Typing of Panton-Valentine Leukocidin-Encoding Phages and lukSF-PV Gene Sequence Variation in Staphylococcus aureus from China

Panton-Valentine leukocidin (PVL, encoded by lukSF-PV genes), a bi-component and pore-forming toxin, is carried by different staphylococcal bacteriophages. The prevalence of PVL in Staphylococcus aureus has been reported around the globe. However, the data on PVL-encoding phage types, lukSF-PV gene variation and chromosomal phage insertion sites for PVL-positive S. aureus are limited, especially in China. In order to obtain a more complete understanding of the molecular epidemiology of PVL-positive S. aureus, an integrated and modified PCR-based scheme was applied to detect the PVL-encoding phage types. Phage insertion locus and the lukSF-PV variant were determined by PCR and sequencing. Meanwhile, the genetic background was characterized by staphylococcal cassette chromosome mec (SCCmec) typing, staphylococcal protein A (spa) gene polymorphisms typing, pulsed-field gel electrophoresis (PFGE) typing, accessory gene regulator (agr) locus typing and multilocus sequence typing (MLST). Seventy eight (78/1175, 6.6%) isolates possessed the lukSF-PV genes and 59.0% (46/78) of PVL-positive strains belonged to CC59 lineage. Eight known different PVL-encoding phage types were detected, and Φ7247PVL/ΦST5967PVL (n = 13) and ΦPVL (n = 12) were the most prevalent among them. While 25 (25/78, 32.1%) isolates, belonging to ST30, and ST59 clones, were unable to be typed by the modified PCR-based scheme. Single nucleotide polymorphisms (SNPs) were identified at five locations in the lukSF-PV genes, two of which were non-synonymous. Maximum-likelihood tree analysis of attachment sites sequences detected six SNP profiles for attR and eight for attL, respectively. In conclusion, the PVL-positive S. aureus mainly harbored Φ7247PVL/ΦST5967PVL and ΦPVL in the regions studied. lukSF-PV gene sequences, PVL-encoding phages, and phage insertion locus generally varied with lineages. Moreover, PVL-positive clones that have emerged worldwide likely carry distinct phages.


INTRODUCTION
Staphylococcus aureus causes a spectrum of diseases from minor skin and soft tissue infections (SSTIs) to life-threatening conditions due to its potential to produce many toxins and efficiency at overcoming antibiotics (David and Daum, 2010;Uhlemann et al., 2014). Panton-Valentine leukocidin (PVL), found in some strains of S. aureus, plays a key role in the leukocytolysis and tissue necrosis (Shallcross et al., 2013). The genes (lukSF-PV) encoding PVL are composed of two cotranscribed open reading frames (lukS-PV and lukF-PV), and located on lysogenized bacteriophages integrated into S. aureus chromosome (Boyle-Vavra and Daum, 2007;Shallcross et al., 2013).
To date, all known S. aureus phages belong to the order Caudovirales, which can be separated into three major families (Podoviridae, Siphoviridae, and Myoviridae) depending on the tail morphology (Xia and Wolz, 2014). At least 10 PVL phages have been described and sequenced, and all of them belong to the Siphoviridae family characterized by double-stranded DNA and a long non-contractile tail (Zhang et al., 2011;Xia and Wolz, 2014). As the mobile genetic elements, phages can be mobilized and transferred among S. aureus strains. Hence, investigating the typing of PVL-encoding phages among S. aureus may help to better understand the evolution of these pathogens. The PCR-based scheme targeting a small part of the phage genome is a cursory means of predicting phages types inexpensively. However, there is no unified pattern for this scheme. To identify as many PVL-encoding phages as possible, this article summarized a strategy by integrating and modifying the expanded PCR-based scheme described previously (Ma et al., 2008;Boakes et al., 2011;Chen et al., 2013;Sanchini et al., 2014). Subsequently, by the usage of this strategy, fifteen-reaction PCR assay was carried out to detect 10 of the PVL-encoding phages ( PVL, 108PVL, tp310-1, Sa2958, Sa2mw, SLT, Sa2USA, TCH60, 7247PVL/ ST5967PVL, and Sa119) in S. aureus from China.
At least 22 single-nucleotide polymorphisms (SNPs) have been identified in the lukSF-PV genes (Wolter et al., 2007;O'Hara et al., 2008;Boakes et al., 2011;Li et al., 2012;Chen et al., 2013;Sanchini et al., 2014). And several PVL protein isoforms carrying non-synonymous mutations have recently been revealed (O'Hara et al., 2008;Li et al., 2012), which may implicate functional significance. As we all know, phages lysogenize into the bacterial chromosome through the integrative pathway. Boakes et al. (2011) analyzed the sequence diversity at the insertion site for the different PVL-encoding phages and found two lineage-specific insertion sites within the S. aureus chromosome: Ins1 and Ins2. Of note, site-specific attachment sequences (attL and attR) are also conserved across lineages of PVL-carrying strains although some variations were found (Boakes et al., 2011;Chen et al., 2013).
Due to the absence of a uniform detection method, complete data on Chinese epidemiology of PVL-encoding phages are limited (Li et al., 2012;Hu et al., 2015). In addition, no details yet about chromosomal PVL-encoding phage insertion sites for S. aureus in China are reported.
In the present study, we aimed to obtain a more complete description of the molecular epidemiology of PVL-positive S. aureus from China by detecting PVL-encoding phage types, analyzing PVL variant alleles and the chromosomal phage insertion junctions, and determining the genetic background. Moreover, the relationships among them were also elucidated.

