Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Intercellular adhesion molecule 1 (ICAM-1) is constitutively expressed at low levels on many cells, including vascular endothelium, keratinocytes, and some subsets of leukocytes; however, protein expression is rapidly up-regulated in response to proinflammatory mediators (Dustin et al., 1986; van de Stolpe and van der Saag, 1996). One of several known roles for ICAM-1 is to increase the association of leukocytes with the luminal surface of endothelial cells in the vicinity of an inflammatory stimulus. Bound leukocytes are activated and participate in the induction of interendothelial gaps that increase vascular leak and enable cellular transmigration into the extravascular tissue (Marcus et al., 1997). Therefore, designing drugs to specifically block leukocyte adherence is a logical approach to the management of inflammatory diseases. In this regard, using antibodies to block the binding between ICAM-1 and its ligands or using antisense oligonucleotides to inhibit the overexpression of ICAM-1 have both been shown to ameliorate inflammatory responses in vivo (Stepkowski et al., 1994; Rothlein and Jaeger, 1995; Bennett et al., 1997).

Antisense oligonucleotides inhibit protein expression by hybridizing to a specific mRNA transcript and disrupting translation through a variety of mechanisms. The antisense molecule used in this study (ISIS 3082) is a 20-base phosphorothioate (PS) oligodeoxynucleotide (ODN) that binds with high affinity to the 3'-untranslated region of murine ICAM-1 (Stepkowski et al., 1994). The DNA/RNA heterodimer formed between the PS ODN and ICAM-1 mRNA is a substrate for RNase H, a ubiquitous intracellular endonuclease, which cleaves the mRNA transcript (Bennett et al., 1994). ISIS 3082 has been shown to reduce ICAM-1 protein and mRNA levels both in vitro and in vivo in a sequence-specific manner (Stepkowski et al., 1994; Kumasaka et al., 1996; Bennett et al., 1997). The human analog (ISIS 2302) is currently in phase III clinical trials for the treatment of inflammatory bowel disease. Earlier phase 1 and 2 trials determined that the drug was well tolerated in patients with active Crohn's disease, and induced remission in 47% of patients treated compared with 20% for a placebo control group. Moreover, the need for steroid therapy was significantly reduced for the group receiving ISIS 2302 (Yacyshyn et al., 1998).

Despite this success, it is generally recognized that the clinical utility of antisense drugs may be improved by better delivery (Gewirtz et al., 1996; Juliano et al., 1999). For example, the sites of action for PS ODN are the cytoplasm and nucleus, but in vitro these molecules can penetrate very few cell types unaided. Intracellular delivery of negatively charged oligonucleotides into cells in culture is commonly achieved by using cationic lipid vesicles (Bennett et al., 1992). Unfortunately, these complexes are unsuitable for systemic delivery of nucleic acids because they are unstable and tend to form large aggregates that are cleared rapidly from the circulation, get trapped in first pass organs, and exhibit hematological and liver toxicities (Litzinger, 1997; Zelphati and Szoka, 1997). There is evidence, both in the literature and from clinical examples, that conventional (mostly nontargeted, nonfusogenic) liposomes can increase the therapeutic index of many different types of drugs, including antisense ODN (Wielbo et al., 1996). This is achieved through a combination of effects, including reduced drug toxicity, enhanced protection of the drug in the biological milieu, altered pharmacokinetics, and disease-site targeting (Chonn and Cullis, 1995).

