2114
Bioconjugate Chem. 2009, 20, 2114–2122
Annexin A5-Functionalized Liposomes for Targeting Phosphatidylserine-Exposing Membranes Boris Garnier,†,‡ Anthony Bouter,† Ce´line Gounou,† Klaus G. Petry,‡ and Alain R. Brisson*,† Molecular Imaging and NanoBioTechnology, IECB, UMR-5248 CBMN CNRS-University Bordeaux1-ENITAB, Avenue des Faculte´s, F-33402 Talence, France, and INSERM EA-2966, University Bordeaux, 146 rue Le´o Saignat, F-33076 Bordeaux, France. Received June 11, 2009; Revised Manuscript Received September 4, 2009
Long-circulating liposomes functionalized with cell-targeting elements and loaded with bioactive compounds present high interest as drug delivery nanosystems. We present here the synthesis and physicochemical characterization of liposomes containing PEGylated lipids covalently linked to oriented Annexin-A5 (Anx5) proteins, and we show that Anx5-functionalized liposomes are able to target phosphatidylserine (PS)-exposing membranes. The covalent coupling of Anx5 to liposomes is almost quantitative, which is mainly due to the high accessibility of the reacting groups. The influence of Anx5 functionalization on liposome aggregation was investigated by dynamic light scattering, showing that Anx5-functionalized liposomes are stable below a threshold density of 250 Anx5 molecules per liposome. Anx5-functionalized liposomes bind PS-containing membranes with very high efficacy, which is mainly due to the controlled orientation of the Anx5 at the liposome surface. A striking result, obtained by quartz crystal microbalance with dissipation monitoring, is that one single Anx5 molecule is able to anchor a liposome to a PS-containing supported membrane. Finally, we show by fluorescence microscopy that Anx5-functionalized liposomes bind PS-exposing apoptotic K562 cells with high specificity. This study demonstrates that Anx5-functionalized liposomes bind specifically to PS membranes and are thus potential candidates to deliver drug or imaging agents to sites of apoptosis or thrombosis.
INTRODUCTION Liposomes are hollow lipid bilayer vesicles which have found practical applications as drug delivery systems, thanks to their intrinsic property of entrapping either hydrophilic compounds within their aqueous interior or hydrophobic compounds within their hydrophobic lipid boundary (1-5). Drug encapsulation provides several advantages over direct administration, such as preventing drug dilution, avoiding potential deleterious side effects, or reducing the volume of distribution of the drug (6, 7)}. Efforts have been developed to design liposomes that achieve long circulation times in the body in order to increase drug accumulation in the desired tissues and organs. Parameters have been identified which are important for increasing the circulation time of liposomes, in particular, a small (∼100 nm) size and the addition of a hydrophilic poly(ethylene glycol) (PEG) coat. Small size strongly reduces elimination of vesicles by body filters, while the attachment of a brush-like layer of PEG molecules helps them escape capture by cells from the reticuloendothelial system and prevents their accumulation in the liver or the spleen (1). The circulation half-life of PEGylated stealth liposomes extends to several hours, compared to a few minutes for uncoated vesicles (8). The surface functionalization of liposomes with molecular recognition elements able to recognize and bind specific targets constitutes a major challenge in the field of drug delivery systems (6, 9-12). Liposomes exposing several types of cell recognition molecules, from small molecules like folic acid to * Corresponding author. Alain R. Brisson. Mailing address: Bat. B8, Avenue des Faculte´s, F-33402 Talence, France, E-mail: a.brisson@ iecb.u-bordeaux.fr; Tel: +33 540003458; Fax: +33 540002200. † Molecular Imaging and NanoBioTechnology. ‡ INSERM EA-2966.
