Novel polyaminolipids enhance the cellular uptake of oligonucleotides.

Two new polyaminolipids have been synthesized for the purpose of improving cellular uptake of oligonucleotides. The amphipathic compounds are conjugates of spermidine or spermine linked through a carbamate bond to cholesterol. The polyaminolipids are relatively nontoxic to mammalian cells. In tissue culture assays, using fluorescent-tagged or radiolabeled triple helix-forming oligonucleotides, spermine-cholesterol and spermidine-cholesterol significantly enhance cellular uptake of the oligomers in the presence of serum. Spermine-cholesterol is comparable with DOTMA/DOPE (a 1:1 (w/w) formulation of the cationic lipid N-[1-(2,3-dioleyloxy)-propyl]-N,N,N-trimethylammonium chloride (DOTMA) and the neutral lipid dioleylphosphatidylethanolamine (DOPE)) in increasing cellular uptake of oligonucleotides, while spermidine-cholesterol is more efficient. The internalized oligonucleotides are routed to the nucleus as early as 20 min after treatment, suggesting that the polyaminolipids increase the permeability of cellular membranes to oligonucleotides. At later times, much of the incoming oligonucleotides are sequestered within punctate cytoplasmic granules, presumably compartments of endosomal origin. Coadministration with polyaminolipids markedly improves the cellular stability of the oligonucleotides; more than 80% of the material can be recovered intact up to 24 h after addition to cells. In the absence of the polyaminolipids, nearly all of the material is degraded within 6 h. These data suggest that the new polyaminolipids may be useful for the delivery of nucleic acid-based therapeutics into cells.

Oligonucleotides that specifically interfere with gene expression at the transcriptional or translational levels have the potential to be used as therapeutic agents to control the synthesis of deleterious proteins associated with viral, neoplastic, or other diseases. It is possible to select single-stranded oligonucleotides that recognize and bind to the major groove of a stretch of double-stranded DNA in a sequence-specific manner to form a triple helix (1)(2)(3). Triple helix-forming oligonucleotides targeted to the promoter region of certain genes have been used to physically block RNA synthesis in cell-free transcription assays (4 -15). Similarly, in vitro translation assays have been used to demonstrate that antisense oligonucleotides can bind mRNA targets and prevent protein synthesis (16,17).
Unfortunately, the biological efficacy of most oligonucleotides is drastically reduced if further testing is conducted in living cells. Factors such as the efficiency of cellular uptake, susceptibility to intra-and extra-cellular nucleases, salt concentration, and nonspecific association with cellular or extracellular components may adversely affect the specific function of oligonucleotides in cells (18).
Typical triple helix-forming and antisense oligonucleotides are large (ϳ5000 -10,000 Da) hydrophilic compounds that do not enter cells or reach their intracellular targets efficiently. Unprotected phosphodiester oligonucleotides are highly susceptible to nucleases in the serum and inside cells, but their t1 ⁄2 in vivo can be increased from a few minutes to several hours by simply blocking the 3Ј end, for example by a propanolamine group (19 -21). More extensive modifications to the backbone (e.g. phosphorothioate or methylphosphonate backbones) or to individual bases can also result in greater in vivo stability, binding affinity, and uptake properties of the oligonucleotides (17,(22)(23)(24)(25)(26)(27). It is believed that nucleic acid internalization occurs mainly via the endocytic pathway (28). Uptake studies in cultured cells using fluorescent oligonucleotides show that the internalized material accumulates inside granular cytoplasmic compartments, most likely of endosomal or lysosomal origin, and only a small fraction reaches the nucleus (29 -31). Interestingly, when oligonucleotides are introduced directly into the cytosol by microinjection, nearly all of the material is quickly transported to the nucleus with a t1 ⁄2 of approximately 5 min (32,33). These observations suggest that for oligonucleotides, movement across the lipid bilayer of the plasma membrane or the endosomal membrane is the rate-limiting step in the import process.
