Impact of the Novel Prophage ϕSA169 on Persistent Methicillin-Resistant Staphylococcus aureus Endovascular Infection

Bacteriophages are viruses that invade the bacterial host, disrupt bacterial metabolism, and cause the bacterium to lyse. Because of its remarkable antibacterial activity and unique advantages over antibiotics, for instance, bacteriophage is specific for one species of bacteria and resistance to phage is less common than resistance to antibiotics. Indeed, bacteriophage therapy for treating infections due to multidrug-resistant pathogens in humans has become a research hot spot. However, it is also worth considering that bacteriophages are transferable and could cotransfer host chromosomal genes, e.g., virulence and antimicrobial resistance genes, while lysogenizing and integrating into the bacterial chromosome (prophage), thus playing a role in bacterial evolution and virulence. In the current study, we identified a novel prophage, ϕSA169, from a clinical persistent MRSA bacteremia isolate, and we determined that ϕSA169 mediated well-defined in vitro and in vivo phenotypic and genotypic signatures related to the persistent outcome, which may represent a unique and important persistent mechanism(s).

genotypic signatures related to the persistent outcome, which may represent a unique and important persistent mechanism(s). KEYWORDS MRSA, bacteriophage, persistent endovascular infection M ethicillin-resistant Staphylococcus aureus (MRSA) is a leading cause of lifethreatening endovascular infections, especially bacteremia and infective endocarditis (IE) (1,2). Persistent MRSA bacteremia (PB; defined as Ն7 days of positive blood cultures in the presence of antibiotic therapy) represents ϳ15 to 30% of such infections (3,4). Most PB strains appear to be susceptible in vitro to standard-of-care anti-MRSA antibiotics (e.g., vancomycin [VAN] and daptomycin [DAP]) by CLSI breakpoints, yet persist in vivo despite seemingly appropriate antibiotic therapy (4)(5)(6). This paradox has fostered a number of investigations in our laboratories focused on a specific molecular mechanism(s) that underlies the in vivo persistent outcome.
Prophage elements have been demonstrated to contribute to the pathogenesis of staphylococcal infections (7)(8)(9). In some cases, prophages can impact bacterial fitness and host-microbe interactions, genetically correlating with well-defined immunomodulatory virulence factors (e.g., lukF-PVL) (7,10), as well as staphylococcal global regulators (e.g., sigB) (9). Importantly, there is growing evidence suggesting prophages might also be involved in bacterial persistence due to their effect on promoting biofilm formation and triggering the stringent response, which is a metabolic signaling pathway activated by nutritional stresses (8,9).
Our recent whole-genome sequence (WGS) analyses identified a novel temperate prophage, SA169, in a prototypical clinical PB strain (300-169) but not present in a genetic-background-matched (clonal complex, agr, and SCCmec types) RB strain (301-188; RB defined as initial MRSA bacteremia resolved within 2 to 4 days of antibiotic therapy) (11). More specifically, the WGS data demonstrated that SA169 was inserted into a chromosomal gene, SAUSA300_1858 (encoding protein YfkAB), in the PB strain; this gene was uninterrupted in the RB strain (Fig. 1A). A transmission electron microscopy image of a viable SA169 particle is presented in Fig. 1B. The genome of prophage SA169 has a size of ϳ44 kb, shows ϳ79% similarity to the genome of phage 11, and can be grouped with SA5 by integrase homology (Fig. 1C) (11,12). Although the phenotypic and genotypic distinctions between PB and RB strains have been well described previously (5,13), the causal interaction between SA169 and the persistent outcome has not yet been studied. Thus, the current investigation was aimed to reveal the role of prophage SA169 in the persistence of MRSA endovascular infection.

RESULTS
Successful transduction of SA169. The titers of SA169 in the recipient strain, RN4220, after exposure to the supernatant of filtered 300-169 stationary-phase cultures with/without mitomycin C (MMC) exposure and plasma from 300-169-infected rabbits with IE were 7 ϫ 10 7 PFU/ml, 3.3 ϫ 10 5 PFU/ml, and 80 PFU/ml, respectively ( Fig. 2A). Similar results were observed following the transduction of SA169 into the RB 301-188 strain. For instance, the titer of SA169 in RB 300-169p with MMC exposure was 3 ϫ 10 7 PFU/ml. PCR results indicated that all study strains, except 301-188 and 301-188p, contained the SAUSA300_1858-int adjunction region and the phage gene dnaC (Fig. 2B). These results suggest that SA169 from the donor strain (PB 300-169p) was successfully integrated into the SAUSA300_1858 gene in the recipient RB strain 301-188.
