In Vitro generation of Panton-Valentine leukocidin (PVL) in clinical Methicillin-Resistant Staphylococcus aureus (MRSA) and its correlation with PVL variant, clonal complex, infection type

The amount of Panton-Valentine leukocidin (PVL) is diverse among Staphylococcus aureus isolates from different geographical regions, and its significance in some infections is disputed. However, data concerning this information in China are limited. Fifty-one lukSF-PV+ methicillin-resistant Staphylococcus aureus (MRSA) isolates gathered from varying infections were used for PVL production using enzyme-linked immunosorbent assay, and the quantity was analyzed in correlation with PVL isoform, genetic background of the isolate, and disease category. All isolates generated PVL with a range of 0.43–360.87 μg/mL, of which 56.9% isolates (29/51) generated 51–200 μg/mL of PVL; 11.8% (6/51) yielded PVL more than 200 μg/mL, and the rest (31.4%, 16/51) produced PVL of ≤50 μg/mL. The amount of PVL was not related to its variant and infection type, although isolates from skin and soft tissue infection had relatively high mean and median. Clonal complex (CC) 22 isolates might be the producer of relatively high concentrations of PVL; however, the difference among CCs was not analyzed due to a small number of CC isolates. The relevance of PVL production with the infection type, toxin isoform, and genetic characteristic of isolates may vary by clone type and also needs to be further evaluated using a large sample size and best concentration on in vivo environment.

(lukSF-PV + ) MRSA isolates and explored whether the quantitative generation of PVL is correlated with specific isoform. In addition, the study also further analyzed the correlation of PVL production with clonal complexes (CCs) of isolates and type of infection.
Among all CCs, the CC22 isolates produced the highest level of PLV (a mean of 116.88 ± 75.98 μg/mL and a median of 118.51 μg/mL), followed by CC59, CC1, CC5, and CC88 isolates ( Table 2). Although the PVL levels of isolates with CC88 and CC5 were far lower than those of CC22 and CC59 isolates, the statistical analysis was not performed to analyze the significance of the difference because of the small number of isolates of most CCs (Table 2). In addition, this study also analyzed the correlation of PVL yields of isolates with CA and HA infections, and PVL H and R isoforms. Unfortunately, no statistical differences were found among them (Table 2).

