* Université Catholique de Louvain, Faculty of Medicine, Unité de Pharmacologie Cellulaire et Moléculaire, Brussels, Belgium; Université Catholique de Louvain, Faculty of Agronomy, Unité de Chimie des Interfaces, Louvain-la-Neuve, Belgium; Faculté Universitaire des Sciences Agronomiques de Gembloux, Unité de Chimie Biologique Industrielle, and Centre de Biophysique Moléculaire Numérique, Faculté Universitaire des Sciences Agronomiques de Gembloux, Gembloux, Belgium; ¶ Université Catholique de Louvain, Louvain-la-Neuve, Faculty of Sciences, Unité de Chimie Structurale et des Mécanismes Réactionnels, Belgium; and || Université Libre de Bruxelles, Faculty of Sciences, Unité de Structure et Fonction des Membranes Biologiques, Brussels, Belgium
An Open Access Article : Biophysical Journal 94:3035-3046 (2008).
Probing drug/lipid interactions at the molecular level representsan important challenge in pharmaceutical research and membranebiophysics. Previous studies showed differences in accumulationand intracellular activity between two fluoroquinolones, ciprofloxacinand moxifloxacin, that may actually result from their differentialsusceptibility to efflux by the ciprofloxacin transporter. Inview of the critical role of lipids for the drug cellular uptakeand differences observed for the two closely related fluoroquinolones,we investigated the interactions of these two antibiotics withlipids, using an array of complementary techniques. Moxifloxacininduced, to a greater extent than ciprofloxacin, an erosionof the DPPC domains in the DOPC fluid phase (atomic force microscopy)and a shift of the surface pressure-area isotherms of DOPC/DPPC/fluoroquinolonemonolayer toward lower area per molecule (Langmuir studies).These effects are related to a lower propensity of moxifloxacinto be released from lipid to aqueous phase (determined by phasetransfer studies and conformational analysis) and a marked decreaseof all-trans conformation of acyl-lipid chains of DPPC (determinedby ATR-FTIR) without increase of lipid disorder and change inthe tilt between the normal and the germanium surface (alsodetermined by ATR-FTIR). All together, differences of ciprofloxacinas compared to moxifloxacin in their interactions with lipidscould explain differences in their cellular accumulation andsusceptibility to efflux transporters.
Due to their ability to accumulate inside phagocytes (1,6–8),fluoroquinolones are also useful for eliminating facultativeintracellular pathogens that resist phagocytic death. We recentlyshowed that fluoroquinolones accumulate in macrophages and showactivity against a large array of intracellular organisms includingListeria monocytogenes and Staphylococcus aureus (9). Quitesignificant differences among closely related derivatives havebeen observed with the following ranking in cellular accumulationand intracellular activity: ciprofloxacin < levofloxacin< garenoxacin < moxifloxacin (9). So far, to our knowledge,this has not received satisfactory explanation.
Characterization of fluoroquinolones uptake by eukaryotic cellssuggested that both passive diffusion and active transport systemsare involved. The transbilayer diffusion of fluoroquinoloneshas been demonstrated (10) and our group reported that ciprofloxacin,but not moxifloxacin, is subject to constitutive efflux in J774macrophages through the activity of an MRP-related transporter(11).
Drug/lipid interactions can modulate not only translocationof the drug through the natural membranes but also its interactionwith efflux proteins (12,13). In this respect, it is well knownthat 1), substrates have to be transported from the lipid bilayerto the transporter protein before a capture mechanism of thedrug by the inner leaflet of the cytoplasmic membrane (14);and 2), the activity of transporter is critically dependenton the surrounding lipid bilayer environment (15,16), whichmay be modified by drugs.
In view of the critical role of lipids for the drug cellularuptake and differences observed for two closely related compounds,ciprofloxacin and moxifloxacin (Fig. 1), we investigated theinteractions of these two fluoroquinolones with lipids, usingan array of complementary techniques. For both ciprofloxacinand moxifloxacin, atomic force microscopy (AFM) reveals an erosionof dipalmitoylphosphatidylcholine (DPPC) domains within dioleoylphosphatidylcholine(DOPC) fluid phase while Langmuir studies show a condensingeffect. Further molecular studies show that fluoroquinolonescan 1), exchange from lipids to aqueous phases (phase transferand molecular modeling studies); 2), decrease the all-transconformation of lipid acyl chain (attenuated total reflectionFourier transform Infra-Red (ATR-FTIR)); and 3), increase thelipid disorder (ATR-FTIR). When the effects of the two fluoroquinolonesare compared, it clearly appears that moxifloxacin has a highercondensing effect related to a lower propensity to be releasedin the aqueous phase from lipid monolayer and to a higher abilityto decrease the all-trans conformation of lipid acyl chain withoutmarked effect in lipid-chain orientation. All together, differencesof ciprofloxacin as compared to moxifloxacin in their interactionswith lipids can be related to differences in their cellularaccumulation and therefore activity against intracellular bacteria.
Ciprofloxacin content was determined by a fluorimetric method(ex, 275 nm; em, 450 nm, using a model No. LS-30 FluorescenceSpectrophotometer; Perkin-Elmer, Beaconsfield, UK) as describedpreviously (11). Under these conditions, our assay had a lowerdetection limit of 5 ng/ml, a linearity (r2 0.99) up to 200ng/ml, and an intraassay reproducibility of 97%.
Fluorescent assay based on the same technique as that used forciprofloxacin but using ex = 298 nm and em = 504 nm; lower limitof detection, 5 ng/ml; linearity up to 450 ng/ml (r2 0.99);intraassay reproducibility, 98%.
