Anti-staphylococcus Antibiotics Interfere With the Transcription of Leucocidin ED Gene in Staphylococcus aureus Strain Newman

Antibiotics have been described to modulate bacterial virulence gene expression. This study aimed to assess the changes caused by anti-Staphylococcus agents in the transcription of leucocidin ED (lukED) gene of Staphylococcus aureus strain Newman in vitro and in vivo and to determine whether the altered expression is agr dependent. The bacteria were exposed to subinhibitory concentrations [1/2, 1/4, or 1/8 minimal inhibitory concentration (MIC)] of 11 antibiotics, and the expression of lukE and agr-effector RNAIII was determined using qRT-PCR. In vivo experiments were performed to evaluate the impact exerted by six representative antibiotics on the transcription of both genes. Molecular analysis showed that in vitro lukE transcription was dramatically promoted in the Newman strain exposed to sub-MICs of vancomycin, trimethoprim–sulfamethoxazole, clindamycin, gentamicin, daptomycin, and ciprofloxacin and considerably reduced when stimulated by cefazolin, erythromycin, rifampicin, tigecycline, and linezolid. In the murine abscess model, tigecycline significantly decreased the transcription of lukE and the bacterial numbers, whereas vancomycin increased them; although cefazolin increased the lukE expression (contrary to the in vitro effect), it had a remarkable role in reducing bacterial load. The correspondence analysis shows that RNAIII expression varied under seven of 11 antibiotics in vitro, and six drugs in vivo were consistent with lukE transcripts. In conclusion, our data show that anti-Staphylococcus antibiotics exert modulatory effects on lukE expression in vitro and/or in vivo, and the changed expression caused by some drugs may be involved with agr activity, thus providing a guide to choose appropriate agents to avoid promoting bacterial virulence in lukED-positive S. aureus infections.


INTRODUCTION
Staphylococcus aureus is a pathogen notorious for its ability to cause many infection-related illnesses ranging from cutaneous infections and food poisoning to toxic shock syndrome, septicemia, and necrotizing pneumonia (Tong et al., 2015). The success of S. aureus infection stems from a repertoire of virulence factors that enable the bacteria to escape from the host immune system (Otto, 2014). Among these factors, leucocidin ED (LukED), a bicomponent pore-forming toxin, plays an important role in S. aureus pathogenicity (Alonzo and Torres, 2014;Balasubramanian et al., 2016).
LukED targets the membrane of various cells such as neutrophils, T cells, myeloid cells, macrophages, dendritic cells, and erythrocytes and elicits β-barrel pores that span the lipid bilayer and lead to osmotic lysis of the host cell (Alonzo et al., 2012Reyes-Robles et al., 2013;Spaan et al., 2015). Epidemiological data and animal infection models show that lukED can be commonly detected in clinical S. aureus strains (approximately 2/3 to 4/5 of isolates) and is closely associated with impetigo, antibiotic-associated diarrhea, and bloodstream infection, among others (Gravet et al., 1998;Arciola et al., 2007;Alonzo et al., 2012;Alonzo and Torres, 2014;He et al., 2018). The accessory gene regulator (Agr)repressor of toxin (Rot) pathway is an important modulatory network of LukED production (Alonzo et al., 2012). The agr operon encodes the regulatory RNA RNAIII, which promotes the transcription of leucocidin genes by negatively controlling the yield of Rot (Benson et al., 2014;Killikelly et al., 2015;Tan et al., 2018).
During treatment, bacteria may be exposed to subinhibitory levels [sub-minimal inhibitory concentrations (sub-MICs)] of antibiotics owing to drug-resistant organisms or the pharmacokinetics of antimicrobial agents (such as short halflife, poor drug distribution and adherence, or interactions between antibiotics) (Cars, 1990;Hodille et al., 2017). Early investigations have shown that sub-MICs of antibiotics may initiate differential expression of virulence genes in S. aureus, which may affect the pathogenesis of infection and result in worse outcomes (Dumitrescu et al., 2007(Dumitrescu et al., , 2008(Dumitrescu et al., , 2011Stevens et al., 2007;Pichereau et al., 2012;Diep et al., 2013;Otto et al., 2013;Yamaki et al., 2013;Rudkin et al., 2014;Turner and Sriskandan, 2015;Hodille et al., 2017;Liu et al., 2018). Therefore, the therapeutic efficacy of antibiotics might also rely on their capacity to prevent the production of virulence factors (Otto et al., 2013). The use of antibiotics that reduce the Panton-Valentine leucocidin (PVL) toxin production is recommended for the treatment of severe infections caused by pvl-positive S. aureus (HPA, 2008;Nathwani et al., 2008). Nevertheless, little is known about the influence of antibiotics on lukED expression.
In this study, we selected common anti-Staphylococcus drugs to evaluate their impact on the expression of lukED in the S. aureus strain Newman in vitro and in vivo. We also analyzed whether the production of RNAIII is associated with variations in the levels of lukED transcripts affected by antimicrobial compounds.

