Strain-specific anti-biofilm and antibiotic-potentiating activity of 3′,4′-difluoroquercetin

Antibacterial properties of 3′,4′-difluoroquercetin (di-F-Q), a fluorine-substituted stable quercetin derivative, were investigated. Even though di-F-Q itself did not show interesting antibacterial activity, treatment of the Staphylococcus aureus strains with di-F-Q resulted in a dose-dependent reduction in biofilm formation with IC50 values of 1.8 ~ 5.3 mg/L. Also, the antibacterial activity of ceftazidime (CAZ) against carbapenem-resistant Pseudomonas aeruginosa (CRPA) showed eightfold decrease upon combination with di-F-Q. Assessment of the antimicrobial activity of CAZ in combination with di-F-Q against 50 clinical isolates of P. aeruginosa confirmed 15.7% increase in the percentages of susceptible P. aeruginosa isolates upon addition of di-F-Q to CAZ. Further mechanistic studies revealed that di-F-Q affected the antibiotics efflux system in CRPA but not the β-lactamase activity. Thus, di-F-Q was almost equally effective as carbonyl cyanide m-chlorophenyl hydrazine in inhibiting antibiotic efflux by P. aeruginosa. In vivo evaluation of the therapeutic efficacy of CAZ-(di-F-Q) combination against P. aeruginosa showed 20% of the mice treated with CAZ-(di-F-Q) survived after 7 days in IMP carbapenemase-producing multidrug-resistant P. aeruginosa infection group while no mice treated with CAZ alone survived after 2 days. Taken together, di-F-Q demonstrated unique strain-specific antimicrobial properties including anti-biofilm and antibiotic-potentiating activity against S. aureus and P. aeruginosa, respectively.


Assessment of antibacterial activity of di-F-Q. Minimum inhibitory concentration (MIC) of di-F-Q
was determined for all isolates in three replicates using broth microdilution with cation-adjusted MHB according to the Clinical and Laboratory Standards Institute (CLSI) guidelines 14 . DMSO stock solution of di-F-Q was serially diluted to the desired concentrations in MHB. Then, 10 μl of the bacterial suspension [5 × 10 5 colony forming units (CFU)/ml] was combined with 200 μl of di-F-Q in 96-well microtiter plates (1% DMSO, final). After incubation of the plate at 37 °C for 24 h, bacterial growth was visibly evaluated by monitoring the turbidity of the resulting suspension, and the MICs were determined as the lowest di-F-Q concentration that inhibited the visible growth of the bacteria.
Anti-biofilm activity. The inhibitory effect of di-F-Q on biofilm formation of various bacterial strains was determined by 96-well plate-based crystal violet assay as described previously 15 . Bacterial strains were cultured with 0.5 × TSB and diluted to 5 × 10 5 CFU/ml in fresh 0.5 × TSB with 1% glucose. In a 96-well plate, 10 μl of di-F-Q was mixed with 190-μl of the bacterial suspension to final concentrations of 0, 1, 5, 10, 25, 50, and 100 mg/l. The plate was incubated for 24 h and gentle aspiration of media was followed by fixing the biofilm by addition of 100% ethanol. Ethanol was then removed, and the plate was air-dried. The wells were treated with 0.1% crystal violet for 10 min. The plates were washed with deionized water (dH 2 O), and the crystal violet-stained biofilms were solubilized with 33% acetic acid for 1 h. Absorbance of the resulting solution at 600 nm was monitored.  Inhibition of β-lactamase. Inhibitory activity of di-F-Q against the β-lactamase activity was explored by using a β-lactamase Inhibition Screening Assay Kit (K804-100) (Biovision, San Francisco, USA) which includes nitrocefin, a β-lactamase substrate generating a colorimetric species, as a marker for β-lactamase activity. Assay was performed in accordance with the manufacturer's instructions, and changes in absorbance A 490 for nitrocefin was monitored. Clavulanic acid was used as a positive control.
