Background

Staphylococcus aureus (S. aureus) continues to be a leading cause of both community-and healthcare-associated infections, including skin and soft tissue infections, bacteremia, pneumonia, osteomyelitis, and endocarditis. Virulence factors play a major role in the pathogenesis of S. aureus, such as Panton-Valentine leukocidin (PVL). PVL is a pore-forming exotoxin composed of LukS-PV and LukF-PV. The PVL-producing strains have been associated with the onset of skin and soft tissue infections (SSTIs) and can also cause severe invasive infections (necrotizing pneumonia, etc) [1]. In addition, S. aureus can form biofilms not only in biological samples and surfaces of medical devices but also in tissues [2]. Biofilm can protect S. aureus from the damage of antibiotics and the host immune system [3]. Subsequently, the successful eradication of S. aureus infections is difficult, rendering biofilm as a vital factor in chronic infections.

The accessory gene regulator (agr)-mediated quorum sensing plays a major role in staphylococcal pathogenesis, which can downregulate the expression of cell surface-associated proteins (microbial surface components recognizing adhesive matrix molecules [MSCRAMMs]) and upregulate the expression of extracellular toxin (hemolysins, enterotoxins, extracellular proteases, etc.) [4]. The regulation of virulence factors by agr is important for disease progression. Several studies demonstrated that genetically engineered agr-knockout strains had attenuated virulence in animal models of skin and soft tissue infections, pneumonia, infective endocarditis, arthritis, and osteomyelitis [5,6,7,8,9,10].

However, recent evidence indicated that agr dysfunction exists extensively in healthcare settings (13–82%) [11,12,13,14], which might be attributed to the agr dysfunction that confers a potential advantage for S. aureus in the current medical environment [15]. For example, agr dysfunction has been linked to attenuated vancomycin activity; both laboratory-derived vancomycin-intermediate S. aureus (VISA) and clinical VISA isolates developed during vancomycin therapy often exhibit as agr dysfunction [16]. In addition, infections, such as bacteremia and bone and joint infection, caused by strains with dysfunctional agr often manifest as a chronic course and result in adverse outcomes [17,18,19]. Thus, understanding the prevalence of agr dysfunction in a specific area is essential.

Herein, we conducted a molecular epidemiology study in Beijing Children’s Hospital in China. The primary objectives of the present study were as follows: (1) to detect the prevalence of agr dysfunction in MRSA and MSSA isolates; (2) to explore the molecular characteristics, pvl carriage rate, biofilm formation ability, and antibiotic susceptibility of S. aureus isolated from pediatric patients; these indicators were compared mainly based on agr functionality.

Methods

Bacterial isolates

This study was approved by the Ethics Committee of Beijing Children’s Hospital affiliated to the Capital Medical University (No. 2016–93, 23/06/2016), and obtained clearance from the Institutional Biosafety Committee (IBSC) ([2017] No.43). S. aureus strains were collected and identified as follows. If the clinical samples (blood, pleural effusion, and joint effusion, etc.) were obtained from steril specimens (blood, pleural effusion, bone marrow, cerebrospinal fluid, joint effusion, seroperitoneum, etc), the bacterial growth in the culture medium could be directly identified by VITEK® MS system (BioMérieux, France). If the clinical samples were obtained from the non-sterile specimens (respiratory tract, skin, etc.), several suspected colonies were selected according to the morphological characteristics and identified by VITEK® MS system. The coagulase test and detection of nuc gene were employed to identify S. aureus as described previously by Kateete et al. [20] and Petersson et al. [21], respectively; these isolates were further confirmed to be S. aureus. The MRSA isolates were screened by cefoxitin disc (30 mg, Oxoid) diffusion test, while the polymerase chain reaction (PCR) was employed for the detection of the mecA gene [22]. In the case of strains isolated from the same patient, if the genotyping studies revealed identical genotype, only one of them was included in the study, which ensured that all strains involved in the current study were non-repetitive. All the strains were preserved at − 80 °C in a bacterial cryopreservation reagent comprised of 2.5% TSB (w/v), 16.7% glycerol (v/v), and 66.7% sterile horse serum (v/v).

The clinical data of children, including age, sex, medical history, isolation site, infection sites, and medication use, were collected. S. aureus infections were categorized as healthcare-associated (HA) or community-associated (CA) according to the epidemiology definitions established previously [23].