WGS further confirmed the transduction of SA169 in the RB 301-188 strain. The WGS data indicated that both 300-169p and 301-188p::SA169 contained identical genomic DNA sequences of SA169 (see Fig. S1 in the supplemental material). SA169 integrated into the SAUSA300_1858 gene in the RB 301-188p strain at the same position as in the genome of the PB 300-169p strain (Fig. S1). Of the sequences other than that of SA169 and the plasmid pALC1766, there were only 3 single-nucleotide polymorphisms (SNPs) demonstrated among 301-188, 301-188p, and 301-188p:: SA169 (Table S1). Of these three SNPs, one synonymous mutation was found in the gene encoding a hypothetical protein in 301-188p::SA169, compared to 301-188 and 301-188p. The other two were identified in genes encoding a siderophore biosynthesis protein and phosphoenolpyruvate carboxykinase in 301-188p versus 301-188, suggesting that these two SNPs might have occurred during the generation of 301-188 carrying pALC1766 (13) and have no substantial impacts on PB-related phenotypes (Table S1).
Transduction of SA169 converted the RB strain into a strain with PB-like phenotypes. The correlation among cap5 promoter activation, bacterial growth, ATP levels, and VAN susceptibility has been well elucidated (13,14). For example, our previous studies indicated that the higher growth rates exhibited by PB versus RB strains correlated with earlier activation of cap5, lower ATP levels, and higher survival rates when exposed to human-simulating VAN concentrations versus RB strains (13). Similarly, in the current studies, we demonstrated that RB 301-188p::SA169, similar to the PB 300-169p strain, had significantly higher cap5 promoter activation and higher growth rates than its parental RB 301-188p strain ( Fig. 3A and B; P Ͻ 0.05). In addition, the 301-188p::SA169 strain, as well as the 300-169p strain, exhibited significantly lower ATP levels and higher survival rates during VAN exposure than the RB 301-188p strain, even with the same VAN MICs ( Fig. 3C; P Ͻ 0.05). To better understand the underlying genetic mechanisms of the faster growth of 301-188p::SA169, we assessed the expression levels of purF in the study MRSA strains; we have previously reported that higher expression of purF results in higher growth rates and subsequently leads to the reduction of VAN susceptibility in PB strains (13). In this study, the 300-169p and 301-188p::SA169 strains showed significantly higher purF expression than the strain 301-188 counterpart ( Fig. 4A; Ͼ2-fold, P Ͻ 0.05), indicating that the faster growth in 301-188p::SA169 was at least partially due to the higher purF expression versus 301-188p.
The RB strain containing SA169 exhibited earlier global regulator transcription profiles than its isogenic RB parental strain without SA169. Our previous studies demonstrated that PB strains exhibited significantly earlier activation of key global regulators involved in pathogenesis of S. aureus, including asp23, sarA, agr RNAIII, and sae versus RB strains (13). Importantly, temporal agr RNAIII expression and agr functional profiles have been used as biomarkers to predict the persistent outcomes in endovascular MRSA infections (5,13,15). In this study, we demonstrated that the transcriptions of these global regulators were higher in early-exponential-growth-phase cells in PB 300-169p and RB 301-188p::SA169 strains versus the parental RB 301-188p strain (Fig. 4B).
The RB strain containing SA169 exhibited significantly reduced VAN treatment efficacy in the IE model. To validate the hypothesized effect of SA169 in the PB outcome in vivo, an experimental IE model was employed. Without VAN treatment, animals infected with 301-188p or 301-188p::SA169 showed slightly lower MRSA counts (not statistically significant) in cardiac vegetation and kidney than the 300-169p counterpart (Fig. 5). Of note, similar to the PB 300-169p strain, 301-188p::SA169infected animals had no response to VAN treatment and had significantly higher MRSA densities in all target tissues than VAN-treated animals infected with the 301-188p counterpart ( Fig. 5; P Ͻ 0.0001).

