An Antibiofilm Vector Based on a Modified β-Cyclodextrin: Complexation Study and Biological Activity

1. Introduction

Staphylococcus epidermidis is a major infective agent in compromised patients, such as drug abusers or immunocompromised patients (for example, patients under immunosuppressive therapy, chronic wound patients, AIDS patients, and premature newborns). The ability to adhere and subsequently form biofilms on indwelling devices is among the potential virulence factors associated with S. epidermidis species [1,2,3]. Indeed, sessile organisms exhibit a high resistance to antimicrobials compared to their planktonic counterparts [4]. The Extracellular Polymeric Substances (EPS) of the biofilm matrix form the scaffold for a three-dimensional architecture, being involved in bacterial adhesion and biofilm cohesion [5]. EPS matrix also contributes to bacterial resistance against nonspecific and specific host defenses and antimicrobial agents [6,7,8,9,10,11,12]. Polysaccharides are a major fraction of this EPS matrix [5,13]. Most of them are linear or branched long molecules. Some of these polysaccharides are homopolysaccharides, including sucrose-derived glucans and fructans [13], but most of them are heteropolysaccharides that consist of a mixture of neutral and charged sugar residues. They can contain organic or inorganic substitutions that significantly affect their physical and biological properties.

Polycationic exopolysaccharides also exist, such as intercellular adhesin, which is composed of β-1,6-linked N-acetylglucosamine with partly deacetylated residues. The matrix-mediated antibiotic resistance is mainly due to a low metabolic activity and oxygen limitation within the biofilm [14]. The role of reduced antibiotic penetration in the drug resistance of biofilms, is highly antibiotic dependent, beta-lactams and vancomycin diffusion being significantly reduced, whereas the diffusion of aminosides and fluoroquinolones seems un-affected [14,15,16]. Some efficient Strategies for biofilms eradication are combinations of bacteria killing and matrix removal by using Detachment Promoting Agents (DPAs). Many of these DPAs were described in the literature, including enzymes and chemical agents [17]. They include in particular chelating agents, for example, EGTA [18], EDTA, NaCl, CaCl₂, or MgCl₂ [17]. Among agents having a promising clinical future, are biofilm-matrix degrading enzymes. Indeed, numerous studies pointed out the ability of these enzymes to inhibit the biofilm proliferation, detach preformed biofilms, and sensitize biofilms toward antimicrobials by depolymerizing either polysaccharides [19,20] or extracellular DNA [21]. DispersinB® (DspB) [22] was discovered in 2003 from Actinobacillus actinomycetemcomitans. Treatment of S. epidermidis biofilms with DspB caused dissolution of the EPS matrix and detachment of biofilm cells from the surface [20,23], and disrupted biofilm formation by Escherichia coli, S. epidermidis, Yersina pestis and Pseudomonas fluorescens [19]. One of the main drawbacks of the enzyme-based anti-biofilm strategy is that dispersal of bacteria from the biofilm may favor bloodstream infections, septic thrombophlebitis, endocarditis, metastatic Infections, and sepsis [2,24,25,26,27]. It is the reason why, most of the time, it is simultaneously or sequentially used in combination with antimicrobial agents.

Several studies also showed that DspB sensitizes biofilm bacteria to antibiotics [28,29,30], and macrophages [31]. A preliminary experiment, here performed, confirmed that DspB enhanced the activity of the ciprofloxacin (CIP) fluoroquinolone against biofilms when used simultaneously.

Synthetic fluoroquinolones and derivatives are broad-spectrum antibiotics that are widely used for the treatment of serious bacterial infections for human and veterinary use [32,33]. The structure-activity relationship of fluoroquinolones has been extensively investigated, and depends on the presence of the carboxyl, carbonyl and fluorine groups at C₃, C₄ and C₆ positions, respectively [34]. Moreover, the antibacterial effectiveness is highly improving if an amino substituent is present at C₅, an alkylated pyrrolidine or piperazine at C₇, a halogen at C₈, and a cyclopropyl group at N1 [34]. An interesting approach is the development of new quinolone-based antibacterial agents by chemical modification on the carboxylic acid for the preparation of quinolone-macrocycle conjugates [35]. Ciprofloxacin is one of the most commonly used fluoroquinolone in clinical field with for examples treatment of respiratory infections [36,37], meningitis [36], septicemia [36], intraabdominal infections [38], So, we designed a new antibiofilm vector, consisting to a DspB-β-cyclodextrin (β-CD)/CIP-Ad complex, combining both enzyme and antibiotic delivery. This vehicle is an “all-in-one” tool, which provides the ability to simultaneously destroy the biofilm matrix and release the antibiotic to achieve a combined effect (Figure 1).

