INTRODUCTION
Staphylococcus aureus is a Gram-positive bacterial species shown to be the most frequent cause of biofilm-associated infections (BAI) (
1) and one of the major causes of morbidity and mortality in hospitals and communities (
2). Unlike planktonic cells, biofilms exhibit specific phenotypic traits allowing them to resist host defenses and antibiotic treatments (
3), which frequently leads to chronic infections such as endocarditis, sinusitis, and osteomyelitis and also to implant-associated infections (
4).
Among the most recent clinically used antibiotics, daptomycin is a cyclic lipopeptide approved for the treatment of serious staphylococcal infections such as bacteremia and implant-related infections (
5). Daptomycin is a calcium-dependent antibiotic that acts by insertion into the Gram-positive cytoplasmic membranes where it forms oligomeric pores, causing potassium ion leakage and subsequent membrane depolarization, leading ultimately to cell death (
6). As is the case for many antibiotics, daptomycin has been shown to exhibit a significant bactericidal activity against planktonic cells (
7–9). However, the eradication of adherent bacteria is rarely achieved despite the large number of
in vitro and animal studies in which daptomycin activity was evaluated (
10–16). Besides the results of the literature that appear controversial (
17), direct comparison between studies is not directly possible, since the biofilm growth and treatment protocols used differ greatly.
To obtain a realistic representation of daptomycin action against
S. aureus biofilms, we developed an
in vitro model built on the basis of our recently published
in vivo study on
S. aureus prosthetic vascular graft infections using the same strains (
18). The interest of this approach was the possibility to use fluorescence imaging techniques that cannot be employed
in vivo (confocal laser scanning microscopy [CLSM], time-lapse microscopy, and fluorescence recovery after photobleaching [FRAP]) to examine the penetration, diffusion, bioavailability, and localization of the fluorescently labeled antibiotic inside the biofilms. To validate this approach, we monitored for 72 h the activity of daptomycin against biofilms formed by methicillin-susceptible and methicillin-resistant clinical and collection strains. In addition, we enriched the culture medium with proteins and calcium to mimic the
in vivo physiological conditions of our mouse model (
18). The same experiments were performed in the presence of rifampin, an antibiotic that can be combined with daptomycin for recalcitrant
S. aureus BAI.
MATERIALS AND METHODS
Bacterial strains.
Four S. aureus strains were tested in the present study: two were collection strains (methicillin-susceptible S. aureus [MSSA] ATCC 27217 and methicillin-resistant S. aureus [MRSA] ATCC 33591), and two others were isolated from patients with S. aureus bloodstream infections (MSSA 176 and MRSA BCB8). All strains were kept at −80°C in tryptic soy broth (TSB) (bioMérieux, France) containing 20% (vol/vol) glycerol. The frozen cells were subcultured twice in TSB (one 8-h culture, followed by an overnight culture) to constitute the stock cultures from which aliquots were kept at −20°C. Bacterial growth and experiments were both conducted at 37°C.
Antimicrobial agents and medium.
Daptomycin and rifampin were both purchased from Sigma (France). The fluorescently labeled antibiotic BODIPY-FL-labeled daptomycin (BODIPY-FL-daptomycin) (BODIPY-FL is fluorescently labeled boron-dipyrromethene) was a kind gift from Cubist Pharmaceuticals (MA, USA), and BODIPY-FL was purchased from Invitrogen (France). According to the manufacturer's instructions, the stock solutions were prepared by diluting daptomycin and BODIPY-FL-daptomycin in dimethyl sulfoxide (1 mg/ml) and by diluting rifampin in sterile water (2 mg/ml), which were then kept at −20°C. Before the solutions were used, they were diluted in TSB enriched with proteins (bovine serum albumin[BSA], 36 g/liter; Sigma, France) and calcium (CaCl2·2H2O, 50 mg/liter; Sigma, France) to mimic in vivo physiological levels. It has been determined that, under these conditions, the final concentration of dimethyl sulfoxide was noncytotoxic for the bacteria. Clinically meaningful concentrations were used in this study: 20 μg/ml for both daptomycin and rifampin. When combined, the antibiotics were mixed together before application to the biofilm surface.
Susceptibility testing.
The MICs of daptomycin and rifampin were determined by the broth microdilution method in cation-adjusted Mueller-Hinton broth (CAMHB), according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST). Media were supplemented with 50 mg/liter Ca2+ for daptomycin.
Characterization of molecular interactions between daptomycin and rifampin by absorption and fluorescence spectroscopies.
