Giving Drugs a Second Chance: Antibacterial and Antibiofilm Effects of Ciclopirox and Ribavirin against Cystic Fibrosis Pseudomonas aeruginosa Strains

Drug repurposing is an attractive strategy for developing new antibacterial molecules. Herein, we evaluated the in vitro antibacterial, antibiofilm, and antivirulence activities of eight FDA-approved “non-antibiotic” drugs, comparatively to tobramycin, against selected Pseudomonas aeruginosa strains from cystic fibrosis patients. MIC and MBC values were measured by broth microdilution method. Time–kill kinetics was studied by the macro dilution method, and synergy studies were performed by checkerboard microdilution assay. The activity against preformed biofilms was measured by crystal violet and viable cell count assays. The effects on gene expression were studied by real-time quantitative PCR, while the cytotoxic potential was evaluated against IB3-1 bronchial CF cells. Ciclopirox, 5-fluorouracil, and actinomycin D showed the best activity against P. aeruginosa planktonic cells and therefore underwent further evaluation. Time–kill assays indicated actinomycin D and ciclopirox, contrarily to 5-fluorouracil and tobramycin, have the potential for bacterial eradication, although with strain-dependent efficacy. Ciclopirox was the most effective against the viability of the preformed biofilm. A similar activity was observed for other drugs, although they stimulate extracellular polymeric substance production. Ribavirin showed a specific antibiofilm effect, not dependent on bacterial killing. Exposure to drugs and tobramycin generally caused hyperexpression of the virulence traits tested, except for actinomycin D, which downregulated the expression of alkaline protease and alginate polymerization. Ciclopirox and actinomycin D revealed high cytotoxic potential. Ciclopirox and ribavirin might provide chemical scaffolds for anti-P. aeruginosa drugs. Further studies are warranted to decrease ciclopirox cytotoxicity and evaluate the in vivo protective effects.


Introduction
In cystic fibrosis (CF) patients, the mutation of the CF transmembrane conductance regulator gene leads to an accumulation of dry and sticky airway secretions, creating the perfect environment for the onset of bacterial pulmonary infections [1]. The altered microenvironment of the CF lung counteracts the inflammation in clearing the infection, thus causing the progression of pulmonary disease towards bronchiectasis and finally death [2].
Pseudomonas aeruginosa is the predominant pathogen, especially in adulthood [3]. Although the aerosolization of tobramycin into CF patients' airways improves outcomes, the lungs of CF patients, even those receiving antibiotic therapy, are persistently colonized by P. aeruginosa [3,4]. Indeed, this microorganism cannot be eradicated because of its ability to grow as biofilm in the CF airway, a functional consortium of sessile bacteria enclosed in an extracellular matrix, making them significantly more tolerant to antimicrobials than with planktonic counterparts [5][6][7]. In addition, another P. aeruginosa adaptative reply to the airway of CF patients is the conversion to a mucoid phenotype due to an overproduction of alginate, leading to the generation of a thicker extracellular polysaccharide matrix [3,8].
As a result, the decreasing number of effective antibiotics raises the urgent need to develop new molecules that possibly target cells within the biofilm and avoid selecting resistant strains. However, the traditional drug discovery process is costly and lengthy, requiring years of experimentation followed by extensive clinical trials [9].
The "drug repurposing" approach recently proposed to reduce the drug discovery time frame is an attractive alternative. This strategy is based on using known and approved drugs for a medical indication other than the one for which it was developed [10]. It can also include using drugs that have reached phase II or III of clinical trials that demonstrate no efficacy for a particular indication but have shown good safety [10]. Since their toxicity and pharmacokinetics have already been studied, the "repurposed" drugs can bypass some clinical trials, save time, and reduce costs [11]. This strategy is a promising tool in treating bacterial infections, as many molecules have secondary mechanisms of action that allow them to be effective against many pathogens. Several non-antimicrobial drugs have demonstrated antibiotic activity [12], but none are currently used in antibacterial therapy.
In this frame, the main aim of the present study was to assess the in vitro activity of eight FDA-approved "non-antibiotic" drugs-namely ribavirin (antiviral), toremifene (nonsteroidal antiestrogen), oxyclozanide (anthelmintic), meloxicam (nonsteroidal antiinflammatory), 5-fluorouracil (antineoplastic), actinomycin D (antineoplastic), furosemide (diuretic), and ciclopirox (antifungal)-against a set of selected P. aeruginosa strains isolated from CF patients. These in-use pharmacological agents, encompassing a wide variety of different chemical structures and mechanisms of action, were selected because they were previously found to have some direct antibacterial activity ( Table 1). The antibacterial, antibiofilm, and antivirulence potential of the non-antibiotic drugs were evaluated compared to tobramycin. Table 1. FDA-approved "non-antibiotic" drugs tested. We selected some in-use pharmaceuticals found to have some direct antimicrobial effects in previously published studies.

Drug
Therapeutic

Selection of P. aeruginosa Strains
First, a collection of 19 P. aeruginosa strains from CF patients was screened for the ability to form biofilm on polystyrene using a microtiter plate crystal violet assay, and the results are shown in Figure 1.

