62) later showed that contact of pyocin with bacterial cells produced changes in membrane permeability. Uratani & Hoshino (63) then showed that R-type pyocin caused a loss of membrane potential and respiration, an event that would result in cell death. Morse et al. (24) showed that bactericidal activity of R pyocin from P. aeruginosa against Neisseria gonorrhoeae resulted in a halt of oxygen uptake and macromolecular synthesis.

Based on these observations, a model for mechanism of action is as follows (Figure 3). Particles first bind to the cell surface lipopolysaccharide via tail fiber RBPs. In one example, R3, calcium is required for binding (64). Once bound, the sheath contracts, forcing the internal core into the cell envelope, led by the iron-tipped tailspike, through the inner membrane. This leaves a channel through which protons and other small ions flow, disrupting membrane concentration gradients. Thus, the mechanism of killing of R-type bacteriocins is related to the mechanism in which Myoviridae phages translocate DNA into the cell after binding. But instead of DNA entering the cell, a pore is created. This mechanism is also probably similar to that described for the bactericidal activity of phage T4 ghosts, which are phages that lack DNA (65). The action of R-type pyocin against N. gonorrhoeae resulted in cell lysis (24). However, this could be due to excess cellular lytic enzymes produced by that species; killing of P. aeruginosa does not result in lysis until many hours after death. Lysis is likely not part of the killing mechanism, but rather occurs due to later degradation of the cells from cellular enzymes.

F-type PTLBs do not have a contractile mechanism; they possess a simple tube that presumably does not penetrate the inner membrane. However, as with R-type PTLBs, a single particle can kill a cell—particularly with F-type monocins (16)—suggesting that a similar channel is formed in the inner membrane, resulting in disruption of respiration. However, this is still only speculation.

All PTLBs, both R-type and F-type, encode a tape measure protein that is part of the structure, probably occupying the tube. This protein, which is also found in both Myoviridae and Siphoviridae phages, is responsible for determining the length of the tube (and sheath). Tape measure proteins have membrane-spanning regions, and it has been recently shown in phages that these proteins insert into the inner membrane and are involved in DNA translocation (66). Perhaps in PTLBs they play a role in forming a pore in the inner membrane, in which case the mechanism of killing of both R- and F-type PTLBs could be quite similar.

There is one report of an R-type pyocin containing single-stranded DNA (67). However, the sequence of this DNA is not related to any pyocin genes; it is instead related to the sequences of the genomes of some filamentous phages. It is possible that this DNA came from a contaminating phage. No evidence of DNA was noted in cryoelectron microscopy studies, and it is assumed that PTLBs in general are composed only of proteins.

EVOLUTIONARY RELATIONSHIP WITH PHAGE TAILS

Very soon after their discovery, PTLBs were shown via electron microscopy to be structures that resembled phage tails, and since that time they have been believed to be evolutionarily related to these phage organelles. Kageyama et al. (68) noted particular structural similarities between R-type pyocins and Pseudomonas phage PS17 in electron microscopy studies. Later they showed serological cross-reactivity between PS17 and R-type pyocins as well as genetic relatedness (69). Shinomiya (70) demonstrated phenotypic mixing of phage PS17 and pyocins and interchangeability with specificity components, most likely tail fibers. It was also found that PS17 tail structures themselves had bactericidal activity, albeit at about 100× less potency (71) (possibly due to a nonideal internal charge; see above). Serological cross-reactivity was also shown with R-type pyocins and phages φCTX and PS20 (72). For F-type pyocins, a phage was discovered, KF1, that had serological cross-reactivity, and it was speculated that F-type pyocins had an evolutionary relationship with Siphoviridae phage tails. The phage-pyocin relationship was solidified when Nakayama et al. (46) sequenced the PAO1 pyocin locus (encoding both R- and F-type pyocins) and concluded that there were evolutionary relationships between R-type pyocins and P2-like phages and between F-type pyocins and λ-like phages.

Based on these observations and the fact that PTLBs can be produced under situations when lysogenic phages are often induced (induction of SOS) and released by lysis, it has often been causally assumed that PTLBs originated from defunct prophages that lost their capsid genes and the ability to replicate DNA. Further adaptions would have then occurred that made them more efficient at disrupting membrane potential and killing cells rather than at transducing DNA. However, structural comparisons of contractile phage tails, R-type pyocins, and type VI secretion systems suggest a common cellular ancestry that very likely predates tailed phages (1, 73, 74). Thus, an alternate model is that these various biological entities coevolved from a common cellular ancestor. Given that genetic exchange between these entities has likely occurred—and continued to occur even after their functional divergence—it may be impossible to determine which might have come first.

