Commercial Biocides Induce Transfer of Prophage Φ13 from Human Strains of Staphylococcus aureus to Livestock CC398

Human strains of Staphylococcus aureus commonly carry the bacteriophage ΦSa3 that encodes immune evasion factors. Recently, this prophage has been found in livestock-associated, methicillin resistant S. aureus (MRSA) CC398 strains where it may promote human colonization. Here, we have addressed if exposure to biocidal products induces phage transfer, and find that during co-culture, Φ13 from strain 8325, belonging to ΦSa3 group, is induced and transferred from a human strain to LA-MRSA CC398 when exposed to sub-lethal concentrations of commercial biocides containing hydrogen peroxide. Integration of ΦSa3 in LA-MRSA CC398 occurs at multiple positions and the integration site influences the stability of the prophage. We did not observe integration in hlb encoding β-hemolysin that contains the preferred ΦSa3 attachment site in human strains, and we demonstrate that this is due to allelic variation in CC398 strains that disrupts the phage attachment site, but not the expression of β-hemolysin. Our results show that hydrogen peroxide present in biocidal products stimulate transfer of ΦSa3 from human to LA-MRSA CC398 strains and that in these strains prophage stability depends on the integration site. Knowledge of ΦSa3 transfer and stability between human and livestock strains may lead to new intervention measures directed at reducing human infection by LA-MRSA strains.


INTRODUCTION
Staphylococcus aureus is an opportunistic pathogen in human and animals, and is one of the leading causes of acute and chronic infections. When treating staphylococcal infections, resistance to βlactam antibiotics is an increasing problem with methicillin resistant S. aureus (MRSA) clones epidemically spreading in the hospitals as well as in the community (DeLeo et al., 2010;Dulon et al., 2011). In recent years, strains belonging to clonal complex 398 (CC398) have been widely detected in pigs and have also been found in other livestock animals such as turkey, chicken and cattle (Smith and Pearson, 2011). In general, livestock-associated MRSA (LA-MRSA) CC398 is regarded less virulent than human-associated MRSA clones due to the absence of several common virulence factors, such as enterotoxins and phage encoded Panton-Valentine leukocidin (PVL) (Schijffelen et al., 2010;Price et al., 2012;Becker et al., 2015). However, LA-MRSA CC398 causes infections in humans with livestock contact (Köck et al., 2011;Pérez-Moreno et al., 2016) and since first detected in the early 2000s (van Loo et al., 2007) the number of clinical cases in humans has increased steadily (van Cleef et al., 2011;Cuny et al., 2015). The infection of humans by LA-MRSA CC398 is primarily regarded as an occupational risk for farmers, veterinarians and meat handlers (Smith and Pearson, 2011) but a secondary risk is the spread of CC398 to the community (Smith and Pearson, 2011;Smith, 2015). In fact, infections of LA-MRSA CC398 in patients without animal contact have been reported both in Europe and the United States (Welinder-Olsson et al., 2008;Wulf et al., 2008;Bhat et al., 2009;Larsen et al., 2015Larsen et al., , 2016Diene et al., 2017).
The ability of S. aureus to colonize host organisms is in part determined by prophages. The majority of humanassociated S. aureus isolates contain β-hemolysin negativeconverting bacteriophages, which are classified as Sa3 (Goerke et al. 2006b). They share an integrase ) and an attachment core sequence (5 ′ -TGTATCCAAACTGG-3 ′ ) in the hlb gene (Coleman et al., 1991). Sa3 phages contain the immune evasion cluster (IEC) encoding the chemotaxin inhibitory protein (CHIPS), staphylococcal complement inhibitor (SCIN), staphylokinase (SAK), as well as some enterotoxins (Coleman et al., 1989;van Wamel et al., 2006;Goerke et al., 2009;McCarthy and Lindsay, 2010), which contribute to adaptation to the human host (de Haas et al., 2004;Rooijakkers et al., 2005;Jung et al., 2016). Recent data suggest that Sa3 phages are occasionally found in livestock-associated S. aureus strains, and the acquisition of Sa3 is important for the host-jump of S. aureus from animals to humans (Price et al., 2012;van der Mee-Marquet et al., 2014;Larsen et al., 2015Larsen et al., , 2016. Induction of prophages generally occurs in response to DNA damage elicited by for example reactive oxygen species and some antibiotics, and is mediated by RecA activation and cleavage of the phage repressor (Frye et al., 2005;Goerke et al., 2006a;Nanda et al., 2015). Thus, extrinsic factors contribute to the mobility of Sa3 phages among S. aureus strains. Disinfection and cleaning are critical steps when maintaining a desired hygiene status in household products, in hospitals and in food animal production. In livestock production, biocides are widely applied for cleaning and disinfection of animalassociated production areas, equipment, during transport and even directly on animals to prevent skin diseases (Quinn and Markey, 2001). Major biocides applied in livestock production include hydrogen peroxide, peracetic acid, glutaric aldehyde, quaternary ammonium compounds, and isopropanol (Kjølholt et al., 2001; SCHENIHR (Scientific Committee on Emerging and Newly Identified Health Risks), 2009). Hydrogen peroxide is a common disinfectant applied in animal production. It is a powerful oxidizing agent that oxidizes thiol groups in enzymes and proteins and leads to free radical production (Kjølholt et al., 2001;Russell, 2003). Here we have applied a prophage 13 from strain 8325 belonging to Sa3 group to examine if sub-lethal concentrations of commercial biocides and biocidal compounds induce phage Sa3 and stimulate its transmission to LA-MRSA CC398 strains.