β-CDs are truncated cone-shaped molecules composed of seven α(1-4)-glucose units. The 7 primary alcohols functions at C-6 position and the 14 secondary alcohols at the C-2 and C-3 positions constitute the primary and secondary faces of the macrocycle, respectively. This conformation forms a hydrophobic central cavity suitable for the inclusion of various organic molecules and a hydrophilic external surface. The main characteristic of CDs is the ability to insert the hydrophobic part of a guest molecule into the cavity. The formation of this host-guest inclusion complex and its stability also depend of the size and functionalization of the molecule [39].

Supramolecular chemistry of CDs has often been used in enzyme technology [40] with, as illustration, works of Villalonga et al. [41,42,43,44,45]. The authors demonstrated that immobilization of CD on surface of enzymes improved its affinity for the substrate and its thermostability while maintaining a good activity.

In order to overcome the impediment due to this weak interaction between the guest antibiotic and the host CD molecules, we synthesized an adamantyl ciprofloxacin derivative (CIP-Ad. Indeed, the high affinity of adamantyl group for the internal cavity of the macrocycle [46] will im prove the stability of the inclusion complex with the ciprofloxacin. Moreover, such supramolecular interaction should enhance the antibiotic solubility, stability, and bioavailability [47,48]. To develop this new concept, we firstly had to form an inclusion complex between the modified antibiotic (CIP-Ad) and various β-CDs, by preserving the antimicrobial activity.

Complexation of the CIP-Ad with native β-CD I and three methylated β-CD derivatives commercially-available, i.e., 2,3-O-dimethyl-β-CD II, 2,6-O-dimethyl-β-CD III, 2,3,6-O-trimethyl-β-CD IV (Figure 2), was thus performed, and investigated using isothermal titration calorimetry (ITC), in order to evaluate thermodynamic parameters of the interactions. The stoichiometry of the obtained complexes was determined using two dimensions Nuclear Overhauser Effect Spectroscopy (NOESY). The antibiofilm efficacy of DspB-CIP and DspB (2,3-O-dimethyl-β-CDs/CIP-Ad) complex was then tested on a mature Staphylococcus epidermidis biofilm model. To reach the one shot strategy, CDs were grafted on DspB via a covalent coupling strategy. In this context, we synthesized three carboxylic permethylated CDs derivatives with different spacer arms lengths at the C-6 position of the CDs I, II and III (Figure 2). The antibiofilm activity of DspB-CDs conjugates was then investigated to evaluate the influence of such modification on the biofilm matrix.

Finally, we have validated our approach by demonstrating that DspB-2,3,6-O-trimethyl-β-CD/CIP-Ad nanovector exhibits an efficient antibiofilm activity.

2. Results

The use of an adamantyl group was chosen as molecular anchor to enhance the intermolecular interactions between the enzyme-conjugated CD and the antibiotic [46]. Furthermore, it was shown that adamantaplatensimycin (bioactive analogous of platensimycin) conserves an antibacterial activity against methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus faecium [49]. The synthesis of CIP-Ad 4 derivative was obtained in three steps

(Scheme 1). After quantitative amino protection step, an efficient peptide coupling reagent was used in order to overcome the low reactivity of C-3 carboxylic acid of quinolone core. After deprotection step, the desired CIP-Ad 4 was obtained quantitatively.