Discussion
Among lukSF-PV + MRSA, 66.7% (34/51) were isolated from young and pediatric patients (≤44 years old), which included 93.1% (27/29) CA MRSA. This was consistent with the reports that CA MRSA mostly affects young individuals and children, and usually presents PVL 18 .
Previous reports showed that the lukSF-PV genes were a frequent genetic marker of CA MRSA isolates with SCCmecIV or SCCmecV 19,20 . Nevertheless, some studies have reported the lukSF-PV + HA MRSA strains [21][22][23][24][25] . This study also found that 43.1% (22/51) lukSF-PV + MRSA were HA MRSA strains ( Table 1). The prevalence of lukSF-PV + MRSA clonal lineages is diverse in different parts of the earth. For example, ST8 (CC8)-SCCmecIV (USA300) clone predominates in USA; ST59 (CC59)-SCCmecV T clone circulates in Asia, ST30 (CC30)-SCCmecIV clone is prevalent in New Zealand, ST93 (CC93)-SCCmecIV clones are found mainly in Australia, ST80 (CC80)-SCCmecIV is the most frequent clone in Europe and the Middle East, ST88 (CC88)-SCCmecIV clone is identified mainly in Africa, and ST22 (CC22)-SCCmecIV and ST772 (CC1)-SCCmecV clones emerge mainly in India 18 . In the present study, ST59 (CC59)-SCCmecIII was the predominant clone (35.3%, 18/51) of lukSF-PV + MRSA isolates, followed by ST59 (CC59)-SCCmecIV (IVa/IVb, 25.5%, 13/51). However, for the former, 61.1% (11/18) of isolates were found to be CA MRSA; for the latter, 38.5% (5/13) of isolates were detected as HA MRSA. Historically, SCCmecIII was considered as one of the epidemiological markers of the main HA MRSA clones, and SCCmecIV was often found in CA MRSA isolates 24 . However, contrary to these views, SCCmecIII and SCCmecIV were found in both HA and CA lukSF-PV + MRSA isolates in this study. This indicated that the lukSF-PV + MRSA isolates carrying SCCmecIII or SCCmecIV could disseminate between the hospital and the community in the region studied. The PFGE types of some isolates with SCCmecIII supported the possibility of the transmission of HA strains of SCCmecIII into the community (Table 1). In addition, the reports by Song et al. 26 and Yao et al. 24 also described this spread of HA and CA MRSA isolates from Taiwan, Hong Kong, Korea, Vietnam, and Wenzhou city of mainland China. Geng et al. 27 reported the existence of ST338 (CC59) CA MRSA isolate in Taiwan and the southern region of China. In the present study, five lukSF-PV + MRSA isolates were also identified to be ST338 type, including three CA isolates and two HA isolates. The investigation results of Song et al. 26 showed that ST30 (CC30) was predominant in Asian areas. However, only one HA isolate was ST30 in this study. Additionally, ST22 (CC22) and ST88 (CC88), predominating in India 28,29 and Africa 3 , respectively, were also found in the isolates in this study. However, the ST22 isolates were SCCmecIII type, and not SCCmecIV detected in India. Although ST8 (CC8) and ST80 (CC80) prevailing in the USA and Europe, respectively, have been described in CA MRSA isolates from Japan and Korea (ST8), and Singapore and Malaysia (ST80), both MRSA clones are apparently scarce in Asian regions 26 and were not found in the isolates in this study (Table 1).
According to the reports of Dumitrescu et al. 12 , O'Hara et al. 8 , and Tong et al. 14 , the R isoform of PVL was found in CC8 (mainly USA300 strains), CC1, and CC93 isolates, which were mainly from the USA and Australia; while the H isoform was mainly observed in non-USA isolates belonging mainly to CC121, CC30, and CC22. A few isolates belonging to CC1, CC5, CC6, CC25, CC59, and CC88 also carried the H isoform. Table 1 H2  III  A1  ST59  CC59   PVL130  HA  H1  III  F2  ST59  CC59   PVL161  CA  H2  NT  K  ST188  CC1   PVL211  HA  H1  III  E1  ST22  CC22   PVL212  HA  H2  IVa  G1  ST59  CC59   PVL214  HA  H2  III  E1  ST22  CC22   PVL217  HA  H2  III  G3  ST59  CC59   PVL219  HA  H2  III  D2  ST338  CC59   PVL220  HA  H2  III  D3  ST338  CC59   PVL222  HA  H2  II  M  PVL can lead to concentration-dependent necrosis and apoptosis of human polymorphonuclear neutrophils and lysis of human monocytes and macrophages 12 , but whether PVL is pathogenic or a marker of specific infection is controversial. Contradictory data may be associated with the quantity of PVL generated by individual isolates. Up to now, the number of S. aureus isolates producing PVL is mainly determined using polymerase chain reaction (PCR) for lukSF-PV genes or quantitative reverse transcription-PCR for lukSF-PV mRNA levels. However, the findings of these methods may not relate to the actual yield of PVL. Therefore, the present study directly detected the PVL toxin production using enzyme-linked immunosorbent assay (ELISA). This study showed that all the lukSF-PV + isolates produced PVL in vitro, and the extent of the toxin expression varied among isolates (range from 0.43 to 360.87 μg/mL), even within specific genetic contexts (Table 2).
Undoubtedly, the presence of the genes encoding PVL does not necessarily imply that the toxin is produced in the setting of human infection. While this is true, it is important to note that neither does in vitro production. Additionally, the production of toxin is varied from culture media. Historical documents have reported the inter-isolate variability for ST8, ST80, ST93, and ST121 isolates 6,11,30 . The same phenomenon was reported for the STs detected in this study. The data showed that two ST59 CA isolates (PVL51 and PVL238) were the strongest producers of PVL (360.87 μg/mL and 339.52 μg/mL, respectively); and three ST59 (PVL8, PVL40, and PVL209) and one ST22 (PVL211) isolates produced 231.74-265.80 μg/mL of PVL (Table 1 and Fig. 1). However, two extremely low producers of PVL also belonged to ST59 and ST22 clones (PVL202 and PVL204, 0.94 μg/mL each) ( Table 1 and Fig. 1). The amounts of PVL yielded by other MRSA isolates (including the remaining ST59 and ST22 isolates) varied between 0.43 and 192.96 μg/mL. The cause of the difference in PVL expression is not clear. Also, the correlation of PVL yield with the genetic background of isolates warrants further investigation. The PVL concentration of 0.33 μg/mL evoked the apoptosis of human granulocytes, and the injection of 0.3 μg/ mL of toxin into the intracutaneous tissue of rabbits caused local inflammation and necrosis 11 . According to the in vitro PVL production in this study (≥0.43 μg/mL), all lukSF-PV + isolates likely had the ability to cause an inflammatory response during infections. Shallcross et al. 31 reported that PVL-producing clinical isolates had a link with SSTIs. In the present study, the mean value and median of PVL production of SSTI isolates were relatively high compared with those of isolates from pneumonia, surgical site infections, and other infections. However, the significant difference in the toxin levels was not found among the four groups (Table 2). This result was not consistent with the conclusion that PVL performed a more significant role in S. aureus SSTIs 31 . In addition, the isolates from invasive infections (such as pneumonia) were even shown to yield low mean and median of PVL ( Table 2). The possible explanation was that the clinical outcome was correlated with host factors, or with the existence of other toxins 30 .
For PVL variants, the H and R isoform isolates generated similar mean concentrations and medians of PVL, and the difference between the two groups was not significant (P = 0.975) ( Table 2). This result suggested that point mutations in the open reading frame of lukSF-PV genes might have no effect on the toxin production. A previous study showed that CA MRSA infection was responsible for rapidly progressive and lethal diseases such as necrotizing pneumonia, severe sepsis, and necrotizing fasciitis. Also, some data sustained the opinion that PVL was accountable at least partially for the enhanced virulence of CA MRSA 32 . In this study, the CA and HA isolates yielded similar levels of PVL (P = 0.754) ( Table 2). This indicated that the lukSF-PV + HA isolates might lead to the severity of clinical infections similar to that caused by lukSF-PV + CA isolates.
Being a retrospective study, this study had several limitations. First, the corresponding clinical samples could not be collected to evaluate the actual concentration of PVL in vivo. Because the PVL production depends on many factors, such as environmental factors, host factors, and internal factors of bacteria, it is not yet known which factor impacts mainly on the protein expression level. Second, the representativeness of isolates from only one hospital was insufficient for determining the generation and isoforms of PVL and the association between PVL production and PVL variants or genetic background or infection types. Third, more clinical information could not be collected to evaluate the correlation of the yield and isoform of PVL with the severity of illness.
In summary, the present experimental data did not sustain the direct correlation of PVL production in vitro with clinical types of infection and its isoform. The in vitro production of PVL by lukSF-PV + MRSA isolates was dramatically changeable, indicating the existence of unknown environmental and host factors in transcriptional and/or translational regulation of gene expression. The regulation mechanisms affecting the differential expression of PVL and the associations among PVL levels and characteristics of isolate and infection require further exploration.

