Interaction with Blood Components Plays a Crucial Role in Asialoglycoprotein Receptor-Mediated in Vivo Gene Transfer by Galactosylated Lipoplex

For effective and safe in vivo gene transfer, the development of targeted gene delivery systems is a promising approach. To achieve targeted gene delivery to hepatocytes, galactose has been shown to be a promising targeting ligand to hepatocytes (liver parenchymal cells) because these cells possess a large number of asialoglycoprotein receptors that recognize the galactose units on the glycoproteins or synthetic galactosylated carriers (Kawakami et al., 2002). Recently, we have developed several types of macromolecular and particulate gene carriers for hepatocyte-selective gene transfection in vivo (Kawakami et al., 2000; Fumoto et al., 2003a, 2004; Morimoto et al., 2003). These include galactosylated cationic liposomes containing Gal-C4-Chol, which can be efficiently recognized by asialoglycoprotein receptors in hepatocytes in vivo (Kawakami et al., 2000; Fumoto et al., 2004). However, a number of possible barriers are associated with in vivo gene delivery (Yang and Huang, 1997; Li et al., 1999; Sakurai et al., 2001a; Fumoto et al., 2003b). Detailed information regarding these barriers is needed to allow the rational design of effective gene carriers.

When galactosylated liposome/pDNA complex (lipoplex) was injected into the portal vein of mice, most of it was taken up by the liver (Kawakami et al., 2000). However, the level of in vivo gene expression was not as high as that expected from the in vitro results. Thus, there must be several barriers associated intrinsically with in vivo situations, such as convective blood flow in the liver, passage through the sinusoids, and tissue interactions. To elucidate these barrier processes, we investigated the hepatic disposition profiles of galactosylated lipoplex in rat liver perfusion experiments (Fumoto et al., 2003b), which allowed us to determine the uptake characteristics of a range of substances under different experimental conditions with the structure of the liver remaining intact. In our study, we demonstrated that the penetration of the galactosylated lipoplex through the hepatic fenestrated endothelium to the parenchymal cells was greatly restricted in perfused rat liver (Fumoto et al., 2003b).

It has been reported that lipoplex is able to interact with various types of biological components (e.g., serum proteins) because of their strong positive charge (Mclean et al., 1999; Sakurai et al., 2001b). The presence of serum proteins has been also thought to be a limiting factor for in vitro transfection by lipoplex. Understanding the interaction with the blood cells as well as serum proteins is crucial for the successful development of an effective gene delivery vector. However, the effects of interaction between the galactosylated lipoplex and blood components on asialoglycoprotein receptor-mediated gene transfer have not been well documented. Because the galactosylated lipoplex must pass through the endothelial cell barriers to reach the hepatocytes, the interaction between blood components and galactosylated lipoplex needs to be examined in detail.

In this study, we evaluated the effects of blood components (whole blood and serum) on physicochemical properties, the in situ and in vivo disposition, and the in vivo transfection efficiency of galactosylated lipoplex.

Materials and Methods

Materials.N-(4-Aminobutyl)carbamic acid tert-butyl ester and DOTMA were obtained from Tokyo Chemical Industry Co. Ltd. (Tokyo, Japan). Chol and Clear-Sol I were obtained from Nacalai Tesque (Kyoto, Japan), and Soluene 350 was purchased from PerkinElmer Life and Analytical Sciences (Boston, MA). Cholesteryl chloroformate and collagenase type IA were obtained from Sigma-Aldrich (St. Louis, MO). Rh-DOPE was purchased from Avanti Polar Lipids (Alabaster, AL). [α-³²P]dCTP (3000 Ci/mmol) was obtained from GE Healthcare (Little Chalfont, Buckinghamshire, UK). Galactosylated bovine serum albumin (Gal-BSA) as a ligand of asialoglycoprotein receptors was synthesized as described in our earlier study (Nishikawa et al., 1995). All other chemicals were of the highest purity available.

Animals. Female 5-week-old ICR mice (20-23 g) and male Wistar rats (170-210 g) were purchased from the Shizuoka Agricultural Cooperative Association for Laboratory Animals (Shizuoka, Japan). All animal experiments were carried out in accordance with the Principles of Laboratory Animal Care as adopted and promulgated by the National Institutes of Health (Bethesda, MD) and the Guidelines for Animal Experiments of Kyoto University.