2.1. NMR Study

The aim of the NMR study was to highlight and describe the inclusion properties and factors affecting the complexation between CIP-Ad 4 with β-CD I. In the structure of β-CDs, H₃ and H₅ protons are located inside the cavity, whereas H₂ and H₄ are outside the torus (Figure 3). The H6 protons of the primary alcohol group are on the narrow side, and the H₁ are in the glycosidic bond plane of β-CD. Therefore, the formation of inclusion complexes is usually highlighted, in the proton NMR spectrum, by the shift and the deformation of H₃ and H₅ NMR signals.

So an assignment of ¹H NMR spectra of the free forms of CIP-Ad 4, β-CD I and the β-CD I/CIPAd 4 complex in D2O/DCl at pH=3 (Figure 3) was achieved by 2D NMR 1H/1H COSY (Figure 1S) as well as ¹H/¹H NOESY (Figure 2S) and the chemical shifts are summarized in Table 1. As expected, β-CD’s H₃, H₅ protons signals were the most affected by the inclusion as their signals were significantly broaden and shifted by Δδ=0.02 ppm and 0.01 ppm respectively. Regarding the CIP-Ad 4 part, the chemical shifts of adamantyl protons H₁₁, H_11’, H₁₂ and the aromatic protons H₂, H₅ and H₈ were modified by the inclusion.

The classic method of Job's Plot, usually used to determine the stoichiometry of this type of complex, was not effective in our case, because of the very low solubility of the CIP-Ad 4 molecule. Nevertheless, from the structural information obtained from the 2D ¹H/¹H NOESY spectrum of the complex, it was possible to determine the orientation of the CIP-Ad 4 molecule in the cavity of the β−CD and to indirectly deduce the stoichiometry of the β-CD I/CIP-Ad 4 complex. The 2D ¹H/¹H NOESY spectrum of the β-CD I/CIP-Ad 4 complex (Figure 4) exhibited cross peaks between the adamantyl protons (H11, H11’, H12) of CIP-Ad 4 and protons H3 and H5 of β-CD I confirming the inclusion of adamantyl group of CIP-Ad 4 in the β-CD cavity. On the contrary, the absence of correlation between H3, H5 of β-CD I and H7a, H7b of CIP-Ad 4, suggested no inclusion of the piperazinyl group.

These results were in favor of a β-CD I/CIP-Ad 4 complex with a 1/1 stoichiometry by the adamantyl side of CIP-Ad 4 (Figure 5).

2.2. ITC studies

ITC experiments were then carried out for the determination of thermodynamic parameters (Ka, ΔH, ΔS) and stoichiometries of CIP-Ad and β-CD I, 2,3-O-dimethyl-β-CD II, 2,6-O-dimethyl-β- CD III and, 2,3,6-O-trimethyl-β-CD IV complexes (Figure 2). Experiments were performed in acidic medium, leading to the protonated form of CIP-Ad 4. The heat generated during the association process was provided in μJ/mol of titrant (host, I, II, III, IV), in the respect of titrant / titrate molar fraction (host, I, II, III, IV / guest CIP-Ad). The experimental titration curve was fitted through a one-site-independent binding model [52]. The value of the apparent binding stoichiometry (n) was extracted by the inflection point of the curve. As all the curves were perfectly sigmoid, an accurate determination of n was obtained.

The thermodynamics parameters were determined at different temperatures (15-60°C) for all β-CDs (I, II, III, IV) hosts. The experimental titration curves of CIP-Ad 4 by β-CD I at 25°C showed an equivalence point for approximately 1/1 molar ratio of β-CD/CIP-Ad, confirming the observed NMR data. Same stoichiometry was obtained for all host molecules (II, III, IV). The fitting of the binding curves gave direct measurement of affinity constant Ka and enthalpy variation ΔH, then, thanks to thermodynamic equations, Gibbs free energy ΔG and entropy variation ΔS (Table 3) were determined [52]. Variation of heat capacity ΔCp values was calculated as the slope of the curve correlated to the enthalpy values versus the temperature.