Selection of P. aeruginosa Strains
First, a collection of 19 P. aeruginosa strains from CF patients was sc ability to form biofilm on polystyrene using a microtiter plate crystal violet results are shown in Figure 1. Biofilm formation by 19 P. aeruginosa strains from CF patients. The am formed on polystyrene following 24 h of incubation at 37 °C was measured by the crystal violet method. Each strain was tested, in quadruplicate, on four different o (A) Results were subtracted by negative control (OD492 = 0.092) and shown as a scat horizontal solid line indicating the mean OD value. The horizontal dotted black lin off value for biofilm formation (ODc = mean + 3 x standard deviation of negative co = 0.143), whereas the blue one indicates the cut-off value for high biofilm-forme 0.572). (B) According to Stepanović et al. [21], each strain was assigned to one groups: non-biofilm-former (OD492 < 0.143), weak biofilm-former (0.143 < OD492 ≤ biofilm-former (0.286 < OD492 ≤ 0.572), and high biofilm-former (OD492> 0.572). Resu the percentage of distribution of each group. Statistical significance at Fisher's exac *** p < 0.001, **** p < 0.0001.
Most strains (14 out of 19, 73.7%) were able to form a biofilm, althoug cant differences for biofilm biomass, measured as optical density at 4 frequence (%) Figure 1. Biofilm formation by 19 P. aeruginosa strains from CF patients. The amount of biofilm formed on polystyrene following 24 h of incubation at 37 • C was measured by the microtiter plate crystal violet method. Each strain was tested, in quadruplicate, on four different occasions (n = 16). (A) Results were subtracted by negative control (OD 492 = 0.092) and shown as a scatter plot, with the horizontal solid line indicating the mean OD value. The horizontal dotted black line shows the cut-off value for biofilm formation (ODc = mean + 3 x standard deviation of negative control wells; ODc = 0.143), whereas the blue one indicates the cut-off value for high biofilm-former class (OD 492 > 0.572). (B) According to Stepanović et al. [21], each strain was assigned to one of the following groups: non-biofilm-former (OD 492 < 0.143), weak biofilm-former (0.143 < OD 492 ≤ 0.286), moderate biofilm-former (0.286 < OD 492 ≤ 0.572), and high biofilm-former (OD 492 > 0.572). Results are shown as the percentage of distribution of each group. Statistical significance at Fisher's exact test: ** p < 0.01, *** p < 0.001, **** p < 0.0001.
Next, the susceptibility of 19 P. aeruginosa strains to eight anti-pseudomonal antibiotics was measured using a disk diffusion agar test, and the results are summarized in Table 2. Amikacin and tobramycin were the most active antibiotics showing a susceptibility rate of 63.1%, followed by colistin (52.6%), netilmicin (31.6%), and ceftazidime (5.2%). Contrarily, ticarcillin and piperacillin/tazobactam showed no activity against the strains tested. According to Magiorakos et al. [22], all P. aeruginosa strains but two (Pa14 and Pa16) were classified as multi-drug resistant (MDR).

Killing Kinetics
The bactericidal or bacteriostatic properties of ciclopirox, 5-fluorouracil, and actinomycin were investigated, compared to tobramycin, by killing kinetics. We choose P. aeruginosa PaPh32 ( Figure 2) and Pa7 ( Figure 3) as representative of tobramycin-resistant and -susceptible strains, respectively.
An evident dose-dependent activity was observed only in the case of actinomycin D, although with different effects depending on the strain tested. It always had a bacteriostatic result towards the Pa7 strain, whereas it yielded a bactericidal effect-i.e., greater than 3 Log-fold decrease in CFUs, equivalent to 99.9% killing of the inoculum, was observedtowards the PaPh32 strain when tested at 1x (after 11 h of exposure) and 2xMIC (after 8 h of exposure). Actinomycin D caused a viability decrease of PaPh32 below the limit of detection (LOD) when tested at 1x (after 14 h of incubation) and 2xMIC (after 11 h of incubation), while a regrowth was found at 1xMIC. A similar trend was observed for ciclopirox that exerted bactericidal activity only when tested towards the PaPh32 strain after 7 h exposure at 4xMIC. The worst activity was shown by 5-fluorouracil, yielding a bacteriostatic effect over 24 h, regardless of the strain tested. By comparison, tobramycin had a bactericidal result against both strains, although to different extents. When tested towards PaPh32, it was bactericidal only at 4x and 8xMIC, although a rapid regrowth was observed in both cases. Contrarily, the effect was more rapid against Pa7, causing a 3 Log reduction within 5 h of incubation at 4x and 8xMIC and after 16 h of incubation at 1xMIC. The cell viability decreased below LOD at concentrations of 2x, 4x, and 8xMIC, although 2xMIC allowed Pa7 regrowth. Overall, the findings from time-kill assays showed that actinomycin D and ciclopirox have the potential for bacterial eradication, although their activity is strain-dependent. Contrarily, 5-fluorouracil showed a bacteriostatic effect regardless of the strain tested.

Killing Kinetics
The bactericidal or bacteriostatic properties of ciclopirox, 5-fluorouracil, and actinomycin were investigated, compared to tobramycin, by killing kinetics. We choose P. aeruginosa PaPh32 ( Figure 2) and Pa7 ( Figure 3) as representative of tobramycin-resistant and -susceptible strains, respectively. multiples, compatibly with the drugs' solubility. The dotted line indicates bactericidal activity, defined as a ≥ 3 Log (CFU/mL) reduction of the initial inoculum size. The limit of detection was 10 CFU/mL. An evident dose-dependent activity was observed only in the case of actinomycin D, although with different effects depending on the strain tested. It always had a bacteriostatic result towards the Pa7 strain, whereas it yielded a bactericidal effect-i.e., greater than 3 Log-fold decrease in CFUs, equivalent to 99.9% killing of the inoculum, was observed-towards the PaPh32 strain when tested at 1x (after 11 h of exposure) and 2xMIC (after 8 h of exposure). Actinomycin D caused a viability decrease of PaPh32 below the limit of detection (LOD) when tested at 1x (after 14 h of incubation) and 2xMIC (after 11 h of incubation), while a regrowth was found at 1xMIC. A similar trend was observed for ciclopirox that exerted bactericidal activity only when tested towards the PaPh32 strain after 7 h exposure at 4xMIC. The worst activity was shown by 5-fluorouracil, yielding a

