Bacteriophages for the treatment of Pseudomonas aeruginosa infections


David R. Harper, Biocontrol Limited, Colworth Science Park, Sharnbrook, Bedfordshire MK44 1LQ, UK. E-mail:


Bacteriophages were first identified in 1915 and were used as antimicrobial agents from 1919 onwards. Despite apparent successes and widespread application, early users did not understand the nature of these agents and their efficacy remained controversial. As a result, they were replaced in the west by chemical antibiotics once these became available. However, bacteriophages remained a common therapeutic approach in parts of Eastern Europe where they are still in use. Increasing levels of antibiotic-resistant bacterial infections are now driving demand for novel therapeutic approaches. In cases where antibiotic options are limited or nonexistent, the pressure for new agents is greatest. One of the most prominent areas of concern is multidrug-resistant Gram-negative bacteria. Pseudomonas aeruginosa is a prominent member of this class and is the cause of damaging infections that can be resistant to successful treatment with conventional antibiotics. At the same time, it exhibits a number of properties that make it a suitable target for bacteriophage-based approaches, including growth in biofilms that can hydrolyse following phage infection. Pseudomonas aeruginosa provides a striking example of an infection where clinical need and the availability of a practical therapy coincide.

The nature of bacteriophages and bacteriophage infections

Bacteriophages are viruses that infect bacteria. They are obligate parasites, replicating within the bacterial cell. Infection with most bacteriophages is restricted to particular strains within a single bacterial species, although others may infect multiple species (or even genera).

More than 90% of bacteriophages have large, double-stranded DNA genomes located in heads with icosahedral symmetry, with tails of varying lengths (Fig. 1). All fall within the order Caudovirales and belong to three major morphological groups. These are the families Myoviridae (with long, rigid, contractile tails), the Siphoviridae (with long, flexible, noncontractile tails) and the Podoviridae (with short, noncontractile tails). The morphology and genome type of the remaining bacteriophage families are highly variable, and they may have DNA or RNA genomes. One notable group with single-stranded DNA genomes (Inoviridae) appears as long filaments (Maniloff 2006).

Figure 1.

 The tailed bacteriophages.

The outcome of infection of the bacterial cell may vary. Many bacteriophages undergo a lytic infectious cycle, with infection resulting in rapid lysis and death of the cell within a very short time. Typically, this will result in the release of hundreds of new, infectious virus particles within minutes or hours, a process that can be repeated as long as their bacterial host is present in sufficient numbers to support replication (Harper and Kutter 2008).

However, two types of bacteriophage life cycle exist (Fig. 2). In the lysogenic cycle, the so-called temperate bacteriophages attach to and invade bacterial cells. They integrate their DNA into the host chromosome and are replicated along with the host cell DNA and transmitted to each daughter cell. This latent, or prophage, DNA can be activated by specific stimuli, such as DNA damage to cause the bacterial host to produce new phage that lyse the host cell – a lytic mode of infection. For some bacteriophages, host chromosomal DNA may be packaged into bacteriophage particles during bacteriophage replication instead of the bacteriophage genome. This can result in high levels of horizontal gene transfer within the bacterial population, a process known as generalized transduction. Specialized transduction can also occur in which particular portions of the genome are mobilized following recognition of specific genetic sequences by phage enzymes. In addition to the lytic and lysogenic types of bacteriophage infection, other types may exist. The filamentous bacteriophages typically cause a persistent infection of bacterial cells that does not kill the host but results in continued excretion of viral particles (Harper and Kutter 2008).

Figure 2.

 Lytic and lysogenic infections.

The use of bacteriophages as highly specific antimicrobial agents is widely documented in the literature (Kutter and Sulakvelidze 2005; Harper and Kutter 2008). For therapeutic uses, obligately lytic bacteriophages are highly desirable. Because they result in rapid killing of their target host cell, bacteriophage numbers increase rapidly and transduction is relatively rare. DNA sequencing of bacteriophage genomes is now used to confirm both identity and the absence of undesirable elements, such as functional lysogenic components or bacterial toxins. Such toxin genes are known to be specifically associated with some bacteriophages, for instance the Shiga toxins of Escherichia coli, which would be of real concern if present in a therapeutic bacteriophage formulation.