Delta-hemolysin expression

RNAIII is the major effector molecule of agr system, and also encodes the gene for delta-hemolysin (hld) [24]. Thus, δ-hemolysin production was used to assess the function of agr operon. The expression of δ-hemolysin was determined using S. aureus RN4420, which produced only β-hemolysin without the interference of α- or δ- hemolysins. Furthermore, β-hemolysin and δ-hemolysin have synergetic effects, while β-hemolysin inhibits lysis by α-hemolysin [25]. Therefore, the presence of enhanced hemolysis within the β-hemolysin zone of RN4220 indicates the production of δ-hemolysin by the test strains.

agr sequencing

The agr locus of non-hemolytic isolates was amplified using the method reported by Robinson et al. [26]. In addition, agr-w1f (5′-CCATTTgCCCAATTCATTC-3′) was used for sequencing the PCR product amplified by agrX1F (5′-TCGTATAATGACAGTGAGGAGAGT-3′) and agrCD2434dn (5′-TAATACCAATACTGCGACTT-3′). The resultant sequences were compared with the known agr sequences of S. aureus strain from the appropriate agr specificity group and sequence types (STs), including NCTC8325 (CC8) (CP000253.1), SA957 (ST59) (NC_022442.1), ST398 (NC_017333.1), N315 (agr group II), and H-EMRSA-15 genome (ST22) (CP007659.1).

Molecular genotyping analysis

Multilocus sequence typing (MLST) was performed as described by Enright et al. [27]. The allelic profiles (allele numbers) and ST types were determined based on the MLST database (http://saureus.mlst.net/). The staphylococcal protein A (spa) gene repeat region was amplified and sequenced as described previously [28], and the sequencing data were submitted to the S. aureus spa type database (http://spaserver.ridom.de) to determine the spa type. Agr typing was assigned by multiplex PCR according to the method described by Gilot et al. [29]. COL (agr I), N315 (agr II), TY114 (agr III), and A920210 were used as positive controls.

The staphylococcal cassette chromosome mec (SCCmec) types of MRSA isolates were determined using a multiplex PCR as described previously [30]. The reference strains used for SCCmec typing included NCTC10442 (SCCmec I), N315 (SCCmec II), 85/2082 (SCCmec III), JCSC4744 (SCCmec IV), and IMVS 67 (SCCmec V).

Detection of pvl gene

The detection of pvl was carried out using primers and conditions as described by Jarraud et al. [31]. ATCC25923 was used as a positive control.

Biofilm formation assays

Tissue culture plate method (TCP) was used to assess the biofilm forming ability of the nonhemolytic strains, as described in our previous study [32]. Briefly, overnight cultures in tryptic soy broth (TSB) (OXOID, USA) containing 0.25% glucose were adjusted to 106 CFU/mL. Then, 0.2 mL cell suspension was inoculated into each well of 96-well flat-bottom plates (Corning Costar #3599, USA) at 37 °C for 48 h. Subsequently, the wells were washed two times with normal saline, fixed by methanol for 15 min, stained with 0.1% crystal violet for 5 min, rinsed, and air-dried. The stained biofilm was solubilized with 33% glacial acetic acid, and the optical density (OD) was measured at 590 nm using on a CLARIOstar Microplate reader (BMG LABTECH, Germany). Each isolate was tested in triplicates. The negative control wells contained only the broth. The cut-off OD value (ODc) was defined as an average OD of negative control with three times of standard deviation. The biofilm formation ability was classified as negative (OD ≤ ODc), weak (ODc < OD ≤ 2ODc, WBF), moderate (2ODc < OD ≤ ODc, MBF), and strong (4ODc < OD, SBF).

Antimicrobial susceptibility testing

Antimicrobial susceptibility testing to 12 antimicrobial agents (penicillin G, oxacillin, erythromycin, clindamycin, tetracycline, gentamicin, chloramphenicol, ciprofloxacin, rifampin, linezolid, vancomycin, mopiroxacin; National Institutes for Food and Drug Control, China) were performed by agar dilution method as described by Wiegand et al. [33] with slight modification. Mueller-Hinton Agar (MHA; OXOID CM0337B, UK) medium without cation-adjustments were prepared and autoclaved according to the manufacture’s instructions. The antibacterial drugs were diluted for usage according to the Clinical and Laboratory Standards Institute (CLSI) guidelines [34]. Appropriate MHA medium was cooled to about 50 °C and poured to the 15 × 100 mm petri dish to produce the required depth of 3–4 mm. The final antibiotic concentration for each drug ranged from 0.032–256 mg/L and three control agar plates were without any antibiotic. The inoculation was carried out using a multipoint inoculator. Bacterial suspension, at a density of 1 × 107 CFU/mL, was inoculated to the agar plates starting from the lowest concentration. Then, the inoculum spots were dried at room temperature before inverting the plates and incubated at 35 °C for 16–20 h before obtaining the minimal inhibitory concentration (MIC). In addition, the E-test method was used to determine the MIC of all isolates to sulphamethoxazole/trimethoprim (SXT) (BioMeriuex, France). MHA plates were inoculated by streaking the standardized inoculums (0.5 McFarland, about 1.5 × 108 CFU/mL with a sterile swab. The SXT E-test strips (BioMeriuex) were placed on the plates, followed by incubation at 35 °C for 16–20 h.

