Introduction
Increasing concerns about antimicrobial resistance in animal and human pathogens and expanding knowledge of the mechanisms and epidemiology of transmission of antibiotic resistance have driven the recent search for novel alternatives to antimicrobial drugs in humans and animals. The need for alternatives for animal use has been further accentuated by regulatory actions such as the recent ban on the use of sub-therapeutic levels of antibiotics in animal production in the European Union.
Although bacteriophage (phage) therapy is one such alternative, it is not novel. Soon after their discovery by Twort (Reference Twort1915) and d'Herelle (Reference d'Herelle1917) phages were used to control avian typhoid caused by Salmonella Gallinarum. Human applications soon followed, and by 1930–1940 phages were commonly used therapeutics, particularly in Georgia, Russia and Poland, and also in the USA (Summers, Reference Summers, Kutter and Sulakvelidze2005). Interest in phage therapy declined in the west following the introduction of antibiotics, but has increased dramatically, particularly with recent research indicating that phage therapy can be effective in treating serious infections caused by antibiotic resistant pathogens such as vancomycin-resistant Enterococci (Biswas et al., Reference Biswas, Adhya, Washart, Paul, Trostel, Powell, Carlton and Merril2002) and methicillin-resistant Staphylococcus aureus (MRSA) (Matsuzaki et al., Reference Matsuzaki, Yasuda, Nishikawa, Kuroda, Ujihara, Shuin, Shen, Jin, Fujimoto, Nasimuzzaman, Wakiguchi, Sugihara, Sugiura, Koda, Muraoka and Imai2003). Details of these recent, as well as early applications of phage therapy in humans, animals and animal models can be found in numerous reviews (see for example, Weber-Drabowska et al., Reference Weber-Dabrowska, Mulczyk and Gorski2000; Sulakvelidze et al., Reference Sulakvelidze, Alavidze and Morris2001; Barrow, Reference Barrow2001; Summers, Reference Summers, Kutter and Sulakvelidze2005; Sulakvelidze and Barrow, Reference Sulakvelidze, Barrow, Kutter and Sulakvelidze2005; Kropinski, Reference Kropinski2006; Parisien et al., Reference Parisien, Allain, Zhang, Mandeville and Lan2008).
The following review is focused on recent research in applications of phages for treatment and prevention of bacterial infections of cattle, swine and poultry. These have evolved largely from the excellent studies by Williams Smith and colleagues on Escherichia coli infections in mice, calves, pigs and lambs. Although these researchers explored phage therapy for pathogens of food animals, many recent veterinary applications have addressed zoonotic pathogens such as Salmonella spp., Campylobacter jejuni and E. coli O157:H7 that have food animals as natural reservoirs. Information on other phage-related therapies such as use of phage-encoded enzymes and bacterial vaccines generated by phage infection or by phage-encoded enzymes can be found in recent comprehensive reviews (Fischetti, Reference Fischetti2005; Sulakvelidze and Barrow, Reference Sulakvelidze, Barrow, Kutter and Sulakvelidze2005; Kropinski, Reference Kropinski2006; Parisien et al., Reference Parisien, Allain, Zhang, Mandeville and Lan2008).
Pertinent aspects of phage biology
Phages are highly plentiful in nature and share a common ecology with their bacterial hosts (Brüssow and Kutter, Reference Brüssow, Kutter, Kutter and Sulakvelidze2005). Consequently they are found in the alimentary tracts of animals and humans, and in foods, soil, water, sewage and associated environmental niches. Because of their ubiquitous presence, they are generally considered safe, an assumption that is strongly supported by the early widespread use of phages in humans, as well as by recent experimental studies in animals and humans.
Phages can be grouped as having either virulent (lytic) or temperate life cycles. Those for therapeutic use are lytic, exhibiting a self-replicating virulent infectious cycle that results in rapid degradation of host cell DNA, phage replication, and lysis with release of tens to hundreds of progeny phages (Guttman et al., Reference Guttman, Raya, Kutter, Kutter and Sulakvelidze2005). In contrast, the life cycle of temperate phages involves integration of the phage genome into the host cell DNA as a prophage, where it replicates along with the host cell DNA without lysing the bacterial host. The prophage can, however, revert to the lytic cycle with release of progeny phages. Clearly, temperate phages do not offer the therapeutic potential as lytic phages. In fact, activated temperate phages may be undesirable because they can transfer virulence and other undesirable genes from one bacterium to another by transduction.
The therapeutically desirable lytic, self-replicating infectious cycle of virulent phages is derived largely from in vitro studies with favorable ratios of phages to the target host bacteria [multiplicity of infection (MOI)]. At an ideal MOI, phages amplify themselves and kill the target hosts by repeated cycles of replication until the host is eliminated. Under other circumstances, the phage–host interaction may have different outcomes and be more self-limiting. At very high MOIs, the host may be killed by ‘lysis from without’, in which attachment of many phages to a single bacterial cell results in lysis without phage replication. In addition, extensive lysis may reduce the numbers of susceptible hosts to below the ‘phage proliferation threshold’ (Payne and Jansen, Reference Payne and Jansen2003), where the numbers of target bacteria are insufficient to sustain phage replication. In other situations, phages and susceptible bacteria may co-exist in an oscillating equilibrium of a predator–prey relationship. Co-existence with relatively stable numbers of hosts and phages may reflect the emergence of phage resistant sub-populations (Mizoguchi et al., Reference Mizoguchi, Morita, Fischer, Yoichi, Tanji and Unno2003) or that phages and bacteria occupy different niches such as the lumen, mucous layer or epithelial surface of the gut (Connerton et al., Reference Connerton, Loc Carrillo, Swift, Dillon, Scott, Rees, Dodd, Frost and Connerton2004; Chibani-Chennoufi et al., Reference Chibani-Chennoufi, Bruttin, Dillmann and Brussow2004).
Another important characteristic that influences the therapeutic potential of phages is their specificity, which determines their host ranges among targeted and non-targeted bacteria. Most phages offer much greater specificity than antimicrobial drugs, targeting only specific subtypes within a species, serovar or serogroup. Others may have a broad host range, such as the Salmonella phage, Felix O1, which infects most Salmonella serovars (Kallings, Reference Kallings1967). Phage specificity is determined largely by the interaction between binding sites on their tail fibers and one or more receptors on the cell surface of the host bacterium, which may include lipopolysaccharides, proteins, capsular polysaccharides, flagella and pili. While a narrow spectrum of activity is desirable to avoid unwanted effects on commensal flora, the ideal therapeutic phage would infect multiple pathogens carrying common surface receptors. A recent report of two phages that lyse multiple serotypes of two different pathogenic groups of E. coli, but few non-pathogenic E. coli (Viscardi et al., Reference Viscardi, Perugini, Auriemma, Capuano, Morabito, Kim, Loessner and Iovane2008) indicates that such phages occur in nature.