Usually, to validate the ITC experiments, Wiseman’s “C” parameter [53] has to be between 10 and 500-1000 [53,54]. In the present study, all C values are in the optimal experimental window, excepted for II / CIP-Ad complex (C value around 2-3) (Table 3). However, Turnbull and Daranas demonstrated that even for low affinity system with weak C value from 0.01 to 10 (Ka =10² M^-1), the ITC results are still valide [55]. In this case, three conditions have to be respected: i) a use a sufficient portion of the binding isotherm for analysis; ii) a precise knowledge of the binding stoichiometry n and of the concentrations of ligand and receptor; iii) an adequate ratio of signal to noise. These criteria are filled for the present study with CIP-Ad/β-CD II complex. The host-guest inclusion complex represents the distribution of species in equilibrium giving an average stoichiometry and a binding constant Ka [39,52]. In the present study, this equilibrium constant Ka was calculated for CIP-Ad/β-CD I complex around 10⁵ M^-1 at 37°C (i.e., ΔG ≈ -30 kJ/mol). This strong association constant is due to perfect fitting of adamantyl moiety inside the β-CD cavity. Consequently, the presence of large CIP structure bound to adamantyl group produced no change in the inclusion process, as reported in the literature with other adamantane derivatives (ΔG° ≈ -20 to -30 kJ / mol) [56,57]. Different values of association constant were obtained for methylated β-CDs (II, III and IV) ranging from 10⁴ to 10⁶ M^-1 at 37°C (i.e., ΔG from -23 to -35 kJ/mol). The highest binding constant was observed with the CIP-Ad/β-CD III (Ka = 72.7 10⁴ M^-1), followed by the CIP-Ad/β-CD I (Ka = 21.3 10⁴ M^-1), the CIP-Ad/β-CD IV (Ka = 8.4 10⁴ M^-1), and CIP-Ad/β-CD II (Ka = 1.0 10⁴ M^-1) complexes. The binding strength is clearly dependent on the host molecule functionalization. Such differences have been already reported in the literature with adamantyl derivatives in favor of β-CD [56,57]. We observed a spontaneaous complexation (ΔG<0) for all CIP-Ad/β-CDs complexes (I to IV) with thermodynamically favorable enthalpy and entropy values (ΔH<0 and ΔS>0). Complexations of CIP-Ad with I and IV are enthalpy-driven (|ΔH|>|TΔS|) on all the temperature range studied, while interactions with II and III are entropy-driven (|ΔH|<|TΔS|) for lower temperatures before becoming enthalpy-driven for higher temperatures. A negative ΔCp was obtained for all CIP-Ad/β-CDs complexes (I to IV). The possible driving forces involved in the formation of inclusion complex with CDs are numerous, such as Van der Waals, hydrophobic, electrostatic, and hydrogen bonding, as well as charge–transfer interactions [39,58]. Moreover, others additional constraints, such as release of conformational strain and exclusion of cavity-bound high energy water, can be implicated [39,58]. In literature, a simple evaluation of enthalpy and entropy changes is usually used to distinguish dominance of hydrophobic or Van der Waals forces during complexation; hydrophobic interactions leading to ΔH>0 and ΔS>0 and Van der Waals interactions to ΔH<0 and ΔS<0. In fact, interpretation of thermodynamic parameters is more complex. One reason is that the formation of an inclusion complex is often accompanied by the release of water molecules from the CD cavity, which are at higher level of energy in comparison with those in external medium. This highly-energetic exclusion phenomenon also contributes to the negative enthalpy and entropy changes observed [58]. Others authors traditionally admit that complexation is governed by hydrophobic interactions when the process is entropy-driven, with larger and positive entropy and smaller enthalpy (|ΔH|<|TΔS| with ΔS>0). As opposed, usually Van der Waals interactions would be enthalpy-driven processes with minor favorable or unfavorable entropies interaction (|ΔH|>|TΔS| with ΔS>0 or ΔS<0). Consequently, hydrophobic and Vander Waals interactions would be the two principal forces leading to the complexation for CIPAd/β-CDs (I to IV) complexes. However, study on a large temperature range permitted to access ΔCp and to discriminate the major force. When a negative ΔCp is obtained, hydrophobic bounds are formed as sequence of breakdown water clathrates around apolar molecules corroborating the desolvatation hypothesis of β-CD [58]. In contrast, hydrogen bond between β-CD and the guest makes a positive contribution to ΔCp [59]. Consequently, the large negative ΔCp values observed here for CIP-Ad/β-CDs complexes, confirmed the contribution of hydrophobic interactions as main driving forces.