Construction and Preparation of pDNA. pCMV-luciferase was constructed by subcloning the HindIII/XbaI firefly luciferase cDNA fragment from pGL3-control vector (Promega, Madison, WI) into the polylinker of pcDNA3 vector (Invitrogen, Carlsbad, CA). pDNA was amplified in the Escherichia coli strain DH5α, isolated, and purified using a QIAGEN Endofree Plasmid Giga Kit (QIAGEN GmbH, Hilden, Germany). Purity was confirmed by 1% agarose gel electrophoresis followed by ethidium bromide staining, and the pDNA concentration was measured by UV absorption at 260 nm. The pDNA for in vivo distribution and in situ liver perfusion experiments was labeled with [α-³²P]dCTP by nick translation (Sambrook et al., 1989).

Synthesis of Gal-C4-Chol. Gal-C4-Chol was synthesized as reported previously (Kawakami et al., 1998). In brief, cholesteryl chloroformate and N-(4-aminobutyl)carbamic acid tert-butyl ester were reacted in chloroform for 24 h at room temperature. A solution of trifluoroacetic acid and chloroform was added dropwise, and the mixture was stirred for 4 h at 4°C. The solvent was evaporated to obtain N-(4-aminobutyl)-(cholesten-5-yloxyl)formamide, which was then combined with 2-imino-2-methoxyethyl-1-thiogalactoside, and the mixture was stirred for 24 h at 37°C. After evaporation, the resultant material was suspended in water, dialyzed against distilled water for 48 h (12 kDa cut-off dialysis tubing), and then lyophilized.

Preparation of Galactosylated Cationic Liposomes. Mixtures of DOTMA, Chol, and Gal-C4-Chol were dissolved in chloroform at a molar ratio of 2:1:1 for galactosylated liposomes, vacuum-desiccated, and resuspended in sterile 5% dextrose solution at a concentration of 4 mg of total lipids/ml. The suspension was sonicated for 3 min, and the resulting liposomes were extruded 10 times through double-stacked 100-nm polycarbonate membrane filters.

Preparation of Galactosylated Lipoplex. Four hundred and twenty microliters of 286 μg/ml pDNA in 5% dextrose solution was mixed with an equal volume of galactosylated cationic liposomes at 1657 μg/ml and incubated for 30 min. The mixing ratio of liposomes and pDNA was expressed as a (±)-charge ratio, which is the molar ratio of cationic lipids to pDNA phosphate residues (Yang and Huang, 1997). A charge ratio of unity was obtained with 2.52 μg of total lipid/μg pDNA for galactosylated liposomes in this study. As far as the charge ratio was concerned, we selected a charge ratio of +2.3 for all experiments to obtain the most effective transfection activity for receptor-mediated gene transfer (Kawakami et al., 2000, 2004) and to prevent any effect of free liposomes (Eastman et al., 1997; Sakurai et al., 2001a). The particle size and ζ-potential of the galactosylated lipoplex were measured using a dynamic light-scattering spectrophotometer (LS-900; Otsuka Electronics, Osaka, Japan) and a laser electrophoresis ζ-potential analyzer (LEZA-500T; Otsuka Electronics), respectively.

Preparation of Serum and Whole Blood. Mouse or rat serum was prepared by the method of Sakurai et al. (2001a). In brief, mouse serum was isolated from fresh whole blood obtained from ICR mice. Blood was collected from the vena cava under anesthesia without heparin treatment and allowed to stand for 3 h at 37°C and then overnight at 4°C. Serum was collected after centrifugation. Inactivated serum was prepared by heating serum for 30 min at 56°C. Whole blood was collected in a heparinized syringe from ICR mice. An erythrocyte suspension was prepared as described in a previous report (Senior et al., 1991) by washing whole blood three times with phosphate-buffered saline (pH 7.4).