large proteins like antibodies, have already been produced and used successfully as drug carriers in Vitro as well as in mice (13). Our aim is to develop long-circulating liposomal nanovectors functionalized with protein-based recognition elements. We present here our initial study which concerns the synthesis and characterization of PEGylated liposomes functionalized with the Annexin-A5 (Anx5) protein. Anx5 is the prototype member of the annexin family, proteins which share the property of binding to membranes containing negatively charged phospholipids in a Ca2+-dependent manner (14). Although the exact biological functions of annexins are still being debated, Anx5 has already found wide application both as a research tool and as a clinical tool for labeling cell membranes exposing phosphatidylserine (PS) molecules. It is indeed well-established that eukaryotic cell membranes present an asymmetric distribution of lipid molecules, with the majority of anionic phospholipids, particularly PS molecules, being located in the inner leaflet of the plasma membrane (15). The exposure of PS molecules on the cell membrane outer leaflet is a hallmark of several basic processes, such as apoptosis or programmed cell death (16) and blood platelet activation (17). The ability of Anx5 to bind PS-exposing membranes in a Ca2+dependent manner explains the use of labeled Anx5 proteins as research tools (18-20) and also as in ViVo diagnostic agents to detect cell death in cancer chemotherapy, organ transplant rejection, or myocardial infarction (21-23). In addition, drugtargeting and drug-delivery strategies based on chimerical Anx5 entities coupled to various types of bioactive moieties have recently been proposed, e.g., for delivering thrombolytic enzymes to thrombotic sites or for delivering compounds intracellularly via Anx5-mediated endocytosis (24, 25). Our first goal is to design liposomes functionalized with oriented Anx5 proteins in such a way that the Anx5 proteins maintain their property of binding to PS-exposing cells. The
10.1021/bc9002579 CCC: $40.75 2009 American Chemical Society Published on Web 10/19/2009
Annexin A5-Functionalized Liposomal Vectors
structures of the soluble form (26, 27) and membrane-bound form of Anx5 (28-31) indicate that this protein presents a slightly bent shape with a convex membrane-binding face and an opposite concave face. On the basis of this structural knowledge, we have designed an Anx5-SH mutant protein harboring a unique sulfhydryl group inserted in a highly accessible loop on the concave face of the protein (32). We present here the synthesis and characterization of Anx5functionalized PEGylated liposomes. Conditions of covalent functionalization were monitored by SDS-PAGE, and the interaction of Anx5-liposomes with PS-exposing model membranes was characterized in a systematic manner by quartz crystal microbalance with dissipation monitoring method (QCMD). The QCM-D method is particularly well adapted for studying the interaction of proteins or proteolipid complexes with model membranes (33-35). The encapsulation property and the ability of Anx5-liposomes to bind PS-exposing apoptotic cells were investigated by fluorescence spectroscopy and microscopy, respectively.
EXPERIMENTAL PROCEDURES Materials. 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), cholesterol, 1,2-dipalmytoyl-sn-glycero-3-phosphoethanolamineN-[methoxy(poly(ethylene glycol))-2000] (PEG2000-DPPE), 1, 2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[maleimide(poly(ethylene glycol))2000] (Mal-PEG2000-DSPE), and 1,2dioleoyl-sn-glycero-3-[phospho-L-serine] (DOPS) were obtained from Avanti Polar Lipids (Alabaster, AL, USA). All other chemicals were of ultrapure grade. Water was purified with a RiOs system (Millipore, St Quentin en Yvelines, France). STI571 (imatinib mesilate) was from Novartis (Basel, Switzerland). Buffer Preparation. A HEPES buffered saline (HBS) solution made of 10 mM HEPES, 150 mM NaCl, and 2 mM NaN3 was prepared in ultrapure water. The pH was set at 7.4, except for Anx5 and liposome solutions used in covalent coupling reactions where pH was set at 6.3. Buffer solutions were supplemented with either 2 mM CaCl2 or 2 mM EGTA for QCM-D and cell experiments, as indicated in the text. Liposome Preparation. Liposomes were prepared by lipid film hydration (36) followed by extrusion, as briefly described here. Lipids were dissolved in chloroform, mixed in desired amounts, and the solvent was evaporated to dryness using a rotary evaporator during 20 min. Lipids were hydrated in HBS, resulting in multilamellar vesicles (MLVs), and homogenized by three cycles of freezing-thawing and subsequent vortexing. Large unilamellar vesicles (LUVs) were obtained by hand extrusion of the resulting lipid dispersion, passing the mixture twenty times through 0.1 µm polycarbonate filters with a LiposoFast extruder (Avestin, Sodexim, France). LUVs were composed of DOPC/cholesterol/Mal-PEG2000-DSPE/PEG2000DPPE in proportions specified in each experiment. Small unilamellar vesicles (SUVs), composed of DOPC/DOPS (weight ratio 4/1), were obtained by sonication of a MLV suspension with a tip sonicator (Misonix, Farmingdale, NY), operated in a pulsed mode at 20% duty cycle for 30 min with refrigeration. Phospholipid concentrations were determined by phosphate analysis according to Rouser (37). Annexin Protein Preparation. Rat recombinant AnxA5-SH and AnxA5-S-S-Anx5 dimers were produced and purified as described (32). The AnxA5-SH protein is a double mutant (C314S, T163C) in which the only cysteine in wt-Anx5 has been replaced by a serine (C314S) and a single cysteine has been introduced, at position 163 (T163C), in a highly exposed region on the concave face of the protein. Covalent Coupling of Anx5-SH to Liposomes. To produce Anx5-SH proteins, Anx5-S-S-Anx5 dimers were reduced by incubation with 10 mM DTT in HBS at pH 6.3 for 30 min.