Strain Collection and Chemical Reagents
Chemicals and biocides used in this study are listed in Table 1 and strains in Table 2. Media used in this study included Mueller-Hinton broth (MHB), tryptic soy broth (TSB) and agar (TSA) from Oxoid. Sheep blood is from Department of Veterinary Disease Biology, University of Copenhagen.

Strain Construction
Phage-cured MW2c was obtained by treating MW2-Sa3mw (Wirtz et al., 2009) with 1 µg/ml mitomycin C for 2 h at 37 • C, 200 rpm in TSB and subsequently with 0.5 mM hydrogen peroxide for 3 h at 37 • C, 200 rpm. Serial dilutions were plated on TSA with 0.5 mM hydrogen peroxide and incubated at 37 • C for overnight. Colonies grew on the plates were selected as phage cured bacteria (MW2c) and checked by PFGE (Goerke et al., 2004) for lacking of the phage.
To identify the 13 attB site in strain 8325-4 (Novick, 1967) and RN4220 (Peng et al., 1988), the original 13 attB site in 8325-4 was predicted by subtracting the 13 sequence (accession No.: NC_004617) from the NCTC 8325 genome harboring 13 (accession No.: NC_007795), and aligning the resulting sequence to the known hlb sequence containing the 13 attB site (Coleman et al., 1991) (accession No.: X61716) for confirmation. To construct a plasmid containing a mutated 13 attB (attBmut), 800 bp upstream sequence of 13 attB site in hlb gene of 8325-4 (Novick, 1967) was amplified by overlapping PCR primer pairs pIMAYhlbattfor and hlbattmutrev (Supplementary Table 1), and 800 bp sequence downstream of 13 attB site was amplified by overlapping PCR primer pairs hlbattmutfor and pIMAYhlbattrev, where the 13 attB mutation from strain 61599 was included in primer hlbattmutfor (Supplementary Table 1). The PCR fragments and pIMAY vector were digested by EcoRI, and ligated together by Gibson Assembly R Master Mixt (BioLabs R ) to form the plasmid named as pIMAY_ 13attmut. This plasmid was electroporated into RN4220 (Kreiswirth et al., 1983) resulting in the strain RN4220pIMAY_ 13attmut, and the 13 attB mutation was introduced into the RN4220 chromosome by homologous recombination (Monk et al., 2012). The mutation at attB in RN4220 was confirmed by sequencing and the strain was named RN4220 13attBmut.