Phage therapy in cattle
Neonatal E. coli infections
In their initial studies in mice, Smith and Huggins (Reference Smith and Huggins1982) showed that a single dose of phage R was at least as effective as one or more doses of antibiotics in preventing death due to E. coli septicemia. Soon after, they explored the use of phages for control of enterotoxigenic E. coli (ETEC) infections in neonatal calves, pigs and lambs (Smith and Huggins, Reference Smith and Huggins1983; Smith et al., Reference Smith, Huggins and Shaw1987a). These studies have been reviewed in detail (Barrow, Reference Barrow2001; Sulakvelidze and Barrow, Reference Sulakvelidze, Barrow, Kutter and Sulakvelidze2005) and key outcomes of their work with ruminants are summarized in Table 1. Briefly, they showed that the severity and mortality of neonatal ETEC diarrheal disease could be prevented by selected phages given orally, sprayed on pen litter or transmitted from pens occupied previously by phage-treated animals. They also found that naturally emerging phage-resistant strains lacking the virulence-associated K antigen were less virulent than the parent strains, but resistant K-positive strains retained their virulence. Factors such as pH, colostral phage antibodies and body temperature on phage viability and efficacy were also investigated (Smith et al., Reference Smith, Huggins and Shaw1987b). The sheer quantity and quality of the work in these reports demands that interested readers consult the original papers.
Table 1. Summary of studies by H. W. Smith and colleagues on phage therapy of neonatal enterotoxigenic E. coli (ETEC) infections in calves and lambs
Furthering the work of Smith and colleagues, Barrow et al. (Reference Barrow, Lovell and Berchieri1998) examined the effect of phage R in experimental E. coli septicemia in colostrum-deprived calves. Whereas two calves inoculated orally with virulent K1+E. coli H247 (O18:K1:H7) rapidly became moribund and required euthanasia 36 h after infection, three of four calves given 1010 PFU of phage R intramuscularly 8 h after the same challenge remained healthy. Phage treatment did not alter numbers of strain H247 in the feces or blood, but delayed its entry into the blood for 1 day as titers of phage R in feces and blood increased to high titers 16–24 h after treatment. One day after challenge, phage resistant K1+ isolates of H247 were identified in the feces of the one treated calf that became ill, and a day later predominated in the blood of this calf. However, the illness in this calf was considered to be due to absorption of toxins from the gut rather than septicemic disease caused by the resistant mutants. Phage treatment 8 h after infection thus appeared to control disease severity but not prevent infection, as also seen in other studies (Smith and Huggins, Reference Smith and Huggins1983; Smith et al., Reference Smith, Huggins and Shaw1987a).
E. coli O157:H7 infection
As the predominant serotype of Shiga toxin-producing E. coli associated with severe disease in humans, E. coli O157:H7 is a major public health concern (Rangel et al., Reference Rangel, Sparling, Crowe, Griffin and Swerdlow2005). Healthy cattle and other ruminants are important natural reservoirs of this organism, shedding it in their feces intermittently or for short periods (Besser et al., Reference Besser, Hancock, Pritchett, McRae, Rice and Tarr1997). Transmission to humans occurs most frequently through meats, milk, water, fresh produce, animals and environmental sources contaminated directly with manure of infected cattle. Hence, there is considerable interest in on-farm interventions to control this organism in cattle, including phage therapy. Several studies of E. coli O157:H7 phage therapy intended for cattle have been conducted in sheep, which serve as a convenient model.
Evaluation of three lytic E. coli O157:H7-specific phages, KH1, KH4 and KH5, in vitro revealed that complete lysis occurred within 8 h at 37°C and 4°C, but only with a mixture of the three phages at high MOI (~103) and incubation with aeration (Kudva et al., Reference Kudva, Jelacic, Tarr, Yourderian and Hovde1999). In a subsequent 21-day trial, phage KH1 was ineffective in sheep when given orally at a dose of 1.3×1011 PFU orally 1, 9, 10 and 11 days after infection with E. coli O157:H7, despite the presence of the phage in feces at 105–106 PFU g−1 during this time (Sheng et al., Reference Sheng, Knecht, Kudva and Hovde2006). Similarly, a high dose (1013 PFU) of phage DC22, which eliminated E. coli O157:H7 from a model rumen fermenter in 4 h, had no significant effect on shedding of E. coli O157:H7 by experimentally infected lambs over 27 days (Bach et al., Reference Bach, McAllister, Veira, Gannon and Holly2008). Fecal phage counts declined from 106 PFU g−1 on day 3 to undetectable levels on day 13.
Sheng et al. (Reference Sheng, Knecht, Kudva and Hovde2006) investigated rectal treatment of cattle infected by rectal inoculation because of reports that the recto-anal junction is the primary site of colonization of cattle by E. coli O157:H7 (Naylor et al., Reference Naylor, Low, Besser, Mahajan, Gunn, Pearce, McKendrick, Smith and Gally2003). On day 0, 7 days after recto-anal inoculation of 10 steers with a four-strain mixture of E. coli O157:H7, five steers were given phages SH1 and KH1 (25 ml of 1010 PFU ml−1) by rectal infusion and swabbing, and in drinking water at 1.8 to 5.4×106 PFU ml−1 for 4 days. Counts of phages and E. coli O157:H7 were monitored for 16 days in recto-anal mucosal swab samples (RAMS). On day 0, counts of E. coli O157:H7 were 104–105 CFU RAMS−1, giving an initial MOI of ~105 to 106. Subsequent average E. coli O157:H7 counts were significantly lower in treated than control calves on days 1–10, by ~1.5 log units RAMS−1. Phage counts peaked at ~106 PFU RAMS−1 on day 3, and then declined steadily to ~102 PFU RAMS−1 on day 16. Despite persistence of the phages and the initial reductions in E. coli O157:H7 counts, four of the five treated calves remained infected. This result differed considerably from that of a mouse experiment included in the same report, in which three oral treatments with SH1 alone or mixed with KH1 eliminated E. coli O157:H7 from feces within 2–6 days. The authors concluded that the mixture of phages KH1 and SH1 can reduce but not eliminate shedding of E. coli O157:H7 by cattle. Although the recto-anal junction is a preferred site of E. coli O157:H7 colonization, rectal phage administration is impractical in cattle, and the phages are likely to be excreted promptly. Also, this route would not be effective initially on the target organism at other sites, although both E. coli O157:H7 and the phages would likely be transmitted at lower numbers via the fecal–oral route.