Early attempts to use bacteriophages for therapy of bacterial infections were compromised by a lack of understanding of the nature of the agents involved. Work in this area dates back to 1919, but as late as 1941, authoritative reports were still challenging the theory that bacteriophages were viral in nature (Krueger and Scribner 1941).

Although a range of companies advertised bacteriophage preparations for sale in the 1920s and 1930s, they were of dubious quality and in addition were often used when no bacterial infection was present (Fig. 3). The introduction of effective chemical antibiotics following detailed clinical evaluations led to the therapeutic use of bacteriophages in the West being curtailed, although clinical use continued in the countries of the former Soviet bloc. Such use has provided a large amount of indicative data supporting the effectiveness of bacteriophage therapy with few reported adverse events. Claims for safety are further supported by the exposure of humans to high levels of bacteriophages via everyday activities because of the ubiquitous nature and high numbers of bacteriophages in the environment (Bergh et al. 1989). However, individual case studies and anecdotal evidence are insufficient to meet current regulatory standards for approval of therapeutic bacteriophages in most jurisdictions, including Europe and the United States.

Figure 3.

 Advertisement for a bacteriophage therapeutic from the 1920s. Reprinted with the kind permission of Dr James Soothill.

The increasing prevalence of multiple antibiotic-resistant pathogens has encouraged a recent re-examination of bacteriophage therapy, and this is thankfully being increasingly supported by work compatible with the requirements of evidence-based medicine (Wright et al. 2009; Hawkins et al. 2010).

Clinical trials

Bacteriophage therapy was used on millions of people prior to the introduction of antibiotics in the 1940s, but unfortunately, efficacy studies that were carried out were largely before the introduction of randomized controlled trials in 1948 (Committee 1948). The evolution and global spread of multidrug-resistant bacterial pathogens has encouraged the re-examination of bacteriophage therapy, increasingly supported by work compatible with the requirements of evidence-based medicine (Wright et al. 2009; Hawkins et al. 2010). However, the extensive, ongoing use of bacteriophage therapy in Eastern Europe has not been subject to such scrutiny, and therefore, data from these areas are not sufficient to support the development of therapeutics in the modern regulatory environment.

Although some lobbyists are pushing for an easing of regulation for bacteriophages on the grounds of their historical use and the resulting evidence for safety and efficacy, it must be remembered that many approaches do have such accumulated evidence and yet new therapeutics are subject to the full exigencies of regulation. The use of vaccines, for example, dates back more than 200 years (and possibly more than 2000), but clinical trials of novel agents are still both demanding and time-consuming. Once bacteriophage therapeutics have established themselves as both safe and effective to the satisfaction of the relevant regulatory bodies, it may be possible to seek some flexibility in the approval process based on proven safety (as with the reformulation of the influenza vaccine), but until that time, the regulatory pathway is likely to remain similar to that of existing medicines.