Determination of Minimal Inhibitory Concentrations (MICs)
MICs of biocides and chemical compounds in strain 8325-4 13 were determined according to the guideline of Clinical and laboratory standards institute (Clinical and Laboratory Standards Institute (CLSI), 2008). Strains from overnight TSA plate was resuspended in 0.9% NaCl to achieve the turbidity of 0.5 McFarland standard and further diluted 100-fold in MHB. The working solution of biocides and chemical compound were benzalkonium chloride (2.67 µg/ml), hydrogen peroxide (3% w/w), Biocide 1 (5% v/v), Biocide 2 (5% v/v), Biocide 3 (5% v/v), and mitomycin C (1 µg/ml). In brief, the working solution of biocides and mitomycin C were prepared with two-fold dilution series in MHB in 96-well microtiter plates with 100 µl volume. Further, 100 µl of cell suspension was added to each well. Positive growth control of wells without biocide, and negative controls of wells with only MHB were included. The microtiter plates were incubated at 37 • C for 24 h. MIC values were determined as the lowest concentration of the compounds that eliminated the visible growth of bacteria.

Induction Assay by Different Biocides and Mitomycin C
To perform the phage induction assay, strain 8325-4 13 was grown to the exponential phase (OD 600 = 0.8) (Wirtz et al., 2009) by shaking at 37 • C, 200 rpm in TSB. Then, different concentration of Biocide 1, Biocide 2, Biocide 3, benzalkonium chloride, hydrogen peroxide and mitomycin C were added into the broth culture respectively, and further incubated for 2 h at 37 • C, 200 rpm. The concentration series of each biocide was determined according to MIC values of strain 8325-4 13 ( Table 3), which included MIC, 5X MIC, 10X MIC, 20X MIC, and 30X MIC. Supernatants were sterilized by 0.45 µm pore diameter membrane filter. The phage induction levels were evaluated by PFU determination as previously described (Goerke et al., 2006b). Briefly, 100 µl of each phage supernatant dilution was mixed 100 µl indicator strain MW2c (OD 600 = 0.1) and incubated for 10 min at room temperature before mixing with top agar and pouring onto a TSA plated with 10 µM CaCl 2 , and further incubated overnight at 37 • C.

Lysogenization Assay of 13 to LA-MRSA
Strain 8325-4 13-kana was used as the phage donor strain. Phage lysate were obtained by growing the strain to exponential phase (OD 600 = 0.8) at 37 • C, 200 rpm, mixing with 1 µg/ml mitomycin C and further incubating for 4 h under the same condition. The culture was centrifuged at 4 • C, 8,500 rpm for 6 min, and sterile filtered by 0.45 µm pore diameter membrane filter before determining the titer. 13 lysogens were obtained by mixing phage 13-kana and LA-MRSA CC398 strains (61599 and 93616) at MOIs of 0.001, 0.01, 0.1, 1, 10, 100, and 1,000 and incubating the mixture at 30 • C for 30 min. Hundred microliter of the mixture was then spread on TSA plate containing 100

Chemical agent MIC
Mitomycin C (mg/L) 0.125 Hydrogen peroxide (w/w) 0.03% Benzalkonium chloride (mg/L) 2.67 µl/ml kanamycin, 10 µg/ml tetracycline and 5% of sheep blood (Kan-Tet plate), and incubated at 37 • C overnight. Lysogens were selected as colonies able to grow on Kan-Tet plate. The plates were stored at 4 • C overnight to detect β-hemolysin activity. Lysogens were verified by colony morphology and hemolysin activity (Supplementary Figure 6) as well as by PCR to detect ahpA3 and sak genes. The lysogenization frequency was evaluated as the ratio of the 13 CC398 lysogen colony count (CFU/ml) on Kan-Tet plate ( 13 CC398 lysogen) to the total recipient colony count (CFU/ml) on TSA plate with 10 µg/ml tetracycline (Tet plate).

PCR Analysis
Bacterial DNA was released by suspending 2-3 colonies in 100 µl Milli Q water and incubated at 95 • C for 10 min. For confirming the transfer of 13 to LA-MRSA CC398 strain, primer pair sakfor/sak-rev for the sak gene and primer pair kanR-for/kanR-rev for the aph gene were used. For checking if the hlb gene was intact, primer pair hlbPhi13attB-for/hlbPhi13attB-rev of hlb gene was used. For primer sequences see Supplementary Table 1. PCR amplification was performed by mixing 12.5 µl DreamTaq Green PCR Master Mix (2X) (Thermo Fisher Scientific), 10.5 µl Milli Q water, 1 µl DNA template and 1 µl (0.1 µM) each primer. For determining the attB core sequence mutation in strain 8325-4 and RN4220, the PCR products of hlb gene were purified by GeneJet PCR purfication kit (Thermo Fisher Scientific) and sequenced by Macrogen Inc.

