**6.3 Failures**

As the history of published successful bacteriophage treatments of enteric disease is reviewed, it is readily evident that such reports, while often dramatic in effect, are relatively sporadic during the last approximately 60 years. Given that experimental failures frequently are not published, as the cause of failure can often not be ascertained, the authors suspect that history is replete with unpublished examples of failures to treat enteric Enterobacteriaceae infections.

Our laboratory, and others, have demonstrated that resistance to bacteriophages selected against *Salmonella* isolates quickly occurs, often in a single passage (Bastias et al, 2010). When bacteriophage cocktails of 71 different bacteriophages selected for treatment of experimental *Salmonella enteritidis* infections in chickens, a brief reduction in enteric

Biswas et al. (2002) successfully cured *Enterococcus faecium*-infected mice with bacteriophage therapy. Mice were treated with bacteriophages just 45 minutes after infection with bacteria. Treatment at a multiplicity of infection (MOI) level of 0.3 to 3.0 was able to cure all of the infected mice. However, lower MOIs of 0.03 to 0.003 resulted in just 60% and 40% survival of mice, respectively. They also noted that bacteriophage treatment could be delayed for up to five hours after infection. However, if treatment was delayed for 18 or 24 hours, only 50%

Berchieri et al. (1991) treated broiler chickens infected with *Salmonella typhimurium* (ST) with bacteriophages and found that the levels of ST could be reduced by several logs, and mortality associated with ST was reduced significantly. However, ST was not eliminated and it returned to its original levels within six hours of treatment. Also, the bacteriophages did not persist in the gastrointestinal tract for as long as the *Salmonella* was present. In fact, bacteriophages persisted only as long as they were added to the feed. In order to be effective, bacteriophages had to be administered in large numbers, and soon

In 1998 Barrow et al. prevented morbidity and mortality in chickens using bacteriophages lytic for *E. coli*. When chickens were challenged intramuscularly with *E. coli* and simultaneously treated with 106 – 108 pfu of bacteriophages the mortality was reduced by 100%. This study also demonstrated that bacteriophages can cross the blood brain barrier, and furthermore that they can amplify in both the brain and the blood. Similarly, a number of other researchers have shown that bacteriophages can be useful for treating non-enteric *E. coli* infections. Extensive research about the effects of bacteriophages on colibacillosis in broiler chickens has shown that bacteriophages can treat respiratory infections (Huff et al, 2002a-b; Huff et al, 2003a-b). Treatment was most successful when bacteriophages were directly applied to the infected area or injected into the bloodstream. This observation is

However, such successes do not necessarily translate into effective enteric treatments. Hostassociated pressure against pathogen infections may predispose systemic bacteriophage therapy toward success. In these cases, where bacteriophage(s) are used to treat systemic or tissue-associated infections, an acute efficacy of merely reducing the infection load by 90% or more, could greatly reduce mortality and reduce the duration and magnitude of disease. In the intestinal lumen, host pressures against the infection may not be as severe and many Enterobacteriaceae are capable of free living status within the gut without eliciting robust acquired immune responses from the infected animal. In these cases, a temporary reduction

As the history of published successful bacteriophage treatments of enteric disease is reviewed, it is readily evident that such reports, while often dramatic in effect, are relatively sporadic during the last approximately 60 years. Given that experimental failures frequently are not published, as the cause of failure can often not be ascertained, the authors suspect that history is replete with unpublished examples of failures to treat enteric

Our laboratory, and others, have demonstrated that resistance to bacteriophages selected against *Salmonella* isolates quickly occurs, often in a single passage (Bastias et al, 2010). When bacteriophage cocktails of 71 different bacteriophages selected for treatment of experimental *Salmonella enteritidis* infections in chickens, a brief reduction in enteric

in enteric colonization may not be as likely to be curative, as discussed below.

recovery was seen.

after infection with ST.

**6.3 Failures** 

Enterobacteriaceae infections.

consistent with previous research discussed above.

colonization was noted during the first 24 hours, but rebound levels were similar to controls within 48 hours, even with repeated or continuous dosage of the bacteriophage cocktail (Higgins et al, 2007). Because of the demonstrated temporary reduction in enteric colonization in these studies, effective bacteriophages were demonstrably able to pass to the lower gastrointestinal tract. As continued treatments failed to maintain this reduction, development of resistance by the enteric *Salmonella enteritidis* is the most likely explanation. In order to potentially deliver higher levels of bacteriophage, several attempts to protect the bacteriophage cocktail through the upper gastrointestinal tract were made in our laboratory.