Absorption spectra of antibiotics alone and in combination were measured with a Varian Cary 300 spectrophotometer (Agilent Technologies, France). The corresponding fluorescence emission spectra were recorded using a Fluorolog-3 (Jobin Yvon, Inc., France) fluorescence spectrophotometer mounted with front-face detection geometry by exciting daptomycin at 360 nm. The measurements were made five times on each sample.
In vitro biofilm preparation and antibiotic activities.
Biofilms were studied in a polystyrene microtiter plate-based assay since it has been shown that this material has physicochemical properties close to that of biomaterial surfaces such as polyethylene terephthalate, which is used in vascular grafts (
19,
20). For the preparation of
S. aureus biofilms, 250-μl portions of an overnight subculture adjusted to an optical density at 600 nm of 0.02 (corresponding to ∼10
8 CFU/ml) were added to 96-well microplates (μClear; Greiner Bio-One, France). After a 1.5-h adhesion period at 37°C, the wells were rinsed with sterile physiological water (150 mM NaCl) in order to eliminate nonadherent cells, refilled with sterile TSB enriched with proteins and calcium (TSBpc) and then incubated for 24 h at 37°C to allow biofilm growth.
To assess antibiotic activities, the 24-h biofilms were rinsed with a 150 mM NaCl aqueous solution before adding the antibiotic solutions diluted in TSBpc as described previously. Viable culturable bacteria were then counted at time points at regular intervals: 0 h (when antibiotics are added), 24, 48, and 72 h after antibiotic injection. For each time point, bacterial cultures were centrifuged 10 min at 7,000 × g in order to eliminate excess antibiotic. The bacterial pellet was suspended in a 150 mM NaCl sterile saline solution, centrifuged again, and suspended in the same conditions. Successive decimal dilutions were then performed. For each dilution, 6 drops (10 μl) were deposited on tryptic soy agar (TSA) plates (bioMérieux, France) and incubated at 37°C during 24 h. CFU were counted and averaged for each dilution at each time point. The detection limit of viable culturable cells was 100 CFU/ml.
Percentages of rifampin-resistant mutants in biofilms.
To determine the percentages of rifampin-resistant mutants, the antibiotic solutions (rifampin alone or combined with daptomycin) were added to 24-h biofilms. CFU were counted on TSA plates or TSA plates containing rifampin (20 μg/ml) at 24 and 48 h after antibiotic injections. The percentages were obtained by calculating the ratio between the number of CFU grown on rifampin-containing TSA plates and the number of CFU grown on rifampin-free TSA plates.
Statistical analysis.
The mean log10 CFU/milliliter and biovolume for each therapy were compared with each other by analysis of variance (ANOVA). They were performed using the Statgraphics software (Manugistics, Rockville, MD, USA). Statistical significance was defined as a P value of less than 0.05 by a Fisher test.
Confocal laser scanning microscopy. (i) Visualization of antibiotic activities against biofilms using LIVE/DEAD staining.
Biofilms grown for 24 h prepared as described previously were observed 24, 48, and 72 h after antibiotic addition using a Leica TCS SP5 confocal laser scanning microscope (Leica Microsystems, France) at the Centre de Photonique Biomédicale (CPBM) (Orsay, France). Prior to each observation, bacteria were stained with 2.5 μM Syto9 (Life Technologies, France), a green cell-permeant nucleic acid dye, and 30 μM propidium iodide (PI) (Life Technologies, France), a red nucleic acid dye that can penetrate cells with compromised membranes (dead cells) only. Syto9 and PI were sequentially excited at 488 nm and 543 nm, respectively, and their fluorescence emissions were collected between 500 and 600 nm for Syto9 and between 640 and 750 nm for PI. Images were acquired using a 63× oil immersion objective with a 1.4 numerical aperture. The size of the confocal images was 512 by 512 pixels (82 by 82 μm2), recorded with a z-step of 1 μm. For each biofilm, four typical regions were analyzed. Images were reconstructed in three dimensions (3D) using ICY software.
(ii) Quantification of biovolumes and maximum biofilm thickness.
Maximum biofilm thickness (in micrometers) was measured directly from xyz stacks. Biovolumes (in cubic micrometers) were calculated by binarizing images with a java script executed by ICY software as described previously (
21). The biovolume was then defined as the overall volume of cells in the observation field. The percentage of dead cells corresponds to the ratio between biovolumes of PI-stained bacteria and Syto9-stained bacteria.
(iii) Antibiotic penetration and localization inside biofilms.
