Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The discovery and development in the last decade of neuroactive peptides and proteins with therapeutic potential for a range of neurological disorders has generated great interest among neuroscientists. Many growth factors have been shown to support the growth of motor neurons in laboratory cultures, and so have become candidates for the treatment of various neurodegenerative diseases (Mobley, 1989). However, the clinical utility of these peptides or protein therapeutic agents in the treatment of neurological disorders is limited by their inability to penetrate the blood-brain barrier (BBB) efficiently after systemic administration.

The BBB is formed by the endothelial cells (ECs) that make up brain capillaries. These brain capillary ECs are sealed together by complex tight junctions and possess few pinocytic vesicles (Reese and Karnovski, 1967). These characteristics, added to the existence of a metabolic barrier, restrict the transport of most small polar molecules and macromolecules from the cerebrovascular circulation into the brain.

One solution is to administer drugs directly into the brain: for peptide and protein drugs this can be done by 1) i.c.v. infusion of the compound into the cerebrospinal fluid (Olson et al., 1992), 2) transplantation into the brain of cells that produce the therapeutic agent (Kordower et al., 1994), 3) implantation in the brain of a polymer matrix impregnated with the therapeutic compound (Krewson et al., 1995), or i.v. delivery of the gene encoding the therapeutic agent to neuronal cells with viral vectors (Suhr and Gage, 1993). However, all these procedures suffer from the major disadvantage of being invasive, and requiring neurosurgery.

If the drug is to be administered noninvasively via the bloodstream, one pathway to the brain is between the cells, by opening tight junctions. Paracellular transport of compounds across the BBB can be increased by means of intracarotid infusion of hyperosmotic saccharide solution (Rapoport, 1988), or with Cereport (RMP-7, receptor mediated permeabilizer-7), an analog of bradykinin (Inumura et al., 1994). A potential drawback to all the methods that involve an increase in BBB permeability is of poor specificity, causing compounds in the circulating blood such as albumin for example to gain access to the brain indiscriminately.

The specific endogenous transport mechanisms through brain capillary EC offer a potential route for the development of brain-specific drug delivery systems for neuroactive compounds. Several studies have already shown that OX-26 monoclonal antibody, a receptor-mediated vector, can be successfully used to deliver therapeutic peptides such as nerve growth factor (Friden et al., 1993) and vasoactive polypeptide (Bickel et el., 1993) into the brain through the BBB. Despite their high specificity and affinity, a major problem with such antibodies has proved to be their failure to reach the target cells in adequate quantities.

To investigate the fundamental mechanisms of BBB transport biology, we developed an in vitro model of the BBB that closely mimics in vivo conditions by culturing brain capillary ECs and astrocytes on opposite sides of a filter (Dehouck et al., 1990, 1992). This model yielded evidence that macromolecules such as low-density lipoprotein, diferric transferrin, and lactoferrin undergo transcytosis through the BBB via a receptor-mediated pathway (Descamps et al., 1996; Dehouck et al., 1997; Fillebeen et al., 1999).

Herein, we report on the ability of nanoparticle carriers to activate this transcytotic EC pathway. The effect of the particles' charge (anionic, cationic, neutral), and the influence of a phospholipid coating were evaluated. To closely mimic the in vivo conditions, different incubation media were evaluated (sera with and without red blood cells).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals and Antibodies

[U-¹⁴C]sucrose (677 mCi/mmol) was obtained from Amersham Laboratories (Les Ulis, France), 1,2-dipalmitoyl-sn-glycero-3-phosphocholine from NOF Corporation (Hyogo, Japan), 5-([4,6-dichlorotriazin-2-yl]amino) fluorescein from Sigma Chemical Co. (Saint Louis, MO), 1-chloro-2,3-epoxypropan (epichlorhydrin) and glycidyltrimethylammonium chloride (hydroxycholine) from Fluka (Saint-Quentin-Fallavier, France), and phosphoryl chloride from Prolabo (Paris, France). The rabbit polyclonal antibody against occludin was from Zymed Laboratories Inc. (South San Francisco, CA). Primary antibody was detected with Bodipy-conjugated goat anti-rabbit IgG from Molecular Probes, Inc. (Eugene, OR). Albumin (bovine albumin; ICN Biomedicals, Inc., Costa Mesa, CA) was radiolabeled with reductive methylation and tritiation of albumin lysine residues with the borohydride method (Tack et al., 1980). The labeled albumin was checked by capillary electrophoresis. The labeled albumin behaved like endogenous albumin. The resulting ³H-labeled proteins were at least 97% trichloroacetic acid precipitable.

