Cystic fibrosis (CF) is the most common inherited genetic disorder in Caucasian populations (Lyczak et al., 2002). The pathology is bought by mutations in the gene for CF Transmembrane Conductance Regulator (CFTR) protein. Initially, CFTR is involved in production of sweat, digestive fluids and mucus. In the CF lungs, the chloride transport defect, caused by non-functional proteins, results in altered airway physiology, impairment of mucociliary clearance and production of thick mucus plugs. Consequently, CF individuals are predisposed to chronic pulmonary microbial infections, which lead to respiratory failure, the primary cause of patient death (Davies et al., 2007; Farrell et al., 2017).

Pseudomonas aeruginosa is the most frequent microorganism establishing chronic lung infections beyond infancy, in people with CF (Bhagirath et al., 2016). Indeed, this opportunistic pathogen possesses features that contribute to its adaptation and persistence in the CF lungs, despite the availability of aggressive use of antibiotics (Høiby et al., 2001). One of the main reason explaining the resistance of P. aeruginosa in the lower tract of CF patients is its ability to grow as a biofilm (Moreau-Marquis et al., 2008b).

This bacterial gathering, associated to the production of a protective polysaccharide matrix, renders microbial cells intrinsically more resistant to environmental agents and antimicrobials. The general principle of biofilm formation is the result of global regulatory modifications within the bacterium. The specific environment found in CF lungs seems to act as a stress signal governing the transition between acute (bacterial planktonic state) and chronic infections (bacterial sessile state) (Ciofu et al., 2015).

The aminoglycosides are an important class of antibiotics used to treat chronic bacterial infections, in particular lung infections. Among them, the superior activity of tobramycin (TOB) against P. aeruginosa makes it the preferred aminoglycoside for treatment of lung infections in CF. TOB by inhalation is often considered in first line for the eradication of early P. aeruginosa infections and for chronic therapies (Elborn et al., 2009). Amikacin (AMK) is also a good candidate as strains that are resistant to multiple other aminoglycosides tend to be susceptible to it. Its antimicrobial activity spectrum is the broadest of the group (Ehsan and Clancy, 2015).

The treatment of biofilm-involving infections is often poorly effective. Antibiotic susceptibility testing is traditionally performed on growing planktonic cells. Thus, antimicrobial selection does not take into account the common sessile growth of bacteria and antibiotic therapies cannot be effective on in vivo microbial biofilms (Ciofu et al., 2015). The development of clinical assays, evaluating the susceptibility of sessile microorganisms to antimicrobials, becomes essential particularly for the treatment of biofilm-associated infections as in CF (Moskowitz et al., 2011; Hurley et al., 2012).

The aim of the study was to evaluate the effects of AMK and TOB on the initiation of the biofilm formation by P. aeruginosa CF isolates, with the Biofilm Ring Test^® (BRT) system. This device allows the evaluation of the bacterial adhesion and the investigation of the antibiotic effect on the biofilm setting up (Antibiofilmogram^®) (Olivares et al., 2016). Interestingly, prevention phenomena of bacterial adhesion were reported with various concentrations of AMK and TOB. Crystal Violet (CV) assay was also used as a comparison (Christensen et al., 1985; Stepanović et al., 2000). Finally, to attest that the delay of the biofilm formation initiation by the two aminoglycosides was not a specific response of the bacterial adhesion on abiotic surfaces, we also used a clinically more relevant cell model. The antibiotic effects were evaluated through a tissue culture-based system allowing the biofilm formation on CF airway epithelium (Anderson et al., 2008; Moreau-Marquis et al., 2010).

Pulmonary infections are the main cause of morbidity and mortality in patients with CF. The lung colonization is essentially dominated by P. aeruginosa, which grows in a sessile state. This biofilm formation leads to a chronic inflammation in the airways and owing to its antimicrobial resistance, makes eradication of the infection impossible (Moreau-Marquis et al., 2008b; Bhagirath et al., 2016). As common antibiotic susceptibility testing do not indicate the tolerance of biofilm-growing bacteria, the development of laboratory assays, taking into account the in vivo sessile state of microorganisms is essential. This kind of diagnostic tools could be easily incorporated in routine clinical laboratory testing to provide complementary data for the antibiotic selection.

Antibiofilmograms^® allowed the study of adhesion kinetics of sessile bacteria and the evaluation of their susceptibility to antimicrobials through the BRT device. In our study, realization of these assays showed that our P. aeruginosa clinical strains are susceptible to various concentrations of AMK and TOB, two molecules considered as antimicrobials of first choice for treatment of pulmonary infections in CF. Both aminoglycosides delayed the initiation of the biofilm formation by bacteria.

These results were confirmed by the conventional CV assay. Finally, the inhibitory effect of aminoglycosides was confirmed with CF-derived cells. Only one discriminating result was noticed between the three applied methodologies: the inhibitory potential of the subMIC concentration of AMK on the bacterial adhesion prevention of the strain #1, when compared to the cell model which detects no antimicrobial effect. Effects of antimicrobials belonging to the families of β-lactams and fluoroquinolones were also tested with BRT, CV and cell co-culture assays. Clearly, these antibiotics did not disturb the spontaneous adhesion ability of bacteria (data not shown).