2.3. Solubility studies

CIP-Ad 4 solubility was measured in water to evaluate its behavior near to physiological pH. Phase-solubility profiles are given in Figure 6A,B for native CD I and derivated II, III, IV, respectively. CIP-Ad is a poorly soluble drug with an intrinsic solubility (S0) in water determined around 1.10-6 M. The solubility of CIP-Ad significantly increased in function of the CD concentration. According Higuchi and Connors classification [60], the solubility curves can be considered as AL-type, that is a linear increase in solubility of CIP-Ad versus CD concentration is obtained. When 1/1 complexation is supposed between drug and CD, a linear correlation is attempted between the total solubility of drug (St) and the total concentration of cyclodextrin ([CD]t) in the aqueous medium (equation 1):

$${\mathrm{S}}_{\mathrm{t}}={\mathrm{S}}_0+\frac{\mathrm{K}1/1\mathrm{S}0}{1+\mathrm{K}1/1S0}\left[CD\right]t$$

where S₀ is the intrinsic solubility of CIP-Ad (i.e. the solubility when non CD is present) and K_1/1 the apparent stability constant of the drug/CD complex. The plot of St versus [CD]_t gives a straight line with a slope K_1/1 S₀ / (1 +K_1/1 S₀)) less than unity and an intercept (S_int) equal to S₀. Next K_1/1 can be determined from the slope and S₀ according the equation 2:

$${\mathrm{K}}_{1/1}=\frac{\mathrm{S}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{e}}{\mathrm{S}0(1-\mathrm{s}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{e})}$$

Eq 2 according to Loftsson et al. [61]

Linear fit with a slope less than unity was obtained for all CDs, in the range of CD concentration studied, confirming the 1/1 stoichiometry of interaction. However, Sint values were far from S0 for all CDs and a negative value was obtained for host CD IV. This negative deviation from linearity at low CD concentration (i.e. Sint < So), reported as A-L-type phase-solubility profile in literature, is currently observed for poorly soluble drugs (S0 < 1.10-4 M) [61]. It could be due to the nonideality of water as a solvent, the self-association of the drug molecules and of the drug/CD complexes, as well as non-inclusion complexation phenomena [61]. The use of incorrect Sint values results in an overestimation of K (or even negative K in the case of negative Sint) [61]. Consequently, the determination of K was here made from experimental value of S0. K values around 66.10⁴ M^-1, 6.10⁴ M^-1, 52.10⁴ M^-1 and 37.10⁴ M^-1 were obtained for I, II, III and IV host molecules respectively. In comparison with ITC results, slightly higher values (excepted for III) were obtained, which could be attributed to the more hydrophobic form of CIP-Ad 4 in water than in acidic buffer (that is, the net charge of molecule is lower).

2.4. Antimicrobial efficacy of Ciprofloxacin

In order to thereafter evaluate the antimicrobial efficacy of CIP against S. epidermidis, Minimal

Inhibitory Concentrations (MICs) values of the antibiotic were measured through the microdilution method based on ISO standard 20776 [62]. Experiments were performed in microplates, filled with Tryptic Soy Broth (TSB), with an initial bacterial concentration of 10⁶ CFU/mL. Values ranging from 0.25 to 4 μg/mL were obtained for CIP depending on the tested strain (Table 4).

Strain 5 was the most sensitive and was consequently used for most of the further investigations. It is well known that biofilms correspond to high cell-density systems. It is the reason why the efficacy of high CIP concentrations was tested against high cell- density planktonic cultures. Experiments were performed in 96-well microplates containing TSB inoculated with an initial bacterial concentration of about 109 CFU/ mL. Bacteria were enumerated by plating out on Tryptic Soy Agar medium after incubation for 24h at 37°C. Time-killing curves are given on Figure 7. The exposure of planktonic cells to 10 x MIC (i.e., 2.5 μg/mL) of antibiotic during 24 hours, allowed a reduction of 3 log of the bacterial population.