Synergy Tests
The activity of tobramycin in combination with actinomycin D, ciclopirox, and 5-fluorouracil was evaluated using a checkerboard assay. P. aeruginosa Pa7 and PaPh32 strains were selected because they are representative of tobramycin-susceptible and -resistant phenotypes, respectively (Table 4).
FICi values indicated additivity, regardless of strain and combination, with the best value of 0.56 for tobramycin + ciclopirox combination against Pa7 strain. Table 4. Activity of tobramycin in combination with other "non-antibiotic" drugs. The fractional inhibitory concentration index (FICi) was calculated as follows, using the checkerboard assay: FIC A + FIC B , where FIC A = MIC of drug A in combination/MIC of drug A alone, and FIC B = MIC of drug B in combination/MIC of drug B alone. The best FICi value and the range of FICi values were reported for each drug combination. All FICi values obtained indicated an additive effect (0.5 < FICi ≤ 4) [25].

Drug Combinations
Best FICi Value (Range) for P. aeruginosa Strain:

In Vitro Activity against Preformed Biofilms
Ciclopirox, 5-fluorouracil, and actinomycin D were tested, comparatively to tobramycin, for dispersal and killing activities against 24 h mature biofilms by P. aeruginosa PaPh32 ( Figure 4) and Pa7 ( Figure 5).  . Dispersal activity against preformed biofilms by P. aeruginosa PaPh32. The efficacy of actinomycin D, ciclopirox, and 5-fluorouracil to disperse 24 h mature biofilms by P. aeruginosa PaPh32 was assessed, comparatively to tobramycin, using crystal violet assay. P. aeruginosa PaPh32 was chosen as representative of tobramycin-resistant strains. Each drug was tested at 0.5x and multiples of MIC value. Results were expressed as mean + SD of the residual biofilm biomass (OD492) after 24 h of exposure. Control samples (CTRL) were not exposed to the drug. Statistical significance at ordinary one-way ANOVA + Holm-Sidak's multiple comparisons post-test: * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001 vs. CTRL. . Dispersal activity against preformed biofilms by P. aeruginosa PaPh32. The efficacy of actinomycin D, ciclopirox, and 5-fluorouracil to disperse 24 h mature biofilms by P. aeruginosa PaPh32 was assessed, comparatively to tobramycin, using crystal violet assay. P. aeruginosa PaPh32 was chosen as representative of tobramycin-resistant strains. Each drug was tested at 0.5x and multiples of MIC value. Results were expressed as mean + SD of the residual biofilm biomass (OD 492 ) after 24 h of exposure. Control samples (CTRL) were not exposed to the drug. Statistical significance at ordinary one-way ANOVA + Holm-Sidak's multiple comparisons post-test: * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001 vs. CTRL.