In opposition to our results, previous works have described the ability of subinhibitory concentrations of aminoglycosides, particularly tobramycin, to induce the biofilm formation in P. aeruginosa (Hoffman, 2005; Elliott et al., 2010). This deleterious effect of antimicrobials is explained as being an adaptive response of microorganisms to antibiotic exposure (Kaplan, 2011). Indeed, antimicrobials were presented as signal agents for the homeostasis regulation of microbial communities. As bacteria coevoluate in the environment under constant weak concentrations of antibiotics, they have developed a protective response by the formation of resistant biofilms (Linares et al., 2006). In our study, the subMIC concentration of TOB did not show deleterious effect on the bacterial adhesion (data not shown). These opposite effects of antimicrobials on microbial adhesion could be explained by the mucoid phenotype of our bacterial isolates. We intentionally used mucoid strains for co-cultures to allow the staining and the observation in fluorescence of the biofilm matrix. Mucoid isolates have been described as able to maintain their antibiotic susceptibility in comparison of non-mucoid strains, which rapidly develop resistance by their hypermutability (Oliver et al., 2000; Ciofu et al., 2001). Kaplan (2011) also introduce the notion of hormetic responses by antimicrobials. In fact, the bacterial biofilm formation exhibits a biphasic response to antimicrobial exposure. In this sense, antibiotics can act as antagonist or agonist according to the concentration tested (Kaplan, 2011). Finally, it seems established that the biofilm formation is mainly governed by adaptive responses. Indeed, transcriptomic analyses have failed to identify specific biofilm regulons suggesting that its formation is dependent on the environmental conditions (Whiteley et al., 2001; Ciofu et al., 2015).

Antibiofilmograms^® revealed that AMK and TOB decrease the bacterial adhesion as paramagnetic beads stay free-moving for these conditions. This tool allows rapid screen of antibiotics and does not require fixation or staining procedures. Its use could be appropriate to precise CF treatments as it is specifically intended for analyze of the early stages of biofilm formation. The major limitation of the conventional CV microtiter plate assay is the bias of the bacterial sessile biomass estimation due to the repetitive washing steps. It is even advisable to add a fixation step with ethanol or heat, just before the dye staining step, to prevent the detachment of adhering microorganisms (Stepanović et al., 2007). Even without this binding step, which usually enhances reproducibility of the assay, we found reproducible results confirming the preventive effect of aminoglycosides. Besides, we introduced our own classification of bacterial cultures according to adhesion potential of microorganisms. Based on our experimental observations, we decided to directly compare the final mean OD value of conditions to the upper OD of the negative control (mean OD + SD). This methodology differs from the one originally described by Stepanović et al. (2007), which defined a cut off OD to compare to final OD of the tested conditions. In terms of experiments carried out on only few hours of incubation, as for our CV assays, this method of classification undervalues the potential of bacteria to adhere to a support. Finally, we adapted a static tissue co-culture system to allow the adhesion, of our clinical isolates, on human CF-derived airway cells. Our clinical strains were able to bind to the cell layer in forming microcolonies and more precisely, 3D-structures. The latter were characterized by the presence of polysaccharide matrix, revealed by a calcofluor staining. In this co-culture model, AMK and TOB showed the ability to decrease the bacterial viability and the formation of a protective matrix. Previous studies have also evaluated the potential toxicity of antimicrobials on the airway cells. No evidence of cytotoxicity was observed for tobramycin treatment of P. aeruginosa biofilms on CFBE cells (until 1000 mg/L) (Anderson et al., 2008; Moreau-Marquis et al., 2009; Yu et al., 2012). Although this kind of in vitro cell model cannot be standardized in diagnostic laboratories, it is closer to the in vivo pathologic reality than microtiter plate assays. Even studies have suggested that in CF airways, the biofilm formation by P. aeruginosa is initiated within the thick mucus, in the early stages of lung colonization, bacteria can directly interact with epithelial cells (Anderson et al., 2008). In that sense, the cell system allows the evaluation of biofilm formation on another support than polystyrene microplates and with a different culture medium. It is well conceivable that according to the adhesion surface, the ability of biofilm formation varies. With tissue culture systems, it has already been observed differences of bacterial responses to antimicrobials according to the epithelial cell lines and even according to the CFTR status of cells, in the case of CF studies (Moreau-Marquis et al., 2008a). In the same manner, with in vitro assays, the medium composition may have a strong impact on the biofilm setting up ability of bacteria. To explore this issue, we have preliminary tested the adhesion ability of our strains, by microscopic observations, in BHI and RPMI + L-Glutamine media. We found a spontaneous adhesion of bacteria to microplates for the two culture media but in a much higher proportion for the RPMI one (data not shown). This medium seems to significantly stimulate the initial adherence of P. aeruginosa. This difference could be due to the poorer nutrients composition in RPMI than in BHI. The nutrient stress is a well-known parameter increasing the biofilm formation. Nevertheless, despite this adhesion stimulation, aminoglycosides could delay the bacterial adhesion in the co-culture experiments.

Taken together, results obtained by three inherently different assays confirm the ability of aminoglycosides to prevent the early adhesion of CF clinical P. aeruginosa strains. Currently, inhalations with TOB are predominantly favored for the maintenance therapy of chronically P. aeruginosa infected patients. Topical administration of antibiotics allows to reach high concentrations, in contrast with systemic treatments. In conclusion, the BRT device provides complementary information about antimicrobial effects on sessile bacterial cells. These data, being more relevant compared to a classical antibiogram, insist on its utility in routine hospital practice. Its robustness was confirmed by conventional and cell assays. Finally, we attested that aminoglycoside family, which is promptly administered to chronic infected CF patients, has a beneficial impact in delaying the initiation of the biofilm setting up by P. aeruginosa strains.

FJ and EO designed the study. EO performed the experiments, analyzed the data and wrote the manuscript. All authors read and approved the final manuscript.

EO, SB-B, CP and TB declare a conflict of interest: SB-B, CP and TB are members of BioFilm Control SAS, Saint-Beauzire, France. A CIFRE fellowship (Ph.D. student) from BioFilm Control, for EO, is gratefully acknowledged. The other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.