Low-density Lipoprotein Receptor-related Protein Mediates the Endocytosis of Anionic Liposomes in Neurons* 210

We have recently demonstrated that anionic liposomes efficiently introduce foreign DNA into postmitotic neurons and other cell types (Lakkaraju, A., Dubinsky, J. M., Low, W. C., and Rahman, Y.-E. (2001) J. Biol. Chem.276, 32000–32007). To investigate the mechanism of liposome uptake, we followed the internalization of anionic liposome-encapsulated Cy3-labeled oligonucleotides (AL-Cy3ONs) by hippocampal neurons using confocal microscopy. Uptake of AL-Cy3ONs was widespread and time- and temperature-dependent, indicative of receptor-mediated endocytosis. The low-density lipoprotein receptor-related protein (LRP) was crucial for anionic liposome endocytosis because the receptor-associated protein or an anti-LRP antibody inhibited internalization, and fibroblasts lacking LRP did not internalize AL-Cy3ONs. Using selective endocytosis inhibitors, we found that liposome endocytosis and intracellular transport required clathrin, dynamin, an intact cytoskeletal network, and phosphatidylinositol 3-kinase activity. Cy3ONs did not significantly colocalize with recycling endosomal/lysosomal markers and entered neuronal nuclei within 1–3 h of incubation. Approximately 50% of the internalized liposomal phospholipids were recycled back to the cell surface, in keeping with the fluidity of their acyl chains. Liposome endocytosis did not require heparan sulfate proteoglycans or cause calcium influx into neurons. Thus, constitutive endocytosis of anionic liposomes by LRP utilizes only one component, in contrast to the more involved heparan sulfate proteoglycan-LRP pathway implicated in the pathogenesis of Alzheimer's disease.

Endocytosis in neurons has mainly been studied in the context of synaptic vesicle recycling (1) and the regulation of neurotransmitter receptor numbers in the post-synaptic membrane (2). Constitutive endocytosis also occurs in neurons, albeit at a slower rate than in non-polarized or mitotically active cells. Growth factors and hormones modulate the rate of constitutive endocytosis of neurotransmitter receptors as well as receptors involved in nutrient acquisition and metabolism. Neurotrophins such as nerve growth factor and bone-derived neurotrophic factor increase both the recruitment of clathrin to the plasma membrane and the rate of constitutive endocytosis of transferrin in hippocampal neurons (3).
Little is known, however, about the mechanisms of internalization of exogenous macromolecules such as phospholipids and nucleic acids in neurons. Insight into the interactions between neurons and these molecules would further our understanding of lipid and DNA transport pathways and provide potential therapeutic advantages. We have recently designed and developed an anionic liposome vector (composed of the anionic phospholipid dioleoylphosphatidylglycerol and the zwitterionic phospholipid dioleoylphosphatidylcholine) for oligonucleotide delivery to neurons. Antisense oligonucleotides targeted to the p53 tumor suppressor mRNA, delivered to neurons via anionic liposomes, efficiently protected hippocampal neurons from excitotoxic death by sequence-specific down-regulation of p53 protein expression (4). Preliminary studies indicated that the uptake of Cy3-labeled oligonucleotides (Cy3ONs) 1 encapsulated in anionic liposomes was rapid and widespread.
This study was undertaken to delineate the mechanism of liposome internalization in neurons. Cy3ONs were encapsulated in anionic liposomes, and the uptake of these molecules by cultured rat hippocampal neurons was studied by confocal microscopy. Each stage in the endocytic pathway was retarded by biochemically interfering with specific proteins to determine the role of that protein in the internalization of liposomes. Here we demonstrate that anionic liposomes utilize constitutive endocytosis of the low-density lipoprotein receptor-related protein (LRP) to enter neurons, followed by intracellular transport and processing via a classical endocytic pathway. In addition to providing a mechanistic basis for oligonucleotide delivery by anionic liposomes, our results have also identified an efficient phospholipid transport pathway in neurons. This pathway appears to be a previously unrecognized, independent component of the more involved LRP-mediated endocytic pathway that has been implicated in the processing of the amyloid precursor protein and the pathogenesis of Alzheimer's disease.