In Vitro Liquid Co-Cultivation Assay
The liquid co-cultivation assay was carried out in TSB with donor strain 8325-4 13-kana resistant to kanamycin and recipient strain LA-MRSA 61599 resistant to tetracycline. In brief, both donor and recipient strains were grown to exponential phase (approximately 10 8 cfu/ml), and mixed with a ratio of 1:1. Different concentrations of hydrogen peroxide, Biocide 1 and mitomycin C were added to the culture, and then incubated at 37 • C, 200 rpm for 4 h. Phage transfer was detected by plating serial dilutions on Kan-Tet plate and Tet plate followed by incubating at 37 • C for 18 h. The 13 CC398 lysogens were counted based on the colony morphology of 13 lysogens (Supplementary Figure 6) and also further confirmed the PCR program of sak and kanR genes. To confirm the lysogens were belonging to CC398 strain and avoid miscounting of donor strain 8325-4 13-kana, a spa-mecA multiplex PCR (Tang et al., 2017) was conducted to CC398 lysogens, which could differentiate CC398 lysogens to donor strain 8325-4 13-kana (Supplementary Figure 5). The transfer ratio of 13-kana was considered as the ratio of the 13 CC398 lysogen colony count on Kan-Tet plates (CFU/ml) to the recipient colony count on Tet plates (CFU/ml).

INFLUENCE OF CHEMICAL AGENTS TO STRAINS AND PHAGE
The influence of applied chemical regents to donor and recipient strains was separately evaluated by growth curve, which culture of both strains at exponential phase (approximately 10 8 cfu/ml) was treated with a series of concentrations of mitomycin C, hydrogen peroxide and Biocide1, and measured OD 600 value every hour till 4 h at 37 • C, 200 rpm. The influence of applied chemical to 13-kana was evaluated by treating the phage stock with a series of concentration of chemical agents and the plaque assay was conducted after 4 h at 37 • C, 200 rpm, with the plaque assay mentioned above. The influence of applied chemical regents to the induction of 13 with the kanamycin resistant cassette was evaluated after 4 h of treatment by 13 induction assay mentioned above. In addition, the influence of chemical compounds present in Biocide 1 to the donor and recipient strains was evaluated individually by growth curve and 13 induction assays as described above in the presence of a series concentrations of peracetic acid, amine oxide and acetic acid.

Characterization of LA-MRSA CC398 13 Lysogens
Stability of the prophages in CC398 13 lysogens was examined in 10 13 lysogens of strain 61599 and passaging them for 20 days. In brief, 5 µl overnight culture of each lysogen was diluted 1000-fold in 5 ml TSB and incubated for 24 h at 37 • C, 200 rpm. After each passage, 100 µl culture was spread on TSA plate containing 10 µg/ml tetracycline and incubated overnight at 37 • C. From this plate, 50 colonies were picked up and streaked on TSA plate containing both 100 µg/ml kanamycin and 10 µg/ml tetracycline and TSA plate containing only 10 µg/ml tetracycline, and then incubated overnight at 37 • C. The stability of each lysogen was determined by the ratio of colonies survived on TSA plate contain both kanamycin and tetracycline to the colonies survived on TSA plate containing only tetracycline. Further, lysogens with different 13 stability characteristics were sequenced by paired-end sequencing (2 × 251 bp) using Nextera XT DNA Library Preparation Kit (Illumina Inc.) on a MiSeq sequencer (Illumina Inc.). Contigs were de novo assembled using CLC-bio assembler (Qiagen).

Lysogenization Assay of 13 attB Mutant
Stains 8325-4 13attBmut and RN4220 13attBmut, and their wild type strains were individually mixed with the 13-kana phage stock at a MOI of 0.1 and were incubated at 37 • C for 4 hrs. After making serial dilutions, 100 µl of each dilution series was spread on TSA plates with 5% sheep blood containing 100 µl/ml kanamycin (Kan-plate) and TSA plate with 5% sheep blood (TSA plate), which were incubated at 37 • C overnight. Lysogens were selected as colonies able to grow on Kan plate. The plates were stored at 4 • C overnight to detect β-hemolysin activity. The lysogen frequency was determined as the ratio of colony count (CFU/ml) on Kan-plate to the total colony count (CFU/ml) on TSA plate. Further, PCR amplification of hlb gene to check if hlb gene was interrupted, PCR amplification of hlb gene including the 13 attB site was conducted for 16 colonies of lysogenized mutants.

