Colistin Enhances Rifampicin’s Antimicrobial Action in Colistin-Resistant Pseudomonas aeruginosa Biofilms

ABSTRACT

The emergence of multidrug-resistant Pseudomonas aeruginosa infections has urged the need to find new strategies, such as the use of combinations of antibiotics. Among these, the combination of colistin with other antibiotics has been studied. Here, the action of combinations of colistin and rifampicin on both planktonic and sessile cells of colistin-resistant P. aeruginosa was studied. Dynamic biofilms were formed and treated with such a combination, resulting in an active killing effect of both colistin-resistant and colistin-susceptible P. aeruginosa in biofilms. The results suggest that the action of colistin on the outer membrane facilitates rifampicin penetration, regardless of the colistin-resistant phenotype. Based on these in vitro data, we propose a colistin-rifampicin combination as a promising treatment for infections caused by colistin-resistant P. aeruginosa.

INTRODUCTION

The increasing emergence of multidrug-resistant (MDR) bacteria has been put in the spotlight, as well as the urgency of finding new treatment strategies (1). A relevant case is the cationic polypeptide colistin, active on Gram-negative bacteria (2, 3). Colistin’s primary targets are the membranes, but the mechanism of action is not fully understood, although it is well established that it firstly attaches to the lipid A segment of the lipopolysaccharide (LPS), displacing divalent cations, and leading to disruption of the membrane bilayer and to death by injuries caused in the cytoplasmic membrane (4, 5).

Colistin has been considered an antibiotic of last-resource as, in some occasions, it is the only option to treat infections caused by MDR bacteria (6). The use of colistin was discarded due to the nephro- and neuro-toxicity besides the appearance of new antibiotics, such as aminoglycosides with similar activities and fewer adverse effects (6) but further rescued to treat MDR infections. Simultaneously, new strategies are envisaged, and among them, combinations containing colistin and other antibiotics (7). The synergistic effect seen in combination of colistin with antibiotics suggested that this could be a basis for new promising therapeutic strategies (8, 9).

P. aeruginosa is responsible of a variety of infections, particularly severe in patients with chronic obstructive pulmonary disease (COPD), burned patients, patients suffering from chronic ulcers, immunocompromised, and cystic fibrosis (CF) patients. In the case of CF and COPD, the colonization of the respiratory tract is frequently caused by antibiotic-resistant strains with remarkable biofilm-forming capacity (10, 11).

Biofilm, a structured community of bacteria embedded in a self-produced polymeric matrix, frequently attaches to surfaces, interfaces, or in the fluids/secretions/tissues like mucus and chronic wounds as small aggregates (12, 13); biofilm bacteria (sessile) differ from the planktonic in many aspects, including the susceptibility to antimicrobials. In fact, bacteria in biofilms become tolerant to common antibiotics as: (i) the antimicrobial have some restriction in penetrating the biofilms, especially the positively charged antibiotics; (ii) the low concentration of oxygen and metabolic activity created in inner regions due to the bacterial consumption in the external layers decrease the susceptibility of bacteria to antibiotics (14), and (iii) growth rates of persistent cells are low, although they can revert to actively growing forms if e.g., hyperbaric oxygen is added to the biofilms (15, 16). Differences between subpopulations in the biofilm are of interest when analyzing the effect of antibiotics. Sans-Serramitjana et al. (17) demonstrate that colistin and nanoformulations containing colistin have remarkable effects on the biofilm subpopulations, as the percentages of dead cells were higher in the inner than in the outer layers of biofilms treated with free colistin and colistin loaded nanoparticles. These results pointed out that colistin is able to reach the deeper layers of the biofilm and kill the dormant subpopulation, independently of the formation of reactive oxygen species (ROS). The prominent role played by biofilms in infections and, particularly, in Pseudomonas respiratory infections makes the study of new perspectives in antbiotic treatment in the biofilm scenarios important.

Colistin has become very useful to treat infections caused by MDR P. aeruginosa infections, as other antibiotics with antipseudomonal activity were no longer useful. Since 1984, all chronically infected Danish CF patients have inhaled 2 million units of polymyxin E (colistin, 1 mg colistin base = 30,000 units) twice daily between the regular intravenous antibiotic courses, in order to maintain pulmonary function and decrease chronic inflammation within the respiratory tract. Moreover, in order to postpone chronic infection, the routine use of early aggressive treatment with inhaled colistin and oral ciprofloxacin for 3 weeks to 3 months for CF patients intermittent colonized with P. aeruginosa was introduced in 1989. We have previously reported on the spread of colistin-resistant non-mucoid P. aeruginosa among chronically infected CF patients in the Copenhagen CF center (18–21).