We have recently completed a characterization of liposome accumulation into sites of inflammation with a murine model of contact hypersensitivity (CHS) (Klimuk et al., 1999). Mice sensitized to 2,4-dinitrofluorobenzene (DNFB) undergo a reproducible and measurable inflammatory response when a solution of the chemical is applied topically to the ear (Young and De Young, 1989). Hapten-specific dermal T lymphocytes are triggered to release proinflammatory cytokines that up-regulate ICAM-1 expression on local endothelial cells, keratinocytes, and antigen-presenting cells in the dermis. Within hours, the ear swells and leukocytes begin to extravasate into the extravascular tissue. Ear thickness and cellular infiltration peak at 24 h and gradually decline to baseline levels over several days (Goebeler et al., 1990; Klimuk et al., 1999). For the first 24 h after epicutaneous contact, the local vasculature is hyperpermeable allowing fluid, cells, and liposomes to accumulate. Herein, we report our initial findings regarding the effects of liposome encapsulation on the biological activity of an ICAM-1-specific PS ODN in this CHS model.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. Egg phosphatidylcholine (EPC) was purchased from Northern Lipids (Vancouver, Canada). Cholesterol (CH), DNFB, and BSA were purchased from Sigma (St. Louis, MO). BioGel A-15 m (200-400 mesh) was purchased from Bio-Rad (Mississauga, Canada). [¹⁴C]Cholesterylhexadecylether ([¹⁴C]CHE), [³H]methylthymidine, and the tissue solubilizer Solvable were obtained from DuPont-NEN (Boston, MA). Dulbecco's PBS (pH 7.4) lacking CaCl₂ and MgCl₂ was obtained from Stemcell Technologies (Vancouver, Canada).

Mice. Female 7-week-old (25 g) ICR mice were obtained from Harlan Sprague-Dawley (Indianapolis, IN) and quarantined for 1 week before use. Each experimental group consisted of at least four mice for pharmacokinetic and tissue distribution studies, and six to eight mice for ODN activity studies.

Antisense ODNs. PS ODNs were synthesized and purified as described in detail elsewhere (Stepkowski et al., 1994; Bennett et al., 1994). The name, base sequence, and target specificity of the three ODNs used in this study are as follows: ISIS 3082 (5'-TGCATCCCCCAGGCCACCAT-3'), murine ICAM-1; ISIS 2302 (5'-GCCCAA- GCTGGCATCCGTCA-3'), human ICAM-1; and ISIS 8997 (5'-TCGCATCGACCCGCCCACTA-3'), scrambled version of ISIS 3082.

Preparation of Empty Liposomes. Liposomes were prepared as described in detail elsewhere (Hope et al., 1993). Briefly, 178 mg of EPC and 72 mg of CH were combined in chloroform in a 16 × 100 mm glass culture tube and solvent was removed under a stream of nitrogen gas. The lipid film was subsequently lyophilized for 3 to 4 h to remove residual chloroform and then hydrated in 1.0 ml of PBS with vigorous mixing. The resulting multilamellar vesicles (MLVs) were transferred to cryovials and subjected to five cycles of freezing in liquid nitrogen and thawing at 40°C. Freeze-thawed MLVs were transferred to a 10-ml capacity thermobarrel extruder (Lipex Biomembranes, Vancouver, Canada) and extruded 10 times through two stacked 100-nm polycarbonate filters (Nuclepore, Pleasanton, CA). Quasi-elastic light scattering was used to assess the size distribution of the extruded liposomes with an NICOMP model 370 submicrometer particle sizer (Pacific Scientific, Santa Barbara, CA). Liposomes were typically 110 ± 30 nm in diameter. In some instances, a trace amount of [¹⁴C]CHE was added to liposome preparations as a nonexchangeable, nonmetabolizable lipid marker (Stein et al., 1980).

Encapsulation of ODNs. Encapsulation of ODN was essentially as described above for the preparation of empty liposomes, with the following modifications. The lyophilized lipid film was hydrated with 1.0 ml of PBS containing 200 mg of ODN. Due to the highly viscous nature of the solution, the sample was extruded 12 times through a single 100-nm filter. Nonencapsulated ODN was usually removed from the extruded sample by size exclusion chromatography (BioGel A-15 m), but anion exchange chromatography (with DEAE-Sepharose CL-6B) was equally effective. Fractions containing lipid were pooled and concentrated by placing the vesicle suspension into presoaked dialysis tubing (12-14,000-mol. wt. cutoff) and covering with Aquacide (Calbiochem, San Diego, CA) until the desired volume was reached. Trapping efficiencies for this procedure were typically 10 to 15%.

