Antibiofilm effects of N,O-acetals derived from 2-amino-1,4-naphthoquinone are associated with downregulation of important global virulence regulators in methicillin-resistant Staphylococcus aureus

Despite the existing antibiotics, antimicrobial resistance is a major challenge. Consequently, the development of new drugs remains in great demand. Quinones is part of a broad group of molecules that present antibacterial activity besides other biological properties. The main purpose of this study was to evaluate the antibiofilm activities of synthetic N,O-acetals derived from 2-amino-1,4-naphthoquinone [7a: 2-(methoxymethyl)-amino-1,4-naphthoquinone; 7b: 2-(ethoxymethyl)-amino-1,4-naphthoquinone; and 7c: 2-(propynyloxymethyl)-amino-1,4-naphthoquinone] against methicillin-resistant Staphylococcus aureus (MRSA). The derivatives 7b and 7c, specially 7b, caused strong impact on biofilm accumulation. This inhibition was linked to decreased expression of the genes fnbA, spa, hla and psmα3. More importantly, this downregulation was paralleled by the modulation of global virulence regulators. The substitution of 2-ethoxymethyl (7b) in comparison with 2-propynyloxymethyl (7c) enhanced sarA-agr inhibition, decreased fnbA transcripts (positively regulated by sarA) and strongly impaired biofilm accumulation. Indeed, 7b triggered intensive autolysis and was able to eliminate vancomycin-persistent cells. Consequently, 7b is a promising molecule displaying not only antimicrobial effects, but also antibiofilm and antipersistence activities. Therefore, 7b is a good candidate for further studies involving the development of novel and more rational antimicrobials able to act in chronic and recalcitrant infections, associated with biofilm formation.


N,O-acetal derivatives.
To obtain N,O-acetals derivatives from 2-amino-1,4-naphthoquinone (6a), a methodology that employs microwave irradiation was used as described previously 11 . The 6a and its derivatives 7a [2-(methoxymethyl)-amino-1,4-naphthoquinone], 7b [2-(ethoxymethyl)-amino-1,4-naphthoquinone] and 7c [2-(propynyloxymethyl)-amino-1,4-naphthoquinone] were characterized using spectroscopic methods, such as 1 H and 13 C NMR 11 . Stock solutions of these derivatives were prepared in DMSO 100%. In the experiments, the final concentrations of DMSO were ≤ 2% (v/v), which showed in previous tests performed in this study to cause no effect in S. aureus planktonic or sessile growth. Susceptibility tests. The antibacterial activity of 6a and its derivatives 7a, 7b, and 7c was tested as previously described using paper disks impregnated with each compound 13 . The derivatives were also analyzed quantitatively using microdilution assays in 96-well inert polystyrene microtiter plates (Nuclon; Nalge Nunc International, Rochester, NY, USA) to determine the minimal inhibitory concentration (MIC) 14 . Biological control, as well as growth conditions, were performed following CLSI recommendation 15,16 . Additionally, the MIC values of the MRSA strains BMB9393 (BEC) and CR15-071 (USA100), and the control strain ATCC25923, for the derivatives 7a, 7b and 7c were also determined using Mueller Hinton agar dilution following CLSI recommendation 16 . Controls were also performed using 2% (v/v) DMSO only. Cytotoxicity assay. The Vero cell line was cultivated to form monolayers on 96-well flat-bottom culture plates with Dulbecco's modified Eagle's medium (DMEM; Thermo Fisher Scientific, Waltham, MA, USA ) supplemented with 5% (v/v) fetal bovine serum (FBS; Thermo Fisher Scientific), 1 µg/mL fungizone, 2.5 Mm l-glutamine (amino acid), and 100 U/mL penicillin, during 24-h incubation period at 37 °C and 5% carbon dioxide (CO 2 ). After that, the supernatant was removed, and the cell monolayers were incubated for 24 h with different concentrations of 6a or its derivatives 7a, 7b, or 7c, ranging from 8 to 512 µg/mL. For comparative purposes, these experiments were also performed with the commercially available antibiotics, tetracycline and trimethoprim-sulfamethoxazole, using the same range of concentrations. All compounds were diluted in DMSO for a final concentration of 1% (v/v) in the cell monolayers. In the experimental controls, DMSO was added to the drug-untreated cell monolayers for a final concentration of 1% (v/v). After incubation, the