The increase of the antibiotic concentration to 20 x MIC did not significantly improve the treatment efficacy. For this reason, the highest tested CIP concentration was set to 2.5 μg/mL for further experiments.

Biofilms were formed in 96-well microplates containing 200 μl of TSB, as previously described [63]. The ability of S. epidermidis species to produce β-(1-6)-N-acetyl-D-glucosamine polymer (PNAG), results from the presence of the ica locus, which controls the production of the polysaccharide intercellular adhesin (PIA) [1]. The detailed structural elucidation of PNAG has been so far performed on different S. epidermidis [64,65]. Strain 5 was isolated from an infected orthopedic prosthesis [66]. Biofilms formed by this strain contained considerable amounts of PNAG as testified by very similar 1H NMR data that observed for the RP62A and 9142 strains [64,65].

The ability of strain 5 to form a biofilm was confirmed by measuring the sessile population in microplates after 24 hours of incubation. The results demonstrated that the strain was able to form a biofilm, the sessile population reaching about 1 x 10⁹ CFU/mL (SE n= 8). This bacterial concentration has been consequently used to test the different compounds on planktonic cells. The antimicrobial effect of CIP and DspB was then investigated on sessile microorganisms (strain 5). After 24h of incubation, wells were washed with Phosphate-Buffered Saline, and filled with 200 μL of TSB enriched with 5 μg/mL DspB and/or increasing amounts of CIP. After 2h of treatment at 37°C, the number of adherent bacteria before and after treatment was compared. Results given in Figure 8 confirmed the high resistance of sessile organisms to CIP since only 1 log kill was observed after exposure for 24 hours to a high antibiotic concentration (12.5 μg/mL, that is 50 x MIC).

DspB used alone as control, did not exhibit an antimicrobial activity. DspB and CIP exerted a synergistic effect against the biofilm, confirming bibliographic data [28,29,30]. Used together at the concentrations of 5 μg/mL DspB and 2.5 μg/ mL CIP (that is 10 x MIC), a bactericidal effect (5 log kill) was indeed observed on sessile organisms after exposure for 24 hours. This result showed that DspB increases the bacteria susceptibility to CIP. Finally, MICs of the CIP-Ad 4 were evaluated on the four S. epidermidis strains. The corresponding MICs are given in Table 5.

Results points out that, for the four strains used, an antibacterial activity of the CIP-Ad 4 was observed, with higher MIC values than those measured for CIP, however (Table 4). Here again, strain 5 was the most sensitive strain.

2.5. Combined effect of Dsp B and β-CD II/CIP-Ad complex against biofilms

The limited bibliographical data pertaining to the CIP interaction with CDs (e.g., dissolution, absorption) makes difficult the prediction of the drug delivery processes. Several mechanisms can contribute to drug release after parenteral administration [67,68], such as drug dilution, competitive displacement by endogenous materials, drug binding to plasma and tissue components, and drug uptake into non targeted tissues. The pharmacokinetic drug properties should be unaffected by the use of CDs, when binding constant is below 105 M–1 [69]. Consequently, two candidates appeared particularly suitable for further studies: the 2,3-O-dimethyl-β-CD (II)/CIP-Ad and the 2,3,6-O-trimethyl-β-CD (IV)/CIP-Ad complexes.

MICs of the two complexes against the 4 strains are given on Table 6. Results showed that complexation of CIP-Ad 4 with the 2,3-O-dimethyl-β-CD (II) did not alter the CIP-Ad derivative activity while complexation with the permethylated β-CD (IV) increased the MIC value on the four tested strains. For these reasons, the 2,3-O-dimethyl-β-CD (II)/CIP-Ad complex was used for further experiments. Again, strain 5 exhibited a higher susceptibility.

The susceptibility of planktonic bacteria against the β-CD II/CIP-Ad complex was then evaluated as described above. The concentrations used corresponded to 2.5 x and 10 x MIC CIP-Ad. Results (Figure 9) exhibited an expected increase in the antimicrobial activity as the complex concentration increased, the 24 hours exposure to the complex at 10 x MIC, leading to more than a 3log unit reduction of the bacterial population.