Complement Assay. The interactions between complement proteins and free or liposome-encapsulated ODN were assessed with a two-step hemolytic assay (CH50) described by Devine et al. (1994). Serial dilutions of sample (100 µl) were incubated with diluted (5×) normal human serum (100 µl) for 30 min at 37°C. Aliquots (100-µl) of the serum incubation were added to hemolysin-sensitized sheep red blood cells (SRBCs) and further incubated for 30 min at 37°C. Hemoglobin released by lysed red blood cells was assessed in the supernatant by measuring the absorbance at 410 nm in a 96-well microplate reader. Reduced hemolysis in the presence of PS ODN was taken to indicate an interaction and disruption of the normal hemolytic process. This analysis did not allow a distinction between direct inhibition of enzymatic reactions and consumption of complement proteins.

Ear Inflammation Model. ICR mice were sensitized to DNFB by applying 25 µl of 0.5% DNFB in an acetone/olive oil vehicle (4:1 v/v) to the shaved abdominal wall for two consecutive days (Young and De Young, 1989). Four days after the second application, ear inflammation was initiated by challenging the mice on the dorsal surface of the left ear with 10 µl of 0.2% DNFB in acetone/olive oil (4:1), whereas the right ear remained untreated. The inflammatory response generated with this model was previously shown to be maximal 24 h after DNFB ear challenge (Klimuk et al., 1999). A topical corticosteroid (Ultravate; active ingredient halobetasol propionate, 0.05%, w/w) was kindly donated by Dr. N. Kitson (Division of Dermatology, University of British Columbia) and was used as a positive control throughout these studies. This corticosteroid, in the form of a gel, was spread evenly over the entire dorsal surface of the left ear 15 min after painting the ear with DNFB.

Measurement of Ear Thickness. Ear thickness was measured on anesthetized mice immediately before ear challenge with DNFB and at various time intervals (usually 24 h) after DNFB challenge, with an engineer's micrometer (Mitutoyo, Tokyo, Japan). Increases in ear thickness were calculated by subtracting prechallenge from postchallenge measurements.

Evaluation of Cellular Infiltration. Mice received a 500-µl i.p. injection of [³H]methylthymidine in sterile saline (1 µCi/g b.wt.) 24 h before ear challenge with DNFB to prelabel bone marrow cells and circulating leukocytes (Young et al., 1984). Ear thickness measurements were recorded and the ears were subsequently removed from the base of the pinna, digested whole in a tissue solubilizer for 12 to 48 h at 60°C, and analyzed for radiolabeled leukocytes by liquid scintillation counting. A relative estimate of cellular infiltration was determined by expressing the ratio of radioactivity observed in the inflamed versus noninflamed ears.

Plasma Elimination and Tissue Distribution. Fifteen minutes after applying DNFB to the left ear, free and encapsulated ODNs were administered i.v. via the lateral tail vein (200 µl) at an ODN dose of 50 mg/kg. At various times, mice were euthanized and plasma aliquots (100 µl) removed for liquid scintillation analysis. Organs also were collected (liver, spleen, and kidneys) and tissue homogenates were made with a Polytron. Aliquots of samples (200 µl) were digested and decolorized (Longman et al., 1995) before liquid scintillation counting to determine the presence of radiolabeled ODN and lipid. ODN distribution in the inflamed and noninflamed ears was determined by digesting the entire ear in 500 µl of tissue solubilizer, followed by liquid scintillation analysis as indicated above.

ODN Integrity in Plasma. ODN integrity during plasma circulation was assessed by polyacrylamide gel electrophoresis analysis at various times after administration of a 50-mg/kg i.v. dose of free or EPC/CH-encapsulated ISIS 3082. Plasma samples were withdrawn by cardiac puncture and 30-µl aliquots were incubated with 30 µl of buffer containing proteinase K (100 mM NaCl, 10 mM Tris, pH 8, 25 mM EDTA, 0.5% SDS, and 1 mg/ml proteinase K) for 1.5 h at 65°C. The incubation was stopped by a 40-µl addition of a formamide stop buffer [93.6 (v/v) formamide, 20 mM EDTA, 0.05% bromophenol blue]. Aliquots (25 µl) of the incubation mixture were analyzed on a 20% denaturing acrylamide gel containing 7 M urea. Visualization of ODN was facilitated by staining with Sybrgreen I nucleic acid gel stain (Molecular Probes, Eugene, OR).

