3334–3341 Nucleic Acids Research, 1999, Vol. 27, No. 16 © 1999 Oxford University Press

Novel cationic amphiphiles as delivery agents for antisense oligonucleotides R. K. DeLong1, Hoon Yoo1,S.K.Alahari1,M.Fisher1,S.M.Short1,S.H.Kang2,R.Kole1,2, V. Janout3,S.L.Regan3 and R. L. Juliano1,*

1Department of Pharmacology and 2Lineberger Cancer Center, University of North Carolina, Chapel Hill, NC 27599, USA and 3Department of Chemistry, Lehigh University, Bethlehem, PA 18015, USA

Received as resubmission April 19, 1999; Revised and Accepted May 7, 1999

ABSTRACT INTRODUCTION There has been great interest recently in therapeutic Currently there is substantial interest in the use of antisense use of nucleic acids including genes, ribozymes and oligonucleotides, ribozymes, high molecular weight DNA and antisense oligonucleotides. Despite recent improve- other types of nucleic acids to modify gene expression for ments in delivering antisense oligonucleotides to therapeutic purposes (1–5). Antisense oligonucleotides are cells in culture, nucleic acid-based therapy is still usually administered in vitro or in vivo either in free form or often limited by the poor penetration of the nucleic complexed with various cationic -based formulations (6–9). acid into the cytoplasm and nucleus of cells. In this Gene therapy has primarily relied on viral vectors (1), but there report we describe nucleic acid delivery to cells is increasing interest in use of non-viral gene delivery systems based on , polymers or combinations of both (10–14). using a series of novel cationic amphiphiles containing The ability of a nucleic acid to achieve a desired function cholic acid moieties linked via alkylamino side within cells is influenced by several critical factors in addition chains. We term these agents ‘molecular umbrellas’ to its own innate biological activity. These include the hetero- since the cationic alkylamino chains provide a ‘handle’ geneity of uptake within the population; the degree to for binding of nucleic acids, while the cholic acid which the nucleic acid reaches appropriate cellular compart- moieties are likely to interact with the ments (cytosol and nucleus); and, when a carrier is used, the allowing the highly charged nucleic acid backbone to extent to which the nucleic acid can dissociate from the carrier traverse across the . Optimal gene and and attain its biologically active form in the target cell. The oligonucleotide delivery to cells was afforded by a commercially available ‘cytofectins’ (cationic liposomes) derivative (amphiphile 5) containing four cholic acid currently widely used as carriers for genes and oligonucleotides moieties. With this amphiphile used as a constituent typically form large, complex aggregates with nucleic acids in cationic liposomes, a 4–5 log increase in reporter (15). Despite some recent progress (16), there is still relatively gene delivery was measured. This amphiphile used little known about the ability of such complexes to enter key alone provided a 250-fold enhancement of oligo- cellular compartments or to release biologically active nucleic nucleotide association with cells as observed by flow acids. In addition, most cationic lipid complexes do not func- cytometry. A substantial fraction of cells exposed to tion well in the presence of plasma , although there are complexes of amphiphile 5 and fluorescent oligo- some exceptions. While some studies with cationic lipids have nucleotide showed nuclear accumulation of the provided evidence for gene expression in vivo (17,18), there are still many problems with access of the complex to desired fluorophore. Enhanced pharmacological effectiveness tissue sites and with attaining and sustaining appropriate levels of antisense oligonucleotides complexed with of gene expression in the target issue. amphiphile 5 was observed using an antisense In vivo studies with therapeutic antisense oligonucleotides splicing correction assay that activates a Luciferase have primarily utilized administration of the oligonucleotide reporter. Intracellular delivery, nuclear localization alone, without complexation to carriers (2,19,20). However, a and pharmacological effectiveness of oligonucleo- substantial amount of in vitro data suggests that carrier tides using amphiphile 5 were similar to those systems, including cytofectins, can substantially increase the afforded by commercial cytofectins. However, in con- delivery of antisense compounds to cells, as well as the con- trast to most commercial cytofectins, the umbrella sequent pharmacological effects (6,7). Thus it is possible that amphiphile showed substantial delivery activity even suitable carrier systems could increase the in vivo efficacy of in the presence of high of serum. antisense oligonucleotides as well.