Determination of Minimal Inhibitory Concentration
Minimal inhibitory concentrations of antibiotics against the S. aureus strain Newman were determined in triplicate by the standard microdilution broth method according to Clinical and Laboratory Standards Institute (CLSI) recommendations (Wayne, 2017).

Growth Kinetics
Overnight liquid cultures of strain Newman were diluted 1:100 into 25 ml of fresh YCP medium, followed by addition of 1/8 MIC, 1/4 MIC, or 1/2 MIC antibiotics. Cultures without antibiotic served as control. Cultures were grown at 37 • C with shaking at 150 r/min. Cell growth was detected by measuring the optical density (OD) at 600 nm every hour using a UV-2102C ultraviolet spectrophotometer (Unico Instruments, Shanghai, China).

In vitro Exposure to Antibacterial Agents
Bacterial culture aliquots for RNA extraction were collected after the early (3 h) and late (5 h) logarithmic growth phases, when transcription of lukED was rising and reached the highest level, respectively (Yang et al., 2019).

Extraction of Bacterial RNA
Bacterial culture samples were centrifuged at 13,000 × g and 4 • C for 10 min; resuspended in TE buffer (10 mM of Tris HCl and 1 mM of EDTA, pH 8.0) with lysostaphin (1 mg/ml, Sangon Biotech, Shanghai, China) and proteinase K (20 mg/ml, TaKaRa, Dalian, China); and incubated at 56 • C for 1 h for cell wall lysis. Total RNA was extracted using the MiniBEST Universal RNA Extraction Kit (TaKaRa, Dalian, China) according to the manufacturer's instructions.

Subcutaneous Abscess in Mice
Female Balb/c nude (nu/nu) mice between 4 and 6 weeks old were prepared for the abscess model. Mice were anesthetized with isoflurane and then injected subcutaneously with 100 µl of phosphate-buffered saline (PBS) containing 3 × 10 8 colonyforming units (CFU)/ml of fresh Newman strain to form the abscess (Turner and Sriskandan, 2015).

In vivo Exposure to Antimicrobials
After 48 h of infection, mice were injected intraperitoneally with 150 µl of 10 mg/kg of clindamycin, 10 mg/kg of linezolid, 25 mg/kg of cefazolin, 30 mg/kg of vancomycin, 4 mg/kg of daptomycin, 1.6 mg/kg of tigecycline, or PBS as a control according to the human therapeutic doses recommended by the Sanford Guide to Antimicrobial Therapy (Gilbert, 2014).

Enumeration and RNA Extraction of Bacteria From Abscess
Following a previously described method (Turner and Sriskandan, 2015), mice were sacrificed after 8 h of treatment, and the abscess tissue was cut. Samples were diluted in PBS and plated for bacteria counting. The remaining abscesses were placed into liquid nitrogen quickly, followed by grinding for extraction of RNA as described above.

Relative Quantitative RT-PCR
Bacterial RNA was quantified using a NanoDrop spectrometer (Thermo Fisher Scientific, Waltham, MA, United States), followed by purification and reverse transcription (1 µg of RNA) using the PrimeScript TM RT reagent Kit with gDNA Eraser (TaKaRa, Dalian, China). Gene transcript levels were determined by quantitative real-time amplification (qRT-PCR, SYBR Premix Ex TaqTM, TaKaRa, Dalian, China) in a 7500 Real Time PCR System (Applied Biosystems, CA, United States). Primers for qRT-PCR are listed in Table 1 (Balasubramanian et al., 2016;Gaupp et al., 2016). The mRNA levels of target genes were standardized against those of the housekeeping gene 16S rRNA. The fold change was determined using the 2 − CT method (Livak and Schmittgen, 2001).

Statistical Analysis
One-way analysis of variance (ANOVA) followed by a posteriori Dunnett's test was used to analyze the results (SAS Institute, Cary, NC, United States). Results were considered statistically significant when p < 0.05.

Minimum Inhibitory Concentrations of Antibiotics
The MIC values of 11 antibiotics, summarized in Table 2, showed that the S. aureus strain Newman was susceptible to all the drugs tested.