In Vivo Transfection Experiments. Before intraportal injection, galactosylated lipoplex was incubated with blood components for 5 min at 37°C. Mice were anesthetized by intraperitoneal administration of 50 mg/kg pentobarbital sodium. An incision was made in the abdomen, and the portal vein was exposed. The lipoplex preincubated with blood components was injected into the portal vein at a volume of 15 ml/kg, and the abdomen was closed with wound clips. Liver samples were taken 6 h after injection, and each sample was homogenized with lysis buffer (0.1 M Tris/HCl containing 0.05% Triton X-100 and 2 mM EDTA, pH 7.8). After three cycles of freezing and thawing, the homogenates were centrifuged at 10,000g for 10 min at 4°C. Twenty microliters of each supernatant was mixed with 100 μl of luciferase assay solution (Picagene; Toyo Ink Mfg. Co. Ltd., Tokyo, Japan), and the light produced was immediately measured using a luminometer (Lumat LB 9507; Berthold Technologies, Bad Wildbad, Germany). The protein content of the samples was determined using a protein quantification kit (Dojindo Molecular Technologies Inc., Gaithersburg, MD). For evaluation of the intrahepatic localization of gene expression, the luciferase activities in the liver parenchymal (PC) and nonparenchymal cells (NPC) were independently determined after centrifugal separation of PC and NPC in collagenase-digested liver as previously described (Kawakami et al., 2000). In the inhibition experiments involving hepatic transfection, mice received intravenous injections of 20 mg/kg Gal-BSA 1 min before the intraportal injection of the lipoplex.

In Vivo Distribution Study.³²P-Labeled galactosylated lipoplex preincubated with blood components was injected into the portal vein of mice at a volume of 15 ml/kg. At each collection time point, blood was collected from the vena cava and mice were killed at the end of the experiment. The liver, kidneys, spleen, heart, and lungs were removed, washed with saline, blotted dry, and weighed. Ten microliters blood and a small amount of each tissue were digested with 0.7 ml of Soluene-350 by incubating overnight at 45°C. After digestion, 0.2 ml of isopropanol, 0.2 ml of 30% hydroperoxide, 0.1 ml of 5 M HCl, and 5.0 ml of Clear-Sol I were added. The samples were stored overnight, and the radioactivity was measured in a scintillation counter (LSA-500; Beckman Coulter, Inc., Fullerton, CA).

Calculation of Organ Uptake Clearance. Tissue distribution data were evaluated using organ uptake clearances as reported previously (Takakura et al., 1987). In brief, the tissue uptake rate can be described by the following equation, where X_t is the amount of ³²P-labeled galactosylated lipoplex in the tissue at time t, CL_uptake is the tissue uptake clearance, and C_b is the blood concentration of ³²P-labeled galactosylated lipoplex. Integration of eq. 1 gives the following, where area under the curve (AUC)_(0∼t) represents the area under the blood concentration time curve from time 0 to t. The CL_uptake value can be obtained from the initial slope of a plot of X_t versus AUC_(0∼t).

Liver Perfusion Experiments and Pharmacokinetic Analysis. In situ liver perfusion studies were carried out as reported previously (Nishida et al., 1989; Fumoto et al., 2003b). In brief, the portal vein was catheterized with a polyether nylon catheter (SUR-FLO i.v. catheter, 16 G′2″; Terumo Co., Tokyo, Japan) and immediately perfused with Krebs-Ringer-bicarbonate buffer supplemented with 10 mM glucose (oxygenated with 95% O₂ and 5% CO₂, adjusted to pH 7.4 at 37°C). The perfusate did not contain serum proteins and blood cells. The perfusate was circulated using a peristaltic pump (SJ-1211; Atto Bioscience, Tokyo, Japan) at a flow rate of 13 ml/min. After a stabilization period of 25 min, ³²P-labeled galactosylated lipoplex preincubated with rat serum or whole blood (30 μg of pDNA/300 μl) was administered via the portal vein using a six-position rotary valve injector (Type 50 Teflon rotary valves; Rheodyne Inc., Cotati, CA). After the addition of 5 ml of Clear-Sol I, the radioactivity of the effluent perfusate was measured in a scintillation counter (LSA-500, Beckman Coulter, Inc., CA). The outflow patterns were analyzed by statistical moment analysis. In brief, the AUC and mean residence time (MRT) were calculated as follows: where t is the time and C is the concentration of ³²P-labeled galactosylated lipoplex. The moments can be calculated by numerical integration using a linear trapezoidal formula and extrapolation to infinite time based on a monoexponential equation (Yamaoka et al., 1978). The t values were corrected for the lag time of the catheter. The recovery ratio (F) and extraction ratio (E) were derived from F = AUC · Q (flow rate) and E = 1 - F, respectively.