P. aeruginosa can develop resistance to colistin by mutations altering the LPS composition and structure, and specifically the lipid A moieties. Expression of the chromosomally encoded genes involved in lipidA modifications is commonly regulated by 2-component regulatory systems (TCS). In P. Aeruginosa, the regulatory systems are complex, and involve PhoPQ, PmrAB, and several others, such as ParRS, CprRS and ColRS. The most common TCS that are mutated are PhoPQ and PmrAB, leading to upregulation of arn operon. Genes pmrAB and phoPQ form a 2-component system that regulates a variety of genes encoding the enzymes responsible for lipid A synthesis and L-Ara4N addition. When the 2-component system is mutated, the lipid A becomes modified, and then colistin does not bind the bacterial outer membrane (22). Thus, the cytoplasmic membrane is protected but not the outer membrane. This is similar, if not identical, to the mechanisms of intrinsic polymyxin-resistance in species, such as Serratia marcescens and Burkholderia species, where polymyxins severely alter the outer membrane but this has no consequences on cytoplasmic ones or on cell viability. Thus, colistin has no antibacterial effect, as it facilitates the entry of rifampicin (23).

Rifampicin is a hydrophobic antibiotic that does not effectively penetrate through the outer membrane of Gram-negative bacteria. However, if associated with antibiotics that permeabilize the outer membrane of Gram-negative bacteria (such as colistin), rifampicin may also be effective for Gram-negative bacteria.

Here, the effect of colistin against biofilms of both colistin-susceptible and colistin-resistant P. aeruginosa CF clinical isolates was investigated. In addition, susceptibility to combinations of colistin and rifampicin on planktonic cells was determined.

RESULTS

Table 1 shows MIC and Fractional Inhibitory Concentration Index (FICI) values of colistin and rifampicin in planktonic growing P. aeruginosa strains. The isolate 41782/98 was fully resistant to colistin, and the rest were susceptible.

TABLE 1.

MIC and FICI values (μg/mL) of the planktonic growing P. aeruginosa strains PAO1, 198848/85, and 41782/98 from CF patients

Bacterial strain	COLa	RIFb	COL/RIFc	FICI COL/RIF
PAO1	2	32	0.5/16	0.75
19848/85	1	64	0.5/32	1
41782/98	256	32	2/1	0.039

^aCol: colistin.

^bRIF: rifampicin.

^cCOL/RIF combination.

Figure 1 shows the time-kill curves at 0.25 μg/mL colistin, 16 μg/mL rifampicin, and combinations of colistin and rifampicin 0.25 and 16 μg/mL, respectively, in planktonically growing susceptible strains (PAO1 and 19848/85), and 1 and 2 μg/mL, respectively, in colR strain (41782/98). While rifampicin alone did not significantly affect the growth kinetics of any of the strains tested, colistin induced a rapid killing of PAO1 and 19848/85, followed by a delayed re-growth. On the contrary, colistin had no effect on the resistant strain 41782/98, as expected. Nevertheless, combinations of both antibiotics succeeded in the complete eradication of susceptible strains (PAO1 and 19848/85) but also in the case of 41782/98, where, at even much lower concentrations, we were able to observe a fully bacteriostatic effect.

FIG 1.

Time-kill curves of planktonically growing P. aeruginosa strains in the presence of colistin, rifampicin, and colistin-rifampicin combination (concentrations were 0.25 μg/mL colisin and 16 μg/mL rifampicin for colistin-susceptible strains [PAO1, 19848/85] and 1 μg/mL colisitn and 2 μg/mL rifampicin for colistin-resistant 41782/98).

Experiments with biofilms also displayed consistent results, as Table 2 shows the results in static biofilms (Calgary device). Colistin-susceptible strains displayed a minimum biofilm eradication concentration (MBEC) of colistin, rifampicin, and the combination of both.

TABLE 2.

Minimal biofilm eradication concentration (μg/mL) of colistin (COL) and rifampicin (RIF), and of the combination (COL/RIF), and the calculated FBECi of P. aeruginosa biofilms

	MBEC
Bacterial strain	COL	RIF	COL/RIF	FBECia COL/RIF
PAO1	64	>512	16/16	<0.281
19848/85	16	>512	4/4	<0.257
41782/98	>1024	>512	32/64	<0.156

^aFBECi, Fractional Eradication Biofilm Concentration Index.