*To whom correspondence should be addressed. Tel: +1 919 966 4383; Fax: +1 919 966 5640; Email: [email protected] Present address: R. K. DeLong, Megabios Corp., Burlingame, CA, USA The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors. Nucleic Acids Research, 1999, Vol. 27, No. 16 3335

In the studies reported here, we have examined novel cationic Preparation of nucleic acid complexes amphiphiles as delivery agents for nucleic acids, particularly Amphiphile/liposome/DNA complexes. Amphiphiles 1–5 dis- for antisense oligonucleotides. These amphiphiles utilize one solved in methanol (1 mg/ml), were combined with an equal or more cholic acid moieties tethered to alkylamino side chains volume of dioleoylphosphatidyl ethanolamine (DOPE) (1 mg/ that are derivatives of spermine and/or spermidine (21). The ml) (Avanti Polar Lipids) in a of chloroform. After amphiphiles were designed to act as ‘molecular umbrellas’ evaporation of the , the was suspended in sterile where the hydrophobic portion of the cholic acid moiety water (1 mg/ml). Each lipid–amphiphile solution (1L–5L) was allows membrane interaction and penetration and the vortexed for 5 min, bath sonicated (30 min) and stored at 4°C µ hydrophilic face of the cholic acid moiety, along with the under N2. Complexes of plasmid (1 g/ml) or oligonucleotide alkylamino side chain, permits interaction with the negatively (10 nM–1 µM) were formed by combining the DNA stock charged nucleic acid. In studies with cultured cells, we have dissolved in 100 µlofOpti-Mem(GibcoBRL)withanequal tested several forms of ‘molecular umbrella’ amphiphiles for volume of the (1L–5L) amphiphile/liposomes (10 µg/ml). The their ability to enhance DNA transfection and for their ability complexes were incubated for 30 min at room temperature and to deliver antisense oligonucleotides. These amphiphiles then added directly to cells in 1 ml cell culture medium. proved to be only moderately effective as DNA transfection agents and were clearly less efficient than the commercially Amphiphile/oligonucleotide complexes. Stocks of oligonucleo- available cytofectins Lipofectin® and Lipofectamine®.How- tide (50–100 µM) were diluted to the appropriate ever, the umbrella amphiphiles were quite effective in promoting for complexation (10 nM–10 µM) with sterile water. The oligo- delivery of antisense oligonucleotides to the nucleus, as nucleotide was then stirred within 0.5 ml glass V-vials (Wheaton) observed by fluorescence microscopy. Further, we found that in 40–100 µl sterile water. Amphiphile at 5–10 µg/ml in 1% the umbrella could enhance the pharmacological methanol/water was then added rapidly via a 0.5 cc U-100 effectiveness of antisense oligonucleotides to a significant Insulin Syringe (Becton Dickinson) and stirred for 15 min at extent. However, in contrast to many lipid cytofectins, the room temperature. The complexes were incubated for 30 min umbrella amphiphile/oligonucleotide complexes retained sub- at room temperature and then added directly to cells in 1 ml stantial activity in the presence of serum. This suggests that cell culture medium. these compounds may have some value for in vivo delivery of antisense oligonucleotides. Cytofectin