Bioconjugate Chem., Vol. 20, No. 11, 2009 2115
Reduced Anx5-SH was purified on a HiTrap desalting column (GE-Healthcare, Brabant, Belgique) equilibrated with HBS at pH 6.3. Anx5-SH was mixed with desired amounts of LUVs containing Mal-PEG2000-DSPE lipids, in HBS at pH 6.3, and incubated at least for 4 h at room temperature. The number of maleimide groups available for covalent coupling was calculated assuming that lipids are equally distributed among the two membrane leaflets and that only maleimide groups located on the outer leaflet are accessible. Characterization of Covalent Coupling of Anx5-SH to Liposomes. A given amount of liposomes composed of DOPC/ cholesterol/Mal-PEG2000-DSPE (74/25/1 mol %) was incubated with Anx5-SH solutions of increasing concentration. The extent of coupling was analyzed by SDS-PAGE (38) using a 10% resolving gel and visualization of protein bands by Coomassie blue staining. Since free maleimides from the inner leaflet are exposed upon solubilization of the liposomes by SDS, the remaining free thiol groups of the suspension were blocked by addition of 5 µM N-ethylmaleimide for 4 h before running the gels. Dynamic Light Scattering (DLS). DLS measurements were performed using an ALV laser goniometer, equipped with a 22 mW HeNe linearly polarized laser operating at a wavelength of 632.8 nm and an ALV-5000/EPP multiple τ digital correlator with 125 ns initial sampling time. The liposome suspensions were maintained at a temperature of 25.0 ( 0.1 °C in all experiments. The counting time was 300 s for each sample. The liposome composition was DOPC/cholesterol/Mal-PEG2000DSPE (70/25/5 mol %). Samples contained 200 µg phospholipids in a total volume of 900 µL, pH 7.4. All measurements were performed at three scattering angles (50°, 90°, 120°); the presented results correspond to 90° scattering. Data were analyzed using ALV Correlator Control software. Quartz Crystal Microbalance with Dissipation Monitoring (QCM-D). QCM-D measurements were performed with a Q-SENSE E4 system (Q-SENSE, Gothenburg, Sweden) (39). Briefly, upon interaction of matter with the surface of a sensor crystal, changes in the resonance frequency, f, related to adsorbed mass (including coupled water), and in the dissipation, D, related to viscous losses in the adlayer are measured. Measurements were performed in flow mode; i.e., the solution was continuously delivered to the measurement chamber (flow speed 150 µL/min) by the aid of a peristaltic pump (ISM935C, Ismatec, Zu¨rich, Switzerland). The working temperature was 24 °C. Variations in resonance frequency (∆f) and dissipation (∆D) were measured at several harmonics (15, 25, 35 MHz) simultaneously. Adsorbed masses, ∆m, are calculated according to the Sauerbrey equation, ∆m ) -C∆fnorm, with the mass sensitivity constant C ) 17.7 ng · cm-2 · Hz1- and ∆fnorm ) ∆fn/ n, with n being the overtone number (40). The presented ∆m and ∆D values correspond to the fifth harmonics (n ) 5, i.e., 25 MHz). Prior to use, the QCM-D sensor crystals were cleaned by two cycles of exposure to a 2% SDS solution for 15 min, rinsing with ultrapure water, blow-drying with nitrogen, and exposure to UV/ozone for 10 min, as described in Richter et al. (41). A supported lipid bilayer (SLB) was formed on silica-coated quartz sensor crystals by deposition of SUVs made of DOPC/ DOPS (4:1, w:w), as described in refs 41-43. After rinsing the chamber with Ca2+-containing HBS to remove the excess of SUVs, solutions of functionalized vesicles were introduced in the chamber. The vesicle composition was DOPC/cholesterol/ Mal-PEG2000-DSPE 74/25/1 mol % for all experiments, except for experiments dealing with the influence of the PEG layer on targeting efficiency. In this latter case, three kinds of formulation
2116 Bioconjugate Chem., Vol. 20, No. 11, 2009
Garnier et al.