Bacterial Isolates
During January 2010 to May 2015, a total of 1175 consecutive, non-duplicate clinical S. aureus isolates were collected from seven hospitals in China, namely, Shanghai General Hospital, Shanghai sixth People's Hospital, Tongren Hospital, Ruijin hospital, Shanghai People's Hospital of Putuo District, Zhejiang Xiaoshan Hospital, and The Central Hospital of Lishui City, Zhejiang province (Figure 1). All the isolates, including 924 methicillin-resistant S. aureus (MRSA) strains and 251 methicillin-susceptible S. aureus (MSSA) strains, were identified by VITEK Systems (BioMérieux, Marcy l ′ Etoile, France), the susceptibility of cefoxitin (30 µg, Oxoid, Basingstoke, UK), and the presence of mecA and mecC genes (Bignardi et al., 1996;Ganesan et al., 2013;Clinical Laboratory Standards Institute, 2014). The presence of lukSF-PV genes was determined by PCR according to previously published method (Lina et al., 1999).

Single-Nucleotide Polymorphisms of the lukSF-PV Genes
All isolates were cultured on blood agar and incubated overnight at 37 • C. Genomic DNA was extracted by TIANamp Bacterial DNA Kit (TIANGEN BIOTECH Co., Ltd., Beijing, China). Three primer pairs were designed to amplify three fragments (654, 718, and 680 nucleotides in length, respectively) of lukSF-PV genes as described by Boakes et al. (2011). All the products of PCR were sequenced in both directions by Shanghai Sangon Biotech.
In case of the fragments amplified >3000 nucleotides, the PCRs were carried out using PrimeSTAR R Max DNA Polymerase (TaKaRa, Dalian, China) adhering to the protocols recommended by manufacturer.

Phage Insertion Locus Sequencing
PCR designed to detect the proximal and distal junctions of PVL-encoding phage insertion sites were performed by the method described previously (Boakes et al., 2011). Sequences were aligned using CLUSTA L X 2.0. Phylip3.6 was used to construct a maximum-likelihood phylogenetic tree with 500 bootstrap replicates for the junction region sequences.
positive amplification for attL sequences, and 71 (91.0%) for attR sequences. This indicated a relatively conserved PVLencoding phage integration site regardless of difference among genetic features of strains, PVL-encoding phage types, and lukSF-PV genes variants. Maximum-likelihood phylogenetic tree analysis of the attR and attL attachment sites showed four clusters (I, II, III, and VI) and two singletons (IV and V) for attR and seven clusters (I, II, III, IV, V, VI, and VIII) and one singleton (VII) for attL (Figure 3, Table 2).