Induction of Sa3 by Biocides
Staphylococcus aureus strain 8325 harbors a Sa3 class of phage termed 13 (Iandolo et al., 2002). In this study, we applied strain 8325-4 13 (Goerke et al., 2006a), which is a derivative of the phage cured strain 8325-4 (Novick, 1967) lysogenized with 13 to evaluate if commercially available biocides can induce Sa3. We evaluated the susceptibility of strain 8325-4 13 to three commercially available biocides in addition to hydrogen peroxide, benzalkonium chloride as well as mitomycin C (Table 1) by MIC (Table 3). Further, we exposed the strain 8325-4 13 to these biocides and chemical reagents ( Table 1) and monitored plaque formation on indicator strain MW2c. Strain MW2c is a derivative of strain MW2-Sa3mw in which all phages have been cured (Wirtz et al., 2009). We found that excision of 13 was induced by mitomycin C (≤3.75 mg/ml), by biocides containing hydrogen peroxide [Biocide 1 (≤0.2% v/v) and 2 (≤ 0.2% v/v)] as well as by hydrogen peroxide (≤0.9% w/w) in a dose-dependent manner (Figures 1A-E). No induction of 13 was observed by benzalkonium chloride or Biocide 3 containing sodium hypochlorite as the active ingredient (Figures 1C,F) or in the absence of stimuli indicating that less than 10 pfu/ml is released.

Transfer of 13 to LA-MRSA CC398
To monitor integration of 13 into LA-MRSA CC398, a 13 phage stock was obtained by mitomycin C induction from strain 8325-4 13-kana, a derivative of 8325-4 13 where the 3'-end of chps and scn in 13 has been replaced by the kanamycin resistance gene, aphA3. This phage stock was used to infect two tetracycline-resistant LA-MRSA CC398 strains 61599 and 93616 isolated from humans in Denmark ( Table 2). Strain 61599 with spa type t034 (Larsen et al., 2015) was obtained from a pig farm worker and strain 93616 with spa type t899 (Larsen et al., 2016) was recovered from a mink farmer. Neither of the strains contained Sa3 phage by sequence analysis (Larsen et al., 2015(Larsen et al., , 2016. After repeated attempts, only strain 61599 could be lysogenised with 13. At multiplicity of infection (MOI) of less than one, we observed colonies resistant to both kanamycin and tetracycline, indicating lysogenization of LA-MRSA CC398 by 13-kana (Supplementary Figure 1). To confirm that the resulting colonies were lysogens and not spontaneous antibiotic resistant mutants, we used PCR amplification of the aphA3 and sak genes of 13-kana (data not shown).

Frontiers in Microbiology | www.frontiersin.org
To examine if 13-kana could transfer directly from S. aureus 8325-4 13-kana (donor strain) to recipient LA-MRSA CC398 strain 61599 in the presence of biocides or mitomycin, we first assessed a series of concentrations of mitomycin C, hydrogen peroxide and Biocide 1 ranging from sub-lethal to lethal with respect to their influence on growth (Supplementary Figure 2) and the induction levels of 13-kana (Supplementary Figure  3). We also confirmed that these concentrations have minimum effect on 13-kana in plaque assays (Supplementary Figure  4). Further, we evaluated MIC of strain 8325-4 13 to acetic acid, amine oxide and peracetic acid (Table 1), which are present in Biocide 1 (Table 3), and exposed the strain to these chemicals to evaluate their effect on growth and 13 induction (Supplementary Figure 5). We found that, in addition to hydrogen peroxide (Supplementary Figure 3B), 13-kana can also be induced by peracetic acid, which is able to spontaneously decompose to hydrogen peroxide and acetic acid (Yuan et al., 1997;Supplementary Figure 6).
We co-cultured the donor strain 8325-4 13-kana with the recipient strain 61599, in the presence of sub-lethal to lethal concentrations of mitomycin C, hydrogen peroxide, and Biocide 1. Colonies of strain 61599 lysogenized with 13kana were detected by plating on blood plates containing kanamycin and tetracycline. The transfer of 13-kana was confirmed by PCR for sak and aphA3 and the identity of the resulting colonies was confirmed by PCR amplification of spa and mecA (Supplementary Figure 7). We observed that 13-kana transferred from strain 8325-4 13-kana to 61599 with a transfer frequency between 10 −5 and 10 −6 in the presence of 0.002-0.1%(v/v) Biocide 1, which contains hydrogen peroxide as the active compound ( Figure 2C). Exposure to mitomycin C and hydrogen peroxide also induced the transfer of 13-kana (Figures 2A,B). Surprisingly, in the absence of DNA damaging agents, we also observed a transfer frequency of 13-kana was around 2.3 × 10 −7 (Figure 2). These results show that sublethal to lethal concentrations of biocides containing hydrogen peroxide and mitomycin C promote transfer of 13 between human and livestock-adapted strains in vitro.