Two brief trials lasting 2–4 days indicated that numbers of E. coli O157:H7 in feces and/or intestinal contents of sheep or cattle were reduced significantly soon after similar high doses of different phages. In four sheep treated once orally with 1011 PFU of phage CEV1 3 days after challenge, E. coli O157:H7 populations 2 days later were reduced significantly by 2–3 log units in cecal and rectal but not rumen contents (Raya et al., Reference Raya, Varey, Oot, Dyen, Callaway, Edrington, Kutter and Brabban2006). In the second short study (Callaway et al., Reference Callaway, Edrington, Brabban, Anderson, Rossman, Engler, Carr, Genovese, Keen, Looper, Kutter and Nisbet2008), a mixture of eight phages (total >1012 PFU) was given by oral gavage to 10 sheep on days 2 and 3 after challenge with E. coli O157:H7. Compared to those in controls, fecal E. coli O157:H7 counts were reduced significantly by 1.5–2 log units g−1 24 h after the first dose, but not at 12 and 24 h after the second dose, suggesting the initial effect was transient. Most sheep continued shedding an average of ~104 CFU g−1 of feces 24 h after two treatments. At autopsy on day 4, reductions in numbers of E. coli O157:H7 were significant in cecal but not rectal or ruminal contents. However, counts in feces collected immediately before euthanasia were ~100 times higher than in rectal contents, suggesting that phage treatment may not have been effective at colonization sites at the recto-anal junction. Also of note was the finding from a second experiment that a MOI of 1 was more effective than MOIs of 0.1, 10 and 100 in reducing counts of the challenge strain in ruminal, cecal and rectal contents.
Both of these brief trials suggest that phage therapy shortly before slaughter may reduce the risk of carcass contamination by lowering the fecal load of E. coli O157:H7. However, such an approach does not address environmental dissemination of the organism during the rearing and growing phases of production, or perhaps contamination of hides, which contributes substantially to carcass contamination.
While none of the above studies resulted in elimination of E. coli O157:H7 from cattle or sheep, some of us (Waddell et al., Reference Waddell, Mazzocco, Johnson, Pacan, Campbell, Perets, MacKinnon, Holtslander and Poppe2000) have had more encouraging results in experimentally infected calves. Weaned 7–8-week-old calves infected orally on day 0 with 3×109 CFU of E. coli O157:H7 were treated or not with a total of 1011 PFU of a mixture of six phages on days −7, −6, −1, 0 and 1. The phages were given in calf milk replacer containing calcium carbonate to buffer stomach acid, and the dose on day 0 was given 4 h before the bacterial inoculum. While most untreated calves shed E. coli O157:H7 in their feces for at least 12–16 days, the treated calves stopped shedding this organism abruptly after day 8, when the concentrations of phages in their feces had increased sharply to 109–1011 PFU g−1. Such dramatic increases in phage numbers did not occur in uninfected control calves given only the phages, indicating extensive replication of the phages in the treated animals. No adverse effects of phage therapy were observed clinically or on monitoring of total fecal coliform and E. coli counts, and no phage-resistant E. coli O157:H7 were isolated over the 14–16 days of the trial.
The very encouraging results with calves suggest that phage therapy can effectively eliminate E. coli O157:H7 under some circumstances. Treatment may have application early in the cattle production cycle, since initial infection of beef cattle often occurs in calves less than 12 weeks of age (Gannon et al., Reference Gannon, Graham, King, Michel, Read, Ziebell and Johnson2002). Additional treatments at periods of peak transmission later in the production cycle may help reduce the overall prevalence and environmental dissemination. However, treatment of young adult beef cattle with a subset of four of these phages reduced but did not eliminate shedding of E. coli O157:H7 from experimentally infected cattle (Niu et al., Reference Niu, Xu, McAllister, Rozema, Stephens, Bach, Johnson and Stanford2008). Possibly, age-related differences in development of the gastro-intestinal tract may in part explain the differences in efficacy in calves and adult cattle. Although the phage dose for the adult cattle was increased at least 10-fold, this would hardly account for dilution in the substantially larger volume of the contents of the adult rumen and intestinal tract. Also, since adult cattle are likely to have experienced greater exposure to E. coli O157:H7 infection (Laegreid et al., Reference Laegreid, Elder and Keen1999; Gannon et al., Reference Gannon, Graham, King, Michel, Read, Ziebell and Johnson2002) they are more likely than calves to have levels of immunity that may limit colonization and replication of E. coli O157:H7 to levels below those required to support phage amplification in the gut.
Bovine mastitis caused by S. aureus
Clinical and sub-clinical mastitis remains a major milk production-associated disease among dairy cattle worldwide. Treatment most often involves repeated intramammary infusions of antibiotics into affected quarters via the teat canal. S. aureus is a common cause of mastitis in cattle, and poses particular challenges because of relatively low cure rates, antimicrobial resistance and sub-clinical persistence in herds (Makovec and Ruegg, Reference Makovec and Ruegg2003; Luby and Middleton, Reference Luby and Middleton2005). Concern also exists because of transmission of antibiotic resistant S. aureus, including MRSA from animals to humans (Lee, Reference Lee2003). Although phage therapy was effective against lethal MRSA infection in mice (Matsuzaki et al., Reference Matsuzaki, Yasuda, Nishikawa, Kuroda, Ujihara, Shuin, Shen, Jin, Fujimoto, Nasimuzzaman, Wakiguchi, Sugihara, Sugiura, Koda, Muraoka and Imai2003), a recent study of phage therapy of S. aureus mastitis in cattle was less encouraging (Gill et al., Reference Gill, Pacan, Carson, Leslie, Griffiths and Sabour2006a). Five daily intramammary infusions with 1011 PFU of S. aureus phage K in udder quarters with sub-clinical S. aureus mastitis resulted in cure of only 3 of 18 quarters, a non-significant effect compared with saline-treated controls (cure rate 0/20 quarters). S. aureus counts in milk fluctuated after treatment, as did the counts in saline treated animals, and recovered S. aureus isolates were susceptible to lysis by phage K. Possible reasons for the lack of efficacy included enzymatic inactivation of phages in the mammary gland, inhibition of the binding of active phages to the target bacteria by milk proteins, and phage aggregation by milk proteins (Gill et al., Reference Gill, Pacan, Carson, Leslie, Griffiths and Sabour2006a, Reference Gill, Sabour, Leslie and Griffithsb). Given these limitations, it would appear that phage therapy of bovine mastitis has limited potential.
Phage therapy in chickens
E. coli infections
Colibacillosis is a serious problem in poultry production causing mortality and carcass condemnations (Barnes et al., Reference Barnes, Vaillancourt, Gross and Saif2003). Initial infection is thought to occur in the respiratory tract as airsaculitis that quickly becomes septicemic, resulting in considerable mortality. In exploring applications of phage therapy, Barrow et al. (Reference Barrow, Lovell and Berchieri1998) found that phage R, active against K1+E. coli (Smith and Huggins, Reference Smith and Huggins1982) was highly effective in both prevention and treatment of experimental E. coli septicemia and meningitis in chickens. Intramuscular (i.m.) inoculation of 106 PFU of phage R fully protected newly hatched and 3-week-old chickens against death due to septicemia following simultaneous i.m. inoculation of 106 CFU of E. coli O18:K1:H7 strain H247. Also, a higher dose (108 PFU) prevented death due to meningitis following intracranial inoculation of 103 CFU of the same bacterium. Titers of E. coli H247 in the blood, spleen and brain were much lower in treated than in untreated birds, and marked increases in titers of phages in samples from treated birds indicated dissemination and replication of the phages after administration. Similar phage treatments given 1–2 days before i.m. challenge were highly effective prophylactically, and if given therapeutically at the onset of clinical signs, reduced mortality by 70%.