Figure 1. Scheme of covalent coupling of Anx5-SH to liposomes containing Mal-PEG2000-DSPE.
were used: PEG2000-DPPE × mol %/Mal-PEG2000-DSPE 1 mol %/cholesterol 25 mol %/DOPC 74 - x mol %, with x ) 0, 2, or 4 mol %. Liposome Leakage Assay. The self-quenching property of the fluorescent dye sulforhodamine B was characterized by measuringtheintensityofdyesolutionsofincreasingconcentrations. The fluorophore was passively encapsulated into liposomes as follows. The dry lipid film was hydrated with a HBS solution containing 50 mM sulforhodamine B at pH 7.4. Vesicles were then prepared and covalently coupled to Anx5 as described above. Nonencapsulated dye molecules were discarded by gel filtration on a Sephacryl S-400 column (Amersham Bioscience, Buckinghamshire, UK). Leakage was followed by measuring the fluorescence intensity of the liposome solutions. Complete release of sulforhodamine B was achieved by addition of Triton X100 (100 µL at 0.5% v/v), and the proportion of fluorophore released was calculated as % sulforhodaminefree )
I t - I0 Ir - I0
where It is the fluorescence intensity after the desired period of time, I0 is the fluorescence baseline, and Ir is the fluorescence after complete fluorophore release. All measurements were realized with a Perkin Ellmer LS55 (Burckinghamshire, UK) at λexc ) 535 nm and λem ) 595 nm with emission and excitation slits of 2.5 nm. Two kinds of formulation were used: PEG2000-DPPE 4 mol %/Mal-PEG2000DSPE 1 mol %/cholesterol 25 mol %/DOPC 70 mol %, and Mal-PEG2000-DSPE 5 mol %/cholesterol 25 mol %/DOPC 70 mol %. Cell Targeting in Vitro. A solution of 12.5 µM sulforhodamine B in HBS at pH 7.4 was passively encapsulated into liposomes, followed by covalent coupling of Anx5. Nonencapsulated dyes and excess Anx5 were discarded by filtration on an Amicon Ultra filter with a 100 kDa molecular weight cutoff (Millipore, Billerica, USA). Bcr/Abl+ K562 leukemic cells (44) were grown in RPMI 1640 medium (Gibco, Cergy Pontoise, France), supplemented with 30% fetal calf serum and a mixture of antibiotics (100 units mL-1 penicillin, 0.1 mg mL-1 streptomycin; Gibco). Cells were grown at 37 °C in humidified atmosphere and 5% CO2. At J0, K562 cells were resuspended at 3 × 105 cells/mL-1. Twenty-four hrs later (J1), 2 µM STI-571 was added to induce apoptosis (45). Twenty-four hrs later (J2), 30 µgphospholipids of liposomes at a concentration of 300 µgphospholipids/mL were added and incubated for 1 h at 37 °C before separation of the cells from unbound vesicles by centrifugation for 5 min at 200 g at 25 °C, followed by microscopy observations. The vesicles’ composition was DOPC/cholesterol/Mal-PEG2000-DSPE/PEG2000DPPE 70/25/1/4.
Figure 2. Quantitative analysis of the covalent coupling of Anx5-SH to liposomes containing Mal-PEG2000-DSPE, by SDS-PAGE Fisrt lane: pure Anx5-SH. Following lanes: a fixed amount of liposomes, corresponding to 20 µg total phospholipid content, was incubated overnight with varying amounts of Anx5-SH. Prior to the SDS-PAGE analysis, a 5-fold excess of NEM was added to the samples to block unreacted thiol groups, and the mixture incubated for 4 h. The amounts of Anx5 are expressed both in micrograms and as the theoretical ratio of the (number of SH groups/number of accessible NEM groups) considering that half of the NEM groups are exposed on the outer lipid layer (from left to right: 0.825 µg to 11.55 µg or 12.5% to 175%).