Characterization of 13 in CC398 Lysogens
Initially, we assessed hemolysin production of nine 13 lysogens (LY01-09) in LA-MRSA CC398 strain 61599 on blood agar plates and observed that all were β-hemolysin positive (Supplementary Figure 8) suggesting that the phage had not integrated in the hlb attachment site commonly preferred in human strains (Goerke et al., 2006b). Subsequently, we evaluated stability of the lysogens over a period of 20 days (corresponding to approximately 960 generations) by plating on tetracycline agar plates and recording the fraction of kanamycin-resistant colonies (Figure 3). Here we observed different stability patterns with 13-kana being completely stabile in six lysogens (LY01-06), and being partially stable in the rest of the lysogens with 44% (LY08) and 29% (LY09) of cells retaining the phage, respectively. In LY07, the percentage FIGURE 3 | 13 stability in LA-MRSA CC398 strain 61599. Nine lysogens were grown over a 20 period with 1000 fold dilution performed daily from the previous day overnight culture. The stability of the phage was monitored by plating on agar plates with tetracycline, and scoring the number that were also kanamycin resistant. (a) filled triangle represents LY01-LY06 of lysogens, which were fully stable with 13 prophages during the 20-day test; (b) opened square represents LY07, which had sporadically lost 13 during 20 days; (c) open circle represent LY08 and (d) filled inverted triangle represents LY09, both lost 13 in the majority of the analyzed colonies after 20 days, respectively.
FIGURE 2 | Transfer of 13 during co-cultivation of LA-MRSA CC398 with 8325-4 13-kana. Strains LA-MRSA CC398 and 8325-4 13-kana were co-cultivated and exposed for 4 h to different concentrations of (A) mitomycin C, (B) hydrogen peroxide and (C) Biocide 1. Open squares represent the 13 transfer ratio calculated as the number of colonies resistant to tetracycline and kanamycin relative to the number resistant to tetracycline. Error bars represent ±SD, n = 3. ***P < 0.001, **P < 0.01 and *P < 0.05, NS, none significant by t-test analysis. of cells carrying the phage varied between 96 and 100% during the experiment indicating sporadic loss. To examine the genetic basis for the different stability patterns we sequenced the genomes of lysogens and observed eight different 13 integration sites in strain 61599 (Table 4). In the six lysogens (LY01-LY06) where 13-kana was completely stable, the phage was integrated at six different locations in the chromosome of 61599. In five of the six lysogens (LY01-05), 13-kana was inserted in annotated genes, whereas in LY06 the phage was integrated in an unrelated intergenic region (Table 4). In LY07, where 13-kana stability varied the phage was integrated in a gene encoding a hypothetical protein located on plasmid JQ861959 that had integrated in the Letter L denotes the position of attL, and letter R denotes the position of attR. **Letters with underline represent the nucleotides in attB that correspond to the core sequence of attB in hlb described by Coleman et al. (1991).
Frontiers in Microbiology | www.frontiersin.org chromosome, which may explain why recovery of the phage varied from plating to plating. In both LY08 and LY09 from which the phage was lost in the majority of cells after the 20 day period, 13-kana was integrated in yozB encoding a putative membrane protein, the function of which is unclear. These results strongly indicate that 13 integrates at alternative sites in LA-MRSA CC398 and that the overall stability of the integrated phage is influenced by its integration site in the chromosome. By sequence analysis, we found that all integration sites in the nine lysogens were partially similar to the attB sequence for 13 in hlb gene originally described by Coleman et al. (1991) (Table 4). However, we were not able to observe other similarities between the integration sites or regions. By further comparing these integration sites in lysogens to the corresponding sequences in strain 61599 and strain 8325-4, we found that only the integration site in LY03 contained one nucleotide variation between the strain 61599 and strain 8325-4, but this variation did not enhance resemblance to the attB in the hlb gene (Supplementary Table 2). In addition, sequence analysis of the hlb gene of strain 61599 revealed variation at two residues in the 14 nucleotide attB sequence when compared to attB in hlb gene of the human derived strain S. aureus COL (Projan et al., 1989). Both nucleotide substitutions are silent leaving the βhemolysin expression intact (Figure 4). This finding suggests that point mutations in hlb of LA-MRSA may drive 13 to integrate elsewhere in the chromosome. To test this hypothesis, we compared the attB core sequence of Sa3 from strain 61599 to the 69 of genome sequenced CC398 isolates from a previous study by Price et al. (2012). We observed that 65 out of 69 CC398 isolates from Price et al.'s study showed the same substitutions of the attB core sequence for 13 in the hlb gene as we found in strain 61599, while four of CC398 isolates had three substitutions compared to the attB core sequence in hlb described by Coleman et al. (1991; Figure 5).
To support our hypothesis that the mutation at attB in hlb of CC398 isolates influences integration of Sa3, we mutated the 13 attB site of hlb both in strain 8325-4 deviated from NCTC 8325 cured of 11, 12, and 13 (Novick, 1967), and in strain RN4220 derived from 8325-4 (Peng et al., 1988). After infecting both attB mutated strains with 13-kana at MOI of 0.1, we observed a significant decrease in lysogenization of 13-kana in both attB mutant strains, when compared to the respective wild type control strains (Figure 6) and confirmed by PCR amplification of hlb (Supplementary Figure 9) and aph3 (Supplementary Figure 10). These results demonstrate that point mutations of attB core sequence in hlb of CC398 isolates strongly influence the integration of Sa3 into the hlb gene, and the mutation favors Sa3 to integrate elsewhere in the bacterial genome and keeps the hlb intact.