The outflow patterns were also analyzed based on a two-compartment dispersion model, where sinusoidal and binding compartments were considered. The mass balance equations involving the axial dispersion in the sinusoidal space are as follows: where C_S(t, z) and C_B(t, z) are the concentrations of drug in the sinusoidal space and binding compartment, respectively, D is the dispersion coefficient, ϵ is the volume ratio of the binding compartment to the sinusoidal space in the liver, k₁₂ and k₂₁ are the forward and backward partition rate constants between the sinusoidal space and binding compartment, k_int is the first-order internalization rate constant from the binding compartment to the intracellular space, v is the linear flow velocity of the perfusate, t is time, and z is the axial coordinate in the liver. The initial and boundary conditions are given as follows: where M is the amount of drug injected into the liver, Q is the flow rate of the perfusate, and f_I(t) has the dimension of the reciprocal of time. Taking the Laplace transform with respect to t, rearranging, substituting the length of the sinusoidal space L with z, and introducing the cross-sectional area of the sinusoidal space A, the following image equation is obtained as follows: where C̃_S(s) and f̃₁(s) denote the Laplace transform of the concentration in the venous outflow and input function f_I(t), respectively, D_C is the corrected dispersion coefficient (D_C = D · A²), V_S is the sinusoidal volume (= L · A), and the flow rate Q is equal to A × v.

Each parameter (D_C, k₁₂, k₂₁, k_int, and V_S) was calculated by curve-fitting of the Laplace-transformed equation to the experimental venous outflow pattern using a nonlinear least-squares program with a fast inverse Laplace transform algorithm MULTI (FILT) (Yano et al., 1989). The damping Gauss-Newton method with no constraint was used for curve-fitting with the MULTI algorithm. Herein, f_I(t) was assumed to be a Δ function, because the lipoplexes were rapidly injected using a six-rotary valve injector.

For evaluation of the intrahepatic localization of the amounts taken up, 30 min after injection of ³²P-labeled galactosylated lipoplex into the isolated perfused liver, the radioactivities in the liver PC and NPC were separately determined after centrifugal separation of PC and NPC in collagenase-digested liver as described previously (Fumoto et al., 2003b).

Observation of Dissociation of pDNA and Lipids from Lipoplex Induced by Mixing with Blood. Carboxy-fluorescein labeling of pDNA was performed using the Label IT Fluorescein Nucleic Acid Labeling Kit (Mirus Co., Madison, WI). Liposomes were labeled with Rh-DOPE at 2% (mol/mol) total lipid. Fluorescent-labeled lipoplex was then prepared by mixing fluorescein-labeled pDNA with rhodamine-labeled liposomes as described above. To observe the dissociation of pDNA and lipids from lipoplex induced by whole blood, fluorescent-labeled lipoplex was mixed with 30% blood and subsequently centrifuged and the erythrocytes were washed twice with phosphate-buffered saline. To investigate the effect of preincubation with serum, fluorescent-labeled lipoplex was mixed with 15% serum for 5 min and then mixed with 15% erythrocyte suspension to adjust the hematocrit. Lipoplex integrity was assessed by fluorescent resonance energy transfer (FRET) from fluorescein-pDNA to rhodamine lipids. Precipitates (i.e., lipoplex bound to blood cells) were mounted on glass slides, covered by slips, and observed by confocal laser-scanning microscope (MRC 1024; Bio-Rad, Hercules, CA). On the other hand, galactosylated lipoplex in the supernatant was measured by spectrofluorophotometry (RF540; Shimadzu Co., Kyoto, Japan). The excitation wavelengths were 480 and 550 nm for fluorescein pDNA and Rh-DOPE.

Agarose Gel Electrophoresis. The pDNA stability of the galactosylated lipoplex in blood was determined by agarose gel electrophoresis (Harvie et al., 2000). The galactosylated lipoplex was preincubated with blood components at 37°C. After incubation, pDNA was extracted from the mixture by phase separation using phenol/chloroform/isoamyl alcohol (25:24:1) followed by precipitation with ethanol. Precipitated pDNA was redissolved with Tris borate-EDTA buffer, pH 8.0, and subjected to agarose gel electrophoresis. Densitometric analysis was performed using a commercially available computer program [CS analyzer; Atto Bioscience and Rise Corporation, Sendai, Japan].