One approach to improve the cellular penetration of oligonucleotides is the coadministration of oligonucleotides with cationic lipids (34,35). This class of uptake enhancer has found wide application in facilitating delivery of DNA (36,37), RNA (38), oligonucleotides (34), and protein (39) into living cells. The mechanism by which these amphipathic compounds promote nucleic acid entry into cells is unknown, but it is possible that the cationic portion interacts with the nucleic acid, while the hydrophobic lipid group associates with membrane lipid bilayers, resulting in fusion with or transient disruption of the membrane and discharge of the nucleic acids into the cytosol. Cell culture assays have shown that the efficiency of cationic lipid-mediated delivery is influenced by numerous factors, including the composition and quantity of the nucleic acid and cationic lipid, cell type, concentration of serum in the medium, and time of exposure to the cell. In addition, many available cationic lipid preparations are toxic to cells at concentrations near their effective doses if exposure times are extended to several hours (34), suggesting that the molecules are not easily metabolized. Thus, there remains a need for less toxic and more efficient delivery vehicles for oligonucleotides and other gene-based therapeutics. To this end, we have synthesized two novel polyaminolipid uptake enhancers, spermine-and spermidinecholesterol, and have investigated their effect on the cellular uptake and distribution of oligonucleotides.

EXPERIMENTAL PROCEDURES
Synthesis of Polyaminolipids-Cholesteryl chloroformate (Aldrich) in methylene chloride was added dropwise to a solution of spermidine or spermine (Aldrich) in methylene chloride and N,N-diisopropylethylamine and stirred at room temperature for 2 h. Conjugation of cholesterol to spermine or spermidine can occur at either end of the polyamine. Because spermidine is an asymmetric molecule, a mixture of two products is expected from its conjugation to a single cholesterol (Fig. 1). The polyamine-(mono)cholesterol conjugates were purified from the reaction mix by chromatography on silica gel columns using methanol: methylene chloride (1:1) as eluent. The final products (white to yellowwhite solids) were obtained by evaporation of the solvent and characterized by NMR. The preparations were approximately 95% pure, the remaining mass presumably consisting of polyamine conjugated to two cholesterols. The reagents were kept desiccated at Ϫ20°C for long term storage, or when used in cell-based experiments, the polyaminolipids were first solubilized to a concentration of ϳ3-5 mg/ml at room temperature and then diluted as necessary in physiological buffers.
Cell Culture-Vero (African green monkey kidney) cells (ATCC CCL 81, American Type Culture Collection, Rockville, MD) were grown in minimal essential medium (MEM) 1 supplemented with 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin (Life Technologies, Inc.) at 37°C in a humidified atmosphere containing 5% CO 2 . Cell cultures were monitored for and maintained free of mycoplasma infection as shown by a solution hybridization assay (GENProbe®, Fisher).
Oligonucleotide Preparation-The oligonucleotides (ODN-1 and ODN-F1) used in the experiments described here are variants of a 36-mer G-rich phosphodiester oligonucleotide (5Ј-gtggttggtggtggtgtgtgggtttggggtgggggg-3Ј) that has been described previously (40). Oligonucleotides were synthesized using standard ␤-cyanoethyl phosphoramidites (PerSeptive Biosystems, Bedford, MA) on an Applied Biosystems model 394 automated DNA synthesizer. To improve nuclease resistance, the oligonucleotides were modified at their 3Ј ends with a propanolamine moiety (21). For this, amino modifier CPG (Glen Research, Sterling, VA) was used as the starting support. ODN-F1 was obtained by tagging the 5Ј end with fluorescein, using fluorescein amidite during synthesis (Pharmacia Biotech Inc.). The deprotected compounds were purified by anion-exchange chromatography (Q-Sepharose HP; Pharmacia), and the eluted full-length compounds were concentrated and desalted on a Sep-Pak C 18 cartridge (Waters, Milford, MA) or by membrane filtration (Filtron Technology Corp., Northborough, MA).
ODN-1 was radiolabeled at the 5Ј end with ␥-35 S-ATP (DuPont NEN) using polynucleotide kinase (Boehringer Mannheim) according to standard procedures (41), followed by Sephadex G-100 exclusion chromatography (Pharmacia) in 5 mM HEPES-KOH, 50 mM NaCl, pH 7.5, buffer to remove unincorporated label. The specific activity of the purified labeled ODN-1 was approximately 150 Ci/mmol. Cellular Uptake of Radiolabeled Oligonucleotide-Cells were seeded at a density of 1 ϫ 10 5 cells/well in Corning 24-well plates (Corning). After 24 h, the medium in each well was replaced with fresh complete medium containing 0.1 M unlabeled ODN-1 and 10 6 cpm 35 S-labeled ODN-1 in the presence or absence of SpdC, SpC (each at 10 g/ml), or DOTMA/DOPE (at 20 g/ml, of which 10 g/ml is DOTMA). In preliminary experiments, we had determined that ϳ1.5-5-fold excess of SpdC over oligonucleotide (by mass) was optimal for uptake experiments. After 0, 3, 6, or 24 h of incubation, cells were washed six times with MEM, harvested by trypsin-EDTA treatment, and washed once more with MEM. The radioactivity associated with cells was quantified by scintillation counting. Intracellular oligonucleotide concentration was calculated based on the specific activity of the radiolabeled oligonucleotide solution and the estimated volume of Vero cells (ϳ2 ϫ 10 Ϫ6 l/cell).