Preparation of liposomes (MLVs, SUVs)
Lipid vesicles were prepared as described previously (17). Briefly,appropriate lipids were mixed in CHCl3/CH3OH 2:1 (v/v), evaporatedunder nitrogen flow, and desiccated under vacuum for at least4 h. The dried films were then resuspended at room temperaturefrom the walls of the glass balloon by vigorous vortexing inaqueous buffer. Lipid in suspension flushed with nitrogen werekept in a water bath for 1 h at 37°C for pure DOPC or 45°Cfor liposomes containing DPPC. This procedure yields multilamellarvesicles (MLVs). The small unilamellar vesicles (SUVs) consistingof DOPC/DPPC (1:1 mol) were prepared from MLVs. The preparation,cooled down by an ice bath, was sonicated to clarity five timesfor 2 min each using a Fisher Bioblock Scientific 750 W sonicator(Avantec, Illkirch, France) set at 35% of the maximal power,and a 13-mm probe. The SUVs preparation was then filtered on0.2 µm Acrodisc filters (Ann Arbor, MI) to eliminate titaniumparticles. The concentration of lipids, the nature of the buffer,and the lipid/drug ratio was adjusted for each type of experiment.
Mica sheets were heated 1 h before fusion at 60°C and cleavedto obtain a flat and uniform surface. The SUV suspension ofDOPC/DPPC (1:1) (10 mg lipids/ml; buffer 10:100:3 mM, pH 7.4),was put into contact with the mica surface for 45 min at 60°Cand the sample was slowly cooled back to room temperature toprevent thermal shock. The excess of SUVs was then eliminatedby four-times rinsing with a Tris/NaCl 10:100 mM buffer, pH7.4. The sample was installed on the microscope without dewetting.The liquid meniscus was completed with the same buffer or solutioncontaining 0.2 mM fluoroquinolones. All AFM measurements werecarried out at room temperature in contact mode using an opticaldetection system equipped with a liquid cell (Nanoscope IV;Digital Instruments, Santa Barbara, CA). Topographic imageswere taken in the constant-deflection mode using oxide-sharpenedmicrofabricated Si3N4 cantilevers (Park Scientific Instruments,Mountain View, CA) with typical curvature radii of 20 nm andspring constant of 0.01 N/m. Scan rate ranging from 4 to 6 Hzwere tested. The applied force was maintained as low as possible(<1 nN) during the imaging. All images were flattened.
Partition of fluoroquinolones—phase transfer assay between aqueous and lipid phases
For the phase transfer assays, 1 ml of the water phase (TrispH7.4, 10mM), containing the fluoroquinolone (1 µM), wasmixed by vortexing for 30 s, to 1 ml of organic phase (CHCl3),with or without lipids (Egg yolk phosphatidylcholine (PC)) froma lipid/drug ratio of from 0.1:1 up to 50:1. Both phases weredecanted overnight at 4°C. The fluoroquinolone recoveredfrom the water, interfacial, and organic phases were quantifiedby fluorimetry (model No. LS30 fluorimeter; Perkin-Elmer).
Interaction between fluoroquinolones and a model membrane by molecular modeling: the IMPALA procedure
The ciprofloxacin or moxifloxacin molecule was inserted intoan implicit simplified bilayer using the IMPALA method describedpreviously (18). This method simulates the insertion of anymolecule into a bilayer by adding energy restraint functionsto the usual energy description of molecules. The lipid bilayerwas defined by C(z), which represents an empirical functiondescribing membrane properties. This function is constant inthe membrane plane (x and y axes), but varies along the bilayerthickness (z axis) and, more specifically, at the lipid/waterinterface corresponding to the transition between lipid acylchains (no water = hydrophobic core) and the hydrophilic aqueousenvironment,
Two restraints were imposed to simulate the lipid membrane:the bilayer hydrophobicity (Epho); and the lipid perturbation(Elip).
The hydrophobicity of the membrane is simulated by Epho,
The perturbation of the bilayer by insertion of the moleculewas simulated by the lipid perturbation restraint (Elip),
The environment energy (Eenv) applied on the drug that insertsinto the membrane becomes equal to
Diagrams showing the restraint values versus the angle betweenthe helix axis and the bilayer normal or versus the penetrationof the mass center are obtained as follows: for each degree(angle) or for each 1/10 Å (penetration), the lowest restraintvalue obtained during the Monte Carlo simulation is taken. Allthe points are then joined to generate a profile of the simulation.
Calculations are performed on an Intel Pentium 4, CPU 3.80 GHz,4.00 GB of RAM. The calculation software has been developedat the CBMN (Gembloux, Belgium). Molecular graphs were drawnusing WinMGM 1.0 (Ab Initio Technology, Obernai, France) andSigmaplot 5.0 (SPSS, Chicago, IL) was used for data analysis.
Surface-pressure isotherms of lipid monolayer—Langmuir trough experiments
An automatically controlled Langmuir trough (KSV Minitrough,KSV Instruments, Helsinki, Finland), equipped with a platinumWilhelmy plate was used to obtain the surface pressure-area(–A) isotherms of monolayers at the air/water interface.The temperature was maintained at 25 ± 0.1°C by anexternal water bath circulation. The volume of the trough was80 ml. The cleanliness of the surface was ensured by closingthe barriers, followed by aspiration of the subphase surface,before each experiment. Each experiment was started when thefluctuation of the surface pressure was <0.1 mN/m duringthe compression cycle. Lipid mixture (DOPC/DPPC (1:1)) and DOPC/DPPCmixture with ciprofloxacin or moxifloxacin at different molarproportions (1:1:0.1, 1:1:0.4, 1:1:1, and 1:1:2) were spreadfrom a 1 mM (1:1:0.1, 1:1:0.4, and 1:1:1 molar ratios) or 2mM (1:1:2) CHCl3/CH3OH (2:1, v/v) solution on a Tris 10 mM subphaseadjusted at pH 7.4. Thirty minutes were allowed for solventevaporation from the interface. The air/water interface wasthen compressed with two Delrin barriers at a rate of 5.8 Å2molecule–1 min–1. The reproducibility of the areavalues remained 7%. The accuracy on surface pressure was within0.1 mN/m.