DISCUSSION
Bacteriophages are ubiquitous in the environment and diverse in size, morphology, and genomic organization (17). Upon invading the bacterial hosts, the phage disrupts bacterial metabolism and ultimately lyses the bacteria, thus indicating its potential be an alternative treatment for human infections, especially due to antimicrobial-resistant pathogens (18). In contrast to bacterial lysis, some bacteriophage could also lysogenize and integrate into the bacterial chromosome (often called prophage) and impact bacterial evolution and virulence through toxin production, biofilm formation, and tissue adhesion (8,9,19). Additionally, bacteriophage is transferable and, in some cases, could cotransfer host chromosomal genes, e.g., virulence (20) and antimicrobial resistance (21) genes, even between different bacterial species (22), raising the concern about its safety in clinical usage (23).
Recent studies have identified bacteriophages as important contributors to persistent outcomes, via regulation of global virulence gene regulators and biofilm formation in other bacteria (8,9). However, the impact of bacteriophages on persistent MRSA endovascular infections has not been previously studied. We recently demonstrated that the prototypical PB strain, 300-169, contains a unique prophage (SA169) which is not present in a genetically matched RB strain, 301-188 (11). This finding led us to hypothesize that this bacteriophage might contribute to distinct PB phenotypes, particularly early-onset activation of global regulators (5,13,15), higher growth rates (13), reduced susceptibility to VAN (13), and greater biofilm formation (24).
In the current investigations, we first noted that viable SA169 particles were observed in both recipient strains (RN4220 and RB 301-188) from not only an in vitro (with transduction) experiment with 300-169 (using pALC1766 as an antibiotic resistance marker for genetic transfer) but also in the plasma of 300-169-infected rabbits with IE. These results indicated that, as a typical phage, SA169 can infect bacteria under both in vitro and in vivo conditions. Not surprisingly, similar results have been reported by Acheson et al. and Cornick et al., showing that viable phage virions were found in Escherichia coli strains in host gastrointestinal tract (25,26). Meanwhile, the WGS data verified the successful transfer of bacteriophage SA169 from PB 300-169 into RB 301-188, generating the 301-188::SA169 strain harboring pALC1766; this further indicated that SA169 is a typical "temperate" phage, which has the ability to integrate into the recipient strain's chromosome (19). The insertion site of SA169 in 301-188p::SA169 is identical to the site in observed in PB 300-169. This is likely achieved by the prophage-encoded integrase (int gene) and the enclosing lysogeny module which processes the site-specific recombination (i.e., integration) of the phage (27). Bacteriophage-mediated spread of chromosomal genes is prevalent in both Gram-negative and Gram-positive bacteria (28), in particular, the spread of bacterial virulence genes (e.g., the S. aureus pathogenicity islands [SaPIs], frequently transferred with staphylococcal phage 80␣ within diverse staphylococcal species) (20). Furthermore, such phage-mediated genetic transduction (e.g., SaPIs) has been documented to occur between S. aureus and Listeria monocytogenes (22). In this regard, the transduction of SA169 may contribute to the distribution of bacteriophage-mediated pathogenesis factors (29).
The current study focused on the role of this novel SA169 in PB signatures in vitro and therapeutic outcomes in vivo. First, the temporal gene transcription profiles revealed that all studied virulence global regulators, including sigB, sarA, agr RNAIII, and sae, activated significantly earlier in 301-188p::SA169 than its isogenic parental RB strain 301-188p. This is in line with the observation by Fernández et al. that the lysogenization of a prophage activated the expression of sigB (9). In addition, greater ␦-hemolysin production and biofilm formation were detected in 301-188p::SA169 than its isogenic RB strain 301-188p, which is consistent with previously published results that key global regulators are responsible for these phenotypes (e.g., early-onset activation of agr RNAIII is responsible for higher ␦-hemolysin production [5], and sigB and sarA regulate biofilm formation [9,30]). Similarly, prophage 11-mediated enhancement of biofilm formation was observed in S. aureus (9). Fernández et al. speculated that phage-promoted biofilm formation might be due to the increased extracellular eDNA, which is likely a consequence of cell lysis by phage (8). It has been widely investigated that bacterial prophages can spontaneously induce bacterial evolution of specific virulence factors, such as platelet-binding ability, Shiga toxin production, biofilm formation, etc. (29). Likewise, curing of all four prophages in S. aureus Newman substantially compromised its virulence (7). Interestingly, the phage gene phi80_gp05 from SA169 has also been identified in 80␣ and 11 and reported to be involved in sigB activation and biofilm formation in S. aureus (9). Further investigations of the SA169 genes' function, including that of phi80_gp05, and their contributions to the PB outcomes are needed. Taken together, SA169 might contribute to the PB outcome by triggering the activation of key virulence global regulators and their respective downstream genes, which might well impact the organism's net fitness and survival potential.