E. coli of serogroup O2 are a frequent cause of airsaculitis and septicemia, prompting some of us, William Huff and colleagues, to investigate phage therapy to prevent and treat infections caused by this serogroup. Phages virulent for a pathogenic E. coli O2:NM strain were evaluated when given by various routes at different times in relation to challenge. In all experiments, the challenge was inoculation of ~104 CFU of an E. coli O2:NM strain into one thoracic airsac (Piercy and West, Reference Piercy and West1976). In initial experiments, different numbers of the selected phage were mixed with E. coli prior to inoculation. The mortality in birds challenged only with E. coli was 85% (Fig. 1), whereas a mixture containing 104 PFU significantly reduced mortality to 35%, and a mixture containing 108 PFU completely protected the birds (Huff et al., Reference Huff, Huff, Rath, Balog, Xie, Moore and Donoghue2002a). Although this experimental design was artificial, it suggested that phages might be used to prevent colibacillosis and it demonstrated the importance of phage dose for therapeutic efficacy. In addition, this model provides a relatively simple method to screen phages for efficacy in vivo.
Fig. 1. Effect of mixing E. coli with bacteriophages prior to challenge Treatments: 1, Control □; 2, E. coli, 104 CFU ▪; 3, E. coli, 104 CFU, mixed with 104 PFU of bacteriophages 
; 4, E. coli, 104 CFU, mixed with 108 PFU of bacteriophages 
. Values represent the means of two replicate pens of 10 birds per pen. Values with different letters differ significantly (P⩽0.05).
Because colibacillosis frequently begins as a respiratory infection, 7-day-old chicks were treated with an aerosol spray of phages as a preventative measure immediately, or 1 or 3 days before challenge. Mortality was significantly lower in all groups of birds treated prior to challenge than in challenged but untreated birds (Fig. 2), suggesting that an aerosol administration of phages could provide protection for at least 3 days (Huff et al., Reference Huff, Huff, Rath, Balog and Donoghue2002b).
Fig. 2. Efficacy of a bacteriophage aerosol spray to prevent colibacillosis. Treatment 1 ▪, birds challenged with E. coli only. Treatments 2 , 3 
, and 4 
, birds sprayed with bacteriophages at 7 days of age prior to challenging with E. coli at 7, 8, or 10 days of age, respectively. Values represent the mean of four replicate pens of 10 birds per pen. Values with different letters differ significantly (P⩽0.05).
To evaluate phages for treatment rather than prevention, they were given by aerosol spray or i.m. inoculation at intervals following challenge of 7-day-old birds (Huff et al., Reference Huff, Huff, Rath, Balog and Donoghue2003a). The aerosol spray was not effective, but i.m. administration of phages given immediately, 24 h or 48 h after challenge significantly reduced mortality compared with that in untreated, challenged birds (Fig. 3). Additional research showed that multiple i.m. doses of phages enhanced phage treatment efficacy (Huff et al., Reference Huff, Huff, Rath, Balog and Donoghue2003b). Blood levels of phages were determined after aerosol or i.m. administration to unchallenged birds, in order to determine if sufficient numbers would reach the blood to combat the septicemic phase of colibacillosis. Few birds treated by aerosol administration had detectable levels of phages in the blood at intervals up to 24 h after treatment. However, after i.m. inoculation, titers of phages in blood remained above 104 PFU ml−1 in all five birds for up to 6 h after administration, and at 24 h, were ~102 PFU ml−1 in four of the five birds (Huff et al., Reference Huff, Huff, Rath, Balog and Donoghue2003a). Intramuscular inoculation of phages is therefore likely to be far more effective than aerosol administration in treatment of the septicemic phase of colibacillosis. Another study (Huff et al., Reference Huff, Huff, Rath, Balog and Donoghue2004) demonstrated that a combination of i.m. inoculation of phages and a low dose of enrofloxacin given in drinking water potentially enhanced the efficacy of treatment of colibacillosis.
Fig. 3. Efficacy of intramuscular administration of bacteriophage to treat colibacillosis. Treatment 1 ▪, birds challenged with E. coli and not treated with bacteriophage. Treatments 2 , 3 , and 4 , birds challenged with E. coli and treated with bacteriophage immediately (7 days of age), 24 h (8 days of age), or 48 h (9 days of age) after E. coli challenge, respectively. Values represent the mean of four replicate pens of 10 birds per pen. Values with different letters differ significantly (P⩽0.05).
Together, these studies of phage therapy of respiratory and septicemic O2 E. coli infections of chickens indicate that prevention and treatment by phage therapy may be feasible, although with potential limitations similar to those noted by Barrow et al. (Reference Barrow, Lovell and Berchieri1998). To be effective preventatively, high numbers of phages are needed at the sites of infection at the time of exposure or systemic infection. High numbers are also likely to be required in the blood within a day or so of infection if phages are to be effective as treatments. In poultry farms, aerosol administration might provide protective levels in the respiratory tract but not the levels required for treatment. Since intramuscular administration is not practical or economically feasible in large modern poultry farms, a focus on preventative use would have the greatest potential.
Salmonella infections
Poultry products are a major source of human salmonellosis, due to widespread sub-clinical carriage of Salmonella by poultry, resulting in high rates of carcass contamination. In addition, certain serovars such as S. Typhimurium and S. Enteritidis can be highly virulent in young chickens and older birds. The emergence of multi-drug resistant subtypes of several Salmonella serovars affecting poultry, other food animals and humans has intensified the search for alternative non-antibiotic control strategies for these important enteric pathogens.
In a brief early report (Taylor and Silliker, Reference Taylor and Silliker1958), treatments of hatching eggs infected with S. Chittagong, S. Pullorum or S. Typhimurium with a broad-range Salmonella phage and/or serovar-specific phages improved the percent hatch of fertile eggs from <45% to >70%. However, despite this apparent success, this technology does not appear to have been explored further, and more recent studies have focused on post-hatch applications of phage therapy.
Berchieri et al. (Reference Berchieri, Lovell and Barrow1991) found that 104 PFU of phage AB2 given orally to newly hatched chicks immediately after challenge with S. Typhimurium F98 and in feed containing 103 PFU g−1 for 7 days was ineffective in reducing cecal content levels of the challenge strain. The phage did however replicate to high titers at 4 days post-infection (p.i.), but was undetectable after the levels of F98 fell below 106 CFU ml−1 of cecal contents. Resistance to phage AB2 was frequent, occurring in 34–82% of 50 isolates of F98 on each of p.i. days 2, 4, 7 and 10. The rough O antigen phenotype of all resistant isolates may have contributed to the persistence of high numbers of S. Typhimurium, although this was not considered to be the case. No phage AB2 neutralizing antibodies were detected in sera collected from these chickens at 32 days p.i. In contrast, a different phage, 2.2, significantly reduced the overall mortality rate over 21 days from 56% to 20%, and was also effective against two other strains. However, this result was only achieved with a dose of 1011 PFU. Reductions of up to >2 log units in counts of the challenge strain in the crop and small intestine occurred during the first 3–6 h, but were transient, and counts in the liver were almost 1 log unit lower than in the controls after 24 and 48 h.