The combination of colistin and rifampicin was much more active and displayed synergistic activity.

Figure 2 shows confocal laser scanning microscopy (CLSM) images of the biofilms formed by the 3 studied strains. While the PAO1 biofilm contained mushroom-shaped microcolonies, the other 2 biofilms were flat. These 5-days-old biofilms were exposed to 25 μg/mL colistin for 24 h, and stained with Live-Dead stain to identify dead cells. A total biofilm eradication (red cells) was achieved in PAO1 and 19848/85, while in the case of 41782/98, a significant number of survivors was found. Similar experiments at higher concentrations (250 μg/mL) displayed similar results (supplementary materials). The biofilms produced by the colistin-resistant strain were not eradicated.

FIG 2.

Exposition of 5-days grown biofilms from P. aeruginosa biofilms to 25 μg/mL of colistin during 24 h. Confocal images shown are from 0 h (left) and 24 h (right) of the treatment from biofilms of (a) PAO1, (b) 19848/85 (colistin-susceptible), and (c) 41782/98 (colistin-resistant) strains. The 3 strains are tagged with gfp (green) and stained with Live-Dead stain; red cells are dead cells.

Earlier studies that used the prodrug colistin methanesulfonate (CMS) did not obtain full eradication of flow-chamber grown PAO1 biofilms, but instead found that the outer part of the biofilm survived CMS treatment. In contrast, the inner part of the biofilm was killed (16). Furthermore, evidence was provided, where the outer part of the biofilm consisted of active bacteria that were able to induce the pmr genes upon gradually increasing colistin exposure, and, therefore, survived the treatment (16). The prodrug CMS is administered intravenously (IV) and is slowly hydrolyzed to liberate increasing concentrations of colistin sulfate. We used increasing colistin concentrations for treatment of the flow-chamber grown P. aeruginosa biofilms, in order to mimick the in vivo situation in patients. Images of CLSM of biofilms treated with increasing concentrations of colistin during the first 6 h (0.39 μg/mL increasing every h to 25 μg/mL), followed by a constant concentration (25 μg/mL) for the next 18 h, are shown in Fig. 3. In agrement with earlier results (16), we found that, for PAO1 and 19848/85, the outer part of the biofilm was able to survive colistin treatment whereas the inner part was killed. However, in the case of 41782/98, none of the cells in the biofilm were killed during the entire experiment.

FIG 3.

Biofilms exposed to increasing concentrations of colistin during the first 6 h, and then a constant concentration (25 μg/mL) until 24 h.

The killing effect of combinations of colistin and rifampicin on the biofilms formed by colistin-resistant strain 41782/98 is shown in Fig. 4, where the combination of colistin (25 μg/mL) and rifampicin (64 μg/mL) succeed in killing cells in the biofilm, while separately, there was no effect at all. As can be concluded from Fig. 5, colistin alone was able to significantly increase the proportion of dead bacteria in PAO1 and 19848/85 biofilms, but not in 41782/85 biofilm. On the contrary, the combination of both antimicrobials significantly enhanced the proportion of dead cells in the biofilm formed by the colistin-resistant strain 41782/98 (Fig. 6).

FIG 4.

Five-day biofilm of colistin-resistant strain 41782/98 was exposed to (a) 25 μg/mL colistin, (b) 64 mg/mL rifampicin and (c) combination of colistin (25 μg/mL) and rifampicin (64 μg/mL). The killing effect of the combination is clear at 24 h.

FIG 5.

Surviving biomass of biofilms treated with increasing concentrations of colistin. There is a clear difference between colistin-susceptible strains PAO1 (blue) and 19848/85 (green), and the colistin-resistant strain 41782/98 (red); the first ones showed increasing ratios, and the second one the same ratio during all the treatment.

FIG 6.

Surviving biomass during treatment of colistin-resistant 41782/98 biofilm. Colistin (blue) and rifampicin (green), and combination of both (red).