DISCUSSION
PVL-positive S. aureus, strongly associated with SSTI and necrotizing pneumonia, has gained great attention in recent years (David and Daum, 2010;Shallcross et al., 2013). In China, the PVL positive rate ranged from 5.5 to 28.6% in HA-MRSA isolates (Yu et al., 2008;Fan et al., 2009;Li et al., 2012;Xiao et al., 2013;Hu et al., 2015). The major clone with lukSF-PV genes was ST59 in MRSA, while ST398, and ST88 in MSSA according to the previous studies (Yu et al., 2008;Fan et al., 2009;Li et al., 2012;Xiao et al., 2013;Hu et al., 2015). In the current study, we also revealed a high percentage of PVL-positive ST59 clones associated with MRSA and ST398 with MSSA strains. A previously published eight-reaction PCR assay was performed to identify five ( Sa2958, Sa2mw, PVL, 108PVL, and SLT) of the PVL-encoding phages in S. aureus (Ma et al., 2008). Subsequently, this scheme was improved to include the detection of Sa2USA (Boakes et al., 2011), TCH60, tp310-1, and7247PVL (Chen et al., 2013). Recently, Sanchini et al. (2014) divided PVL-encoding phages into three groups including the elongated-head group, icosahedral-head group I, and icosahedral-head group II, and incorporated Sa119 and ST5967PVL belonging to icosahedral-head group II into the scheme. However, this PCR-based scheme could not discriminate ST5967PVL from 7247PVL (Chen et al., 2013;Sanchini et al., 2014), because the genomes of both phages showed more than 99% identity (Zhang et al., 2011). Therefore, these two phages were marked as " 7247PVL/ 5967PVL" in this study.
Applying the PCR-based strategy, the distribution of PVLencoding phages was investigated all over the world. Two morphologically distinct phages ( PVL and Sa2958) were found to be predominant among Japanese PVL-positive MRSA (Ma et al., 2008), and Sa2USA was the most prevalent phage in Italy (Sanchini et al., 2014), while in United Kingdom, the most common phage types were Sa2USA and Sa2mw (Otter et al., 2010;Boakes et al., 2011). In this study, the results revealed that 7247PVL/ ST5967PVL, and PVL were the major PVL phage types in the isolates studied. Notably, all the untypeable isolates were positive for PCR-3 and belonged to ST30 and ST59 strains, indicating that both the clones may harbor novel unknown icosahedral-head group II phage type(s) in China. In line with the previous observations (Ma et al., 2008;Otter et al., 2010;Boakes et al., 2011;Chen et al., 2013;Sanchini et al., 2014), lineage-specificity of PVL-encoding phages were noted in this study. For instance, 92.3% of 7247PVL/ ST5967PVL-carrying isolates belonged to CC59 lineage, harboring cluster I attR, and attL, and 60% of Sa2USA-hosting isolates pertained to ST398. However, the PVL-carrying isolates showed more variability of genetic characterization, with CC22, CC25, CC59, CC88, and CC9 detected within this group. Sa119, recently identified in CC5 lineage by Sanchini et al. (2014), was detected in CC59 lineage. Sa2958, associated with MRSA belonging to CC30 (Ma et al., 2008), was related to CC22 and CC59 in the present study.
These findings are similar with those reported in China previously (Li et al., 2012;Hu et al., 2015). However, 108PVL, which was the predominant phage type in CA-MRSA from children in mainland China (Li et al., 2012), and SLT, which was prevalent in 74 health care-associated PVL-positive MRSA strains (Hu et al., 2015), are not common in the present study. The difference of regions across China and sources of isolates may account for the dissimilarity of the predominant PVL-encoding phages among these studies.
The correlation between the presence of PVL in S. aureus infections and the clinical outcome has been controversial because of some conflicting data from epidemiological study or in vivo animal models (Hermos et al., 2010;Otto, 2011). Corresponding protein models of two lukSF-PV sequence variants, named R and H variant, may indicate certain functional significance (Wolter et al., 2007;John and Lindsay, 2008;O'Hara et al., 2008). In our study, the H-and R-PVL isoforms could be identified in three morphological phage groups. Of note, the R1 isoform was mainly associated with Sa2USA, similar to the previous study (Chen et al., 2013). Of the five SNPs in the present study, two were non-synonymous, including a previously described arginine to histidine replacement at amino acid residue 176 (nucleotide 527) and a valine to isoleucine replacement at amino acid 340 (nucleotide 1022). The latter amino acid change presented in only one sequence (LS2074) and decreased the predicted molecular size of LukF from 36962Da to 36948Da, but did not influence the theoretical isoelectric point (9.1). The nucleotide sequence of the lukSF-PV genes has been deposited in GenBank under the Accession no. KX443594. Further work is required to confirm whether this non-synonymous replacement results in clinical significance.
Another interesting finding was that all the attR cluster Iharboring isolates possessed the same agr type (agr I) and were mostly observed within CC59 lineage. With the exception of one strain (strain 148), all of the attR sequences of cluster II strains were identified in CC22 lineage. The CC88 isolates were completely associated with attR cluster III and attL cluster IV. The main variations in attL were located in the phage binding sites (P1 and P2) and those in attR were located in the phage binding site P3 and chromosome binding sites B2. These indicated that the diversity of junction sequences was mainly due to differences in the phage DNA, which was different from the previous report that the main variations in attR were located in the 29-bp chromosome binding sites (B2) (Chen et al., 2013).
There were two limitations in this study. First, the identification of PVL-encoding phages was on the basis of PCR assays targeting only a part of the phage genome. Since phages tend to have recombination events, positivity by PCR should be supposed to be a family of PVL-encoding phages with variable genomic portions, instead of being taken as a specific phage (Sanchini et al., 2014;Xia and Wolz, 2014). Second, the proportion of phage-untypable PVL-positive isolates is a little high (32.1%), although the rate is within the range of 16.4-77.6% reported previously (Ma et al., 2008;Boakes et al., 2011;Li et al., 2012;Hu et al., 2015) by the PCR-based scheme. This could be because some unknown or newly published PVLencoding phages such as 7401PVL (Mariem et al., 2013) have not being included in the scheme. Therefore, unknown PVL-encoding phages remain to be discovered and new primer pairs are expected to be designed to further expand this scheme. Whatever, next-generation sequencing may provide us with a way to achieve perfect epidemiological picture of diversity of PVL-encoding phages, which seems to represent a trend in the future.

CONCLUSION
This study characterized PVL-encoding phages, the chromosomal phage insertion sites, the polymorphism of lukSF-PV genes, and the genetic background of PVL-positive S. aureus clinical isolates from China, and found the existence of some correlation among them. Our findings may contribute to the understanding of the epidemiology and evolution of PVL-positive S. aureus, and add the evidence that PVL-positive strains disseminating worldwide likely carry distinct PVL phages.

AUTHOR CONTRIBUTIONS
HZ, FH performed the experiments; QL designed and conceived the study; HZ, QL, CH analyzed the clinical data and wrote this manuscript; QL, SJ, XX, YZ, BD, and FG collected the clinical samples. All authors read and approved the final manuscript.