Methods
Clinical isolates. Fifty-one consecutive, nonduplicate lukSF-PV + MRSA isolates from various clinical specimens of individual patients were obtained from Zhejiang Xiaoshan Hospital (a tertiary hospital), Hangzhou City, China. The isolates were collected between January 2010 and December 2011 and identified using VITEK 32 instrument (bioMérieux, Marcy l′ Etoile, France). Methicillin resistance was determined using the Kirby-Bauer method with cefoxitin disk (30 µg, Oxoid, Basingstoke, UK) and PCR for mecA 33,34 . All isolates had been previously described by PVL isoform detection and PFGE-SCCmec-MLST typing (including CC) ( Table 1) 35 . The strains ATCC 49775 and MRSA N315 were used as positive and negative controls for the PVL production, respectively.

Quantitative detection of PVL.
For quantifying the production of PVL in vitro, a certain amount of lukSF-PV + MRSA (1 to 3 × 10 6 CFU/mL) was cultured at 37 °C in 5 mL of casein-casein-yeast extract broth in which the optimal production of PVL could be acquired, with shaking at 200 rpm for 20 h. At this time, all growth isolates had reached a survival plateau of 1 to 4 × 10 9 CFU/mL 6,11 . Culture supernatants were acquired by centrifugation at 8000 g for 15 min. The quantification of secreted PVL was performed using ELISA, according to the kit instructions (Shanghai Y-J Biotechnology Co., Ltd., China). Each sample was measured in triplicate. The ELISA detection limit of PVL was 0.3~10 μg/mL. The samples with a concentration of more than 10 μg/mL were diluted for the redetection of PVL. A previous study reported that the intracutaneous injection of 0.3 μg/mL PVL led to local inflammation and necrosis in rabbits, which, like humans, are sensitive to similar PVL concentration 11 . Therefore, a concentration of 0.3 μg/mL was used in the present study, just the lower detection limit of the ELISA kit, as the cutoff value to determine whether the supernatant was PVL positive.
Definitions. (1) SSTI was described as the existence of a skin abscess or other symptoms of inflammation and a positive MRSA culture 36 . (2) Pneumonia was described as the existence of fever with imaging changes corresponding to pulmonary infection and MRSA-positive respiratory tract specimens (sputum, endotracheal aspirate, bronchial washing, and bronchoalveolar lavage) or blood cultures 36  Clinical information. The clinical data of the patients were collected retrospectively from the case notes. The data included gender, age, outpatient or inpatient, clinical symptoms, sample type, date of strain isolation, clinical diagnosis, and other medical histories related to HA or CA infections.

Statistical analysis.
In view of the uneven distribution of PVL production among isolates, the comparisons of two and ≥3 groups were analyzed using the Wilcoxon rank-sum test and the Kruskal-Wallis rank-sum test (SPSS 22.0), respectively. A two-sided P value < 0.05 was considered statistically significant.
Data availability. Most of the experimental data acquired/analyzed during this study have been included in this published version. Information on rest of the data can be obtained from the corresponding author on request.
Ethics approval and consent to participate. The ethics committee of Zhejiang Xiaoshan Hospital provided ethical approval for this study. The approval ID number was XSYY2015012. All study participants provided an informed consent before participation, and patient information was anonymized.