supernatant was removed, and 100 µL fresh medium and 20 µL 3-[4,5-dimethylthiazole-2-y]-2,5-diphenyltetrazolium bromide (MTT; 1 mg/mL) were added into each well. Plates were incubated for 1 h, the supernatant again removed, and 200 µL dimethyl sulfoxide (DMSO) were added to each well to dissolve the formazan crystals for 10 min. Absorbance was measured at 570 nm using an ELISA reader (Thermo Plate, TP-READER, Thermo Fisher Scientific). The cytotoxic concentration 50% (CC 50 ) value was defined as the concentration of a derivative required ADMET properties. The in silico pharmacokinetic properties and toxicity profiles (ADMET) of 6a and its derivatives 7a, 7b and 7c were evaluated using pkCSM-pharmacokinetic webserver (https ://biosi g.unime lb.edu.au/pkcsm /) 20 , Pro-Tox-II webserver (https ://tox.chari te.de /protox_II/) 21 and SwissADME webserver 22 . These results were also compared with the profiles of an antimalarial (atovaquone) and six antimicrobials (cefoxitin, ciprofloxacin, furazolidone, nitrofurantoin, tetracycline, and vancomycin). The theoretical pharmacokinetic properties evaluated were absorption, distribution, metabolism, and excretion (ADME). The absorption parameters were tested through simulation modelling of human intestinal absorption using Caco-2 permeability, human intestinal absorption, P-glycoprotein substrate, and P-glycoprotein I/II inhibition. For drug distribution, it was analyzed the steady-state volume of distribution (Vss) and the blood-brain barrier penetration (BBBP) predictions. Drug metabolism was analyzed by the theoretical activity of cytochrome P450 (CYP) enzymes. Lastly, excretion parameters including theoretical total clearance and renal OCT2 substrate were evaluated. Toxicological analyses comprised toxicity target, hepatotoxicity, hERGI/II inhibitors, and toxicity endpoints including carcinogenic, mutagenic, immunotoxicity, and cytotoxicity parameters.

Effect of N,O-acetal derivatives on biofilm development.
Biofilms were formed on polystyrene surfaces using the BEC representative, strain BMB9393. It had previously been shown that this method has a good correlation with an in vivo biofilm model 23 . The compound 6a (1MIC = 128 µg/mL) and its derivatives 7a (1MIC = 256 µg/mL), 7b (1MIC = 128 µg/mL), and 7c (1MIC = 64 µg/mL) were added, separately, into each well of a polystyrene microtiter plate (Nunc) at concentrations of 1/4MIC 1/8MIC, or 1/16MIC in trypticase soy broth with 1% glucose (TSB-Glu; 2× final concentration). The same volume of the bacterial culture in TSB-Glu (37 °C/18 h; at 250 rpm) diluted 1/100 in the same broth was added to the wells. After incubation (37 °C/24 h), supernatants were gently removed, rinsed with distilled water, and treated with crystal violet solution as described 23 . The adhered biomass was suspended in ethanol PA, and the OD read at 570 nm using the SpectraMax Plus 384 (Absorbance Microplate; Molecular Devices, San Jose, CA, USA). Eight independent tests with three replicates each were performed. Controls were inoculated with the same bacterial inoculum but using only TSB-Glu with 2% (v/v) DMSO, final concentration.

Confocal laser scanning microscopy (CLSM).
Because derivative 7b showed the stronger reduction in biofilm accumulation, this compound was elected for these and other experiments designed to unveil the mechanisms by which this molecule may affect biofilm development. For comparative purposes and validation of the experiments performed here, the derivative 7c, which exhibited a lesser but still important reduction on biofilm accumulation, was also included in these studies. For CLSM experiments, after 24 h of incubation, the supernatant containing planktonic cells was gently removed, and biofilm treated with 25 nM SYTO 9 DNA-intercalating stain (Invitrogen; Carlsbad, CA, EUA), as previously described 24 . Before visualization, SYTO 9 solution was removed and the material visualized using a confocal laser scanning microscope (Model LSM510 Carl Zeiss Meditec; Jena, Germany). The microscope was configured with the argon laser (458 nm/477 nm/488 nm/514 nm). The images were randomly captured with the Neorlu-Plan 406/06 Korr. Controls were performed using 2% (v/v) DMSO only (final concentration). Three independent experiments were performed in quadruplicate.