complexes. Complexes of oligonucleotides with Lipofectin® or Lipofectamine® were prepared as previously described (6), essentially according to the manufacturer’s MATERIALS AND METHODS directions. Cholic acid–polyamine conjugates (molecular umbrellas) Plasmid transfection assay Specific procedures that were used to prepared the cholic acid– For transfections, ECV304 or MDR3T3 cells were seeded in 6- polyamine conjugates 1, 3, 4 and 5 were essentially those well dishes ~24 h prior to transfection and used at 60–80% previously described (21). Compound 2, which was prepared confluence. Cells were transfected for 4 h in serum-free by similar methods, exhibited the expected 1H NMR spectrum. medium with ~1 µgofpGL3plasmid(Promega)andeither 20 µg/ml of Lipofectamine® (Gibco BRL) or amphiphile (5–10 µg/ Cell culture ml); cells were then placed in serum-containing growth Handling and maintenance of multi-drug resistant NIH 3T3 medium and allowed to recover for 24 h. Transfected cells cells has been previously described (6). Human endothelial were lysed in a buffer containing 0.25 M Tris (pH 7.8), 1% Triton cells (ECV304; ATCC) were routinely grown in Medium 199 X-100 and 1 mM dithiothreitol (DTT). Lysate concen- (Gibco) supplemented with 10% heat inactivated fetal bovine trations were determined using the bicinchonic acid assay (Pierce Chemical Co.). Cell lysates were stored at –70°Cuntil serum (FBS; Hyclone), 2 mM L-glutamine and penicillin (50 U/ml)/ µ ° used. Luciferase activity was measured in cell lysates using streptomycin (50 g/ml) in 5% CO2/95% air at 37 C. ECV304 2 reagents from Analytical Luminesence Laboratory (San Diego, cultures were grown to confluence in 185 cm Nunc tissue culture CA). Briefly, using the automatic injector of the Monolight® 2010 flasks and typically split at a ratio of 1:6. HeLa cells containing µ µ β Luminometer (Analytical Luminescence), 100 l (~200 g) of a modified Luciferase gene interrupted by a -globin intron cell extract was mixed sequentially with 100 µl of Luciferase (HeLa/705) were maintained as described (22). buffer (3 mM ATP, 15 mM MgSO4,30mMTricine,10mMDTT, µ Nucleic acids pH 7.8) and 100 l of luciferase substrate (1 mM D-Luciferin); the light output was measured for 10 s. Determinations of protein A2'-O-methyl-phosphorothioate oligonucleotide complementary concentration were used to convert raw activity values to to an aberrant splice site in the β-globin intron (5'-CCUCU- specific activities for all samples. All transfections were per- UACCUCAGUUACA-3') was prepared as decribed (22). formed in triplicate. FITC or cyanine-5 fluorophore derivatives were coupled to the 5'-end of oligonucleotides as phosphoramidites (Glen Research) Toxicity and growth assays during the final step in automated synthesis. Luciferase plasmid Cytotoxicity studies were performed by plating MDR3T3 cells pGL3 (Promega), under the control of the SV40 promoter, was into 24-well plates (Nunc) at 8 × 104 cells/well. Cells were maintained as a 1 mg/ml stock in -buffered saline incubated with complexes of nucleic acid and umbrella (PBS) and stored at –4°C. amphiphile or with nucleic acids and commercial cytofectins, 3336 Nucleic Acids Research, 1999, Vol. 27, No. 16