Before investigating the antimicrobial activity of the association DspB / β-CD II/CIP-Ad complex, on sessile organisms, the efficacy of the alone complex was evaluated (Figure 10).

Results showed that this complex exhibited the same activity against sessile organisms than the CIP (see Figure 8). For example, about 1 log unit decrease of the bacterial population was observed for 50 x MIC. When used in association with 5 μg/mL DspB, a reduction of 2 log unit of the sessile population was observed after 24 hours of exposure (Figure 10). These data confirmed the positive effect of the association, a control experiment confirming the absence of antimicrobial activity of DspB when used alone.

2.6. Synthesis of CD derivatives CD1-3

In order to elaborate DspB-β-CDs conjugates, we synthesized three carboxylic CDs derivatives with different spacer arm lengths, i.e., with 2, 4 and 10 atoms of carbon, named CD1, CD2 and CD3, respectively (Scheme 2). We focused this preliminary study on permethylated cyclodextrins, because easier to purify in large quantity. From this 6-O-monofunctionalization, the DspB was grafted via an amidation reaction (Scheme 3). The strategy used for the functionalization of the β-CD core involved the synthesis of the well-known mono-6-O-tosyl-β-CD precursor 5 [70]. After the introduction of the azide function, the macrocyle 6 was permethylated as described in the literature. A reduction step afforded the monoamino derivative 8 [70], ready for a coupling reaction with succinic anhydride [71], monomethyl adipic acid ester a or monomethyl dodecanedioic acid ester b at room temperature [45]. The corresponding 6-monoalkylcarboxyl-β-CDs derivatives, CD1, CD2, and CD3, were obtained after a saponification step.

The immobilization of the DispersinB® on the three macrocycles (CD1, CD2 and CD3) was performed with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDAC) and Nhydroxysuccinimide (NHS) in borate buffer (Scheme 3). After ultrafiltration of enzymatic solution, the modified DspB was purified using a 2-D Clean-Up Kit (GE Healthcare).

Mass spectrometry analyses confirmed the expected masses of 69994; 48106 and 51171 g/mol for CD1, CD2 and CD3 with maximum rates of 17, 4 and 9 molecules per enzyme, respectively (data not shown). Native DspB was never found on MS spectra (data not shown), suggesting high grafting rate of DspB onto modified CDs.

2.7. Enzymatic biofilm dispersion assays by DspB-β-CDs conjugates

Tests were adapted from the method implemented by O’Toole and Kolter [72]. Briefly, bacteria were grown for 24 h in TSB at 37°C in microplates. Unattached cells were removed by rinsing the wells thoroughly with PBS, and biofilms were subsequently stained by incubation with Crystal Violet (CV). CV was then solubilized by adding ethanol and the OD of the solution measured at 550 nm.

The antibiofilm activity of the three DspB-β-CDs conjugates (that is bearing the 3 different arms) is given in Figure 11. Results showed that the conjugate bearing the shorter arm (CD1), exhibited the lower activity (highlighted by high OD values), whereas no alteration of the enzyme activity was observed with the two other conjugates (CD2 and CD3), as compared with DspB alone. As expected, this activity was concentration dependent. The enzymatic alteration observed with CD1 is probably due to a steric effect, the high number of CDs grafted onto the enzyme (17 per enzyme molecule) and the proximity of CDs restricting substrate accessibility to the active sites.

These results were then confirmed on the three other S. epidermidis strains (Figure 12). Results showed that the conjugate was efficient on all biofilms, confirming that the three strains produced PNAG. Here again, the DspB-CD1 conjugate exhibited the lowest antibiofilm activity.