Expression of biofilm-associated genes. Total RNA was obtained using the RNeasy Kit (Qiagen; Hilden, North Rhine-Westphalia, Germany) from sessile cells treated with 1/4MIC of the N,O acetal derivatives 7b (32 µg/mL) or 7c (16 µg/mL). Gene expression analyses were performed using real-time RT-qPCR with Power SYBR Green RNA-to-CTTM 1-Step Kit (Applied Biosystems, Foster City, CA, USA). The platform used was a One Step Real-Time PCR System (Applied Biosystems). The genes chosen were those well characterized as associated with biofilm development, including fnbA, spa, mecA, hla, and psmα3-encoding fibronectinbinding proteins A (FnBPA), staphylococcal protein A (Spa), penicillin-binding protein 2a or 2′ (PBP 2a or 2′), α hemolysin, and phenol-soluble modulin α3, respectively. To confirm the data from real-time RT-qPCR for the biofilm-associated genes, and to further understand the mechanisms by which 7b caused a profound effect on MRSA biofilm, the same experiments were performed with the derivative 7c. Additionally, these data were also validated by testing important biofilm and global virulence regulators, involved in the modulation of the biofilm-associated genes, such as RNAIII (the main effector molecule of the Agr quorum sensing), sarA (a global Effect of 7b and 7c derivatives in S. aureus autolytic system. The autolytic activity in the presence and absence of the derivatives 7b or 7c was measured by quantifying atlA cDNA by real-time qRT-PCR, as described above. To validate the RT-qPCR data for derivative 7b, the same experiment was performed for derivative 7c. Additionally, to confirm these data, quantification of extracellular DNA (eDNA) was performed in the supernatant of bacterial growth treated and untreated with 1/4MIC of these two derivatives (32 µg/mL and 16 µg/mL, respectively), using a previously described method 25 . Controls were performed using 2% (v/v) DMSO only. Three independent experiments were performed in triplicate.

Effect N,O-acetal derivatives against subpopulations of MRSA cells persistent to vancomycin.
For this test, we choose the MRSA strain CR15-071 representative of the ST5-SCCmecII lineage (NY/ Japan-USA100 clone), since this MRSA has replaced the ST239-SCCmecIII (the Brazilian clone) in this country, and is now the most frequent MRSA lineage in Brazilian hospitals 1 . A high bacterial load (~ 10 9 CFU/ml) was inoculated on the surface of the TSA plate containing 8 µg/mL vancomycin alone, corresponding to 4MIC, or containing 4MIC vancomycin plus the derivatives 7b or 7c at concentrations of 1MIC (128 µg/mL), 1/2MIC (64 µg/mL), 1/4MIC (32 µg/mL), or 1/8MIC (16 µg/mL). The plates were incubated at 37 °C/24 h before being examined. For controlling the activity of the antimicrobials, plates containing vancomycin or each derivative alone at low bacterial load (~10 6 CFU) were also assayed. Bacterial growth was determined by removing persistent cells from the plates and determining UFC/ml. Controls were also performed using 2% (v/v) DMSO only. Three independent experiments were performed in triplicate.

Statistical analysis.
All experiments were independently performed in triplicate with at least three technical replicates. The results shown are the mean ± standard deviation (SD). Statistical significance was determined by unpaired two-tailed Student's t-test for all experiments, except for the biofilm data that were analyzed by Tukey's test following One-Way ANOVA, considering a significance level of α = 0.01.

Results
Antibacterial activity. The N,O acetals derived from 2-amino-1,4-naphthoquinone (6a), the compounds 7a, 7b, and 7c, showed antimicrobial effects against the representatives of major MRSA international clones (Supplementary Table S2). It is important to highlight the activity of these derivatives-inhibition zones ranging from 8 ± 2.0 to 13 ± 1.0 mm-against the strain BMB9393 (ST239-SCCmecIII). This MRSA lineage shows resistance to several classes of antimicrobials besides β-lactams, including aminoglycosides, quinolones, and tetracycline 2 , but showed to be susceptible to all derivatives tested here. The antimicrobial properties of these compounds were also evaluated using the minimal inhibitory concentration (MIC) method. The MIC of the derivatives using broth microdilution for the reference strain ATCC 25923 and MRSA clones ranged between 16 to 128 μg/mL and 64 to 256 μg/mL, respectively ( Table 1). The MIC values for the strains BMB9393 (BEC) and CR15-071 (USA100) determined by broth microdilutions were also confirmed by agar dilution method.