DISCUSSION

Multi-drug-resistant P. aeruginosa infections have increased their prevalence in both hospitals and communities. Patients, particularly immunocompromised, burned, patients, and CF patients, colonized with this bacterium are at high risk of treatment failure. Biofilm formation has been confirmed as a factor of antibiotic tolerance, increasing the resistance, and making it challenging to obtain bacterial eradication. It has been pointed out that even the use of last-resource antibiotic options, such as colistin, may be problematic due to the emergence of resistance phenomena. Colistin resistance mechanisms are not fully understood, although it has been pointed out that both the 2-component system, pmrAB and phoPQ, and biofilm formation play prominent roles. Characteristics of biofilm formation and biofilm structure vary from one bacterium to another. More efforts should be done to find new compounds and strategies. Among them, the combination of antimicrobials is worthy of being explored, since, except in the case of allergy, side effects are dose-dependent (i.e., colistin) and the combinations may be active at low concentrations. The 3 strains tested displayed MICs of rifampicin of 32 or 64 μg/mL; although no breakpoints for rifampicin are available, they may be considered moderately susceptible to rifampicin (24).

Values of FICI of planktonically growing bacteria pointed out a strong synergism of colistin and rifampicin in the colistin-resistant strain, while indifference was displayed in the case of 2 susceptible strains. It has been proposed that resistance to rifampicin in Gram-negative bacteria is, at least partially, caused by a natural membrane barrier that limits the penetration of the antibiotic. If this is so, the role of colistin may be that of a door opener, despite being unable to kill the bacterium, as it facilitates the penetration of rifampicin and subsequently its antimicrobial action (24).

The physiological state of bacteria living in biofilms is highly variable. Thus, one may define at least 2 subpopulations: the first one is formed by individuals inhabiting the external layers of the biofilm, as they are metabolically active, and the exchange with the medium is easy. The second is a dense subpopulation of bacteria located in the deeper part of the biofilm, less active metabolically.

Our results obtained in flow-cell biofilms (Fig. 3) are in agreement with those of Pamp et al. (25), who demonstrated that cells at the inner layers of the biofilm, which had lower metabolic activity than those being at external layers, are metabolically active and are more capable of activating mechanisms of reversible colistin tolerance at low concentrations. Colistin concentrations of 25 μg/mL succeeded in completely eradicating biofilm, although after 24 h of incubation, a few individuals still persisted and may regrow. The colistin-resistant isolate (41782/98) was able to survive after treatment with colistin, even at concentrations as high as 250 μg/mL (Fig. 1 and Supplementary materials). The use of dynamic biofilms and their behavior, when submitted to increasing concentrations of colistin, may greatly help in the elucidation of the mechanisms by which sessile bacteria tolerate the antibiotic.

Recently, AMPs have gained interest in research since they expose a need for new antimicrobials or new applications of traditional antimicrobials. Promising results of combinations formed by subinhibitory concentrations of colistin with uncommon antibiotics have been reported (7). On the other hand, studies using colistin combined with more than one antibiotic are scarce. The eventual success of such a strategy may allow the reduction of antibiotic concentrations without losing efficiency (26–29). The use of 3 (or more) antimicrobials may, in principle, be regarded as a way to: (i) enlarge the bacterial spectra by targeting different structures, (ii) enhance the efficacy and activity of antibiotics, (iii) overcome resistance emergences, and (iv) decrease the toxicity since concentrations are much lower (30).

Rifampicin was previously reported to be highly effective on Gram-negatives when combined with colistin (31). Still, our study demonstrates, for the first time, that this is also the case with biofilm-growing bacteria. Linezolid’s effect is also enhanced by colistin since linezolid resistance depends upon the efflux, and efflux machinery is inhibited by colistin (9).

When comparing the results in planktonic and sessile cells, we found that the combination of colistin and rifampicin had a bactericidal effect on colistin-resistant P. aeruginosa biofilms. Humphrey et al. (32) showed that colistin permeabilizes the outer membrane, although not the cytoplasmic membrane. Thus, the role of the outer membrane barrier would disappear due to the effect of colistin, allowing the entry of small molecules, such as rifampicin, through the injured outer membrane, which may be the mechanism underlying the synergistic effect of colistin and rifampicin. Furthermore, a higher effect was observed when the combinations were assayed in Escherichia coli colistin-resistant strains (32), as we see in both planktonic and biofilm cells of colistin-resistant P. aeruginosa. Further studies are needed to understand the mechanism of action of the studied combinations in detail.