Light Microscope Autoradiography-Cells were seeded at a density of 7500 cells/chamber on eight-chamber microscope slides (Nunc, Naperville, IL) in 200 l of medium. After 24 h, medium in each chamber was replaced with fresh medium containing 0.1 M (4.44 ϫ 10 5 cpm) 35 Slabeled ODN-1 in the presence or absence of SpdC at a 1:5 mass ratio of oligomer to cationic lipid. Cells were incubated with the oligonucleotide formulations for 30 min and then rinsed four times with ice-cold Dulbecco's PBS (D-PBS) containing 0.1% sodium azide and fixed with Ϫ20°C methanol for 6 min. After fixation and rinsing with D-PBS, the slides were air-dried for approximately 15 min. Microscopy autoradiography procedure was adapted from methods provided by Eastman Kodak Co. and described in Freshney (42). Slides were coated with Kodak NTB2 emulsion (in the darkroom), dried for 1 h, placed in light-tight plastic boxes with desiccant, sealed with electrical tape, and allowed to expose for 6 days at 4°C. Subsequently, slides were developed in Kodak Dektol developer (diluted 1:1 with distilled water) for 2 min, rinsed 15 s in distilled H 2 O, fixed for 5 min in Kodak Fixer solution, and rinsed 5 times in distilled H 2 O (ϳ1 min/wash). Slides were covered with coverslips, mounted with water, and viewed by bright field or phase contrast microscopy on a Nikon Labophot 2 system (Melville, NY). Images were recorded on Kodak Ektachrome ASA400 film.
Intracellular Stability of Radiolabeled Oligonucleotide-Cells were seeded at a density of 4 ϫ 10 5 cells/well in six-well plates (Corning). After 24 h, the medium in each well was replaced with fresh medium containing 1 M (4 ϫ 10 6 cpm) 35 S-labeled oligonucleotide and 20 g/ml SpdC. After 0, 3, 6, or 24 h of incubation, cells were washed six times with MEM, harvested by trypsin-EDTA treatment, and washed once more with MEM. The radioactivity associated with whole cells was quantified by scintillation counting of an aliquot. Oligonucleotides were then isolated from the remaining cells by phenol extraction and ethanol precipitation and electrophoresed on 12% acrylamide/7 M urea gels. The gels were fixed, soaked in intensifying solutions (Entensify; DuPont NEN), dried, and exposed to film. The amount of radioactivity associated with individual bands on gels was quantified by a ␤-Scope analysis (Endogen, Cambridge, MA).
Fluorescence Microscopy-Cells were seeded at a density of 2.5 ϫ 10 5 cells/dish in 60-mm dishes containing several glass coverslips. After 24 h, the medium in each dish was replaced with fresh medium containing 0.1 M fluorescein-tagged oligonucleotide ODN-F1 in the presence or absence of SpdC (1:5 mass ratio of oligonucleotide to SpdC). At 20 min, 1 h, 5 h, or 24 h after the addition of oligonucleotide, the cells were washed four times with D-PBS containing 0.1% sodium azide, fixed with Ϫ20°C methanol at for 6 min (fixation by a 2% paraformaldehyde solution also gave identical results), and washed four more times with D-PBS. Coverslips were mounted on glass slides using Vectashield mounting medium for fluorescence (Vector Laboratories, Burlingame, CA), sealed with clear nail polish, and observed by phase contrast or fluorescence microscopy on a Nikon Labophot 2 system, using 100ϫ objective.

RESULTS AND DISCUSSION
Polyaminolipids Are Simple and Relatively Nontoxic Cationic Lipids-Several methods of enhancing oligonucleotide uptake have been established, including lipophilic modification of oligonucleotides (23,25,(43)(44)(45)(46)(47)(48)(49)(50), incorporation of oligonucleotides into liposomes (51)(52)(53), and the coadministration of oligonucleotides with cationic lipids (34,35). We have focused on cationic lipids as uptake enhancers because of ease of use, feasibility in screening large numbers of oligonucleotides, and potential versatility for delivery of other nucleic acid therapeutics. Although several cationic lipid preparations have been described (54 -56), most lose their activity in the presence of serum, and many are toxic to cells after less than 24 h of exposure. In an effort to improve upon available compounds, two novel polyaminolipids, spermine-and spermidine-cholesterol (SpC and SpdC, respectively), have been synthesized (Fig. 1).