Figure 5.
Dispersal activity against preformed biofilms by P. aeruginosa Pa7. The efficacy of actinomycin D, ciclopirox, and 5-fluorouracil to disperse 24 h mature P. aeruginosa Pa7 biofilms was assessed, comparatively to tobramycin, using crystal violet assay. P. aeruginosa Pa7 was chosen as representative of tobramycin-susceptible strains. Each drug was tested at 0.5x and multiples of MIC value. Results were expressed as mean + SD of the residual biofilm biomass (OD492) after 24 h of exposure. Control samples (CTRL) were not exposed to the drug. Statistical significance at ordinary one-way ANOVA + Holm-Sidak's multiple comparisons post-test: * p < 0.05, ** p < 0.01, **** p < 0.0001 vs. CTRL.
The activity of each drug against mature biofilm was not dependent on the strain tested. Ciclopirox was the most active, causing a significant reduction of biofilm biomass, regardless of concentration and strain tested. Conversely, the exposure to 5-fluorouracil and actinomycin D never caused biofilm reduction but even stimulated its biomass, although at different extents depending on strain and concentration, particularly in the case of 5-fluorouracil. Tobramycin significantly reduced biofilm biomass formed by both strains only when tested at the maximum concentration of 8xMIC, whereas at 0.5xMIC it constantly stimulated biomass formation.
Since the antibiofilm effect could be specific and not related to antibacterial activity, the dispersal activity was also evaluated for drugs, which were not active against planktonic P. aeruginosa cells ( Figure 6). Ribavirin was the only drug able to significantly decrease biofilm biomass, effective against both strains tested, although to different extents Figure 5. Dispersal activity against preformed biofilms by P. aeruginosa Pa7. The efficacy of actinomycin D, ciclopirox, and 5-fluorouracil to disperse 24 h mature P. aeruginosa Pa7 biofilms was assessed, comparatively to tobramycin, using crystal violet assay. P. aeruginosa Pa7 was chosen as representative of tobramycin-susceptible strains. Each drug was tested at 0.5x and multiples of MIC value. Results were expressed as mean + SD of the residual biofilm biomass (OD 492 ) after 24 h of exposure. Control samples (CTRL) were not exposed to the drug. Statistical significance at ordinary one-way ANOVA + Holm-Sidak's multiple comparisons post-test: * p < 0.05, ** p < 0.01, **** p < 0.0001 vs. CTRL.
The activity of each drug against mature biofilm was not dependent on the strain tested. Ciclopirox was the most active, causing a significant reduction of biofilm biomass, regardless of concentration and strain tested. Conversely, the exposure to 5-fluorouracil and actinomycin D never caused biofilm reduction but even stimulated its biomass, although at different extents depending on strain and concentration, particularly in the case of 5-fluorouracil. Tobramycin significantly reduced biofilm biomass formed by both strains only when tested at the maximum concentration of 8xMIC, whereas at 0.5xMIC it constantly stimulated biomass formation.
Since the antibiofilm effect could be specific and not related to antibacterial activity, the dispersal activity was also evaluated for drugs, which were not active against planktonic P. aeruginosa cells ( Figure 6). Ribavirin was the only drug able to significantly decrease biofilm biomass, effective against both strains tested, although to different extents (biomass removal: 53.5% and 35.1%, respectively, for Pa7 and PaPh32; p < 0.001). Conversely, exposure to oxyclozanide significantly improved biofilm formation by about 100%, although only in the case of the PaPh32 strain.
(biomass removal: 53.5% and 35.1%, respectively, for Pa7 and PaPh32; p < 0.001). C versely, exposure to oxyclozanide significantly improved biofilm formation by a 100%, although only in the case of the PaPh32 strain. Figure 6. Biofilm dispersal activity of drugs not active against planktonic cells. The efficacy of d in disrupting 24 h mature biofilms was assessed using a crystal violet assay. Each drug was te at the maximum concentration tested in MIC assays: 1.024 µg/mL for all drugs except for Torem (256 µg/mL). Results are expressed as mean + SD of the residual biofilm biomass (OD492) after of exposure. Control samples (CTRL) were not exposed to the drug. Statistical significance at nary one-way ANOVA + Holm-Sidak's multiple comparisons post-test: * p < 0.05, **** p < 0.000 CTRL.
Next, the activity of each drug against the viability of preformed biofilms was te after exposure to MIC and its multiples, and the results are summarized in Figures 7 8. All drugs caused a significant reduction of biofilm viability, regardless of strain concentration tested, although at different levels. Ciclopirox was the most active, show a dose-dependent activity regardless of the strain considered (viability reduction; PaPh32 biofilm biomass (OD 492 ) Figure 6. Biofilm dispersal activity of drugs not active against planktonic cells. The efficacy of drugs in disrupting 24 h mature biofilms was assessed using a crystal violet assay. Each drug was tested at the maximum concentration tested in MIC assays: 1.024 µg/mL for all drugs except for Toremifene (256 µg/mL). Results are expressed as mean + SD of the residual biofilm biomass (OD 492 ) after 24 h of exposure. Control samples (CTRL) were not exposed to the drug. Statistical significance at ordinary one-way ANOVA + Holm-Sidak's multiple comparisons post-test: * p < 0.05, **** p < 0.0001 vs. CTRL.
Next, the activity of each drug against the viability of preformed biofilms was tested after exposure to MIC and its multiples, and the results are summarized in Figures 7 and 8. All drugs caused a significant reduction of biofilm viability, regardless of strain and concentration tested, although at different levels. Ciclopirox was the most active, showing a dose-dependent activity regardless of the strain considered (viability reduction; Pa7: 98.7% at 2xMIC, and 99.8% at 4xMIC; PaPh32: 97.5% at 2xMIC, and 99.9% at 4xMIC). Tobramycin exhibited comparable activity to ciclopirox, causing a significant, concentration-dependent reduction ranging from 99.7% (2xMIC) to >99.9% (8xMIC) for Pa7 and from 88.5% (2xMIC) to >99.99% (8xMIC) for PaPh32.
Our findings indicated ciclopirox as the most effective against mature biofilms, active on biofilm biomass-consisting of extracellular polymeric substance (EPS) and cells-and viability. Although active on biofilm viability, the other drugs probably stimulate EPS production, as suggested by increased biofilm biomass after exposure. Of interest is the specific activity exhibited by ribavirin, which is not dependent on bacterial killing. Our findings indicated ciclopirox as the most effective against mature biofilms, activ on biofilm biomass-consisting of extracellular polymeric substance (EPS) and cells-an viability. Although active on biofilm viability, the other drugs probably stimulate EP production, as suggested by increased biofilm biomass after exposure. Of interest is th specific activity exhibited by ribavirin, which is not dependent on bacterial killing.

Effect on P. aeruginosa Virulence Genes Expression
The effect of 20 h of exposure to actinomycin D, 5-fluorouracil, ciclopirox, and ribav rin at 1/4xMIC on the expression of selected virulence genes of P. aeruginosa PaPh32 wa evaluated, comparatively to tobramycin, by real-time RT-qPCR, and the results are show in Figure 9. Killing activity against preformed biofilms by P. aeruginosa Pa7. The efficacy of actinomycin D, ciclopirox, and 5-fluorouracil on the viability of 24 h mature biofilms by P. aeruginosa Pa7 was assessed, comparatively to tobramycin, using cell viable count assay. Each drug was tested at multiples of MIC value. Results are expressed as mean + SD of the residual biofilm viability [Log (CFU/well)] after 24 h of exposure. Control samples (CTRL) were not exposed to the drug. Statistical significance at ordinary one-way ANOVA + Holm-Sidak's multiple comparisons post-test: ** p < 0.01, *** p < 0.001, **** p < 0.0001 vs. CTRL.