Serum IFN-γ Concentration Measurement. Blood was collected 6 h after intraportal injection of galactosylated lipoplex, simultaneously with the in vivo transfection experiment, and was subsequently allowed to stand for 3 h at 4°C. Serum was collected after centrifugation and frozen at -80°C until measurement. The serum IFN-γ concentration was determined using commercially available enzyme-linked immunosorbent assay kits (OptEIA mouse IFN-γ set; BD Biosciences, San Jose, CA).

Statistical Analysis. Statistical comparisons were performed by an unpaired Student's t test for two groups or Dunnett's test for multiple comparisons with a control group. Statistical comparisons in the transfection experiments and serum IFN-γ measurements were performed by the Mann-Whitney test for two groups or Steel's test for multiple comparisons with a control group because of the heterogeneity of the variance evaluated by the F test and Bartlett test, respectively.

Results

Effect of Blood Components on the Physicochemical Characteristics of Galactosylated Lipoplex. To investigate the effect of serum protein on the physicochemical characteristics of galactosylated lipoplex, the particle size and ζ-potential were measured after exposure to mouse serum because these parameters affect the hepatic disposition of galactosylated lipoplex (Fumoto et al., 2003b). Mixing with serum [30% (v/v)] significantly enlarged the particle size of the lipoplex (Table 1). The ζ-potential of the galactosylated lipoplex was significantly reduced by mixing with serum, and the charge became negative, suggesting that negatively charged serum proteins covered much of the galactosylated lipoplex surface. These results are consistent with our previous report about the conventional lipoplex (Sakurai et al., 2001a).

After mixing galactosylated lipoplex with erythrocyte suspension, hemagglutination was observed. However, mixing galactosylated lipoplex with whole blood did not induce any obvious hemagglutination (data not shown), suggesting that the presence of serum components prevents hemagglutination. Recently, Eliyahu et al. (2002) also reported similar results with the conventional lipoplex. Thus, galactosylated lipoplex was mixed with whole blood to evaluate the effect of blood cells on transfection activity.

Transfection Activities of Galactosylated Lipoplex Preincubated with Blood Components. To study the effect of blood components on in vivo transfection activity, galactosylated lipoplex was preincubated with serum or whole blood before administration and the transfection activities in the liver were evaluated 6 h after intraportal injection to mice. When the galactosylated lipoplex was preincubated with serum, the hepatic transfection activities were enhanced approximately 20- to 70-fold (Fig. 1A). However, incubation with whole blood [30% (v/v)] reduced the transfection activity in the liver by 97% (Fig. 1B). These results show that the interaction with blood cells markedly inhibits the hepatic transfection activity of galactosylated lipoplex. In addition, higher concentration of blood components exhibited more marked effect on transfection activity. In in vivo condition, blood volume is larger than the volume of the lipoplex solution. Thus, the result of 30% blood or serum would be more close to in vivo condition than the result of 15% blood components. Therefore, we applied 30% blood components as the experimental condition for further studies.

In Vivo Distribution of Galactosylated Lipoplex Preincubated with Serum. To investigate why the transfection activity of galactosylated lipoplex was enhanced by incubation with serum, the biodistribution of galactosylated lipoplex preincubated with serum was evaluated using ³²P-labeled galactosylated lipoplex (Fig. 2). However, similar distribution patterns in the liver and other organs were observed with or without incubation with serum and, accordingly, this result did not correlate with the enhanced hepatic transfection activity of the galactosylated lipoplex preincubated with serum.