Preparation of Light-Biovectors (L-BV)

Polysaccharide particles were prepared from US Pharmacopoeia maltodextrin (Glucidex, Roquette, Lille, France) as described previously (Betbeder et al., 1996; Prieur et al., 1996). Briefly, 100 g of maltodextrin was dissolved in 2 N sodium hydroxide with magnetic stirring at room temperature. Addition to the crude mixture of 1-chloro-2,3-epoxypropane (epichlorhydrin, neutral) (4.52 ml), or of a mixture of epichlorhydrin (4.72 ml) and glycidyltrimethylammonium chloride (hydroxycholine, cationic ligand) (31.18 g), or of phosphoryl chloride (anionic ligand:phosphate) (28.46 ml) yielded neutral, cationic, and anionic polysaccharide gels, respectively. The gels were then neutralized with acetic acid and sheared under high pressure in a Minilab homogenizer (Rannie; APV Baker, Evreux, France). The 60-nm neutral, cationic, and anionic polysaccharide nanoparticles obtained were ultrafiltered on an SGI Hi-flow system (hollow fiber module: 30 UFIB/1 S.6/40 kD; Setric Génie Industriel, Toulouse, France) to remove low-molecular weight reagents and salts.

L-BVs were prepared in a Minilab homogenizer by mixing polysaccharide nanoparticles (Major et al., 1997), 1,2-dipalmitoyl-sn-glycero-3-phosphocholine, and cholesterol at a temperature above the gel-to-liquid phase transition temperature of the phospholipid (Woodle and Papahadjopoulos, 1989). Polysaccharide and phospholipid concentrations were 1.0 and 0.3 mg/ml, respectively. Phospholipid concentration was determined by the method of Bartlett (1959) for nonionic and cationic L-BVs. Additionally, a phospholipid enzymatic colorimetric test PL/MRP2 from Boehringer Mannheim GmBH (Mannheim, Germany) was used for anionic L-BVs. Cholesterol was analyzed with an enzymatic assay. The mean diameter of biovectors was determined by laser light scattering with the N4 MD Coulter nanoparticle analyzer (Coultronics, Margency, France). With this technique, the mean diameter obtained was 60 ± 15 nm for all the particles studied (neutral, anionic, and cationic).

Fluorescence Labeling of L-BVs

Labeling of the polysaccharidic core with fluorescein was achieved by adding 10 mg of a 5-([4,6-dichlorotriazin-2-yl]amino) fluoresce water solution (2 mg/ml) to 100 mg of polysaccharidic particles at pH 10, with magnetic stirring. These labeled particles were washed and purified by ultrafiltration on an SGI Hi-flow system (30 UFIB/1S.6/40 kD) with 1 M NaCl, and then with demineralized water until no fluorescein was detected in the ultrafiltrate. The fluorescein-labeled polysaccharidic particles (1 mg/ml) were stored in sterile tubes after filtration through a 0.2-µm filter. This type of linkage of fluorescein with polysaccharides is classically used for in vivo studies (Arfors et al., 1973). The stability of the linkage of fluorescein with the polysaccharide backbone was found to be very high in solution in the presence of ionic ligands (>1 year at 4°C).

Cell Culture

Bovine Brain Capillary ECs. ECs were isolated and characterized as described by Dehouck et al. (1990). The use of cloned ECs afforded a pure EC population without contamination by pericytes. The cells were cultured in the presence of Dulbecco's modified Eagle's medium supplemented with 10% (v/v) heat-inactivated calf serum and 10% (v/v) horse serum (Gibco Life Technologies, Rockville, MD), 2 mM glutamine, 50 µg/ml gentamycin, and basic fibroblast growth factor (1 ng/ml, added every other day).

Rat Astrocytes. Primary cultures of mixed astrocytes were grown from newborn rat cerebral cortex. After removing the meninges, the brain tissue was gently forced through a nylon sieve. Astrocytes were plated on six-well dishes (Nunclon; Nunc A/S, Roskilde, Denmark) at a concentration of 3 × 10⁴ cells/ml in 2 ml of Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (Gibco Life Technologies), and the medium was changed twice a week. Three weeks after seeding, the astrocyte cultures were stabilized, and were free of any oligodendrocyte contamination.