Pre-treatment of infected poultry with antacid preparations designed to reduce the acidity of the proventriculus (true stomach) were successful in increasing the number of administered bacteriophage that successfully passed into the intestinal tract, but this treatment did not improve the outcome of bacteriophage treatment of *Salmonella enteritidis* infection (Higgins et al, 2007).

An alternative approach is to select for alternative non-pathogenic bacteriophage hosts which could potentially "carry" bacteriophage through the gastrointestinal tract and, with continuous dietary administration of the non-infected alternative host bacterium, provide a means of amplification within the gut of the host (Bielke et al., 2007a). Bielke and co-workers demonstrated that non-pathogenic alternative hosts can be selected for some bacteriophages that were originally isolated using a *Salmonella enteritidis* target (2007b). This approach, which has potential utility for amplification of large numbers of phage without the necessity to thoroughly separate bacteriophage from a pathogenic target host, was also used to create a potential "Trojan Horse" model for protecting the bacteriophages through the upper gastrointestinal tract, thus potentially providing a vehicle for enteric amplification of those surviving bacteriophages. In these studies, neither the Trojan Horse approach, nor the continuous feeding of the alternative host bacteria as a source of enteric amplification, were effective in producing even more than a transient reduction in enteric *Salmonella* infections.

Through these failures, many investigators have concluded that the escape of even a minority of target bacteria within the enteric ecosystem allows for almost immediate selection of resistant target bacteria and rebound to pre-treatment levels of infection may even exceed the levels of non-treated controls in some cases.

#### **6.4 Potential strategies to overcome failures**

Bacteriophage resistance is an important component of therapy to overcome before bacteriophages can really be a viable antimicrobial for infection. The generation time for bacteria is typically short enough that mutants with bacteriophage resistance can emerge within hours (Higgins et al, 2007; Lowbury and Hood, 1953). One possibly strategy to overcome this problem is administration of multiple bacteriophage isolates for treatment, but resistance is difficult, if not impossible, to predict and combining the correct cocktail of bacteriophages to overcome resistance would be a blind guess in most cases.

The most success is likely to come from treating points in the system that are continually bombarded with bacteria that have not been previously subjected to the bacteriophages being used for treatment. Also important for this system is keeping exposure of the bacteria to bacteriophages to a minimal amount of time. If the bacteriophages interact with the bacteria for long periods of time, the bacteria will become resistant. Food and meat processing facilities are an excellent example. As live animals enter a processing facility, the bacteria have not likely been exposed to the bacteriophages used to treat the infection. This greatly increases the chances of success.

Alternative Strategies for *Salmonella* Control in Poultry 269

number of *Salmonella* serovars, each with individual epitopes that do not elicit crossprotection against other serovars, there has been little traditional emphasis on development of generic *Salmonella* vaccines. Primarily, killed vaccines, which generally must be administered parenterally (through injection), have been applied to protect against systemic infections, and although they have been shown to reduce colonization and shedding, the protection provided by these vaccines has limited ability to stop intestinal colonization. They predominantly stimulate both humoral (circulating IgM and IgG) and cell-mediated responses, but are quite ineffective at generating mucosal immunity as secretory IgA antibody stimulation is very low through this type of vaccination. This is important because, whereas both systemic (humoral and cell-mediated) and mucosal immunity can reduce the chances of disease and mortality, only the mucosal portion of this adaptive immune response is capable of protecting animals from infection. The key to inducing both an adaptive systemic and mucosal response has traditionally been through the use of the mucosa as a "portal of entry" for live but weakened (attenuated) vaccines. However, the use of such vaccines for protection against *Salmonella* infection have been tremendously limited due to the very large number of different antigens presented by the more than 200 serotypes that can infect domestic animals and man, with more than 38 of these commonly infecting

poultry within the United States, as discussed below (Hargis et al., 2010).

2007; Huang et al., 2008; Liu et al., 2008; Ceragioli et al., 2009; Deguchi et al., 2009).