The above study illustrates several of the issues in phage therapy. Both phages were virulent in vitro, but AB2 was not effective in vivo, despite evidence of replication in the gut. Phage 2.2 was effective in vivo, but only at a high dose. The need for high doses and the early and transient nature of the decreases in target organism perhaps suggest lysis from without or only one cycle of replication. Treatment also appeared to limit invasion and colonization of the liver. However, S. Typhimurium was not eliminated and its persistence at increasing titers up to 24 h after treatment may reflect that the Salmonella had escaped from phage predation by establishing intracellular infection.
A recent study evaluated three broad host range phages against S. Enteritidis, S. Typhimurium and S. Hadar as a pre-harvest treatment in 36-day-old broiler chickens (Atterbury et al., Reference Atterbury, Van Bergen, Ortiz, Lovell, Harris, De Boer, Wagenaar, Allen and Barrow2007). Two days after oral challenge of groups with S. Enteritidis, S. Typhimurium or S. Hadar, they were treated orally with 109 PFU (trial 1) or 1011 PFU (trial 2) of corresponding phages 151, 10 or 25, respectively, in an antacid buffer. Counts of S. Enteritidis and S. Typhimurium in cecal contents over 6 days were reduced significantly by >4 and >2 log units, respectively, but only at the higher dose (1011 PFU). These substantial reductions could potentially reduce the risk of product contamination at slaughter. Phage 25 had no effect on counts of S. Hadar, despite its strong lytic activity against this strain in vitro. The frequencies of phage resistant colonies isolated from treated birds in trial 1 (S. Enteritidis, 21%; S. Typhimurium, 10%; S. Hadar, 23%) were about twice those of isolates from untreated infected birds, and except for S. Hadar, were even greater after treatment with the higher phage dose. Tested resistant isolates did not retain their resistant phenotype when sub-cultured five times, and colonized the ceca of chickens as effectively as the original strains. Overall, these findings indicated that a high phage dose was potentially effective as a pre-harvest intervention to reduce the levels of S. Enteritidis and S. Typhimurium in cecal contents of broiler chickens shortly before the age of slaughter. Although the high phage dose generated higher proportions of phage resistant strains, resistance did not persist for long in the absence of phage in vitro or in vivo, and perhaps could be circumvented by treatments with a mixture of phages.
The substantial reductions in Salmonella reported above were rarely or inconsistently achieved in several other studies, although evidence for an effect of phage therapy was noted. Interpretation of results is somewhat challenging however, due to different methods, doses and modes of administration, age of chickens and outcomes measured. For example, after chickens challenged with S. Enteritidis were treated orally with two phage cocktails, either alone or in combination, rates of isolation of S. Enteritidis from cecal tonsils after 24 h were reduced significantly from 100% in untreated controls to 45–70%. However, the effect was transient since isolation rates were not significantly lower in treated (85–100%) compared with control chickens (100%) 48 h after treatment (Andreatti Filho et al., Reference Andreatti Filho, Higgins, Higgins, Gaona, Wolfenden, Tellez and Hargis2007).
In a longer term study, counts of S. Enteritidis in cecal homogenates 14 days after challenge of day-old chickens were reduced by ~0.3–1.3 log units following five trials of various oral treatments with four individual phages 3 h after challenge (Sklar and Joerger, Reference Sklar and Joerger2001). However, these reductions were significant in only two trials, in which counts of S. Enteridis in cecal contents exceeded 104 CFU g−1. Similarly, an oral dose of 1011 PFU of a mixture of three phages reduced S. Enteritidis counts in cecal contents up to 3.5-fold over 25 days, and rates of colonization of the liver and spleen, but the reductions were not significant (Fiorentin et al., Reference Fiorentin, Vieira and Barioni2005). In another study, oral treatment of chickens with a low dose of a mixture of three broad host range phages 3 days before and 3 days after challenge with S. Typhimurium resulted in marginally improved weight gain and moderate but inconsistent reductions in numbers of the challenge strain in ileal and cecal samples (Toro et al., Reference Toro, Price, McKee, Hoerr, Krehling, Perdue and Bauermeister2005). Also, the same authors found that a commercial competitive exclusion product (Protexin) was at least as effective as phage treatment alone, and had no synergistic effect when combined with phage treatment. Similar results were obtained when a commercial probiotic, Floramax-B11, was compared to a mixture of 45 phages (WT45) either alone or in combination with the phages in treatment of day-old chicks challenged with S. Enteritidis (Andreatti Filho et al., Reference Andreatti Filho, Higgins, Higgins, Gaona, Wolfenden, Tellez and Hargis2007).
Campylobacter infections
Campylobacteriosis, particularly caused by C. jejuni, is the most common food-borne bacterial enteritis in developed countries. Campylobacters colonize the intestinal tracts of healthy animals of many species, including foods animals such as poultry, pigs and cattle. Poultry meat poses a major public health risk since broiler chickens are heavily colonized with C. jejuni at slaughter, resulting in high rates of carcass contamination during processing, frequently at levels greater than 105 CFU per carcass (Jorgensen et al., Reference Jorgensen, Bailey, Williams, Henderson, Wareing, Bolton, Frost, Ward and Humphrey2002). Quantitative risk assessment (Rosenquist et al., Reference Rosenquist, Nielsen, Sommer, Norrung and Christensen2003) indicates that the incidence of human Campylobacter infections could be reduced substantially if the numbers of Campylobacter on poultry meat are reduced by 2 log units. Recent evidence suggests that this objective might be achieved with pre-harvest phage therapy. Moreover, the presence of naturally occurring Campylobacter phages in commercial broiler flocks correlates with a significant >1 log unit reduction in the numbers of C. jejuni in the cecal contents of broiler chickens (Atterbury et al., Reference Atterbury, Dillon, Swift, Connerton, Frost, Dodd, Rees and Connerton2005). These and many other aspects of Campylobacter phages have been reviewed in detail by Connerton et al. (Reference Connerton, Connerton, Barrow, Seal, Atterbury, Nachamkin, Szymanski and Blaser2008).