Coculture of ECs and Astrocytes. Filters for coculture were prepared as follows: culture plate inserts (Millicell PC 3 µm, 30-mm diameter; Millipore Corp, Bedford, MA) were coated on the upper side with rat tail collagen prepared by a modification of the method of Bornstein (1958).

Fluorescence Microscopy

ECs grown on porous filters were fixed and permeabilized with cold ethanol (20°C). The samples were washed three times with Tris-buffered saline (20 mM Tris-HCl, 0.5 M NaCl, pH 7) and soaked in the blocking solution [Tris-buffered saline containing 5% ovalbumin (wt/vol)] for 10 min at room temperature. They were then incubated for 1.5 h in a moist chamber with the rabbit antioccludin pAb in the blocking solution. After rinsing, the cells were incubated for 1 h with the secondary antibody, Bodipy-conjugated goat anti-rabbit IgG, in the blocking solution. Following a final rinse, the filters and their attached monolayers were mounted on glass microscope slides with Mowiol mountant (Hoechst, Frankfurt, Germany). The specimens were visualized and photographed with a Leica fluorescence microscope.

Transendothelial Transport Studies

On the day of the experiment, the inserts containing confluent EC monolayers and inserts only coated with collagen were washed twice with buffered Ringer's solution. One insert containing a confluent monolayer of ECs was transferred to the first well of a six-well plate filled with 2.5 ml of buffered Ringer's solution in each compartment. Then, 1.5 ml of buffered Ringer's solution supplemented with 10% heat-inactivated calf serum (with and without 4-5 million red blood cells/ml) containing the different fluorescein-labeled L-BVs or polysaccharide cores (PC) (100 µg/ml), was placed at time zero in the upper compartment. The incubations were performed for 4 h at 37°C on a rocking platform. Triplicate inserts coated with collagen seeded and unseeded with ECs were incubated for 12 days with astrocytes, and used for each compound. At the end of the experiment, amounts of fluorescein-labeled compounds in the lower compartments were measured with a Hitachi F-2000 spectrofluorimeter. For the fluorescence excitation and emission spectra of fluorescein, the wavelengths _ex and _em were 495 and 520 nm, respectively. For each test compound, results were expressed as percentage transport across the brain capillary EC monolayer alone, obtained from the transport across the inserts coated with collagen and seeded with ECs, and the transport across the inserts coated only with collagen. Each point was done in triplicate and the data are represented as means ± S.E. Sets of data were then compared with the Mann-Whitney test. At the end of the experiments, cells incubated with PC and L-BV QAE (cationic) were washed five times with ice-cold buffered Ringer's solution and fixed in paraformaldehyde. The filters and their attached monolayers were mounted on glass microscope slides with Mowiol mountant. The specimens were visualized and photographed with a Leica fluorescence microscope.

Using the same procedure, the integrity of the brain EC monolayers was checked by adding [¹⁴C]sucrose in the upper compartment containing the different test compounds. Amounts of radiotracers in the lower compartment were measured in a liquid scintillation counter (Wallac 14110; Pharmacia, Piscataway, NJ). The endothelial permeability coefficient (Pe in cm/min) was calculated as previously described (Dehouck et al., 1992).

The same experiments were performed with [³H]albumin and biovectors loaded with [³H]albumin. The formulations were made up by simple mixing of [³H]albumin with premade L-BV with stirring. The albumin/liter-BV ratio was 1:10 (w/w). The rate of adsorption was analyzed with a plasmon resonance experiment (BIACORE, Uppsala, Sweden).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

EC Characterization. Fig. 1A depicts the typical structure of confluent brain capillary ECs cocultured for 12 days with astrocytes on an insert coated with rat tail collagen. The cells form a monolayer of nonoverlapping and contact-inhibited cells. The monolayer was homogeneous; no contamination by pericytes was observed. Immunofluorescent staining of the tight-junction integral protein occludin showed preferential cortical membrane localization (Fig. 1B). This continuous network of occludin labeling suggests that the tight junction barrier is well established. This and numerous previous results concerning the high electrical resistance (500-800  · cm²), low permeability for sucrose and inulin (Dehouck et al., 1990), and particularly the correlation between drug permeabilities obtained in vitro with our model and in vivo with the Oldendorf technique (Dehouck et al., 1992) support this coculture model as a legitimate model of the BBB.