Impacts of Sub-Minimum Inhibitory Concentrations of Antibiotics on Staphylococcus aureus Growth
We generalized the impact of 11 antibiotics at sub-MICs on strain Newman growth ( Figure 1). As can be seen, graded concentrations of vancomycin, trimethoprim-sulfamethoxazole, clindamycin, gentamicin, daptomycin, and tigecycline triggered no significant growth defects over the entire growth curves compared with those of the control without drugs; in contrast, ciprofloxacin, cefazolin, erythromycin, rifampicin, and linezolid caused growth inhibition (p < 0.05) at 1/2 MIC. Because of this inhibition, we excluded these five antibiotics at 1/2 MIC from subsequent experiments of in vitro measurement of transcription to eliminate possible effects from antibioticinduced growth impairment.
In vivo, RNAIII transcript levels were remarkedly reduced by tigecycline by 5.37-fold (p = 0.004) and increased by vancomycin and cefazolin by 5.58-fold (p < 0.001) and 2.05-fold (p = 0.002), respectively ( Figure 3B). Table 3 shows the correspondence between the transcription levels of lukE and RNAIII in vitro and in vivo. Our data demonstrate that the expressional variations of RNAIII had a consistent trend with those of lukE when exposed to clindamycin at 1/8 to 1/2 MICs for 5 h; tigecycline at 1/4 MIC for 3 h; vancomycin at 1/8 to 1/2 MICs for 3 h and 1/2 MIC for 5 h; trimethoprim-sulfamethoxazole at 1/8 MIC and 1/2 MIC for 3 h and at 1/8 MIC and 1/4 MIC for 5 h; and ciprofloxacin, cefazolin, and rifampicin at 1/4 MIC for 5 h. In the animal abscess model, the expression levels of RNAIII were strongly consistent with those of lukE after exposure to tigecycline, clindamycin, daptomycin, linezolid, vancomycin, and cefazolin.