Mean molecular area (A) of the components at the interface wascalculated taking into account the percentage of fluoroquinoloneremaining at the interface after 30 min. The following equationwas used:
Release of fluoroquinolones from lipid monolayer to aqueous phase—Langmuir experiments
The determination of the release of fluoroquinolones from lipidmonolayer into the subphase (Tris buffer 10 mM, pH 7.4) wasperformed as described previously. In these experiments, mixedsolutions of DOPC/DPPC/fluoroquinolone at different molar proportions(1:1:0.1, 1:1:0.4, 1:1:1, and 1:1:2) were spread from a CHCl3/CH3OH(2:1, v/v) solution on the subphase until a surface pressureof 11.6 ± 0.8 mN/m was reached. Although this pressureis well below the estimated surface pressure of a biologicalmembrane (31–34 mN/m (19,20)), it allows an accurate determinationof the kinetics of the transfer of fluoroquinolones to the subphase.Immediately after spreading and every 5 min, two aliquots of500 µl were taken from the subphase with a micropipette.Homogenization of the subphase was provided by a gentle constantstirring. Each experiment was replicated at least three times.The presence of lipids was detected by phospholipid assay andfluoroquinolones were assayed by fluorimetry.
Conformation and orientation of lipids in interaction with fluoroquinolones—ATR-FTIR spectroscopy
Attenuated total reflection Fourier transform infra-red (ATR-FTIR)is particularly well suited for the study of membranes and tocharacterize the effect of drug interacting with lipids on conformationand orientation of acyl chains of phospholipids (21). This techniqueis based on internal reflection of the infrared light withinan internal reflection germanium plate, which creates an evanescentfield at the surface of the plate where the lipid bilayer (andeventually the bound proteins or drugs) resides (22). Afterdeposit of lipids on the germanium plate, while evaporating,capillary forces flatten the membranes which spontaneously formoriented multilayer arrangements (23). The internal reflectionelement was a 52 x 20 x 2 mm trapezoidal germanium ATR plate(ACM, Villiers St. Frédéric, France) with an apertureangle of 45° yielding 25 internal reflections.
Infra-red spectra were obtained on a model No. IFS55 FTIR spectrophotometer(Bruker, Ettlingen, Germany) purged with N2 (as described previously(24)). Spectra were recorded with 2 cm–1 spectral resolutionwith a broad-band MCT detector provided by Bruker between 4000and 800 cm–1; 128 scan were averaged for one spectrum.A modified continuous flow ATR setup was equipped with a polarizerthat can be oriented parallel or perpendicular to the incidenceplane. Fifteen microliters of the sample containing DPPC withdifferent lipid/antibiotic molar ratios (DPPC/drug ratio: 1:0,1:0.2, 1:0.5, 1:1, and 1:2) were dried under a stream of nitrogenon one side of the germanium internal reflection element usingan incident angle of 45° at 20°C. An elevator undercomputer control made it possible to move the whole setup alonga vertical axis (built for us by WOW Company, Nannine, Belgium).The software used for data processing was written under MatLab6.5 (The MathWorks, Natick, MA). All spectra were correctedfor water vapor contribution and CO2 and finally apodized ata resolution of 4 cm–1.
To analyze the conformation of lipids, nonpolarized spectrawere recorded. The hydrocarbon chain in -position of DPPC inthe gel state is in all trans from the ester group down to themethyl group. This conformation allows a resonance to occurbetween the ester group and the CH2 groups of the chain, givingrise to the so-called w(CH2) progression between 1200 and 1350cm–1 (peaks at 1200, 1221, 1246, 1266, 1286, 1309, and1330 cm–1). The proportion of the -chains in the all-transconformation was evaluated from the area of the band at 1200cm–1 relative to the C-H stretching vibrations of theCH2 and CH3 at 3000–2800 cm–1.
To get information about the orientation of lipids and chainordering, dichroic spectra of DPPC and DPPC in interaction withciprofloxacin or moxifloxacin were obtained by subtracting thespectrum measured with a perpendicular polarization of the incidentlight (A) from the spectrum of absorbance measured with a parallelpolarization of the incident light (A//). The angle betweenthe molecular axis and the membrane normal was calculated asreviewed in Goormaghtigh et al. (21) using the STDWAVE programdeveloped in the laboratory of one of the authors.
In the absence of drug, classical images previously published(13) were obtained. They displayed two discrete height levelsreflecting phase separation between gel phase DPPC and liquid-crystallinephase DOPC. The DPPC gel domains were well defined and homogenous,with a size ranging from 0.15 to 1.5 µm. The height differencebetween DPPC domains and the fluid DOPC matrix was 1.10 ±0.05 nm.
When bilayers were incubated with either ciprofloxacin (toppanels) or moxifloxacin (bottom panels), we observed a decreaseof the size of DPPC domain with time (Fig. 2). The height differencesbetween gel phase DPPC and fluid phase DOPC remains constantduring the incubation. We note that for all incubation timesthe bilayer surface was devoid of defects, i.e., holes in theupper monolayer or in the bilayer were never observed. To evaluatethe kinetics of the erosion process, a plot of the average domainareas was represented as a function of time. Fig. 3 shows thatwithin a few hours, the area of DPPC domains decreased from114.1 to 75.6 µm2 and 83.3 to 42.3 µm2 for ciprofloxacinand moxifloxacin, respectively. Thus, ciprofloxacin induceda decrease of the surface occupied by the DPPC domain of 27%.This erosion process was more marked with moxifloxacin sinceit reached a value of 58%. The kinetic trend was also differentfor ciprofloxacin compared to moxifloxacin, as reflected bythe linear and the exponential-like processes, respectively.This suggests that, in contrast with ciprofloxacin, differentregimes of erosion had to be distinguished with moxifloxacin.To assess whether such an alteration of DPPC domains could bedue to mechanical perturbation by the scanning tip, we performedthe same measurements on a control bilayer that was not incubatedwith drugs. We obtained an area decrease of 3% indicating thatthe time-dependent erosion of the DPPC gel domains is due tothe action of the antibiotics rather than to a simple scanningeffect.