Inhibition of efflux pumps. The efflux activity of CRPA was assessed by determining the accumulation of ethidium bromide (EtBr) as described previously 18 . Overnight culture of CRPA strain was inoculated in MHB. After incubation at 37 °C for 4 h, the bacterial cells were centrifuged (3,000 rpm, 15 min) and the pellet was resuspended in PBS. After adjusting the OD 600 to 0.1, the bacterial suspension (186 µl) was added to each well of the flat bottomed, black 96-well plate. EtBr (10 µl) was also added to each well to a final concentration of 2.5 µM. Finally, 4 µl of di-F-Q or a positive control (carbonyl cyanide m-chlorophenyl hydrazine, CCCP) was added to the plate and the bacteria were incubated at 37 °C. The fluorescence emission from EtBr was taken (λ ex 535 nm/ λ em 600 nm) for 30 min with a 1-min interval using Cytation 5 imaging multi-mode reader (BioTek Instruments, Inc., Winooski, VT, USA).

In vivo antibacterial activity of CAZ in combination with di-F-Q.
Animal experiments were carried out in accordance with established practices as described in the National Institutes of Health Guide for Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee of the Asan Institute for Life Sciences, Seoul, Korea. Six-week-old female outbred immunocompetent CD-1 mice were randomly distributed into 2 groups (10 mice per group), which were infected with 1.0 × 10 7 CFU/mouse of P. aeruginosa ATCC 27853 (CSPA) or IMP carbapenemase-producing MRPA admixed with 5% porcine mucin. The mice of each group were further allocated in two subgroups (n = 5 for each subgroup) and, after 2 h, the mice in each subgroup were treated with CAZ (10 mg/kg/day) or CAZ-(di-F-Q) (10 mg/kg/day-40 mg/kg/day) for 7 days.
Statistical analysis. Data are representative of at least three independent experiments and expressed as means ± SD. Tests used for nonparametric data included one-way analysis of variance (ANOVA) with Tukey's post hoc test (GraphPad, Prism 5).

Results and discussion
Antimicrobial activity of di-F-Q. The antibacterial activity of di-F-Q against various gram-positive and gram-negative bacterial strains was investigated ( were also tested for their susceptibility to di-F-Q by using the broth microdilution checkerboard (CB) method ( Table 1). The gram-positive strains were weakly sensitive to di-F-Q (MIC, 8-32 mg/L), but gram-negative strains were highly resistant (MIC, > 128 mg/L) ( Table 2). This difference may be explained by the structure and composition of bacterial cell wall. Gram-positive bacteria consist of a relatively simple envelope, and its Table 2. MICs (mg/l) of di-F-Q against various gram-positive and gram-negative bacteria. www.nature.com/scientificreports/ cytoplasmic membrane is surrounded by peptidoglycan layer. Previously, sub-MICs of quercetin and its derivatives were shown to disrupt or alter cytoplasmic membrane to exhibit antibacterial activity against S. aureus 19,20 . Based on the structural similarity, di-F-Q is presumed to have the similar mode of action as quercetin against gram-positive bacteria. On the other hand, gram-negative bacterial cell envelope is more complex and relatively impermeable structure due to their distinctive outer membrane containing lipopolysaccharide 21 . As a result, gram-negative bacteria are protected from many antibiotics, especially large and hydrophobic ones, to which gram-positive bacteria are susceptible. The lack of antibacterial activity of di-F-Q against the gram-negative bacteria might be attributed to its hydrophobic nature (cLogP = 3.78) and resulting inability to cross the bacterial cell wall. In this context, it is worth to note that Wang et al. reported that higher concentration of quercetin (50 × MIC) was required to damage the membrane of E. coli compared with S. aureus (10 × MIC) 20 .