These results open the door to design future treatments for colistin-resistant bacteria, even when growing as a biofilm by using combinatory therapy and, therefore, increasing the perspectives to successfully treat such infections. Combinations of colistin and rifampicin have also higher killing effect on colistin-resistant A. baumannii than on colistin-susceptible strains (33). Similarly, we report here that colistin-resistant P. aeruginosa are fully susceptible to rifampicin-colistin in planktonic and sessile cells. Results on dynamic biofilms also showed the efficacy of the colistin-rifampicin treatment on colistin-resistant biofilms, and, accordingly, this may be a therapy option for such infections. Although the intimate mechanism of action is not discussed here, it is worth noting that the outer membrane of bacteria, such as Pseudomonas and Acinetobacter, is particularly impermeable, and that numerous efflux pumps have been described in both genera. Therefore, the use of “door openers” and pump inhibitors, such as colistin (8), should play an active role in the sensitization of other antibacterials. The destruction of the biofilm at 25 μg/mL of colistin was practically complete when combined with rifampicin. However, this concentration of colistin is actually achieved in serum by i.v. dosing and in sputum by inhalation, and is used in Danish CF patients (34–36). It is relevant to propose that, in spite of colistin strictly having no antimicrobial activity against biofilms formed by colistin-resistant P. aeruginosa, it has therapeutic interest when used in combinatory strategy because it may sensitize bacteria to other antimicrobials. Harvested cells of dynamic biofilm from flow-cell of a colistin-resistant strain showed that there was no large reduction in the quantitative count (Supplemental materials), which may be caused by reversion of injuries in the outer membrane caused by colistin. This is also the reason why the dead/alive ratio was lower in the resistant strain, which may reside in the fact that rifampicin behaves bacteriostatic on P. aeruginosa.

It has been shown that colistin-based double and triple antimicrobial combinations may offer promising alternatives in the treatment of infections by MDR bacteria, even when using antibiotics toward the targeted bacterium is intrinsically resistant, such as linezolid in Gram-negatives (27). When CAMPS (like colistin) are combined with other antimicrobials (rifampicin, in our case), they made otherwise fully and constitutively resistant bacteria susceptible, but also enhance susceptibility. This could suggest that the outer membrane permeability could play a larger role, even than it has been ascribed to it in the literature, and that any strategy tending to increase the concentration of antimicrobial in the periplasmic space would result in increased susceptibility; in this paper we confirm this conclusion in the case of bacteria when they are part of the biofilm. This is reinforced by the observation that, apart from its direct effect on the permeability of the outer membrane, colistin also has an inhibitory effect on efflux pumps (8).

MATERIALS AND METHODS

Bacterial strains, growth conditions, and chemicals.

P. aeruginosa PAO1 and 2 isogenic sequential isolates from the CF patient affected by cystic fibrosis from the Copenhagen CF Center (19848/85 and 41782/98) were used. The whole-genome sequence of the resistant isolate 41782/98 showed that colistin-resistant phenotype was due to a deletion1089_1094delCCTGGG in the phoPQ 2-component system as well as 61C > T in OprH (see Fig. 5, Table 1, and Supplementary material).

All strains were tagged with eGfp in a mini-Tn7 construct (37).

Routine cultures were done in LB agar at 37°C. Susceptibility tests and time-killing curves were done in Müller-Hinton Broth (MHB). In experiments, to determine MBEC and for flow-cell biofilm experiments, minimal medium (MM) was used. Minimal media was based on trace metals and salts supplemented with 10% A10 phosphate buffer and 3 mM glucose as described in Tolker-Nielsen et al. 2014 (38). Rifampicin was purchased by Sigma-Aldrich chemicals. Colistin sulfate and propidium iodide (PI) were from Sigma-Aldrich.

Susceptibility testing and checkerboard assay.

Following EUCAST recommendations (39), microdilution method was used to determine susceptibility to colistin and rifampicin.

Interaction between the 2 antimicrobials was quantitatively determined by checkerboard experiments, and calculated as FICI. FICI is the sum of FIC A (MIC of antimicrobial A combined/MIC of A) and MIC B (MIC antibiotic B in combination/MIC antibiotic B).

Time-kill curves.

The killing effects of colistin and rifampicin alone, and in combinations, were plotted as time-kill curves. Starting inoculum was 5x10⁵ CFU/mL (CFU/mL). Antimicrobials were added to 5 mL of cultures at exponential phase of growth, and incubated at 37°C. Samples were retrieved aseptically at 0.15, 0.30, 0.45, 1, 2, 4, 6, 24, and 48 h. An antibiotic was considered active when a reduction of ≥ 1 log₁₀ from the initial inoculum was done, in accordance with Lora-Tamayo et al. (26). When the combinations reduced ≥ 2 log₁₀, with respect to the most active antibiotic, they were considered synergistic. Concentrations of colistin and rifampicin must be carefully adjusted because the bacterial-killing effect of colistin is fast. For example, for the colistin-resistant strain, concentrations of 1 μg/mL of colistin and 2 μg/mL of rifampicin were used.