In Situ Hepatic Disposition of Galactosylated Lipoplex Preincubated with Blood Components. We performed pharmacokinetic analyses using a single-pass rat liver perfusion experiment that allowed us to evaluate the local disposition of the carrier systems. The venous outflow profile of the complexes (Fig. 3) after a bolus input into the isolated perfused liver was analyzed by a two-compartment dispersion model to quantitatively evaluate the difference in each kinetic process. Table 2 shows the results of moment analysis for the venous outflow pattern of galactosylated lipoplex preincubated with rat blood components. Incubation with serum significantly reduced the extraction ratio of galactosylated lipoplex, whereas the effect of incubation with rat whole blood on the extraction ratio was minor. However, the MRT of the galactosylated lipoplex in the liver was significantly reduced after incubation with serum. To clarify each kinetic process in perfused liver, the binding and internalization processes were evaluated by a two-compartment dispersion model. As shown in Fig. 3, the simulation curves using the model were in good agreement with the observed data. The validation of two-compartment dispersion model was discussed in our previous report (Fumoto et al., 2003b). The association rate (k₁₂) of the galactosylated lipoplex was reduced by half after incubation with serum (Table 3). However, the dissociation rate (k₂₁) was reduced by 80% by serum. As a consequence, the tissue binding affinity (k₁₂/k₂₁) was 2.6-fold higher than that without preincubation. On the other hand, incubation with whole blood significantly increased the dissociation rate while maintaining the high association rate. As a result, the tissue binding affinity (k₁₂/k₂₁) was reduced by 60% after incubation with whole blood.

Assessment of Galactosylated Lipoplex Stability in Blood. To evaluate the differences in galactosylated lipoplex stability with or without incubation with serum, we performed FRET analysis. Galactosylated lipoplex was labeled with both fluorescein-labeled pDNA (green fluorescence) and rhodamine-labeled lipid (red fluorescence). A mixture of galactosylated lipoplex and mouse blood components was centrifuged after a 5-min incubation, and subsequently, the erythrocyte compartment and supernatant were examined using confocal laser-scanning microscopy and spectrofluorophotometry, respectively. After incubation with whole blood, the signal of the erythrocyte compartment was found to be red (Fig. 4A), whereas the energy transfer from fluorescein-labeled-pDNA to rhodamine-labeled lipid in the supernatant was reduced (Fig. 4C), suggesting that the pDNA and lipids in the lipoplex had dissociated. On the contrary, preincubation with serum before incubation with erythrocyte suspension did not induce such dissociation of pDNA and lipids in the lipoplex (Fig. 4, B and C). The dissociation of pDNA from cationic liposomes might cause degradation by serum nuclease; therefore, agarose gel electrophoresis was performed to analyze the degradation of pDNA (Fig. 5). Preincubation with serum before mixing with erythrocyte suspension markedly inhibited pDNA degradation, indicating that preincubation with serum stabilized the lipoplex.

Hepatic Cellular Localization of Galactosylated Lipoplex with or without Incubation with Serum. The hepatic cellular localization of ³²P-labeled galactosylated lipoplex was investigated after bolus injection into perfused rat liver (Fig. 6A). When the radioactivity associated with PC and NPC per unit cell number was measured, the PC/NPC ratio for the galactosylated lipoplex with pre-exposure to serum was 1.0, which was comparable with the galactosylated lipoplex without pre-exposure to serum (PC/NPC ratio: 1.1). As for the hepatic cellular localization of transfection activity in mice, gene expression in PC and NPC of galactosylated lipoplex with preincubation with serum was one order of magnitude higher than the values for the lipoplex without preincubation (Fig. 6B).

Inhibition Experiment of Hepatic Transfection Activity of Galactosylated Lipoplex. To confirm whether galactosylated lipoplex exposed to serum is recognized by asialoglycoprotein receptors on hepatocytes, we performed an inhibition experiment involving predosing with Gal-BSA (Fig. 7A). The hepatic transfection activity was significantly inhibited by Gal-BSA, suggesting that the serum protein-bound galactosylated lipoplex is recognized by asialoglycoprotein receptors.

Effect of Inactivation of Serum on Transfection Activity and Biodistribution of Galactosylated Lipoplex. Opsonin activity in serum might affect the in vivo distribution and subsequent transfection activity of the lipoplex. To confirm the effect of opsonin activity in serum, serum was inactivated by heating before incubation with the lipoplex. Figure 7B shows the effect of inactivation of mouse serum on the hepatic transfection activity of the lipoplex 6 h after intraportal injection to mice. Approximately 85% of the hepatic transfection activity of the galactosylated lipoplex was attenuated by inactivation of serum, suggesting that some factor, such as complement component, is involved in the hepatic transfection activity of the galactosylated lipoplex.