High values for volume of distribution (Vss) − log Vss > 0.45-suggest that the drug will be distributed in body tissue rather than plasma. On the other hand, lower values are considered when log Vss < −0.15. The intermediate values suggest that the drug presents appropriate plasma distribution profile 20 . In this analysis, the compounds showing values below − 0.15 were the derivative 7a, cefotoxin, ciprofloxacin, furazolidone, nitrofurantoin, and vancomycin. In addition, all derivatives tested were able to pass through blood brain barrier (BBB) according SwissADME webserver (Table 3).
According to the results obtained, none of the compounds evaluated was considered a substrate of CYP2D6, except atovaquone. The derivatives 6a, 7a and 7b did not inhibit CYP2C19, CYP2C9, CYP2D6, and CYP3A4, but showed inhibitory properties against CYP1A2. 7c derivative did not inhibited CYP2C9, CYP2D6, and CYP3A4, despite their inhibitory effect on CYP1A2 and CYP2C19. Among the referential drugs, only atovaquone showed inhibitory effects on CYP1A2, CYP2C19, and CYP2C9 ( Table 3).
The pkCSM webserver was used to analyze the Log (CLtot) values and the possibility of these compounds acting as substrates for organic cation transporter 2 (OCT2). The values of Log (CLtot) of all derivatives ranged from − 1.273 to 0.987, and thus none of them would apparently be a substrate of renal OCT2 ( Table 3).
The toxicity parameters of the derivatives studied were predicted using PkCSM and Pro-Tox II webservers ( Table 4). The oral rat acute toxicity (LD50) for 6a, 7a, 7b and 7c derivatives ranged from 1.776 mol/kg to 2.192 mol/kg. The rat oral chronic toxicity (LOAEL) varied from 2.503 log mg/kg b.w. per day to 2.776 mg/kg b.w. per day, and the maximum tolerated dose ranged from 0.516 log mg/kg b.w. per day to 0.778 log mg/kg b.w. per day (Table 4). Hepatotoxicity has been and remains a major reason for drug withdrawal from clinical use 26 . Considering pkCSM prediction, the compounds showing hepatotoxic potential were just amongst referential drugs including cefoxitin, ciprofloxacin, doxorubicin, and vancomycin (Table 4).
Concerning the cardiotoxic potential, all N,O-acetal derivatives were classified as non-hERG I and II inhibitors. It is noteworthy that atovaquone, doxorubicin, and vancomycin were all classified as hERG II inhibitors, according with platform used (Table 4). Similar to doxorubicin, tetracycline, and vancomycin, the derivatives 7a and 7c were predicted to be immunotoxic. Moreover, 6a, 7a and 7c derivatives were expected to be carcinogenic as well as furazolidone and nitrofurantoin. The theoretical mutagenic effect was predicted for the derivatives 6a, 7a and 7b. However, this effect was also observed for doxorubicin, furazolidone, and nitrofurantoin, which are drugs of current clinical use ( Table 4).

Impact of N,O acetals derived from 2-amino-14-naphthoquinone on biofilm formation. The
concentration of 32 μg/ml (1/4MIC) of 7b was enough to reduce the accumulated biofilm matrix by 87.4% ± 1.9% while 16 μg/mL (1/4MIC) of 7c reduced by 75.6% ± 3.7%. This inhibition was more specific for biofilm development since such a strong impact in growth reduction was not observed for planktonic cells at these concentrations (Supplementary Figure S2). In addition, 1/4MIC 6a (32 μg/ml) and 7a (64 μg/ml), although showing antimicrobial effects, were much less effective at reducing biofilm development compared with 7b or 7c ( Fig. 2A). It is notable that vancomycin, one of the last resources to treat MRSA severe infections, was not able to reduce biofilm development ( Fig. 2A). CLSM was used to visualize the biofilm architectures and clearly document the inhibitory effects of 1/4MIC 7b and 7c (Fig. 2B). The images obtained endorse the results observed in a micro- www.nature.com/scientificreports/ plate-based assay, demonstrating that 7b and 7c were able to decrease biofilm accumulation massively compared with untreated biofilms (Fig. 2B). In these assays, 7b also showed superior ability in reducing biofilm formation in relation to 7c (Fig. 2B).