or with control medium for 24 h, rinsed twice with PBS, ATP, 15 mM MgSO4, 30 mM Tricine, 10 mM DTT, pH 7.8) followed by the addition of 2 ml OPTI-MEM (Gibco BRL) and and 100 µl of Luciferase substrate (1 mM D-luciferin). The incubation for a further 24 h. The surviving fraction was deter- light emission is quantified for 10 s using a Monolight® 2010 mined by the MTT dye assay, measuring absorbance at 540 nm luminometer (Analytical Luminescence Laboratory). Luciferase with an automated microplate reader (BioTech EL 340) as activity is expressed as relative light units (RLU) per well. Light described (23). emission is normalized to the protein concentration of each sample, determined according to the bicinchonic acid assay Flow cytometry (Pierce Chemical Co.) for protein concentration. The flow cytometry technique for oligonucleotide–cell asso- ciation has been previously reported (6,24). Briefly, 500 000 Oligonucleotide/amphiphile 5 complexes in serum. Effects of cells/well were plated in a Nunclon 6-well plate and allowed to serum on the oligonucleotide uptake mediated by umbrella attach for 18–24 h. Amphiphile–oligonucleotide complexes amphiphile was performed by plating cells into a 24-well tray formed as above were made with FITC-phosphorothioate oligo- (Nunc) at 4 × 104 cells/well. Cells were incubated for 12 h at nucleotide (100 nM–1 µM). Complexes were added to cells in 37°C. A 100 µl aliquot of an oligonucleotide at a given concen- a 40–100 µl volume and incubated for 18 h at 37°C. Cells were tration in Opti-MEM was mixed with 100 µl of Opti-MEM then rinsed with PBS, trypsinized, pelleted, and resuspended in containing 6.5 µg(3.0µM) of amphiphile 5.Afterbeing PBS (100 000 cells/ml) prior to flow cytometry. briefly mixed, the preparation was left at room temperature for ~15 min, followed by to 1 ml with a given percentage Confocal microscopy of serum (FBS) in Opti-MEM. A 200 µlaliquotofthecomplex For confocal microscopy, cells were grown on glass coverslips solution was plated in a Nunclon 24-well tray. The cells were (Corning) coated with fibronectin (10 µg/ml) and treated with incubated for 6 h and subsequently the medium was replaced complexes of cyanine-derivatized oligonucleotide, essentially with the same percentage of serum containing DMEM. An as previously described (6,24). additional 18 h later, the cells were rinsed with PBS and lysed in 40 µl of lysis buffer on ice for 15 min. Following centrifugation Antisense assays (13 000 r.p.m., 2 min), 10 µl of supernatant cell extract was Splicing correction assay (β-globin system). In the thalassemic mixed with 100 µl of luciferase assay buffer to measure the human β-globin gene, a T→G mutation at position 705 of luciferase activity as described in the antisense splicing correction intron 2 (IVS2-705) improves the match of the surrounding assay. sequence to the consensus donor (5') splice site (ACTGAT/ GTAAGA to ACTGAG/GTAAGA; slash indicates the splice RESULTS site). In the transcribed IVS2-705 pre-mRNA, the presence of the created 5' splice site activates an acceptor (3') splice site Amphiphile structure 126 nt upstream, resulting in incorrectly spliced β-globin mRNA containing a fragment of the intron. In a novel method The structures of the cholic acid-based amphiphiles used in this developed by Sierakowska et al.(25)andextendedbyKanget al. study are shown in Figure 1 (21). Molecular umbrellas 2–5 22), HeLa-based cell lines were created that have a human β- contain two or more cholic acid moieties covalently coupled to globin intron 2 with the 705 mutation inserted into a Luciferase spermine and/or spermidine moieties to provide a cationic reporter gene. In the absence of treatment, the presence of the ‘tail’ or ‘handle’ to which anionic oligonucleotides can bind. intron prevents correct splicing and thus blocks Luciferase The design principle involves the linking of two or more rigid expression. When the cells are effectively treated with a 2'-O- amphiphilic ‘walls’ or ‘ribs’ to the cationic ‘handle’ to form an methyl-phosphorothioate oligonucleotide targeted to the aber- ‘umbrella’. The flexible nature of the linkages allows the cholate rant splice site, the oligonucleotides restore correct splicing residues to be in a compact conformation (‘umbrella‘) in a and thus permit expression of the Luciferase protein. This hydrophobic membrane environment, thus possibly shielding approach provides a basis for an excellent assay of antisense the bound oligonucleotide, and in an extended conformation activity and delivery, since only the presence of active oligo- (‘inverted umbrella‘) in a hydrophilic environment. For purposes nucleotide within the nuclei of living cells will permit correct of comparison, a ‘single walled’ analog (1) has been included. splicing. This assay was used, essentially as described (22), to Cytotoxicity studies evaluate the ability of the umbrella amphiphiles to deliver oligo- nucleotides. Typically a 100 µl aliquot of oligonucleotide at a Using the MTT assay (23) a systematic comparison of the given concentration in Opti-MEM is mixed with 100 µlof cytotoxicity of the various umbrella amphiphiles was performed. Opti-MEM containing various concentrations of delivery Figure 2 shows a representative acute toxicity versus concen- agent (amphiphile 5 or commercial cytofectin). After being tration plot for amphiphile 5.AsshowninTable1,therewasa briefly mixed, the preparation is left undisturbed at room tem- range of approximate LD10 and LD50 values for compounds 1–5. perature for ~15 min, followed by dilution to 1 ml with Opti- In general, complexation with oligonucleotides tended to modestly MEM before being layered on the cells. The cells are incubated reduce the toxicity of the amphiphiles; however, a variety of for 6 h and subsequently the medium is replaced with 10% effects were observed. In further studies, the various delivery FBS/DMEM. An additional 18 h later, the cells are rinsed with agents were used at equitoxic concentrations. PBS and lysed in 100 µl of lysis buffer (200 mM Tris–HCl, pH 7.8, 2 mM EDTA, 0.05% Triton X-100) on ice for 15 min. Gene expression Following centrifugation (13 000 r.p.m., 2 min), 5 µl of supernatant The ability of the umbrella compounds 1–5 to deliver plasmid cell extract is mixed with 100 µl of Luciferase assay buffer (3 mM DNA was evaluated using luciferase expression in fibroblastic Nucleic Acids Research, 1999, Vol. 27, No. 16 3337