To investigate the inactivation of serum further, we evaluated the in vivo distribution after intraportal injection of ³²P-labeled galactosylated lipoplex. Figure 8 shows the blood concentration and tissue accumulation (liver and lung) at the indicated time points. Inactivation with serum significantly reduced the accumulation not only in the liver but also in the lung, whereas an increased blood concentration was observed at the early time points. To estimate the effect of preincubation with inactivated serum on organ uptake of galactosylated lipoplex, we calculated organ uptake clearance (Table 4). The hepatic uptake clearance was the highest of all of the tissues. The hepatic uptake clearance was markedly reduced by incubation with inactivated serum. These differences in distribution caused by inactivation of serum will be involved in the difference seen in the hepatic transfection activity of galactosylated lipoplex preincubated with inactivated serum.

Effect of Blood Components on Serum Cytokine Level after Intraportal Injection of Galactosylated Lipoplex. It is well known that lipoplex induces production of inflammatory cytokines via the unmethylated CpG motif in pDNA (McLachlan et al., 2000). To study the contribution of the interaction with each blood component to cytokine production, we evaluated the serum IFN-γ level at 6 h after intraportal injection of galactosylated lipoplex preincubated with blood components (Fig. 9). Preincubation with serum did not affect the serum IFN-γ level, whereas preincubation with whole blood significantly reduced it. Preincubation with inactivated serum also significantly reduced the serum IFN-γ level, suggesting that an interaction between galactosylated lipoplex and some serum components is involved in inflammatory cytokine production.

Discussion

A number of possible barriers for targeted gene delivery to hepatocytes are thought to be limiting factors for in vivo transfection, including 1) nonspecific interactions with erythrocytes, serum, and nontarget cells, 2) aggregation of lipoplex, and 3) penetration of endothelial cells. It is necessary to discover the fate of the gene carrier in order to improve and/or develop an effective gene carrier system. In the present study, we evaluated the effect of blood components (serum and whole blood) on asialoglycoprotein receptor-mediated in vivo gene transfer using galactosylated lipoplex. Preincubation with serum greatly enhanced the hepatic transfection activity of the lipoplex, whereas preincubation with whole blood reduced it (Fig. 1); this indicates that interaction with blood components plays a crucial role in in vivo gene transfer.

We initially evaluated the effect of the interaction between galactosylated lipoplex and blood components on the basic physicochemical properties. Because particle size is an important factor that determines the delivery efficiency of various particulates to liver parenchymal cells (Rahman et al., 1982; Ogawara et al., 1999), we measured the change in particle size of the lipoplex produced by exposure to serum. As we expected, exposure of the lipoplex to serum increased the particle size (Table 1). However, the particle size of the galactosylated lipoplex after exposure to serum was comparable with the size of the fenestrae (100-200 nm) in the liver sinusoidal endothelium (Wisse, 1970; Wisse et al., 1985; Gatmaitan et al., 1996). Indeed, the intrahepatic distribution (PC/NPC ratio) of galactosylated lipoplex after pre-exposure to serum was similar to that of galactosylated lipoplex without pre-exposure to serum (Fig. 6A). It was also demonstrated that the transfection activity of galactosylated lipoplex after pre-exposure to serum was equally enhanced in both PC and NPC; as a consequence, the PC/NPC ratio of the transfection activity had a similar value (Fig. 6B).

The reduced ζ-potential after incubation with serum (Table 1) suggests that the surface of the galactosylated lipoplex is extensively covered by serum protein, because most serum proteins have negative charges. However, a reduced surface charge is expected to reduce the liver accumulation. In fact, the in situ liver perfusion experiments showed that incubation with serum reduced the extraction ratio (Table 2); however, the extraction ratio (65% of the dose) was high in absolute terms despite a single passage through the perfused liver. Furthermore, two-compartment dispersion model analysis demonstrated that incubation with serum reduced not only the association rate but also the dissociation rate. As a consequence, the tissue binding affinity of galactosylated lipoplex preincubated with serum was increased (Table 3). These results indicate that preincubation with serum reduces the initial amount of lipoplex bound to tissue, although the binding of retained lipoplex to tissue is stronger than that in the case of no incubation. As shown in Fig. 7A, the hepatic transfection activity of galactosylated lipoplex preincubated with serum was significantly inhibited by predosing with Gal-BSA, suggesting that asialoglycoprotein receptor-mediated endocytosis markedly contributes to the high tissue binding affinity. In contrast, incubation with whole blood did not reduce the association rate but increased the dissociation rate in the liver perfusion experiments (Table 3). Under in vivo conditions, dissociation by blood cell would be more marked than in perfused liver because of the abundance of blood cells in the continuous blood flow. This analysis might explain why the liver accumulation of lipoplex without incubation after intraportal injection (Fig. 2) was relatively lower than the extraction ratio in the liver perfusion experiment (Table 2). As a result, the liver accumulation of the lipoplex, with or without incubation with serum, exhibited a similar profile.