The MIC reading for both E-test and agar dilution method was conducted independently by a senior experimenter, with the result confirmed by a second reader. The results of MIC were interpreted according to the CLSI breakpoints for Staphylococcus spp. [34]. S. aureus ATCC29213 was used as quality control. Multidrug resistance (MDR) was defined as isolates resistant to ≥3 classes of non-β-lactam antimicrobials for MRSA, and resistant to ≥3 classes of antibiotics including β-lactam antibiotics for MSSA.

Statistical analysis

SAS JMP Statistical Discovery v11.0 was used for statistical analysis. Chi-squared (χ2) test or Fisher’s exact test was used to analyze the categorical variables, and Wilcoxon rank sum test was used to compare the biofilm formation ability between the two groups. P < 0.05 was considered as statistically significant.

Results

Clinical characteristics

A total of 269 non-duplicate S. aureus clinical isolates were collected from Beijing Children’s Hospital, including 211 MRSA from September 2010–2017 and 58 MSSA from February 2016–2017. These strains were isolated from several clinical sources, including respiratory tract (6 from throat swab, 69 from sputum, and 37 from bronchial alveolar lavage fluid), skin and soft tissue (38 from pus, 20 from secretions of omphalitis, 10 from skin secretions, 17 from wound surface, 6 from eye secretions, and 2 from ear secretions), sterile sites (48 from blood, 6 from pleural effusion, 3 from bone marrow, 3 from cerebrospinal fluid, 2 from joint effusion, and 1 from seroperitoneum), and midstream urine (1 isolate). The characteristics of patients (107 females and 162 males; median age: 11.9 months) from whom samples were collected are shown in Table 1. Approximately 72.86% (196/269) of the patients were < 3-years-old. The proportion of children with community-associated (CA) infections and hospital-associated infections was roughly equivalent (50.19% vs. 49.81%). Then, 9/10 agr dysfunctional isolates caused healthcare-associated (HA) infections. A total of 29.74% (80/269) children presented invasive infections. All children included in this study were treated with antibiotics. Vancomycin and linezolid were used in 24.54% (66/269) and 16.73% (45/269) patients, respectively. The median hospital stay was 15 (interquartile range, IQR: 10–24) Days.

Table 1 Characteristics of patients for samples

Molecular typing and virulence characteristics

Among the 269 isolates, only 3.79% (8/211) MRSA and 3.45% (2/58) MSSA isolates were identified with agr dysfunction (no apparent hemolytic activity as shown in Fig. 1), and the overall prevalence rate was 3.71% (10/269).

Fig. 1
figure 1

δ-hemolysin assays. a δ-hemolysin-negative; b δ-hemolysin-positive (LAC USA 300). The arrows showed RN4220

The molecular characteristics of MRSA and MSSA stratified by agr functionality were shown in Table 2 and Table 3, respectively. ST59-SCCmec IV-t437/t441-agr I (49.29%, 104/211) and ST239-SCCmec III-t030/t037-agr I (9.00%, 19/211) were the most common MRSA clones. The prevalence rate of agr dysfunction in ST239-SCCmec III isolates (17.39%, 4/23) was significantly higher than in ST59-SCCmec IV (1.69%, 2/118) and other genotype strains (2.86%, 2/70) (P = 0.006) (Table 1). For MSSA isolates, the top 4 genotypes were ST22-t309-agr I (15.52%, 9/58), ST398-t571-agr I (10.34%, 6/58), ST5-t002-agr II (6.90%, 4/58), ST6-t701-agr I (6.90%, 4/58), and ST188-t189-agr I (6.90%, 4/58).

Table 2 Genotypic characteristics of MRSA isolates stratified by agr functionality
Table 3 Genotypic characteristics of MSSA isolates stratified by agr functionality

The detection rate of pvl was similar in MRSA (35.55%, 75/211) and MSSA (29.31%, 17/58) isolates (P = 0.4359). Among MRSA, the pvl prevalence rate of ST239-SCCmec III isolates (4.35%, 1/23) was significantly lower than that of ST59-SCCmec IV (37.3%, 44/118) and other strains (42.86%, 30/70) (P = 0.0031). All the ST22 isolates were pvl-positive. Among the ten agr dysfunctional isolates, only one MRSA belonging to the ST59-SCCmec IV-t437-agr I and one MSSA belonging to ST22-t310-agr I harbored pvl.