Effect on P. aeruginosa Virulence Genes Expression
The effect of 20 h of exposure to actinomycin D, 5-fluorouracil, ciclopirox, and ribavirin at 1/4xMIC on the expression of selected virulence genes of P. aeruginosa PaPh32 was evaluated, comparatively to tobramycin, by real-time RT-qPCR, and the results are shown in Figure 9. Figure 9. Effect of drugs exposure on the expression of selected P. aeruginosa virulence genes. The effects of exposure to actinomycin D, 5-fluorour and ribavirin on the expression levels of P. aeruginosa PaPh32 virulence genes aprA (alkaline protease), lasI (quorum sensing), mexA, mexB, an pumps), toxA (exotoxin A), and algD (alginate) were assessed, comparatively to tobramycin, by real-time RT-qPCR assay. Each drug was tested at h, while controls (CTRL) were not exposed to the drug. The relative expression of each gene was normalized on the housekeeping proC gene. Res as means + SDs (n = 6) of fold change (FC: 2 −ΔΔct ) on a log2 scale. Statistical significance at ANOVA followed by Tukey's multiple comparisons post ** p < 0.01, *** p < 0.001, and **** p < 0.0001 vs. CTRL. Figure 9. Effect of drugs exposure on the expression of selected P. aeruginosa virulence genes. The effects of exposure to actinomycin D, 5-fluorouracil, ciclopirox, and ribavirin on the expression levels of P. aeruginosa PaPh32 virulence genes aprA (alkaline protease), lasI (quorum sensing), mexA, mexB, and mexC (efflux pumps), toxA (exotoxin A), and algD (alginate) were assessed, comparatively to tobramycin, by real-time RT-qPCR assay. Each drug was tested at 1/4xMIC for 20 h, while controls (CTRL) were not exposed to the drug. The relative expression of each gene was normalized on the housekeeping proC gene. Results are shown as means + SDs (n = 6) of fold change (FC: 2 −∆∆ct ) on a log 2 scale. Statistical significance at ANOVA followed by Tukey's multiple comparisons post-test: * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001 vs. CTRL.

Gene expression (log 2 FC)
The gene expression pattern was dependent on the drug and gene considered. The pattern observed after exposure to actinomycin D and 5-fluorouracil was nearly the same. Both drugs downregulated aprA, codifying for the alkaline protease (fold change: −5.53 for both; p < 0.001 vs. unexposed control), whereas they upregulated the expression of the QS mediator lasI (fold change: 5.07 and 8.19, respectively; p < 0.001) and all efflux pump-related genes mexA (fold change: 4.77 and 4.22, respectively; p < 0.001), mexB (fold change: 3.57 and 2.66, respectively; p < 0.001), and mexC (fold change: 20.95 and 115.04, respectively; p < 0.001). In addition, actinomycin D decreased the expression of algD (fold change: −1.38; p < 0.05), codifying for GDP-mannose 6-dehydrogenase and involved in the alginate polymerization. The exposure to ciclopirox caused the worst expression pattern, namely the hyperexpression of all genes but mexC. Finally, ribavirin significantly increased mexA, mexC, and toxA genes.
The cytotoxic potential of actinomycin D, 5-fluorouracil, ciclopirox, and ribavirin was investigated, in comparison with tobramycin, using a cell-based MTS assay. Each drug was tested at the maximum concentration active against planktonic and/or biofilm P. aeruginosa cells (Figure 10).  MTS tetrazolium-based colorimetric assay showed 5-fluorouracil, ribavirin, and tobramycin were not toxic for IB3-1 cells, allowing a cell growth comparable to untreated control cells ( Figure 10A). On the contrary, a significant (p < 0.0001 vs. CTRL) cytotoxic MTS tetrazolium-based colorimetric assay showed 5-fluorouracil, ribavirin, and tobramycin were not toxic for IB3-1 cells, allowing a cell growth comparable to untreated control cells ( Figure 10A). On the contrary, a significant (p < 0.0001 vs. CTRL) cytotoxic effect was observed for ciclopirox and actinomycin D, although to a different extent. Indeed, when tested at lower concentrations, ciclopirox was revealed to be more toxic, causing a higher reduction of IB3-1 cells survival (ranging from 47.7% at 1/4xMIC to 99% at 1/2xMIC) ( Figure 10B) compared with actinomycin D (ranging from 38% at 1xMIC to 56.7% at 2xMIC) ( Figure 10C).