Figure 1. Structure of the umbrella amphiphile series. The structures of the five amphiphiles used in this study are shown. The cholic acid ring system is indica ted as a gray and black rectangle with the black portion representing the non-polar portion of this so-called ‘facial amphiphile’. The nitrogens of the spermidine and spermine residues provide positive charge at physiological pH that permits binding of anionic oligonucleotides.

Table 1. Toxicity of umbrella amphiphiles

Amphiphile –Oligo +Oligo µ µ µ µ LD10 ( g/ml) LD50 ( g/ml) LD10 ( g/ml) LD50 ( g/ml) 1 nt nt nt nt 2 2.4 15.0 1.7 4.0 3 2.0 8.0 7.0 20.0 4 1.7 16.0 3.0 15.0 5 0.3 4.0 2.7 20.0

Data are based on the MTT assay. nt, no toxicity observed at highest concentration tested, 10 µg/ml. A single dose–toxicity curve was run for each condition tested. The individual points for the curve were the means and standard errors of three replicates.

and endothelial cell lines, as shown in Figure 3. When complexed umbrella/DOPE formulations were still many-fold less than with DNA and formulated with the DOPE, several that attained with Lipofectin® or Lipofectamine®. Endothelial amphiphiles were quite active, giving luminescence readings cells were consistently less efficiently transfected than fibro- 4–5 log units higher than DNA alone. Interestingly, a significant blasts either with the Lipofectamine® positive control or with amount of expression was also obtained in the absence of the umbrella compounds and showed a greater dependence on added DOPE. However, the best transfections attained with the thepresenceofDOPE. 3338 Nucleic Acids Research, 1999, Vol. 27, No. 16

Table 2. Quantitation of oligonucleotide uptake and nuclear localization by confocal microscopy

Condition Nuclear (%) Nuclear fluorescence Total cell fluorescence Free oligo, 1 µM1(n = 95) 0.4 3.4 ± 0.26 Lipofectin + oligo 59 (n =121) 5.6±1.6 37±7.84 1 +1µMoligo 1(n =107) 1.6±0.44 13±1.1 2 +1µMoligo 5(n = 57) 2.6 ± 0.32 15 ± 14.7 3 +1µMoligo 4(n = 76) 2.0 ± 0.92 11 ± 1.83 4 +1µM oligo 13 (n = 50) 6.9 ± 1.2 39 ± 14.7 5 +1µM oligo 27 (n = 135) 6.5 ± 2.6 49 ± 11.6

Nuclear, percentage of cells with nuclear fluorescence. Units = average gray value/µm2 × 1000. These observations allow a relative comparison of nuclear or total cell fluorescence within the experiment. Results from two to four experiments/compound. Standard errors are shown for experiments where n >3.

Figure 2. Toxicity profile of amphiphile 5 alone or amphiphile 5:oligo complex. Data are expressed as percent cell number compared to untreated 3T3cellsbasedontheMTTassay(n = 3 experiments per concentration).

Association of oligonucleotide with cells The umbrella compounds’ ability to enhance oligonucleotide uptake by cells was estimated using flow cytometry, as shown in Figure 4 (24). Relative fluorescence (RFU) from cells treated with amphiphile or amphiphile/lipid complexed with FITC-derivatized phosphorothioate oligonucleotide was measured. Amphiphiles 4 and 5 stimulated oligonucleotide association with cells by as much as 100- to 200-fold relative to oligonucleotide alone. In contrast to gene delivery, formulation Figure 3. Reporter gene expression. Transfection activity of the amphiphiles into either fibroblastic (A) or endothelial (B) cells is shown. Relative luciferase with DOPE provided no significant advantage in promoting gene expression levels was quantitated by the number of luciferase units oligonucleotide–cell association. (RLU)/µg of cellular protein after incubation with DNA complexed with amphiphiles 1–5 either with (1L–5L) or without (1–5) 1:1 wt% DOPE. Cells Intracellular distribution treated with Lipofectin® or Lipofectamine®:DNA were used as positive controls. All values of means and standard errors reflect three to five independent Figure 5 depicts the intracellular distribution of oligonucleo- measurements. tide in live cells visualized by confocal fluorescence micro- scopy. Red color is derived from a cyanine fluor directly conjugated to the oligonucleotide. The cell is highlighted in green via immunostaining with an FITC anti-MDR antibody. amphiphile 5 complexes (Fig. 5C) are seen to have a high Free oligonucleotide (Fig. 5A) shows only a slight degree of degree of cell-associated material and a significant fraction of cell association and no evidence for nuclear accumulation. The the cells show nuclear fluorescence. Quantitation of total cell positive control, a Lipofectin®/oligo complex (Fig. 5B) demon- and nuclear fluorescence from fluorescent oligonucleotides strates a high level of cell-associated fluor as well as many cells complexed with the amphiphile series is summarized in showing nuclear localization. Cells treated with oligonucleotide/ Table 2. Amphiphile 5 produced a consistently higher fraction Nucleic Acids Research, 1999, Vol. 27, No. 16 3339