Biofilm formation

Table 4 shows that 88.63% (187/211) of MRSA and 56.90% (33/58) of MSSA isolates were strong biofilm formers. However, the biofilm formation ability of MRSA was significantly higher than that of MSSA isolates (P < 0.0001) (Fig. 2a). Interestingly, no significant difference was detected between agr dysfunction and agr functional isolates regarding the biofilm formation ability (P = 0.4972) (Fig. 2b); nevertheless, all the eight agr dysfunctional MRSA isolates and one MSSA isolate showed strong biofilm formation. Any association between pvl and biofilm formation in both MRSA and MSSA isolates was not observed (P = 0.4004 and P = 0.0509, respectively) (Figs. 2c and d).

Table 4 Biofilm formation ability of MRSA and MSSA regarding agr functionality [N (%)]
Fig. 2
figure 2

Biofilm formation ability of S. aureus. a Comparison between MRSA and MSSA isolates. b Comparison between agr dysfunctional and agr functional isolates among MRSA. c Comparison between pvl (+) and pvl (−) isolates among MRSA. d Comparison between pvl (+) and pvl (−) isolates among MSSA. OD590, optical density at 590 nm

Antimicrobial resistance

The results of antimicrobial susceptibility test are shown in Table 5. In this study, the S. aureus isolates were sensitive to vancomycin and linezolid, but highly resistant to penicillin (97.03%, 261/269) and erythromycin (98.89%, 266/269). Approximately, 80.09% (169/211) of MRSA and 74.14% (43/58) of MSSA isolates were MDR strains. The non-susceptibility rate of MRSA to oxacillin, clindamycin, tetracycline, and rifampin was significantly higher than that of MSSA isolates (P < 0.05). In addition, the non-susceptibility rates to ciprofloxacin, gentamicin, and SXT among MRSA isolates were significantly higher in agr dysfunctional isolates than in those with functional agr.

Table 5 Non-susceptibility rates of S. aureus in pediatric population in China [N (%)]

agr sequencing

Two isolates that belonged to ST239 had no mutations in agr locus, and another ST239 isolate harbored a synonymous mutation in agrA. Inactivating mutations were detected in other 5 MRSA and 2 MSSA isolates. The variants were characterized by nonsynonymous changes (n = 5) and frameshift mutations (insertions, n = 2), which mainly occurred in agrC and agrA (Table 6).

Table 6 agr mutations for non-hemolytic isolates

Discussion

Recent evidence indicated that agr dysfunction was common among S. aureus clinical isolates, especially MRSA [11,12,13,14, 17, 18]. However, the dysfunction was unusual in pediatric populations in China according to the current study. Moreover, the data showed that infections caused by agr-dysfunctional strains were always HA. Similarly, the study by Shopsin et al. [35] showed that the carriage of an agr-defective strain was associated with hospitalization. Butterfield et al. [13] also indicated that agr dysfunction was closely associated with prior administration of β-lactam and fluoroquinolone. Therefore, the high antibiotic selection pressure in the medical environment might lead to the emergence of agr dysfunction, which might be due to the agr-controlled virulence that is energy-consuming and needs to be balanced with the expression of antibiotic resistance in a healthcare environment filled with antibiotics [36].

In the current study, MRSA strains showed strong homology, and ST59-SCCmec IV and ST239-SCCmec III were the most prevalent clones. This result was consistent with that of a previous study conducted by Qiao et al. in Chinese children [37]. These results showed that agr dysfunction was more common in ST239-SCCmec III than in ST59-SCCmec IV isolates, the phenomenon was consistent with a previous study, which indicated that SCCmec IV/IVa MRSA (3%, known as CA-MRSA clone) was associated with lower rates of agr dysfunction as compared to SCCmec I-III MRSA (43%, known as HA-MRSA clone) [38]. Thus, the low prevalence of agr dysfunction in this study might be attributed to more than half of the MRSA isolates belonging to ST59 harboring the SCCmec IV.

In addition, infections caused by PVL-producing ST22-t309-agr I clone should be under intensive research. The current literature demonstrated that ST22-MRSA isolates mainly carry SCCmec IV [39, 40]. However, in this study, ST22-t309-agr I-MRSA strains were classified as SCCmec V. Moreover, ST22-t309-agr I present in MSSA isolates indicated that ST22-t309-MRSA probably arose from ST22-t309-MSSA.