Moving on from the historical work, some clinical trials have now been conducted to modern standards in both the European Medicines Agency and the United States Food and Drug Administration jurisdictions (Table 1). Human safety (phase 1) trials have been undertaken in Europe and in the United States using bacteriophages specific for a range of bacterial infections, including E. coli, Staphylococcus aureus and (most commonly) Pseudomonas aeruginosa (Bruttin and Brussow 2005; Merabishvili et al. 2009; Rhoads et al. 2009). These have been reported in varying degrees of detail, while an apparent 1999 trial of bacteriophages targeting vancomycin-resistant Enterococcus does not seem to have been reported formally. However, like live vaccines, bacteriophages have the property of multiplying in vivo, generating therapeutic doses locally and to levels which can be considerably higher than the input dose. As a result, the value of safety trials in which the bacteriophages cannot replicate (because of the lack of bacterial target) is limited because an in vivo therapeutic dose level is not attained. Trials to date have used between 105 and 3 × 109 plaque-forming units of individual bacteriophages, with from one to eight bacteriophages administered (Bruttin and Brussow 2005; Merabishvili et al. 2009; Rhoads et al. 2009; Wright et al. 2009). This is, in absolute terms, a very small amount of material. For example, in the trial against Ps. aeruginosa conducted by Wright et al. (2009), the input dose of 6 × 105 bacteriophages corresponded to only 2·4 ng of protein. Even at the highest doses used to date, the level of bacteriophage material administered is far more like the levels used in vaccines than for conventional chemical antibiotics, where individual doses usually range from milligrams to grams.

Table 1.   Trials of bacteriophage therapeutics in Western Europe or USA
  1. A large efficacy trial begain under local regulation in Bangladesh in 2009, conducted by the food multinational Nestlé.

  2. MRSA, Methicillin-resistant S. aureus.

Investigator-led (bacteriophages from Intralytix)2009USAPseudomonas, MRSA, Escherichia coliLeg ulcersRhoads et al. (2009)
Belgian military2009BelgiumPseudomonas, MRSABurnsMerabishvili et al. (2009)
Phase 2
 Biocontrol (1/2)2007UKPseudomonasEar infectionWright et al. (2009)

Moving on from trials that provide supportive safety data, the first clinical trial of bacteriophages as a veterinary therapeutic, in this case to treat ear infections in companion dogs, reported no safety concerns and demonstrated short-term efficacy (Soothill et al. 2004; Hawkins et al. 2010). Using the same mixture of bacteriophages, the first fully regulated, placebo-controlled, double-blind, randomized phase II clinical trial of the efficacy of a bacteriophage therapeutic was completed in 2007 and reported a successful outcome against long-term infections with Ps. aeruginosa, despite using only a single dose of input bacteriophages (Wright et al. 2009). Both these human and veterinary trials were based on earlier work targeting Ps. aeruginosa infection, which reported bacteriophage replication following treatment (Marza et al. 2006). Thus, there is now a significant body of work up to and including clinical trials that support the use of bacteriophage therapeutics for Ps. aeruginosa infections.

This leads to the question of why this therapeutically intractable organism has been chosen so often for such work (Table 1), and what is it that makes it suitable for such a therapeutic approach?

Why choose Pseudomonas aeruginosa as a target for bacteriophage therapeutics?

Antibiotic resistance

The main driver in the development of novel therapeutics for Ps. aeruginosa infections is the declining efficacy of conventional chemical antibiotic therapy, generating both clinical need and commercial opportunity. Pseudomonas aeruginosa has been described as a ‘phenomenon of bacterial resistance’ (Strateva and Yordanov 2009). It routinely exhibits multiple mechanisms of resistance, including efflux pumps, antibiotic degrading or modifying enzymes of a variety of types, and limited membrane permeability. Added to these is its growth in dense biofilms, reducing the efficacy of multiple antibiotics.

If an infection is routinely and adequately controlled by existing therapies, then there is little need for new approaches and little motive for their development. In the case of Ps. aeruginosa (and the multiple antibiotic-resistant Gram-negative infections as a whole), there is both a shortage of effective antibiotics and a dearth of new agents coming through the development pipeline (Theuretzbacher 2009). Thus, both the clinical and commercial drivers for the development of novel approaches are in place.

Biofilm formation

In addition to the antibiotic resistance mechanisms mentioned above, Ps. aeruginosa is highly active in the formation of biofilms. It is known that these can decrease the activity of antibiotics to the point where effective concentrations cannot be achieved in vivo (Moskowitz et al. 2004). Antibiotic resistance in biofilms is driven by a number of mechanisms. These include impermeability, sequestration of antibiotics by components of the biofilm (Mah et al. 2003) and the presence of metabolically inactive ‘persister’ cells within the biofilm matrix (Keren et al. 2004). In some cases, biofilm formation appears to be induced by antibiotics, apparently in a protective response by the bacteria (Hoffman et al. 2005; Elliott et al. 2010).