The first report of phage therapy for C. jejuni in broiler chickens employed both preventative and therapeutic phage administration (Wagenaar et al., Reference Wagenaar, Van Bergen, Mueller, Wassenaar and Carlton2005). In the preventative trial, broiler chickens 10 days old were challenged with C. jejuni strain C356 on day 4 of a 10-day course of oral gavage with 0.4 to 2×1010 PFU of phage 71. In the therapeutic trial, the same dose of phage 71 was given daily for 6 days commencing 5 days after challenge at 15 days of age. Counts of phages and the challenge strain were monitored in cecal contents until the chickens were 39 days old. Preventative treatment delayed but did not prevent colonization. Levels of C. jejuni were initially 2 log units lower than in controls, and then stabilized at ~1 log unit lower than in the controls. Phage and C. jejuni counts showed alternating fluctuations, consistent with alternating cycles of host replication and clearance with phage amplification, as occurs in naturally infected flocks (Atterbury et al., Reference Atterbury, Dillon, Swift, Connerton, Frost, Dodd, Rees and Connerton2005). After therapeutic phage treatment, counts of C. jejuni were immediately reduced by >3 log units for several days then stabilized at ~1 log unit lower than in the controls. Based on this greater efficacy following therapeutic dosage, a similar approach was applied to chickens challenged 10 days before the usual age of slaughter (~42 days). Birds were given a mixture of two phages, 71 and 69, orally for 4 days commencing 7 days after challenge. Following an initial drop of 1.5 log units, counts stabilized at 1 log unit lower than in control birds. These results were interpreted as encouraging for the use of phage therapy immediately pre-slaughter as a means of reducing the risks of human campylobacteriosis. To overcome the tendency towards the phage–host equilibrium and development of a resistant host subpopulation under farm conditions, it was suggested that pre-harvest treatment would be more effective with fresh lytic phages rotated in different production cycles.
Loc Carillo et al. (Reference Loc Carillo, Atterbury, el Shibiny, Connerton, Dillon, Scott and Connerton2005) reported a comprehensive study of phage therapy to reduce the C. jejuni burden in chickens. Groups of chickens infected with C. jejuni strains HPC5 or GIIC8 that became heavily colonized within 5–7 days were treated at 25 days of age with two candidate phages, CP8 or CP34, given as a single oral dose of 105, 107, or 109 PFU in an antacid buffer containing calcium carbonate. In the first 24 h, doses of 105 and 107 PFU of CP 8 were more effective than 109 PFU in reducing cecal colonization. With 107 PFU, levels of the challenge strains in contents of the upper intestine, ceca and lower intestine fell 0.5 to >5 log units, then rose after 2–3 days to levels ~1–2 log units lower than in controls. Phage CP34 was clearly more effective than CP8 against strain HPC5, despite in vitro evidence that CP8 was virulent for this strain. CP8 replicated in the intestinal tract without significantly affecting the overall population of strain HPC5. However, phage CP8 proved very effective against strain GIIC8, with reductions of 5.6 log units 24 h after treatment. After 5 days, levels of the challenge strain in cecal and lower intestinal contents were still significantly lower than in controls by ~2 log units. Resistance to phage CP34 was detected less frequently in isolates of HPC5 from chickens (4%) compared with HPC5 isolates from in vitro experiments (11%). Two of the CP34-resistant HPC5 isolates from chickens were less able to colonize chickens, and 97% of 90 isolates recovered from birds infected with these resistant strains had reverted to their phage sensitive state. As in the study by Wagenaar et al. (Reference Wagenaar, Van Bergen, Mueller, Wassenaar and Carlton2005), these findings suggest that phage therapy holds promise for control of C. jejuni in poultry as a food-borne pathogen. The results emphasize the need for well-designed in vivo as well as in vitro evaluation of candidate phages, phage dose, and the merit in investigating the emergence and characteristics of phage-resistant hosts. Most importantly, it demonstrates the substantial influence of individual phage–host combinations on the effectiveness of phage therapy.
Phage therapy in pigs
ETEC infections
The first of the few studies of phage therapy in pigs were the excellent experiments conducted by Williams Smith in the United Kingdom on treatment of experimental ETEC diarrhea in neonatal pigs (Smith and Huggins, Reference Smith and Huggins1983). They assessed the efficacy of a mixture of two phages against diarrhea induced in neonatal pigs by infection with ETEC strain P433 (O20:K101:987P+). One phage, P433/1, used the K101 antigen as a receptor and lysed strain P433 and the other, P433/2, was lytic for K101-negative mutants of strain P433 that arose spontaneously at high frequency in vitro and were resistant to phage P433/1. Both phages were highly lytic in vitro, with as few as nine particles of P433/1 and four particles of P433/2 required to completely lyse broth cultures of their respective hosts.
In the first experiments, 14 newborn pigs fed bovine colostrum were inoculated orally at 6 h of age with E. coli P433 along with a non-pathogenic E. coli and a Lactobacillus. Seven of the pigs were treated with 1010 PFU of each of the two phages at the onset of diarrhea, 13–16 h after infection. The untreated pigs developed severe diarrhea; four died after 26–65 h and the remaining three pigs had severe diarrhea lasting 44 to 84 h. Fecal counts of ETEC P433 increased to ~109 CFU g−1 within 3–7 h, remained at this level for 24 h then declined to ~108 CFU g−1 at the end of the study (96 h). In contrast, none of the treated pigs died, diarrhea was mild and lasted 7–13 h, and fecal counts of ETEC P433 declined rapidly from ~108 CFU g−1 at the time of treatment to 105 CFU g−1 5 h after treatment. Fecal titers of the phages, predominantly P433/1, remained at 106 PFU g−1 or higher throughout the study. Phage 433/1-resistant K-negative mutants were detected in several treated pigs, but always at low numbers compared with ETEC P433. Interestingly, in subsequent experiments, phage P433/1 was essentially as effective alone as it was when given with P433/2, whereas phage P433/2 alone was ineffective. Also, the K-negative phage-resistant mutants did not induce diarrhea in pigs. Smith and Huggins (Reference Smith and Huggins1983) noted that because several K antigens are associated with ETEC in pigs, phages that target colonizing pili (F4, F5, F6 or F18) would cover a wider range of porcine ETEC and might be more practical.
Interest in phage therapy for neonatal ETEC infections declined after the development of effective control with pilus-based vaccines administered to pregnant sows. However, ETEC infections in weaned pigs continue to be a major economic problem for pig producers. Furthermore, widespread resistance of ETEC to antimicrobial agents has lead to renewed interest in phage therapy. Recently, Jamalludeen et al. (Reference Jamalludeen, Johnson, Friendship, Kropinski, Lingohr and Gyles2007) isolated phages active against O149 ETEC, now the most common and widespread porcine ETEC, worldwide. Following characterization of the phages, nine were selected for further evaluation. Six phages (GJ1–GJ6) lysed 99–100% of 85 strains of O149:H10 ETEC but only 0–12% of 42 O149:H43 ETEC strains. Phage GJ1 was completely sequenced and shown to lack genes for toxins and for lysogeny (Jamalludeen et al., Reference Jamalludeen, Kropinski, Johnson, Lingohr, Harel and Gyles2008). Three other phages (GJ7–GJ9) that were selected on an O149:H43 ETEC host strain lysed 86–98% of 42 O149:H43 ETEC and 2–53% of O149:H10 ETEC. The nine phages were not specific for O149 E. coli as they lysed 6–68% of the 72 strains of the ECOR collection and 3–41% of 37 non-O149 ETEC.