Anti-biofilm activity of di-F-Q. The anti-biofilm activity of di-F-Q was then investigated because, by using this protective mechanism 22,23 , bacteria are known to become significantly more resistant to antibiotics 24 . The bacterial strains tested above were monitored for their biofilm-formation activity in the presence of increasing concentrations of di-F-Q, and a noticeable change was observed only in the S. aureus strains which are well characterized for pronounced ability to form biofilms 25 (Fig. 2). Treatment of the six S. aureus strains with di-F-Q resulted in a noticeable dose-dependent reduction in biofilm formation with IC 50 values of 1.8 ~ 5.3 mg/l ( Table 3). As mentioned above, quercetin and its derivatives more easily disrupt the membrane of gram-positive bacteria compared with that of gram-negative bacteria. In addition, previous studies showed that quercetin and its derivatives inhibit hemolysis and biofilm formation in S. aureus by reducing alpha-toxin secretion [26][27][28][29] , which might explain the S. aureus-specific anti-biofilm effect of di-F-Q. On the other hand, quercetin is also known to inhibit the aqr quorum-sensing system which modules biofilm formation in S. aureus 28 . However, in current investigation, anti-biofilm effect of di-F-Q was observed in both agr-functional MRSA strains (AMC-SA 5013, AMC-SA 5016) and agr-dysfunction MRSA strain (AMC-SA 3416). In addition, although quercetin exerted strong anti-biofilm effects on P. aeruginosa stain PAO1 by repressing expression levels of quorum-sensing associated genes 30 , di-F-Q did not exert anti-biofilm effect on our P. aeruginosa strains (data not shown). Taken together, it seems that inhibition of quorum-sensing system may not be essential for anti-biofilm effect of di-F-Q.  (Table 4). In general, the antibacterial activity of the antibiotics was not significantly affected by combination with di-F-Q. To our surprise, however, MICs of CAZ against the gram-negative carbapenem-resistant strains such as CRPA and CRAB showed significant decreases (eight-and four-fold, respectively) upon combination with di-F-Q (MIC/4, Table 4). The remarkable effect of di-F-Q on the selective restoration of the antibacterial activity of CAZ against carbapenem-resistant gram-negative strains such as CRPA and CRAB is of critical interest because carbapenems are used as last resorts for the treatment of infections caused by multidrug-resistant gram-negative bacteria 31 . Thus, the antimicrobial activity of CAZ alone and in combination with di-F-Q was further assessed against 50 clinical isolates of P. aeruginosa using broth microdilution method. The MIC distributions of CAZ against P. aeruginosa isolates were obtained in the absence and presence of di-F-Q, which showed significant shifts to lower values upon combination with di-F-Q (Fig. 3). The ceftazidime-potentiating activity of di-F-Q and thereby sensitization of the P. aeruginosa isolates to CAZ was found to be concentration-dependent, and the shifts in MICs were more pronounced with increasing concentrations of di-F-Q (Fig. 3).

Antibiotics-potentiating activity of di-F-Q.
Based on the Clinical and Laboratory Standards Institute (CLSI) susceptibility interpretive criteria (breakpoints) for CAZ against P. aeruginosa 14 Table 4. Also included in Table 4 are the MIC 50 and the MIC 90 values, the lowest concentration of CAZ at which 50% and 90% of the isolates were inhibited, respectively. CAZ-(di-F-Q) activity (MIC 50/90 , 2/8 mg/l; 94.1% susceptible at 8 mg/l) against all 50 P. aeruginosa isolates (with the di-F-Q concentration fixed at 32 mg/l) was noticeably enhanced in comparison to the CAZ single treatment (MIC 50/90 , 4/ > 32 mg/l; 78.4% susceptible at 8 mg/l), which amounts to 15.7% increase in the percentages of susceptible P. aeruginosa isolates upon addition of di-F-Q to CAZ ( Table 5).
The significantly increased sensitization of P. aeruginosa isolates to CAZ by di-F-Q is reminiscent of CAZ-avibactam (AVI) combination (MIC 50/90 , 2/8 mg/l; 96.3% susceptible at 8 mg/l) which showed 7.4 ~ 16.3% increases in the percentages of the susceptible P. aeruginosa isolates compared with those of CAZ single treatment 32,33 . As the CAZ-potentiation activity of avibactam is attributed to its β-lactamase inhibitory activity 33 , we evaluated the effect of di-F-Q on the enzymatic activity of β-lactamase. However, the β-lactamase activity was not affected by treatment with di-F-Q (Fig. 4).