We next assessed the stability of the galactosylated lipoplex in blood. It is known that mixing lipoplex with anionic liposomes induces the dissociation of pDNA from the lipoplex (Xu and Szoka, 1996; Sakurai et al., 2000). Blood cells have a negative surface charge; thus, they might induce dissociation of pDNA from the galactosylated lipoplex. After incubation with whole blood, dissociation of pDNA from the galactosylated lipoplex was observed (Fig. 4, A and C). On the contrary, incubation with serum inhibited such dissociation of pDNA from the galactosylated lipoplex (Fig. 4, B and C). In addition, the stability of pDNA in the galactosylated lipoplex was improved by preincubation with serum (Fig. 5). It has been reported that prolonged incubation with serum also induces pDNA dissociation and degradation (Li et al., 1999). Furthermore, we have demonstrated that the presence of blood cells together with galactosylated lipoplex enhances pDNA dissociation and degradation. The injected lipoplex is immediately mixed with blood before contact with the liver; therefore, this result partially supports the enhanced transfection activity of the galactosylated lipoplex. Taking these results into consideration, not only ligand modification (i.e., galactosylation) of the lipoplex but also controlling the stability of the galactosylated lipoplex is important for achieving cell-selective gene transfection under in vivo conditions.

Inactivation of serum before incubation with galactosylated lipoplex significantly reduced both liver accumulation (Fig. 8) and hepatic transfection activity (Fig. 7). It is known that the opsonin activity in serum affects the disposition of liposomes (Chonn et al., 1992; Patel, 1992), suggesting that the disposition of galactosylated lipoplex would also be affected by opsonization. The complement component C1q, factor B, and fibronectin are heat-labile (Hamuro et al., 1978; McManus and Nakane, 1980; Rivedal, 1982); thus, both classical and alternative pathways of complement are no longer able to work after inactivation of serum. As a result, the uptake via recognition of these factors bound to the galactosylated lipoplex was expected to fall. In fact, the liver and lung accumulation of galactosylated lipoplex with inactivated serum was significantly reduced (Fig. 8; Table 4). Barron et al. (1998) reported that complement depletion by intraperitoneal injection of cobra venom factor and anti-C3 antibodies did not affect the distribution and gene expression of lipoplex after intravenous injection. Cobra venom factor depletes factor B and component C3; therefore, other factors apart from factor B and component C3 might be involved in the reduced accumulation of galactosylated lipoplex preincubated with inactivated serum. Further studies are needed to clarify how complement component(s) affect the disposition and transfection activity of the lipoplex.

It is known that lipoplex initiates inflammatory cytokine production via pDNA containing the unmethylated CpG motif (McLachlan et al., 2000). When the galactosylated lipoplex was preincubated with serum, there was no significant change in the serum IFN-γ level; therefore, incubation with serum may be a useful method for enhancing transfection activity without increasing the inflammatory response. On the contrary, incubation with both whole blood and inactivated serum significantly reduced the serum IFN-γ level. Although the mechanism is unclear, this information may useful for development of safe gene carrier with less inflammatory response.

In summary, we have shown that the interaction with blood components plays a crucial role in in vivo gene transfer. Although incubation of galactosylated lipoplex with whole blood reduced the hepatic transfection activity, incubation of the lipoplex with serum enhanced it. The stability of pDNA in blood rather than its in vivo distribution partially explains this difference. In contrast, inactivation with serum reduced the hepatic transfection activity, suggesting that other factor(s) in serum are involved in the hepatic transfection activity of galactosylated lipoplex. Hence, the information in this study will be valuable for the future use, design, and development of galactosylated lipoplex for in vivo asialoglycoprotein receptor-mediated gene transfer.