Subsequently, phages GJ1–GJ7 were evaluated individually and in some combinations for prophylaxis and treatment of experimental infection of pigs with an O149:H10 ETEC strain (Jamalludeen et al., submitted for publication). Parameters that were measured were severity and duration of diarrhea, weight change, and excretion of the challenge ETEC. Oral administration of 109 PFU of phages GJ1–GJ6 individually in prophylactic mode, 15 min before challenge, significantly decreased the composite diarrhea score (a combination of duration and severity), excretion of the challenge ETEC and weight loss in treated compared with untreated, challenged pigs. Prophylactic treatment with a mixture of phages GJ1, GJ2 and GJ7 (109 PFU of each phage) also resulted in a significant reduction in the composite diarrhea score and a reduction in shedding of the challenge ETEC, but the decrease in the challenge ETEC was not significant. In therapeutic mode, three doses of a mixture of phages GJ1 and GJ6 (108 PFU of each phage) were given orally at 6-h intervals beginning 24 h after challenge with ETEC O149:H10. Treatment was associated with significant decreases in diarrheal scores and excretion of the challenge ETEC as well as with weight gain compared with weight loss in the control pigs. Levels of the phages in feces increased to as high as 1011 PFU g−1 within 1–2 days after treatment and gradually declined to ~103 PFU g−1 by the end of the experiment on day 6. In a separate trial in unchallenged pigs, the levels of phages in feces increased markedly when sodium bicarbonate was given orally just prior to the phages, suggesting that antacids protected the phages from the low pH of the stomach.
Additional work is needed to optimize doses, prepare the phages in a suitable formulation for field applications, and evaluate their effectiveness under field conditions. Post-weaning ETEC diarrhea is a good target for alternatives to antibiotics because of the high frequency of multiple drug resistance of ETEC (Maynard et al., Reference Maynard, Fairbrother, Bekal, Sanschagrin, Levesque, Brousseau, Masson, Lariviere and Harel2003) and the association of virulence and drug resistance genes on plasmids (Boerlin et al., Reference Boerlin, Travis, Gyles, Reid-Smith, Janecko, Lim, Nicholson, McEwen, Friendship and Archambault2005). Smith's notion of seeking phages that use the colonizing pili as receptor has particular merit for ETEC in post-weaning diarrhea, since F4 and F18 pili are the only colonizing pili that are frequently associated with ETEC that cause this disease in pigs.
Salmonella infections
The only reported evaluation of phage therapy for Salmonella in pigs was a pre-harvest food safety intervention to reduce contamination of pork products that might occur following the dissemination of S. Typhimurium during shipping, lairage and slaughter (Lee and Harris, Reference Lee and Harris2001). Pigs challenged intranasally with 5×108 CFU of S. Typhimurium were treated 3 h later by both oral and i.m. routes with either 2×109 PFU of phage Felix-O1 or a placebo. Six hours after treatment, enumeration of the challenge strain in tonsils, ileocecal lymph nodes, lung, liver, spleen, cecum and rectal contents revealed S. Typhimurium was still present at high numbers in the intestinal contents of pigs in both groups, but levels in the tonsils and ceca were reduced significantly. Further details in a patent (Harris and Lee, Reference Harris and Lee2003) suggest that multiple dose variations on this treatment regime in the 24 h before shipping and slaughter can reduce the levels of S. Typhimurium in these and other tissues. As in other species, i.m. inoculation of phages may not be feasible in modern swine production.
Related applications of phages
Phage control of zoonotic pathogenic bacteria in foods
While there has been little or no adoption of phage therapy at the farm level, considerably greater interest has evolved in the food processing industry, where control of zoonotic pathogens immediately pre-harvest and during processing is a major focus. For example, pre-slaughter treatments of animal hides with phages against E. coli O157:H7 and Salmonella to reduce their entry into the food chain are acceptable to the US Food and Drug Administration (see http://omnilytics.com/news). Phages also reduce the load of zoonotic pathogens in some food products during or after processing, although efficacy has varied (Greer, Reference Greer2005; Higgins et al., Reference Higgins, Higgins, Guenther, Huff, Donoghue, Donoghue and Hargis2005; Hudson et al., Reference Hudson, Billington, Carey-Smith and Greening2005; Sulakvelidze and Barrow, Reference Sulakvelidze, Barrow, Kutter and Sulakvelidze2005; Hagens and Loessner, Reference Hagens and Loessner2007). Significant reductions of C. jejuni, Salmonella and other bacteria that do not replicate at refrigeration temperatures required high MOIs, suggesting that bacterial killing was by lysis from without. High doses of phages against Listeria monocytogenes, which replicates at refrigeration temperatures, have also proved effective in some foods (Carlton et al., Reference Carlton, Noordman, Biswas, de Meester and Loessner2005) and the US Food and Drug Administration recently approved use of a commercial phage preparation to control Listeria spp. in ready-to-eat meat and poultry products (http://www.cfsan.fda.gov/~lrd/fr060818.html).
Current status of phage therapy in cattle, poultry and pigs
Research into phage therapy in cattle, poultry and pigs has addressed two major applications; reducing the impact of infections caused by animal pathogens on animal health and production, and control of zoonotic human pathogens by reduction of the bacterial load spread from animal reservoirs to foods and other vehicles of transmission to humans. While the objectives of these applications differ in some respects, both include potential preventative and therapeutic uses of phages, and they share a number of common requirements for successful development. Perhaps the most important requirement is a clear understanding of the epidemiology and pathogenesis of infection, the diversity of the target organisms, and the ability of candidate therapeutic phages to match this diversity in their common ecology.
Currently, the most promising animal health applications of phage therapy appear to be for acute intestinal, respiratory and systemic infections of young cattle, poultry and pigs caused by E. coli and Salmonella. Despite the rigor and early success in these studies, phage therapy for these and other animal diseases has not been developed further. Although this is largely due to the adoption of alternatives such as improved management practices, new vaccines and prophylactic use of antimicrobial drugs, it may also reflect some of the limitations of phage therapy, as pointed out by Barrow (Barrow et al., Reference Barrow, Lovell and Berchieri1998; Barrow, Reference Barrow2001). In addition to knowledge of the epidemiology of the agent, these included: the need for broad host range phages, the efficacy of phages in fluid as opposed to other matrices, the importance of timing phage use to critical points in infection and pathogenesis, and the need for practical and economic modes of administration. Some of these challenges are discussed below.
Interest in phage therapy for zoonotic pathogens in animals intensified more recently, when heightened awareness of the impact of food-borne infections with S. Enteritidis, E. coli O157:H7 and C. jejuni coincided with increasing concerns about antimicrobial resistance (Van den Bogaard and Stobberingh, Reference Van den Bogaard and Stobberingh2000). As a result, the studies of this application of phage therapy are relatively recent and vary widely in scope, designs and outcome. Also, most used newly isolated, partially characterized phages against one or only a few different strains of these very diverse target organisms. While these differences make interpretation difficult, there are several consistent observations. High doses of phages given orally can reduce intestinal levels of the target organisms by >1–2 log units for 1–3 days. Although transient, this effect can substantially reduce the pathogen load entering the food chain if treatment is administered immediately prior to slaughter. Currently, such applications appear to be most successful for C. jejuni in poultry. Treatment immediately pre-slaughter does not, however, impact farm prevalence, environmental dissemination of the organism, or hide or feather contamination. To effect control at the farm level, the same or greater reductions, and preferably elimination of the target organism would be required. However, elimination appears to be an unlikely outcome of phage therapy, as for most other interventions.