Figure 4. Oligonucleotide uptake. Total cell association of oligonucleotide was quantitated by flow cytometry. MDR3T3 cells treated with amphiphile (1–5)or amphiphile/DOPE (1L–5L) complexed with FITC-phosphorothioate oligo- nucleotide were analyzed for total cell-associated fluorescence. The ordinate expresses the mean values of fluorescence in arbitrary units relative to free oligo- nucleotide (n ≥ 4 measurements per compound or formulation). Light scatter parameters were used to gate viable cells and exclude dead or damaged cells. of cells with oligonucleotide in the nucleus (27%) relative to Figure 5. Confocal microscope images of live cells treated with fluorescent ® phosphorothioate oligonucleotide. MDR3T3 cells were cultured overnight in the other amphiphiles, yet lower than the case for Lipofectin thepresenceof1µM phosphorothioate oligonucleotide derivatized with a (59%). cyanine fluorophore. The fluorescent oligonucleotide was presented either in free form (A), complexed with Lipofectin® (B) or complexed with amphiphile Oligonucleotide activity assays 5 (C). Cells were rinsed and the cell surface was immunostained with anti-P- glycoprotein antibody and an FITC-conjugated secondary antibody (Materials The ability of the amphiphiles to deliver oligonucleotides in and Methods). The cells were then allowed to recover for 1–2 h in complete pharmacologically active form was examined. A powerful test medium lacking dye indicator, placed on the microscope stage and imaged for the biological activity of a 2'-O-methyl-oligonucleotide within several minutes. The green color shows the cell membrane while the red involves its ability to correct improper splicing of an intron color indicates the cellular location of the oligonucleotide. Yellow color indicates β areas where the cyanine and FITC fluorophores are in close proximity and derived from the -globin gene (25). Using a recently devel- probably represents binding of the oligonucleotide at the cell surface or in oped reporter assay, this effect is manifested as an increase in endosomes. Similar results were observed with HeLa cells, but 3T3 cells Luciferase activity (22). Oligonucleotide complexed with spread extensively and are better subjects for obtaining fluoresence images. Lipofectin® or Lipofectamine® was quite effective, with a substantial increase in Luciferase activity. Oligonucleotide complexed with amphiphile 5 was also effective, attaining shield between the highly polar nucleic acid and the non-polar approximately two-thirds the effect seen with Lipofectamine® interior of the lipid bilayer membrane. Initially, it was thought (Fig. 6A). The Luciferase activity showed progressive dose– that the umbrella compounds might form bi-molecular com- response relationships both with respect to oligonucleotide plexes with oligonucleotides. However, current investigations concentration (Fig. 6B) and with respect to the concentration have indicated that the first generation umbrella amphiphiles of the amphiphile (Fig. 6C). The response at the highest level form large complexes with oligonucleotides that can be of amphiphile 5 was attenuated, probably due to cytotoxicity. sedimented by high speed centrifugation; the precise physical The biologically active form of the OD-705/amphiphile 5 characterisation of these complexes is now being pursued in complex was shown to be particulate in nature since it could be other studies. sedimented by high speed centrifugation (data not shown). The first generation umbrella compounds reported here have In contrast to most commercial cytofectins, amphiphile 5 displayed significant activity both for transfection of high was also effective in the presence of substantial amounts of molecular weight DNA and for delivery of oligonucleotides. In serum protein, as seen in Figure 7. Thus, at 10% serum the the former case, complexation with the umbrella surfactants effects of Lipofectamine® were diminished almost to baseline, alone provided only modest increases in DNA transfection while a reduced but significant effect was still seen with efficiency compared to a ‘naked’ plasmid. The umbrella amphiphile 5. The ability of the umbrella molecule to deliver amphiphiles could be mixed with the non-bilayer forming oligonucleotide and thus activate Luciferase expression was phospholipid DOPE and in this form reasonably high transfec- detectable even in 50% serum. tion efficiencies were attained; however, this approach offers little advantage over cationic lipid-based ‘cytofectins’ that are already in use. Presumably the umbrella compounds, acting DISCUSSION alone, cannot form a cationic lipid–DNA complex that is Umbrella amphiphiles were designed to complex with the efficient in destabilizing membranes sufficiently to permit the negatively charged backbone of nucleic acids and to provide a delivery of functional high molecular weight DNA complexes. 3340 Nucleic Acids Research, 1999, Vol. 27, No. 16

Figure 6. Oligonucleotide activity (correction of splicing). HeLa/Luc 705 cells were treated with various concentrations of a 2'-O-Me phosphorothioate oligonucleotide complementary to the β-globin splice site junction (ON-705). Various complexes were used as delivery formulations. Luciferase activity was measured as described in Materials and Methods. (A and B) Activation of the Luciferase reporter using umbrella 5 to deliver antisense oligonucleotide. (A) Luciferase expression at various concentrations of umbrella amphiphile 5 was compared with Lipofectamine® (8 µg/ml) and Lipofectin® (10 µg/ml); (B) Luciferase expression at constant concentration of the umbrella amphiphile (3.0 µM) with varying concentrations of oligonucleotide; (C) Luciferase expression at constant molar ratio of amphiphile to oligonucleotide. The activity of Luciferase was normalized on total cellular protein and is presented in relative luminescence units (RLU)/µg of protein. Vertical bars indicate means and standard errors (n = 3).