Discussion
Repurposing or repositioning FDA-approved pharmacotherapies for off-label use has recently supplied alternative approaches to identifying new classes of antibiotics and scaffolds to combat infections caused by MDR pathogens [26]. Herein, we screened eight FDA-approved "non-antibiotic" drugs for their potential use in treating lung infections caused by P. aeruginosa in CF patients. With this aim, we tested six P. aeruginosa CF strains selected because of MDR and high-biofilm production. First, we evaluated the activity of each drug against P. aeruginosa planktonic cells. Ciclopirox, followed by 5-fluorouracil, and actinomycin D were the only ones showing antibacterial activity.
A literature survey revealed that although these compounds had reported bioactivities, no study was focused on the antibacterial activity against P. aeruginosa in CF patients. Ciclopirox is an off-patent, broad-spectrum antifungal agent used, as an olamine salt, in various formulations to treat superficial fungal infections [27]. It has recently been found to have considerable potential to act against both Gram-positive and Gram-negative bacterial pathogens [20,28]. The synthetic fluorinated pyrimidine 5-fluorocytosine is used as an antimycotic drug with the brand name Ancobon. It has successfully been used to treat fungal infections in CF patients, including a case of pulmonary candidiasis, without causing side effects [29]. Actinomycin D, an antitumor antibiotic that inhibits transcription, is one of the oldest chemotherapy drugs used to treat various types of cancer, such as Wilms tumor, rhabdomyosarcoma, Ewing's sarcoma, trophoblastic neoplasm, testicular cancer, and certain types of ovarian cancer [30].
Our results showed that ciclopirox has antibacterial activity against P. aeruginosa CF strains, with MIC values ranging from 128 to 512 µg/mL. Our findings are consistent with previous studies focused on Acinetobacter baumannii, Escherichia coli, and Klebsiella pneumoniae clinical isolates, although the effect was more potent as indicated by an MIC range of 5-15 µg/mL [20]. Despite the multiple potential uses of ciclopirox, very little is known about its antibacterial mechanism, thus warranting further studies. In this frame, Conley et al. observed that ciclopirox acts against E. coli interfering with galactose metabolism and blocking LPS synthesis [28]. E. coli and K. pneumoniae preferentially use glucose, metabolizing it through the Embden-Meyerhof-Parnas pathway. In contrast, Pseudomonas species use glucose as a secondary carbon source and catabolize it using the Entner-Doudoroff pathway [28]. This differential use of glucose as a carbon source might explain why we observed P. aeruginosa has higher ciclopirox MICs than E. coli and K. pneumoniae. Ribavirin, oxyclozanide, meloxicam, furosemide, and toremifene did not show any activity at the tested concentrations, although higher than those previously revealed as effective [13,14,16,17].
Tobramycin, along with colistin and aztreonam, is the inhaled antibiotic currently used for treating P. aeruginosa CF lung infections [31]. However, when antibiotic therapy is administered continuously and for long periods, it can select for resistant P. aeruginosa strains [32]. The use of bactericidal rather than bacteriostatic agents as first-line therapy is therefore recommended because the eradication of microorganisms serves to curtail, although not avoid, the development of bacterial resistance. The results we obtained from time-kill analyses indicated that actinomycin D and ciclopirox exhibit a strain-dependent bactericidal activity. Indeed, both drugs were bacteriostatic against the Pa7 strain but bactericidal against the tobramycin-resistant PaPh32 strain, even until causing a decrease in the viability under the limit of detection without regrowth over monitored time. Conversely, 5-fluorouracil and tobramycin were bacteriostatic and allowed for regrowth, regardless of the strain tested.
Current CF treatment regimens involve not only aggressive use of high-dose antibiotics but also combination therapy. However, the limited number of available antibiotics makes it difficult to know what combination would be most effective in any clinical situation. In this frame, several non-antibiotic compounds were reported to synergize with antibiotics, offering a new direction for fighting emergent drug-resistant pathogens such as P. aeruginosa. In this regard, Ejim et al. [33] found that benserazide (a DOPA decarboxylase inhibitor to treat Parkinson's disease) and loperamide (an opioid receptor agonist used to treat diarrhea) restore minocycline susceptibility in MDR P. aeruginosa strains. In another study, the antipsychotic agents levomepromazine and chlorpromazine exhibited synergy with polymyxin B against P. aeruginosa, providing potential chemical scaffolds for further drug development [34]. In the present study, we tested the activity of tobramycin combined with ciclopirox, actinomycin D, and 5-fluorouracil against both tobramycin-resistant and -susceptible P. aeruginosa CF strains. Unfortunately, no synergistic effect was found. Each combination indeed resulted in an additive effect, regardless of tobramycin-resistant and -susceptible strains.
The treatment of pulmonary bacterial infections in CF patients is significantly affected by adaptative strategies that allow pathogens to survive despite repeated, broad-spectrum courses of antibiotics [35]. Among these strategies, biofilm formation provides bacterial communities with both physical protection and reservoirs of phenotypically distinct subpopulations that withstand antimicrobials and immune responses [35]. Though existing antibiotic therapies have improved CF patients' lung function, survival, and quality of life, the biofilm lifestyle represents a barrier limiting benefits. In the present study, the antibiofilm potential of all eight drugs was evaluated, for the first time, against 24 h old preformed, mature P. aeruginosa biofilms. Scientific literature is inconsistent and fragmented in this regard. Previous studies reported the efficacy of actinomycin D in preventing biofilm formation by S. epidermidis and methicillin-resistant S. aureus strains [36,37]. Conversely, ciclopirox was previously shown to be not active on biofilm formation by MDR A. baumannii, E. coli, and K. pneumoniae clinical isolates [28]. Toremifene prevented biofilm formation and eradicated preformed biofilms by the oral bacteria Porphyromonas gingivalis and Streptococcus mutans [15]. No study was focused on 5-fluorouracil, ribavirin, furosemide, and oxyclozanide. Overall, our findings indicated ciclopirox as the most effective, in some cases even more active than tobramycin. Ciclopirox always caused a significant dispersion of biofilm biomass, consisting of EPS and cells, along with a reduction of biofilm viability nearly to eradication. Similarly, actinomycin D and 5-fluorouracil reduced biofilm viability, although to a lesser extent; furthermore, they stimulated EPS production as suggested by increased biofilm biomass after exposure. The same trend was observed for tobramycin when tested at sub-inhibitory concentrations. The EPS hyperproduction could prevent eradicating cells persisting within biofilms despite repeated rounds of antibiotic treatment due to reduced antibiotic penetration and inhibition of phagocytosis and complement activation [38]. It is worth noting the activity of ribavirin, the only one among the drugs without influence on planktonic cell growth that caused a significant dispersion of preformed biofilm by P. aeruginosa. The lack of published studies warrants further studies to elucidate the underlying mechanism(s) of action. No antibiofilm activity was observed for meloxicam, previously reported to significantly inhibit P. aeruginosa PAO1 biofilm formation in a dose-dependent manner [17].
Although repurposing existing drugs offers the advantage of known safety and pharmacokinetic profiles, the cytotoxicity of these compounds is still to be investigated for novel applications. Therefore, the potential cytotoxic effect of the biologically active drugs ciclopirox, actinomycin D, 5-fluorouracil, and ribavirin was investigated, in comparison with tobramycin, against human bronchial CF cells. Ribavirin and 5-fluorouracil did not show any toxicity against CF bronchial cells, similarly to tobramycin. Despite earlier studies showing that ciclopirox has an excellent safety profile toxicity in several animal models after several types of administration [39], we found it has potential for cytotoxicity at concentrations lower than 1% commonly used for topical administration. Actinomycin D, a potent inducer of apoptosis, showed a cytotoxic potential comparable to ciclopirox, in agreement with the previously observed hepatic, blood, gastrointestinal, and immune-system-related toxicity [40]. Developing a novel aerosolized formulation of ciclopirox or actinomycin D would be beneficial since it could significantly reduce the amount of a drug related to a clinically relevant outcome, as already shown for NSAIDs pulmonary delivery [41]. Moreover, nanoparticulate formulations using biocompatible and biodegradable polymers could improve the residence time of the drug in the lung by supplying a depot delivery to the lung following nebulization. Specifically, the incorporation of PEG has been shown to increase the diffusion of nanoparticles through human mucoid surfaces [42], which might be particularly relevant in CF patients.
A drug repurposing strategy has also been successfully used to find antivirulence compounds able to decrease the potential damage produced by the pathogens to the host [43]. Unlike conventional antimicrobials, they act without affecting bacterial growth, reducing the chances of developing resistance. In this context, actinomycin D was previously considered a potential antivirulence agent against S. aureus due to the hemolysis inhibition [37], while ciclopirox was found to inhibit pyocyanin, although it increased pyoverdine production in P. aeruginosa [20]. Therefore, for the first time, we tested the effects of the biologically active drugs on the expression levels of selected P. aeruginosa virulence genes using real-time RT-qPCR. The exposure to each drug at 1/4xMIC generally caused increased expression of most genes tested, although to different extents. Specifically, the most advantageous pattern was observed for ribavirin, which increased only mexA, mexC, and toxA expression, while the worst one was associated with ciclopirox, which provoked hyperexpression of all genes but mexC. It is worth noting that actinomycin D caused down expression of aprA and algD codifying for the main virulence traits of P. aeruginosa, the alkaline protease and GDPmannose 6-dehydrogenase involved in the alginate polymerization needed for adhesion and subsequent biofilm formation, respectively. Similarly, we observed that 5-fluorouracil and tobramycin reduced aprA and algD expression, respectively. The anticancer drug 5-fluorouracil was recently proposed for repurposing as a quorum sensing inhibitor in P. aeruginosa [44]. However, the isolation of drug-insensitive spontaneous mutants indicates that resistance mechanisms can emerge even under in vitro conditions where the targeted virulence factor(s) is not required for growth [44]. Conversely, we found that exposure to 5-fluorouracil caused the down expression of the QS mediator lasI, probably due to the high variability in antivirulence activities previously observed among CF strains [44].