DISCUSSION
The S. aureus LukED toxin is able to trigger the damage of host cells and plays a vital role in controlling infection progress Reyes-Robles et al., 2013;Spaan et al., 2015). Therefore, this toxin may be established as a novel potential target of antitoxin therapy for S. aureus diseases (Nocadello et al., 2016). Antimicrobial treatment for most infections can promote rapid bacterial damage. However, sometimes, elimination of the pathogen does not occur quickly enough to prevent the harmful impact of virulence factors (Hodille et al., 2017). Thus, antibiotics-mediated reduction of virulence factor production was suggested for the treatment of toxin-mediated diseases (Nathwani et al., 2008). Here, we explored the effects of anti-Staphylococcus antibiotics commonly used in the clinic on lukED expression in vitro and in vivo using the S. aureus strain Newman, a good producer of LukED.
Clindamycin, linezolid, erythromycin, gentamicin, and tigecycline, protein synthesis inhibitor compounds, block mRNA translation at the level of the ribosome to suppress the production of staphylococcal exotoxin protein (Hodille et al., 2017). Therefore, these drugs can exhibit broad anti-virulence traits. In this study, we discovered that these drugs at sub-MICs also modulated the mRNA levels (increase or decrease) of lukE (Figure 2). Possible interpretations for this observation are that protein synthesis inhibitors specifically disturb the expression of regulator(s) or two-component signal transduction system(s) that regulate transcription or translation of virulence determinants or that the activities of proteases and RNases affect     the formation of the in-process product of the translational complex (Otto et al., 2013;Hodille et al., 2017). Previous reports showed that vancomycin has a minor effect on pvl, hla, and protein A (spa) mRNA levels (Dumitrescu et al., 2007(Dumitrescu et al., , 2008Otto et al., 2013;Hodille et al., 2017). Nevertheless, our findings exhibited a significant impact of vancomycin on lukE expression in vitro. This suggests that a cell wall-disrupting agent has the ability to induce some virulence gene expression at subinhibitory levels. It is believed that SOS response, leading to upregulation of an ensemble of DNA repair and recombination genes, can be activated by subinhibitory concentrations of trimethoprim and fluoroquinolones (Bisognano et al., 2004;Goerke et al., 2006;Hodille et al., 2017). In this investigation, the reason for the patently increased lukE expression regulated by trimethoprimsulfamethoxazole and ciprofloxacin may be related to the SOS response. A previous study reported that transcription of lukE is remarkedly stimulated by low concentrations of penicillin and cefalotin (Subrt et al., 2011). However, sub-MICs of cefazolin strongly inhibited this gene transcription in this work. Cefalotin and cefazolin both belong to the first-generation cephalosporins binding to penicillin-binding protein 1 (PBP-1).
The PBP-1-specific blockage by β-lactams can also initiate the SOS response (Hodille et al., 2017). However, this SOS-based mechanism of gene activation does not seem suitable to explain our observations. Here, we demonstrate an increased effect and a reduced effect on lukE transcription when S. aureus was exposed to sub-MICs of daptomycin and rifampicin, respectively. A published study showed that daptomycin also induces pvl mRNA, but the effect on hla mRNA level is varied and strain dependent (Otto et al., 2013). Rifampicin inhibits bacteria by suppressing the synthesis of mRNA; therefore, it is not surprising that this drug had an anti-LukED effect at sub-MIC. It is well known that in vitro treatment with antimicrobials does not sufficiently correlate to clinical exposure to drugs during disease. In contrast to the in vitro data, we measured a pronouncedly increased level of lukE transcript in mice exposed to cefazolin but a significant reduction in the S. aureus burden (Figures 2, 3). Exposure to the protein synthesis inhibitors clindamycin and linezolid or the lipopeptide antibiotic daptomycin in vivo (no effect) was also in contrast to the effects of antibiotics on the transcription of lukE in vitro (Figures 2, 3). However, tigecycline, also a protein synthesis inhibitor, not only inhibited lukE expression in vitro and in vivo but also reduced the bacterial load (Figures 2, 3). The same observation for tigecycline was also reported in a rat burn model (Nosanov et al., 2017). The superior ability of tigecycline in vivo may be correlated with tigecycline-induced differential modification of matrix metalloproteinase-9, which can recruit leukocytes to the site of infection for the elimination of the bacteria (Simonetti et al., 2012).
Previous data from animal model showed that vancomycin was inferior to linezolid (Yanagihara et al., 2009;Martinez-Olondris et al., 2012). In the present investigation, we found that vancomycin had a poorer ability in reducing bacterial level and a stronger role in elevating lukE expression in vivo than had linezolid (Figures 2, 3). The increase or decrease in bacterial counts when using antibiotics may have a significant effect on the production of total virulence factors, which may affect the progress of disease. Therefore, simple in vitro experiments cannot accurately represent the final results, and thus more in vivo experiments are needed to evaluate the effect of antibiotics.
The expression of S. aureus virulence genes is controlled by complicated mechanisms (Pichereau et al., 2012). Many global modulators fine-tune virulence factor expression in response to outside signals such as host defenses and antibacterial agents. A regulator of this kind is the agr quorum-sensing system (Alonzo et al., 2012;Hodille et al., 2017). So far, there have been plenty of studies on the response of agr operon to antibiotics (Joo et al., 2010;Otto et al., 2013;Cazares-Dominguez et al., 2015;Jin et al., 2018). In those studies, the tested antibiotics, such as cephalosporins, penicillin, ciprofloxacin, tetracycline, clindamycin, and tigecycline, induced RNAIII transcription, but aminoglycosides and mupirocin had an inhibitory role. Here, we also detected the levels of RNAIII transcript. Table 3 shows that agr activity (RNAIII expression) was modified by 10 of the antibiotics tested (except erythromycin) in a concentration-and/or time-dependent manner in vitro, and lukE transcript levels varied under seven of them (except gentamicin, daptomycin, and linezolid) with a consistent trend. This suggests that most antibiotics tested at sub-MICs may modify lukE expression by affecting agr activity. Moreover, the in vivo experiments of six representative antibiotics also suggested the same conclusion ( Figure 3B). Nevertheless, the mechanism by which antibiotics affect agr activity is unclear, and further investigation is needed. It is worth mentioning here that in our study, gentamicin, ciprofloxacin, cefazolin, and tigecycline did not show a completely consistent effect on RNA III expression, compared with those drugs mentioned by the above references. We speculate that the reason may be associated with the difference of antibiotic concentration, testing time point, and experimental strain.
The variations in lukE transcript levels may not necessarily translate to a difference in toxin production. Regrettably, we were not able to measure LukED toxin in the present study owing to the lack of a corresponding antibody. In addition, whether the in vitro and in vivo impacts on the Newman strain are applicable to other strains remains to be determined. Despite these shortcomings, our findings may still provide a clue to select suitable antibiotics for the treatment of lukED-positive S. aureus infections.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/supplementary material.

ETHICS STATEMENT
The animal study was reviewed and approved by the Shanghai General Hospital ethical committee on animal experiments: 2018DW003.

AUTHOR CONTRIBUTIONS
HY, SX, and KH carried out the experiments and wrote the manuscript. XX and FH provided the laboratory for making experiments. CH, WS, and ZW analyzed the data and interpreted the results. FG and CZ revised the manuscript critically. QL designed the experiments and corrected the manuscript. All authors read and approved the manuscript.