EXPERIMENTAL PROCEDURES
Oligonucleotides and Liposomes-An antisense oligonucleotide to the p53 mRNA was designed as previously described (4). Oligonucleotides were synthesized and labeled at the 5Ј-end with Cy3 and purified by reverse-phase high performance liquid chromatography to remove free dye (Integrated DNA Technologies). The oligonucleotides were reconstituted in sterile, nuclease-free Tris/EDTA buffer (pH 7.2) and stored at Ϫ20°C. Encapsulation of oligonucleotides in anionic liposomes (87.5 mol % dioleoylphosphatidylcholine and 12.5 mol % dioleoylphosphati-* This work was supported by the Pharmaceutical Research Fund (to Y.-E. R.), by a grant from the Huntington's Disease Society of America, and by National Institutes of Health Grant NS39414 (to J. M. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. dylglycerol), their characterization, and complexation of oligonucleotides with cationic liposomes were all performed as previously detailed (4). Liposomes labeled with N-Rh-DOPE were prepared in a manner identical to that described for liposomes encapsulating Cy3ONs, except that the lipid films contained 1-2.5 mol % N-Rh-DOPE, and the liposomes were prepared with buffer alone (i.e. without oligonucleotides).
Cell Culture and Treatments-Primary cultures of hippocampal neurons were prepared from neonatal rat pups (postnatal day 1 or 2) as previously described (4,5) and cultured on 22-mm square glass coverslips or eight-chambered glass slides (LabTek II, Nalgene Nunc) in serum-free Neurobasal medium with vitamin B 27 supplements unless otherwise stated. Neurons were incubated with 2 M Cy3ONs (free, encapsulated in anionic liposomes, or complexed with cationic liposomes) for 30 min at 37°C unless stated otherwise. For low-temperature studies, neurons were incubated at 4°C with AL-Cy3ONs for 10 min. The LRP-deficient mouse embryonic fibroblast cell line PEA-13 and its wild-type counterpart MEF-1 (6) were obtained from American Type Culture Collection. These cells were cultured on eight-chambered glass slides in Dulbecco's modified Eagle's medium containing penicillin and streptomycin with 10% Cosmic calf serum (Hyclone Laboratories) and incubated with AL-Cy3ONs for either 1 or 3 h at 37°C. At the end of the incubation periods, uninternalized oligonucleotides, liposomes, or complexes were removed by several washes with Leibovitz L-15 medium and fixed in 4% paraformaldehyde.
The following agents were used to manipulate specific steps of the endocytic cycle: 0.45 M hyperosmolar sucrose (ICN Biochemicals); 1 M FK506 (Calbiochem); and 100 nM wortmannin, 5 g/ml nocodazole, 10 g/ml cytochalasin D, 100 g/ml heparin, and 100 g/ml protamine sulfate (all from Sigma). The LRP inhibitor receptor-associated protein (RAP) and the LRP-blocking antibody R2629 were used at concentrations of 500 nM (7) and 100 g/ml (6), respectively. Cells were treated with the drugs, RAP, or R2629 for 10 min prior to incubation with AL-Cy3ONs for 30 min at 37°C, fixed, and imaged.
Colocalization Experiments-Intracellular fates of endocytosed liposomes and oligonucleotides were determined by comparing their intracellular distributions with those of Oregon Green 488-transferrin (OG-Tf) and Alexa 488-dextran (Molecular Probes, Inc.), used as markers for the recycling and lysosomal compartments, respectively. Neurons were incubated with AL-Cy3ONs and either 100 g/ml OG-Tf for 0.5, 1, or 3 h or 1 mg/ml Alexa 488-dextran for 3 h. For phospholipid transport experiments, neurons were incubated with liposomes labeled with N-Rh-DOPE and either OG-Tf for 0.5, 1, or 3 h or 1 mg/ml Alexa-dextran for 3 h. These time points were chosen based on earlier work on the kinetics of transferrin and dextran endocytosis in hippocampal neurons (3,8,9). At the end of the incubation periods, cells were rinsed and fixed for imaging as described above.