Partition of ciprofloxacin and moxifloxacin measured by phase transfer
The differences in behavior of ciprofloxacin and moxifloxacinobserved by AFM experiments might be related to their abilityto partition between aqueous and hydrophobic environments. Tothis end, we followed the transfer of the ciprofloxacin andmoxifloxacin from an aqueous to a lipidic phase, using egg yolkphosphatidylcholine dissolved in chloroform (Fig. 4). Mostly,egg yolk phosphatidylcholine is a mixture of C16 and C18 saturatedalkyl chains at C-1, and C18 unsaturated alkyl chain at C-2.Its thickness and degree of hydration, as well as mean acyl-chainarea, are well known (25,26). Egg yolk phosphatidylcholine iscommonly used to mimic lipid membranes and shows close characteristicsof synthetic lipids used in this study. For example, the thicknessof egg yolk phosphatidylcholine bilayer was estimated to be30 Å, a close value to the thickness of DPPC (29.3 Å)and DOPC (30 Å; deduced from the thickness of DSPC, whichhas the same carbon number in the alkyl chain as DOPC (26)).In the absence of lipid, more than 40% of ciprofloxacin wasdetected in the aqueous phase, while only 5% of moxifloxacinewas found in these conditions. A huge amount of this latterwas found at the interface. The addition of lipids to the organicphase, with a lipid/drug molar ratio up to 50:1 did not changesignificantly the drug phase transfer.
Transfer of fluoroquinolones from lipid monolayer to aqueous phase
To get more insight on the partition of fluoroquinolones betweenlipid and aqueous phases, we investigated the ability of fluoroquinolonesto be released from a lipid monolayer to an aqueous phase, byusing the Langmuir trough technique. The amount of fluoroquinolonefound in the subphase increased with the initial quantity offluoroquinolone in the monolayer (data not shown). For a DOPC/DPPC/fluoroquinoloneratio 1:1:2, a plateau value was reached within 10 min for ciprofloxacinand 15 min for moxifloxacin (Fig. 5). At this equilibrium state,the percentage of fluoroquinolone detected in the subphase wasclearly higher for ciprofloxacin (70%) as compared to moxifloxacin(at 40%). In the experimental conditions used, no lipid wasdetected by phospholipid assay in the buffer that supports thelipid-fluoroquinolone monolayer.
Conformational analysis of the interactions between fluoroquinolones and lipids
In an attempt to correlate our experimental data with molecularmodeling, the interaction of ciprofloxacin and moxifloxacinwith a model membrane was calculated using the IMPALA method.This procedure was used to study the membrane behavior of bothmolecules when crossing the bilayer from the hydrophilic environmentto the hydrophobic.
Fig. 6 A shows the most stable position of each molecule intothe membrane. Both fluoroquinolones are clearly located at thehydrophilic-hydrophobic interface. The molecules were embeddedinto the membrane, with their mass center near the phospholipidheadgroup/acyl-chain interface (13 Å from the bilayercenter), as shown on the plot of the mass center position versusthe restraints (Fig. 6 B). It should be noted that differenceswere seen between ciprofloxacin and moxifloxacin. The interactionof moxifloxacin notably appeared more favorable than that ofciprofloxacin, since the restraint value of the most stableposition was 1.5 kcal/mol lower for moxifloxacin as comparedto ciprofloxacin.
Effect of fluoroquinolones on lipid monolayer—surface-pressure isotherms
To investigate the ability of fluoroquinolones to modify thesurface pressure versus area isotherms of DOPC/DPPC (1:1) monolayers,we investigated the effect of increasing amounts of antibioticson these isotherms curves.
For the sake of accuracy, we took into account the release offluoroquinolones from the lipid to the aqueous phases in thedetermination of the quantity of drug remaining in the monolayerat the air-water interface, on the monolayers isotherms. Wetherefore recalculated the monolayer compression isotherms usingthe proportion of fluoroquinolones remaining in the monolayerafter 30 min. Results are illustrated in Fig. 7.
The curve corresponding to pure DOPC/DPPC (1:1) is in perfectagreement with the one already reported (27). As already evokedby Montero et al. (28), pure ciprofloxacin or moxifloxacin doesnot form a film at the air-water interface. In presence of fluoroquinolones,the isotherms were shifted toward the small molecular areas.This effect is more pronounced with moxifloxacin (Fig. 7 B)than ciprofloxacin (Fig. 7 A).
In addition, fluoroquinolones also affected the collapse pressure:46.0 mN/m for DOPC/DPPC; 45.9 mN/m, 44.8 mN/m, and 37.7 mN/m,for initial proportions of DOPC/DPPC/moxifloxacin of 1:1:0.4,1:1:1, and 1:1:2, respectively; and 46.0 mN/m, 40.5 mN/m, and37.8 mN/m, for initial proportions of DOPC/DPPC/ciprofloxacinof 1:1:0.4, 1:1:1, and 1:1:2, respectively. This disruptionof the lipid monolayer stability at a high compression levelis more pronounced at a high level of fluoroquinolone in themixed monolayer.
Effect of fluoroquinolones on lipid conformation—ATR-FTIR
Because the lower area occupied by lipids in presence of fluoroquinolonesmight be partly due to a change in the orientation of lipidat the interface by straightening up their fatty acid chains,we used ATR-FTIR to investigate the effect of ciprofloxacinand moxifloxacin on conformation and orientation of acyl chainof lipids.
Nonpolarized ATR-FTIR spectra of supported layers of DPPC, drug(ciprofloxacin or moxifloxacin), and DPPC/drug at a molar ratioof 1:1 were recorded (Fig. 8, top panel). As the drug proportionincreased, the drug spectrum appeared in the DPPC/drug mixturespectrum, notably at 1630 cm–1. Interestingly, in DPPC/drugspectra, the DPPC (C=O) band at 1736 cm–1 was modifiedin terms of frequency and shape, suggesting a modification ofthe interfacial lipid carbonyl groups. Analysis of the lipidC-H wagging (w(CH2)) allowed us to get information on lipidchain conformation and proportion of the chains in the all-transconformation (23). Here the wagging band at 1200 cm–1was selected because it has little overlap with other lipidor drug absorption. As shown in Fig. 8, bottom panel, area evolutionof DPPC peak at 1206–1193 cm–1 as function of increasingamounts of fluoroquinolones decreased by up to 60 and 72% forciprofloxacin and moxifloxacin, respectively. These data indicateda loss of all-trans conformation and the appearance of a kinksomewhere between C-2 and C-6 of the chain.