Second, we demonstrated that like the PB strain, the SA169-transducted RB strain 301-188p::SA169 exhibited earlier activation of cap5 than its isogenic parental RB strain 301-188p, which suggests the RB strain containing the phage reached stationary growth phase earlier than its parental RB strain. Accordingly, significantly higher growth rates and shorter generation times were observed in PB 300-169p and 301-188p::SA169 than in the parental RB strain 301-188p. These results are in line with our previous studies demonstrating that earlier activation of cap5 positively correlates with higher growth rates (13,14). Similar results have been observed by Bossi et al. showing that prophage could improve the fitness of Salmonella enterica serovar Typhimurium (31). The link among higher growth rates, lower ATP levels, and higher survival rates during VAN exposure in PB than in RB strains has been well established (13,14). Indeed, this same linkage was observed in 301-188p::SA169 versus its isogenic parental RB strain 301-188p. In this regard, this led us to hypothesize that reduced VAN susceptibility mediated by the prophage SA169 might offer a survival advantage in the presence of VAN exposure in vivo.
Third, similar to the PB strain 300-169p, the 301-188p::SA169 strain showed significantly higher expression levels of the purine biosynthesis pathway gene purF than did its isogenic parental RB strain. This genotype was reflected in faster growth in PB 300-169p and SA169-transducted RB strain 301-188p::SA169, as purine biosynthesis is essential for cell growth via nucleotides synthesis (32,33). Similar results were also reported by Fernández et al., demonstrating that expression of purine biosynthesis pathway genes (e.g., purH) was upregulated after the lysogenization of 11 or IPLA-RODI in S. aureus (8,9). In addition, Edlin et al. showed that prophages can benefit E. coli fitness under glucose-limited conditions by accelerating its metabolic activity (34). The precise mechanism(s) of prophage-mediated elevation of purine biosynthesis pathway gene expression in S. aureus is under current investigations in our laboratory.
Most importantly, the in vivo relevance of SA169 in PB outcomes was demonstrated in the experimental IE model. Animals infected with the SA169-containing RB strain 301-188p::SA169, similarly to those infected with its genetically matched PB strain 300-169p, exhibited no response to VAN treatment versus animals infected with its isogenic parental RB strain. The PB outcome might be due to a combination of SA169-mediated phenotypic and genotypic profiles, such as earlier onset of key global regulators and higher growth rates, subsequently enhanced biofilm formation, and reduced VAN susceptibility. Global regulators have been shown to impact transcription of key virulence genes and antimicrobial susceptibility in vivo, including IE (35)(36)(37). In addition, the role of sigB in promoting bacterial intracellular persistence by defending S. aureus from invading immune cells has been well studied (38). We, along with other investigators, have shown that sarA is crucial in persistent MRSA infection via its effect on the host defense system and biofilm formation, within which bacteria become more resistant to antibiotics (16,39).