View larger version (162K): [in this window] [in a new window]   Fig. 1.   Confluent monolayer of bovine brain capillary ECs demonstrates homogeneity in phenotype (A). Bar, 200 µm. Bovine brain capillary ECs grown on filters were fixed and stained for the tight junction protein occludin (B). Bar, 50 µm.

Transendothelial Transport Studies. L-BVs were prepared from 60-nm PCs enveloped in a lipid bilayer (Major et al.,1997). These PCs were neutral (N), cationic (QAE), or anionic (P), depending on the meshing agent and on additional groups used for the synthesis. Transport studies of the different PCs (neutral and ionically charged) and their corresponding L-BVs across the brain capillary EC monolayer were performed as described in Materials and Methods. To study the luminal-to-abluminal transport, fluorescein-labeled PCs or L-BVs were added to the luminal chamber of the coculture system and the transfer of these nanoparticles across the cell monolayer was observed after a 4-h incubation.

View larger version (23K): [in this window] [in a new window]   Fig. 2.   Luminal-to-abluminal transport of PCs, neutral and ionically charged, and their corresponding L-BVs. EC monolayers and filters coated with collagen were incubated with PCs, neutral and ionically charged (open column) or their corresponding L-BVs (hatched column) for 4 h at 37°C on a rocking platform. Results are expressed as percentage of each compound passage through the EC monolayer. Each point is the mean of three different filters and the histograms are representative of three series of independent experiments.

3

3

3

3

3

3

3

View larger version (70K): [in this window] [in a new window]   Fig. 3.   Luminal uptake of fluorescein-labeled PC QAE (A) or L-BV QAE (B) by brain capillary ECs. We added 100 µg/ml of each compound to the luminal side of the cells for 4 h at 37°C. After washing in ice-cold buffered Ringer's solution, cells were fixed in paraformaldehyde. The filters and their attached monolayers were mounted and visualized with Leica fluorescence microscope. Bar, 50 µm.

3

3

3

3

3

3

3

View larger version (27K): [in this window] [in a new window]   Fig. 4.   Luminal-to-abluminal transport of PCs, neutral or ionically charged, and their corresponding L-BVs in presence of red blood cells. EC monolayers and filters coated with collagen were incubated with PCs, neutral and ionically charged, or their corresponding L-BVs in a cell culture medium with (hatched column) or without (open column) red blood cells for 4 h at 37°C on a rocking platform. Results were expressed as percentage of each compound passage through the EC monolayer. Each point is the mean of three different filters.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

By growing primary ECs on one side of a porous filter and astrocytes on the other, we can reconstruct the environment that exists in vivo and so induce most of the BBB characteristics (Dehouck et al., 1992). Of these characteristics, one is of major importance for studying drug transport to the brain; in vivo, brain capillary ECs connected by tight junctions form a continuous lining in the capillary wall that prevents paracellular flux of solutes. Occludin expression can be a determinant of tight junction permeability (McCarthy et al., 1996). Our immunofluorescent studies clearly demonstrate that brain capillary ECs in these culture conditions express occludin continuously at cell-to-cell contacts. As described by Hirase et al. (1997), this continuous staining of occludin at the cell border indicates that these cells are highly differentiated ECs of cerebral origin. These and previously published results demonstrating high electrical resistance (500-800  · cm²) and low permeability (Dehouck et al., 1990), indicate that we have a highly differentiated BBB model that is suited to the study of drug transport. Furthermore, with this in vitro model, we found evidence for an endogenous transcytotic pathway in ECs for the transport of different receptor-mediated proteins. Such a pathway seems to be a feature of capillary ECs of cerebral origin because it has already been described for three blood-borne molecules: LDL, transferrin, and lactotransferrin (Descamps et al., 1996; Dehouck et al., 1997; Fillebeen et al., 1999). All these macromolecules follow an intracellular pathway bypassing the lysosomal compartment, and thus are not degraded within the cells to be delivered to the abluminal side of the barrier. This endogenous pathway is therefore a potential route for molecule delivery to the brain. We set out to evaluate whether 60-nm L-BV nanoparticles could cross the BBB. We evaluated the effects of core charge and coating with a lipid bilayer. We observed that by enveloping a charged PC with a phospholipid bilayer, it was possible to increase its transport through the EC monolayer. No modification of the paracellular permeability was observed during the incubation of cells with L-BV, so this increase was not due to a breakdown of the barrier. These results are confirmed by the immunofluorescent study demonstrating that fluorescein-labeled L-BVs are distributed throughout the cytoplasm. This intracellular localization is characteristic of transcellular transport, as described for LDL and transferrin (Descamps et al., 1996; Dehouck et al., 1997). In contrast, the perinuclear localization of fluorescein PCs shows an intracellular accumulation of these nanoparticles in a degradation compartment, as already observed for acetylated LDL. Taken together, these results demonstrate that the intracellular pathway of PCs differs according to whether the phospholipid bilayer is present, with coated PCs being transcytosed through the ECs.

