Staphylococcus aureus is an opportunistic bacterial pathogen forming biofilms, which are known to be involved in a number of infective chronic diseases (Petti and Fowler, 2003; Ziran, 2007; Baldoni et al., 2009; Foreman and Wormald, 2010; Singhal et al., 2010, 2011; Jervis-Bardy and Wormald, 2012). These include osteomyelitis (Ziran, 2007), endocarditis (Petti and Fowler, 2003), as well as infections of indwelling devices (Baldoni et al., 2009). In addition, sinonasal bound biofilms of S. aureus are known to impact negatively in chronic rhinosinusitis (CRS). The presence of such infections in CRS reportedly leads to more frequent out-patient visits (Singhal et al., 2011), increased risk of recurrent infections and antibiotic use (Jervis-Bardy and Wormald, 2012) as well as poorer post-operative progression (Foreman and Wormald, 2010; Singhal et al., 2010). Such biofilms are up to 1000-fold more tolerant of current antibiotic therapies than their planktonic counterparts (Anwar et al., 1990). Further, increased levels of antibiotic resistance observed in pathogenic bacteria around the globe (Roca et al., 2015) also limit the success of antibiotic therapies. It is important that new therapies which effectively treat such infections are identified.

One alternative to antibiotics, originally described in the early 1900s and coming back into focus, is bacteriophage “phage” therapy (Carlton, 1999). Beneficially, phages are not only effective against planktonic infections, but can also infect and lyse biofilm-bound cells (Doolittle et al., 1995; Corbin et al., 2001; Tait et al., 2002; Sillankorva et al., 2004; Curtin and Donlan, 2006; Cerca et al., 2007; Lu and Collins, 2007; Carson et al., 2010; Fu et al., 2010). Recently, we have shown that topical phage therapy has potential against pathogenic S. aureus bacterial biofilms, using an animal model of rhinosinusitis (Drilling et al., 2014a). Our previous work demonstrated the safety of once-daily phage application into the frontal sinuses of sheep for 3 days (Drilling et al., 2014a). The first aim of the current work was to examine the safety of phage administration for a substantially longer period of time. In the current study we investigated the effects of a longer-term (20 day) application of twice-daily frontal sinus bacteriophage flushes in sheep. A cocktail of two phages against S. aureus, designated NOV012, was used. NOV012 contains two highly-characterized phages, K710 and P68. Both the parental version of phage K710, phage K (O'Flaherty et al., 2004), and phage P68 (Vybiral et al., 2003) have had their genomes completely sequenced, and they have been shown to lack any known genes that could increase the virulence of S. aureus or confer resistance to antibiotics (Vybiral et al., 2003; O'Flaherty et al., 2004). Phages K710 and P68 have been shown to be active against a wide range of S. aureus isolates from the United Kingdom and Europe (Takac and Blasi, 2005; J Clark, personal communication). The second aim of this work was to extend these observations by examining the susceptibility of local (Australian) S. aureus isolates to the phages in NOV012.

Phage therapy has almost a 100 year history of human application for treating bacterial infections (Carlton, 1999). Accompanying this long history is an outstanding record of safety. There are numerous reports of the safety of phage therapy in animal (Soothill, 1994; Biswas et al., 2002; Wills et al., 2005; McVay et al., 2007; Hawkins et al., 2010; Park et al., 2014) and in human trials (Markoishvili et al., 2002; Bruttin and Brussow, 2005; Rhoads et al., 2009; Wright et al., 2009; Sarker et al., 2012; McCallin et al., 2013). Our previous work showed that short-term bacteriophage application is safe when applied topically to the sheep sinonasal region (Drilling et al., 2014a). This study aimed to extend this work by assessing longer-term (20 days) sinonasal phage application. This safety data is needed to support the further preclinical development of phage therapy in general and NOVO12 specifically as it is likely that phage therapy would be used in patients for an extended period of time, at least 2 weeks.

It is recognized that phage may interact with some aspects of the host immune system, and it is important to ensure they do not elicit adverse immune responses (Kaur et al., 2012). Supporting our previous work (Drilling et al., 2014a), this study shows that 20 days of topical phage therapy did not modify or damage the architecture of the sinus mucosal lining. Phage application for this extended period did not appear to increase or alter the profile of immune cells in the sinus mucosa. A limitation of this study is that only the presence of the cells was examined, whereas stimulation of the cells in relation to immune effector molecules such as cytokines was not investigated. Previous work has investigated this parameter, applying phage T4 or purified phage T4 proteins to mice and humans (Miernikiewicz et al., 2013). It was shown that such products did not stimulate the production of inflammatory related cytokines and reactive oxygen species (ROS; Miernikiewicz et al., 2013). In contrast, it has been suggested that through complex bacterial-phage-host interaction, phage can reduce ROS production (Przerwa et al., 2006; Miedzybrodzki et al., 2008). Recent research has also investigated the effect on phage infectivity when exposed to cell lines that mounted an inflammatory response to the phage. This study found that the phage still retained infectivity against bacterial cells (Khan Mirzaei et al., 2016).

It is not known whether phages can traverse the sinonasal mucosal barrier to gain access to the bloodstream. During the 20 days of treatment, no phages were detected in the bloodstream. A limitation of this study, which relied on plaque assays, is that it could not determine whether inactivated phage or phage genomes entered the bloodstream. A further limitation is that this study did not address the question of whether the sheep developed antibodies against phages. It has been shown that neutralizing antibodies may be produced following phage treatment (Kucharewicz-Krukowska and Slopek, 1987), and our future work will examine this possibility. This is of interest because it is not yet confirmed whether development of anti-phage antibodies will have a negative impact on phage therapy (Sulakvelidze et al., 2001). Recent research however has shown positive results, showing that development of antibodies may not necessarily strongly impact the clinical success of phage therapy (Zaczek et al., 2016; Lusiak-Szelachowska et al., 2017).

In addition to examining the safety profile of phage cocktail NOV012, this study also investigated the NOV012 host range in the context of CRS S. aureus infections. The findings of this study build on our previous work (Drilling et al., 2014a) that suggests using a cocktail of phage assisted in overcoming issues of matching phage to bacteria. We found that phage K710 was effective against 59% of the S. aureus strains in our panel of isolates, which was increased to 85% with addition of phage P68. Using cocktails rather than a single phage has the additional benefit of reducing the rate of generation of bacteria resistant to phage infection (Chan et al., 2013).

This work has implications beyond the treatment of CRS. For example, nasal colonization with S. aureus increases the risk of surgical wound infection (Perl and Roy, 1995). S. aureus decolonization reduces such risks (Bode et al., 2010; Chen et al., 2013), and phage have potential to achieve this. Other potential applications include topical treatment of burn wounds (Mousa, 2005) and treatment of indwelling catheter infections (Piraino, 2000). Importantly, this therapy can be broadened to target other bacterial pathogens.

This work confirms that the NOV012 phage cocktail infects a broad range of S. aureus isolates, including a number of MRSA isolates, from CRS patients. Further, we find that longer term (20 day) topical application of the cocktail is safe for sheep sinonasal application. This safety data supports the potential for the use of phage as a topical antimicrobial treatment in CRS and will help build the profile of the product to lead to the ability to use the product in clinical trials and eventually commercially.

AD executed experiments, wrote first draft. MO assisted with experiments, edited first draft. DM assisted with experiments. CJ assisted with data interpretation. PS and SV helped with experiment design, data interpretation, manuscript writing. JC helped with data interpretation. PW designed the study, manuscript writing.