It is quite remarkable that multiple PTLB types exist, each having common ancestry with a different phage tail structure type. For example, R-type pyocins are related to PS17-like phage tails, F-type pyocins to λ-like phage tails, F-type monocins to TP901-1-like phage tails, and diffocins to φCD119-like phage tails. It appears that multiple structural motifs of phage tails and PTLBs have evolved in parallel. Another interesting observation is that a Mu-like PTLB has been found in P. syringae (50). This PTLB is in a genome position analogous to that of the R- and F-type pyocins; this particular locus appears to be a hot spot for acquisition of these structures. When considering the tremendous amount of diversity and the dynamic nature of the bacteriophage/PTLB populations, it may be expected that tail and bacteriocin functions have evolved many times, further clouding the origin question.

BIOLOGICAL FUNCTION

While non–phage tail–like bacteriocins have been studied in some detail, the biological role of PTLBs has received relatively little attention. PTLBs do have potent bactericidal activity; in several cases it has been shown that a single particle is sufficient to kill a single cell. This means that in spite of their large size they are more efficient at killing bacteria per unit mass than are smaller bacteriocins. However, the target bacteria are typically very specific; PTLBs do not serve as general antibacterial weapons. Producing them likely provides the population a more subtle biological advantage. Cells are typically resistant to the PTLBs produced by their sister cells. However, an individual cell must sacrifice itself to produce them because they can be released into the medium only by lysis. It can be viewed as a form of altruism; cells produce PTLBs to provide sister cells a competitive advantage. Given the high potency of PTLBs, it may be that only a small fraction of a population needs to lyse and release particles to protect kin and give the population a selective advantage.

There are a few experimental studies. In P. aeruginosa it was shown in mixed culture experiments that R-type pyocin production conferred a competitive growth advantage (75). Also, both R- and F-type pyocins (as well as S-type pyocins) were upregulated in mixed biofilm experiments, and PTLB production likely shaped the population dynamics (76).

Perhaps the most interesting biological studies of a PTLB were performed with the R-type xenorhabdicin of Xenorhabdus nematophila (77). This bacterium forms a mutualistic relationship with a nematode that can infect insect guts. Photorhabdus luminescens is a bacterium that can potentially antagonize the nematode growth and can therefore be a competitor to X. nematophilia. P. luminescens is sensitive to xenorhabdicin; production of this PTLB was required for X. nematophilia to gain a competitive advantage over P. luminescens. An R-type PTLB from X. bovienii also provided a similar competitive advantage (78).

A recent study showed that lysis of P. aeruginosa caused by the induction of R- and F-type pyocins is important for production of biofilms and membrane vesicles (79). Extracellular DNA released in this event is important for formation of the biofilm matrix. Interestingly, it is only lysis that is important in this model; the bactericidal activity of the pyocins does not seem to be relevant. However, in a natural environment where competing strains may be present, both the formation of extracellular matrix and the killing of neighboring bacteria may provide a double selective advantage.

APPLICATIONS OF PHAGE TAIL–LIKE BACTERIOCINS

The first example that PTLBs can be effective as anti-infectives was provided in 1969 when Bird & Grieble (80) showed that R-type pyocins could rescue chick embryos infected with a sensitive strain of P. aeruginosa. Three years later, Merrikin & Terry (81) showed that R-type pyocins had efficacy against systemic P. aeruginosa infections in mice. In these studies, animals were infected intravenously with high doses (10⁷ CFU) and treated intravenously. Pyocin could be administered as late as 6 h after infection and still rescue animals. Efficacy of pyocin treatment was noted on two strains, but a third was nonresponsive, even though it was sensitive to the pyocin; perhaps the very high inoculum was too difficult to overcome if a strain was even slightly more virulent. Haas et al. (82) performed a prophylactic study in 1974 in mice. A single intraperitoneal pyocin dose could protect animals from a lethal intraperitoneal P. aeruginosa challenge when administered up to 4 days prior to infection. Protection was not seen in an insensitive strain, ruling out any effect not due to bactericidal activity.

More than 30 years passed before the next in vivo study showing that an R2 pyocin could be used to treat infectious disease in a lethal P. aeruginosa mouse peritonitis model (83). In this study, which was performed using purified and better quantified material, mice could be rescued by either intraperitoneal or intravenous administration of single doses of R2 pyocin. A dose-response curve indicated that microgram quantities of material were sufficient to rescue animals. A time-of-administration experiment showed that beyond 4 h after infection treatment began to fail in terms of rescuing animals; however, it did extend survival, suggesting that a multidose regimen could be effective. Neutralizing activity to pyocin was noted in serum of treated animals after several weeks, suggesting that such a treatment could only be used once, perhaps in salvage therapy applications.