Laser Scanning Confocal Microscopy. Live K562 cells (1.2 × 105 cells/cm2) were mixed with liposomes containing sulforhodamine-B in 8-well Labteck plates (Nalc Nunc International). Image acquisition was performed with a Zeiss LSM510 laser scanning confocal microscope equipped with HeNe laser and with a Plan-Apochromat 63.0 × 1.40 oil objective lens (Carl Zeiss). Cells immerged in 0.5 mL of culture medium were excited with the 543 nm laser line (intensity set at 60% of maximal power) and their emission observed through a 560 nm long-pass filter. Photomultiplier detector voltage values (offset and gain) were adjusted on cells containing nonfluorescent samples for each set of experiments. The liposomes composition was DOPC/cholesterol/Mal-PEG2000-DSPE/PEG2000-DPPE 70/ 25/1/4.
RESULTS Covalent Coupling of Anx5-SH to Liposomes. The covalent coupling of Anx5-SH proteins to liposomes containing lipid molecules exposing a maleimide group (referred to hereafter as Mal-liposomes) at their PEG distal end was achieved on preformed vesicles. The ligation reaction, represented schematically in Figure 1, was monitored by SDS-PAGE (Figure 2). For the protein alone, the major Anx5-SH band is accompanied by a minor band, of variable intensity, which corresponds to disulfide-bonded Anx5 dimers. After incubating Mal-liposomes with Anx5-SH, an additional band was observed, which is
Annexin A5-Functionalized Liposomal Vectors
Figure 3. Dispersity of Anx5-functionalized liposome suspensions, by DLS size distribution profiles of Anx5-functionalized liposomes, determined by DLS at 90°. Densities of Anx5 per liposome: (A) 250; (B) 2000.
located slightly above the Anx5 monomer, as expected for covalent Anx5-PEG2000-DSPE complexes. The molecular mass determined by MALDI-TOF mass spectrometry was 38 578 Da, in agreement with the theoretical value of 38 538 Da. Figure 2 shows that maximal coupling was obtained for a ratio between sulfhydryl groups and exposed maleimide groups close to a 1:1 stoichiometry. In these conditions, 80% of Anx5-SH was covalently coupled to Mal-liposomes. The fact that about 20% Anx5 remains uncoupled is attributed to partial hydrolysis of the maleimide group and to the spontaneous oxidation of Anx5SH into Anx5 dimers, as indicated above. Complementary experiments indicated that maximal coupling was reached in 4 h at µM protein concentration and that the coupling efficiency was not influenced by the addition of nonmaleimidated PEG2000DPPE molecules (up to 4%) (data not shown). Dispersity of Liposome Suspensions. The dispersity of liposome suspensions was monitored by DLS (Figure 3). Nonfunctionalized liposomes were uniform in size, with a mean radius of 60 ( 4 nm, as expected for 100 nm extruded liposomes (data not shown). We found that the liposome dispersity was highly dependent on the extent of Anx5 coupling. Up to 250 Anx5/vesicle, liposomes showed a homogeneous population with a radius of 62 ( 6 nm (Figure 3A). At higher values of Anx5/vesicle (g250 Anx5/vesicle), aggregation occurred and precipitates were observed in the samples. The precipitates were found to be polydisperse with particle sizes ranging from hundreds of nanometers to several micrometers (Figure 3B). Binding of Anx5-Functionalized Liposomes to Model Membranes, Monitored by QCM-D. In order to characterize the ability of Anx5-functionalized liposomes to bind to lipid membranes exposing PS molecules, we investigated the interac-
Bioconjugate Chem., Vol. 20, No. 11, 2009 2117
tion of Anx5-liposomes with SLBs made of DOPC/DOPS (4: 1, w:w), by the QCM-D method. Figure 4A presents a reference experiment consisting of three steps: (1) the formation of a PS-containing SLB by adsorption and cooperative rupture of SUVs (arrow 1), followed by (2) the Ca2+-dependent binding of Anx5 (arrow 2) and (3) the EGTA-induced desorption of Anx5 (arrow 3). The values obtained for the adsorbed mass and dissipation shift of a SLB, namely, 460 ng × cm-2 and ∼0 × 10-6, respectively, and for a saturating Anx5 monolayer, namely, 275 ng × cm-2 and