Because biofilms are now thought to be involved in the majority of infections (Flemming 2008), it is perhaps unsurprising that bacteriophages have evolved to infect bacteria in this context. The advantages of bacteriophages over chemical antibiotics in treating biofilm infections are several. First, while antibiotics passing over the surface of a biofilm will only affect bacteria at that surface, bacteriophages infecting cells at the surface will then replicate, generating a high concentration of antibacterial agent locally at the site of the infection, which can then progressively destroy the biofilm. Second, bacteriophages can carry (Verma et al. 2010) or, in some cases, induce the expression, by the bacterial host (Hanlon 2007), of enzymes that dissolve the biofilm matrix. In the latter case, this appears to be a flight response to invasion of the biofilm by bacteriophage. Finally, bacteriophages are able to infect persister cells and remain with them until they become metabolically active, at which point the bacteriophage can also become active and then destroy the newly revived cell (Pearl et al. 2008). The ability of bacteriophages to weaken and even destroy the biofilm matrix can also aid the penetration of other agents, raising the possibility of synergistic combination with antibiotics (Verma et al. 2010). Because Ps. aeruginosa is important in producing the biofilm matrix, such synergy could also extend to other species present within a polymicrobial biofilm. While they might not be targeted by the bacteriophage, they could become more susceptible to antibiotics as a result of bacteriophage-induced biofilm disruption.

Pathological significance of the target infection

Bacteriophages are highly specific. The vast majority infect a single species of bacterium and usually only a subset of strains within that species (requiring the use of a mixture of bacteriophages even for use against a single target species). This means that it is necessary to know that one or a small number of bacteria are responsible for significant pathological effects and that clearing that infection alone has the potential to result in clinical benefit. With Ps. aeruginosa, there are a number of conditions where this is the case, including late-stage otitis media (Wright et al. 2009), cystic fibrosis lung infections (Murray et al. 2007), ventilator-associated bacterial pneumonia (Paterson 2006) and burns (Kerr and Snelling 2009). In the case of Ps. aeruginosa, it has also been proposed that it has a key role in the perpetuating chronic wound infections (Bjarnsholt et al. 2008) opening up additional therapeutic modalities.

Limited geographical variation

Pseudomonas aeruginosa is a ubiquitous organism found in myriad ecological niches around the world. In consequence, while there is a high degree of variation in the susceptibility of isolates from a particular area to individual bacteriophages, individual strains of bacteriophages targeting Ps. aeruginosa seem to be globally widespread (Ceyssens and Lavigne 2010). As a result, there is little evidence that bacteriophages from one location will be significantly less effective against those from elsewhere. This may not be the case for other species.

With Staph. aureus, for example where most bacterial variation appears to be locally generated (Nubel et al. 2008; Harris et al. 2010), it could be expected that bacteriophage from one locale may be unable to infect hosts from another. If this were shown to be the case, then the prospect for globally effective bacteriophage therapeutics for Staph. aureus disease would be poor.

Availability of bacteriophages

For some bacterial species, it can be both technically challenging and extremely time-consuming to isolate bacteriophages. For example, isolation of bacteriophages targeting Mycobacterium tuberculosis can be technically demanding because of the slow growth of the host (McNerney 1999), while isolation of bacteriophages for Streptococcus mutans has historically proven difficult (Bachrach et al. 2003).

Lytic nature of bacteriophages

It is generally conceded that bacteriophages intended for use as therapeutics need to be obligately lytic in nature, both to ensure killing of the target bacteria and to reduce transfer of bacterial genes (Harper and Kutter 2008). For some bacterial species, even though bacteriophages may be isolated, they are rarely lytic. For example, bacteriophages specific for Clostridium difficile appear to be predominantly if not entirely lysogenic in nature (Goh et al. 2005). In contrast, for Ps. aeruginosa, lytic bacteriophages are readily isolated from environmental sources.