tides with the tetra-walled umbrella compound 5 resulted in complexes that could convey oligonucleotides into the nucleus of viable cells. Differences in the ability to deliver small oligo- nucleotides versus high molecular weight plasmids may not be surprising, since the physical characteristics of the two types of complexes are likely to be different. The evidence for nuclear delivery includes (i) the visualization of fluorescent oligo- nucleotides in the nucleus by confocal microscopy; and (ii) an increase in the pharmacological effectiveness of an antisense oligonucleotide that modifies splicing of a β-globin intron inserted into a Luciferase reporter gene. These observations, particularly the latter, provide convincing proof that umbrella compounds can deliver oligonucleotides to key target compart- Figure 7. Effect of serum on the effectiveness of oligonucleotide/amphiphile 5 ments in living cells. The splice site correction assay has some complexes. Complexes of ON-75 and amphiphile 5 were formed and added to HeLa cells stably transfected with plasmid pLuc/705 in 24-well trays as important advantages for evaluating the delivery and effective- described in Materials and Methods. Cells treated with Lipofectamine® (8 µg/ml)/ ness of oligonucleotides. First, it is very sensitive since it oligonucleotide complex were used for comparison. Additional control samples provides a positive readout rather than relying on a difference underwent the same oligonucleotide treatment, but in the absence of any delivery measurement (the inhibition of message levels) as is the case agent. The abcissa indicates the percentage of fetal calf serum present during for standard antisense assays. Second, splicing can only occur the incubation with oligonucleotides. Vertical bars indicate means and standard errors (n =3). within the nucleus of a living cell; thus the effects measured on splice site selection are clearly associated with the oligonucleotide functioning in viable cells and cannot be ascribed to toxicity. In most experiments the delivery efficiency of the oligo- In contrast, the umbrella amphiphiles seem to work quite nucleotide/umbrella amphiphile complexes in serum-free medium well when applied to the delivery of relatively low molecular was somewhat less than that provided by commercial cytofectins weight oligonucleotides. Thus complexation of oligonucleo- (cationic liposomes) such as Lipofectin®.However,thematerials Nucleic Acids Research, 1999, Vol. 27, No. 16 3341 studied here were first generation compounds; it seems likely ACKNOWLEDGEMENTS that modification of the size/charge of the cationic tail or mod- This work was supported by NIH grants CA47044 to R.L.J. ification of the cholic acid moieties of the umbrellas might and GM51814 to S.L.R. and by an NIH fellowship CA72208 confer more desirable delivery properties on this class of to R.K.D. molecules. In addition, our understanding of the most appropriate way to formulate oligonucleotide/umbrella complexes is very limited; we have not yet systematically optimized these REFERENCES complexes for size, charge or delivery efficiency. Despite this, 1. Verma,I.M. and Somia,N. (1997) Nature, 389, 239–242. even the first generation umbrella amphiphiles seem to offer an 2. Crooke,S.T. (1997) Adv. Pharmacol., 40, 1–49. important advantage in that they retain the ability to deliver oligo- 3. Roush,W. (1997) Science, 276, 1192–1193. nucleotides in the presence of substantial amounts of serum 4. Ho,P.T. and Parkinson,D.R. (1997) Semin. Oncol., 24, 187–202. 5. Stull,R.A. and Szoka,F.C.J. (1995) Pharm. Res., 12, 465–483. proteins, whereas many commerical cytofectins show signifi- 6. Alahari,S.K., Dean,N.M., Fisher,M.H., DeLong,R., Manoharan,M., cantly reduced efficacy in the presence of low concentrations Tivel,K.L. and Juliano,R.L. (1996) Mol. Pharmacol., 50, 808–819. of serum. There have been previous reports of effective oligo- 7. Bennett,C.F., Chiang,M.Y., Chan,H., Shoemaker,J.E. and Mirabelli,C.K. nucleotide delivery in the presence of serum. For example, (1992) Mol. Pharmacol., 41, 1023–1033. 