Confocal Microscopy and Image Analysis-Imaging was performed on a Leica TCS 4D confocal microscope equipped with a mercury/xenon lamp and an argon/krypton laser. Cells were excited using the 488 nm laser line to detect OG-Tf and Alexa 488-dextran, and the emitted fluorescence was collected using a 515 nm long-pass filter. The 568 nm laser line was used to excite Cy3 and N-Rh-DOPE (LP590 emission). Cells were imaged at a plane midway between the substrate-attached plasma membrane and the top of the cell, such that neuronal nuclei were clearly identifiable. In some cases, the entire volume of the cell was scanned in 0.5-m increments. For experiments with Alexa 488dextran, planes in which the Alexa 488 label was most visible were chosen to enable identification of late endosomal/lysosomal structures. Optimal images were obtained by averaging 16 images in the line-scan mode at the same fixed gains for all experiments. For the colocalization experiments, the cell outlines for each set of fields were traced out manually in the corresponding bright-field image and then used to mask the fluorescence images (Metamorph, Universal Imaging Corp.). In each cell, the total fluorescence intensity was measured, and the percent of Cy3 or rhodamine label that colocalized with the Oregon Green 488 or Alexa 488 label was calculated. All the representative fluorescence images shown in the figures were equally contrast-enhanced using Adobe Photoshop ® .
Calcium Imaging-Hippocampal neurons cultured on 35-mm glassbottomed Petri dishes (5) were loaded with 4 M fura-2/AM (Molecular Probes, Inc.) for 30 min. The cultures were mounted in recording solution containing 139 mM NaCl, 3 mM KCl, 10 mM NaHEPES, 1.8 mM CaCl 2 , 0.8 mM MgCl 2 , 5 mM glucose, 15 mM sucrose, and 0.1 mM glycine (pH 7.4) on the stage of an inverted Nikon Diaphot microscope. The recording solution also contained 1 M tetrodotoxin to inhibit spontaneous firing of the neurons. Images were captured with a ϫ40 objective every 20 s (Metafluor, Universal Imaging Corp.) using the following wavelengths: fura-2, excitation at 340 Ϯ 15 and 380 Ϯ 15 nm attenuated with a 10% quartz neutral density filter and emission at 525 Ϯ 25 nm; and N-Rh-DOPE, excitation at 565 Ϯ 15 nm and emission at Ͼ590 nm (10). All images were corrected by subtracting a wavelength-specific background from an unpopulated portion of the dish. In preliminary experiments, the addition of recording solution or liposomes by manual pipetting caused a transient increase in intracellular Ca 2ϩ . To avoid this artifact, anionic liposomes labeled with N-Rh-DOPE were perfused onto the neurons at a rate of 1 ml/min for 10 -15 min.
Protein Binding to Liposomes-Neurons were incubated with 100 l of liposomes in recording solution (as described above) for 3 h with or without 500 nM RAP. The medium overlying the cells was first centrifuged at 320 ϫ g for 8 min to remove cell debris, followed by centrifugation at 100,000 ϫ g for 15 min at 25°C in a Beckman ultracentrifuge to recover the liposomes (11). The liposome pellet was resuspended in 10 mM HEPES (pH 7.4) containing 150 mM NaCl, and an aliquot was analyzed using the highly sensitive 3-(4-carboxybenzoyl)quinoline-2carboxaldehyde (CBQCA) protein assay kit (Molecular Probes, Inc.) to determine whether any proteins secreted into the medium bound to liposomes. Fluorescence associated with any protein bound to the liposomes labeled by the CBQCA reagent was detected using a PerkinElmer Life Sciences LS55 fluorometer.