Efficacy of ODN Formulations. Fifteen minutes after ear inflammation was initiated, ODN was administered i.v. at doses ranging from 5 to 50 mg/kg. Control mice received 200-µl injections of PBS or empty lipid vesicles (300-mg/kg lipid). Ear measurements were recorded 24 h later to assess the extent of edema before ears were collected to determine the level of ³H-labeled leukocyte infiltration.

ICAM-1 Immunohistochemistry. Twenty-four hours after initiating ear inflammation, specimens from the central ear region were isolated, mounted between two slices of calf liver, and frozen in isopentane cooled by liquid nitrogen. Transverse serial cryosections (5 µm in thickness) were collected on polylysine-coated slides and examined within 24 h of sectioning. Slides were preincubated with PBS/BSA (PBS containing 1% BSA) for 10 min to reduce nonspecific binding of the antibody. ICAM-1 expression was detected by incubating the slides for 1 h at room temperature in a humidified chamber with a rat monoclonal anti-mouse ICAM-1 affinity-purified antibody (Seikagaku Corp., Tokyo, Japan) diluted 1:50 with PBS/BSA. After washing with PBS/BSA (3 × 15 min), a biotin-conjugated goat anti-rat secondary antibody (Caltag, San Francisco, CA) diluted 1:100 was applied to the slides for 1 h. For visualization, sections were incubated for 45 min with fluorescein isothiocyanate-conjugated streptavidin (Caltag) diluted 1:200 in PBS/BSA. Slides were rinsed well with buffer and examined immediately with a Zeiss Axiophot photomicroscope equipped with epifluorescence. Fluorescein iosothiocyanate fluorescence was detected with a band pass 450 to 490 exciter filter, and randomly chosen sections were photographed on Ektachrome P1600 (35 mm) color slide film.

Statistical Analysis. All data values are presented as the mean ± S.D. A standard one-way ANOVA was used to determine the significance of the difference of the means. When significant differences among the means were indicated, a Dunnett's test was used to determine which treatment groups differed significantly from the control. P < .05 was considered significant.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Encapsulation of Antisense ODNs. The encapsulation of PS ODN in EPC/CH (55:45 mol/mol) vesicles was achieved with standard passive entrapment techniques in which a dry lipid film was hydrated with an aqueous solution of ODN. The resulting MLVs were subsequently sized by extrusion to generate a population of vesicles with a mean diameter of 110 ± 30 nm as indicated by quasi-elastic light scattering analysis. To maximize the ratio of encapsulated ODN to lipid as well as the overall encapsulation efficiency, high concentrations of ODN and lipid were used (see Experimental Procedures). Despite using these extreme concentrations, the encapsulation efficiency achieved rarely exceeded 10 to 15%, with ODN/lipid ratios on the order of 0.1 (w/w). This was expected given the absence of a cationic lipid in our formulation, which would provide an electrostatic attraction for the anionic ODN and therefore enhance the encapsulation efficiency. Preparations were physically stable for at least 5 days at both 4 and 37°C, with no change in vesicle diameter or loss of [³H]ODN (data not shown).