On the other hand, overexpression of efflux pumps is generally considered as the major mechanism for intrinsic and acquired antibiotic resistance in P. aeruginosa 34 ; a diverse array of antibiotics including β-lactams, quinolones, tetracyclines, macrolides, linomycins chloramphenicols and novobiocins are known to suffer from efflux pump-mediated resistance by P. aeruginosa 35 . Thus, efflux pump inhibitory activity of di-F-Q was investigated by using ethidium bromide (EtBr) as the fluorescent substrate of efflux pumps in P. aeruginosa (Fig. 5) 36   www.nature.com/scientificreports/ and thereby intracellular retention of EtBr; significant efflux pump inhibitory activity of di-F-Q was observed at concentrations 32 mg/l or above, which is in line with the eightfold reduction of the MIC of CAZ against CRPA by the same amount of di-F-Q (Table 3). Carbonyl cyanide m-chlorophenyl hydrazine (CCCP) is known to inhibit the efflux pump in P. aeruginosa 36 , and we observed that treatment of CRPA with CCCP (16 mg/l) increased fluorescence from EtBr (Fig. 5). These results collectively suggest that di-F-Q (32 mg/l) and CCCP (16 mg/l) are almost equally effective in inhibiting antibiotic efflux by P. aeruginosa. The MexAB-OprM multidrug efflux pump system is known to pump out mostly lipophilic and amphiphilic drugs, which include EtBr. Therapeutic efficacy of CAZ-(di-F-Q) combination against P. aeruginosa was then evaluated in mice infected with CSPA or multidrug resistant P. aeruginosa (MRPA) (Fig. 6). Thus, six-week-old female outbred immunocompetent CD-1 mice were infected with 1.0 × 10 7 CFU/mouse of P. aeruginosa ATCC 27853 (CSPA) (Fig. 6a) or MRPA (Fig. 6b) admixed with 5% porcine mucin and, after 2 h, treated with CAZ (10 mg/kg/day) or CAZ-(di-F-Q) (10-40 mg/kg/day) for 7 days (n = 5 for each treatment group). The CSPA-or MRPA-infected mice without antibiotic treatment all died after 1 day (infected control, Fig. 6). Also, in both CSPA and MRPA infection   www.nature.com/scientificreports/  www.nature.com/scientificreports/ groups, no mice treated with CAZ alone survived after 2 days (CAZ, Fig. 6). However, 20% of the mice treated with CAZ-(di-F-Q) survived after 7 days in CSPA as well as MRPA infection group [CAZ + (di-F-Q), Fig. 6]. Interestingly, between 3 to 5 days after infection, CAZ-(di-F-Q) was more effective in protecting MRPA-infected mice (60 ~ 40% survival rate, Fig. 6b) than the CSPA-infected ones (20% survival rate, Fig. 6a).
In summary, we evaluated antibacterial properties of di-F-Q, a fluorinated quercetin analogue. Di-F-Q was not effective in killing bacteria but showed potent anti-biofilm activity against S. aureus with IC 50 values of 1.8 ~ 5.3 mg/l. Also, di-F-Q potentiated the antibacterial activity of the 3rd generation cephalosporin CAZ against the gram-negative carbapenem-resistant strains such as CRPA and CRAB (eight-and four-fold decreases in MICs, respectively). Assessment of the antimicrobial activity of CAZ in combination with di-F-Q against 50 clinical isolates of P. aeruginosa confirmed a remarkable enhancement in antibacterial activity (MIC 50/90 , 2/8 mg/l; 94.1% susceptible at 8 mg/l) in comparison to the CAZ single treatment (MIC 50/90 , 4/ > 32 mg/l; 78.4% susceptible at 8 mg/l), which amounts to 15.7% increase in the percentages of susceptible P. aeruginosa isolates upon addition of di-F-Q to CAZ. Further mechanistic studies revealed that di-F-Q affected the antibiotics efflux system in CRPA but not the β-lactamase activity. Thus, di-F-Q was almost equally effective as CCCP in inhibiting antibiotic efflux by P. aeruginosa. In vivo evaluation of the therapeutic efficacy of CAZ-(di-F-Q) combination against P. aeruginosa showed 20% of the mice treated with CAZ-(di-F-Q) survived after 7 days in MRPA infection group while no mice treated with CAZ alone survived after 2 days. Taken together, di-F-Q demonstrated the peculiar strain-specific anti-biofilm and antibiotic-potentiating activity against S. aureus and P. aeruginosa, respectively.