Effect of fluoroquinolones on lipid orientation—ATR-FTIR
To get information on molecular orientation in the absence orin the presence of both fluoroquinolones, we took advantagefrom the fact that, in an ordered membrane deposited on thegermanium crystal (oriented multilayers), all the moleculeshave the same orientation with respect to a normal to the germaniumplate. Measuring the spectral intensity with two orientationsof the incident-light electric field obtained with a polarizerallowed us to obtain information on several chemical groupsof the lipid molecule. The dichroic spectrum of pure DPPC andDPPC/drug (molar ratio 1:1) mixture were obtained by subtractingthe spectrum recorded with perpendicular-polarized light fromthat recorded with the parallel-polarized light using the lipid(C=O) band at 1780–1700 cm–1 as a reference (Fig. 9)(29). Interestingly the dichroic spectra of DPPC/drug mixturedisplayed strong dichroism for bands assigned to the drug, notablyat 1630 and 1465 cm–1, suggesting a well-organized, well-definedorientation of the drug in the DPPC bilayer. The orientationof the lipid acyl chain can be estimated from the wagging band(w(CH2)). The dipole of this transition is oriented parallelto the all-trans chain (23). In turn, positive deviations ofthe dichroism spectrum demonstrate that the chains are mainlyperpendicular to the germanium surface, i.e., perpendicularto the membrane plane since AFM recording demonstrated thatmembranes orient themselves parallel to the germanium surface,even when natural membranes are used (30). In Fig. 9, bottompanel, we plotted the area evolution of the wagging peak integratedbetween 1206 and 1193 cm–1 as a function of the DPPC/drugmolar ratio. Both fluoroquinolones induced a marked and similardecrease of the area when they were added at low concentration(1:0.2 molar ratio). When the amounts of fluoroquinolones wereincreased, the area decreased further in presence of ciprofloxacinbut remained almost stable for moxifloxacin. This observationwas similar for the four wagging peaks (indicated by arrows;Fig. 9, top panel) (1275–1261 cm–1, 1253–1240cm–1, 1229–1216 cm–1, and 1206–1193cm–1).
To quantify the orientation of DPPC all-trans chains, we measuredthe dichroic ratio for wagging band at 1200 cm–1. RATR(A///A) was 6.8 with an isotropic dichroic ratio of 1.33 (calculatedfrom (C=O) band at 1755–1750 cm–1 (29)). On thebasis of this determination, the angle between the acyl chainsof DPPC and the normal at the germanium surface was found tobe 21°. The same calculation was done in the presence offluoroquinolones at a lipid/drug ratio of 1:1. The angle was27° and 20° in the presence of ciprofloxacin and moxifloxacin,respectively. These data suggested that in contrast to ciprofloxacin,moxifloxacin had no effect on the orientation of the acyl chainsand did not induce additional disorder.
The activity of efflux pumps in general, and those involvedfor ciprofloxacin or moxifloxacin cellular accumulation, inparticular, is closely related to the interaction of the drugwith lipids, regardless of the model proposed for their activity(i.e., flippase or hydrophobic vacuum cleaner). We thereforelooked at the influence of the small changes in chemical structurebetween moxifloxacin and ciprofloxacin on their interactionwith lipids. To address this crucial question, the interactionsbetween model lipid membranes and the two fluoroquinolones wereprobed using a variety of techniques.
Nanoscale investigations by AFM revealed different behaviorsfor ciprofloxacin and moxifloxacin. AFM imaging showed a reductionof the size of the DPPC gel phase domains in presence of fluoroquinolones.The erosion process is greater for moxifloxacin compared tociprofloxacin, and follows an exponential-like function-decreasefor moxifloxacin and a linear-decrease for ciprofloxacin. Consideringboth molecules, the difference in their ability to erode theDPPC gel phase domains might be due to a better insertion ofthe moxifloxacin than the ciprofloxacin into the fluid matrix,and a more marked decrease of the line tension at the boundaryof the DOPC/DPPC phases resulting from a fluidification of theDPPC.
This marked effect of moxifloxacin on lipids as compared tociprofloxacin was confirmed by conformational analysis. In accordancewith experimental data obtained for ciprofloxacin (37) or grepafloxacin(38), the fluoroquinolones are located at the lipid-water interface,near the first carbons of the acyl chains. Both molecules showeda minimum of energy when they are at the phospholipids headgroup/acyl-chainsinterface, and the interaction energy rose markedly when themolecule was forced into the hydrophobic domain. This energyincrease was less marked for moxifloxacin as compared to ciprofloxacin,suggesting a higher affinity of moxifloxacin for lipid phase.
Taken together, our data suggest that ciprofloxacin and moxifloxacininteract in a very different way with lipids. The major challenge,however, is to understand the mechanism, at a molecular level,unraveling the interaction between lipids and fluoroquinolonesand the path of these antibiotic molecules through lipid layers.Previous data reported by Montero et al. (28) showed a shiftof the surface pressure-area isotherms of monolayer toward alower area per molecule in the presence of ciprofloxacin. Weextended these data to one major fluoroquinolone used in clinics,moxifloxacin. To go further in the mechanism involved, we monitoredthe amount of fluoroquinolones in the subphase after the monolayercompression. In agreement with the hypothesis based on a dissolvingeffect in the subphase (39), we did find fluoroquinolones thereinwith a higher proportion of ciprofloxacin as compared to moxifloxacin.Taking into account the amount of fluoroquinolones present insidethe monolayer, we corrected the surface-pressure-area isothermsof the monolayer and again observed a shift toward a lower areaper molecule in the presence of fluoroquinolones. This effectwas more marked with moxifloxacin as compared to ciprofloxacin.The dissolving effect in the aqueous phase is therefore probablyessential, although not fully sufficient, to explain the condensingeffect of fluoroquinolones.
Thus, we investigated a change in the lipid chain conformationand orientation using ATR-FTIR technique (21,40). Indeed, thedrug-induced area condensation of lipids can derive from theacyl ordering attained when trans-gauche isomerization aboutthe carbon-carbon bonds is reduced. The trans conformation isthe most stable and has an estimated energy barrier of 3.5 kcal/molto rotate past the eclipsed configuration to the gauche form.The all-trans conformation allows the chain to be maximallyextended, whereas a gauche bond alters the direction of thechain inducing a kink in the chain. Our results clearly indicatedthat moxifloxacin has a higher ability than ciprofloxacin tomarkedly decrease the number of all-trans conformation. Therelated change in the packing of the acyl chains might allowmoxifloxacin to be located in the pocket created by the presenceof a kink in the acyl chain. In contrast, with ciprofloxacin,the appearance of a kink from the all-trans chain conformationwould be less marked, suggesting a less important change inlipid packing. Interestingly, both condensing effects (lowerarea of mixed monolayer lipids/fluoroquinolones) have also beendescribed when cholesterol was added to fluid-phase phosphatidylcholine(41–43).