Encapsulation Prevents Serum Protein Interactions with PS ODN. Although PS ODNs are more resistant to endogenous nucleases than natural phosphodiesters, they are not completely stable and are slowly degraded in blood and tissues (DeLong et al., 1997). Yeast S1 nuclease rapidly cleaves PS ODN free in solution but encapsulation in EPC/CH vesicles provides complete protection (data not shown), a property also observed in vivo as we describe in later sections. A limitation in the clinical application of PS antisense drugs is their propensity to activate complement on i.v. administration (Henry et al., 1997a). To determine how effectively vesicles prevented this interaction, we used an antibody-sensitized SRBC assay to measure the extent to which free and encapsulated PS ODN interacted with complement factors in human plasma (Devine et al., 1994). The two-step process involves first incubating PS ODN in freshly prepared human serum, followed by mixing with sheep erythrocytes coated with anti-sheep antibody. The complement cascade in naive serum then lyses erythrocytes by formation of the membrane attack complex. Reduction of hemolysis in this assay can be attributed to consumption of complement factors or to direct inhibition of enzymatic processes. As shown in Fig. 1, human serum samples preincubated for 30 min at 37°C with free PS ODN at concentrations >10 µg/ml are not capable of hemolysis. Therefore, complement factors in these samples have been activated and used, or are bound and unable to participate in the cascade that results in the formation of attack complexes (Devine et al., 1994). However, preincubating human serum with encapsulated PS ODN at concentrations as high as 500 µg/ml has no measurable effect on the ability of human serum to hemolyze cells, indicating that the PS ODN was prevented from interacting with complement proteins.

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Liposome encapsulation of a PS ODN inhibits interactions with complement components. Serial dilutions of free and encapsulated ISIS 2302 were incubated with 5× diluted human serum before 100-µl aliquots were removed and incubated with hemolysin-sensitized SRBCs as described in Experimental Procedures. Aliquots (100-µl) were then removed to determine hemoglobin release from lysed red blood cells by measuring absorbance at 410 nm. Free ODN () used complement factors to a much greater extent than the liposomal formulation ().

Plasma Clearance and Stability of Encapsulated ODN. Large unilamellar vesicles (LUVs) containing [³H]ISIS 2302 PS ODN and ¹⁴C-labeled lipid (see Experimental Procedures) were administered as a 200-µl bolus injection via the tail vein at an ODN dose of 50 mg/kg and a lipid dose of ~300 to 400 mg/kg. Free PS ODN was cleared rapidly from the circulation (Fig. 2A) with biphasic kinetics suggesting distribution followed by elimination phases. The half-life of distribution is on the order of 2 to 5 min, which is consistent with published murine pharmacokinetic data for PS ODN administered at a comparable dose (DeLong et al., 1997). In contrast, the same dose of encapsulated ODN exhibits an almost monophasic plasma clearance curve with a half-life of 6 to 8 h. This kinetic is identical with that observed for the LUV alone (data not shown) and is in the same range observed previously for neutral lipid vesicles ~100 nm in diameter and administered at a dose of 300 mg/kg in a mouse (Rodrigueza et al., 1993).

View larger version (17K): [in this window] [in a new window]   Fig. 2.   Clearance of free and encapsulated PS ODN from plasma and the retention of PS ODN by lipid vesicles while in circulation. Mice received a bolus i.v. injection of [3H]ISIS 2302 in the free form or encapsulated in EPC/CH LUV containing the lipid marker [14C]CHE. At the times indicated, plasma was obtained and analyzed by liquid scintillation to determine the circulating levels of ODN and lipid. ODN in the free form () exhibited rapid clearance (t1/2 = ~2-5 m). The circulation half-life of formulated ODN () was greatly increased (t1/2 = ~8 h) with no apparent loss of ODN from LUV recovered from the plasma (B). Data are expressed as means ± S.D. (n = 4).

View larger version (53K): [in this window] [in a new window]   Fig. 3.   Lipid encapsulation protects PS ODN from degradation by plasma nucleases. ODN (ISIS 3082) was administered i.v. either encapsulated in EPC/CH LUV (A) or in the free form (B). At the times indicated, plasma was removed and incubated with buffer containing proteinase K as described in Experimental Procedures. ODNs were separated on 20% polyacrylamide gel electrophoresis gels, followed by Sybrgreen I nucleic acid gel staining to visualize individual bands. Lane 1 represents 1 µg of native, full-length ODN (ISIS 3082). The remaining lanes represent intact ODN and metabolites that were present in the plasma of mice at the time points indicated. Each lane represents a different animal.