The antibiofilm effect of N,O-acetals derived from 2-amino-14-naphthoquinones was linked
to decreased expression of biofilm-associated genes. Some S. aureus genes are well known for their role in biofilm development. Therefore, due to the strong biofilm inhibition promoted by the derivative 7b, we investigated the role played by this compound in the expression of important biofilm-associated genes, to get some insights on the mechanisms by which this molecule strongly impairs biofilm production. Additionally, the derivative 7c, which also affects biofilm accumulation, but in lesser extent, was also included for comparison purpose. Among these genes, the mecA encoding PBP2A was not significantly affected by 1/4MIC of 7b (32 µg/ mL) or 7c (16 µg/mL) (Fig. 3A). The expression of fnbA encoding FnbpA protein-an important adhesin in S. aureus biofilm development-was reduced by 89.3% ± 1.5% after treatment with derivative 7b (p < 0.0001). However, for the 7c derivative, the reduction for this gene was only 24.64% ± 3.5% (p < 0.01) (Fig. 3A). The expression of hla encoding hemolysin/α-toxin (Hla) was considerably reduced by 90.0 ± 0.6% and 86.5 ± 1.0% after treatment with derivatives 7b and 7c, respectively (p < 0.0001). Also, an important reduction-by 95.4 ± 0.9% for 7b and 74.82 ± 4.9% for 7c (p < 0.0001)-was observed for the spa gene, which encodes the staphylococcal protein A. The expression of another important biofilm-associated gene, the psmα3 gene encoding the phenol soluble modulin alpha 3-was reduced by 97.75 ± 0.6% for 7b and 87.73 ± 4.3% for 7c (p < 0.0001) (Fig. 3A).

The role of global staphylococcal regulators in the decreased expression of biofilm-associated genes.
The effect of 1/4MIC of the derivatives 7b (32 µg/mL) and 7c (16 µg/mL) in the expression of rnaIII (encoding the RNAIII, the main effector molecule of the quorum-sensing Agr), sarA (encoding the transcriptional regulator SarA), and sigB (encoding an alternative RNA polymerase sigma subunit) were tested (Fig. 3B). A pronounced decrease was observed for the rnaIII expression, which regulates the temporal expression of several S. aureus virulence genes, such as hla, spa, fnbpA and psmα3. The number of RNAIII transcripts was reduced by 88.0 ± 1.1% for derivative 7b and 70.6 ± 2.2% for 7c (p < 0.0001). Derivative 7b also reduced importantly sarA gene expression at 65.5 ± 6.0% (p < 0.001). However, no important effect on sarA expression was observed for the derivative 7c. Moreover, derivative 7c increased sigB transcripts at 57.22 ± 5.2% (p < 0.01) while 7b did not affect the expression of the sigB transcription factor (Fig. 3B).

Effect of N,O-acetals derived from 2-amino-1,4-naphthoquinone in the S. aureus autolytic
system. The role played by 1/4MIC 7b (32 µg/mL) on the autolytic activity was evaluated in the biofilm cell growth of the strain BMB9393. An important increase of 3.5-times (p < 0.0001) in the atlA transcripts was observed. This result was validated by the fact that 1/4MIC 7c (16 µg/mL) led to three-fold increase in the atlA expression (p < 0.0001) (Fig. 4A). In agreement with this data, the amount of eDNA measured in the bacterial supernatant increased, after treatment of bacterial cultures with 7c. Indeed, the bacterial cell lysis was even higher (about two-fold) for the derivative 7b (p < 0.0001), which triggered an increased expression of atlA in comparison with 7c (Fig. 4B).   www.nature.com/scientificreports/ sistent cells was achieved for derivative 7b (p < 0.0001), and an important reduction of 88.3% (p = 0.0002) in the total of persistent cells were caused by the association with the derivative 7c (Fig. 5). Accordingly, the N,O acetal derivative 7b was not only able to inhibit bacterial biofilm, but also showed an important and impressive ability to kill persistent cells at MIC concentrations, which were totally refractory to 4MIC vancomycin.