Neuronal Uptake of Anionic Liposomes Occurs by Clathrinmediated Endocytosis-Incubation of hippocampal neurons
with anionic liposomes containing 2 M Cy3ONs for 30 min at 37°C resulted in the localization of the labeled oligonucleotides in vesicular cytoplasmic structures, but not in the nucleus (Fig.  1a). After a 1-h incubation, diffuse Cy3 fluorescence was observed in the nucleus (Fig. 1b). The intensity of the diffuse nuclear label increased after 3 h, and portions of the cytoplasm often contained uniform Cy3 fluorescence, in addition to the punctate label (Fig. 1c). Virtually all the neurons examined under the microscope and/or imaged exhibited Cy3 fluorescence 30 min after incubation with anionic liposomes. Uptake of anionic liposomes was greatly reduced at 4°C, with Cy3 fluorescence seen only at the cell surface, indicating binding of liposomes to the plasma membrane, but not internalization (Fig. 1d). The time-and temperature-dependent uptake of anionic liposomes containing Cy3ONs suggested that liposome internalization into neurons occurred by endocytosis and was possibly receptor-mediated. To identify the cellular elements required for liposome uptake, we inhibited various steps in the endocytic pathway. Fig. 2 depicts the incidence of intracellular Cy3 fluorescence in neurons following various experimental manipulations.
Cell-surface receptors concentrated in clathrin-coated pits mediate the endocytosis of many macromolecules. To determine whether clathrin-coated pits are involved in the uptake of anionic liposomes, hyperosmolar sucrose was used to disrupt clathrin assemblies (12). Treatment of neurons with 0.45 M sucrose completely prevented internalization of anionic liposomes containing Cy3ONs (Fig. 3b) compared with cells treated with AL-Cy3ONs alone (Fig. 3a). This confirmed that neuronal uptake of liposomes was receptor-mediated because hyperos-molarity inhibits receptor-mediated endocytosis, but does not interfere with nonspecific fluid-phase endocytosis (13).
Clathrin-dependent endocytosis involving accessory proteins such as the GTPase dynamin, amphiphysin, and synaptojanin plays a critical role in synaptic vesicle recycling at nerve terminals (14). Dynamin and amphiphysin need to be dephosphorylated by the Ca 2ϩ /calmodulin-dependent phosphatase (calcineurin) to interact with one another and the lipid bilayer (15). We used a calcineurin inhibitor (FK506) to study the role of dynamin in anionic liposome internalization. Treatment of neurons with FK506 prior to the addition of AL-Cy3ONs significantly decreased liposome endocytosis (Fig. 3c), providing further evidence that liposomes are internalized in neurons via clathrin-coated pits.
Liposome Endocytosis Occurs via LRP-LRP, a member of the low-density lipoprotein receptor gene family, is highly expressed in the mammalian central nervous system and has been implicated in the endocytosis of several unrelated ligands (7). A major function of lipoprotein receptors is the regulation of lipoprotein uptake and metabolism (16). Immunofluorescence staining with an antibody to LRP confirmed that this receptor was highly expressed in our hippocampal neuronal cultures (see Supplemental Material). To determine whether LRP is involved in the endocytosis of anionic liposomes, we blocked LRP using RAP, which is a potent inhibitor of all known ligand interactions of LRP (7). RAP binds with high affinity to the heavy chain of LRP on multiple ligand-binding domains and induces a conformational change in the receptor, thus interfering with ligand binding (17). When neurons were incubated with AL-Cy3ONs in the presence of RAP, both binding and internalization of anionic liposomes were inhibited. Note the complete absence of Cy3 fluorescence both on the cell surface and within the RAP-treated neurons (Fig. 4b) compared with treatment with AL-Cy3ONs alone (Fig. 4a). Further evidence for the involvement of LRP in anionic liposome endocytosis was obtained when preincubation of neurons with the neutralizing LRP antibody R2629 (6) also prevented AL-Cy3ON uptake (Fig. 4c).