Disease Site Targeting and Biodistribution of Encapsulated ODN. We have recently characterized the extravasation and subsequent accumulation of lipid vesicles in the inflamed ears of mice hypersensitized to DNFB (Klimuk et al., 1999). Using the same model herein (described in Experimental Procedures), we show that EPC/CH vesicles can deliver a significantly larger dose of antisense to the site of inflammation over 24 h compared with an equivalent dose (50 mg/kg) of PS ODN administered in the free form (Fig. 4). Moreover, at 24 h the amount of [³H]ODN detected in the inflamed ear was 10- to 20-fold greater than that measured in the contralateral control ear for free or encapsulated ODN. This is an example of disease-site targeting by LUV, and is considered to be a key factor in the ability of lipid vesicles to enhance the therapeutic activity of encapsulated drugs (Chonn and Cullis, 1995). In this example, [³H]ISIS 2302 (specific for human ICAM-1) was encapsulated instead of murine-specific ICAM-1 ODN (ISIS 3082) because the human version was not expected to show antisense activity in this model.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Accumulation of PS ODN at the site of inflammation. Fifteen minutes after initiating ear inflammation, mice received a bolus i.v. injection of free (,) or EPC/CH encapsulated [3H]ISIS 2302 (, ) at an ODN dose of 50 mg/kg. At the times indicated, mice were euthanized and inflamed (, ) and untreated (, ) ears removed for analysis of ODN accumulation as described in Experimental Procedures. Data are expressed as means ± S.D. (n = 4). Lipid encapsulation of the ODN enabled greater quantities to accumulate at the inflammation site compared with drug administered in the free form.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   Tissue distribution of free and encapsulated ODN. Mice received an i.v. injection of liposome-encapsulated () or free () [3H]ISIS 2302 at ODN doses of 50 mg/kg. One hour (A) and 24 h (B) later, organs were removed and digested to determine ODN accumulation as described in Experimental Procedures. Encapsulation inhibited the accumulation of ODN in the kidneys and increased delivery to the liver and spleen.

Anti-Inflammatory Activity of ICAM-1-Specific ODNs. The anti-inflammatory activity of an ODN specific for murine ICAM-1 (ISIS 3082) has been demonstrated in several chronic inflammatory models (Stepkowski et al., 1994; Bennett et al., 1997) as well as in acute, endotoxin-induced inflammation (Kumasaka et al., 1996). The CHS model used in this study provides an acute inflammatory response initiated within minutes after topical application of DNFB to the dorsal ear surface. Ear swelling and cellular infiltration are maximal 24 h after initiating inflammation (Klimuk et al., 1999); however, swelling subsides by 48 to 72 h, whereas lymphocyte levels in the affected ear do not return to normal for ~2 weeks (Goebeler et al., 1990). Consequently, all the measurements reported herein were made at the 24-h time point.

View larger version (39K): [in this window] [in a new window]   Fig. 6.   Anti-inflammatory activity of ICAM-1 ODN. Three different PS ODN (ISIS 3082, 2302, and 8997) were each encapsulated in 100 nm EPC/CH LUV. Fifteen minutes after initiating ear inflammation, mice received a bolus injection of ODN at 50 mg/kg. Mice were treated with the following: 1) PBS, 2) empty LUV, 3) free ISIS 3082, 4) encapsulated ISIS 2302, 5) encapsulated ISIS 8997, and 6) encapsulated ISIS 3082. A control group received a topical application of the corticosteroid halobetasol propionate (7), applied to the ear 15 min after eliciting inflammation. Twenty-four hours later, ear thickness measurements were recorded (A) and ears removed for determination of cellular infiltration by analysis of [3H]thymidine-labeled leukocytes (B). *, data significantly different from that obtained from control group administered PBS only (P < .05).