Discussion
The N,O acetals derived from 2-amino-1,4-naphthoquinone, 7a, 7b, and 7c showed antibacterial activity not only for the methicillin-susceptible strain but also for internationally spread MRSA lineages, including ST239-SCCmecIII, which shows high multiresistance level, homogeneous resistance to methicillin, and increased ability to accumulate biofilm 2 . Divergent MIC values were observed for some of these derivatives in relation to previous studies with the strain ATCC 25923 11 . However, this may be attributed to the use of different methods in these two studies. Although we did not compare broth microdilution with broth macrodilution method, the latter used in the previous study 11 , the MICs for the derivatives 7a, 7b, 7c performed here, using Mueller Hinton agar dilution, confirmed the MIC data obtained by broth microdilution of the MRSA strains BMB9393 (BEC) and CR15-071 (USA100). Actually, previous studies have already demonstrated disagreement in MIC data obtained with different recommended methods for some drug/strain combinations, including for S. aureus 27 . The MRSA lineages (ST1-SCCmecIV, ST5-SCCmecII, ST8-SCCmecIV, ST30-SCCmecIV, and ST239-SCCmecIII) tested for these derivatives are currently causing the majority of hospital and community infections worldwide, including bacteremia and biofilm-associated infections 1 . Notably, despite the infection control measures, bacteremia caused by MRSA remains at high frequency and accounts for a mortality rate of about 25% 28 . The in vitro toxicity assays revealed that the derivatives 7a, 7b, and 7c did not display a hemolytic profile, according to Dobrovolskaia and collaborators 19 . The toxicity tests using monolayers of the Vero cell line revealed that the 7b derivative showed better SI value indicating that 7b displayed not only antibacterial effect but also higher selectivity against MRSA strain than toxicity to eukaryotic cells. Theoretical pharmacokinetic properties are important parameters to assess potential compounds during drug discovery process. All N,O-acetal derivatives showed good predictions for absorption values in the human intestine, similar to atovaquone, ciprofloxacin, and furazolidone. Indeed, the compounds 7b and 7c showed absorption values greater than 90%, close to that of ciprofloxacin, which is a good estimative for oral bioavailability 29 .
Interestingly, only N,O-acetal derivatives and atovaquone showed permeability to Caco-2 cells, which has been used as a model for human intestinal absorption of drugs 30 . The results also suggested that none of the N,Oacetal derivatives seem to affect P-glycoprotein, which is involved in drug exclusion 31 . All together, these data reveal an optimal in silico prediction of oral bioavailability for the N,O-acetal derivatives. Derivatives 6a, 7b and 7c showed intermediate Vss values, representing good predictions for adequate plasma distribution profiles. In addition, the analysis of BBB values, according to SwissADME calculation, suggested that N,O-acetal derivatives could cross blood brain barrier 32 .
It has been estimated that CYPs could be able to metabolize around 75% of the commercially available drugs 33 . None of the derivatives tested was substrate for cytochrome P450 isozyme CYP2D6 and CYP3A4. Moreover, these derivatives were only presumed to be inhibitors of CYP1A2, except 7c derivative that was predicted to inhibit CYP2C19. Interestingly, CYP1A2 is one main xenobiotic metabolizing enzyme in humans, and a recent study associated this enzyme with the bioactivation of procarcinogens, including 4-(methylnitrosamino)-1-(3pyridyl)-1-butanone (NNK), a tobacco specific and potent pulmonary carcinogen 34 .
Drug clearance was measured by determining Log(CLtot). Except for 6a, all derivatives tested presented high Log(CLtot) values, but still acceptable. Actually, these values were lower than those observed for referential drugs Note that 1MIC (128 µg/mL) of the derivative 7c was able to eliminate vancomycin-persistent cells completely. Data were represented by the mean of three independent experiments with triplicates. The bar represents the standard deviation. NS not significant; *p < 0.01; **p < 0.001; and ***p < 0.0001.