Components of the polyaminolipids are ubiquitous in mammalian cells. The intracellular concentration of spermine, spermidine, and related polyamines is estimated to be in the millimolar range (57). The polyamines participate in the structural organization of the genome by binding to the nucleic acids with high affinity and neutralizing their negative charge, allowing the compact packaging of DNA (58). Cholesterol is an abundant constituent of animal cell membranes with a major role in the modulation of membrane fluidity (59). The conjugates of polyamines and cholesterol, joined through a carbamate linkage, are expected to be biodegradable and nontoxic. To test this, the cytotoxic effect of the polyaminolipids on cell proliferation was assayed using a sensitive, nonradioactive cell proliferation assay (see "Experimental Procedures"). As shown in Fig. 2, SpdC and SpC had no discernible effect on cells when administered at levels below 25 g/ml for up to 4 days, and the cells appeared healthy and viable even at concentrations as high as 100 g/ml. Light microscopic examination of treated cells revealed no gross changes in cellular morphology. In contrast, DOTMA/DOPE was highly toxic to cells above approxi-mately 20 g/ml (10 g/ml DOTMA), and at 100 g/ml there were no viable cells left after 4 days of continuous treatment. Similar viability assays were performed with mixtures of SpdC and oligonucleotides, and again, no cytotoxic effects were observed at below 20 g/ml of each (data not shown). The half-life of the polyaminolipids in vivo is not yet known, but the carbamate linkage in these carriers is designed to be biodegradable soon after completing delivery of the cargo, thus reducing the potential of observing toxic effects. The carbamate bond is already known to be chemically unstable at high pH and elevated temperatures (60). The oligonucleotides begin to have some cytotoxic effects of their own at extracellular concentrations above about 50 M (61). In experiments described in this paper, SpC and SpdC were used at the relatively nontoxic concentrations of 20 g/ml or less and oligonucleotides at 1 M or less.
Polyaminolipids Enhance the Cellular Uptake of Oligonucleotides-The amount of radiolabeled ODN-1 taken up by Vero cells in the presence or absence of various uptake enhancers was quantitated by scintillation counting. Cells were treated with 0.1 M 35 S-ODN-1 in the presence or absence of SpdC, SpC, or DOTMA/DOPE. Oligonucleotides were labeled with 35 S instead of 32 P, because data from preliminary studies showed that terminal 35 S is approximately 10-fold more resistant to cleavage by enzymes in the growth medium (data not shown). Incubation of Vero cells with 0.1 M ODN-1 in the absence of cationic lipids resulted in cellular oligonucleotide concentration of approximately 0.3 M (Fig. 3). Coadministration of the same oligonucleotide with cationic lipids enhanced cellular uptake by 30 -65-fold after 24 h of incubation. Improved oligonucleotide uptake by cells in the presence of polyaminolipids was also confirmed by light microscopy autoradiography studies. Vero cells were incubated with 0.1 M radiolabeled ODN-1 in the presence of SpdC or SpC for 30 min and processed for autoradiography, as described under "Experimental Procedures." Observation of dark silver grains in and around the nuclear region indicated that the polyaminolipids had markedly enhanced the uptake of ODN-1 by cells (data not shown). In contrast, untreated cells contained few or no silver grains, confirming the idea that there is very little unaided uptake of oligonucleotides by cells. Among the uptake enhancers, SpdC was significantly more efficient at improving oligonucleotide entry into cells than either SpC or DOTMA/DOPE. Improved in Vivo Stability of Oligonucleotides Coadministered with SpdC-To examine the intracellular stability of the oligomers taken up in the presence of cationic lipid, Vero cells were incubated with 1 M radiolabeled ODN-1 coadministered with 20 g/ml SpdC. At 0, 3, 6, or 24 h after addition, oligonucleotides were isolated from the cells by phenol extraction and ethanol precipitation and fractionated by denaturing polyacrylamide gel electrophoresis (Fig. 4). The absence of significant amounts of degradation products in the cellular oligonucleotide samples suggested that the internalized oligonucleotides are somehow protected from nucleases. Quantification of intact oligonucleotide by ␤-scope analysis of the gel confirmed that greater than 80% of the radiolabeled material recovered from cells after a 24-h incubation is intact. The phosphodiester oligonucleotides used in these studies were blocked at their 3Ј end with a propanolamine moiety to confer greater resistance against serum exonucleases (19 -21). Interestingly, the 5Ј end seems to require no such protection. Oligonucleotide-3Ј-propanolamine conjugates have been administered to mice and found to remain intact in the circulation and to be stable in tissues for at least eight hours (21). SpdC appears to increase the in vivo stability to an even greater extent, partly because binding of the polyaminolipid to the