DISCUSSION
Mitomycin C and hydrogen peroxide are known to induce prophages in bacteria by causing DNA damage, which activates FIGURE 5 | Variation of attB core sequence for 13 in hlb gene in Sa3-negative CC398 isolates compared to the attB in hlb gene of S. aureus COL (accession no. X13404) described by Coleman et al. (1991). CC398 isolate information was obtained from the study of Price et al. (2012). The gray shaded letters represent the variations in the attB of CC398 isolates.
FIGURE 6 | Lysogenization of 8325-4 13attBmut and RN4220 13attBmut by 13. 13-kana were mixed with 8325-4 13attBmut, RN4220 13attBmut, and the corresponding wild type strains with MOI of 0.1 and lysogen frequency was monitored as ratio of CFU on TSA plates with 100 µg/ml kanamycin and 5% of sheep blood ( 13-kana LA-MRSA CC398 lysogen) relative to the total CFU count on TSA plates with 5% of sheep blood. Horizontal lines represent mean value, error bar represents ±SD, n = 3. ***P < 0.001 by t-test analysis. the SOS response and in turn cleaves the phage repressor (Frye et al., 2005;Goerke et al., 2006a;Loś et al., 2010;Nanda et al., 2015). Here, we show that commercial biocides containing hydrogen peroxide can induce phage 13, a phage belonging to Sa3 phage group that is known to encode human immune evasion genes and likely to be important for human colonization (Iandolo et al., 2002). Additionally, when co-culturing the human originating strain, 8325-4 13-kana with the livestock MRSA CC398 strain 61599 (spa type t034), the transfer frequency of 13 to strain 61599 was significantly increased in the presence of sublethal concentrations of hydrogen peroxide or commercial biocides containing hydrogen peroxide when compared to the transfer frequency without any treatment. Integration of 13 was also attempted in another LA-MRSA CC398 strain namely 93616 (spa type t899), but in this strain, we did not obtain lysogens. The inability of 13 to lysogenize strain 93616 could be caused by strain variations such as the restriction modification systems that restricts DNA transfer between staphylococcal lineages (Sadykov, 2016), or differences in wall teichoic acid glycopolymers that are the receptors of 13 and are highly strain-specific (Xia et al., 2010).
Hydrogen peroxide is a commonly applied bactericidal compound for disinfection in livestock production (Kjølholt et al., 2001). Our study indicates that this compound can act as an extrinsic factor contributing to the induction of Sa3 by triggering excision and propagation of the prophage, and subsequent transfer from human to livestock-associated S. aureus strains. In contrast, neither sodium hypochlorite that is known to affect the bacterial cell wall (Maillard, 2002), nor benzalkonium chloride that is a membrane-active agent (SCHENIHR (Scientific Committee on Emerging and Newly Identified Health Risks), 2009), were able to induce 13. In the absence of external stimuli, no spontaneous induction of 13 was observed in concordance with a previous study (Goerke et al., 2006a). Interestingly, we did observe transfer of 13 from strain 8325-4 13-kana to strain 61599 when the two strains were co-cultured. The most likely reason for this is that 13 propagates on strain 61599 resulting in more phages that in turn increase the chance of phage integration in strain 61599.