Scientific Reports
| (2020) 10:19631 | https://doi.org/10.1038/s41598-020-76372-z www.nature.com/scientificreports/ such as ciprofloxacin, doxorubicin, furazolidone, and nitrofurantoin. None of the N,O-acetal derivatives tested was predicted to show hepatotoxicity. Also, they were not expected to inhibit hERG I and II channels. Drug-hERG channel interactions have been considered a therapeutic challenge as a major cause acquired-long QT syndrome. Actually, a relationship between chronic heart failure (CHF) and S. aureus bacteremia was suggested. CHF patients with S. aureus bacteremia showed a significantly higher mortality rate compared with patients with normal heart functions 35,36 . This observation reinforces the need of non-cardiogenic effect for anti-MRSA drug candidates and highlights this promising profile of our derivatives. Data from the ProTox-II platform revealed that the 7b derivative showed good toxicological parameters, except for the mutagenic profile. This feature could be related to the mechanism of action of these derivatives. Besides oxidative stress, another mechanism of action generally associated with naphthoquinones is the ability to bind to topoisomerase complex causing damage to the DNA replication process 37 . Despite that, in vitro studies using V79 Chinese hamster cells did not confirm a mutagenicity effect for 1,4-naphthoquinone 38 .
Besides the antimicrobial activity against MRSA and the improvement of pharmacological parameters, the introduction of the 2-ethoxymethyl radical generating the 7b derivative resulted in the best antibiofilm and antipersistence effects. Furthermore, contrarily to the substitution by 2-methoxymethyl or 2-propynyloxy, the 2-ethoxymethyl substituted amino-1,4-naphtoquinone did not inhibit four of five CYPs enzymes tested, was not a substrate of CYP3A4 and CYP2D6, and also was predicted to have the best intestinal absorption, comparable to ciprofloxacin.
It is well known that biofilm development enhances S. aureus ability to cause infections and persist into the host 5 . In the present study, the N,O acetal derivatives from 2-amino-1,4-naphthoquinone, 7b and 7c, were able to reduce biofilm accumulation in 88% and 75%, respectively, at 1/4MIC without causing similar massive reduction in viability of planktonic cells. Oja and collaborators 39 also found that other naphthoquinone derivatives [biosynthetic pyranonaphthoquinone (PNP) polyketides] impaired biofilms formed by methicillin-susceptible S. aureus (MSSA), ATCC 25923, and Newman strains. It is important to remark that vancomycin, considered one of the last resources to treat serious infections by MRSA, did not impair biofilm development in the study model chosen. Not only that, induction of biofilm by vancomycin had already been observed for some S. aureus strains 40 .
In one of the first stages of biofilm development, bacteria adhere to biotic or abiotic surfaces, in a mechanism mediated by bacterial surface adhesion molecules, such as the fibronectin-binding protein A (FnBPA) encoded by the fnbA gene 5 . The expression of fnbA is upregulated by the sarA in an agrRNAIII-independent mode 41 . The derivative 7b, that showed a drastic reduction in biofilm accumulation in 1/4MIC, also presented a deeper decrease in the expression of the fnbA gene paralleled by an also important decline in sarA transcripts. Actually, it is well known that inhibition of sarA also leads to impairment of biofilm development in S. aureus 42 , which might explain the important reduction observed in the biofilm accumulation caused by the 7b derivative. In agreement with these data, the decrease of fnbA transcripts for MRSA treated with 7c was less pronounced when compared with 7b. Indeed, 7c also presented a lower ability to impair biofilm development compared with 7b. Moreover, the poorer effect of 7c in the decreasing sarA expression could be explained by the increase in the expression of sigB caused by 7c. SigB is a sarA-positive regulator that has also been implicated in the modulation of the maturation phase of S. aureus biofilms 43,44 .
Protein A is an antiphagocytic protein encoded by the spa gene that can also provide S. aureus with selfaggregation property, which is an important condition for biofilm development 45,46 . The MRSA treatment with both 7b and 7c derivatives caused an important reduction in the spa expression, which was more pronounced for 7b, and thus it might also account for the stronger biofilm impairment caused by these derivatives. The control of spa expression is complex and involves an intricate regulatory network 47 .