Last, we compared the endocytosis of AL-Cy3ONs in immortalized mouse embryonic fibroblast cell lines that either expressed LRP (MEF-1) or lacked the receptor (PEA-13). After a 1-h incubation, almost all MEF-1 cells displayed robust Cy3 fluorescence (Fig. 5a), in contrast to the faint signal seen in PEA-13 cells (Fig. 5b). Following a 3-h incubation, Cy3 fluorescence was visible in the PEA-13 cultures at lower intensity than in the MEF-1 cells, indicating that liposomes were being taken up by the PEA-13 cells, albeit with very slow kinetics. In contrast to the MEF-1 cultures, where all the cells examined had the Cy3 label, only 50 -60% of the PEA-13 cells exhibited Cy3 fluorescence after 3 h (Fig. 5, compare c and d).
Endocytosis of AL-Cy3ONs via LRP Is Independent of HSPGs and Does Not Alter Neuronal Calcium Influx-Several LRP ligands, including ␣ 2 -macroglobulin, apoE, and HIV Tat protein, bind HSPGs on the cell surface prior to being internalized by LRP. Consequently, RAP does not inhibit binding of HIV Tat to the plasma membrane, but inhibits its internalization and subsequent degradation (18). To determine whether HSPGs are necessary for liposome endocytosis by LRP, we incubated neurons with AL-Cy3ONs along with 100 g/ml heparin or protamine sulfate. Heparin is a specific inhibitor of HSPGs, and protamine competes with LRP ligands for HSPG-binding sites (19). Neither heparin (Figs. 2 and 6a) nor protamine (Fig.  2) altered the level of Cy3 fluorescence within neurons after 30 min of incubation, indicating that HSPGs are not required for anionic liposome endocytosis by LRP.
To determine whether endogenous proteins secreted by neu- rons can bind liposomes and act as intermediaries between liposomes and LRP, we measured protein binding to liposomes after a 3-h incubation with neurons. Incubations were carried out in either the absence or presence of 500 nM RAP to increase the possibility of protein-bound liposomes being recovered from the medium. The amount of protein detected by the CBQCA assay did not significantly differ between the untreated and RAP-treated controls and liposome-treated conditions (one-way analysis of variance, p ϭ 0.5) ( Table I). As a positive control, liposomes were incubated with poly-L-lysine (lysine/lipid phosphate charge ratios of 0.6 and 2), and 100% of the added polylysine was detected in the liposome pellet (data not shown). As the amine moiety on the choline head group of dioleoylphosphatidylcholine was found to interact with the CBQCA dye, standard curves with bovine serum albumin were constructed in solutions containing liposomes, and the samples were diluted to minimize lipid interference. We were able to detect 10 ng of exogenously added bovine serum albumin (data not shown), indicating that within the limits of sensitivity of this assay, no endogenous proteins from cultured neurons bound liposomes.
Recent studies on cortical neurons and hippocampal slices have suggested a role for LRP in synaptic neurotransmission (20,21). The addition of activated ␣ 2 -macroglobulin to cortical neurons caused a Ca 2ϩ influx that was both spatially and temporally discrete. Only ligands that bind LRP at multiple sites were capable of eliciting this calcium response, indicating that receptor dimerization was essential. To examine whether the endocytosis of anionic liposomes via LRP causes Ca 2ϩ influx into neurons, we studied neuronal calcium influx during a continuous perfusion of liposomes labeled with N-Rh-DOPE. Anionic liposomes did not evoke a calcium response (Fig. 6b), although they were endocytosed as evidenced by rhodamine fluorescence in the neurons after liposome perfusion (Fig. 6c).

Intracellular Trafficking of Anionic Liposomes Is Associated with the Cytoskeleton and Requires PI 3-Kinase Activity-Micro-
tubule-dependent movement is a predominant means of axonal and dendritic transport in neurons (22). To determine whether intracellular trafficking of AL-Cy3ONs requires an intact microtubule network, we used nocodazole to depolymerize microtubules in hippocampal neurons. When neurons were incubated with anionic liposomes in the presence of nocodazole, the Cy3 label was found only at the edges of the cell and on the plasma membrane (Fig. 7b).