Lack of gene transfer and associated toxins

In a number of cases, bacteriophages are associated with the transfer of bacterial toxins. By enhancing the virulence of their bacterial host, lysogenic bacteriophages may be able to confer a selective advantage on their host favouring their reproduction. Examples of such bacteriophages include the filamentous bacteriophage CTXphi, which transfers the exotoxin in Vibrio cholerae (Davis and Waldor 2003), the six bacteriophages that carry the Shiga toxins carried by E. coli (Gyles 2007) and the bacteriophages that mediate lateral gene transfer of virulence determinants in the Shigellae (Yang et al. 2005). In addition, the Staph. aureus pathogenicity islands (SaPis), which are mobile pathogenicity elements of Staphylococcus spp., are transferred through an intimate relationship with specific bacteriophages that adapt their structure to carry these small (15–17 kb) DNAs (Novick and Subedi 2007).

These findings provide a powerful driver to avoid the use of temperate bacteriophages in therapeutic applications. Thankfully, the absence of lysogenic cassettes can be routinely confirmed by both microbiological and sequence-based approaches. The latter also allow confirmation of the absence of any toxin genes or virulence determinants at the most basic level. The ease and low cost of genomic sequencing means that all bacteriophages that would be offered in human clinical trials could be expected to have their DNA sequenced and examined for lysogenic cassettes and bacterial toxin genes.

In the case of Ps. aeruginosa, one bacteriophage is known to carry a toxin gene (Ceyssens and Lavigne 2010). This is ΦCTX, a temperate member of the tailed Myoviridae. The toxin forms pores in cell membranes and has been shown to enhance virulence in neutropenic mice (Baltch et al. 1994). However, it would appear that such toxin carriage in bacteriophages of Ps. aeruginosa is neither widespread not associated with the powerful clinical effects seen in cholera, dysentery and toxigenic E. coli infections. Considering this aspect of bacteriophage infection more widely, the filamentous bacteriophage Pf4 has been shown to be associated with biofilm formation in Ps. aeruginosa (Rice et al. 2009), but as this type of bacteriophage does not kill its bacterial target, it must be regarded as of limited relevance to phage therapy applications.

A ‘perfect storm’ for bacteriophage therapy

Many bacterial infections present promising targets for phage therapy. However, issues do exist with many of these as noted above.

The key to the progression of phage therapy as a way to control bacterial disease is the performance and proper reporting of the outcomes of high-quality clinical trials. Multiple trials of both safety and efficacy conducted to date have targeted Ps. aeruginosa, and this article has attempted to explain the reasoning behind such a selection.

Work in animal models has given promising results (Soothill 1992; Soothill et al. 2004; Wang et al. 2006; Vinodkumar et al. 2008; Debarbieux et al. 2010) as have reports from Eastern Europe (Sulakvelidze and Morris 2001) and similar work elsewhere (Abul-Hassan 1990). Early work targeting Ps. aeruginosa including single animal and patient treatments has also given promising results (Marza et al. 2006). Following on from this, multiple phase 1 trials of bacteriophage mixtures have now demonstrated safety – which given the dosing levels used is perhaps unsurprising (Merabishvili et al. 2009; Rhoads et al. 2009) – and efficacy trials have given evidence of efficacy in both animals (Soothill et al. 2004; Hawkins et al. 2010) and humans (Wright et al. 2009).

Progression into advanced trials is now under way, and it may well be that the first phage therapy to market in Europe and the USA will indeed target Ps. aeruginosa, providing a valuable new control for this intractable organism. The combination of antibiotic resistance, clinical need and the availability of suitable bacteriophages combines to make this a potential ‘perfect storm’ with which to progress this new (but also very old) therapeutic approach into the clinic.