Notably, the derivatives 7b and 7c showed antibiofilm effects associated not only with a strong inhibition of fnbA and spa genes but also with an important attenuation in the expression of cytotoxicity-associated genes, such as hla encoding the staphylococcal α-hemolysin, hld encoding δ-hemolysin (whose coding sequence is localized inside the sequence of agr-rnaIII), and psmα3 encoding phenol-soluble modulin alpha 3 peptide 47,48 . It is very well known that PSMs are multifactorial molecules in staphylococcal pathogenesis. PSMs are broadly cytolytic, inducing the killing of different cell types. Furthermore, PSMs are important for the architecture of the S. aureus biofilms 49 . The α-hemolysin (Hla) is the prototype for small β-barrel pore-forming cytotoxins, which induces lysis of different host cells, including monocytes, neutrophils, pneumocytes, and endothelial cells. Besides its action promoting biofilm development 50 , it was observed that Δhla mutants lead to attenuated bloodstream infections in animal models 51 . It is notable that the inhibitory effects in toxin-associated genes were also more pronounced for 7b than 7c. Due to the importance of these genes for S. aureus pathogenesis, although we did not perform studies using animal models, it seems logical to suppose that the attenuation of cytotoxic genes, caused by these derivatives, is likely to result in antivirulence effects 49,51 .
It was previously found that agr-RNAIII can positively or negatively regulate biofilms depending on the background of the strains, and that agr-RNAIII is a positive biofilm regulator for strains with the ST239-SCC-mecIII background, such as BMB9393 24 . In fact, agr-RNAIII was impaired by both derivatives 7b and 7c, and again 7b caused a deeper effect by diminishing RNAIII transcripts. The impairment of RNAIII agrees with the reduction observed in hla and psmα3, since RNAIII is a positive regulator for these genes. AtlA is the main S. aureus muramidase that has an important role in cell wall biogenesis and the septation process during bacterial cell division 52 . It was found that small fractions of bacterial cell lysis enhance biofilm formation due to the ligation of eDNA to proteins and polysaccharides of the biofilm matrix 53 . However, it was demonstrated that the uncontrolled expression of atlA causes increased cell lysis, resulting in a defective biofilm formation 54 . Thus, the triggering of the autolytic system by 7b and 7c is likely to play a role in the impact of these naphthoquinone derivatives in MRSA biofilm development.
Therapeutic failures during the administration of vancomycin have been described, and in some cases, involved patients with MRSA infections 55 . A hypothesis suggested to explain these failures was linked to a Scientific Reports | (2020) 10:19631 | https://doi.org/10.1038/s41598-020-76372-z www.nature.com/scientificreports/ decreased penetration of vancomycin into the biofilm matrix 56,57 . Also, it was proposed that stress conditions in the hyper-populated environment, such as those encountered in biofilm growth or high bacterial load, might induce the generation of persistent cells 58 . However, the exact mechanisms through which bacteria become refractory/persistent to different antimicrobials in the biofilm environment are still to be completely clarified 5 .
In this study, we found that 7b derivative at 1MIC (128 µg/mL) was able to eradicate subpopulations of persistent cells that survived to concentrations of 4MIC (8 µg/mL) vancomycin. The ability of these derivatives to trigger cell autolysis might be involved in the killing of persistent cells. It was previously shown that the deletion of the operon msaABCR of S. aureus increased the processing of the major S. aureus autolysin by proteases, leading to the activation of the muramidase activity. In addition, it was found that msaABCR deletion enhanced the effectiveness of antibiotics against persistent cells, which was attributed to the increased cell lysis 54 . Thus, it is interesting to further investigate the role played by the autolytic system in the killing of vancomycin-persistent cells by these naphthoquinone derivatives and other active molecules.

Conclusions
In conclusion, the accelerated evolution and dissemination of antimicrobial resistance in S. aureus are outpacing the development of completely new antibiotics, and antibiofilm agents might provide an interesting alternative. In this context, the 2-(ethoxymethyl)-amino-1,4-naphthoquinone (7b) derivative, besides the antimicrobial action, also showed strong antibiofilm and antipersistence effects against multidrug-resistant MRSA strains. These effects were paralleled by the ability of 7b to concomitantly affects (directly or indirectly) important S. aureus virulence regulators (agr-RNAIII and sarA) and biofilm-associated genes (spa, fnbA, hla and psmα3). Finally, these properties associated with the biological and pharmacological aspects of this compound identified 7b derivative as an interesting model for the design of potentially promising and more effective drugs against MRSA (Fig. 6).