Drugs
Ribavirin, toremifene, oxyclozanide, meloxicam, 5-fluorouracil, actinomycin D, furosemide, and ciclopirox were tested comparatively with the aminoglycoside antibiotic tobramycin commonly prescribed for the inhalation therapy of P. aeruginosa infection in CF patients. All drugs tested were purchased from Merck KGaA (Darmstadt, Germany). According to the manufacturer's recommendations, stock solutions were prepared in dimethyl sulfoxide (DMSO; Merck KGaA) or reagent-grade water, aliquoted, and stored at −80 • C until use.

Bacterial Strains and Standardized Inoculum Preparation
Nineteen clonally distinct P. aeruginosa strains isolated from sputum samples of CF patients were initially enrolled in the study. Each strain was identified using MALDI-TOF mass spectrometry and then stored at −80 • C until it was cultured twice on Mueller-Hinton agar (MHA; Oxoid, Milan, Italy) to restore the original phenotype. A standardized inoculum was prepared for each strain, depending on the use.

Drug Susceptibility Assays of Planktonic Cells
The in vitro susceptibility of P. aeruginosa was evaluated, comparatively to tobramycin, using several assays.

Disk Diffusion Assay
The susceptibility of P. aeruginosa isolates to several antibiotics (i.e., amikacin, ticarcillin, piperacillin/tazobactam, ceftazidime, gentamicin, netilmicin, levofloxacin, and tobramycin) was evaluated by the disk diffusion technique according to the CLSI guidelines [24] and using Multodisc Pseudomonas (Liofilchem Srl, Roseto degli Abruzzi, Italy). A strain was defined as multidrug-resistant (MDR) if non-susceptible to at least one agent in three or more antimicrobial categories among those tested (aminoglycosides, penicillins, cephalosporins, carbapenems, and fluoroquinolones) [22]. E. coli ATCC25922 and P. aeruginosa ATCC27853 were used as quality control strains.

MIC and MBC Measurements
MIC and MBC values of each "non-antibiotic" drug were measured against 6 P. aeruginosa isolates, selected as representatives for MDR and HBF phenotypes. MIC was measured by the broth microdilution technique, according to the CLSI guidelines [24]. E. coli ATCC25922 and P. aeruginosa ATCC27853 were used as quality control strains. MBC was evaluated by plating onto MHA (Oxoid) 10 µL of broth culture from wells showing no visible growth at MIC determination. Following incubation at 37 • C for 24 h, the MBC value was defined as the minimum antibiotic concentration needed to eradicate 99.9% of the starting inoculum. Differences between MIC or MBC values were significant for discrepancies ≥2 log 2 concentration steps.