We demonstrated previously that this pathway was induced by the binding of a ligand to its receptor (apoB-LDL receptor; ferrotransferrin-transferrin receptor). The degradative pathway also was induced by the binding of the degraded apoB to the scavenger receptor. In contrast, the results with L-BVs indicated that the lipid bilayer could activate the transcytotic pathway by a nonspecific process that does not involve their binding to a receptor. Because cationic lipid-coated nanoparticles were found to best cross the BBB, we loaded BSA onto these particles, assuming BSA loading would not interfere with their membrane properties. Unlike the ECs of peripheral endothelium, ECs from brain capillaries do not express albumin receptors in vivo or in vitro (Pardridge et al., 1985). Consequently, albumin transport through the ECs was extremely low, consistent with the results of Smith and Borchardt (1989), as indicated by the calculated Pe (0.03 × 10³ cm/min). By loading albumin on L-BV QAE, we were able to increase its transport 27-fold. Recent studies have demonstrated that caveolae are involved in the first steps of LDL and transferrin transcytotic pathways (L.F., R.C., G. Torpier, unpublished data). Caveolae are small noncoated plasmalemmal vesicles that are particularly abundantly expressed in many endothelia. All recent studies suggest these vesicles are involved in both endocytosis and transcytosis (Ghitescu and Bendayan, 1992). Moreover, endothelial caveolae have key proteins that mediate different aspects of vesicle formation, docking, and fusion, including vesicular soluble NSF attachment receptor, vesicle-associated membrane protein, and cellubrevin; small and large GTP-binding proteins; the calcium-dependent lipid-binding proteins annexin II and VI; and the N-ethylmaleimide-sensitive factor N-ethylmaleimide-sensitive fusion protein along with S-nitroso-N-aceytlpenicillamine (Schnitzer et al., 1995). All these results demonstrate that caveolae are indeed genuine trafficking organelles capable of budding from the plasmalemma to form discrete carrier vesicles containing the molecular machinery necessary for regulated specific transport.

From our previously published results (Dehouck et al., 1997; Fenart et al., submitted), which agree with those of Schnitzer and Oh (1994), caveolae can be divided into subpopulations depending on their cellular destination. Similar results were obtained in this study with PCs and L-BVs; PCs are delivered to the degradative compartment, whereas L-BVs are directed to the abluminal side of the cell. Different signaling molecules have to be activated to induce each of these pathways. Although kinase and phosphatase have been reported to regulate caveolae internalization (Smart et al., 1995), nothing is known about the different intracellular traffic pathway regulation. These two different vectors could be used to discriminate between the intracellular signals leading to the transcytotic or degradative pathway.

We studied the influence of red blood cells on the ability of these nanoparticles to cross the BBB model. We observed a strong inhibition of the crossing, probably due to nanoparticle-red blood cell interactions. Red blood cells particles interactions are well described in the literature. Cationic particles were found to strongly bind to human erythrocytes (Weiss et al. 1972) Liposomes also were found to bind to erythrocytes and their lipid content to be incorporated in the cell membrane (Eytan et al. 1982). These interactions could explain why the transcytosis of BV nanoparticles is inhibited in the presence of red blood cells. To decrease the nonspecific binding to red blood cells, the principle of steric stabilization of colloid particles was first introduced by Napper and Netchey (1971); it was later found (Illum and Davies, 1983) that i.v administration of nanoparticles could be improved by coating them with poloxamer. On the basis of these results, we expect to improve the ability of BV particles to cross the BBB in the presence of red blood cells.