This type of vaccine uses a genetically modified bacterium to express a heterologous antigen. Oral live attenuated *Salmonella* vaccine vectors expressing recombinant foreign antigens have previously been shown to stimulate systemic, mucosal, humoral, and cellmediated immune responses against *Salmonella* (Mollenkopf et al., 2001; Koton and

One approach to solving the problem of serotype variation among the common paratyphoid strains of *Salmonella*, which are often not a disease-causing problem for poultry but rather create a source of foodborne illness for consumers, is the identification of "universal epitopes" that are shared among all *Salmonella* isolates. This concept has been established for a number of pathogens and is based on the identification of a minor surface structure (antigen or epitope) which does not cause robust immune reaction during infection, but which can be targeted for protection if the antigen is presented in a way that tricks the animal into responding robustly. Some of these are relatively minor antigens which are highly conserved among related organisms – usually because they involve biological function. Since small peptide sequences that are biologically functional cannot vary in sequence, organisms that carry a mutation for such sequences are often either lethal or sufficiently detrimental to cause these to not be successful over time (Neirynck et al., 1999). A well-described example of this phenomenon is a small 23 amino acid peptide on the surface of Type A Influenza viruses named M2e. This peptide is part of an ion transport channel which is necessary for viral activation. Mutations in this sequence undoubtedly occur frequently, but since the 1918 Spanish Influenza outbreak, all Type A Influenza isolates share a highly conserved core sequence for this peptide (Layton et al., 2009). Although natural influenza infection does not result in a robust immune response to this peptide sequence, tricking the animal into producing a robust response has resulted in protective immunity in several animal species (Neirynck et al., 1999; Mozdzanowska et al., 2003; Fiers et al., 2004; Zou et al., 2004). In recent years, the rapid increase in molecular biological techniques has led to the development of more sophisticated vaccines, of which live recombinant bacterial vectored vaccines are one of the most promising (Ashby et al., 2005; Zhang et al., 2006; Duc et al., 2007; Kajikawa et al., 2007; Uyen et al., 2007; Yang et al.,

Higgins and co-workers (2005) successfully treated turkey carcasses at a processing facility with bacteriophages specific to the *Salmonella* to which they were infected. This process was effective when either an autogenous bacteriophage treatment targeted to the specific *Salmonella* strain infecting the turkeys was used, or a cocktail of nine wide host-range *Salmonella*-targeting bacteriophage were used. Similarly, a bacteriophage treatment for cattle carcass contamination has been effective at reducing the *E. coli* 0157:H7 load at processing has been developed and commercially licensed in the United States. These successes avoid development of bacteriophage resistance by applying treatment at a single point during production, in an environment where proliferation of the target organism is extremely limited. In this way, since the target organism is never intentionally exposed twice to the same treatment, resistance is unlikely to ever increase beyond the naturally-occurring resistance to the bacteriophage (or cocktail) used.

One of the most well documented successes of published treatment of enteric Enterobacteriaceae infections with bacteriophages was the study of Smith and Huggins (1983) as described above. It is notable that in this successful study, the bacteriophage cocktail used was a combination of two bacteriophages, but the second was isolated using the target organism which was resistant to the first bacteriophage. This approach of selecting for bacteriophage isolates using target bacteria that are resistant to sequential bacteriophage treatments was not used in the work of Higgins et al (2007), or in several other published studies. Higgins and co-workers (2007) used a collection bacteriophages, independently isolated from different sources and with several different plaque morphologies, suggesting that a number of different bacteriophages were employed – and failed to persistently reduce enteric colonization.

It is possible that one of the most notable exceptions to the many failures to treat enteric Enterobacteraceae infections during recent years, that of Smith and Huggins (1983), provides a singular clue as to the potential for enhancing the likelihood of enteric Enterobacteriaceae efficacy. It is possible that selection of multiple bacteriophages for the same target cell phenotype results in selection of bacteriophages that are effective through identical mechanisms of adhesion, penetration, replication, and release. When new bacteriophages are isolated for efficacy against sequentially resistant isolates of the target bacteria, and these are combined for administration as a cocktail, the ability of the target cell to shift phenotype may be severely limited, resulting in a much larger proportion of target cell reduction, thereby increasing the probability of elimination or cure.

Clearly, widespread bacteriophage treatments with Enterobacteriaceae have not been adopted for any animal species during the last 60 years and successful research in this area has been modest and sporadic. Nevertheless, the occasional reports by reputable scientists in solid journals must indicate that there is potential for improved therapeutic efficacy of bacteriophages for this purpose. With the diminution of new antimicrobial pharmaceuticals and the widespread resistance among many pathogenic enteric Enterobacteriaceaes, a breakthrough in this area is sorely needed.