The disorder in the lipid chains revealed by the decrease ofall-trans conformations has also been analyzed in terms of orientationand tilt between the molecular axis (the membrane normal) andthe transition dipole moments. In this analysis, the ciprofloxacinshowed an additional cause of disorder, because it modifiesthe orientation of the acyl chain in relation to its higherability to be released in an aqueous phase after monolayer compression.
Differences in the charge distribution of the molecule at thephysiological pH could also explain changes for drug membranelocation and bound hydration shell surrounding the headgroupof membrane lipids, which, in turn, could partly explain themore condensing effect of moxifloxacin as compared to ciprofloxacin.Moreover, the determinations were made in aqueous environment,whereas the condensing effect of moxifloxacin involved the presenceof the drug with a lipidic phase.
All together, we showed that the condensing effect of fluoroquinoloneson lipid layer resulted not only from a dissolving mechanismbut also from an alteration of the intramolecular acyl-chainorder in relation to a reduction in trans-gauche isomerizationabout the carbon-carbon bonds, and change in the average moleculartilt of lipid acyl chain of DPPC. The two fluoroquinolones investigatedshowed difference in their effects. Ciprofloxacin had a lowerability to decrease the all-trans conformation of lipid chainsthan moxifloxacin but showed a higher capacity to affect theorientation of lipid chains and to disorder the membrane. Theseeffects might explain its higher ability to be released fromthe lipid monolayer to aqueous phase and its lower effect onsurface pressure-area isotherms of monolayers. In contrast,moxifloxacin has a lower capacity to induce membrane disorderand does not change the tilt between the molecular axis andthe transition dipole moment. Moxifloxacin has also a highertendency to decrease the number of all-trans conformations withincrease of kink, creating a pocket in which moxifloxacin canbe located. This can explain why the amount of moxifloxacinin the aqueous phase was lower than that found for ciprofloxacinand why the mean molecular area of lipids/fluoroquinolones monolayersafter compression is significantly lower in the presence ofmoxifloxacin as compared to ciprofloxacin.
This model is entirely compatible with the physico-chemicalcharacteristics of the two fluoroquinolones. It suggested thatsmall structural differences among fluoroquinolones (notablyoverall molecular hydrophobicity (Papp = 0.089 vs. 0.031 forciprofloxacin and moxifloxacin, respectively (45)), bulkiness,and/or the internal dynamics of the C-7 substituent, could beimportant for drug lipid interactions and lipid packing. Thediazabicyclonyl-ring at position 7 of moxifloxacin, by aligningthe sn-2 chain, probably contributes to the higher tendencyof this antibiotic to induce a decrease of all-trans configurationas compared to ciprofloxacin. This is in line with data reportedwith n-alkyl-piperazinyl-ciprofloxacin (39).
In conclusion, we provided a comprehensive picture of the interactionof the two major fluoroquinolones ciprofloxacine and moxifloxacinewith lipids, and elucidate fundamental issues such as the relationshipbetween lipid chain conformation and orientation with changesin membrane properties as determined by Langmuir studies andthe ability of drugs to diffuse through membranes. All theseparameters might be related to the activity of membranous proteins.Our work notably showed that an increase in drug lipophilicityand addition of a bulky moiety (moxifloxacin versus ciprofloxacin)produced marked changes in the packing of lipids. This was concomitantwith a lower release of the more lipophilic drug from lipidmonolayer and with a potential inefficient activity of effluxproteins which could be involved in a kind of futile cycle resultingin an increase in cellular accumulation (1). So far, progressin understanding the structure-function relationships of membranesand understanding of the lipid-drug interaction appears to beof crucial importance in understanding the mechanisms involvedin cellular drug accumulation.
R.B., E.G., L.L., M.D., and Y.D. are, respectively, Research Directors and Research Associates of the National Foundation for Scientific Research. H.B. is an assistant and doctoral fellow of the Université Catholique de Louvain. The support of the Région Wallonne, of the National Foundation for Scientific Research, of the Université Catholique de Louvain (Fonds Spéciaux de Recherche, Actions de Recherche Concertées), and of the Federal Office for Scientific, Technical and Cultural Affairs (Interuniversity Poles of Attraction Program), is gratefully acknowledged. We also thank Bayer Healthcare AG for providing us fluoroquinolone antibiotics.
Hayet Bensikaddour and Nathalie Fa contributed equally to the work.
Nathalie Fa's present address is Campden & Chorleywood Food Research Association Group, Chipping Campden, Gloucestershire, GL55 6LD, UK.
Editor: Petra Schwille.
Submitted on June 15, 2007; accepted for publication October 3, 2007.
2. Bakker-Woudenberg, I. A., M. T. ten Kate, L. Guo, P. Working, and J. W. Mouton. 2001. Improved efficacy of ciprofloxacin administered in polyethylene glycol-coated liposomes for treatment of Klebsiella pneumoniae pneumonia in rats. Antimicrob. Agents Chemother. 45:1487–1492.
3. Bakker-Woudenberg, I. A., M. T. ten Kate, L. Guo, P. Working, and J. W. Mouton. 2002. Ciprofloxacin in polyethylene glycol-coated liposomes: efficacy in rat models of acute or chronic Pseudomonas aeruginosa infection. Antimicrob. Agents Chemother. 46:2575–2581.
4. Wong, J. P., H. Yang, K. L. Blasetti, G. Schnell, J. Conley, and L. N. Schofield. 2003. Liposome delivery of ciprofloxacin against intracellular Francisella tularensis infection. J. Control. Release. 92:265–273.