Furthermore, S. aureus can lead to chronic infections by forming the biofilm on the surface of medical implants [41]. Herein, a majority of the S. aureus isolates were strong biofilm formers, especially MRSA, which should be brought to the attention of Chinese pediatricians. In addition, agr dysfunction has been linked to increased biofilm formation and enhanced colonization ability previously [11]; however, this phenomenon was inconsistent with that in the current study and could be attributed to the small sample size of agr dysfunctional isolates.

Both PVL and biofilm are major virulence factors of S. aureus. Previous studies found that some secreted virulence factors are closely related to biofilm formation, such as phenol-soluble modulin α (PSMα) [42]. Thus, we tried to explore the correlation between PVL and biofilm formation. However, no such association was found in both MRSA and MSSA isolates. A previous meta-analysis demonstrated that PVL strains are rare in colonizing isolates as compared to isolates causing skin and soft-tissue infections [43]. These results indicated that PVL was primarily associated with disease rather than biofilm formation and colonization.

Intriguingly, we found that agr dysfunctional isolates were more resistant than isolates with functional agr. However, at present, agr dysfunction is the result or cause of drug resistance remains unclear, while methicillin resistance might lead to agr dysfunction. Rudkin et al. [44] found that the expression of mecA (major resistance determinant of MRSA) could subtly affect the peptidoglycan structure or its interaction with other cell wall-associated proteins and prevented the detection of autoinducing peptide (AIP), thereby resulting in an unresponsive agr system and the subsequent low-level toxicity. When the type II SCCmec element or only mecA was deleted from an isolate with dysfunctional agr, the agr activity was restored. However, Tsuji et al. [45] suggested that agr dysfunctional might directly influence the acquisition of intermediate resistance to vancomycin after subtherapeutic exposure.

Nonetheless, all S. aureus strains, including the 10 agr dysfunctional isolates, were sensitive to vancomycin. Kim et al. [46] showed that 2/12 initial agr-functional isolates acquired agr dysfunction during vancomycin therapy for persistent bacteremia, but were still sensitive to vancomycin. In addition, all the 4 strains developed from vancomycin-susceptible S. aureus (VSSA) to heterogeneous VISA (hVISA) were initially agr dysfunctional strains. Therefore, we speculated that agr dysfunctional isolates can more easily adapt to glycopeptide selection pressure than agr functional isolates. Taken together, further studies assessing the correlation between agr dysfunction and antibiotic resistance, especially vancomycin resistance, are imperative.

The agr quorum sensing system has become a new target for developing new antibiotics. Hitherto, many natural and synthetic compounds have been found to interfere with the functions of agr [47]. However, the prevalence of agr dysfunction among S. aureus, and the potential correlation between agr dysfunction and antibiotic resistance indicated that isolates could withstand drug interference with agr functionality emerge rapidly. Strikingly, the promotion effects of agr dysfunction on biofilm formation may pose a great threat for patients using indwelling devices. Thus, additional studies are needed to explore the feasibility of agr system as a new target for antimicrobial agents.

Herein, we found that the agr sequence was not only associated with the agr group but also to the MLST types. Thus, the agr group and MLST types were considered while detecting the occurrence of mutation in the current agr dysfunctional isolates. Furthermore, agr dysfunctional isolates detected mutations that mainly occurred in agrC and agrA, leading to the inactivation of the auto-activation circuit and RNAIII expression [48]. However, no mutation or synonymous mutation or no mutations were also detected in our study, which suggested the involvement of additional mechanisms in agr dysfunction. Interestingly, mecA expression [49, 50] and abnormal expression of regulators of agr (sarA) can also be the cause of agr dysfunction [11].

Nevertheless, the current study had limitations. First, the sample size was relatively small. Second, this was a single-center study, and all the isolates were collected from one hospital. However, the hospital serves the whole of China, and > 60% of the hospitalized children come from all over the country. Therefore, the present study is still representative in China. Third, some MRSA strains used in this study have been sub-cultured 2 or 3 times since first collected, and multiple subcultures and long-term cryopreservation might have affected the characteristics of the strain.

Conclusions

In summary, agr dysfunction was not common in pediatric populations in China, and ST59-SCCmec IV was associated with a low prevalence of agr dysfunction as compared to the ST239-SCCmec III isolates. Agr dysfunctional isolates were always healthcare-associated and multidrug resistant. Except for the gene mutations, other mechanisms might also be involved in agr dysfunction. The agr dysfunction might be a major adaption mechanism of S. aureus to antibiotic selection pressure. Thus, an in-depth understanding of agr dysfunction is an urgent requirement for the development of new antibiotics that target agr expression.