2.8. Effect of the DspB-β-CD3/CIP-Ad conjugate on biofilm of S. epidermidis

Finally, we tested the antimicrobial activity of the nanovector DspB-β-CD3/CIP-Ad conjugate on strain 5 biofilms. A CIP concentration of 5 x MIC, i.e., 10 μg / mL was used. A significant activity of the vector was observed (Figure 13) on a 24 h-old biofilm, since about a 3 log reduction of the sessile bacterial population was observed after 24 hours of exposure. This activity was logically lower than that recorded with the association DspB/CIP (Figure 8) as a lower CIP concentration was used. The efficiency of the nanovector is similar to that obtained when the 2,3-O-dimethyl-β-CD (II)/CIP-Ad complex was used in association with the DspB (Figure 10) which demonstrates that chemical immobilization of CDs on DspB doesn't alter its antibiofilm properties.

3. Discussion

Complexation assays between adamantyl group grafted on ciprofloxacin, CIP-Ad, and four β- cyclodextrins derivatives, i.e, β-CD, 2,3-O-dimethyl-β-CD, 2,6-O-dimethyl-β-CD, and 2,3,6-O-trimethyl-β-CD, were investigated. A phase solubility study confirmed an increase of the CIP-Ad solubility with CD concentration, pointing out to a complex formation. NMR and ITC experiments showed a stoichiometry 1/1 of the β-CD/CIP-Ad complexes with an affinity depending on the type of methylation of β-CD. Due to their moderate affinity constant, 2,3-O-dimethyl-β-CD/CIP-Ad and permethylated-β-CDs/CIP-Ad complexes were selected for the antibacterial studies. The evaluation of the antibacterial activity of both CIP-Ad and β-CD derivatives/CIP-Ad complexes was performed on different S. epidermidis strains. MICs values showed that the 2,3-O-dimethyl-β-CDs did not alter the CIP-Ad activity, while complexation with permethylated-β-

CDs increased MIC values. Antibiofilm assays, then performed on S. epidermidis biofilms, confirmed the synergistic effect observed with the association of DspB/CIP which was partly maintained in presence of the complex formed between the modified antibiotic CIP-Ad and the 2,3-Odimethyl-β-CD. DspB was then modified with different CDs conjugates through chemical covalent coupling with carboxylic permethylated CDs derivatives with various spacer length. We showed that theses modified DspB with the CD conjugates maintained their antibiofilm activities. We validated our approach by demonstrating that the DspB-permethylated-β-CD/ciprofloxacin-Ad nanovector exhibited an efficient antibiofilm activity. Biofilm-associated infections are particularly problematic because sessile bacteria are able to withstand host immune defense response and are drastically more resistant to antimicrobials than their planktonic counterparts [4,73].

This is all the more true in the context of implanted medical device (for example, prosthesis or catheters), nocosomial, or depressed immune system related infections. Among the Staphylococcal species that appear in the list of leading etiologic agents, S. epidermidis is the second after Staphylococcus aureus [1,2,3,73], hence our choice. In the last decade, several innovative strategies have been published to fight against biofilms, some of them being patented. Among these strategies, modifications of surfaces to prevent adhesion and / or kill adherent bacteria have been proposed [74,75]. However, these approaches are not completely satisfactory due to their limited efficiency. Consequently, the main anti-infection tool used today in the medical field remains the antibiotherapy approach with often highly limited efficiency due to the high resistance of sessile bacteria to most antibiotics [4]. Faced with this assessment, anti-biofilm enzyme based approaches (as the nanovector presented here) constitute a promising alternative. Due to their ability to degrade the biofilm polymer matrix, they highly increase the efficiency of some antibiotics by both modifying the environmental conditions within the biofilm (for example, dissolved oxygen concentration levels) and by increasing drug diffusivity. Some of these enzymes are already marketed. Thus, the pulmozyme® (dornase alpha) a recombinant human deoxyribonuclease I (rhDNase), is inhaled to improve lung function in cystic fibrosis [76] and mechanically ventilated patients [77]. However, as already mentioned, the main drawback of this enzymatic antibiofilm strategy is the risk of live bacteria dispersal, hence the use of antibiotics in association. Though future investigations are required in order to improve the efficacy of the approach, the present concept presents the high advantage of being an “all-in-one” tool, and constitutes a real progress compared to conventional antibiotherapy treatments used by clinicians today.