5. Rolston, K. V., D. Yadegarynia, D. P. Kontoyiannis, I. I. Raad, and D. H. Ho. 2006. The spectrum of Gram-positive bloodstream infections in patients with hematologic malignancies, and the in vitro activity of various quinolones against Gram-positive bacteria isolated from cancer patients. Int. J. Infect. Dis. 10:223–230.
6. Bounds, S. J., R. Nakkula, and J. D. Walters. 2000. Fluoroquinolone transport by human monocytes: characterization and comparison to other cells of myeloid lineage. Antimicrob. Agents Chemother. 44:2609–2614.
7. Hirota, M., T. Totsu, F. Adachi, K. Kamikawa, J. Watanabe, S. Kanegasaki, and K. Nakata. 2001. Comparison of antimycobacterial activity of grepafloxacin against Mycobacterium avium with that of levofloxacin: accumulation of grepafloxacin in human macrophages. J. Infect. Chemother. 7:16–21.
8. Hara, T., H. Takemura, K. Kanemitsu, H. Yamamoto, and J. Shimada. 2000. Comparative uptake of grepafloxacin and ciprofloxacin by a human monocytic cell line, THP-1. J. Infect. Chemother. 6:162–167.
9. Seral, C., M. Barcia-Macay, M. P. Mingeot-Leclercq, P. M. Tulkens, and F. Van Bambeke. 2005. Comparative activity of quinolones (ciprofloxacin, levofloxacin, moxifloxacin and garenoxacin) against extracellular and intracellular infection by Listeria monocytogenes and Staphylococcus aureus in J774 macrophages. J. Antimicrob. Chemother. 55:511–517.
10. Fresta, M., S. Guccione, A. R. Beccari, P. M. Furneri, and G. Puglisi. 2002. Combining molecular modeling with experimental methodologies: mechanism of membrane permeation and accumulation of ofloxacin. Bioorg. Med. Chem. 10:3871–3889.
11. Michot, J. M., F. Van Bambeke, M. P. Mingeot-Leclercq, and P. M. Tulkens. 2004. Active efflux of ciprofloxacin from J774 macrophages through an MRP-like transporter. Antimicrob. Agents Chemother. 48:2673–2682.
12. Fernandez-Teruel, C., I. Gonzalez-Alvarez, V. G. Casabo, A. Ruiz-Garcia, and M. Bermejo. 2005. Kinetic modeling of the intestinal transport of sarafloxacin. Studies in situ in rat and in vitro in Caco-2 cells. J. Drug Target. 13:199–212.
13. Berquand, A., N. Fa, Y. F. Dufrene, and M. P. Mingeot-Leclercq. 2005. Interaction of the macrolide antibiotic azithromycin with lipid bilayers: effect on membrane organization, fluidity, and permeability. Pharm. Res. 22:465–475.
15. Hinrichs, J. W., K. Klappe, I. Hummel, and J. W. Kok. 2004. ATP-binding cassette transporters are enriched in non-caveolar detergent-insoluble glycosphingolipid-enriched membrane domains (DIGs) in human multidrug-resistant cancer cells. J. Biol. Chem. 279:5734–5738.
17. Laurent, G., M. B. Carlier, B. Rollman, F. Van Hoof, and P. M. Tulkens. 1982. Mechanism of aminoglycoside-induced lysosomal phospholipidosis: in vitro and in vivo studies with gentamicin and amikacin. Biochem. Pharmacol. 31:3861–3870.
19. Demel, R. A., W. S. Geurts van Kessel, R. F. Zwaal, B. Roelofsen, and L. L. van Deenen. 1975. Relation between various phospholipase actions on human red cell membranes and the interfacial phospholipid pressure in monolayers. Biochim. Biophys. Acta. 406:97–107.
24. Fa, N., S. Ronkart, A. Schanck, M. Deleu, A. Gaigneaux, E. Goormaghtigh, and M. P. Mingeot-Leclercq. 2006. Effect of the antibiotic azithromycin on thermotropic behavior of DOPC or DPPC bilayers. Chem. Phys. Lipids. 144:108–116.
25. Komatsu, H., H. Saito, S. Okada, M. Tanaka, M. Egashira, and T. Handa. 2001. Effects of the acyl chain composition of phosphatidylcholines on the stability of freeze-dried small liposomes in the presence of maltose. Chem. Phys. Lipids. 113:29–39.
26. Hara, M., H. Yuan, Q. Yang, T. Hoshino, A. Yokoyama, and J. Miyake. 1999. Stabilization of liposomal membranes by thermozeaxanthins: carotenoid-glucoside esters. Biochim. Biophys. Acta. 1461:147–154.
27. Vie, V., N. Van Mau, E. Lesniewska, J. P. Goudonnet, F. Heitz, and C. Le Grimellec. 1998. Distribution of ganglioside G(M1) between two-component, two-phase phosphatidylcholine monolayers. Langmuir. 14:4574–4583.
30. Ivanov, D., N. Dubreuil, V. Raussens, J. M. Ruysschaert, and E. Goormaghtigh. 2004. Evaluation of the ordering of membranes in multilayer stacks built on an ATR-FTIR germanium crystal with atomic force microscopy: the case of the H+,K+-ATPase-containing gastric tubulovesicle membranes. Biophys. J. 87:1307–1315.
34. Sun, J., S. Sakai, Y. Tauchi, Y. Deguchi, J. Chen, R. Zhang, and K. Morimoto. 2002. Determination of lipophilicity of two quinolone antibacterials, ciprofloxacin and grepafloxacin, in the protonation equilibrium. Eur. J. Pharm. Biopharm. 54:51–58.
35. Neves, P., A. Leite, M. Rangel, B. de Castro, and P. Gameiro. 2007. Influence of structural factors on the enhanced activity of moxifloxacin: a fluorescence and EPR spectroscopic study. Anal. Bioanal. Chem. 387:1543–1552.
36. Furet, Y. X., J. Deshusses, and J. C. Pechere. 1992. Transport of pefloxacin across the bacterial cytoplasmic membrane in quinolone-susceptible Staphylococcus aureus. Antimicrob. Agents Chemother. 36:2506–2511.
38. Rodrigues, C., P. Gameiro, S. Reis, J. L. F. Lima, and B. de Castro. 2002. Interaction of grepafloxacin with large unilamellar liposomes: partition and fluorescence studies reveal the importance of charge interactions. Langmuir. 18:10231–10236.