Sa3 group is known as hlb-converting phages that integrate in the hlb-gene at the attB attachment site (5 ′ -TGTATCC AAACTGG-3 ′ ) recognized by the phage integrase (Coleman et al., 1991). A likely explanation for the atypical integration of Sa3 in strain 61599 is the two silent point mutations in the attB core sequence located within hlb. We speculate that these mutations in attB may be the reason that we did not observe phage integration in hlb but rather at numerous other locations in the LA-MRSA CC398 strain 61599. These integration sites share homology to the attB core sequence for 13 (Coleman et al., 1991). In particularly seven of the eight integration site sequences contain a four nucleotides sequence (5 ′ -CTGG-3 ′ ) at the 3 ′ -end, which is present in the 13 attB core sequence in hlb gene (5 ′ -TGTATCCAAACTGG-3 ′ ) (Coleman et al., 1991), but is not in the attB core sequence in hlb from CC398 strains (5 ′ -TGTATCCGAATTGG-3 ′ and 5 ′ -TATATCCGAATTGG-3 ′ ). We speculate that this sequence similarity is important for the integration of Sa3 into CC398. It was previously reported that 13 can be integrated at different locations in S. aureus (Goerke et al., 2006b;Kraushaar et al., 2017), but the reason for the atypical integration of Sa3 had not been studied before. Here, we exchanged the core sequence of 13 attB in hlb from both strain 8325-4 and strain RN4220 to the attB core sequence from CC398 strain 61599. After infecting with 13, we observed a significant decrease in lysogenization frequency in the 13 attB mutants compared to cells carrying the intact hlb gene. This result demonstrates that the mutations at the attB core sequence in hlb gene of CC398 strain cause the low transfer frequency of 13 to CC398 strains and also drives the atypical integration of 13 in CC398 strains.
In conclusion, we show that in the presence of commercially available biocides containing hydrogen peroxide, Sa3 is transferred from a human-associated S. aureus donor strain to a strain, belonging to the CC398 complex. Previous studies have shown that the CC398 ancestor was adapted to humans but jumped to animals by loss of prophage Sa3 during the close human-livestock activities (Price et al., 2012). Recently, it was shown that Sa3 may be re-introduced into livestock-associated CC398 strains in a single horizontal gene transfer event from human-associated S. aureus and that it can be maintained stably as a prophage in the CC398 lysogens (Larsen et al., 2016). Our study highlights the importance of environmental factors in transfer of Sa3 between S. aureus strains and that just a four nucleotides sequence may be enough to guide integration of the phage LA-MRSA strains. Future studies will be needed to determine if and how integration frequencies vary from strain to strain and to identify factors and conditions that may prevent Sa3 transmission.

AUTHOR CONTRIBUTIONS
YT, JH, LN, CWo, and HI designed the study, YT and AH conducted the microbiological analysis, CWi and YT constructed strains, YT, JL, JH and HI analyzed the results. YT wrote the manuscript and YT, LN, AH, JH, CWi, PA, JL, CWo, and HI reviewed the manuscript.