MBEC determination.

The effect of colistin and rifampicin was determined in static biofilms using the Calgary device method by Moskowitz et al. (40), and by Minimal biofilm eradication concentration (MBEC) determinations. Briefly, bacterial biofilms were formed by immersing pegs of a modified polystyrene microtiter lid (catalog No. 445497; Nunc TSP system) into 96-well microtiter plates filled with 200 μL/well of CAMHB, and incubated at 37°C 24 h in static conditions, which contrasts the continuous culture conditions in the flow cells. After incubation, Pegs were gently rinsed in 0.9% NaCl solution, and biofilms were exposed to the different concentrations of antimicrobials for 24 h at 37°C. Pegs were again rinsed with 0.9% NaCl solution, and biofilms were removed by sonication. Recovered bacteria were incubated for 24 h at 37° C. Optical densities at 620 nm wavelength were measured to determine MBEC, and defined as the lowest concentration of an antimicrobial that prevented bacterial regrowth. Experiments were performed in triplicate.

The effect of combinations of colistin and rifampicin was also determined. The Fractional Biofilm Eradication Concentration index (FBECi) was calculated as follows: FBECi = BEC A + BEC B; where BEC A is the result of dividing MBEC A in combination by MBEC A alone, and BEC B is the result of dividing MBEC B in combination by MBEC B alone (41). The criteria were analogue to that of FICI; that is, values of FBECi < 0.5 indicate synergism, while > 0.5 and < 4 no interaction; and > 4 < antagonism.

Flow-cell biofilm experiments.

The effect of colistin on continuous culture conditions biofilms was studied by using a flow-cell biofilm setup. Flow cells consisted of 3 parallel and independent channels of 1 × 4×40 mm each. The channels were covered with a glass coverslip (Knittel Gläser) as substratum for biofilm attachment and formation. MM was pumped continuously for 1 h to stabilize the system. Biofilm growth took place after inoculation with 300 μL at OD₄₅₀ between 0.05 and 0.02, and then fresh MM was pumped (Watson Warlow 250S peristaltic pump), creating a continuous flux of medium. To allow initial bacterial attachment to the cover slip, the chamber was left without flow for 2 h, then a constant flow of medium was set at 0.2 mm/s for the rest of the experiment. Once inoculated, the flow-cell chambers were left for 5 days in all experiments at room temperature, until a grown biofilm was observed. Simultaneously with the treatment, PI was added at 0.3 μM (PI stains only dead cells) to count dead cells. All experiments were performed in triplicate.

Variable concentrations of colistin.

To determine the possible effect of colistin in the activation of adaptive behavior (tolerance), clinical conditions were mimicked by performing experiments in which increasing concentrations of colistin were added to the system, thus doubling the concentration every hour starting at 0.39 μg/mL; after 6 h, the system was left at final concentration (25 μg/mL) up to 24 h.

Effects of combinations.

Based on the results obtained from colistin-resistant strain in MBEC experiments, and taking into account the clinical doses at which colistin is used, concentrations of 25 μg/mL of colistin and 64 μg/mL of rifampicin were chosen for experiments. Susceptible strains were used as controls. The concentration of colistin was chosen from our previous study on the optimized use of colistin in CF patients (34).

The survival rates were determined by colony counting of bacteria recovered from treated and untreated biofilms. First, sessile bacteria were harvested after carefully removing the liquid content with sterile syringes. Then, bacteria were harvested by vigorously shaking with 1 mL of 0.9% NaCl containing 50 μg of glass beads (ø ≤ 106 μm) (Sigma). The harvested biofilm was mixed and vortexed, and bacterial suspensions were diluted and spread onto LB agar plates incubated 24 h at 37°C, and the number of CFU was counted.

Microscopy imaging.

Both the control and treated biofilms were visualized by using a Zeiss LSM 800 confocal laser scanning microscope (Carl Zeiss), every 30 min. The microscope was equipped with an argon and NeHe laser and detectors, and filters for GFP (excitation 488 nm, emission 517 nm) and PI (excitation 543 nm, emission 565 nm) detection. The optic used was a 63X/1.40 oil immersion objective. Image processing and grouping was performed with Imaris 8.2 Software (Bitplane AG). Biofilm flow-cell channels were placed under the microscope during the 24 h of treatment, and images of 3 different fields of each channel were taken each hour. Ratios between the GFP-fluorescence and PI were analyzed using the program GraphPad Prism version 6.0c.