Although an actin-based framework is required for the organization of clathrin-coated pits at the cell surface, the role of actin in receptor-mediated endocytosis is still unclear (23,24). The involvement of the actin cytoskeleton in liposome endocytosis was studied using cytochalasin D to depolymerize actin filaments. In contrast to nocodazole-treated neurons, no Cy3 fluorescence was detected in neurons incubated with AL-Cy3ONs in the presence of cytochalasin D (Fig. 7c). This con- firms previous reports that the cytochalasin D-sensitive step precedes the nocodazole-sensitive step in receptor-mediated endocytosis (25).
Activation of the PI 3-kinase family of lipid kinases is involved in the rearrangement of cytoskeletal proteins, vesicle sorting, and receptor recycling during endocytosis. Specific inhibitors such as wortmannin have been widely used to study the potential sites of PI 3-kinase function in the endocytic pathway (26). To determine whether PI 3-kinase activity is necessary for neuronal endocytosis of anionic liposomes, neurons were incubated with wortmannin prior to the addition of AL-Cy3ONs. A low level of cell-associated Cy3 fluorescence was observed in neurons pretreated with wortmannin for 10 min (Fig. 7d), and increasing wortmannin exposure time beyond 10 min not only abolished the internalization of liposomes, but also caused formation of vacuoles associated with the plasma membrane (data not shown). Other studies have also documented a temporal correlation between exposure to wortmannin and drastic changes in organelle morphology (27).
Cytoplasmic Cy3ONs Do Not Significantly Colocalize with Organelles Containing Transferrin or Dextran-To determine the identity of vesicular structures containing the Cy3 label, we incubated neurons with AL-Cy3ONs along with either Oregon Green 488-transferrin as a marker for the recycling endosomal pathway or Alexa 488-dextran as a fluid-phase marker for the lysosomal degradative pathway for different time periods (Fig. 8,  a and b; and Table II). After 30 min of co-incubation, 15% of the total intracellular Cy3 label was present in the same organelles as transferrin. The proportion of total Cy3 that colocalized with transferrin did not increase beyond 25% even after 3 h of incubation. Only 20% of the total cell-associated Cy3 was present in compartments containing dextran. The lack of significant colocalization between Cy3ONs and transferrin suggested that either Cy3ONs do not undergo recycling or that recycling does occur, but with kinetics that are far slower than that of transferrin. As only 20% of Cy3ONs were present in lysosomal compartments after 3 h, the bulk of the cargo delivered by the endocytosed anionic liposomes was available to the cell.

Liposomal Lipids Are Preferentially Sorted into Recycling
Compartments-Recent evidence suggests that lipids endocytosed from the plasma membrane are sorted into either recycling endosomes or late endosomes based on the length and degree of unsaturation in their acyl chains (28). To determine whether the dioleoylphospholipids (two 18-carbon acyl chains with one cisdouble bond each) dioleoylphosphatidylcholine and dioleoylphosphatidylglycerol used in our studies were sorted according to this model, we used the head group-labeled lipid N-Rh-DOPE to tag the liposomes. As head group-labeled lipids do not spontaneously transfer between membrane leaflets, they can be expected to label liposomal lipids reliably during membrane trafficking after internalization (29). Fluorescence resonance energy transfer measurements between unlabeled and N-Rh-DOPE-labeled liposomes demonstrated that self-quenching of rhodamine within the bilayer was relieved only by calcium-induced liposome aggregation and fusion and not by simple mixing of labeled and unlabeled liposomes (data not shown). Neurons were incubated with N-Rh-DOPE-labeled liposomes and the recycling endosomal or lysosomal markers for various time periods (Fig. 8, c and d). Approximately 50% of the internalized liposomal lipid colocalized with transferrin, suggesting that the fluid nature of the phospholipids used in this study led to their preferential sorting into recycling compartments (Table II).