There are some important limitations to this study. Besides the remarkable PB outcomes of the lysogenized RB strain, a new question also comes along with the exciting findings: does curing prophage SA169 have a reserve effect on the PB strain? Gaining prophage SA169 results in acquisition of PB phenotypes; therefore, it is possible that curing SA169 may ultimately lead to loss of PB phenotypes. In addition, we studied only one clinical genetic-background-matched PB-RB strain pair. It would be interesting to study if the SA169 transduction in RB strains with other genetic backgrounds has a similar PB phenotype. Interestingly, we recently carried out genome analysis using 542 S. aureus genome databases published in the National Center for Biotechnology Information (NCBI), including 313 MRSA and 229 methicillin-susceptible S. aureus (MSSA) strains, by assembling nucleotide sequences of SA169. This analysis was performed with online tools available on the NCBI website (https://www.ncbi.nlm .nih.gov/guide/taxonomy/) and demonstrated that there are 205 genomes (141 MRSA and 64 MSSA) showing Ն95% identity on Ն75% length of the SA169 sequences. These results suggest that the SA169-like sequences are present in S. aureus genomes. However, there are no data available in the NCBI database regarding what proportion of these S. aureus strains is PB versus RB. Therefore, further investigations are needed to define whether the transduction with SA169 is a common and determinative event in the S. aureus strains for the PB outcome.
In summary, the present findings are, to our knowledge, the first to investigate the importance of the novel prophage SA169 in the PB outcomes. Although the mechanism(s) of bacteriophage-mediated PB outcome is not well understood, these data underscore this prophage as a critical factor in persistent MRSA endovascular infections.

MATERIALS AND METHODS
Bacterial strains, plasmids, and growth medium. Bacterial strains and plasmids used in this study are listed in Table 1. PB strain 300-169 was isolated from a patient with 16 days of persistent MRSA bacteremia, while RB strain 301-188 was obtained from a patient with 2 days of MRSA bacteremia (5,15,40). Both strains represent the initial bloodstream isolate. The two MRSA strains have a similar genetic background (e.g., CC45, agr I, and SCCmec IV) and are susceptible to VAN based upon in vitro CLSI breakpoints (5,15,40). All study strains were routinely grown at 37°C in tryptic soy broth (TSB; Difco) or on tryptic soy agar (TSA) plates if not otherwise specified.
Isolation of SA169 from the PB 300-169 strain. The supernatant of stationary-phase 300-169 cells with/without MMC (1 g/ml; Sigma; as an agent to induce the liberation of temperate phages) (41) exposure and plasma from 300-169 strain-infected rabbits in an experimental infective endocarditis (IE) model (see detailed description under "Experimental IE model" below) were filtered (0.22 m; Millex; Millipore Corp.) to remove bacterial cells, diluted in phage buffer (42), and mixed with a recipient strain, RN4220 (a well-studied prophage-free reference S. aureus strain). The mixtures were plated on TSA plates using a well-established double-layer technique (42,43) and incubated at 37°C overnight or until plaques developed.
Transduction of SA169 into a genetic-background-matched RB strain, 301-188. To transduce SA169 into RB strain 301-188, PB 300-169 carrying pALC1766 (this strain is named 300-169p) was employed as the donor strain (13). The plasmid pALC1766 (chloramphenicol [CHL] resistance) was used as a marker during the transduction of SA169 into the recipient RB 301-188 strain for the selection of CHL-resistant colonies. Briefly, the filtered supernatant from the MMC-induced strain of 300-169p cells was mixed with the recipient strain 301-188. The mixture was then plated on TSA plates containing CHL (10 g/ml; Sigma). After overnight incubation at 37°C, transductant colonies were collected from the culture plates for further verification (44).
Determination of VAN MICs. VAN MICs were determined by a standard Etest method according to the manufacturer's recommended protocols (bioMérieux, La Balme-les-Grottes, France).
Growth curves and generation time. To evaluate the impact of SA169 on the growth rate, overnight cultures of the study MRSA strains were adjusted to a density of 1.0 McFarland standard in phosphate-buffered saline (PBS) and diluted 1:100 with 50 ml TSB in 500-ml Erlenmeyer flasks. Samples were then incubated at 37°C with shaking at 200 rpm for 24 h. Cell growth was monitored by measuring OD 600 (13). Generation time during the exponential phase was calculated based on the growth curve (38). NCTC8325-4, ␣-hemolysin negative, ␤-hemolysin positive 5 SH1000 rsbU-positive derivative of NCTC8325-4, agr-I 5 SH1001 SH1000 agr::tet(M) Tet r 5 pALC1766 Derivative of pALC1484 in which the cap5 promoter was cloned upstream of gfp UVR reporter, Chl r 13 a PB, persistent MRSA bacteremia strain; RB, resolving MRSA bacteremia strain. b CC, clonal complex.