8. Skorski,T., Perrotti,D., Nieborowska-Skorska,M., Grayaznov,S. and Lewis et al. (26) used a novel cationic lipid formulated with the Calabretta,B. (1997) Proc. Natl Acad. Sci. USA, 94, 3966–3971. fusogenic lipid DOPE to attain significant antisense effects in 9. Dean,N.M. and McKay,R. (1994) Proc. Natl Acad. Sci. USA, 91, 10% serum. In the present case, however, no accesssory lipid 11762–11766. was used, but nonetheless significant antisense effects were 10. Zabner,J. (1997) Adv. Drug Deliv. Rev., 27, 17–28. 11. Felgner,J.H., Kumar,R., Sridhar,C.N., Wheeler,C.J., Tsai,Y.J., Border,R., obtained at up to 50% serum concentration. Ramsey,P., Martin,M. and Felgner,P.L. (1994) J. Biol. Chem., 269, The mechanism of action of amphiphile 5 and other umbrella 2550–2561. amphiphiles is unclear at this point. Unlike most of the cationic 12. Wagner,E., Zatloukal,K., Cotten,M., Kirlappos,H., Mechtler,K., surfactants used in commercial cytofectins, the umbrella com- Curiel,D.T. and Birnstiel,M.L. (1992) Proc. Natl Acad. Sci. USA, 89, 6099–6103. pounds do not require a neutral lipid such as DOPE to allow 13. Mahato,R.I., Rolland,A. and Tomlinson,E. (1997) Pharm. Res., 14, oligonucleotide delivery to cells. Centrifugation experiments 853–859. indicate that the biologically active form of the oligonucleotide/ 14. Gao,X. and Huang,L. (1996) , 35, 1027–1036. amphiphile 5 complex is a sedimentable particle and not a low 15. Zabner,J., Fasbender,A.J., Moninger,T., Poellinger,K.A. and Welsh,M.J. (1995) J. Biol. Chem., 270, 18997–19007. molecular weight moiety. These particles bind to cell membranes, 16. Xu,Y. and Szoka,F.C.J. (1996) Biochemistry, 35, 5616–5623. as shown by flow cytometry, and may be taken into endo- 17. Nabel,G.J., Nabel,E.G., Yang,Z.Y., Fox,B.A., Plautz,G.E., Gao,X., somes, as suggested by the appearance of material in vesicular Huang,L., Shu,S., Gordon,D. and Chang,A.E. (1993) Proc. Natl Acad. structures seen by confocal microscopy. Whether the actual Sci. USA, 90, 11307–11311. 18. Stribling,R., Brunette,E., Liggitt,D., Gaensler,K. and Debs,R. (1992) delivery of oligonucleotide to the cytoplasm and nucleus Proc. Natl Acad. Sci. USA, 89, 11277–11281. occurs at the plasma membrane or in endosomes and whether 19. Bennett,C.F., Kornbrust,D., Henry,S., Stecker,K., Howard,R., Cooper,S., it involves particles or soluble complexes of oligonucleotide Dutson,S., Hall,W. and Jacoby,H.I. (1997) J. Pharmacol. Exp. Ther., 280, and umbrella amphiphiles remains to be determined. 988–1000. 20. Akthar,S. and Agrawal,S. (1997) Trends Pharmacol. Sci., 18, 12–18. An interesting possibility for future investigation will be to 21. Janout,V., Lanier,M., Deng,G. and Regen,S.L. (1997) Bioconjugate link oligonucleotides directly to the cholic acid umbrella Chem., 8, 891–895. framework, rather than using a cationic ‘handle’ to promote 22. Kang,S.H., Cho,M.J. and Kole,R. (1998) Biochemistry, 37, 6235–6239. ionic interactions. If a bio-reversible covalent linkage were 23. Carmichael,J., DeGraff,W.G., Gazdar,A.F., Minna,J.D. and Mitchell,J.B. (1987) Cancer Res., 47, 936–942. employed, this might allow development of non-cationic delivery 24. Alahari,S.K., DeLong,R., Fisher,M.H., Dean,N.M., Viliet,P. and agents that could carry the oligonucleotide through the cell Juliano,R.L. (1997) J. Pharmacol. Exp. Ther., 286, 419–428. membrane as a mono-molecular species and then release it 25. Sierakowska,H., Sambade,M.J., Agrawal,S. and Kole,R. (1996) Proc. within the cell in biologically active form. This might permit Natl Acad. Sci. USA, 93, 12840–12844. 26. Lewis,J.G., Lin,K.Y., Kothavale,A., Flanagan,W.M., Matteucci,M.D., efficient in vivo delivery without some of the problems associated DePrince,R.B., Mook,R.A.,Jr, Hendren,R.W. and Wagner,R.W. (1996) with use of cationic lipids. Proc. Natl Acad. Sci. USA, 93, 3176–3181.