Trafficking of Anionic Liposomes in Hippocampal
Neurons-A classical endocytic pathway involving LRP was responsible for the uptake and intracellular transport of anionic liposomes (see Supplemental Material for a schematic of the liposome uptake pathway). Three lines of evidence indicate that LRP is crucial for liposome endocytosis (Figs. 4 and 5): pretreatment of neurons with (i) RAP or (ii) a neutralizing LRP antibody inhibited binding and subsequent internalization of liposomes, and (iii) uptake of AL-Cy3ONs in fibroblasts lacking LRP was markedly reduced compared with those expressing the receptor. Although RAP is known to block ligand binding to all members of the low-density lipoprotein receptor family, the neutralizing antibody is specific for LRP. Furthermore, previous studies have demonstrated no difference between the LRP- expressing and LRP-deficient fibroblasts in low-density lipoprotein receptor function (6), indicating that LRP, and not another member of the low-density lipoprotein receptor family, is responsible for anionic liposome uptake.
Internalization of the receptor-liposome complex was inhibited at low temperatures ( Fig. 1) because interaction of the receptor's cytoplasmic internalization signal with the clathrin adaptor protein AP2 is temperature-sensitive (30). Random dispersal of receptors on the membrane by treatment with hyperosmolar sucrose (12) or interference with scission of the coated pit by treatment with FK506 (15) both inhibited anionic liposome endocytosis (Fig. 3). An intact cytoskeleton and PI 3-kinase activity were also required for anionic liposome endocytosis (Fig. 7). Treatment with wortmannin greatly reduced Cy3 fluorescence in neurons, in agreement with reports that recruitment of LRP to the cell surface from endosomal storage pools is almost completely inhibited by wortmannin (31).
Colocalization experiments with endosomal/lysosomal markers indicated that Cy3ONs were neither recycled nor rapidly degraded within neurons (Fig. 8, a and b). Alternatively, oligonucleotide recycling may occur with slower kinetics compared with transferrin, like glycosylphosphatidylinositol-anchored proteins that are recycled approximately three times more slowly than transferrin (32). The length and fluidity of the acyl chains determined the intracellular fate of liposomal phospholipids (Fig.  8, c and d). The dioleoyllipids used in our study preferentially partitioned into fluid membrane domains with concave curvature and were sorted into transferrin-containing tubulovesicular recycling endosomes. The differential colocalization of oligonucleotides and liposomal lipids with transferrin indicated divergence in their intracellular paths 30 -60 min after endocytosis. After receptor-ligand dissociation, proteins of the annexin family on the luminal surface of endosomes may bind to the anionic liposomes. Annexins cause lateral segregation of phosphatidylglycerol in mixed bilayers of phosphatidylcholine and phosphatidylglycerol in the presence of physiological concentrations of Ca 2ϩ (33). Liposome destabilization by annexins would then provide a conduit for oligonucleotides into the cytoplasm, from where they can freely diffuse into the nucleus, accounting for the differential sorting of the lipids and oligonucleotides.