View larger version (18K): [in this window] [in a new window]   Fig. 7.   Dose-response curves for the anti-inflammatory activity of encapsulated ICAM-1 PS ODN. The murine-specific PS ODN for ICAM-1 (ISIS 3082) was encapsulated in 100 nm EPC/CH LUV. The encapsulated sample was diluted with empty EPC/CH LUV to yield the ODN doses of 5, 10, 30, and 50 mg/kg. Liposomal ODN was administered as a single bolus i.v. injection 15 min after initiating ear inflammation. Twenty-four hours later, ear thickness measurements were recorded (A) and ears removed for determination of 3H-labeled leukocyte infiltration (B).

View larger version (52K): [in this window] [in a new window]   Fig. 8.   Immunohistochemical evaluation of ICAM-1 protein expression in ear tissue. Fifteen minutes after initiating ear inflammation, mice were injected i.v. with PBS (A), free ISIS 3082 at a 50-mg/kg dose (B), or encapsulated ISIS 3082 at a 50-mg/kg dose of ODN (C). D, constitutive ICAM-1 expression detected in normal ear tissue. Twenty-four hours after challenge, mice were euthanized and representative ears excised and stained with ICAM-1 antibody as described in Experimental Procedures. The results demonstrate a reduction in ICAM-1 expression in DNFB-challenged ears from mice treated with encapsulated ISIS 3082 only. Original magnification, 100×.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

In this study, we have demonstrated that the acute anti-inflammatory activity of antisense ODN targeted against murine ICAM-1 mRNA is dramatically enhanced after encapsulation in LUV. In general, liposomes enhance the therapeutic index of conventional drugs by reducing drug toxicity, increasing residency time in the plasma, and delivering more active drug to disease tissue by extravasation of the carriers through hyperpermeable vasculature associated with tumor and inflammation sites (Mayer et al., 1990; Jain et al., 1997). All of these positive attributes of drug delivery are shown to be present in the model described herein. For example, the clinical concerns most commonly associated with rapid systemic administration of PS ODN are complement activation and prolonged blood clotting times (Galbraith et al., 1994; Henry et al., 1997a,b). Disruption of normal hematological homeostasis in this way is related to the polyanionic nature and stability of the PS backbone and is considered a class toxicity of PS ODN molecules and not the result of a sequence specific antisense effect (Galbraith et al., 1994). Herein, we confirm that nonencapsulated PS ODN can interact with complement factors in human serum in vitro and interfere with the normal hemolytic process, whereas ODN encapsulated in EPC/CH vesicles do not (Fig. 1). Therefore, we would expect encapsulated PS ODN to be less toxic on bolus injection compared with the free drug. However, mice are relatively insensitive to factors that activate complement, and therefore it is not usually a limiting toxicity in rodent models. This is consistent with the fact that we do not observe any gross sign of distress in the mice after bolus injection of any of the three PS ODNs examined in these studies, despite the relatively high doses used.

Significant changes to the pharmacokinetics and biodistribution of PS ODN were brought about by lipid encapsulation; specifically, circulation half-life and ODN targeting to the inflammation site were considerably enhanced (Figs. 2-5). These may be the principle reasons for the enhanced anti-inflammatory activity of the murine-specific ODN observed in the CHS model. It is important to note that there are many examples in the literature demonstrating biological activity of antisense drugs in animal models (Akhtar and Agrawal, 1997); however, care must be taken when interpreting the data and defining a mechanism of action (Bennett, 1998). It has been proposed that, in vivo, PS ODNs access cells by first binding to cell surface receptors or soluble receptor ligands present in plasma (Geselowitz and Neckers, 1995; Bijsterbosch et al., 1997). Presumably, their resistance to nucleases enables PS ODNs to survive subsequent endocytosis and degradation in the lysosomal compartment or they may be carried into cells via nonendosomal routes. Encapsulation may enhance these processes in several ways, as discussed below.