nucleic acid may prevent the nucleases from recognizing and cleaving the oligonucleotide and partly because the SpdC-ODN complex may be stored within intracellular compartments that are not easily accessible to nucleases (see below). Vero cells were treated with ODN-F1 in the presence or absence of SpC or SpdC and the intracellular localization of fluorescein was examined by microscopy ( Fig. 5; only the data using SpdC are shown). In the absence of uptake enhancers, oligonucleotide was taken up by cells at a low level, with much of the internalized material in perinuclear cytoplasmic compartments. Coadministration of SpC or SpdC with oligonucleotide resulted in brighter nuclear fluorescence by 20 min after addition (at earlier time points, from 5 to 15 min, the cellassociated fluorescence was very weak and nuclear fluorescence was barely discernible). The nuclear fluorescein was expected to be still attached to intact oligonucleotides, because studies by Fisher et al. (62) have shown that fluorescein, if detached from oligonucleotides, is quickly discarded from cells. Cellular and nuclear uptake as early as 20 min after administration of SpdC/oligonucleotide mix to cells suggests that internalization is occurring within the time frames expected of endocytic events. Furthermore, entry into the nucleus indicates that the polyaminolipids can somehow increase the permeability of cellular membranes to the oligonucleotides. At approximately 5 h after treatment, cells exhibited nuclear fluorescence as well as some punctate cytoplasmic staining. Every cell on the slide contained fluorescent material, although there was considerable variability in the amount of oligonucleotide internalized by cells, even within the same microscopic field. Some decrease of nuclear fluorescence was observed after 24-h incubation, concomitant with an accumulation of large fluorescent granules in the cytoplasm. This loss in nuclear fluorescence over time has been seen in other studies (34,62) and may represent some degradation of the fluorescein or the oligonucleotides in the nucleus, followed by the export of fluorescent metabolites. The composition of the large cytoplasmic granules is currently unknown, but may represent endosomal accumulation of the oligonucleotides. In an effort to confirm this, and to determine whether incubation of cells with SpdC/oligonucleotide leads to altered endosomal morphology, we probed the treated and untreated cells with a monoclonal antibody to the human transferrin receptor (a marker for endosomes), followed by incubation with Texas Red-conjugated goat anti-mouse antibody. Subsequent examination of cells by fluorescence microscopy showed the colocalization of transferrin receptor and internalized oligonucleotides in numerous cytoplasmic compartments (images not shown). In the case of SpdC/ODN-F1treated cells, the compartments became larger and brighter with increasing time of incubation, confirming that much of the internalized oligonucleotide was sequestered within the endosome-derived vesicles (Fig. 5), protected from cellular nucleases (Fig. 4).

Subcellular Localization of Fluorescent Oligonucleotides-
Further studies of the polyaminolipids will be required to determine their mechanism of action and how well these compounds improve the actual biological activity of internalized oligonucleotides. Factors that may influence uptake include cell type, ratio of cationic lipid to the oligonucleotide, extracellular pH, and protein and salt concentrations. Moreover, the difference in toxicity and activity between SpC and SpdC would have been difficult to predict based on simple structural information. It was discussed earlier that SpdC described here is really a mixture of two isoforms (Fig. 1). It is not known whether one or both of the isoforms are equally effective at enhancing uptake. Synthesis of individual isoforms would require more complex regioselective synthesis, but may lead to a more effective uptake enhancer. Other polyamine-lipid conjugates (63) have been widely used for improving the cellular uptake of nucleic acids, but in those compounds the lipids are fatty acyl chains linked through an internal site on the polyamine. That may, in part, account for some of the major differ-ences in the toxicity and uptake properties.
The polyamine-cholesterol conjugates described in this report are relatively nontoxic compounds that markedly enhance the uptake of oligonucleotides into cells. It is likely that the activity of the polyaminolipids can be optimized by properly formulating the compounds, e.g. with fusogenic lipids or peptides and/or by chemically altering the polyamine or steroid portions of the molecule. In addition, the delivery system can be further fine-tuned for different types of cells and different types of cargo. It is possible that SpC, SpdC, and a related family of compounds will have a much broader application than for just improving the cellular uptake of oligonucleotides. They may be useful for enhancing the delivery of other nucleic acid-based therapeutics and perhaps even acidic peptides or proteins or other polyanionic therapeutics.