LRP-mediated Endocytosis for Macromolecule Delivery to Neurons-In marked contrast to the rapid and widespread uptake of Cy3ONs delivered by anionic liposomes, neurons incubated with Cy3ONs either free or complexed with the cationic lipid dimethylaminoethane carbamoylcholesterol/dioleoylphosphatidylethanolamine exhibited minimal intracellular Cy3 fluorescence ( Fig.  2 and Supplemental Material). LRP is a receptor that is concentrated in coated pits in the absence of any stimulus (34) and is constitutively endocytosed, irrespective of ligand binding. In Chinese hamster ovary cells, ϳ60% of cell-surface LRP is endocytosed within 5 min, and ϳ50% of the internalized LRP is recycled within 30 -60 min (35). The high sequence variability between the 31 ligand-binding sites of LRP endows different domains with unique charge densities and hydrophobic patches, resulting in distinct ligand recognition sites (36,37). Initial binding of many LRP ligands with basic residues occurs via HSPGs, which concentrate ligand on the cell surface and present it to LRP for internalization (38). In contrast, endocytosis of anionic liposomes was independent of HSPGs ( Fig. 6a), indicating that liposomes may interact directly with LRP. The observation of uptake during liposome perfusion in protein-free solution and the analysis of liposomes harvested after incubation with neurons ( Fig. 6 and Table I) suggest that proteins either present in or secreted into the medium do not bind or mediate liposome endocytosis via LRP. The spatially restricted calcium influx reported in primary neurons following multivalent ligand binding to LRP (20) was not FIG. 6. Anionic liposome endocytosis by LRP is independent of HSPGs and does not alter neuronal calcium contents. a, neurons were treated with 100 g/ml heparin prior to incubation with AL-Cy3ONs. Scale bar ϭ 5 m. b, fura-2 ratios in hippocampal neurons were not altered during perfusion of anionic liposomes labeled with N-Rh-DOPE, but increased in response to 100 M N-methyl D-aspartate. Data are from a representative field of 23 neurons from among 110 neurons imaged in five experiments. c, an image taken at 24 min (asterisk in b) indicated the uptake of N-Rh-DOPE within the same neurons in the field. An image taken at comparable gains and wavelengths prior to the anionic liposome perfusion was blank (not shown). observed in our experiments (Fig. 6, b and c), suggesting that liposomes do not induce LRP multimerization and that liposome endocytosis by LRP is distinct from that of other LRP ligands. The widespread expression of LRP (39) should enable anionic liposomes to deliver nucleic acids and possibly proteins to a broad spectrum of cell types (4) and tissues. In light of recent evidence that stressful external stimuli increase the rate of endocytosis (40), the LRP-dependent pathway of anionic liposome uptake into neurons holds important implications for therapeutic approaches in a number of diseases.
Physiological Relevance of LRP-mediated Phospholipid Up-take in Neurons-It is well established that the synthesis and metabolism of cholesterol and phospholipids in the central nervous system are compartmentalized from those in the plasma by the blood brain barrier (41). Astrocytes package cholesterol into apoE-containing lipoprotein particles that are then taken up by neurons through LRP-mediated endocytosis (42). Recent studies demonstrate that apoE lipoproteins, internalized via LRP, promote neurite outgrowth (43) and synaptogenesis (44). Furthermore, Alzheimer's disease is associated with altered phospholipid metabolism and marked reductions in the phosphatidylcholine content of synaptosomes and  plasma membranes of Alzheimer's disease hippocampi (45). Levels of apoE and phosphatidylcholine precursors increase in parallel with neurite sprouting in the lesioned hippocampus, probably enabling membrane synthesis and dendritic rearrangement (46). The importance of LRP-mediated endocytosis in the growth and maintenance of neuronal structural plasticity is underscored by the rapid mobilization of LRP to the plasma membrane from intracellular storage pools following treatment with nerve growth factor (47). Growth factor-induced increase in the endocytosis of lipoproteins may be a way of providing neurons with the lipids and proteins necessary for growth and regeneration.
LRP-ligand interactions and apoE-delivered cholesterol have been implicated in numerous intracellular signal transduction events (48). In addition, multivalent ligand binding to LRP causes local influx of Ca 2ϩ (20) and affects synaptic neurotransmission (21), probably by triggering N-methyl D-aspartate receptor clustering because the cytoplasmic tail of LRP physically interacts with N-methyl D-aspartate receptors via PSD95 (49). Modulation of downstream signaling cascades and neurotransmission by the apoE-HSPG-LRP pathway have been hypothesized to play a role in Alzheimer's disease pathogenesis (48). However, our results demonstrate for the first time that constitutive endocytosis of liposomes by LRP utilizes only one component of the LRP endocytic pathway, without involving HSPGs or altering intracellular calcium levels. The existence of an endogenous phospholipid ligand in the central nervous system that might activate LRP-mediated endocytosis independent of HSPGs remains to be determined.