ATP levels. ATP levels of the late-exponential cultures (6-h incubation, corresponding to the time point at which the PB 300-169 strain showed significantly higher global regulator activation than the parental RB 301-188 strain) (13) were measured using a BacTiter Glo kit (Promega) (45); results were given as ATP concentrations normalized to CFU.
In vitro VAN killing assay. A starting inoculum of ϳ10 8 CFU/ml of late-exponential MRSA cells (the same incubation time as the ATP assay above) was exposed to 15 g/ml VAN (to mimic targeted serum trough VAN concentrations for severe MRSA infections in humans) (5) in cation-adjusted Mueller-Hinton broth for 24 h. Survival rates were expressed as the number of surviving cells divided by the initial inoculum (5,13).
Delta-hemolytic activity. ␦-Hemolysin activity was determined by perpendicular streaking of test strains with the ␤-hemolysis-producing S. aureus reference strain, RN4220, on 5% sheep blood TSA plates after overnight incubation at 37°C (5). The ␦-hemolysis was denoted by an enhanced area of hemolysis at the intersection of RN4220 and test strain streaks (5). Strains SH1000 and SH1001 (agr mutant of SH1000) were used as positive and negative controls, respectively.
RNA isolation and target gene expression by RT-qPCR. Total RNA of the study MRSA cells from 2, 3, 4, 6, and 24 h of incubation (representing early-, mid-, late-, post-exponential, and stationary phases, respectively) was isolated by using the RNeasy kit (Qiagen) (13,46). DNase-treated RNA (1 g) was transcribed into cDNA. Real-time quantitative PCR (RT-qPCR) was performed using an ABI Prism 7000 instrument (Applied Biosystems) and a SYBR green PCR master kit (Applied Biosystems). Global regulators asp23 (a surrogate for sigB expression), sarA, agr RNAIII, and sae, as well as the purine biosynthesis pathway gene purF, were amplified using appropriate primers as described previously (13,35,36,46); gyrB was used to normalize the transcript quantification. Relative quantification was calculated by the threshold cycle (ΔΔC T ) method (35).
Determination of cap5 promoter activation by flow cytometry. Expression of the cap5 promoter denotes the onset of the stationary phase of growth (13,14). The temporal activation of the cap5 promoter in the 300-169p, 301-188p, and 301-188p::SA169 strains was detected as described previously (13). Briefly, overnight MRSA cultures were 1:100 diluted into TSB and incubated overnight at 37°C with shaking at 200 rpm for 6 h. Samples were obtained at 6 h of incubation to assess the profile of cap5 promoter activation by flow cytometry (35,47).
Experimental IE model. A well-characterized rabbit model of catheter-induced aortic valve IE was used to define the potential role of SA169 in persistent in vivo outcomes during VAN treatment (5,13). At 24 h after aortic catheterization, animals were infected intravenously with one of the following strains: 300-169p, 301-188p, or 301-188p::SA169. For infection, we used 10 5 CFU/animal, a 95% infective dose (ID 95 ) dose previously established for these strains (5). Twenty-four hours after infection, animals were randomly assigned to receive either no therapy (control group) or VAN at 15 mg/kg of body weight intravenously twice daily for 3 days. Control animals were sacrificed at 24 h postinfection (starting time point for VAN treatment). VAN-treated animals were euthanized 24 h after the last treatment to avoid VAN carryover effects. The cardiac vegetation, kidneys, and spleen were removed and quantitatively cultured (5,13). MRSA counts in the target tissues were calculated as the mean log 10 CFU/g of tissue (Ϯ standard deviation [SD]). Rabbits were cared for in accordance with the American Association for Accreditation of Laboratory Animal Care criteria. The Institutional Animal Care and Use Committee (IACUC) of the Lundquist Institute at Harbor-UCLA Medical Center approved all animal studies.
Statistical analysis. All in vitro experiments were performed in triplicate and repeated at least twice. The two-tailed Student t test was employed to analyze the in vitro data, as well as differences in tissue MRSA counts in the study groups. P values of Ͻ0.05 were considered statistically significant.
Data availability. All data supporting the findings of this study are available either within the article or in the supplemental material.