Challenges and opportunities
Further development of phage therapy for animal applications faces a number of practical and biological challenges that will require considerable research. The dose and route of administration are important economic and practical considerations in agriculture. The most practical route for all species is via feed or drinking water. Given the sensitivity of many phages to the low pH of the stomach, encapsulation of phages with pH protectants (Waddell and Johnson, Reference Waddell, Johnson and Mazzocco2004; Ma et al., Reference Ma, Pacan, Wang, Xu, Huang, Korenevsky and Sabour2008) would reduce inactivation of phages after ingestion and hence the total required phage dose. If required, i.m. inoculation may be feasible for calves, pigs and dairy cattle, but not for large-scale treatment of beef cattle. Large-scale i.m. inoculation of poultry, for example, during outbreaks of colibacillosis, is also not feasible, except perhaps in small, valuable breeder flocks. However, preventative injection of hatching eggs and spray treatment of the hatchery could possibly prevent the early onset of colibacillosis and colonization by Salmonella. Also, spray treatment of litter with phages (Smith et al., Reference Smith, Huggins and Shaw1987b) may reduce horizontal transmission of these and other organisms.
On a cautionary note, on-farm phage treatments will contribute to the diverse populations of host-specific phages that exist in these environments (Atterbury et al., Reference Atterbury, Dillon, Swift, Connerton, Frost, Dodd, Rees and Connerton2005; Hansen et al., Reference Hansen, Rosenquist, Baggesen, Brown and Christensen2007; Ott et al., Reference Oot, Raya, Callaway, Edrington, Kutter and Brabban2007; Callaway et al., Reference Callaway, Edrington, Brabban, Anderson, Rossman, Engler, Carr, Genovese, Keen, Looper, Kutter and Nisbet2008) and promote the emergence of phage resistant mutants or new genotypes of the target organism. This may reduce the effectiveness of the therapeutic phages, requiring alternating use of different phage cocktails in successive flocks or herds. Also, natural exposure and/or repeated use of phages, especially parenterally, may lead to development of phage neutralizing antibodies, reducing phage efficacy (Smith et al., Reference Smith, Huggins and Shaw1987b; Huff et al., unpublished data).
The selection of therapeutic phages is crucial for their efficacy in the field. The high specificity of phages is a major advantage over antibiotics with respect to preservation of the normal flora, but it also has limitations. Most lytic phages are specific for a narrow range of subtypes within target pathogens. Consequently, phage cocktails may be required, particularly in preventative applications against potentially diverse animal or zoonotic pathogens, and to combat the emergence of phage resistant strains. Alternatively, phages with a broad host range can be selected by extensive screening to identify those that lyse a range of pathogens with common surface receptors, such as those that target multiple isolates of two different pathogenic groups of E. coli (Viscardi et al., Reference Viscardi, Perugini, Auriemma, Capuano, Morabito, Kim, Loessner and Iovane2008). Such phages could also be engineered to target multiple receptors (Kropinski, Reference Kropinski2006). Fortunately, in most studies where phage resistant mutants have been sought and characterized, they have been less virulent or do not become predominant. Furthermore, phage resistant mutants can be used to select the most effective phages (Yoichi et al., Reference Yoichi, Morita, Mizoguchi, Fischer and Tanji2004).
Since in vivo and in vitro virulences of phages frequently do not correlate, informative evaluations of candidate phages require well-planned and rigorous in vivo studies in appropriate models. Poultry, pigs and sheep are reasonably manageable for controlled experimental trials, except with pathogens such as E. coli O157:H7 that require higher levels of biocontainment. The use of cattle however has obvious practical and economic challenges. Mice have been used in several studies of phage therapy for E. coli O157:H7 (see, for example, Sheng et al., Reference Sheng, Knecht, Kudva and Hovde2006), but their suitability as a model for cattle is questionable because they differ so greatly from ruminants in gastro-intestinal anatomy, physiology and microbiology. Also, because of age-related development, the age of experimental subjects within a species, would best match the intended age of application, as has been done in some of the above studies of phage therapy for C. jejuni in chickens. Additional challenges in animal experimentation in some cases include the selection and availability of the most appropriate infection models. For example, in control of E. coli O157:H7 in cattle, naturally occurring ‘super shedders’ are likely the best candidates for phage therapy (Matthews et al., Reference Matthews, McKendrick, Ternent, Gunn, Synge and Woolhouse2006) but are difficult to obtain in sufficient numbers.
The selected phages and the phage formulations for therapeutic use must be safe. Phages are generally considered innocuous, although their ability to carry and transduce virulence related and other unwanted genes are a concern. This possibility can be largely eliminated by testing phages for undesirable genes or by multiplex PCR or currently economical whole genome sequencing. Most of the animal studies in cattle, poultry, and pigs have used crude phage lysates produced in the target pathogenic hosts without any untoward effects, despite the presence of endotoxins, bacterial debris, in some cases, exotoxins. For large-scale commercial production, it would be safer to propagate phages in non-pathogenic or at least toxin-negative hosts. Also, for parenteral use, removal of endotoxin would reduce any associated risk, as would the use of lysis-deficient phages (Matsuda et al., Reference Matsuda, Freeman, Hilbert, Duff, Fuortes, Stapleton and Daly2005).
Conclusions
Phage therapy has potential for control of both animal and zoonotic pathogens in cattle, pigs and poultry. The information gained from recent studies provides a basis for much needed comprehensive and rigorous investigation of this potential. Hopefully, this will include more complete characterization of candidate phages, efficacy studies with administration by feasible routes of administration, and measured outcomes appropriate for the intended application. Additionally, thorough investigation of phage resistance and other dynamics of phage–host interactions in animal environments is essential. If the potential of phage therapy holds, its acceptance will depend on its efficacy, safety and cost relative to other interventions, and its complementarity with them. Complementarity is an important consideration, as it is likely that effective control of animal and zoonotic pathogens will require multiple interventions. Should these criteria be met, phage therapy will also require regulatory approval. As mentioned previously, the US Food and Drug Administration and the US Department of Agriculture have shown some flexibility in approving the use of phages for control of L. monocytogenes in ready-to-eat meat and poultry products, and not objecting to pre-harvest phage treatment of hides to reduce Salmonella and E. coli O157:H7 contamination. Whether this flexibility will extend to use in animals in the US and elsewhere remains to be seen. Most opinions acknowledge that the regulatory requirements in most Western countries are not geared to deal with therapeutic agents that multiply and the prompt introduction of new phages when resistance develops, as has been practiced in Eastern Europe for decades.