Lipid encapsulation prevents the PS ODN from interacting with potential protein-binding sites while in circulation. Although this includes productive binding to ligands and receptors that could lead to intracellular delivery, it also prevents binding to unproductive sites that are probably responsible for the rapid decrease of free-drug concentration in circulation (Fig. 2). ODN encapsulation also results in higher concentrations of unbound PS ODN being delivered to specific organs, cells, and diseased tissue. At these sites, the phospholipid delivery vehicle degrades in the extracellular space or in lysosomal pathways after endocytosis. PS ODNs released extracellularly would be available to bind with cell surface receptors or soluble ligands and therefore enter cells in the same manner as free drug. Alternatively, vesicles taken into cells by endocytosis would be processed in a similar manner to drug bound to receptors or proteins. The net result being more active drug reaching intracellular target sites.

The observation that modifying the biodistribution and pharmacokinetics of ICAM-1 antisense enhances the therapeutic activity of the drug in the CHS model is consistent with the data of Kumasaka et al. (1996). They demonstrated that free ISIS 3082 inhibits endotoxin-induced increases in ICAM-1 mRNA and neutrophil emigration into lung tissue 24 h after instillation of Escherichia coli endotoxin into the distal airways of mice. However, to observe these effects at 24 h, the drug had to be administered i.v. at a dose of 100 mg/kg 2 h pre- and 4 h postinstillation, indicating that activity may be dependent on maintaining high plasma and tissue concentrations of ODNs.

In the CHS model, there are at least two cellular targets where increased ODN delivery is expected due to encapsulation of the drug in LUV, and where inhibition of ICAM-1 expression could produce an anti-inflammatory effect. Endothelial cells in the inflamed ear become active in response to proinflammatory cytokines released after the topical application of DNFB. Extensive remodeling results in the formation of interendothelial gaps, which are necessary for cellular emigration. This requires cell movement that may involve increased turnover of plasma membrane and therefore endocytosis (Jain et al., 1997), a process that could result in greater uptake of vesicles in the blood and vesicles trapped at the site as a result of extravasation. Consequently, the sequence of events is consistent with ODN delivery to the site followed by endothelial cell processing, the result being reduced ICAM-1 expression that inhibits further leukocyte adhesion and therefore an escalation of the inflammation response. In this regard, less accumulation of liposome-encapsulated murine ICAM ODN (ISIS 3082) was observed in DNFB-challenged ears compared with the encapsulated human ICAM ODN (ISIS 2302) sequence (data not shown). This was presumably because hyperpermeability of the endothelium did not develop at all, or was short-lived, because the inflammatory response was attenuated by the activity of the liposome-encapsulated murine ICAM-1 ODN.

Cellular targets for antisense specific activity also include dermal antigen-presenting cells (APCs). Mice sensitized to DNFB have memory T cells ready to release proinflammatory cytokines when DNFB hapten antigens are appropriately presented again. This requires T cell binding to major histocompatibility complex proteins and the participation of costimulatory factors, including APC expression of ICAM-1 (van de Stolpe and van der Saag, 1996). Dermal APC (Langerhans cells and macrophages) can be expected to phagocytose vesicles containing ICAM-1-specific ODN, especially because the vesicles are rapidly and nonspecifically opsonized by a surface coating of plasma IgG shortly after administration, which increases binding to Fc receptors on the surface of APCs (Semple et al., 1998). It is possible that the encapsulated ODN inhibits ICAM-1 expression on the surface of dermal APCs, thus preventing T-cell activation through a lack of costimulation. Experiments are underway to further clarify the site of action. It is important to note, however, that the murine ICAM-1 ODN sequence does not contain a CpG motif, whereas both inactive control ODN do (see Experimental Procedures for sequences). This supports the conclusion that the anti-inflammatory effects observed for encapsulated ISIS 3082 are not the result of nonantisense mitogenic effects (Chu et al., 1997).

In summary, we have demonstrated that ISIS 3082, a PS ODN specific for murine ICAM-1, is significantly more potent as an anti-inflammatory agent in a CHS acute inflammation model, when it is encapsulated in EPC/CH LUV. Although the formulation process is inefficient, the extent of the increase in activity justifies further investigation into the application of lipid-based delivery systems for antisense drugs.