To study the penetration of BODIPY-FL-daptomycin alone and combined with rifampin within 24-h biofilms, we employed time-lapse microscopy, as described before (
22), using the same Leica TCS SP5 confocal microscope. Briefly, the fluorescence intensity evolution over time was observed in a defined focal plane (5 μm above the substratum surface). As soon as the TSBpc-diluted solutions of BODIPY-FL-daptomycin alone or combined with rifampin were added to the biofilm, fluorescence intensity images were acquired every second for 15 min. Simultaneously, transmission images were acquired to ensure that no structural alteration of the biofilm occurred during this process. The labeled antibiotic was excited with a continuous argon laser line at 488 nm through a 63× oil immersion objective, and the emitted fluorescence was recorded within 500 and 600 nm.
The corresponding diffusive penetration coefficients (
Dp) through the biofilms were determined according to the relationship previously described by Stewart (
23):
where
L is the biofilm thickness and
t90 is the time required to attain 90% of the equilibrium staining intensity at the deeper layers of the biofilm.
To observe the localization of the fluorescently labeled daptomycin within biofilms, bacteria were counterstained with the FM4-64 dye (Life Technologies, France): this dye was also excited at 488 nm, but its fluorescence emission was collected within the range 640 to 750 nm. The images (512 by 512 pixels) of both fluorophores were simultaneously recorded with a z-step of 1 μm.
(iv) Antibiotic diffusion and bioavailability inside biofilms using FRAP experiments.
Image-based fluorescence recovery after photobleaching (FRAP) measurements was used to assess local diffusion and bioavailability of the fluorescently labeled daptomycin. Briefly, FRAP is based on a brief excitation of fluorescent molecules by a very intense light source in a user-defined region to irreversibly photobleach their fluorescence. Fluorescence recovery is then probed over time at a low light power in the same photobleached region (
22,
24). All time-resolved measurements were obtained using the same confocal microscope. The time course of fluorescence intensity recovery was analyzed with mathematical models, giving us the quantitative mobility of the fluorescent molecules and allowing us to determine the diffusion coefficients. For all FRAP experiments, the fluorescence intensity image size was fixed to 512 by 128 pixels with an 80-nm pixel size and recorded using a 12-bit resolution. The line scan rate was fixed at 1,400 Hz, corresponding to a total time between frames of ∼265 ms. As determined previously, the full widths at half maximum in xy and z (along the optical axis) of the bleached profile were 0.8 μm and 14 μm, respectively, allowing us to neglect diffusion along the axial/vertical axis and thus to consider only two-dimensional diffusion. Each FRAP experiment started with the acquisition of 50 images at 7% of laser maximum intensity (7 μW) followed by a 200-ms single bleached spot at 100% laser intensity. A series of 450 single-section images was then collected with the laser power attenuated to its initial value (7% of the bleach intensity). The first image was recorded 365 ms after the beginning of bleaching.
DISCUSSION
The choice of antibiotics to treat
S. aureus BAI remains a challenge for the medical community. In this context, the ambivalence of the published results on daptomycin activity is a relevant example. Despite increasing data about daptomycin as an option to treat implant-associated
S. aureus infections, as many failures (
18,
27) as successes (
7–9,
12) have been reported both in clinical practice and in laboratory models. This highlights that
S. aureus BAI resistance/tolerance mechanisms to antimicrobials deserve more attention.
The biofilm-associated exopolymeric matrix may be considered to act as a shield to the antimicrobial diffusion reaction (
28–32) by delaying its penetration and/or reducing its bioavailability. To verify this hypothesis noninvasively, we took advantage of dynamic fluorescence imaging methods: confocal microscopy, time-lapse imaging, and FRAP. For our biofilm model and whatever the bacterial strain, no failure of daptomycin penetrability or bioavailability was observed. The opposite finding described by Siala et al. (
31) may be related to the conditions of fluorescence acquisition that were not well adapted to BODIPY-FL fluorescence. In this study, time-lapse fluorescence imaging experiments demonstrated that daptomycin rapidly reached the biofilm's deepest layers, while section views of fluorescence intensity images presented in
Fig. 3 ascertain the presence of the fluorescently labeled antibiotic through the whole biofilm structure. Furthermore, FRAP results ascertained that only ∼20% of the antibiotic molecules were immobilized. Thus, the majority of the antibiotic molecules were in free movement and could be bioavailable through the biomass (∼80% of nonimmobilized molecules).