PTLBs have very focused bactericidal spectra, mainly targeting competing strains within the species of the producer bacteria. This narrow spectrum is ideal in situations where one needs to target a specific pathogen without harming beneficial flora. However, in the case of natural PTLBs, this narrow spectrum is perhaps too limiting, and it is unlikely that a natural bacteriocin can be identified that kills enough isolates of a given pathogen to be useful. This limitation has been circumvented by engineering the natural spectrum. The tail fiber RBPs that confer specificity to the cellular receptor can be retargeted by genetic engineering to bacterial species of choice. This was first shown in 2008 through the creation of fusions between the N terminus of R2 tail fiber and the C terminus of the tail fiber of the myobacteriophage P2, which has a tail structure closely related to R2 pyocin (60). Pyocin particles that incorporated these fusion proteins killed the Escherichia coli strains that served as hosts to P2.

The idea was later expanded to specifically target E. coli O157:H7 (84). In this case the E. coli O antigen was targeted using an RBP from a bacteriophage (φV10) specific for this strain. φV10 is a Podoviridae bacteriophage and encodes as its RBP a tailspike protein that specifically binds and degrades O antigen. This demonstrated that RBPs from phage tail structures unrelated to the PTLB could be adapted to function on the bacteriocin. This approach was further demonstrated by targeting E. coli O104 (85). In this case the PTLB was engineered with only the genome sequence of the target bacteria in hand; a phage RBP was identified from a prophage in the sequenced genome, and the coding sequence was synthesized to make a pyocin-RBP fusion. This study was notable in that the agent was generated within weeks after identifying a new disease outbreak strain. It also showed that new targeting motifs can be mined from bacterial genomic data. A rabbit enterohemorrhagic E. coli infection model of E. coli O157 was used to demonstrate efficacy in treating full-blown diarrhea using orally administered bacteriocin (86). In this study, pathogen counts were reduced by several logs, and symptoms of disease were greatly reduced. Also shown in this study was that the killing by the PTLB did not result in production and release of Shiga toxin, a problem encountered when treating these infections with traditional antibiotics.

R-type PTLBs from C. difficile (diffocins) have been similarly engineered for target specificity. This was first accomplished by switching the natural RBPs from two different diffocins, thus changing the spectra (56). It was later shown that PTLB–phage RBP fusions could be made by using an approach similar to that used for R2 pyocin. Engineered diffocins have been shown in a mouse model to prevent transmission of disease and colonization of C. difficile when bacteriocin was administered in drinking water (57).

There is one published example of engineering an F-type bacteriocin, a monocin of L. monocytogenes. In this case, the related bacteriophage A118 served as the RBP donor to make the fusion (16). This engineered monocin has a spectrum related to that of the host range of A118 and is distinct from any natural monocin. A combination of this hybrid monocin with a wild-type monocin covers all of the relevant L. monocytogenes clinical isolates, and it was suggested that this cocktail could be used in food safety applications.

The above studies demonstrate that PTLBs can be engineered to target very specific bacterial pathogens without harming normal, beneficial microbiota. They also show that, if administered correctly, they can be used as highly effective therapeutics. These bacteriocins could potentially be used to treat a wide range of bacterial infections, especially those that are now resistant to traditional antibiotics. Although systemic administration of these large proteinaceous structures is likely to result in the production of neutralizing antibodies, which could limit their usefulness, oral administration to target pathogens residing in the gut has great potential.

Because of their strain specificity, PTLBs have been used for bacterial typing. This was particularly developed for P. aeruginosa and L. monocytogenes (52–54, 87). However, many of the studies relied on lysates of strains that likely consisted of mixtures of PTLBs and other bacteriocins, and possibly phages. Typing systems have also been examined using purified R- and F-type pyocins (88). Other groups have discovered that R-type pyocins, which have bactericidal activity against some Neisseria spp., can be used to type these bacteria (26, 89).

CONCLUDING REMARKS

PTLBs are widespread and diverse among bacteria. They likely have evolved multiple times in parallel, and the fact that some (monocins in particular) have diverged considerably from any known phage suggests that they could be quite ancient. These specialized bacteriocins must provide a strong selective advantage for retention; they are not just phage remnants that still possess a side activity. The selective advantage that they provide to cells likely involves subtle interactions with related bacteria in tight communities; it is unlikely that they are a general defense mechanism. A considerable amount of work will be needed to fully understand the biology of PTLBs, and it is likely that they play different roles among different bacterial species depending on the lifestyle of the bacteria.

In spite of their large size, PTLBs are highly potent. A single particle can kill a cell, and microgram quantities can rescue mice in experimental infection models. Although their large size presents some pharmacokinetic issues, and immunogenicity needs to be managed, there are clearly indications where they can be efficacious in treating infection. Particularly intriguing are oral applications where a highly targeted agent is needed to avoid disruption of the normal microflora, and salvage therapy applications in a world that is entering the postantibiotic age due to multidrug resistance.

disclosure statement

The author is a stockholder of AvidBiotics Corp.