Time-Kill Assay
Kill kinetics of actinomycin, 5-fluorouracil, ciclopirox, and tobramycin against selected P. aeruginosa strains were evaluated by broth macrodilution. Briefly, the standardized inoculum (1-2 × 10 5 CFU/mL) was exposed to several concentrations of each drug in CAMHB and incubated at 37 • C. At prefixed times (1,2,3,4,5,6,12,16,20, and 24 h), a cell viable cell count was performed, and the results were expressed by plotting Log (CFU/mL) over time, considering 10 CFU/mL as the LOD. Control samples were prepared similarly but were not exposed to drugs. The carry-over antibiotic effect was not observed. Bactericidal activity was defined as a ≥ 3 Log (CFU/mL) reduction.

Checkerboard Microdilution Assay
The activity of tobramycin combined with 5-fluorouracil, ciclopirox, or actinomycin D was assessed against selected P. aeruginosa strains by the checkerboard microdilution method [25].

In Vitro Activity against Preformed Biofilms
Biofilms were grown for 24 h in a 96-well microtiter plate as previously described in "Biofilm formation assay". Next, they were exposed to each drug tested at the desired concentrations prepared in CAMHB. Following 24 h of exposure at 37 • C under static conditions, the effect against mature biofilms was evaluated in terms of biofilm biomass dispersion (crystal violet assay, as described in "Biofilm formation assay") and residual viability. In the latter case, non-adherent bacteria were removed after drug exposure by washing once with sterile PBS, then biofilm samples were scraped following a 5 min exposure to 100 µL trypsin-ethylenediaminetetraacetic acid 0.25% (Merck KGaA), and finally the suspension underwent to viable cell count on MHA. The percentage of inhibition of biofilm formation or dispersal of preformed biofilms following drug exposure was calculated as follows: (i) (1 -OD 492 of test/OD 492 of untreated control) ×100, in the case of crystal violet assay; (ii) [(CFU/well of the test)/(CFU/well of untreated control)] ×100, in the case of plate count assay.

Gene Expression Assay
The effect of drug exposure on the transcription levels of algD, toxA, lasI, aprA, mexA, mexB, and mexC virulence genes by P. aeruginosa PaPh32 was assessed by real-time reverse transcription quantitative PCR (RT-qPCR). Planktonic cells were exposed to each drug at 1/4xMIC for 20 h at 37 • C, washed with PBS, and then harvested in Qiazol (Qiagen; Milan, Italy). RNA was extracted by the phenol-chloroform technique, treated with DNase I (Merck KGaA), and checked for purity and quantity by NanoDrop-2000 spectrophotometer (Thermo Fisher Scientific Italia Inc., Monza, Italy). Strand cDNA was synthesized from 2 µg of RNA using a high-capacity cDNA reverse transcription kit (Thermo Fisher Scientific Italia), and gene expression was then evaluated using 10 ng of cDNA by RT-qPCR assay on QuantStudioTM 7 Pro Real-Time PCR System (Applied Biosystems) using the PowerTrack SYBR Green Master Mix (Thermo Fisher Scientific Italia Inc.). Primers were designed using as a reference the genome of P. aeruginosa strain NDTH9845 (GenBank accession number: CP073080.1) ( Table 5).
Specificity was assessed in silico with BLAST and by PCR endpoint under the same real-time RT-qPCR conditions. The ∆∆Ct method was applied to evaluate the relative gene expression in exposed vs. unexposed cells after normalizing on the proC housekeeping gene expression. The modulation of expression levels was shown as fold change. Table 5. Primer sequences used in real-time reverse transcription quantitative PCR analyses. The expression of selected virulence genes by P. aeruginosa PaPh32 was evaluated after 20 h of exposure to FDA-approved drugs by real-time RT-qPCR using oligonucleotides designed on the sequence of P. aeruginosa strain NDTH9845 (GenBank accession number: CP073080.1). The gene proC was used as housekeeping [46].

Statistical Analysis
Each experiment was carried out at least in triplicate and repeated on two different occasions (n ≥ 6). Statistical analysis was performed using GraphPad software (ver. 8.0; GraphPad Inc., San Diego, CA, USA). Data distribution was assessed using the D'Agostino and Pearson normality test, and then the differences in the biofilm biomass (OD 492 ) were evaluated using: (i) ANOVA + Tukey's multiple comparisons post-test for datasets normally distributed; (ii) ordinary one-way ANOVA + Holm-Sidak's multiple comparisons post-test in case datasets did not pass the normality test. Differences between percentages were assessed using Fisher's exact test. The significance level was set at p < 0.05.

Conclusions
Overall, our findings indicated ciclopirox and ribavirin as attractive candidates for repurposing as anti-P. aeruginosa agents in CF patients. Ciclopirox exhibited relevant antibacterial and antibiofilm activities, although further studies are needed to decrease its cytotoxic potential. At safe concentrations, ribavirin, a guanosine analog with broadspectrum virustatic activity, showed a specific antibiofilm effect since it was not related to antibacterial activity. In addition, it has already received FDA approval as an aerosol formulation, although for respiratory syncytial virus-infected infants. Ciclopirox and ribavirin have also been shown to have anti-inflammatory [47] and immunomodulatory [48] properties, respectively, which is highly relevant in CF patients where an exuberant, acute inflammatory response leads to pulmonary tissue damage and failure in clearing the infection [1,2].
Ciclopirox and ribavirin might therefore represent a good starting point for traditional medicinal chemistry providing potential chemical scaffolds for further drug development. Pre-clinical studies are warranted to evaluate the protective effect in animal models and pharmacodynamics/pharmacokinetics in humans.