We further addressed the question of whether or not daptomycin reached its bacterial target. Fluorescence intensity images provided interesting information, showing that the majority of fluorescently labeled antibiotic was distributed in the extracellular matrix rather than in the bacterial cell membranes (
Fig. 3). This is in agreement with the well-known property of daptomycin to have a very high degree of protein binding, especially with serum albumin (90 to 93%) (
26,
33) which is naturally present in physiological conditions. Nevertheless, the fluorescence recovery curves obtained by FRAP experiments in free medium and in the biofilms strongly suggested the reversibility of daptomycin protein binding (
33,
34): the equilibrium between the bound and unbound states may conserve the apparent mobility of the antibiotic. Additional experiments were performed in a protein-free medium (a saline solution supplemented with calcium ions). In this case, bacterial cell membranes appeared as hot spots on fluorescence images, consistently with the described antibiotic interaction with its target (
6). Surprisingly, whether the medium was protein-free or not, daptomycin exhibited the same lack of effectiveness, as revealed by time-kill studies performed by fluorescent LIVE/DEAD staining and conventional plating on agar (data in the absence of proteins not shown). Thus, the interaction with the matrix components cannot explain biofilm tolerance to the antibiotic.
Thus, the particular physiology of embedded bacteria should be considered, and more specifically, cells with low metabolic activity should be investigated. Previous studies using a bromodeoxyuridine (BrdU) immunofluorescent labeling technique demonstrated that the large majority of staphylococcal cells in a biofilm were actually in a low metabolic state (
35,
36). Additionally, in the present study, the comparison of cell viability results obtained by CFU counts and fluorescence imaging highlighted a significant proportion of viable cells detected by LIVE/DEAD staining but not by CFU measurements. This subpopulation may be considered viable but nonculturable (VBNC), a subpopulation known to have a slow metabolism (
37–39). Moreover, it has been demonstrated that daptomycin is poorly effective against bacteria in stationary stage (
7,
27). One can thus reasonably suggest that for bacteria with low metabolic activity, the cell membrane dysregulation induced by daptomycin may be slower and/or more difficult to attain due to structural modifications of the cell membrane. This assumption is supported by the reported data revealing that daptomycin displays a concentration-dependent bactericidal activity against dormant cells (
7,
9,
12). In the present study, we tested a higher concentration (double) of daptomycin (40 μg/ml) on the different
S. aureus biofilms, almost leading to bactericidal effects after 24 h of drug exposure (see Fig. S3 in the supplemental material) and showing no cell regrowth over time. However, biofilm clearance was not reached. This achievement was reported to occur at a daptomycin concentration equal to or greater than 100 μg/ml but may not be relevant in clinical practice (
7,
9,
12).
Two distinctive findings in this study concern the BCB8 clinical isolate, which discriminated itself by a twofold-higher penetration coefficient compared to the other strains tested and a greater susceptibility as revealed by the observation of a larger proportion of dead cells over the whole biofilm thickness, including the basal layer in contact with the substratum. These results are in line with those obtained
in vivo (
18), which demonstrated a strain-dependent activity of daptomycin against
S. aureus biofilms. In view of the antibiotic mechanism of action which is supposed to target the plasma membrane, the observed variable response depending on the bacterial strain may be due to a change in membrane composition or conformation from a strain to another.
Facing the lack of daptomycin efficiency in treating recalcitrant
S. aureus BAI, the addition of rifampin has raised great interest (
10–13,
15,
40–42). In this study, we demonstrated that the combined therapy was indeed highly efficient against
S. aureus biofilms but did not allow total bacterial clearance. Both antibiotics have been shown not to cross-react with each other, as evidenced by steady-state fluorescence spectroscopy. Moreover, the penetration, diffusion, and localization of the fluorescently labeled daptomycin were not affected by the presence of rifampin. We also proved here that rifampin-resistant mutants emerged when biofilms were treated with rifampin alone, but not when treated with the antibiotic combination. Altogether, the data presented here confirm that daptomycin prevents the emergence of rifampin-resistant mutants, allowing the bactericidal activity of rifampin to occur quickly, regardless of the cell physiological state.
In conclusion, consistently with the previous
in vivo study aiming at evaluating the antibiotic efficacy in
S. aureus prosthetic vascular graft infections (
18), we demonstrated in the present
in vitro model a strain-dependent lack of daptomycin activity toward biofilms. Dynamic fluorescence microscopy allowed discarding a lack of antibiotic availability and interaction with bacteria. Given the mode of action of daptomycin, these observations suggest a membrane-dependent factor of tolerance in such biofilms. Therefore, to provide a better understanding of the reduced activity of daptomycin against biofilms, the composition of the membrane should be analyzed.