39. Vazquez, J. L., M. T. Montero, S. Merino, O. Domenech, M. Berlanga, M. Vinas, and J. Hernandez-Borrell. 2001. Location and nature of the surface membrane binding site of ciprofloxacin: a fluorescence study. Langmuir. 17:1009–1014.
40. Pare, C., M. Lafleur, F. Liu, R. N. Lewis, and R. N. McElhaney. 2001. Differential scanning calorimetry and 2H nuclear magnetic resonance and Fourier transform infrared spectroscopy studies of the effects of transmembrane -helical peptides on the organization of phosphatidylcholine bilayers. Biochim. Biophys. Acta. 1511:60–73.
41. Smaby, J. M., M. Momsen, V. S. Kulkarni, and R. E. Brown. 1996. Cholesterol-induced interfacial area condensations of galactosylceramides and sphingomyelins with identical acyl chains. Biochemistry. 35:5696–5704.
42. Smaby, J. M., M. M. Momsen, H. L. Brockman, and R. E. Brown. 1997. Phosphatidylcholine acyl unsaturation modulates the decrease in interfacial elasticity induced by cholesterol. Biophys. J. 73:1492–1505.
43. Bin, X., S. L. Horswell, and J. Lipkowski. 2005. Electrochemical and PM-IRRAS studies of the effect of cholesterol on the structure of a DMPC bilayer supported at an Au111 electrode surface, part 1: properties of the acyl chains. Biophys. J. 89:592–604.
44. Sun, J., S. Sakai, Y. Tauchi, Y. Deguchi, G. Cheng, J. Chen, and K. Morimoto. 2003. Protonation equilibrium and lipophilicity of olamufloxacin (HSR-903), a newly synthesized fluoroquinolone antibacterial. Eur. J. Pharm. Biopharm. 56:223–229.
(A) Structural formula of ciprofloxacin (3-quinolinecarboxylic
and (B) moxifloxacin (3-quinolinecarboxylic acid,
(Click image to enlarge)
AFM height images (20 µm x 20 µm (top panels) or 15 µm x 15 µm (bottom panels),
z-scale: 5 nm) of a mixed DOPC/DOPC (1:1, mol/mol) bilayer recorded in
Tris 10 mM, NaCl 100 mM buffer, pH 7.4 containing 1 mM of ciprofloxacin
(top panels) or moxifloxacin (bottom panels) at increasing incubation time.
(Click image to enlarge)
Evolution of the area of the DPPC domain (Fig. 2) with time for
DOPC/DPPC bilayers incubated in Tris 10 mM, NaCl 100 mM buffer, pH 7.4,
containing 1 mM of ciprofloxacin (solid squares) or moxifloxacin (solid circles).
(Click image to enlarge)
Phase transfer of ciprofloxacin (1 µM, top panel) and moxifloxacin (1 µM, bottom panel) in Tris buffer pH 7.4 against increasing amounts of PC (from 0.1:1 up to a lipid/drug ratio of 50:1) in chloroform. (Open bars) Drug in aqueous phase. (Hatched bars)
Drug at interface (calculated from the difference between initial drug
concentration and measured drug in aqueous and organic phase). (Solid bars) Drug in organic phase. Experiments were reproduced at least three times with similar results.
(Click image to enlarge)
Kinetics of the release of fluoroquinolones from the mixed
lipids/fluoroquinolones monolayer to the subphase (10 mM Tris pH 7.4,
25°C). Results are expressed as the percentage of fluoroquinolones
initially present in the monolayer. Ciprofloxacin, ; and moxifloxacin, •. Molar proportion of DPPC/DOPC/fluoroquinolones (1:1:2).
(Click image to enlarge)
Molecular modeling of the interactions between ciprofloxacine and moxifloxacine with an implicit membrane. (A) Most favorable position of the ciprofloxacin (left) and moxifloxacin (right) in a lipid bilayer. (B) Restraints versus the position of the ciprofloxacin (up) and moxifloxacin (down) in the bilayer. Constraints are expressed in kcal/mol and the positions are expressed in Ångstroms.
(Click image to enlarge)
Surface pressure-molecular area isotherms of DOPC/DPPC in the presence of ciprofloxacin (A) and moxifloxacin (B),
on a subphase of 10 mM Tris at pH 7.4 and 25°C. Mean molecular area
were corrected to take into account the percentage of fluoroquinolone
remaining at the interface. DOPC/DPPC/drug molar ratio were 1:1:0 (continuous line), 1:1:0.4 (discontinuous line), 1:1:1 (dotted line), and 1:1:2 (dash-dotted line).
(Click image to enlarge)
Fluoroquinolones effect on the conformation of DPPC monolayer as revealed by the infrared absorbance spectra. (Top panel) ATR-FTIR spectra of moxifloxacin (a), ciprofloxacin (b), DPPC (c), DPPC/ciprofloxacin (d), and DPPC/moxifloxacin (e). DPPC was used at 50 mg/ml and the molar ratio of lipid/drug was 1:1. (Bottom panel) Evolution of the peak area at 1200 cm–1 (wagging w(CH2) band; integrated between 1206 and 1193 cm–1) as a function of increasing lipid/drug ratio: 1:0, 1:0,2, 1:0,5, 1:1, and 1:2. Ciprofloxacin, ; and moxifloxacin, •.
(Click image to enlarge)
Fluoroquinolones effect on the orientation of DPPC monolayer as revealed by ATR-FTIR dichroic spectra. (Top panel) Polarized ATR-FTIR spectra of DPPC (a), DPPC/ciprofloxacin (b), and DPPC/moxifloxacin (c). DPPC was used at 50 mg/ml and the molar ratio of lipid/drug was 1:1. Wagging w(CH2) bands are indicated by arrows. (Bottom panel) Area evolution of dichroic peak of 1200 cm–1 of DPPC in the presence of fluoroquinolones. Integration area of dichroic w(CH2) band (integrated between 1206 and 1193 cm–1) was plotted versus DPPC/drug molar ratio as indicated, in the presence of ciprofloxacin, ; and moxifloxacin, •.
(Click image to enlarge)